key: cord-0774373-03g8ly6x
authors: Belew, Ashton Trey; Meskauskas, Arturas; Musalgaonkar, Sharmishtha; Advani, Vivek M.; Sulima, Sergey O.; Kasprzak, Wojciech K.; Shapiro, Bruce A.; Dinman, Jonathan D.
title: Ribosomal frameshifting in the CCR5 mRNA is regulated by miRNAs and the NMD pathway
date: 2014-07-09
journal: Nature
DOI: 10.1038/nature13429
sha: 0ce959882a912b65a604fa42e75c0a16ff830f56
doc_id: 774373
cord_uid: 03g8ly6x

Programmed −1 ribosomal frameshift (−1 PRF) signals redirect translating ribosomes to slip back one base on messenger RNAs. Although well characterized in viruses, how these elements may regulate cellular gene expression is not understood. Here we describe a −1 PRF signal in the human mRNA encoding CCR5, the HIV-1 co-receptor. CCR5 mRNA-mediated −1 PRF is directed by an mRNA pseudoknot, and is stimulated by at least two microRNAs. Mapping the mRNA–miRNA interaction suggests that formation of a triplex RNA structure stimulates −1 PRF. A −1 PRF event on the CCR5 mRNA directs translating ribosomes to a premature termination codon, destabilizing it through the nonsense-mediated mRNA decay pathway. At least one additional mRNA decay pathway is also involved. Functional −1 PRF signals that seem to be regulated by miRNAs are also demonstrated in mRNAs encoding six other cytokine receptors, suggesting a novel mode through which immune responses may be fine-tuned in mammalian cells. SUPPLEMENTARY INFORMATION: The online version of this article (doi:10.1038/nature13429) contains supplementary material, which is available to authorized users.

Programmed 21 ribosomal frameshift (21 PRF) signals redirect translating ribosomes to slip back one base on messenger RNAs. Although well characterized in viruses, how these elements may regulate cellular gene expression is not understood. Here we describe a 21 PRF signal in the human mRNA encoding CCR5, the HIV-1 co-receptor. CCR5 mRNA-mediated 21 PRF is directed by an mRNA pseudoknot, and is stimulated by at least two microRNAs. Mapping the mRNA-miRNA interaction suggests that formation of a triplex RNA structure stimulates 21 PRF. A 21 PRF event on the CCR5 mRNA directs translating ribosomes to a premature termination codon, destabilizing it through the nonsense-mediated mRNA decay pathway. At least one additional mRNA decay pathway is also involved. Functional 21 PRF signals that seem to be regulated by miRNAs are also demonstrated in mRNAs encoding six other cytokine receptors, suggesting a novel mode through which immune responses may be fine-tuned in mammalian cells.

Viral programmed ribosomal frameshift events typically produce carboxy-terminally extended fusion proteins. However, computational analyses predict that .95% of 21 PRF events on cellular mRNAs direct ribosomes to premature termination codons (PTC), suggesting that 21 PRF may be used by cells to regulate gene expression by destabilizing mRNAs through the nonsense-mediated mRNA decay (NMD) pathway 1 . Whereas a role for 21 PRF has been shown in yeast 2, 3 , it has not been tested in higher eukaryotes so far. In yeast, mutants and drugs that globally affect 21 PRF generally promote deleterious phenotypes 4 , and global dysregulation of 21 PRF may contribute to human disease 3, [5] [6] [7] . How sequence-specific regulation of 21 PRF might be achieved has been the central unanswered question in the field.

A 21 PRF signal in the CCR5 receptor mRNA CCR5 is a cytokine receptor which is exploited by HIV-1 as a coreceptor for entry into CD4 1 T-cells 8 . A strong candidate 21 PRF signal beginning at nucleotide 407 in the human CCR5 mRNA was identified computationally 9 . This sequence is .99% conserved among the great apes and is highly conserved among the higher primates (Extended Data Fig. 1 ). Using dual luciferase reporters (Extended Data Fig. 2a) , the CCR5 sequence promoted 9-11% 21 PRF in HeLa (Fig. 1a ) and 4.5-6.3% in Chinese hamster ovary (CHO) or Vero cells (Extended Data Fig. 2b) . Mutagenesis of the slippery site from UUUAAAA to GCGCGCG reduced 21 PRF to ,1% (Fig. 1a) . Introduction of an inframe termination codon (PTC control) 59 of the CCR5-derived sequence, or placing the firefly luciferase reporter out of frame with respect to Renilla reduced 21 PRF levels by more than two orders of magnitude (Fig. 1a) , ruling out the possible presence of either a splicing donor site or an internal ribosome entry signal (IRES). An in vitro translation assay revealed a peptide consistent with a CCR5 21 PRF event at levels comparable to that promoted by the HIV-1 21 PRF signal (Fig. 1b) . Liquid chromatography dual mass spectroscopic analysis of an affinity purified CCR5-b-gal fusion protein (Extended Data Fig. 2c ) unambiguously identified the predicted 21 frameshift peptide harbouring the junction between the 02 and 21 frame encoded CCR5 sequence (Fig. 1c, Extended Data Fig. 2d ). Analysis of published ribosome profiling data from human cells 10 revealed a sizable fraction of ribosomes paused at the CCR5 21 PRF signal, 9/59 (,15%) of which were shifted into the 21 reading frame (Fig. 1d) . These four lines of inquiry demonstrate that this sequence in the human CCR5 mRNA promotes efficient 21 PRF.

Computational analyses predicted the presence of two nearly equivalent downstream mRNA pseudoknots or a tandem stem-loop structure immediately 39 of the slippery site (Extended Data Fig. 3a ). Analyses of chemical modification assays of a CCR5 runoff transcript (Extended Data Fig. 3b , c) were consistent with the presence of a two-stemmed mRNA pseudoknot (Fig. 2) . Whereas the slippery site distal region of Stem 1 is stable, the proximal region is conformationally dynamic, consistent with single-molecule optical trap experiments revealing a complex network of folding pathways for this element 11 . The weak slippery site proximal half of stem 1 coupled with the internal bulge is reminiscent of the HIV-1 21 PRF signal solution structure 12 and is consistent with the emerging view of conformational complexity as a critical feature of recoding pseudoknots 13, 14 . Stem 2 contains four semi-helical segments (labelled a, b, d, e in Fig. 2a) , plus a small segment in the middle (c), all separated by unpaired bases. The unpaired bases may allow the entire structure to bend, enabling U23 to bridge the gap between C22 and U24. The 'best fit' conformer diagrammed in Fig. 2a was used as the basis for molecular-dynamics-based simulation of the CCR5 21 PRF stimulatory mRNA structure (Fig. 2b, Extended Data Fig. 3d, e) . The root mean squared deviation (r.m.s.d.) average structure was calculated for the last 12 ns of an 80-ns long molecular dynamics simulation, where the r.m.s.d.s of the full structure and its two sub-domains SL1 (nucleotides 8-22 and 55-75) and SL2 (nucleotides 24-53 and 76-103) are most stable. The total energy for this structure is 224,296 kcal mol 21 .

As cellular gene expression tends to be regulated, it is reasonable to hypothesize that 21 PRF might be regulated in a sequence-specific manner. This could be achieved through base-pairing interactions between specific small noncoding RNAs (ncRNAs) and 21 PRF signals, a hypothesis supported by the ability of antisense oligonucleotides to stimulate 21 PRF (reviewed in ref. 1) . Computational searches revealed miR-1224, miR-711 and miR-141 as potential interacting partners with the CCR5 21 PRF signal (Extended Data Fig. 4a ). Transfection of HeLa cells with a miR-1224 precursor revealed concentration-dependent enhancement of CCR5-mediated 21 PRF (Fig. 3a) . miR-1224 did not affect HIV-1-mediated 21 PRF. Addition of a miR-1224 antagomir (anti-miR-1224), or short interfering RNA knockdown of argonaute 1 reversed the effect of miR-1224 on CCR5-mediated 21 PRF (Fig. 3b) . Although anti-miR-1224 seemed to stimulate CCR5 21 PRF, the effect was not significant (P 5 0.15). siRNA knockdown of mRNAs encoding proteins involved in miRNA processing inhibited CCR5-mediated 21 PRF, but stimulated HIV-1-driven 21 PRF (Fig. 3c ), supporting the model of sequence-specific regulation of 21 PRF through interactions between miRNAs and 21 PRF signals. Neither miR-141 nor miR-711 affected 21 PRF in a HeLa-cells-based assay (Extended Data Fig. 4b) , perhaps owing to the presence of endogenous miR-711, and/or miR-141. However, miR-141 specifically stimulated CCR5-mediated 21 PRF in CHO cells (Extended Data Fig. 4c ).

Two different in vitro electrophoretic mobility shift assays (EMSAs) were used to probe the interactions between the CCR5 21 PRF signal and miR-1224. In one, the RNAs were mixed and incubated at physiological temperature ('native'), whereas in the second, they were codenatured at high temperature and slowly annealed ('refolded'). Both a, Measurement of 21 PRF in HeLa cells. 21 PRF efficiency was monitored in HeLa cells using dual luciferase reporters. Error bars denote an approximation of standard errors. ***P , 0.001 compared to the out of frame control (Student's two-tailed t-test). b, Efficient 21 PRF promoted by the CCR5 sequence in vitro. Autoradiogram of in vitro translation reaction using mRNAs harbouring CCR5-or HIV-1-derived 21 PRF signals. Green arrows denote 0-frame encoded products. Red arrows denote 21 PRF encoded peptides. RTC indicates the readthrough control. Percentage 21 PRF promoted by CCR5 and HIV-1 frameshift signals is indicated below the lanes. c, Liquid chromatography with tandem mass spectrometry (LC-MS/MS) spectrum of a proteolytic fragment containing the CCR5 frameshift peptide. N-terminally acetylated leader peptide sequence is coloured blue, CCR5-derived 0-frame sequence beginning at V94 is red, and CCR5 21 frame encoded sequence beginning after L101 is coloured green. d, Ribosomes accumulate at the CCR5 21 PRF signal. Data mined from ref. 10 . Top, locations of the 21 PRF signal and first 21 frame termination codon are indicated. Bottom, profiling data at the slippery site (indicated in capital letters) at single nucleotide resolution. Ribosomes arrested in the three different reading frames are colour-coded. a, HeLa cells were transfected with 0-30 nmol of miR-1224 miRNA expressing constructs and with HIV-1 or CCR5 21 PRF reporters. b, PRF assays of HeLa cells mock-transfected (0), or transfected with scrambled miRNA (Scr), a miR-1224 antagomir (anti-1224), miR-1224, miR1224 1 anti-miR-1224, or miR-1224 plus an siRNA directed against argonaute 1. c, Ablation of the miRNA processing machinery affects 21 PRF promoted by the HIV-1 and CCR5 frameshift signals. 21 PRF assays were performed using cells transfected with siRNAs targeting Argonaute 1 (AGO1), Argonaute 2 (AGO2), DGCR8, exportin 5 (XPO5) or scrambled sequences (Scr). Error bars denote standard error. *P , 0.05, **P , 0.01 (Student's two-tailed t-test).

reactions were resolved through native polyacrylamide gel electrophoresis (PAGE; Extended Data Fig. 5a, b) . Although miR-1224 interacted with the CCR5 sequence with sub-nanomolar dissociation constants in both conditions, its affinity was approximately twofold higher in the 'native' context ( Fig. 4a) . miR-1224 enhanced the appearance of multiple pre-existing conformers, particularly in the 'refolded' context, consistent with the structurally complex nature of the pseudoknot. miR-1224 did not interact with a transcript containing the HIV-1 21 PRF signal (Extended Data Fig. 5c, d) .

An affinity capture assay to probe CCR5-miR-1224 interactions in HeLa Tzm-BL cells expressing CCR5 15 revealed an approximately threefold enrichment for CCR5 mRNA relative to cells transfected with a scrambled control (Fig. 4b) . In a parallel experiment in HeLa cells, the CCR5 21 PRF signal containing dual-luciferase reporter mRNA was enriched more than 2,000-fold compared to no-miRNA controls (Fig. 4c) , whereas the HIV-1 21 PRF reporter was only enriched about tenfold. These findings demonstrate that miR-1224 specifically interacts with the CCR5 21 PRF signal in live cells. Selective 29-hydroxyl acylation analysed by primer extension (SHAPE) did not reveal differences in RNA modification patterns in the presence of miR-1224 (Extended Data Fig. 5e ), suggesting that miR-1224 does not function to create any new conformation(s) of the CCR5 21 PRF signal per se. Rather, it may stabilize a pre-existing structure(s) promoting efficient 21 PRF.

Mapping the miR-1224 binding site CCR5-derived transcripts harbouring mutations in the predicted miR-1224 interacting sequences (mutants M1-M3, Extended Data Fig. 6a) were assayed by EMSA. The 59 proximal mutant (M1) yielded the same K D (0.76 nM) as the wild-type sequence under 'refolded' conditions, the central sequence mutant (M2) promoted the same dissociation constant as the wild-type sequence assayed under 'native' conditions (0.36 nM), and the 39 proximal binding site mutant (M3) caused an approximately 100-fold increase in K D (42 nM) (Fig. 4d , Extended Data Fig. 6b, c) . These findings suggest that miR-1224 may participate in a triple helical interaction with subdomains a-d of Stem 2 under native conditions (modelled in Fig. 4e and Extended Data Fig. 6c ). The predicted triple-base interaction between miR-1224 and Stem 2 is consistent with the stable 39 end of the pseudoknot identified in the molecular dynamics simulation (Extended Data Fig. 3d , e). The 'torsional restraint' model of 21 PRF posits that ribosomes are directed to pause over the slippery site by Stem 2-induced supercoiling of Stem 1 16 . The miR-1224 mapping data are consistent with this model: increased stability of the Stem 2 by the mRNA-miRNA interaction renders this structure even more difficult to resolve, further increasing the fraction of paused ribosomes. miR-141 is predicted to interact with the same region of the CCR5 21 PRF signal whereas miR-711 is not, suggesting that miR-141 enhances CCR5-directed 21 PRF in a similar manner to miR-1224. The sequence of the mature miR-1224 is 100% conserved among higher primates (Homo, Pan, Pongo and Macaca) as is its binding site with the 39 end of their respective CCR5 21 PRF signals, suggesting that miR-1224-mediated regulation of CCR5 21 PRF is evolutionarily conserved. It is also notable that the miR-1224/CCR5-interacting sequences do not conform to established seed sequences for miRNAs.

Ribosome profiling data also revealed a cluster of ribosomes paused at the first 21 frame termination codon after the CCR5 slippery site (Fig. 1d) . A series of rabbit b-globin-derived reporters (Extended Data Fig. 7a ) were used to assess the effects of the CCR5 21 PRF signal on mRNA steady-state abundance and stability. Steady-state abundance of the CCR5 21 PRF-containing reporter mRNA was about 38% of the readthrough control and was further decreased upon addition of miR-1224 (,10% of readthrough control), consistent with an inverse correlation between 21 PRF efficiency and mRNA abundance 2 (Fig. 5a ). An in-frame PTC strongly decreased mRNA abundance (,1% of readthrough control). A reporter with the tumour necrosis factor (TNF)-aderived AU-rich element (ARE) in its 39 untranslated region (UTR) 17 reduced mRNA abundance to ,22%. In combination with the CCR5 21 PRF signal, mRNA abundance was reduced to ,6%, consistent with NMD and ARE-mediated decay operating independently of one another. The CCR5 slippery site mutant (SSM) decreased reporter mRNA abundance to ,64%, and addition of miR-1224 decreased this to ,47%. The former finding suggests that the stable mRNA pseudoknot has mRNA destabilizing activity independent of frameshifting, perhaps through the no-go mRNA pathway as described in yeast 18 . Its stabilization by miR-1224 may enhance this process. Alternatively, miR-1224 may promote accelerated mRNA turnover through canonical miRNA-mediated translational repression 19 . However, if this were true, miR-1224 should have reduced SSM mRNA abundance to the same extent as the native sequence. Abundance of the CCR5 21 PRF signal containing reporter mRNA was increased by about 4.4-fold by partial siRNA knockdown of NMD (Fig. 5b) . Abundance of the SSM construct was not increased by NMD ablation, consistent with its mRNA Biotinylated miR-1224 precursor or a scrambled biotinylated control (Scr) were transfected into HeLa TZM BL cells expressing CCR5. Fold enrichment of affinity purified mRNAs were analysed by quantitative PCR with reverse transcription (qRT-PCR) using CCR5-or GAPDH-specific primer sets. c, HeLa cells were co-transfected with dual-luciferase plasmids containing either the CCR5 or HIV-1 21 PRF signal sequences and affinity-purified mRNAs were analysed as in b. d, EMSA assays were performed using miR1224 and M1, M2 and M3 variants of the CCR5 signal using native conditions. Single site binding isotherms generated from these data are plotted. K D values are indicated. For a and d, n 5 6 for each sample (three times each of two technical replicates). For b and c, n 5 9 for each sample (three times each for three biological replicates. Error bars denote standard deviation. *P , 0.05, ***P , 0.001 (Student's two-tailed t-test). e, Conceptual model of CCR5 pseudoknot complexed with miR-1224 (purple).

destabilizing activity being independent of 21 PRF-directed NMD. A transcriptional arrest time course experiment showed that the CCR5 21 PRF signal rendered the reporter mRNA a direct substrate for NMD: its half-life was reduced to about 180 min whereas NMD ablation increased the half-life to about 380 min (Extended Data Fig. 7b ).

In HeLa Tzm-BL cells the abundances of both CCR5 mRNA and CCR5 protein increased proportionally to the extent of NMD abrogation (Fig. 5c, d, siRNA_SMG1) . Conversely, addition of miR-1224 decreased both CCR5 mRNA and CCR5 protein abundance. Abrogation of miRNA processing by siRNA knockdown of AGO1 or DGCR8 resulted in increased abundance mRNA and protein, consistent with inhibition of CCR5-mediated 21 PRF under these conditions. Transcriptional arrest time-course experiments demonstrated that the CCR5 mRNA is a direct substrate for NMD (Extended Data Fig. 7c ). Whereas miR-1224 decreased the abundance of CCR5 mRNA, this effect was abrogated by addition of an anti-miR-1224 antagomir, but antagomir alone had no effect (Extended Data Fig. 7d ). siRNA knockdown of SMG1 was epistatic to miR-1224, consistent with the mRNA destabilization activity of the miRNA being NMD-dependent. This is also consistent with findings that human UPF1 may participate in RNA silencing 20 , with the caveat that miR-1224 may also promote mRNA degradation by a mechanism that is independent of the NMD machinery, for example, No-go decay. Combinations involving human SMG1 siRNA knockdown plus miR-1224, human SMG1 siRNA knockdown plus the antagomir, or all three together were also supportive of this model.

To our knowledge before the current study, only three 21 PRF signals were known in mammalian genomes, all thought to be remnants of ancient retroviral insertional events [21] [22] [23] . Potential 21 PRF signals in seven additional interleukin receptor subunit mRNAs were assayed in the presence of either a scrambled siRNA control or an siRNA targeting argonaute 1. Efficient 21 PRF (.1%) was elicited by six of these ( Fig. 5e ). siRNA knockdown of argonaute 1 stimulated 21 PRF in some cases and inhibited it in others, consistent with sequence-specific regulation of 21 PRF by miRNAs. Ribosome profiling data 10 revealed ribosomes paused and directed to new reading frames at three of these signals (Extended Data Fig. 8 ). Single nucleotide polymorphisms (SNPs) capable of disrupting frameshifting activity were identified in all these 21 PRF signals (Extended Data Fig. 8 ). These may account for disease phenotypes associated with SNPs that do not alter the primary amino acid sequences of their encoded proteins.

To summarize, precise regulation of 21 PRF is accomplished by sequence-specific interactions between individual 21 PRF signals and naturally occurring miRNAs. That global ablation of miRNA processing differentially affected 21 PRF promoted by many different signals suggests that miRNA-mediated control of 21 PRF is the biologically significant norm. This confers sequence specificity, and is energetically less expensive than producing new, or modifying pre-existing ribosomes. It may also enable rapid regulation of 21 PRF on specific mRNAs within individual cells or intracellular compartments. This solves the central question, unanswered until now, of how 21 PRF may be regulated in a sequence-specific manner, and suggests a novel mode through which 21 PRF signals may be targeted for therapeutic intervention. To our knowledge, this is also one of the few demonstrations of an miRNA affecting the expression of a cellular gene through an interaction with its ORF. The discovery of 21 PRF signals in the mRNAs encoding cytokine receptors has a potentially profound impact on our understanding of immune homeostasis. Although a robust immune response is critical for limiting and preventing infection, left uncontrolled, it can rapidly result in pathology and death. Despite a large body of literature describing how expression of small-peptide mediators of the immune response are regulated at the level of mRNA stability, this only provides a global mechanism of immune regulation by controlling production of effector molecules. In contrast, the ability to control expression of cytokine receptors through 21 PRF induced NMD, and how rates of 21 PRF in turn may be controlled by miRNAs, represents a way for individual recipient cells to modulate their responses to cytokines; this would provide the means to fine-tune immune responses, and suggests a novel molecular mechanism underlying immune desensitization. The studies described here also have consequences for directing antiviral efforts. RNA viruses such as retroviruses, coronaviruses, alphaviruses and totiviruses require extremely stringent levels of 21 PRF for their propagation 1 . We suggest that their 21 PRF promoting structural elements may have evolved in two different ways so as to ensure set The effects of transfected RNA species on CCR5 mRNA steady-state abundance were assayed by qRT-PCR. d, Quantitative sandwich enzyme-linked immunosorbent assay (ELISA) of samples from c. Cell lysates (16 mg protein per sample) were assayed and total amounts of CCR5 protein were determined relative to standards. e, Computationally identified putative 21 PRF signals assayed in HeLa cells transfected with an siRNA directed against AGO1 or a scrambled siRNA control. Numbers in human IL8Ra denote the nucleotide positions of the beginning of the slippery sites in the native mRNA. a-c, n 5 9 (three times on three independent biological replicates). d, n 5 8 (quadruplicate assays of two independent biological replicates). Error bars denote standard error. *P , 0.05, **P , 0.01 (Student's two-tailed t-test).

rates of 21 PRF. First, either their 21 PRF stimulatory elements should not interact with any ncRNAs present in the cells in which they replicate, or their 21 PRF signals may have evolved in the presence of trans-acting ncRNAs specific to their host cells. If the latter is true, as suggested by stimulation of HIV-1-promoted 21 PRF in response to siRNA knockdown of argonaute, this may define a new parameter governing host cell permissiveness, presenting a novel therapeutic targeting opportunity. Thus, while the discovery of operational 21 PRF signals in cellular mRNAs suggests that global targeting of 21 PRF may not be the wisest approach, discovery and subsequent targeting of specific cellular miRNAs required by viruses to ensure proper rates of 21 PRF may present a more narrowly targeted therapeutic option.

HeLa, HeLa Tet-Off, HeLa TZM-BL, CHO and Vero cells were cultured according to suppliers instructions. Insertions were amplified using PCR and ligated into appropriate backbone plasmids. 21 PRF was assayed in live cells using dual-luciferase assays. Liquid chromatography with tandem mass spectrometry (LC-MS/MS) analysis of affinity purified CCR5/b-galactosidase fusion protein digested with Asp-N was performed at the University of Maryland Proteomics Core Facility. Data generated by chemical modification assays were used for in silico threedimensional modelling. Sandwich enzyme-linked immunosorbent assay (ELISA) was used to monitor CCR5 protein expression, and quantitative PCR with reverse transcription (qRT-PCR) analyses were used to monitor mRNA steady-state abundance and half-lives. In vivo affinity capture used a double stranded miR-1224-5p RNA containing a sense strand 59 biotin modification and mismatch transfected into HeLa and HeLa TZM-Bl cells. Electrophoretic mobility shift assays used a synthetic hsa-miR-1224-5p and transcripts harbouring the CCR5 or HIV-1 21 PRF signals.

Online Content Methods, along with any additional Extended Data display items and Source Data, are available in the online version of the paper; references unique to these sections appear only in the online paper. Sample sizes were determined following the rigorous criteria previously described 25 .

A minimum of 15 (three independent biological replicates assayed in quintuplicate) were performed per sample, or until normal distributions were obtained. Statistical analyses were performed as previously reported 25 .

To monitor 21 PRF in vitro, linear DNA templates were prepared by PCR reactions using plasmids pJD175f, pJD187, pJD827, pJD1078, and the T7-Kozak-Renilla/PolyA-Stop-firefly primer set. PCR products were purified by agarose gel electrophoresis. Capped mRNAs were synthesized using the mMESSAGE mMACHINE Kit (Ambion). Transcription reactions (40 ml) were assembled containing 2 mg linearized DNA templates, incubated at 37 uC for 4 h, and mRNAs were purified using a MEGAclear mRNA purification kit (Applied Biosystems). mRNA concentrations were calculated from OD 260 nm readings. In vitro translation reactions were assembled in a total volume of 25 ml containing 0.2 to 5 mg capped mRNAs using the Retic Lysate IVT Kit (Applied Biosystems). Reactions contained [ 35 S]methionine (1,175 Ci mmol 21 , Perkin Elmer) and the 203 -Met Translation mix provided by the kit. Reactions were incubated at 30 uC for 90 min. Translation products were resolved through 12% SDS-PAGE, and translation products were visualized using a phosphorimager. Production and purification of CCR5/b-galactosidase 21 PRF fusion protein in yeast. Yeast strain JD1585 was created by transformation of yeast strain JD1370 with pJD1930 and selection on defined medium lacking tryptophan (2Trp). JD1585 was inoculated from overnight grown -Trp plate into 2 ml of -Trp and incubated at 30 uC in shaker overnight. This culture (0.5 ml) was used to inoculate 50 ml of -Trp in a 200 ml baffled flask and grown overnight at 30 uC. Five ml of this culture was then used to inoculate 500 ml -Trp in a 2 l volume baffled flask. Five litre of culture (10 flasks of 500 ml each) was incubated overnight at 30 uC with shaking (250 r.p.m.). Cells (35 g wet weight) were collected by centrifugation (10 min at 4,000g at 4 uC). Cells were washed and suspended in lysis buffer (0.5 g ml 21 ) and disrupted with glass beads (0.5 mm, MiniBeadbeater, Biospec). Fos-choline-12 (Affymetrix) detergent was added to 0.05% and homogenate was brought to 1.5 M NaCl by adding 5 M stock NaCl. Cells and large debris were removed by centrifugation at 4,000g for 10 min. The supernatant was cleared by centrifugation at 30,000g for 30 min and loaded on 4-aminophenyl-b-D-thiogalactopyranosideagarose 4B (Sigma) column (1 ml bed volume). The column was washed with 50 volumes of lysis buffer containing 1.5 M NaCl. Recombinant CCR5-b-galactosidase protein was eluted with 0.1 M sodium borate buffer (pH 10.0) and concentrated on Ultracel-50K filter units (Amicon). b-galactosidase activity was monitored during purification using b-galactosidase Assay Kit (Pierce). Eluted proteins were fractionated through 8% SDS-PAGE. A protein band of expected size (124.7 kDa) was cut from gel and analysed by mass spectrometry at the UMD Mass Spectroscopy Core Center. Mass spectroscopic analysis of CCR5/b-galactosidase 21 PRF fusion protein.

Sequencing grade Asp-N was purchased from Promega (Madison, WI). Triethylammonium bicarbonate, and iodoacetamide were purchased from Supelco. DTT was from Sigma. Formic acid, optima grade water and acetonitrile are from Fisher Scientific. In-Gel digestion was carried out following manufacturer's protocol. Briefly, gel was cut into pieces ,1 mm 3 , destained with 50% acetonitrile (ACN) in water, dehydrated with ACN, then rehydrated with 25 mM DTT, incubated at 65 uC for 20 min, washed, dehydrated with ACN. Samples were rehydrated with 50 mM iodoacetamide, incubated at room temperature in the dark for 25 minutes, washed with water, and dehydrated with ACN. Gel was then rehydrated with 10 ng ml 21 Asp-N solution, and incubated overnight at 37 uC. Peptides were extracted twice by sequential addition of 100 ml 50% ACN, 100 ml 20% formic acid, and 100 ml. Extracts were combined and concentrated with speedvac to dryness. Liquid chromatography with tandem mass spectrometry (LC-MS/MS) analysis. Peptides were redissolved in 50 ml solvent A and loaded into a trapping cartridge (0.3 3 5 mm, Agilent, Palo Alto, CA) with autosampler, and desalted with 100% solvent A at 10 ml min 21 for 10 min. Peptides were separated using a Zorbax 300 B-C18 nano column (3.5 mm, 0.075 3 150 mm, Agilent Technologies, Palo Alto, CA) with a binary gradient consisting of A: 0.1% formic acid with 2.5% acetonitrile and B: 0.1% formic acid and 97.5% acetonitrile at a flow rate of 200 nl min 21 . A gradient was run from 5% B to 25%B over 60 min, followed by a 50-min gradient to 50% B, and 10-min gradient to 80%B. The gradient was held at 80% B for 5 min before returning to 5% B. Positive ion mass spectra of Nano LC eluents were acquired with a Thermo Finnigan LTQ Orbitrap XL mass spectrometer with data dependent analysis in which a full Scan FT analysis of m/z 400-1,800 with resolution at 60,000 (m/z 400) in the Orbitrap is followed by up to 5 MS/MS analyses in the linear ion trap at unit mass resolution. Peptides eluting from the HPLC column that have ions above 10,000 arbitrary intensity units and charges higher than 1 trigger the ion trap to isolate the ion and perform an MS/MS experiment scan after the MS full scan. Dynamic exclusion was set at 1 repeat count and 60 s exclusion time. Data processing. A common contaminants protein database was downloaded from http://maxquant.org. Sequence of CCR5-LacZ was inserted into the database before the database was imported into the in-house Mascot Server and Proteome Discoverer 1.4 program. Raw data from LTQ Orbitrap was searched against the modified common contaminant database using Sequest HT and Mascot with AspN as digestion enzyme with up to 1 missed cleavage. Peptide mass tolerance was 6 20 p.p.m., and fragment mass tolerance was 6 0.8 Da. Carbamidomethyl (C), Deamidated (NQ), Oxidation (M), Acetyl (N-term) are set as variable modifications. Decoy search was performed and false discovery rate of ,1% required for positive Identification. Search results showed 36% sequence coverage of the CCR5-LacZ protein. The N-terminal peptide was unambiguously identified, with loss of the initial methionine and acetylation of the N-terminal threonine. Quantification of protein levels by ELISA. HeLa TZM-BL were transiently transfected with siRNAs against SMG1, AGO1, DGCR8, miRNA precursor of hsa-miR-1224-5p, scrambled sequence or mock transfected using HiPerFect transfection reagent (Qiagen). The cells were lysed using 0.5% Fos-choline-12 (Affymetrix) in 50 mM Tris-HCl, pH 8.0 lysis buffer. Protein concentrations of cell lysates were determined by Bradford assay (Bio-Rad). Concentrations of CCR5 protein were measured in 16 mg total protein of cell lysates using the CCR5 sandwich ELISA kit (US Biological Life Sciences), according to the manufacturer's protocol. All assays were repeated in triplicate. Quantitative real-time reverse-transcription PCR. For qPCR analyses of the b-globin based reporters, assays were performed as previously described 26 with the following modifications. The dual luciferase readthrough control was used for RESEARCH ARTICLE co-transfections rather than EGFP. RNA samples for qPCR were isolated using the RNAqueous kit (Ambion), digested with rDNase (Ambion) and analysed using agarose gel electrophoresis and/or OD 260/280 measurements. The remaining samples were reverse transcribed using the iScript cDNA kit (Bio-Rad). The resulting complementary DNAs were diluted to 1:50-10,000 depending on mRNA concentration. Reactions were performed using 10 ml of LightCycler 480 SYBR Green I Master mix (Roche), 0.2-0.3 mM of each oligonucleotide, 2 ml of cDNA, and water to 20 ml per well. Samples were assayed for genomic DNA contamination by performing the assay using wells containing 1-2 ml of digested mRNA instead of cDNA. Reactions were amplified using either a Roche 480 LightCycler or a Bio-Rad CFX 96 thermocycler as follows: 25 uC for 10 s, 95 uC for 5 min, followed by 45-60 cycles of 95 uC for 10 s, 52 uC for 15 s, and 72 uC for 15 s. Melting curves were monitored by taking readings every 0.5 uC from 52-95 uC. The time-course qPCR analyses were performed with 53 uC and 54 uC annealing temperatures and 20 s extension time with no significant changes in results. For qPCR analyses of the full length CCR5 mRNA, assays were performed as described for the b-globin assays, but using oligonucleotides specific for b-micoglobulin and/or GAPDH and CCR5 27 . Reactions were amplified using the same conditions as for the b-globin constructs except all reactions used 20 s at 55 uC for extension. All assays were performed at least three times. Time course assays. mRNA decay time course assays using the tetracycline repressible rabbit b-globin reporter were performed as previously described with minor changes 26 . To monitor time-dependent decay of the CCR5 mRNA, HeLa Tzm1 cells were first transfected with either scrambled or hSMG1 siRNAs as described above. Forty eight hours after transfection, cells were treated with actinomycin D (10 mM). In all experiments, RNA isolations were performed immediately at each time point after transcriptional arrest using the RNAqueous kit (Ambion) rather than after freezing samples on dry ice. RNAi assays. Cells were transfected with RNA oligonucleotides specific to UPF1, UPF2, SMG1, argonaute 1, argonaute 2, DGCR8, exportin 5, or random oligonucleotides using the HiPerFect transfection reagent (Qiagen). Initial transfections were performed at 1, 5, 10 nM and 20 nM for optimization. Final transfections were performed at 5 nM. The MAPK cell-death positive control was used for optimization as well as qPCR quantification of the targeted mRNA. Most final transfections were performed at 5 nM. Transfections were performed into 30,000-40,000 cells (via haemocytometer) in 500 ml of DMEM1FBS using 100 ml of DMEM without FBS and 3 ml of HiPerFect reagent after incubating for 15-20 min at room temperature. Media was replaced with fresh DMEM1 FBS after 8-12 h. Assays were performed 36-72 h after siRNA transfection. When other plasmids were also transfected, they were performed separately 24-48 h after siRNA transfection using the FuGENE 6 (Roche) reagent. miRNA transfection. Cells were transfected with the following miRNA precursors: hsa-miR-141, hsa-miR-711, hsa-miR-1224-5p, and hsa-miR-1205 using either siPORT, lipofectamine, or HiPerfect reagent (Applied Biosystems/Ambion). When performing miRNA transfections with all miRNAs, 10-30 nM was used, depending on cell viability. When performing the miRNA titration, four 1:10 dilutions were used starting at 5 nM. Transfections were performed into 20,000-40,000 cells in 500 ml DMEM1FBS using 25 ml of DMEM without FBS and 1 ml of siPORT reagent after incubating for 20 min at room temperature per well. Media was replaced with fresh DMEM1FBS after 8-12 h. Dual luciferase plasmid transfections were either performed at the same time or 24 h later using the FuGENE 6 reagent. When HiPerfect was used, the conditions followed those used for siRNA transfections. Affinity purification of miRNA targets. Double stranded miR-1224-5p RNA containing a sense strand 59 biotin modification and mismatch were purchased from IDT. Pull-down experiments were performed as previously described 28 with the following modifications: Streptavidin agarose beads were pre-washed as described in 500 ml aliquots and stored for up to a week at 220 uC. Five washes with lysis buffer were performed rather than 3; after the final wash, 450 ml buffer was removed, samples were incubated for 5 min at 80u and quenched on ice for 2 min before isolating RNA. RNA isolations were performed using the RNAqueous kit (Ambion). qRT-PCR was used to observe mRNA isolation as previously described, using oligonucleotide primers specific for CCR5, Renilla luciferase, firefly luciferase and GAPDH for normalization. miR-1224 electrophoretic mobility shift assays. Transcripts harbouring the CCR5 21 PRF signal (247 nucleotides), mutants thereof, or HIV-1 21 PRF signal (315 nucleotides) were synthesized from DNA templates using T3 RNA polymerase using MEGAscript, and purified using MEGAClear kits (Ambion). HPLC purified miR-1224-5p (59GUGAGGACUCGGGAGGUGG39) RNA oligonucleotide was purchased from Integrated DNA Technologies, and was 59-[ 32 P]labelled by using the KinaseMax kit (Ambion). Small amounts of the CCR5 and HIV-1 derived mRNAs were also 59-[ 32 P]-labelled and used as markers. CCR5 or HIV RNA dilutions at 23 final concentration were mixed with equal volumes of 1.0 nM 59-[ 32 P]-labelled 1224-5p RNA. Samples were incubated at 37 uC for 30 min in HB buffer (50 mM Tris, pH 7.5, 0.1 mM EDTA, 10 mM NaCI, 10 mM MgCI 2 , 3% glycerol, 0.05% bromophenol blue) and immediately separated through 10% native polyacrylamide gels. For experiments with RNA refolding the RNA mix was incubated at 90 uC for 5 s, cooled quickly to 60 uC and then slowly to 37 uC (0.02 uC per s). The electrophoresis buffer was 34 mM Tris-66 mM HEPES pH 7.5, 0.1 mM EDTA, 10 mM MgCl 2 . To confirm that multiple bands of CCR5 in native gel were RNA conformers, labelled CCR5, miR-1224-5p RNA, and the mix of these RNAs were separated through an 8% denaturing gel and visualized using a phosphoimager. Single site binding isotherms were generated using GraphPad Prizm. Chemical modification assays. Dimethylsulphate, kethoxal and CMCT were used to probe the solvent accessibility of individual bases 29, 30 , while NMIA was used to probe ribose 29-OH groups 31 in [ 32 P]-labelled run-off transcripts. In a separate experiment, synthetic CCR5 (139 nM) and miR-1224 (1.1 mM) RNAs were annealed at 37 uc for 30 min in 33 mM HEPES, pH 8.0, 33 mM NaCl, 10 mM MgCl 2 . Structure probing with NMIA and reverse transcription reactions were subsequently performed as described 31 . Products were separated through 8% denaturing polyacrylamide gels, and visualized using a phosphorimager. Molecular modelling and molecular dynamics simulation protocol. Three dimensional modelling based on the secondary structure shown in Fig. 2 was performed using programs RNAComposer 32,33 and RNA2D3D 34 . The preliminary 10 alternative structures were obtained from the RNAComposer web server (http://euterpe.man.poznan.pl/Home). However, most of the models produced were knotted in 3D and the best two alternatives had the backbones of loop L (see Fig. 2 ) pass through a helix in S1. RNA2D3D was used to separate the tangled structures. The corrected models (a total of six alternatives) were minimized, equilibrated in solvent with some restraints meant to maintain the base pairs strained in the preliminary models, and subjected to short (25 ns) molecular dynamics runs without any restraints, after which the best model (average structure with the lowest minimized energy) was subjected to extended molecular dynamics simulation lasting a total of 80 ns. Given how structurally tight the initial models were and the extent of manual editing required, the extended simulation was necessary to stabilize the model and obtain an accurate average structure.

Molecular dynamics simulations were performed with Amber 12 with the ff10 Cornell force field for RNA. The Particle Mesh Ewald (PME) summation method was used to calculate the electrostatic interactions [35] [36] [37] . Following minimization, the RNA models (103 nucleotides long) were solvated in TIP3P waters with 102 neutralizing Na 1 ions and additional Na 1 /Clion pairs added to the solvent box to achieve a relative salt concentration of 0.1 M (from 51 to 57 depending on the system). The multi-step equilibration protocol started with solvent equilibration (minimization, heating and short dynamics stages) with the RNA being subject to slowly released motion restraints (holding). A periodic boundary condition was used in the simulation. The entire system was equilibrated at 300K using the Berendsen thermostat 38 , a cut-off of 9 Å was used with the non-bonded interactions and SHAKE was applied to all hydrogen bonds in the system. Pressure was maintained at 1.0 Pa using the Berendsen algorithm 39 . The last phase of the equilibration was performed for 2.0 ns with distance restraints placed on the hydrogen bonds of six base-pairs that were affected by the manual editing of the preliminary RNA models (base pairs: G8-U75, G12-U73, G13-C72, A14-U63, G16-U61, G17-C60). Following equilibration, the production simulation was performed with 2 fs time steps to obtain short trajectories of 25 ns. The total sizes of the systems subjected to molecular dynamics ranged from 71,462 to 79,346 atoms, including the 3,290 RNA atoms of the CCR5 model. The solvent boxes had a clearance distance of 10 Å (also named 'buffer' in Amber, that is, the minimum distance between the solute and the solvent box wall). Analyses of the molecular dynamics results excluded the equilibrations and were performed using the ptraj module of Amber. Multiple sequence alignment of CCR5 coding sequences. Sequences for CCR5 coding sequences were chosen for 39 primates and a naive BLAST search extended these candidates to 45 sequences with the CCR5 mRNA from Danio rerio chosen as an outgroup. Sequences were aligned with Clustal W 40 using default parameters (gap opening penalty: 15, extension penalty: 6, IUB matrix). The resulting alignment was manually edited using seaview 41 to trim especially long 39 UTR and long 59 UTR sequences (notably Danio rerio, sheep, goat, rat, and mouse). A guide tree was constructed using the default parameters of BioNJ 42 which was used as a starting tree for PhyML 43 . The GTR model was used with a nearest neighbour and SPR search strategy, 100 bootstrap replicates were performed. Branch lengths and bootstrap support are supplied above branches where appropriate. Analysis of ribosome profiling data. Sequence reads from Guo et al. 10 and Hsieh et al. 44 were downloaded from the Gene Expression Omnibus 45 . The ribosomal footprinting data was extracted and aligned against a library of human coding sequences from the Mammalian Gene Collection 46 using Bowtie2 47 . Papio_hamadryas   TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Gorilla_gorilla  TTTAAAAGCCAGGACGGTCACGTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Macaca_fascicularis  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Pan_troglodytes  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Trachypithecus_francoisTTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Trachypithecus_phayrei TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Pygathrix_roxellana  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Pygathrix_nemaeus  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Hylobates_leucogenys  TTTAAAAGCCAGGACGGTCACCTTTGGGGTAGTGACAAGTGTGATCACTTGGGT  Cercocebus_torquatus  TTTAAAAGCCAGGACAGTCACCTTTGGGTTGGTGACAAGTGTGATCACTTGGGT  Cercopithecus_aterrimusTTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Papio_hamadryas  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Hylobates_concolor  TTTAAAAGCCAGGACGGTCACCTTTGGGGTAGTGACAAGTGTGATCACTTGGGT  Pongo_pygmaeus  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Saguinus_sp.  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGTT  Callithrix_jacchus  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Alouatta_caraya  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGACCACTTGGGT  Aotus_trivirgatus  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Colobus_guereza  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Lemur_catta  TTTAAAAGCCAGGACGGTCACCTTTGGGGTAGTGACAAGTGGGGTGACCTGGAT  Varecia_variegata  TTTAAAAGCCAGGACGGTCACCTTTGGGGTAGTGACAAGTGGGGTGACCTGGAT  Mandrillus_sphinx  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Alouatta_seniculus  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGACCACTTGGGT  Symphalangus_syndactyluTTTAAAAGCCAGGACGGTCACCTTTGGGGTAGTGACAAGTGTGATCACTTGGGT  Miopithecus_talapoin  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Callicebus_moloch  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Pan_paniscus  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Lophocebus_aterrimus  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Hylobates_moloch  TTTAAAAGCCAGGACAGTCACCTTTGGGGTAGTGACAAGTGTGATCACTTGGGT  Theropithecus_gelada  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Hylobates_agilis  TTTAAAAGCCAGGACAGCCACCTTTGGGGTAGTGACAAGTGTGATCACTTGGGT  Oryctolagus_cuniculus CTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGCGGGGTCACCTGGGT  Leontopithecus_chrysomeTTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Sylvilagus_floridanus TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGCGGGGTCACCTGGGT  Oryctolagus_cuniculus CTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGCGGGGTCACCTGGGT  Saimiri_sciureus  TTTAAAAGCCAGGACGGTCACGTTTGGGCTGCTGACAAGTGTGATCACTTGGGT  Homo_sapiens  TTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Equus_asinus  TTTAAAAGCCAGGACGGTCACCTTTGGGCTGATGACAAGTGGGGTCACTTGGGC  Loxodonta_africana  TGTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGGGGTCACCTGGGT  Ovis_aries  TTTAAAAGCCAGGACAGTCACCTTTGGGGCAGTGACAAGTGGGGTCACGTGGGT  Sylvilagus_brasiliensisCTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACAAGCGGGGTCACCTGGGT  Lepus_townsendii  CTTAAAAGCCAGGACGGTCACCTTTGGGGTGGTGACGAGTGTGATCACCTGGGT  Capra_hircus  TTTAAAAGCCAGAACAGTCACCTTTGGGGCAGTGACAAGTGGGGTCACGTGGGT  Bos_taurus  TTTAAAAGCCAGAACAGTCACCTTTGGGGCGGCGACAAGTGTGGTCACCTGGGT  Canis_lupus  TTCAAAAGCCCGGACAGTCACCTTTGGGGTGGTGACAAGTGGGATCGCCTGGGT  Gallus_gallus  TTTAAAAGCTAGGACAGTTACCTACGGCATCCTCACCAGCATTGTCACGTGGGC  Equus_caballus  TTTAAAAGCCAGGACGGTCACCTTTGGGCTGATGACAAGTGGGGTCACTTGGGC  Felis_catus  TTTAAAGGCCAGGACGGTCACCTTTGGGGCGGTGACAAGCGCGGTCACCTGGGC  Papio_anubis  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Mus_musculus  TTTAAAAGTCAGAACGGTCAACTTTGGGGTGATAACAAGTGTAGTCACTTGGGC  Rattus_norvegicus  TATAAAAGCCAGAACAGTCAACTTTGGGGTAATAACAAGTGTAGTCACTTGGGT  Macaca_mulatta  TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Cercopithecus_aethiops TTTAAAAGCCAGGACAGTCACCTTTGGGGTGGTGACAAGTGTGATCACTTGGGT  Danio_rerio  AAATAAAAACCGCAGAAGCGTCTACGCTGCATCGTTATCTGTGGCCGTCTGGAT   GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCGTCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTATTTGCCTCTCTCCCAGGAATCATCTTTACCAGGTCTCAAAAAGA  GGTGGCTGTATTTGCCTCTCTCCCAGGAATCATCTTTACCAGGTCTCAAAAAGA  GGTGGCTGTATTTGCCTCTCTCCCGGGAATCATCTTTACCAGGTCTCAAAAAGA  GGTGGCTGTACTTGCCTCTCTCCCAGGAATCATCTTTACCAGGTCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  AGTGGCTGTGTTAGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  AGTGGCTGTGTTAGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTATTTGCCTCTCTCCCGGGAATCATCTTTACCAGGTCTCAAAAAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTATTTGCCTCTCTCCCGGGAATCATCTTTACCAGGTCTCAAAAAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCATCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGCGGCCATCTTGGTGTCTCTCCCAGACATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTATTTGCCTCTCTCCCAGGARTCATCTTTACCAGGTCCCAAAAAGA  GGCGGCCATCTTGGTGTCTCTCCCAGACATCATCTTTACCAGATCTCAAAAAGA  GGCGGCCATCTTGGTGTCTCTCCCAGACATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTATTCGCCTCTCTCCCAGGAATCATCTTCACCAGGTCTCAAAAAGA  GGTGGCTGTGTTTGCGTCTCTCCCAGGAATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTGTGTTTGTCTCTCTCCCAGGAATCATCTTTACCAAATCCCAAAAAGA  GGTGGCTGTGCTTGTCTCTCTCCCAGGGATCATCTTTACCAGATCCCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATTATCTTTACCAAATCCCAAAAAGA  GGCGGCCATATTTGCCTCGGTCCCGGGAATCATGTTTACCAGGTTCCAAAAAGA  GGCGGCCATCTTGGTGTCTCTCCCGGGCATCATCTTTACCAGATCTCAAAAAGA  GGTGGCTATGTTTGCCTCTCTCCCAGGAATTATCTTTACCAAATCCCAAAAGGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATTATCTTTACCAAATCCCAAAAAGA  GGTGGCCGTGCTCGCCTCTTTCCCCAGGATCATCTTCACCAGATCCCAAAAAGA  TGTTGCTCTTTTTGCCTCTGTTCCTGGGATAGTATTTCACAAAACTCAACAGGA  GGTGGCTGTGTTTGTCTCTCTCCCAGGAATCATCTTTACCAAATCCCAAAAAGA  GGCGGCCGTGGTCGCCTCTCTGCCGGGGTGCGTCTTTAACAAGTTGCAGTGGGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGAAATAATCTTTACCAGATCTCAGAAAGA  GGTGGCTGTGTTTGTCTCTCTCCCAGAAATAATCTTTATGAGATCTCAAAAAGA  GGTGGCTGTGTTTGCCTCTCTCCCAGGAATCATCTTTACCAGATCTCAGAGAGA  GGTGGCTGTGTTTGCCTCTCTCCCAAGAATCATCTTTACCAGATCTCAGAGAGA  TATCAGCCTTCTCGAAAGCCTCGAATATCTCATCCGTTTC- a, Computationally predicted and best fit 2-dimensional mRNA structures with chemical modification data. Stems 1, 2, and the loop are indicated as S1, S2 and L. The five different segments of stem 2 are labelled a-e. Nucleotide bases showing low levels of reactivity with DMS, CMCT, and kethoxal, and ribose sugars whose 29-OH groups with NMIA were similarly non-reactive are noted as open circles. Moderately reactive sugars and bases are denoted by grey filled circles. Strongly modified (strongly reactive) sugars and bases are represented as black filled circles. Circles proximal to the bases denote reactivity of the bases, while circles distal to the bases denote reactivity of the ribose sugars. From left to right: mRNA pseudoknot best fit to data; mRNA pseudoknot predicted by Pknots; mRNA pseudoknot predicted by NUPAK; tandem stem-loops predicted by mFold. Red bases represent chemical modification patterns inconsistent with computational predictions. b, c, Chemical modification experiments. Autoradiograms of reverse transcriptase primer extensions performed on RNA transcribed by T7 RNA Pol from template PCR amplified from CCR5 21 PRF signal containing plasmid. Bands correspond to strong readthrough control (RT) stops 1 nucleotide 59 of bases modified by chemical reagents. b, The CCR5 mRNA was either left unmodified (un), or modified with 3 increasing concentrations of dimethyl sulphate (DMS, reacts with A and C), 1-cyclohexyl-(2morpholinoethyl)carbodiimide metho-p-toluene sulfonate (CMCT, reacts with U), or 1,1-dihydroxy-3-ethoxy-2-butanone (kethoxal, reacts with G) respectively. c, Primer extension reactions were performed on unmodified samples and samples incubated with 30, 65, and 110 nM NMIA (N-methyl isatoic anhydride). These are labelled 1, 2 and 3, respectively, beneath each sample, and 'un' denotes untreated RNA. d, The all atom r.m.s.d. plot of the 80 ns-long molecular dynamics simulation states against the reference structure (the starting state of the simulation) at the end of a 2 ns-long equilibration. 

U C U C-G C-G G-C G-C G G U U C C U U C U C C 3' 3' C A A A A A G-C G-C G-U G-U A A C A-U A-U U- U-A C- C-G A A A A U U C C C C C U A U A U U U 5' UUUAAAAC-G C-G C-G C-G A-U A-U C-G C-G U-G U-G G-C G-C G-U G-U C- C-G A- A-U | G | G G- G-U G- G-U A-U U C-G C-G | C | C C-G C-G G-U G-UU-A U-A U-A U-A G-C G-C G-U G-U G-C G-C G-U G-U | A | A | G | G U-A U-A G-C G-C G-C G-C U-A U-A G-U G-U A-U A-U C | C | A | A | A-U A-U G-C G-C U | U | G-U G-U U-A U-A G-C G-C A-U A-U U-A U-A C A C-G C-G U-G U-G U-A U-A G-C G-C G-C G-C G-C G-C U U S2 L a b c d e

L C-G C-G C-G C-G A-U A-U C-G C-G U-G U-G G-C G-C G-U G-U C- C-G A- A-U G G G- G-U A-U U C U C U C-G C-G G-C G-CU-A U-A U-A U-A G-C G-C G-U G-U G-C G-C G-U G-U | A | A | G | G U-A U-A G-C G-C G-C G-C U-A U-A G-U G-U A-U A-U C | C | A | A | A-U A-U G-C G-C U | U | G-U G-U U-A U-A G-C G-C A-U A-U U-A U-A C A C-G C-G U-G U-G U-A U-A G-C G-C G-C G-C G-C G-C U U

Mechanisms and implications of programmed translational frameshifting

Yeast telomere maintenance is globally controlled by programmed ribosomal frameshifting and the nonsense-mediated mRNA decay pathway

Bypass of the pre-60S ribosomal quality control as a pathway to oncogenesis

Recoding Expansion of Decoding Rules Enriches Gene Expression

A conserved eEF2 coding variant in SCA26 leads to loss of translational fidelity and increased susceptibility to proteostatic insult

Exome sequencing identifies mutation in CNOT3 and ribosomal genes RPL5 and RPL10 in T-cell acute lymphoblastic leukemia

rRNA pseudouridylation defects affect ribosomal ligand binding and translational fidelity from yeast to human cells

HIV-1 entry into CD4 1 cells is mediated by the chemokine receptor CC-CKR-5

PRFdb: a database of computationally predicted eukaryotic programmed -1 ribosomal frameshift signals

Mammalian microRNAs predominantly act to decrease target mRNA levels

Single-molecule measurements of the CCR5 mRNA unfolding pathways

Solution structure and thermodynamic investigation of the HIV-1 frameshift inducing element

An equilibrium-dependent retroviral mRNA switch regulates translational recoding

Programmed 21 frameshifting efficiency correlates with RNA pseudoknot conformational plasticity, not resistance to mechanical unfolding

Effects of CCR5 and CD4 cell surface concentrations on infections by macrophagetropic isolates of human immunodeficiency virus type 1

Torsional restraint: a new twist on frameshifting pseudoknots

Regulation of mRNA stability in mammalian cells

Endogenous ribosomal frameshift signals operate as mRNA destabilizing elements through at least two molecular pathways in yeast

A parsimonious model for gene regulation by miRNAs

Human UPF1 participates in small RNA-induced mRNA downregulation

Characterization of the frameshift signal of Edr, a mammalian example of programmed 21 ribosomal frameshifting

A functional 21 ribosomal frameshift signal in the human paraneoplastic Ma3 gene

Mammalian gene PEG10 expresses two reading frames by high efficiency 21 frameshifting in embryonic-associated tissues

Assays of adenylate uridylate-rich element-mediated mRNA decay in cells

A dualluciferase reporter system for studying recoding signals

Ribosomal frameshifting efficiency and Gag/Gag-pol ratio are critical for yeast M1 double-stranded RNA virus propagation

Systematic analysis of bicistronic reporter assay data

Massive infection and loss of memory CD4 1 T cells in multiple tissues during acute SIV infection

Isolation of microRNA targets using biotinylated synthetic microRNAs

Structural features required for the binding of tRNATrp to avian myeloblastosis virus reverse transcriptase

Chemical probes for higher-order structure in RNA

Selective 2'-hydroxyl acylation analyzed by primer extension (SHAPE): quantitative RNA structure analysis at single nucleotide resolution

Automated 3D structure composition for large RNAs

RNA FRABASE 2.0: an advanced web-accessible database with the capacity to search the three-dimensional fragments within RNA structures

RNA2D3D: a program for generating, viewing, and comparing 3-dimensional models of RNA

How well does a restrained electrostatic potential (RESP) model perform in calculating conformational energies of organic and biological molecules?

A smooth particle mesh Ewald method

Molecular-dynamics with coupling to an external bath

Sequence evolution of the CCR5 chemokine receptor gene in primates

CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice

SeaView version 4: A multiplatform graphical user interface for sequence alignment and phylogenetic tree building

BIONJ: an improved version of the NJ algorithm based on a simple model of sequence data

New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0

The translational landscape of mTOR signalling steers cancer initiation and metastasis

NCBI GEO: archive for high-throughput functional genomic data

The completion of the Mammalian Gene Collection (MGC)

Fast gapped-read alignment with Bowtie 2

RESEARCH ARTICLE

The following options were used: The resulting SAM output was parsed with a simple Perl script (available at: https://github.com/abelew/prfdb/tree/master/ingolia) which extracted reads and sorted them by position and apparent reading frame. The resulting data structure was translated into JSON and plotted with the flot library. Genome ontology analysis. Homo sapiens accessions were collected from the PRFdb which lie more than one standard deviation from mean with respect to predicted MFE value and randomized Z-score. This population of 1,846 accessions was provided to the FuncAssociate 48 analysis tool.

FLuc (-1 frame) HeLa-TZM BL cells transcriptionally arrested with actinomycin D. Cells were transfected with SMG1 miRNA or scrambled miRNA control. d, Effects of various RNAs on CCR5 mRNA steady-state abundance. HeLa-TZM BL cells expressing CCR5 were transfected as follows: scrambled siRNA control (scr), human SMG1 siRNA (Smg), miR-1224 (1224), a mIR-1224 antagomir (anti), and combinations thereof. b-d, n 5 9 (three times on three independent biological replicates). Error bars denote standard error. *P , 0.05, **P , 0.01 (Student's two-tailed t-test).