Posttranscriptional Gene Regulation by Spatial Rearrangement of the 3′ Untranslated Region

Translation termination at premature termination codons (PTCs) triggers degradation of the aberrant mRNA, but the mechanism by which a termination event is defined as premature is still unclear. Here we show that the physical distance between the termination codon and the poly(A)-binding protein PABPC1 is a crucial determinant for PTC recognition in human cells. “Normal” termination codons can trigger nonsense-mediated mRNA decay (NMD) when this distance is extended; and vice versa, NMD can be suppressed by folding the poly(A) tail into proximity of a PTC or by tethering of PABPC1 nearby a PTC, indicating an evolutionarily conserved function of PABPC1 in promoting correct translation termination and antagonizing activation of NMD. Most importantly, our results demonstrate that spatial rearrangements of the 3′ untranslated region can modulate the NMD pathway and thereby provide a novel mechanism for posttranscriptional gene regulation.


Introduction
Nonsense-mediated mRNA decay (NMD) represents a translation-dependent posttranscriptional mRNA quality control process that selectively degrades mRNAs containing premature termination codons (PTCs), thereby preventing the synthesis of truncated, potentially deleterious proteins [1,2]. Because one-third of all known disease-causing mutations are predicted to generate a PTC, NMD serves as an important modulator of genetic disease phenotypes in humans [3,4]. Hence, understanding the molecular mechanism of NMD will facilitate the future development of genespecific therapies for many genetic diseases. Interestingly, NMD affects 3%-10% of the transcriptome of Saccharomyces cerevisiae, Drosophila melanogaster, and mammals, indicating that NMD, in addition to its quality control function, is also involved in regulating the expression of many physiological transcripts (reviewed in [5]).
The three Upf (Up-frameshift) proteins-Upf1, Upf2, and Upf3-work at the heart of the NMD pathway in all organisms studied. The Upf proteins were first discovered in genetic screens in S. cerevisiae and Caenorhabditis elegans, and orthologs have subsequently been identified in other eukaryotes (reviewed in [6]). Upf1 is an ATP-dependent RNA helicase, and a mutation in the ATPase domain abolishes its 59-to-39 helicase activity and its function in NMD [7][8][9]. Human Upf2 contains three conserved middle of eIF4G-like (MIF4G) domains, multiple putative nuclear localization signals in its N-terminus, and a putative nuclear export signal. Upf2 interacts with Upf1 and Upf3, and the three proteins can be isolated as a complex [10][11][12][13]. Upf3 is the least conserved component among the Upf proteins [6]. Humans contain two different UPF3 genes, encoding Upf3a and Upf3b (also known as Upf3X since the corresponding gene maps to the X chromosome, respectively) [11,13]. In addition to Upf1-3, metazoans contain additional NMD factors (Smg1, Smg5, Smg6, and Smg7) that are involved in regulating the phosphorylation state and therewith the activity of Upf1 [14].
While the phenomenon of NMD and its impact on gene expression are well documented, the understanding of the underlying molecular mechanisms is still fragmented. A central question is how PTCs are recognized and discriminated from natural termination codons (TCs). The current models for NMD differ remarkably between mammals and other eukaryotes. While all models agree that translation is required for NMD, studies of mammalian genes indicate that NMD in higher eukaryotes also depends on pre-mRNA splicing. The current model for mammalian NMD postulates that PTC recognition requires an interaction between an exon junction complex (EJC) bound to the mRNA downstream of the TC and the terminating ribosome [15,16]. However, several examples of NMD in mammals have been reported that are inconsistent with this EJC-dependent NMD model ( [17] and references therein). Recently, we showed that PTCs in the terminal exon of Ig-l minigenes (minil) elicit NMD dependent on the length of their 39 untranslated region (UTR) [17]. This is reminiscent of the situation in D. melanogaster, C. elegans, S. cerevisiae, and plants, where PTC recognition occurs independently of splicing and EJC factors, but where instead 39 UTR length and the poly(A)-binding protein (PABP) were found to play an important role [18][19][20]. The ''faux 39 UTR'' model, which is based on studies with yeast, postulates that proper translation termination requires an interaction between PABP and eRF3 bound to the terminating ribosome, and that the absence of this positive signal leads to aberrant termination and NMD as a consequence [21]. We report here that our results obtained with human cell lines are consistent with the yeast ''faux 39 UTR'' model, providing strong evidence for an evolutionarily conserved basic mechanism of PTC recognition in which PABPC1 acts as an NMD antagonizing factor and a role for the mammalian EJC as an NMD enhancer. But most importantly, our data show that mRNA half-lives can be regulated by altering the spatial configuration of their 39 UTRs. This represents a novel, potentially widespread mechanism for posttranscriptional gene regulation by NMD.

UTR Extensions Reduce the mRNA Half-Life by NMD
Comparison of relative mRNA levels of minil constructs with PTCs at different positions indicated that PTCs located toward the 59 and the 39 ends of the mRNA induce gradually less efficient NMD ( Figure 1A). Because mRNAs were shown to adopt a circular conformation that positions the 59 end close to the 39 end by eIF4G bridging the cap-bound factors (eIF4E or CBC) with the poly(A)-binding protein PABPC1 bound to the 39 end [22][23][24], and based on our previous results [17], we hypothesized that the distance between the TC and the poly(A) tail might be a crucial determinant to identify a TC as premature. Supporting this hypothesis, extension of this distance by insertion of prokaryotic sequence into the 39 UTR redefines the normal TC as NMD-triggering PTC ( Figure   1B-1D and [17]). Prokaryotic sequence was chosen for these 39 UTR extensions to minimize the risk of unintentionally inserting binding sites for mammalian RNA-binding proteins that could affect transcript stability. Extending the 39 UTR of a minil construct with the full-length coding region (minil C3/C4 WT) from 300 to 900 nucleotides reduced the mRNA level to 40%, and an extension to 1,500 nucleotides led to a further reduction to 7% ( Figure 1B). Judged from the Northern blot analysis, insertion of these sequences into the 39 UTR did not interfere with pre-mRNA splicing or 39 end formation. We further determined the decay kinetics of the minil WT and minil WT þ1,200 mRNA using the Tet-Off Advanced System in HeLa cells. The 39 UTR extension of 1,200 nucleotides caused a reduction of the mRNA's half-life from 4.16 h (minil WT) to 2.16 h (minil WT þ1,200; Figure  1C). Additionally, the mRNA reduction of minil WTþ1,200 can be suppressed to variable extent by RNAi-mediated knockdown of Upf1, Upf2, and Upf3b, indicating that it is caused by bona fide NMD ( Figures 1D and S1). Depletion of Upf1 resulted in a 16-fold, depletion of Upf2 in a 4-fold, and depletion of Upf3b in an 8-fold increase of minilWTþ1,200 mRNA, respectively. These differences in the extent of NMD suppression most likely reflect different knockdown efficiencies and/or different minimal concentrations of these proteins required to sustain NMD.
A similar extension of the 39 UTR of a b-globin reporter gene from 128 to 705 nucleotides ( Figure S2), and extension of the minil 39 UTR by a different sequence [17], also caused an Upf1-dependent mRNA reduction, suggesting that the length of the 39 UTR is an important and general determinant to define a TC as premature and trigger NMD independent of the sequence context. Because in all these examples (i) no exon-exon junction and hence no EJC is located downstream of the PTC, and (ii) several lines of evidence suggest that the upstream EJCs have been removed by the first translating ribosome [25], we refer to this form of NMD as ''EJC-independent.'' Reducing the Physical Distance between the TC and the Poly(A) Tail Suppresses EJC-Independent NMD Because many transcripts in higher eukaryotes have long 39 UTRs [26,27], simply the number of nucleotides between the TC and the poly(A) tail is unlikely to represent the signal for defining a TC as premature. Instead, we hypothesized that it may be rather the physical distance between the TC and the poly(A) tail that bears the kinetic and regulatory potential to distinguish a proper translation termination event from aberrant termination. According to this model, it should be possible to suppress NMD by reducing this distance. We tested this with so-called ''foldback'' constructs, in which 26 nucleotides complementary to the sequence located about 50 nucleotides downstream of the PTC were inserted into minil immediately upstream of the poly(A) signal. Base pairing of this complementary sequence positions the poly(A) tail in the vicinity of the PTC in these transcripts ( Figure 2A). Northern blot analysis ( Figure S3A) and reverse transcriptase (RT)-PCR (unpublished data) confirmed that introduction of this intramolecular base pairing did not interfere with splicing or 39 end processing. For this EJC-independent NMD reporter (minil C3/H4 ter440 [17]), a mRNA half-life of 2.59 h was observed with the minil ter440 ''no foldback'' (NFB) control construct, whereas the mRNA half-life of the

Author Summary
Correct expression of the genetic information is essential for life, and several quality control systems have evolved to ensure accurate protein synthesis. One of these processes, termed nonsensemediated mRNA decay (NMD), detects inappropriate termination of mRNA translation at premature termination codons (PTCs) and triggers degradation of the aberrant mRNA. Although the occurrence of NMD is well documented in yeast, worms, flies, mammals, and plants, the mechanism by which a termination event is defined as premature is still unclear, and different models have been proposed for different species. For mammals, the current prevailing view is that a termination codon is identified as premature and elicits NMD when it is located upstream of the 39-most exon junction complex. However, well-documented examples of NMD triggered by PTCs in the last exon challenge this ''mammalian NMD model.'' Here we show that the physical distance between the termination codon and the poly(A)-binding protein PABPC1 is a crucial determinant for PTC recognition in human cells, indicating an evolutionarily conserved function of PABPC1 in promoting correct translation termination and antagonizing activation of NMD. Most importantly, our results demonstrate that spatial rearrangements of the 39 untranslated region can modulate the NMD pathway and thereby provide a novel, translation-dependent mechanism for posttranscriptional gene regulation.
corresponding ''foldback'' (FB) construct was with 7.74 h very similar to the half-lives of the control constructs WT NFB (8.44 h) and WT FB (6.96 h, Figure 2B). Furthermore, the steady-state mRNA level of the minil ter440 NFB control construct was reduced to 30% in an Upf1-dependent manner, whereas the mRNA level of the corresponding FB construct was only marginally reduced compared to the WT and was not affected by RNAi-mediated Upf1 depletion ( Figure 2C and 2D). To confirm that the observed NMD suppression of ter440 FB mRNA was specifically dependent on the base  pairing between the inserted sequence near the poly(A) tail and the complementary region about 50 nucleotides downstream of ter440, we mutated seven nucleotides in this region to abolish the base pairing potential (ter440 mutFB, Figure  S4A). The Upf1-dependent mRNA reduction observed with ter440 mutFB shows that this mRNA is a substrate for NMD ( Figure S4B and S4C). This indicates that NMD suppression of ter440 FB requires the actual formation of the predicted intramolecular base pairs, because abolishing of this base pairing potential renders the transcript NMD-sensitive. Collectively, these results demonstrate that folding the poly(A) tail into the vicinity of a PTC in the terminal exon suppresses EJC-independent NMD.

mRNA Stability of FB Constructs Gradually Decreases with Increasing Distance between TC and Poly(A) Tail
Next we wanted to determine up to which maximal distance from the TC the poly(A) tail is able to suppress NMD. We generated additional FB constructs analogous to minil C3/H4 ter440 FB ( Figure 2) by inserting different sequences into the poly(A) signal proximal Spe1 restriction site that are complementary to different regions downstream of ter440 ( Figure 3A). The complementary sequences are between 20 and 30 nucleotides long and designed to have a melting temperature of about 60 8C when base pairing to their target sequence (see Materials and Methods). In HeLa cells transiently transfected with these pTRE-tight FB reporter constructs, we stopped transcription by addition of doxycycline and analyzed the decay kinetics of the FB transcripts. These experiments revealed a gradual destabilization of the mRNAs with increasing distance of the poly(A) tail from the PTC ( Figure 3B and 3C). Noteworthy, the two independently determined half-lives for minil C3/H4 ter440 FB mRNA differ by less than 1% (7.74 h in Figure 2B and 7.81 h in Figure 3B), indicating that these half-life measurements are highly reproducible and precise. Whereas the half-life of FB mRNA is similar to the half-lives of WT NFB and WT FB ( Figure 2B), indicating a complete suppression of NMD, the half-life of FB5 (2.54 h, Figure 3B) is similar to the half-life of ter440 NFB (2.59 h, Figure 2B), indicating a complete loss of NMD suppression. The half-lives of FB2, FB3, and FB4 mRNA fall in-between and indicate a partial loss of NMD suppression. From these results, we conclude that the poly(A) tailmediated NMD-suppressing activity functions in a distancedependent manner, as manifested by the gradual decrease of mRNA stability upon increasing distance between TC and poly(A) tail ( Figure 3C).

Reducing the Physical Distance between the TC and the Poly(A) Tail Also Suppresses EJC-Enhanced NMD
Our previous results suggested that in mammals, the EJC has adopted a NMD-enhancing function when present downstream of a TC, presumably by increasing the local concentration of the two essential NMD factors Upf2 and Upf3b [17]. We therefore tested if such EJC-enhanced NMD of a minil mRNA can also be suppressed by folding back the poly(A) tail (Figures 4 and S5). To this end, we generated a FB construct with a PTC located in the fourth of six exons (minil ter310 FB, Figure 4A). After splicing, the mRNA of this minigene construct is expected to harbor two EJCs downstream of the PTC, and the corresponding NFB control mRNA (ter310 NFB) should therefore be subject to efficient EJC-enhanced NMD. Indeed, the mRNA half-life of control construct ter310 NFB was only 1.92 h ( Figure 4B) and the steady-state mRNA level only 1.1% of the corresponding WT NFB mRNA ( Figure 4C), indicative of efficient EJC-enhanced NMD. In contrast, the mRNA of the PTC-containing FB construct (ter310 FB) was reduced only to 30%-40% of WT NFB, and significantly stabilized, indicated by an average halflife of 4.69 h ( Figure 4B). Importantly, the intramolecular base pairing by itself did not stabilize the mRNA when it did not position the poly(A) tail close to the PTC (ter310 SL), and mutations in the base pairing region of ter310 FB reverted the transcript back into an NMD substrate ( Figure S4D-S4F). Rather than fitting a simple exponential decay curve, we noticed that ter310 NFB and ter310 SL mRNA exhibit a fast initial decay rate that slows down at lower mRNA levels. The reason for this apparently bi-phasic decay kinetics is currently not known. Because it is not observed with EJCindependent NMD reporters, it may reflect mechanistic differences between EJC-independent and EJC-enhanced NMD (see Discussion). Consistent with the result from the decay assay, NMD inhibition by RNAi-mediated Upf1 depletion ( Figure 4C and 4D) or treatment of the cells with the translation inhibitor cycloheximide ( Figure S5) elevated the mRNA levels of the NMD substrates ter310 NFB and ter310 SL by a factor of 10-20; ter310 FB behaved like WT NFB. We conclude that positioning of the poly(A) tail near a PTC efficiently suppresses both EJC-independent and EJCenhanced NMD. The latter result shows that the NMDinhibiting signal (poly(A) tail proximity) efficiently competes with the NMD-promoting signal (the downstream EJC), resulting in a strong attenuation of NMD.

PABPC1 Is an NMD Antagonizing Signal
But which constituent of the poly(A) tail has the capacity to suppress NMD? It was recently shown for S. cerevisiae [18] and for Drosophila cell lines [19] that the poly(A)-binding protein inhibits NMD when tethered near the PTC on a NMD reporter transcript. To test if tethered poly(A)-binding protein inhibits NMD also in human cells, we tested both the nuclear and the cytoplasmic poly(A)-binding proteins (PABPN1 and PABPC1) expressed as N-terminal fusions to an HA-tagged variant of the MS2 coat protein and assessed their effect on minil and TCR-b NMD substrates that harbor six MS2 binding sites either about 50 nucleotides downstream of the PTC (constructs A) or further away as a control (constructs B, Figure 5). Western blotting confirmed comparable expression levels of the different fusion proteins. Tethering of PABPC1 caused a strong increase in both reporter mRNAs when tethered nearby the PTC, indicative for NMD suppression. In contrast, tethering of PABPN1, the MS2 domain alone, PABPC1 expression without the MS2 domain, or a fragment of b-galactosidase with similar mass as PABPC1 did not significantly stabilize the reporter mRNAs. Thus, as in D. melanogaster, only the cytoplasmic but not the nuclear PABP inhibited NMD when located in the vicinity of the PTC [19].

Evidence for an Evolutionarily Conserved Mechanism for PTC Recognition
In summary, our results demonstrate that in human cells the proximity of PABPC1 provides an important signal for defining a translation termination event as ''correct'' and prevents degradation of the mRNA by NMD. Vice versa, translation termination too distant from PABPC1 lacks this signal, and as a consequence, NMD ensues. Our results are entirely consistent with models previously proposed for NMD in S. cerevisiae [18,21,28] and strongly argue that the basic mechanism for PTC recognition is much more conserved among eukaryotes than previously assumed. In essence, the current data suggest that the two antagonizing signals (PABPC1 proximity and Upf1-3 recruitment, respectively) determine if a translation termination event is defined as premature or correct. The recent characterization of PTC124, a small chemical entity that selectively induces ribosomal readthrough of premature but not normal TCs  [29], further supports the idea of a mechanistic difference between translation termination at a PTC and termination at a ''normal'' TC.
Our finding that NMD suppression mediated through the poly(A) tail gradually declines with increasing distance between the TC and the poly(A) tail ( Figure 3) is consistent with evidence suggesting that Upf1 and PABPC1 competes for interaction with release factor eRF3 bound to the ribosome at the TC [30]. Furthermore, our ''distance model'' provides a possible explanation for the reported distance effect of PABPC1 tethering to a b-globin NMD reporter transcript [31]. Consistent with our results on minil ( Figures  3 and 5), NMD of this b-globin mRNA was suppressed more efficiently by tethering PABC1 45 nucleotides downstream of the PTC than tethering it 132 nucleotides downstream of the PTC [31]. The postulated requirement for correct translation termination to occur within a certain maximal physical distance from PABPC1 could also explain why PTCs near the start codon fail to trigger efficient NMD ( Figure 1A and [32]), assuming that the start codon and the poly(A) tail are located in spatial proximity due to the interaction between eIF4G and PABPC1 [22].

Mammalian EJCs Have Evolved to Function as NMD Enhancers
Our data further show that in mammals, the EJC is not required for PTC recognition, as in C. elegans and D. melanogaster [20,33]. But unlike C. elegans and D. melanogaster, where the EJC does not appear to affect NMD at all, the EJC plays an important role as an enhancer of NMD in mammals. The comparison between NMD of minil ter440 and the minil ter440 C3/C4 (39-most intron deleted) represents an example for such an EJC-mediated enhancement of NMD [17]. The simplest mechanistic explanation for the NMDenhancing effect of EJCs is that they accelerate SMG1/Upf2/ Upf3-dependent phosphorylation of Upf1 by locally concentrating Upf2 and Upf3 [34]. We demonstrated here that even in a situation of such EJC-enhanced NMD, positioning the poly(A) tail between the PTC and the EJC still strongly suppressed NMD (Figure 4), indicating that the translation termination-promoting signal in this position still efficiently competed with NMD-promoting events. Confirming our result and showing that PABPC1 is necessary for this NMD suppression, Ivanov and colleagues found that artificially inducing NMD by tethering the EJC factor Y14 into the 39 UTR of a b-globin reporter mRNA was suppressed by tethering PABPC1 between the TC and Y14 [35].
Noteworthy in the context of suppressing EJC-enhanced NMD, the mRNA decay kinetics of the reporter constructs in Figure 4B do not fit a simple exponential decay curve, but rather these transcripts seem to be degraded in an apparently bi-phasic mode. The initially fast decay rate might reflect the NMD enhancing effect of the EJC during the pioneer round of translation [25], whereas the second, somewhat slower decay rate would signify EJC-independent decay. We hypothesize that in mammals, where a large number of nonsense transcripts is produced by extensive alternative splicing [36], the EJC has evolved to function as an enhancer of NMD by locally concentrating Upf2 and Upf3b nearby terminating ribosomes and thereby tilting the balance of the two antagonizing signals toward NMD.

Implications for Disease-Associated Mutations
Collectively, these data have potentially important clinical implications. Our findings predict that mutations leading to extended 39 UTRs, such as poly(A) site mutations or sequence insertions into the 39 UTR, constitute a so far overlooked group of NMD substrates that may explain the molecular mechanism of certain genetic diseases. For such mutations, treatments with readthrough-promoting drugs like PTC124 [29] would not be suitable, because PTC124 does not stabilize the mRNA and would lead to the synthesis of C-terminally extended wild-type protein. In contrast, mRNA stabilization by a FB strategy as described here would augment wild-type protein levels and therefore represent a putative genespecific therapeutic approach, provided the FB can be induced in trans.

Posttranscriptional Gene Regulation by Means of NMD
It has not escaped our notice that the new, unified NMD model we postulate immediately suggests a possible mechanism for posttranscriptional regulation of a wide variety of genes by NMD. It is well documented that 39 UTRs, many of which comprise several thousand nucleotides in mammals, serve as binding sites for numerous factors that regulate mRNA translation or stability [26,27]. We postulate that by binding to their target transcript, many of these factors alter the tertiary structure of the 39 UTR, thereby changing the local environment for translation termination (i.e., the physical distance between the TC and the poly(A) tail), which in turn will amend the transcript's half-life ( Figure 6). 39 UTRbinding factors can change the 3-D 39 UTR configuration by masking mRNA sequences otherwise engaged in intramolecular base pairing, or by interacting with each other and thereby looping out mRNA sequence in-between. Proteinprotein and protein-RNA interactions can be regulated through signal transduction pathways by posttranslational modification of the involved RNA-binding proteins. Furthermore, this NMD-dependent posttranscriptional gene regulation can also be modulated through transcript-specific RNA-binding proteins with intrinsic NMD-promoting or translation termination-promoting activity that binds into the proximity of the TC. For example, the RNA-binding protein Staufen has been reported to bind the 39 UTR of a few specific mRNAs and to induce their rapid degradation by directly recruiting Upf1 [37,38]. Although it remains to be further investigated to which extent cells use this gene regulation pathway, the surprisingly large number of physiological transcripts detected by microarray analysis that raise in levels upon Upf1 knockdown indicates that it might be widespread [39][40][41][42][43]. A central prediction of this NMDmediated gene regulation mechanism is that it depends on ongoing translation and that one would expect the set of transcripts affected by Upf1 depletion to vary in a tissuespecific manner, during development and differentiation, and by environmental cues in general. This might explain why the sets of transcripts affected by Upf1 depletion in the different microarray studies showed only limited overlap [39][40][41][42][43]. To test the postulated mode of gene regulation by spatial remodeling of the 39 UTR more directly, development of techniques that allow in vivo measurements of the physical distance between two molecules and improved predictions of mRNA folding in the presence of RNA-binding proteins will be necessary.
Cell culture and cycloheximide treatment. HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM, Invitrogen), supple-mented with 10% heat-inactivated fetal calf serum (FCS), 100 U/mL penicillin, and 100 lg/mL streptomycin (Amimed). For the experiment in Figure S5, 44 h after transfection of the reporter plasmids, 100 lg/mL cycloheximide was added for 4 h before RNA isolation.
Half-life measurements. The Tet-Off Advanced Transactivator (tTA-Advanced) was stably integrated into the genome of HeLa cells according to the manufacturer's protocol (Tet-Off Advanced Inducible Gene Expression System, Clontech) and a cell clone with high tTA-Advanced expression was selected for further experiments.
To determine decay kinetics of the minil reporter mRNAs, 2 3 10 5 per well of the tTA-Advanced expressing cells were seeded in 6-well plates. The next day, two wells for each time course were transfected with 100 ng pTRE-Tight minil reporter plasmid and 100 ng pBS bglobin WT plasmid (for normalization) per well, using 2 ll DreamFect (OZ Biosciences) according to the manufacturer's protocol. On the following day, the cells of these two wells were split into six wells. Time course was started 40 h after transfection by adding 1 lg/ml doxycycline (Sigma) to each well and harvesting the cells after 0, 1, 2, 4, 6, and 8 h.
RNAi. Knockdown of hUpf1, hUpf2, and hUpf3b was induced by transfection of pSUPERpuro plasmids targeting two different sequences in hUpf1 [48], one sequence in hUpf2 [43], or two different sequences in hUpf3b (see above), respectively. Starting 24 h after transfection, untransfected cells were eliminated by culturing the cells in the presence of 1.5 lg/mL puromycin for 48 h. Cells were then washed in PBS and incubated in puromycin-free medium for another 24 h. Total cellular RNA was isolated and whole cell lysates for Western blotting were prepared 96 h post transfection. The efficiency of the knockdown was assessed on the mRNA level by realtime RT-PCR (unpublished data) and on the protein level by Western blotting.
Northern blot analysis. Total cellular RNA (10 lg) was separated on a 1.2% agarose gel containing 13 MOPS and 1% formaldehyde. RNA was transferred to positively charged nylon membrane (Roche) in 0.53 MOPS by 1-h wet blotting in a genie blotter (Idea Scientific). Following UV crosslinking of the RNA to the nylon filter, prehybridization and hybridization of the blot in Figure 1B was carried out in 63 SSC, 53 Denhardt's reagent, and 0.5% SDS with 50 lg/mL denatured salmon sperm DNA and 100 lg/mL denatured calf thymus DNA at 60 8C. For hybridization, 100 ng lter310 and 20 ng b-globin DNA was labeled with a-32 P-dCTP using the Ready-To-Go DNA-Labeling Kit (Amersham). The blot in Figure S3 was hybridized with an in vitro-transcribed, a-32 P-UTP-labeled antisense minil RNA probe in ULTRAHyb buffer (Ambion) at 68 8C. After overnight hybridization, membranes were washed twice with 23 SSC/0.2% SDS and twice with 0.23 SSC/0.1% SDS at 60 8C before exposure to a PhosphorImager screen.
Immunoblotting. Whole cell lysates corresponding to 0.37 3 10 4 -2 3 10 5 cells per lane were electrophoresed on a 10% SDS-PAGE. Proteins were transferred to Optitran BA-S 85 reinforced nitrocellulose (Schleicher and Schuell) and probed with 1:2,500 diluted polyclonal rabbit anti-hUpf1, anti-hUpf2, or anti-hUpf3b antiserum [11], 1:1,000 diluted monoclonal mouse anti-lamin A/C (Santa Cruz Biotechnology) or anti-HA antibody (Roche), 1:400 diluted supernatant of the mouse hybridoma cell line Y12, which produces a monoclonal antibody against the human Sm B/B'proteins [51]. 1:2,500 diluted HRP-conjugated anti-rabbit IgG or HRP-conjugated antimouse IgG (Promega) was used as secondary antibody. ECLþ Plus Western blotting detection system (Amersham) was used for detection and signals were visualized on a Luminescent Image Analyzer LAS-1000 (Fujifilm). Figure S1. Monitoring of the Knockdown Efficacy of Upf1, Upf2, and Upf3b at the mRNA Level From the RNA samples of Figure 1D, relative mRNA levels of Upf1, Upf2, and Upf3b, normalized to endogenous GAPDH mRNA, were measured by RT-qPCR using the TaqMan assay Hs00161289_m1, Hs00210187_m1, Hs00224875_m1, and 432-6317E from Applied Biosystems. Average values of two qPCR runs are shown. Found at doi:10.1371/journal.pbio.0060092.sg001 (214 KB PDF).     Figure  4A, normalized and displayed as in Figure 4C, from cells treated (þCHX) or not (control) with 100 lg/mL cycloheximide for 4 h before RNA isolation. Average values and SD of five qPCR measurements from two independent experiments are shown. Found at doi:10.1371/journal.pbio.0060092.sg005 (241 KB PDF).