SMALL ORGAN4 Is a Ribosome Biogenesis Factor Involved in 5.8S Ribosomal RNA Maturation1[OPEN]

Rosa Micol-Ponce,a,2,3 Raquel Sarmiento-Mañús,a,2 Sara Fontcuberta-Cervera,a Adrián Cabezas-Fuster,a Anne de Bures,b,c Julio Sáez-Vásquez,b,c and María Rosa Poncea,4,5 Instituto de Bioingeniería, Universidad Miguel Hernández, Campus de Elche, 03202 Elche, Alicante, Spain Centre National de la Recherche Scientifique, Laboratoire Génome et Développement des Plantes, Unité Mixte de Recherche 5096, 66860 Perpignan, France Universite Perpignan Via Domitia, Laboratoire Génome et Développement des Plantes, Unité Mixte de Recherche 5096, 66860 Perpignan, France

The eukaryotic 80S cytoplasmic ribosome was first described in the mid-1950s (Palade, 1955) and is now considered a paradigm for our understanding of complex molecular machines (Dinman, 2009). The structure and biogenesis of the 80S ribosome are similar in all eukaryotes. The ribosomal RNAs (rRNAs) and ribosomal proteins (RPs) involved in these processes are also highly conserved. The function of the 80S ribosome in mRNA translation exhibits high-level evolutionary conservation, similar to that of other essential cellular functions (Wilson and Doudna Cate, 2012).
Approximately 80 RPs and four rRNAs form the 80S ribosome. These rRNAs are produced via processing of the primary transcripts of the repeated 5S ribosomal DNA (rDNA) genes in all eukaryotes, and 47S, 45S, and 35S rDNA genes in animals, plants, and yeast, respectively. The processing of the 5S pre-rRNA primary transcript to produce mature 5S rRNA is a straightforward process. By contrast, the processing of 47S/45S/ 35S pre-rRNA to produce mature 5.8S and 18S rRNAs in all eukaryotes, mature 25S rRNA in yeast and plants, and mature 28S rRNA in animals is a complex, multistep process (Supplemental Figs. S1-S3; Wilson and Doudna Cate, 2012). The biogenesis of the 80S ribosome in eukaryotes is best characterized in the yeast Saccharomyces cerevisiae. Indeed, the individual functions of many plant and animal ribosome biogenesis factors (RBFs) and their interactions have been inferred based on their homology with yeast putative orthologs (Tomecki et al., 2017;Sáez-Vásquez and Delseny, 2019).
The exosome, an evolutionarily conserved complex in eukaryotes, has 39 to 59 exoribonuclease activity, which is required for the metabolism of many RNA species, such as mRNAs, rRNAs, small nucleolar RNAs, and small nuclear RNAs. The exosome facilitates the degradation, surveillance, precursor processing, and degradation of processing by-products of these RNA species (Kilchert et al., 2016). In yeast, Nucleolar protein53 (Nop53; Thomson and Tollervey, 2005) acts as an adaptor protein that targets the mRNA transport4 (Mtr4) ATP-dependent RNA helicase (Kilchert et al., 2016) to preribosomal particles for exosome processing of the 39 end of the 7S pre-rRNA, a 5.8S rRNA precursor (Supplemental Fig. S1; Thoms et al., 2015). The loss of Nop53 function perturbs 5.8S and 25S rRNA biogenesis but not that of 18S rRNA, leading to severely stunted growth and an imbalance in the 40S:60S ribosomal subunit ratio (Granato et al., 2005;Sydorskyy et al., 2005;Thomson and Tollervey, 2005). Yeast Mtr4, also termed Dependent on eIF4B, is also essential; the loss of Mtr4 function reduces 5.8S rRNA production (de la Cruz et al., 1998).
Knockdown of human (Homo sapiens) MTR4 leads to the strong accumulation of 34S pre-rRNA (an aberrant early precursor of 18S rRNA, produced by the inhibition of the cleavage at the A9, A 0 , and 1 endonucleolytic sites; Supplemental Fig. S2) and the overaccumulation of 26S and 18S-E pre-rRNAs (late precursors of 18S rRNA; Supplemental Fig. S2; Tafforeau et al., 2013). Arabidopsis (Arabidopsis thaliana) mtr4 mutants overaccumulate precursors of the 18S and 5.8S rRNAs, but the levels of the mature rRNAs are not affected in these mutants (Supplemental Fig. S3; Lange et al., 2011). The Arabidopsis ortholog of yeast Nop53 is encoded by AT2G40430, which was previously named SMALL ORGAN4 (SMO4) by Zhang et al. (2015); these authors characterized SMO4 as a nuclear protein that affects cell proliferation. However, to date, a role for SMO4 in ribosome biogenesis has not been established.
The human ortholog of yeast Nop53 is the nucleolar protein glioma tumor-suppressor candidate region gene2 (GLTSCR2; Lee et al., 2012), which is also known as GSCR2, p60, and protein interacting with carboxyl terminus1 (PICT1). GLTSCR2 is an essential protein whose loss of function stabilizes p53 and induces p53dependent G1 cell cycle arrest and apoptosis (Sasaki et al., 2011). GLTSCR2 is also considered to be a key regulator of p53-mediated responses to the nucleolar stress caused by impaired ribosomal biogenesis or function (Suzuki et al., 2012). Knockdown of GLTSCR2 causes the overaccumulation of 47S pre-rRNA, the primary transcript of 47S rDNA genes, and of 18S-E pre-rRNA, a late intermediate of 18S rRNA maturation, as well as 12S pre-rRNA, a 5.8S rRNA precursor (Supplemental Fig. S2; Tafforeau et al., 2013).
Here, we investigated the action and interactions of Arabidopsis SMO4. The genetic interactions of smo4 alleles found in this study confirm the functional relationship of SMO4 with MAS2 and MTR4. The morphological, cytological, and molecular phenotypes caused by the loss of function of SMO4 shed light on the role of SMO4 in ribosome biogenesis, specifically in 5.8S rRNA maturation, as it has been shown for its yeast and human orthologs.
In yeast, an N-terminal LFXwD arch interaction motif (AIM; where X is any amino acid and w is a hydrophobic amino acid) of Nop53 interacts with the arch domain of the exosome cofactor Mtr4 (Jackson et al., 2010;Thoms et al., 2015). The AIM in yeast Nop53, human GLTSCR2, and Arabidopsis SMO4 is well conserved, despite the relatively low shared identities between the full-length proteins. The AIM sequence of Arabidopsis SMO4 and yeast Nop53 is LFHVD, and that of human GLTSCR2 is LFFVD (where V is Val, a hydrophobic residue; Supplemental Fig. S4; Thoms et al., 2015). This motif is present in species representative of all major plant clades, as shown in a multiple sequence alignment of plant putative Nop53 orthologs, where w is V except for soybean (Glycine max; which harbors E [Glu], a charged residue, at this position; Supplemental Fig. S5). Arabidopsis MTR4 also contains an arch domain (Jackson et al., 2010;Lange et al., 2014). Mutational analysis in yeast has shown that the arch domain of Mtr4 is necessary for the 39 processing of 7S pre-rRNA, to yield mature 5.8S rRNA, and in the degradation of the by-product of 59-ETS (Supplemental Fig.  S1; Jackson et al., 2010). The full-length MTR4 is more conserved than NOP53, including the arch domain, which is not the most highly conserved region among eukaryotic MTR4 proteins (Supplemental Figs. S6 and S7;Lange et al., 2014).

Isolation of Novel Alleles of SMO4
In yeast, the lack of Nop53 function is lethal or associated with significant growth defects (Granato et al., 2005;Sydorskyy et al., 2005;Thomson and Tollervey, 2005). The smo4-1 mutant, the first smo4 mutant identified in Arabidopsis, was isolated based on its phenotype: reduced plant size resulting from delayed cell cycle progression during leaf development, which eventually reduces cell number (Zhang et al., 2015). The smo4-1 allele carries a 14-bp deletion in the 12th exon of SMO4 (Fig. 1A), which is predicted to produce a truncated protein 12 amino acids shorter than the wild-type version. In two insertional mutants, the transfer DNA (T-DNA) insertions disrupted SMO4 in its 14th exon (smo4-2; SALK_012561; Zhang et al., 2015) and 12th intron (smo4-3; SALK_071764; Micol- Ponce et al., 2018;Fig. 1A). Under our growth conditions, smo4-2 and smo4-3 plants, observed 14 days after stratification (das), had very mild morphological phenotypes, with slightly dentate and pointed rosette leaves; these mutant plants were nearly indistinguishable from the wild type at bolting (Fig. 1, B-D). These smo4 alleles might be hypomorphic, based on their weak mutant phenotypes and because the T-DNA insertions in smo4-2 and smo4-3 and the deletion mutation in smo4-1 are located near the 39 end of the AT2G40430 coding region (Fig. 1A). Hence, we also included in our study the GABI_082H04 line (hereafter referred to as smo4-4), which carries a smo4 allele with a T-DNA insertion disrupting the first exon of SMO4 (Fig. 1, A and E). Homozygous smo4-2, smo4-3, and smo4-4 plants displayed indistinguishable phenotypes, suggesting that each mutant allele confers the same loss of function of SMO4 and that they all are likely to be null. The SMO4.2 (AT2G40430.2) gene model, which corresponds to the splice variant that is predicted to produce the largest SMO4 protein, is shown. The start (ATG) and stop (TGA) codon positions are also indicated. Black and white boxes represent exons and 59 and 39 untranslated regions, respectively. Lines between boxes represent introns, and triangles indicate T-DNA insertions. Red arrows indicate the positions of the 14-bp deletion of smo4-1 and the single-base substitution of den2. B to M, Rosette morphological phenotypes of Col-0 (B), smo4-2 (C), smo4-3 (D), smo4-4 (E), Ler (F), den2 (G), smo4-2 SMO4 pro :SMO4 (H), smo4-3 SMO4 pro :SMO4 (I), den2 SMO4 pro :SMO4 (J), smo4-2 SMO4 pro :SMO4:GFP (K), smo4-3 SMO4 pro :SMO4:GFP (L), and den2 SMO4 pro :SMO4:GFP (M) plants. All plants were homozygous for the mutant alleles and the transgenes shown. Photographs were taken 14 das. Bars 5 3 mm. N to P, Confocal laser-scanning micrographs of cells from the root elongation zone of plants homozygous for the SMO4 pro :SMO4:GFP transgene in the Col-0 background. Fluorescent signals correspond to Hoechst 33342 (N), GFP (O), and their overlay (P). Bars 5 5 mm.
In a large-scale ethyl methanesulfonate (EMS) mutant screening performed in the laboratory of José Luis Micol, 58 denticulata (den) pointed-leaf mutants were isolated and fell into 17 complementation groups (Berná et al., 1999). A single den2 mutant allele was subsequently isolated and mapped at low resolution to chromosome 2 (Robles and Micol, 2001). Here, we delimited a 3.8 Mb genomic region candidate to harbor the den2 mutation, using iterative linkage analysis to molecular markers (Supplemental Table S1), and performed whole-genome sequencing of DNA from den2 plants. This analysis revealed a C→T base substitution in 12th exon of AT2G40430, which is placed within the candidate genomic region. This base change is predicted to be a nonsense mutation (Arg372→stop), producing a truncated protein 71 amino acids shorter than the wild-type protein (Fig. 1A), losing the most conserved region among yeast, human, and Arabidopsis NOP53 orthologs (Supplemental Fig. S4), which suggests that den2 is a null allele of SMO4. A smo4-3 3 den2 cross confirmed that these two mutants are allelic (Supplemental Fig. S8).
To further examine if the mutant alleles of SMO4 under study are null, we used several primer pairs, each flanking one of the smo4 insertions or hybridizing in the T-DNA and the SMO4 gene, to perform semiquantitative reverse transcription PCR (RT-PCR) analyses of smo4-2, smo4-3, smo4-4, and den2 transcription (Supplemental Fig. S9; Supplemental Table S2). We detected transcripts including exons located upstream of the insertions of smo4-2 and smo4-3, and downstream of those of smo4-3 and smo4-4 (Supplemental Fig. S9B). Our results for smo4-2 are in accordance with those obtained by Zhang et al. (2015). In addition, we obtained bands with sizes that may correspond to chimeric transcripts, including T-DNA and SMO4 sequences, in all smo4 mutants (Supplemental Fig. S9B). Since these alleles cause almost identical phenotypes and are predicted to produce aberrant proteins, irrespective of the positions of the mutations at the beginning (the smo4-4 insertion) or the end (the smo4-2 and smo4-3 insertions and the den2 point mutation) of the coding region, we concluded that probably all of them carry null alleles of SMO4. The morphological phenotype of den2 (Fig. 1, F and G) is stronger than those of the smo4 mutants ( Fig. 1, C-E); this is likely to be due to the Landsberg erecta (Ler) genetic background of den2, as already shown for other mutants, including mutants affected in genes related to ribosome biogenesis (Horiguchi et al., 2011;Rosado et al., 2012).
To confirm that the mutant phenotypes of smo4 and den2 are caused by a lack of SMO4 activity, we created the SMO4 pro :SMO4 construct, which was transferred into smo4-2, smo4-3, den2, Columbia-0 (Col-0), and Ler plants. The SMO4 pro :SMO4 transgene complemented the mutant phenotypes caused by the smo4 and den2 mutations (Fig. 1, H-J). We also created the 35S pro :SMO4 construct, which had no visible morphological effects when transferred into the Col-0 and Ler wild types (Supplemental Fig. S10).

SMO4 Localizes to the Nucleolus and Nucleoplasm
In addition to 5.8S rRNA maturation (Supplemental Fig. S1), yeast Nop53 is required for the nuclear export of the 60S preribosomal particle, which matures in the cytoplasm into the 60S subunit of the 80S ribosome (Thomson and Tollervey, 2005). To play this dual role in ribosome biogenesis, we reasoned that SMO4 must be present in both the nucleolus and nucleoplasm, similar to yeast Nop53 (Sydorskyy et al., 2005;Thomson and Tollervey, 2005). To test this hypothesis, we generated the SMO4 pro :SMO4:GFP construct, which produced a functional protein that complemented the mutant phenotypes of smo4-2, smo4-3, and den2 ( Fig. 1, K-M). SMO4 is a nuclear protein (Zhang et al., 2015), but whether it localizes to the nucleolus and/or nucleoplasm is unknown. To visualize the nucleus, we stained the roots of Col-0 SMO4 pro :SMO4:GFP plants with Hoechst 33342, a dye that strongly binds to doublestranded DNA but not to RNA, the primary nucleic acid in the nucleolus. GFP fluorescence was detected in a diffuse pattern in the nucleolus and nucleoplasm (Fig. 1,. In agreement with these findings, SMO4 has also been identified in the nucleolar proteome of Arabidopsis (Montacié et al., 2017).
To ascertain if SMO4 genetically interacts with MTR4 and MAS2, we crossed smo4-3 to mtr4-2 and mas2-1 (Fig. 2, B, C, and G). We also crossed smo4-3 to mutants carrying alleles of the paralogous Arabidopsis genes NUC1 and NUC2 (encoding NUCLEOLIN), which antagonistically act in the control of 45S rDNA transcription (Pontvianne et al., 2007;Durut et al., 2014). NUC1 is expressed in all tissues, and its mutations cause nucleolar disorganization, chromatin decondensation at the nucleolar organizer regions, and the up-regulation of 45S rDNA. NUC2 is expressed at much lower levels than NUC1 in wild-type plants but is induced in the nuc1 mutant background (Durut et al., 2014). NUC1 loss of function causes aberrant leaf development and vein patterning. This phenotype was described for the null alleles of NUC1 named parallel1 (parl1; Fig. 2D); parl1 mutants accumulate 35S A123B pre-rRNA due to failed cleavage at the P site (Petricka and Nelson, 2007;Pontvianne et al., 2007) of the 59-ETS of the 35S A123B pre-rRNA (Supplemental Fig. S3). The only effects of mutations in NUC2 on the morphological phenotype are an increased rosette size and mild late flowering ( Fig. 2E; Durut et al., 2014).
Palisade mesophyll cell size increases in the smo4 mutants (Zhang et al., 2015), and leaf venation is aberrant in parl1 mutants (Petricka and Nelson, 2007). We performed a morphometric analysis of these phenotypes ( Fig. 2, M-Z) and found that palisade mesophyll cells were larger in smo4-3 than in Col-0 ( Fig. 2, S, T, and Y), but to a lesser extent than that previously described (Zhang et al., 2015), perhaps due to differences in culture conditions and/or the node chosen for study. Palisade mesophyll cells were also enlarged in mas2-1 compared with Ler ( Fig. 2, N, O, and Y), but were smaller in mtr4-2 and much smaller in parl1-2, compared with Col-0 ( Fig. 2, P, Q, S, and Y). mas2-1 and nuc2-2 double mutant combinations with smo4-3 exhibited increased palisade mesophyll cell size, particularly smo4-3 mas2-1 (Fig. 2, O, R, T, U, X, and Y). These results indicate that the reduction of leaf and rosette area in these double mutants is caused by decreased cell proliferation but not cell expansion. This hypothesis is supported by the finding that the number of palisade mesophyll cells per leaf was much lower in the double mutants than in the single mutants (Fig. 2Z). We did not obtain results from parl1-2 or smo4-3 parl1-2 leaves, except for the cellular area (Fig. 2, W and Y), as they became extremely fragile after depigmentation treatment. In line with these findings, the use of the SMO4 pro :GUS transgene revealed the strongest expression of SMO4 in proliferating tissues (Supplemental Fig. S12), as previously described (Zhang et al., 2015).
smo4 Mutants Show Defects in 18S and 5.8S rRNA Maturation As mentioned above, yeast Mtr4 and Arabidopsis MTR4 interact with the nuclear exosome for 5.8S rRNA maturation, and yeast Nop53 acts as an adaptor protein for this interaction (Lange et al., 2011(Lange et al., , 2014Thoms et al., 2015). In Arabidopsis mtr4 plants, unprocessed 18S and 5.8S rRNAs, as well as 59-ETS by-products (downstream of the P site), overaccumulate (Lange et al., 2011). To visualize the intermediates of 45S pre-rRNA processing (Supplemental Fig. S3), we performed gel-blot analyses of RNA extracted from smo4-2, smo4-3, and den2 plants using the previously described S2, S7, and S9 probes, which are complementary to a Whiskers represent Q1 2 1.5 3 IQR (bottom) and Q3 1 1.5 3 IQR (top), where the interquartile range (IQR) is Q3 2 Q1. Black diamonds 5 means; black lines 5 medians; red crosses 5 outliers; and red circles 5 extreme minimum (less than Q1 2 3 3 IQR) or maximum (greater than Q3 1 3 3 IQR) outliers. Asterisks indicate values significantly different from the corresponding wildtype or parental line (indicated by color) by Student's t test (*P , 0.05; **P , 0.01; and ****P , 0.0001). More than 20 rosettes and 10 first-node leaves collected 21 das were analyzed per genotype. N to X, Diagrams of the subepidermal layer of palisade mesophyll cells from first-node leaves collected 21 das. Bars 5 40 mm. Y, Box plot showing the distribution of cell sizes in the subepidermal layer of palisade mesophyll cells from first-node leaves. Z, Number of palisade mesophyll cells of the subepidermal layer per leaf. Ten leaves collected 21 das were studied per genotype in Y and Z, and more than 230 cells were analyzed per genotype in Y. Error bars indicate SD. Asterisks indicate values significantly different from the corresponding wild-type or singlemutant parental line (indicated by color) by Student's t test (*P , 0.05; **P , 0.01; ***P , 0.001; and ****P , 0.0001). segment of the 59-ETS, internal transcribed spacer 1 (ITS1), and ITS2, respectively ( Fig. 3A; Lange et al., 2011).

, A and B; Supplemental
Using the S7 probe, accumulation of P-A 3 pre-rRNA, the first 18S rRNA precursor in the ITS1-first pathway, was detected in smo4 mutants, to a lesser extent than the 5.8S pre-rRNAs, but not in smo4-3 SMO4 pro :SMO4 and mtr4-2 plants ( Fig. 3E; Supplemental Fig. S3; Supplemental Table S5). As previously reported (Lange et al., 2011), we detected the accumulation of P9-A 3 in mtr4-2, as in den2, but not in the other smo4 mutants ( Fig. 3E; Supplemental Table S5). P9-A 3 is generated by cleavage at the P9 site of the P-A 3 pre-rRNA or at the A 3 site of the 33S(P9) pre-rRNA (Supplemental Fig. S3). In mtr4 mutants, the levels of the 18S rRNA remained unaltered, but the P9-A 3 and 18S-A 3 pre-rRNAs, which are polyadenylated, accumulated, indicating that these . Visualization by RNA gel blots of 45S pre-rRNA processing in the smo4-2, smo4-3, den2, and mtr4-2 single mutants. A, Diagram (modified from Hang et al. [2014]) illustrating the pre-rRNA processing intermediates that can be detected in RNA gel blots using the S2, S7, or S9 probes. The precursor regions hybridizing with the probes are highlighted in green (S2 probe), blue (S7), and red (S9). Vertical bars indicate the endonucleolytic cleavage sites relevant to this study (P, P9, A 2 , A 3 , E9, and C 2 ). B to E, Visualization of the processing of 5.8S (B and C) and 18S (D and E) rRNA precursors using RNA gel blots. Total RNA was separated on formaldehyde-agarose (B, D, and E) or polyacrylamide-urea (C) gels, transferred to a nylon membrane, and hybridized with the S9 (B and C), S2 (D), or S7 (E) probe. Two views of the bands visualized from smo4-3, smo4-2, and mtr4-2 RNA with the S9 probe are provided in C, one of which corresponds to a very short exposure time, which allowed 7S and 5.8S170 pre-rRNAs to be distinguished. EtBr, Ethidium bromide-stained gels, visualized before blotting, which served as loading controls. Similar results were obtained in at least two independent experiments. Relative quantification of the bands visualized with the S9, S7, and S2 probes is shown in Supplemental Table S5. pre-rRNAs are targets of the exonucleolytic activity of the exosome. The P9-A 3 and 18S-A 3 pre-rRNAs could be by-products of an irregular processing of the 18S rRNA precursors at their 39 end (Lange et al., 2011). The differential accumulation of 18S rRNA precursors, such as P-A 3 in smo4, and P9-A 3 and 18S-A 3 in mtr4-2, was confirmed in three RNA gel blots performed using the S7 probe, in each one of which a different biological replicate per genotype was used. Using the r5 1 r6, r5 1 r7, and r5 1 r8 primer sets for circular RT-PCR analyses, we obtained bands of the same sizes than those previously described (Hang et al., 2014;Liu et al., 2016), which corresponded to the P-A 3 , P9-A 3 , 18S-A 2 , 18S-A 3 , 33S, 32S, and 27SA 2 precursors, as we confirmed by Sanger sequencing (Fig. 5; Supplemental Table S7). These circular RT-PCR results reconfirmed our findings on the differential accumulation of the P-A 3 pre-rRNA in the smo4 mutants.
To ascertain if the accumulation of the P-A 3 pre-rRNA in the smo4 mutants is caused by defective processing, we cloned the products that we obtained by circular RT-PCR using the r5 1 r6 primer set (Fig. 5A). Sanger sequencing of these molecules showed that their 59 and 39 extremities (29 clones were analyzed from Col-0 and 27 from smo4-3) corresponded to the correct processing at the P and A 3 endonucleolytic sites and that the polyadenylation status of their 39 ends was similar for both genotypes (Supplemental Fig. S13). Polyadenylation of pre-rRNAs occurs mainly after cleavage and stimulates its exonucleolytic trimming by the exosome (Slomovic et al., 2006). Our results suggest that SMO4 is not required for the correct cleavage and polyadenylation of the P-A 3 pre-rRNA. However, the observed P-A 3 pre-rRNA accumulation suggests its delayed 59 end processing at the P9 site. These results are in agreement with the existence of a mechanism of quality control to ensure a balanced synthesis of the 40S and 60S ribosomal subunits, . Visualization by RNA gel blots of 45S pre-rRNA processing in the smo4-3 mas2-1 and smo4-3 mtr4-2 double mutants. RNA gel blots using the S9 (A and B), S2 (C), and S7 (D) probes are shown. Total RNA was separated on formaldehyde-agarose (A, C, and D) and polyacrylamide-urea (B) gels, transferred to a nylon membrane, and hybridized with the corresponding probe. Two views of the bands visualized from smo4-3 mtr4-2 RNA with the S9 probe are provided in B, one of which corresponds to a very short exposure time, which allowed 7S and 5.8S170 pre-rRNAs to be distinguished. EtBr, Ethidium bromide-stained gels, visualized before blotting, which served as loading controls. Similar results to those shown here were obtained in at least two independent experiments. Relative quantification of the bands visualized with the S9, S7, and S2 probes from smo4-3 mtr4-2 RNA is shown in Supplemental Table S6. as previously proposed for mtr4 mutants (Lange et al., 2011), which show an inefficient 5.8S rRNA maturation, as smo4 mutants do.
Using the S2 probe, we detected in mtr4-2 plants strong accumulation of the P-P9 fragment ( Fig. 3D; Supplemental Table S5), a by-product generated by the early processing of the 59-ETS 45S pre-rRNA, or the P-A 3 precursor of the 18S rRNA, as previously described (Lange et al., 2011). The overaccumulation of the P-P9 by-product was similar between smo4-3 mtr4-2 Figure 5. Visualization by circular RT-PCR amplification of 45S pre-rRNA processing in the smo4-2, smo4-3, den2, and mtr4-2 mutants. Ethidium bromide-stained agarose gels visualizing products from the circular RT-PCR amplifications performed are shown. RNA was extracted and circularized with T4 RNA ligase and reverse transcribed using the rt1 (A-D) or rt2 (E) primer, and the resulting cDNA was PCR amplified with the r5 1 r6 (A), r5 1 r7 (B), r5 1 r8 (C), r5 1 r2 (D), and r1 1 r2 (E) primer pairs. The full names of the rt1 and rt2 primers were 18c and 5.8SrRNA_R, respectively. Diagrams illustrate all (D) or part (A-C and E) of the 45S pre-rRNA, represented in black and gray (see Fig. 3A), with indication of the positions of the primers used for circular RT-PCR amplifications. Circular RT-PCR products are shown in red. Given that the primers used are divergent, part of the cDNA obtained from each circularized rRNA precursor (dotted red lines) is absent from the final circular RT-PCR products. and mtr4-2 ( Fig. 4C; Supplemental Table S6), whereas the levels in wild-type plants, smo4-2, smo4-3, den2, and smo4-3 SMO4 pro :SMO4 were similar ( Fig. 3D; Supplemental Table S5).
Taken together, these results indicate that MTR4 and SMO4 act together in 5.8S rRNA but that SMO4 does not participate in the degradation of by-products derived from the 59-ETS, as MTR4 does (Lange et al., 2011).
18S and 5.8S pre-rRNA Species Accumulate in the Nucleoli of smo4 Plants Given that 45S pre-rRNA processing is defective in smo4-2, smo4-3, and den2 plants, we investigated whether any mature or precursor rRNA species accumulate in the nucleolus or nucleoplasm in these mutant lines. We first performed RNA fluorescence in situ hybridization (RNA-FISH) using the S9 and S2 probes, which hybridize with the 5.8S and 18S pre-rRNAs, respectively, but not with their corresponding mature rRNAs (Fig. 3A). We detected fluorescence mainly in the nucleolus for both wild types and mutant lines, but with high differences in intensities among them (Fig. 6, A-D1). Using the S9 probe, we observed increased fluorescence in smo4-3 nucleoli (1.46-fold compared with Col-0), den2 (1.59-fold compared with Ler), and mtr4-2 (1.32-fold compared with Col-0; Fig. 6E1). Such increases were much less pronounced than the accumulation of the pre-5.8S rRNA species that we observed in RNA gel blots (Fig. 3, B and C; Supplemental Table  S5). These results suggest that immature 5.8S rRNAs are incorporated to the 60S subunit and exported to the cytoplasm.
With the S2 probe, we found higher fluorescence levels relative to the wild types in smo4-3 (2.13-fold) and den2 (1.64-fold), which accumulate the P-A 3 pre-rRNA, and in mtr4-2 (2.59-fold), which accumulates the P-P9 by-product (Fig. 6F1; Supplemental Table S5). These results suggest that these two RNA species, P-A 3 and P-P9, are retained within the nucleolus, and that they are the major contributors to the high fluorescence level detected with the S2 probe in smo4 and mtr4-2 mutants, respectively. It is worth mentioning here that relative fluorescence levels are underestimated in the mutants, mainly those of den2, because the detector of the confocal microscope was saturated for many dots. Saturation was unavoidable because of the extreme differences in fluorescence intensities between the smo4 and mtr4-2 mutants and their corresponding wild types; raising the level of detection resulted in undetectable signals from the wild types.
Our RNA-FISH experiments revealed hypertrophy of the nucleolus in smo4 mutants but not in mtr4-2, and it was particularly pronounced in den2 (Fig. 6). Then, we quantified the nucleolar and nuclear areas in Col-0, Ler, smo4-3, and den2 plants. To delimitate the nucleolus, we carried out immunolocalizations using an antibody against the nucleolar marker fibrillarin, and stained the samples with DAPI to visualize the nucleoplasm (Fig. 7, A-L). Both the nucleoplasm and nucleolus were larger in smo4-3 and den2 than in Col-0 or Ler, especially the nucleolus (Fig. 7, M-P). These quantitative results confirm the notion that the loss of SMO4 activity causes nucleolar hypertrophy. In addition, fibrillarin exhibited nonuniform staining in Col-0 and Ler, which was more diffuse and with a more granular appearance in smo4-3 and den2 (Fig. 7, A-L), pointing to some degree of disorganization of the nucleolus.

Ribosome Profiles Are Slightly Imbalanced in smo4 Plants
Finally, to evaluate whether ribosome assembly was impaired in the smo4 mutants, we obtained ribosome profiles of Col-0, smo4-2, smo4-3, and smo4-3 SMO4pro :SMO4 seedlings. Specifically, we fractionated purified cell extracts through sucrose gradients to separate the 40S and 60S ribosomal subunits, the 80S monosome, the 90S preribosome, and polysomes. We did not observe major defects in the 60S and 80S peaks, but the 40S particle overaccumulated in both mutants, which was normalized in smo4-3 SMO4 pro :SMO4 plants (Fig. 8), according to its wild-type phenotype (Fig. 1I). An excess of the 40S subunit has also been found in yeast cells lacking Nop53 activity, which causes an imbalanced 40S:60S ratio (Sydorskyy et al., 2005). In addition, we detected a peak corresponding to particles that sedimented slightly more rapidly than 80S monosomes, which might have been 90S particles (Fig. 8). The nature of the peak that sedimented more slowly than 40S particles is unknown. Neither of these peaks showed major changes in the mutants compared with Col-0 and served as internal controls.

Evolutionary Conservation and Divergence of SMO4 Function in Ribosome Biogenesis
80S ribosome biogenesis is an essential, evolutionarily conserved process that has diverged among fungi, plants, and animals. The extent of this conservation is demonstrated by the existence of one or more Arabidopsis orthologs for 179 of the approximately 250 RBFs in yeast (Simm et al., 2015). Evidence of the divergence in 80S ribosome biogenesis is provided by the observations that many human orthologs of yeast RBFs are involved in pre-rRNA processing but have evolved different functions and that 74 human RBFs lack orthologs in yeast (Wild et al., 2010;Tafforeau et al., 2013).
Based on their homology with yeast and human proteins, several hundred Arabidopsis proteins are annotated as encoding putative components of the ribosome biogenesis machinery, including RPs and RBFs (Sáez-Vásquez and Delseny, 2019). Mutants representing more than 20 mutated RP-encoding genes have been isolated in screens for morphological aberrations or embryonic lethality (Byrne, 2009;Horiguchi et al., 2011). A recent survey of plant RBFs at the mutational and molecular levels revealed 28 individual proteins, 27 of which have yeast orthologs. For all but two of these plant RBFs, gene mutations result in developmental defects, and mutants in 16 of Figure 6. Subcellular localization of 5.8S and 18S rRNA precursors in the smo4-3, den2, and mtr4-2 mutants. A to D1, RNA-FISH assays in palisade mesophyll cells from first-node leaves of Col-0 (A-F), smo4-3 (G-L), mtr4-2 (M-R), Ler (S-X), and den2 (Y-D1). Fluorescent signals correspond to 49,6-diamidino-2-phenylindole (DAPI; A, D, G, J, M, P, S, V, Y, and B1), which was used as a nuclear marker (in blue); S9 probe labeled with Cy3 (in red; B, H, N, T, and Z); S2 probe labeled with FAM (in green; E, K, Q, W, and C1); and the overlay of the previous signals (C, F, I, L, O, R, U, X, A1, and D1). Photographs were taken from plants collected 14 das. Bars 5 25 mm. E1 and F1, Relative fluorescence intensity from the S9 (E1) and S2 (F1) probes, measured in 10 nuclei per leaf from six leaves per genotype. Error bars indicate SD. Asterisks indicate values significantly different from the corresponding wild type (indicated by color) by Student's t test (***P , 0.001 and ****P , 0.0001). these RBF-encoding genes overaccumulate pre-rRNAs, indicating that rRNA maturation is defective. The wildtype versions of 18 of these Arabidopsis RBF-encoding genes have been transferred into yeast strains, carrying mutations causing the absence or depletion of the corresponding RBF ortholog; however, no complementation of the yeast RBF mutant phenotype was observed for 10 of the 18 Arabidopsis RBF-encoding genes examined (Weis et al., 2015); one of these Arabidopsis RBFencoding genes is MTR4, which did not complement mtr4 mutations in yeast (Lange et al., 2014). It is thus not surprising that Arabidopsis SMO4 did not complement a Nop53 lack-of-function mutation in yeast (Zhang et al., 2015).
Our results provide experimental evidence for the role of SMO4 as an RBF in 45S pre-rRNA processing. Indeed, in the smo4 and den2 mutants, we detected accumulation of three 5.8S rRNA precursors that were nearly absent in wild-type and smo4-3 SMO4 pro :SMO4 plants: the 7S (5.8S 1 120 nucleotides), 5.8S170, and 6S (5.8S 1 11/12 nucleotides) pre-rRNAs (Fig. 3, B and C; Supplemental Table S5). These mutants also accumulated P-A 3 pre-rRNA, an 18S rRNA precursor that is only produced by the ITS1-first pathway. In addition, den2 accumulates all rRNA precursors that we have detected in RNA blots, including the P9-A 3 pre-rRNA, which also accumulates in mtr4 mutants ( Fig. 3E; Supplemental Fig. S3; Supplemental Table S5). These results suggest that the processing of the 45S pre-rRNA is delayed in smo4 mutants.
Yeast Nop53 is a single-copy, essential gene, as is expected from its key role in rRNA maturation (Granato et al., 2005). SMO4 is also a single-copy gene in Arabidopsis, and previous studies with two smo4 alleles, including smo4-2, have shown that SMO4 is required but not essential for cell cycle progression and survival (Zhang et al., 2015). Our work with two additional insertional alleles, and den2, which does not produce a full-length protein, suggests that the absence of SMO4 causes mild morphological phenotypes that become less apparent during later stages of plant development (Zhang et al., 2015;this work). In fact, as previously noted (Zhang et al., 2015), the C-terminal region of the wild-type SMO4 protein, which is predicted to be absent from the DEN2 protein, is the most conserved part among yeast, human, and Arabidopsis orthologs (Supplemental Fig. S4).
We have observed a dramatic hypertrophy of the nucleolus in smo4 mutants, mainly in the den2 mutant, as well as disorganization of the nucleolus, revealed by an atypical distribution of fibrillarin. Morphometric analysis of smo4 mutants has shown that SMO4 positively regulates cell proliferation (Zhang et al., 2015;this study). It is possible that, similar to GLTSCR2, SMO4 participates in the control of the progression of the cell cycle in response to nucleolar stress.

SMO4 and MAS2 Are Functionally Related
The number of proteins known to bind to RNA or to participate in processes involving RNA molecules has increased over the past decades, together with the number of known RNA functions. Some of these proteins play multiple roles in RNA metabolism pathways, many of which appear to be ancient. For example, metazoan NKAPs are multifunctional factors involved in the regulation of diverse processes, such as cellular differentiation, proliferation, and apoptosis. Human NKAP immunoprecipitates together with several RBFs, most of which are involved in 18S rRNA maturation, and it can associate with different spliceosomal complexes and to premRNAs and spliced mRNAs as well as to small nuclear RNAs, small nucleolar RNAs, rRNAs, and long intergenic noncoding RNAs (Burgute et al., 2014).
Small RNAs are involved in PTGS pathways and the epigenetic regulation of gene transcription. To further explore the role of AGO1, the main RNase that functions in PTGS pathways mediated by microRNAs, we previously performed a screen for suppressors of the morphological phenotype of the ago1-52 hypomorphic and viable allele (Micol-Ponce et al., 2014). One suppressor gene that we identified was MAS2, but the mas2 alleles that we isolated, which were dominant and likely antimorphic, act as informational suppressor mutations (Sánchez-García et al., 2015). Comparable to its NKAP orthologs, MAS2 is essential and multifunctional, acting in processes such as splicing and ribosome biogenesis. In a search for physical interactors of MAS2, we identified two RBFs, RRP7 (Micol-Ponce et al., 2018) and SMO4 (this work), as well as RPS24B (encoded by AT5G28060), an RP whose yeast and human orthologs also act as RBFs for 18S rRNA maturation (Ferreira-Cerca et al., 2005;Choesmel et al., 2008). The SMO4, RRP7, and MAS2 promoters share regulatory motifs that are enriched in genes encoding factors involved in translation, including RBFs (Micol- Ponce et al., 2018). The epistatic interaction of rrp7-1 on smo4-3 found in smo4-3 rrp7-1 plants and the synergistic phenotypes of the rrp7-1 mas2-1 (Micol- Ponce et al., 2018) and smo4-3 mas2-1 (this work) double mutants indicate that SMO4, RRP7, and MAS2 are functionally related. We did not detect alterations in 45S pre-rRNA processing in mas2-1 plants or in the smo4-3 mas2-1 double mutant, which was the same as that in smo4-3 (Fig. 4), indicating that the synergistic effect in the double mutants was not due to the role of SMO4 in 45S pre-rRNA processing. Figure 7. (Continued.) corresponding wild types. Between 287 and 554 cells were studied per genotype, from the roots of five seedlings of each genotype, collected 5 das. Nuclei and nucleoli size distributions of smo4-3 and den2 were significantly different from the corresponding wild type in a Kolmogorov-Smirnov test (P , 0.0001).
We have not yet identified in silico yeast orthologs of Arabidopsis MAS2 or human NKAP. However, Dictyostelium discoideum has an NKAP ortholog that also appears to be a multifunctional protein that coimmunoprecipitates with several RPs and RNA-binding proteins, including RBFs such as bystin (Burgute et al., 2016). S. cerevisiae possesses a bystin ortholog named Essential nuclear protein1 (Enp1), which is required for efficient nuclear export of the pre-40S ribosomal particle (Seiser et al., 2006). In the enp1 and mtr4 yeast mutants, mRNAs are retained within the nucleolus. Nuclear retention of mRNAs is also caused by mutations in genes involved in rRNA biogenesis or premRNA splicing, ribosome quality control (including genes encoding components of the nuclear exosome), and mRNA nuclear export (Paul and Montpetit, 2016).
SMO4 Is Functionally Related to MTR4 and NUC1 Yeast Nop53 acts as an adaptor protein that interacts with the arch domain of Mtr4 via its AIM for the recruitment of exosomes to rRNA precursors (Thoms et al., 2015). Notwithstanding the low level of similarity observed between SMO4 and its yeast Nop53 and human GLTSCR2 orthologs, the AIM in these proteins is conserved, as in comparable proteins in many other plant species (Supplemental Figs. S4 and S5). The conservation of the SMO4 AIM and the MTR4 arch domain in Arabidopsis and other plant species (Supplemental Fig. S7) suggests that the mechanism of exosome recruitment to 45S pre-rRNA is also conserved in the plant kingdom. Taken together, these findings and the similar morphological phenotypes of mtr4-2, smo4-2, smo4-3, and den2 plants indicate that MTR4 and SMO4 play related roles in the 45S pre-rRNA processing pathways.
Most bands that we visualized in RNA blots using the S9 probe were similar in size and intensity in smo4-3, smo4-2, and den2, as well as in mtr4-2 and the smo4-3 mtr4-2 double mutant. These results suggest that SMO4 facilitates the exonucleolytic trimming of 5.8S rRNA precursors, a process comparable to that of yeast Mtr4 and Arabidopsis MTR4 in mtr4 plants (Lange et al., 2011). However, unlike MTR4, SMO4 seems to not participate in the degradation of by-products of the early steps of 45S pre-rRNA processing, suggesting that MTR4 associates with other protein(s). Two partners of yeast Mtr4 have been identified: Nop53 and U three protein18 (Utp18), a component of the small subunit processome, a large ribonucleoprotein complex that participates cotranscriptionally in early 35S pre-rRNA processing, to generate pre-18S rRNA (Phipps et al., 2011). Interactions between Mtr4 and both Nop53 and Utp18 are required to recruit the exosome to the precursors of 5.8S rRNA and 59-ETS, respectively (Bernstein et al., 2004;Falk et al., 2017). Utp18 orthologs, including Arabidopsis UTP18 (encoded by AT5G14050), harbor the conserved AIM (Thoms et al., 2015), suggesting that Arabidopsis UTP18 could be a second partner of MTR4, required for 59-ETS degradation. However, SMO4 and UTP18 were not identified among the proteins that copurified with MTR4-GFP (Lange et al., 2014), perhaps because Figure 8. Ribosome profiles of the smo4 mutants. Extracts from the aerial organs of wild-type (A), smo4-2 (B), smo4-3 (C), and smo4-3 SMO4 pro :SMO4 (D) plants were collected 18 das and then lysed and fractionated through 15% to 60% sucrose gradients by ultracentrifugation. The percentage of the full scale of absorbance was monitored at 254 nm. Peaks corresponding to 40S and 60S ribosomal subunits, 80S monosomes, 90S preribosome, and polysomes are indicated. The asterisks indicate an unknown particle. they do not interact or because the GFP tail in the fusion protein prevented their interaction.
The weak morphological phenotypes of smo4 mutants and their ribosomal profiles suggest that mature, functional 18S and 5.8S rRNAs are produced in these plants, as has been observed in mtr4 mutants, suggesting that ribosome biogenesis is delayed. It also appears that the overaccumulation of 5.8S and 18S rRNA precursors in the smo4 and den2 mutants has no effect on viability or fertility and only mildly affects development. The synergistic phenotype of smo4-3 parl1-2 provides genetic evidence for the functional relationship between SMO4 and NUC1, both of which function as RBFs in 45S pre-rRNA processing.
Based on the morphological phenotype of the double mutants generated in this study, mtr4-2 appears to be epistatic to smo4-3, since rosettes were indistinguishable between smo4-3 mtr4-2 and mtr4-2, but clearly different from that of smo4-3 (Fig. 2, C, G, and I). mtr4-2 leaves are more pointed than those of smo4-3, and mtr4-2 rosettes are smaller than those of smo4-3 (Fig. 2, C, G, and L; Supplemental Table S3). These findings are consistent with the results of our molecular analysis of 45S pre-rRNA processing in the single and double mutants, indicating that both MTR4 and SMO4 act together in 5.8S rRNA maturation. In addition, our analysis using the S2 probe showed that MTR4 but not SMO4 participates in the degradation of 59-ETS byproducts of 45S pre-rRNA processing.
Seed sterilization and sowing, plant culture, and crosses were performed as previously described (Ponce et al., 1998;Berná et al., 1999), except for plants used for ribosome profiling (see below). When required, culture media were supplemented with hygromycin (15 mg mL 21 ).

Molecular Characterization of Mutations
To characterize den2, a candidate interval of 3,827 kb was delimited by iterative linkage analysis as previously described (Ponce et al., 1999(Ponce et al., , 2006 using PCR amplification and the polymorphic markers listed in Supplemental Table  S1. The den2 point mutation was ultimately identified by whole-genome, nextgeneration sequencing, looking for the transitions typically caused by EMS (G→A or C→T) within the candidate interval. Only one EMS-type mutation (G→A) was found in a coding region and verified by Sanger sequencing. The raw data have been deposited in the Sequence Read Archive (https://www. ncbi.nlm.nih.gov/sra/) database under accession number SRP103180.
The presence of T-DNA insertions in the SMO4, NUC1, NUC2, and MTR4 genes was verified by PCR using the primers shown in Supplemental Table S2. Discrimination between the wild-type MAS2 and mutant mas2-1 alleles was carried out by PCR amplification followed by restriction analysis, as described by Sánchez-García et al. (2015).
For massive sequencing, DNA was extracted from plants using a DNeasy Plant Mini Kit (Qiagen), and sequencing was performed using the Ion Proton platform following the manufacturer's instructions (Applied Biosystems, now Thermo Fisher Scientific). The resulting reads were mapped to the TAIR10 version of the Col-0 reference genome with Torrent Suit Software v5.2.1 (Thermo Fisher Scientific), which was also used to compile a list of the mutations identified. Sanger sequencing was performed with ABI PRISM BigDye Terminator Cycle Sequencing kits on an ABI PRISM 3130xl Genetic Analyzer (Applied Biosystems).

Construction and Analysis of Transgenes and Transgenic Lines
Constructs for Gateway cloning were generated and transferred into plants as described by Sánchez-García et al. (2015). The pGEM-T Easy221 entry vector and the pMDC32, pMDC83, pMDC107, and pMDC164 destination vectors (Curtis and Grossniklaus, 2003) were used. Inserts were generated by PCR using primers that included attB1 and attB2 sequences, as detailed in Supplemental Table S2. Chemically competent Escherichia coli DH5a cells were transformed with BP or LR Gateway cloning reaction products using the heat shock method. Agrobacterium tumefaciens C58C1 cells carrying the pSOUP helper plasmid were transformed with the constructs by electroporation.
To obtain the 35S pro :SMO4 and 35S pro :SMO4:GFP overexpression constructs, the full-length coding sequence (stop codons were removed to obtain all GFP translational fusions) was PCR amplified from Col-0. The 1,317-bp region upstream of the translation start codon of SMO4 was PCR amplified and used as the promoter driving the SMO4 pro :GUS transgene. This same promoter region was present at one end of the 4,156-bp segment amplified to isolate the entire 2,839-bp SMO4 transcription unit, which was used to create the SMO4 pro :SMO4 and SMO4 pro :SMO4:GFP constructs. The fidelity of all constructs was verified by Sanger sequencing before they were transferred into plants via the floral dip method (Clough and Bent, 1998).
The subcellular localization of SMO4 was visualized in the roots of plants in the Col-0 background carrying the SMO4 pro :SMO4:GFP or 35S pro :SMO4:GFP transgene, with plant tissue collected 10 das. GUS activity was analyzed in homozygous plants, and photographs of three plants were taken from each of three independent lines per genotype.

Plant Morphometry, Histology, and Microscopy
Photographs of Arabidopsis plants were taken with a Nikon SMZ1500 stereomicroscope equipped with a Nikon DXM1200F digital camera. To obtain high-resolution images from large rosettes, four to five partial photographs from the same plant were taken and assembled using the Photomerge tool of Adobe Photoshop CS3 software. Some empty spaces resulting from the assembly were further equalized with the black background, using Adobe Photoshop CS3. Measurement of rosette sizes was performed 21 das with the NIS Elements AR 3.1 image-analysis package (Nikon) from photographs of plants in petri dishes taken with a Canon PowerShot S315 camera.
For morphometric analysis of leaf lamina, palisade mesophyll cells, and vein patterns, the ImageJ processing program (https://imagej.net/ImageJ), the NIS Elements AR 3.1 (Nikon) image-analysis package, and the phenoVein software (http://www.plant-image-analysis.org/) were used, respectively, as described previously (Robles et al., 2010;Pérez-Pérez et al., 2011). In brief, fully expanded first-node leaves were collected 21 das, cleared with ethanol and chloral hydrate, and mounted on slides. Micrographs of the internal tissues were obtained using a Leica DMRB microscope equipped with a Nikon DXM1200 digital camera. Diagrams of the palisade mesophyll cells and leaf venation patterns were drawn on the screen of a Cintiq 18SX Interactive Pen Display (Wacom) using Adobe Photoshop CS3 software.
All fluorescence and confocal laser-scanning microscopy images were generated using a D-Eclipse C1 confocal microscope equipped with a DS-Ri1 camera and digitally processed with EZ-C1 operation software (Nikon), with band-pass filters. For observing DAPI and Hoechst 33342 nuclear staining, laser excitation was carried out at 408 nm and their emissions collected at 450/35 nm. GFP and FAM were excited at 488 nm and their emissions collected at 515/30 nm. Tetramethylrhodamine-5-isothiocyanate and Cy3 (Cyanine 3) were excited at 544 nm and their emissions detected at 605/75 nm.

Chlorophyll and Anthocyanin Extraction and Measurements
To quantify chlorophyll and anthocyanin content, 30 and 10 independent biological replicates were used, respectively. Each biological replicate included three rosettes collected 18 das, which were pooled, weighed, and homogenized in a MixerMill 400 (Retsch) automatic mixer. Chlorophylls were extracted with cold 80% (v/v) acetone for 30 min in the dark with agitation. Cell debris was removed by centrifugation at 3,000 rpm for 15 min at 4°C, and the pigment concentration in the supernatant was spectrophotometrically determined as described by Arnon (1949). Anthocyanins were extracted in 45% (v/v) methanol and 5% (v/v) acetic acid buffer, and the samples were centrifuged twice for 5 min at 12,000 rpm to remove cellular debris. Pigment concentration in the supernatant was spectrophotometrically measured as described by Mancinelli (1990).
Semiquantitative RT-PCR, RNA Gel-Blot Analysis, and Circular RT-PCR Assays Total RNA was extracted from the aerial tissues of plants collected 15 das using TRI RNA Isolation Reagent (Sigma-Aldrich). Each biological sample contained RNA extracted from three plants, each grown on a different plate.
For semiquantitative RT-PCR experiments, retrotranscription was carried out with random hexamer primers and Maxima Reverse Transcriptase (Invitrogen), following indications of the manufacturer. RNA was treated with Turbo DNase (Invitrogen) prior to the reverse transcription step. The ACTIN2 (ACT2) housekeeping gene was used as an internal control. Sequences of primers for PCR amplifications of ACT2 and SMO4 cDNAs are indicated in Supplemental Table S2.
For RNA gel-blot analysis, 3 mg of total RNA was used per gel blot, and the samples were loaded onto 1.2% (w/v) agarose/formaldehyde or 6% (w/v) polyacrylamide (a ratio between acrylamide and bisacrylamide of 29:1)/8 M urea gels. The polyacrylamide gel was run for 1 to 3 h at 180 V in 0.53 Trisborate/EDTA buffer. RNA was visualized following ethidium bromide staining and transferred and cross-linked onto a Hybond N 1 nylon membrane (Thermo Fisher Scientific). Hybridization and detection were carried out as previously described (Micol-Ponce et al., 2018) using digoxigenin-labeled probes and Lumi-Film Chemiluminescent Detection Films (Roche), which were exposed for 20 min, 1 h, or overnight. The S2 probe was synthetized by PCR using digoxigenin-11-dUTP, genomic DNA as template, and the S2 fw and S2 rev primers; the S7 and S9 probes were 59 end digoxigenin-labeled oligonucleotides; these probes were synthetized by Eurofins Genomics. The sequences of the primers and oligonucleotides used to obtain all probes are described in Supplemental Table S7 and were taken from Lange et al. (2011). To compare band intensities between samples, film images were quantified by densitometry with the Vision-Capt software (Vilber Lourmat) using as an internal loading control the 18S rRNA band visualized with ethidium bromide. Circular RT-PCR was performed as described by Hang et al. (2014) and Micol- Ponce et al. (2018) using the primers described in Supplemental Table S7. In short, 5 mg of total RNA was circularized with T4 RNA ligase and reverse transcribed with Maxima Reverse Transcriptase (Thermo Fisher Scientific) and the rt1 or rt2 primers (Supplemental Table S7). cDNA was amplified using the r5 primer combined with r2, r6, r7, or r8 or with the r1 1 r2 primer set (Supplemental Table S7). Products corresponding to P-A 3 pre-rRNA obtained from Col-0 and smo4-3 with the r5 1 r6 primer set were gel excised, purified, and cloned into the pCR2.1 vector (Invitrogen), and several clones were sequenced using the M13F universal primer. The remaining products obtained with the r5 primer combined with r2, r6, r7, or r8 were not cloned; instead, they were gel excised, purified, and sequenced using the r5 primer.

RNA-FISH and Immunolocalization
RNA-FISH was carried out as described by Parry et al. (2006). Oligonucleotides, which were synthetized and labeled at their 59 ends (Supplemental Table  S7) by Eurofins Genomics, were used as probes. The S2 probe was labeled with FAM and the S9 probe with Cy3. Approximately 100 cells per genotype from the first-node leaves of 10 plants collected 14 das were analyzed and were fixed in glass vials at 600 mbar for 25 min. Probes were used in a 0.5 mg mL 2 1 hybridization solution. The samples were mounted on slides with a drop of Vectashield antifade mounting medium (Vector Laboratories) containing 0.01 mg mL 21 DAPI. Fluorescence intensity was measured with the EZ-C1 3.90 FreeViewer software (Nikon).
Immunolocalization was carried out as described by Pasternak et al. (2015). The assays were performed on 24-well microplates using the roots of seedlings collected 5 das. Samples were fixed for 40 min at 37°C with a solution containing 2% (v/v) paraformaldehyde in 13 microtubule-stabilizing buffer (50 mM PIPES, 5 mM MgSO 4 , and 5 mM EGTA, pH 6.9) and 0.1% (v/v) Triton X-100. A 1:250 dilution of the mouse monoclonal primary anti-fibrillarin antibody [38F3] (Abcam) and a 1:500 dilution of the tetramethylrhodamine-5-isothiocyanate-conjugated anti-mouse IgG secondary antibody (Sigma-Aldrich) was used. Nuclei were stained for 10 min with 0.2 mg mL 21 DAPI and washed for 5 min before mounting the samples on slides.
To quantify nuclei and nucleoli sizes, their outlines were drawn from confocal images on a Wacom DTF-720 Pen Display using Adobe Photoshop CS3 software. Nuclei and nucleoli areas were measured from these diagrams with the NIS Elements AR 3.1 (Nikon) image-analysis package. The ratio between the areas of nuclei and nucleoli was obtained from 287 to 554 cells from five seedlings per genotype. The number of nuclei and nucleoli, which were classified into 16 and 14 groups, respectively, depending on their size, were determined and their percentages calculated. Statistical analysis of the distribution of nuclei and nucleoli areas was performed with the XLSTAT statistical software (Addinsoft).

Ribosome Profiling
Seeds were sown on petri dishes containing 13 Murashige and Skoog medium supplemented with 1% (w/v) sucrose and stratified for 2 d. Plants were grown under a 16-h-light (120 mE m 22 s 21 )/8-h-dark cycle, at 21°C for 18 d, and relative humidity of 55% in light or 60% at night. Aboveground organs were collected 18 das, shock frozen in liquid nitrogen, and ground to a fine powder. Each 250-mg sample was resuspended in 750 mL of extraction buffer (0.2 M Tris-HCl, pH 9, 0.2 M KCl, 0.035 M MgCl 2 , and 0.025 M EGTA, pH 8) supplemented with 1% (w/v) sodium deoxycholate, 1% detergent mix (5 g of Brij-35, 5 mL of Triton X-100, 5 mL of Igepal, and 5 mL of Tween 20), 2% (v/v) polyoxyethylene (10) tridecyl ether, 5 mM DTT reducing agent, 10 mM MG132 proteasome inhibitor, 50 mg mL 21 chloramphenicol, 100 mg mL 21 cycloheximide, and 100 mL of protease inhibitor cocktail for plant cell and tissue extracts (Sigma-Aldrich) per 10 mL of buffer. After 10 min of incubation on ice, the samples were centrifuged at 17,000g for 10 min at 4°C and the supernatant clarified through a 0.45-mm filter.
For ribosome profiling, 360 mL of clarified supernatant was layered onto a 9-mL linear 15% to 60% (w/v) sucrose gradient. After centrifugation in a Beckman SW41 rotor at 38,000 rpm for 3 h at 4°C, the gradient was analyzed using the Type 11 Optical Unit (Teledyne ISCO) attached to a UA-6 UV/VIS Detector (Teledyne ISCO) for continuous measurement of the A 254 . The assay was repeated up to three times with consistent results.

Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Overview of 35S pre-rRNA processing in yeast.
Supplemental Figure S2. Overview of 47S pre-rRNA processing in humans.
Supplemental Figure S3. Overview of 45S pre-rRNA processing in Arabidopsis.
Supplemental Figure S4. Sequence conservation among putative SMO4 orthologs in Arabidopsis, humans, and yeast.
Supplemental Figure S6. Sequence conservation among MTR4 orthologs in Arabidopsis, humans, and yeast.
Supplemental Figure S9. Molecular effects of the smo4 and den2 mutations on the expression of the SMO4 gene.
Supplemental Figure S10. Morphological and molecular effects of the expression of the 35S pro :SMO4 transgene in the Col-0 background.
Supplemental Figure S11. Pigment concentration and leaf venation pattern in the double mutants studied in this work.
Supplemental Figure S13. Sequences of the extremities of P-A 3 pre-rRNA molecules found in Col-0 and smo4-3 RNA.
Supplemental Table S1. Primer sets used for the fine-mapping of den2.
Supplemental Table S2. Primer sets used in this work.
Supplemental Table S3. Morphometry of first-node leaf venation in the single mutants studied in this work.
Supplemental Table S4. Morphometry of first-node leaf venation in the double mutants studied in this work.
Supplemental Table S5. Relative quantification of several bands visualized with the S9, S7, and S2 probes in the RNA blots shown in Figure 3.
Supplemental Table S6. Relative quantification of several bands visualized with the S9, S7, and S2 probes in the RNA blots shown in Figure 4.
Supplemental Table S7. DNA probes and primers used for rRNA analysis.