Ribosomal Frameshifting Selectively Modulates the Biosynthesis,

,


34
Cystic fibrosis (CF) is a disease of protein homeostasis (proteostasis) that arises from the defective biosynthesis, 35 folding, and/ or function of a chloride channel known as the cystic fibrosis transmembrane conductance regulator 36 (CFTR). 1,2 The most common CF mutation (ΔF508) induces CFTR misfolding by decoupling the folding of its 37 subdomains during translation. 3,4 This cotranslational misassembly reaction enhances the retention and 38 degradation of the CFTR protein within the endoplasmic reticulum (ER) and ultimately reduces the trafficking of 39 the functional protein to the plasma membrane. This lapse in protein quality control (QC) coincides with the 40 remodeling of CFTR proteostasis network and changes in the dynamics of CFTR translation. 5-11 Nevertheless, 41 the precise chain of events involved in the crosstalk between the conformational state of nascent CFTR, the 42 activity of the translation machinery, and the interaction of the nascent chain with various components of the 43 cellular proteostasis network is not fully understood. 44 45 CFTR QC begins during its cotranslational folding at the ribosome-translocon complex. 12 Vectorial folding of its 46 five subdomains is orchestrated, in part, by positional variations in translation kinetics that are coordinated by 47 rare codons, 9 the relative abundance of certain translation factors, 5,7 and the presence of mRNA secondary 48 structure. 13 Together, these effectors help the ribosome tailor its activity to the kinetic constraints of 49 cotranslational CFTR folding. Indeed, the effects of CF mutations on CFTR folding and assembly are highly 50 sensitive to changes in translational dynamics. For instance, suppressing translation initiation reduces the 51 density of ribosomes on the ΔF508 transcript in a manner that partially rescues its stability, cellular trafficking, 52 and function. 7,11 The cotranslational misfolding of ΔF508 CFTR also appears to modulate translation in cis by 53 enhancing ribosome collisions and triggering "preemptive quality control" pathways. 11 Though the later finding 54 suggests the intrinsic activity of the ribosome is sensitive to conformational transitions in the nascent chain, it 55 remains unclear whether cotranslational misfolding events are capable of directly altering translation in real time. 56 57 We recently found that cotranslational folding mechanically modulates a translational recoding mechanism 58 known as -1 programmed ribosomal frameshifting (-1PRF) during viral polyprotein biosynthesis. 14,15 These 59 findings reveal that, under certain circumstances, the ribosome exhibits an enhanced propensity to slip into 60 alternative reading frames in response to conformational transitions in the nascent polypeptide chain. 16 In the 61 following, we evaluate whether similar feedback occurs during CFTR biosynthesis. We first identify a structured 62 segment within the region of the CFTR transcript that stimulates ribosomal frameshifting (RF). We then show 63 that the disruption of this motif selectively remodels the interactome of ΔF508 CFTR in a manner that partially 64 restores its expression and function. These modifications, which appear to have no effect on WT CFTR, also 65 enhance the pharmacological rescue of ΔF508 CFTR by the leading CF therapeutic Trikafta. Our results suggest 66 this RF site acts a QC-mediated translational "kill switch" that selectively promotes the premature termination of 67 translation in response to the cotranslational misfolding of the nascent chain. 68 69

71
Discovery of an Active - Given that ribosome collisions can occur during CFTR translation 21 93 and that they are known to modulate -1PRF, 22 we hypothesized that 94 RF could occur at one or more of these sites. To determine whether 95 efficient -1 RF occurs at any of these positions, we generated a 96 series of bicistronic reporters in which this CFTR segment is flanked 97 by an upstream Renilla luciferase (rLuc, expression control) and a 98 downstream -1 firefly luciferase (fLuc, -1 RF reporter, Fig. 2A). 23 We 99 generated three versions of this reporter that produce fLuc in 100 response to frameshifting at any of these slip-sites (SSABC), or 101 conditionally at B or C (SSBC), or at C only (SSC) by knocking out 102 various -1 frame stop codons downstream of these slip-sites ( Fig.  103 2A orange). Transient expression of the SSABC reporter in a human bronchial epithelial cell line (CFBE41o-) 104 generates a detectable fLuc frameshift signal 4-fold over the no-insert control baseline. rLuc intensities are 105 similar for all three reporters, suggesting all three reporters exhibit similar expression under these conditions 106 (Fig. 2B). fLuc intensities are also quite similar for all three reporters (Fig. 2C), which implies most frameshifting 107 occurs at SSC. Consistent with this interpretation, we find that the fLuc signal is ablated by mutations that disrupt 108 SSC (SSC mut , Fig. 2C). We note that we ruled out splicing artifacts 24 by RT-PCR and found that the frameshift 109 signal can also be partially suppressed by stop codons in either upstream of SSC in the 0-frame (5ʹTer) or 110 downstream of SSC in the -1 frame (3ʹTer, Figs. 2D & S2), which confirms that these signals arise from a genuine 111 RF. By normalizing the rLuc and fLuc signals, we estimate ribosomal frameshifting occurs with an efficiency of 112 2.9 ± 0.2% at SSC, which is comparable to levels achieved during translation of the canonical HIV gag-pol motif 113 (5.2 ± 0.4%, Fig. 2D). Similar frameshifting also occurs in HEK293T cells (Fig. S3). Together, these results 114 identify an active -1 RF motif within the human CFTR transcript. 115 116 117

119
Unlike viral -1PRF events that generate functionally distinct proteins, a -1 frameshift at SSC should only fuse five 120 non-native residues onto nascent CFTR prior to translational termination and truncation in the middle of NBD2 121 (Fig. 1A). Based on this consideration, we hypothesized that the RF motif serves to promote frameshifting and 122 the premature termination of translation in response to cotranslational CFTR misassembly. To determine 123 whether this motif is involved in assembly, we assessed how this RNA structure influences the interactomes of 124 WT and ΔF508 CFTR. Briefly, we designed a series of mutations to disrupt the slip-sites and weaken the stem-125 loops within NBD2 while preserving the amino acid sequence of CFTR in the 0-frame (RF mut , Fig. 1B). We then 126 incorporated these modifications into the full length CFTR (Fig. 1C) and used affinity purification-mass 127 spectrometry (AP-MS) based interactome profiling to determine the effects of these silent modifications on the 128 interactions formed by WT and ΔF508 CFTR. 5,25,26 In agreement with previous reports, 5,25 the ΔF508 mutation 129 significantly enhances the interaction of CFTR with numerous proteostasis factors ( Fig. 3A & Table S1). By 130 comparison, the silent mutations in NBD2 have minimal impact on the WT CFTR interactome (WT-RF mut , Fig  131  3A). In contrast, these same mutations significantly alter the propensity of ΔF508 CFTR to interact with several 132 translation factors and quality control proteins (ΔF508-RF mut , Fig. 2A). Consistent with our hypothesis, the 133 disparate effects of this transcript modification on WT and ΔF508 suggests the native RNA structure in NBD2 134 selectively alters CFTR translation and assembly in response to misfolding. We note that the ΔF508 mutation is 135 ~2.4 kb upstream of the ribosomal frameshift site in NBD2 (Fig. 1C), which suggests these differences likely 136 arise from the effects of the ΔF508 mutation on the nascent CFTR protein rather than perturbation of the mRNA 137 structure. 138 139 There are several key differences in the interaction profiles of ΔF508 and ΔF508-RF mut CFTR (Table S1). We protein in the 0-frame, we suspect these common interactions may arise from the post-translational effects of 143 the ΔF508 mutation on the CFTR protein. Nevertheless, ΔF508-RF mut exhibits significantly attenuated 144 associations with eight ribosomal subunits (Fig. 3B, Table S1), which implies the frameshift motif normally 145 creates ribosome pile-ups on the ΔF508 transcript. These silent mutations also appear to modulate QC, as 146 ΔF508 RF mut exhibits both weaker interactions with certain chaperones and co-chaperones (e.g. CANX, HspA6, 147 BAG2, and DNAJB6) and stronger interactions with others (e.g. HspB1, HspA4, and DNAJA3 & Fig. S5). These 148 QC modifications also coincide with a decrease in the interaction of ΔF508-RF mut with the UBE3C ubiquitin ligase, 149 the ER-PHAGY protein RTN3, and various components of the 19S proteasome complex (Table S1), which 150 suggests the RF motif ultimately modifies CFTR degradation. Finally, we note these silent mutations enhance 151 the interaction of ΔF508 CFTR with two components of the eIF3 complex (i.e. eIF3c and eIF3e, Cluster 3, Fig.  152 S4, Table S1), which is known to modify the proteostatic effects of the ΔF508 mutation. 7 Overall, these results 153 suggest the ribosomal frameshift site in NBD2 acts as a junction where cotranslational misfolding triggers 154 changes in CFTR translation, QC, and degradation. 155 156

Impact of Ribosomal Frameshifting on CFTR Expression
To determine how these changes in cotranslational assembly impact CFTR proteostasis, we compared the 159 effects of the RF mut modification on ΔF508 and WT CFTR expression. Quantitative western blot analyses reveal 160 Figure 3. Impact of ribosomal frameshifting on the CFTR interactome. Affinity purification-mass spectrometry was employed to compare the effects of the silent RFmut mutations on the interactome of WT and ΔF508 CFTR. A) A heatmap indicates the log2 (foldchange) in interactions of CFTR variants relative to WT. Increases in the association of interactors with CFTR variants relative to WT are indicated in yellow while blue indicates a decrease in relative abundance of interactors. Black indicates no change relative to WT. Scale shown in log2 fold change over WT abundance. Proteins are organized into 3 clusters according to a hierarchical clustering analysis that groups interactors based on changes in abundance. B) A network map depicts the relationships between interactors in cluster 1, which includes interactors the exhibit the largest changes in the context of ΔF508 RFmut relative to ΔF508. Lines indicate known protein-protein interactions in the String database (human). Line widths indicate the strength of data support. The colors of nodes reflect the identity of sub-clusters from K-means clustering of the interactors within hierarchical cluster 1. The labels for each color summarize the most common classes of proteins within each sub-cluster. Interactors in gray were missing from the String database and were manually added in as isolated nodes. that modifications of the RF motif have minimal impact on the maturation of WT or ΔF508 CFTR in both 161 CFBE41o-and HEK293T cells as judged by the relative abundance of the mature (C band) and immature (B 162 band) glycoforms (Figs. 4 A-C). This observation likely reflects the fact that the RF mut constructs generate the 163 same full-length protein as those containing the unmodified RNA structure. Nevertheless, integrated intensities 164 suggest the ΔF508 RF mut transcript generates a modest (~20%), yet statistically insignificant increase in the total 165 CFTR expression relative to the ΔF508 CFTR transcript (Figs. 4 A,B, and D). Consistent with this observation, 166 flow cytometry measurements suggest this modest uptick in expression arises from a 19 ± 11% increase in the 167 intracellular CFTR generated by ΔF508-RF mut relative to the unmodified ΔF508 (p = 0.007, Fig. 4 E). We note 168 that there is no corresponding increase in surface CFTR levels generated by ΔF508-RF mut (Fig. 4F), which may 169 again reflect the fact that this transcript still generates the same unstable ΔF508 CFTR protein. Notably, the 170 intracellular immunostaining intensity of ΔF508-RF mut is indistinguishable from that of WT (Fig. 4E), which 171 suggests this motif reduces the accumulation of ΔF508 CFTR in the ER. We note that all trends were 172 quantitatively consistent in CFBE41o-and HEK293T cells (Figs. 4 & S6). Though these effects are subtle, 173 together these results reveal that the RF motif selectively reduces the expression of ΔF508 CFTR potentially by 174 promoting the premature termination of translation in response to misfolding. 175 176

178
To determine how the RF motif impacts CFTR function, we compared the effects of the RF mut modification on 179 the ΔF508 and WT CFTR-mediated quenching of a cellular halide-sensitive yellow fluorescent protein (hYFP)-180 a common CFTR activity assay. 27 Briefly, we generated a series of recombinant stable HEK293T cell lines that 181 inducibly express a single CFTR variant off a bicistronic transcript containing a downstream internal ribosomal 182 entry site (IRES) that produces the hYFP sensor and a fluorescent mKate reference protein (Fig. 5A). We then 183 used flow cytometry to track the loss of hYFP intensity relative to that of mKate at the single-cell level following 184 activation of CFTR (Fig. 5B). Half-life values extracted from a global fit of hYFP quenching reveals that the RF mut 185 modification has no impact on the function of WT CFTR (Fig. 5C). This observation is consistent with the limited 186 impact of this modification on the interactome and expression of WT CFTR (Figs. 3A & 4). As expected, ΔF508 187 CFTR exhibits a sizable increase in quenching half-life relative to WT, which reflects its impaired channel 188 conductance function (Fig. 5C). In contrast to WT, the RF mut mutations decrease the half-life of ΔF508 CFTR-189 mediated hYFP quenching by 46 ± 6% (Fig. 4 B & C). Given that the disruption of the RF motif does not appear 190 to increase ΔF508 CFTR levels at the plasma membrane (Fig. 4C), the enhanced conductance of ΔF508-RF mut 191 implies that the changes in its interactome enhance the functionality of the mature channel protein. Together,192 these cumulative findings reveal that the RF motif selectively modulates the assembly of ΔF508 CFTR in a 193 manner that ultimately attenuates its expression and function. 194 A GAPDH loading control is included for reference. B) A representative western blot depicting the relative abundance of the mature (band C) and immature (band B) CFTR glycoforms of each indicated variant in HEK293T cells is shown. A GAPDH loading control is included for reference. C) A bar graph depicts the average C: B band intensity ratio relative to WT in CFBE41o-cells expressing the indicated variants as determined by western blot (n = 3). Error bars reflect the standard deviation. D) A bar graph depicts the average total CFTR intensity (C+B) relative to WT in CFBE41o-cells expressing the indicated variants as determined by western blot (n = 3). Error bars reflect the standard deviation. E) A bar graph depicts the average intracellular CFTR immunostaining intensity relative to WT among HEK293T cells expressing the indicated variants as determined by flow cytometry (n = 6). Error bars reflect the standard deviation. F) A bar graph depicts the average surface CFTR immunostaining intensity relative to WT among HEK293T cells expressing the indicated variants as determined by flow cytometry (n = 6). Error bars reflect the standard deviation.

197
The functional expression of ΔF508 and various other CF variants can be partially restored by Trikafta, an FDA-198 approved cocktail of two correctors that stabilize the CFTR protein (VX-661 + VX-445) and a potentiator that 199 activates it (VX-770). 28-30 To determine whether ribosomal frameshifting potentially influences the effects of these 200 compounds, we assessed whether the RF motif impacts the pharmacological rescue of ΔF508 CFTR. As 201 expected, flow cytometry-based immunostaining measurements reveal that both WT and ΔF508 exhibit 202 enhanced accumulation at the plasma membrane and within the secretory pathway upon stabilization by VX-203 661 + VX-445 ( Fig. 6 A & B). A western blot analysis also confirms that these correctors enhance the maturation 204 of both variants (Fig. 6C, Fig S7). However, similar gains in expression and maturation were observed for WT-205 RF mut and ΔF508-RF mut (Fig. 6 A-C, Fig. S7), which demonstrates that the RF motif has minimal impact on ΔF508 206 expression upon stabilization by corrector molecules. Next, we assessed functional rescue of CFTR conductance 207 using our hYFP quenching assay. For this purpose, the potentiator VX-770 was added in combination with the 208 correctors VX-661 and VX-445 to evaluate the functional rescue by Trikafta. Despite the comparable expression 209 of ΔF508 and ΔF508-RF mut in the presence of correctors, HEK293T cells expressing ΔF508-RF mut that were 210 treated with VX-661 + VX-445 + VX-770 exhibit a is 41 ± 5% decrease in hYFP quenching half-life relative to 211 cells expressing ΔF508 CFTR under the same conditions (Fig. 6D). Together, these results suggest that, while 212 stabilization by correctors minimizes the impact of the RF motif on ΔF508 CFTR expression, disrupting the RNA 213 structure that promotes frameshifting ultimately enhances its functional rescue. 214 215

217
The ΔF508 mutation promotes cotranslational CFTR misfolding and degradation in a manner that reduces its 218 functional expression. 1 Nevertheless, its proteostatic effects also stem from aberrant translational dynamics.

219
ΔF508 perturbs over 20 translation factors and RNA processing proteins involved in CFTR biosynthesis. 5 220 However, it remains unclear how the conformational effects of ΔF508 result in direct modifications of translation 221 and vice versa. In light of the recent discovery that conformational transitions in the nascent chain can stimulate 222 -1PRF, 14,15,31 we searched the CFTR transcript for structural features that might promote ribosomal frameshifting. 223 Our findings identified a slippery sequence (SSC) and stem-loop (SL2) that stimulate ribosomal frameshifting 224 and the premature translational termination (Fig. 2). To determine how this RF motif impacts CFTR expression 225 and function, we designed a series of silent mutations that maintain the native amino acid sequence in the 0-226 frame while disrupting these putative RNA structures (RF mut , Fig. 1). While this modification has minimal impact 227 on WT CFTR, disrupting the RF motif significantly remodels ΔF508 CFTR interactome (Fig. 3). These changes 228 in ΔF508 assembly enhance the accumulation of ΔF508 CFTR within the secretory pathway and increase its 229 functional conductance at the plasma membrane (Figs. 4 & 5). Finally, we show that disrupting the RF motif also 230 Figure 5. Impact of ribosomal frameshifting on CFTR function. The functional conductance of stably expressed CFTR variants was compared in HEK293T cells by measuring the time-dependent quenching of a halide-sensitive yellow fluorescent protein (hYFP). A) A cartoon depicts a diagram of the stably expressed genetic cassette (top) and a schematic for the CFTR activity assay (bottom). Upon activation, iodide ions flow through the CFTR channel and quench the hYFP fluorophore, which is monitored by the decrease of hYFP intensity relative to that of the fluorescent mKate standard. B) HEK293T cells stably expressing WT (white), WT RFmut (gray), ΔF508 (red), and ΔF508 RFmut (blue) were stimulated with 25 µM forskolin (0.06% DMSO Vehicle) to activate CFTR prior to measurement of the change in cellular hYFP: mKate intensity ratio measurements over time by flow cytometry. Single cell intensity ratios are plotted against the time and the global fits of the decay are shown for reference. C) A bar graph depicts the globally fit half-life for hYFP quenching enhances the pharmacological rescue of ΔF508 CFTR function (Fig. 6). Given that the disruption of this motif 231 has no impact on the assembly, expression, function, or pharmacological profile of WT CFTR (Figs. 2-5), we 232 conclude that this RNA structure stimulates RF and the premature termination of translation in response to 233 cotranslational CFTR misfolding. Together, our findings suggest the RF motif acts as a kill switch that down-234 regulates the synthesis of defective proteins in real time. Future investigations are needed to determine how this 235 translational regulation influences the biosynthesis and pharmacological response of other known CF variants.

237
Interactome profiles suggest this RF motif may represent an assembly junction where the conformational state 238 of the nascent chain modulates the interplay between translation and QC. We identify several ΔF508-specific 239 chaperone interactions that are lost and others that are gained upon disruption of the RF motif (Table S1, Fig.  240 3). Moreover, weakening this RNA structure the context of the ΔF508 transcript significantly reduces the co-241 immunoprecipitation of several ribosomal subunits (Table S1, Fig. 3), which implies ribosomes typically stall 242 within this region during ΔF508 translation. Based on these observations, we propose that the misassembly of 243 nascent ΔF508 CFTR stimulates ribosomal frameshifting and directs the ribosome to a stop codon in the -1-244 frame (Fig. 7). Without the prompt intervention of ribosome quality control, 32 this premature termination causes 245 trailing ribosomes to collide (Fig. 7). Such crosstalk may explain the link between cotranslational misfolding and 246 ribosome collisions that occurs during pre-emptive QC. 11 Ribosomal pausing 11 and/or heightened frameshifting 247 during ΔF508 CFTR translation could potentially explain how this mutation enhances associations with UPF1 248 and other nonsense-mediated decay (NMD) proteins 5 that typically promote the degradation of nonsense 249 transcripts. 33,34 Knocking down UPF1 and other associated RNA binding proteins partially restores ΔF508 250 expression, 5 which until now has been a perplexing observation given that ΔF508 preserves the native reading 251 frame. Nevertheless, the stimulation of frameshifting by ΔF508 would promote premature termination via stop 252 codons in the -1 frame. Such a link between ribosomal frameshifting and NMD has been previously described, 35 253 and certain ribosome-associated molecular chaperones have been found to influence -1PRF. 36 Nevertheless, to 254 our knowledge, previous investigations have not established a direct role of ribosomal frameshifting in protein 255 QC. We note that these interpretations remain speculative given that we are unable to compare ribosomal 256 frameshifting efficiencies during CFTR biosynthesis and that our interactome measurements do not reveal a 257 clear link to NMD. Additional investigations are needed to gain mechanistic insights into the interplay between 258 cotranslational misfolding, ribosomal frameshifting, and RNA surveillance pathways.   Baltimore, MD) and harvested with 1X 0.25% Trypsin-EDTA (Gibco, Grand Island, NY).

297
CFBE41o-cells were grown in minimal essential medium (Gibco, Grand Island, NY) containing 10% fetal bovine 298 serum (Corning, Corning, NY) and a penicillin (100U/ml)/streptomycin (100µg/ml) (Gibco, Grand Island, NY) 299 antibiotic supplement (Gibco, Grand Island, NY) in a humidified incubator containing 5% CO2 at 37°C on culture 300 plates coated with PureCol purified bovine collagen (Advanced BioMatrix, Carlsbad, CA). CFBE41o-cells were 301 transfected using Lipofectamine 3000 (ThermoFisher Scientific, cat #L3000015). Two days post-transfection, 302 cells were washed with 1X hepes buffered saline (Gibco, Grand Island, NY) and harvested with 1X 0.25% 303 Trypsin-EDTA (Gibco, Grand Island, NY). 304 305 CFTR function measurements were carried out in recombinant stable cells were made from genetically modified 306 HEK293T cells grown in 10 cm dishes in complete media as was previously described. 38,39 Briefly these 307 HEK293T cells contain a Tet-Bxb1-BFP "landing pad" where recombination results in the integration of a plasmid 308 into the gDNA of the cell. 38 These cells were co-transfected with plasmid and a Bxb1 recombinase expression 309 vector using Fugene 6 (Promega, Madison, WI). Doxycycline (2µg/mL) was added one day after transfection 310 and the cells were grown for 3 days at 33°C. The cells were then incubated at 37°C for 24 hours prior to 311 passaging for further experiments. 312 313 Figure 7. Proposed role of ribosomal frameshifting in the selective downregulation of ΔF508 CFTR biosynthesis. A cartoon depicts the proposed role of ribosomal frameshifting in CFTR proteostasis. Ribosomes pause at the ribosomal frameshift site at the point in which MSD1 and MSD2 undergo domain swapping. Synthesis proceeds with modest frameshifting when correct assembly occurs. However, failure to achieve the correct conformation and/ or chaperone interactions stimulates ribosomal frameshifting and the premature termination of termination of translation. Premature termination causes prolonged stalling that leads to collisions with trailing ribosomes and a failure to clear RNA surveillance proteins that trigger mRNA decay.