The Ribosome-Sec61 Translocon Complex Forms a Cytosolically Restricted Environment for Early Polytopic Membrane Protein Folding*

Background: Mechanisms that guide membrane protein folding in the endoplasmic reticulum membrane remain unresolved. Results: During aquaporin-4 synthesis, extracellular peptides loops enter the endoplasmic reticulum lumen sequentially, whereas delivery of cytosolic loops is actively delayed. Conclusion: The assembled ribosome translocon complex (RTC) shields large regions of the protein from the cytosol throughout synthesis. Significance: Early membrane protein folding occurs in a proteinaceous environment provided by the RTC. Transmembrane topology of polytopic membrane proteins (PMPs) is established in the endoplasmic reticulum (ER) by the ribosome Sec61-translocon complex (RTC) through iterative cycles of translocation initiation and termination. It remains unknown, however, whether tertiary folding of transmembrane domains begins after the nascent polypeptide integrates into the lipid bilayer or within a proteinaceous environment proximal to translocon components. To address this question, we used cysteine scanning mutagenesis to monitor aqueous accessibility of stalled translation intermediates to determine when, during biogenesis, hydrophilic peptide loops of the aquaporin-4 (AQP4) water channel are delivered to cytosolic and lumenal compartments. Results showed that following ribosome docking on the ER membrane, the nascent polypeptide was shielded from the cytosol as it emerged from the ribosome exit tunnel. Extracellular loops followed a well defined path through the ribosome, the ribosome translocon junction, the Sec61-translocon pore, and into the ER lumen coincident with chain elongation. In contrast, intracellular loops (ICLs) and C-terminalresidues exited the ribosome into a cytosolically shielded environment and remained inaccessible to both cytosolic and lumenal compartments until translation was terminated. Shielding of ICL1 and ICL2, but not the C terminus, became resistant to maneuvers that disrupt electrostatic ribosome interactions. Thus, the early folding landscape of polytopic proteins is shaped by a spatially restricted environment localized within the assembled ribosome translocon complex.

ing aquaporin-4 (AQP4), utilize such a strategy and encode TMs with alternating SA and ST activities that establish topology of each successive TM as the nascent polypeptide emerges from the ribosome (34 -36). Other proteins, however, lack a well ordered SA-ST-SA arrangement and, as a result, transiently acquire a non-native topology during their synthesis that is modified during subsequent stages of tertiary folding and helical packing (37)(38)(39)(40)(41)(42). This latter process may involve subtle readjustment of TM boundaries (40,43) or inversion of TMs or groups of TMs as native helical contacts are established (1, 38, 39, 41, 42, 44 -46). It remains unknown, however, where these early folding events take place and how they might be influenced by the physical environment of the nascent polypeptide.
Two possible models for TM integration provide differing views of cellular mechanisms of PMP folding. One model predicts that TMs rapidly partition into the lipid bilayer through energetically favorable hydrophobic interactions (8,11,32,(47)(48)(49). Such a model supports a two-step folding process in which TMs first equilibrate with membrane lipids as they emerge from the lateral translocon gate, prior to forming native intramolecular TM-TM contacts (50,51). Alternatively, biochemical, cross-linking, and FRET experiments have shown that TMs can be transiently retained, either within the translocon or adjacent to translocon components, before release (2,28,33,36,(52)(53)(54)(55)(56). Although the precise mechanism of TM retention remains unknown, evidence suggests that charged residues and/or reduced TM hydrophobicity may delay release, thereby favoring protein-protein over protein-lipid interactions during early stages of folding (54,55). These findings raise key questions regarding the role of the RTC in PMP biogenesis. Does the RTC act primarily as a passive channel for partitioning of TMs into a lipid environment (11), or does TM retention provide a specialized environment to influence early stages of helical packing and/or tertiary folding (1,38,54,55)?
To address these issues, we monitored the cytosolic and lumenal accessibility of multiple peptide loops during cotranslational biogenesis of the six-spanning AQP4 water channel. Our results show that the RTC shields the nascent AQP4 peptide from the cytosol upon exiting the ribosome. Extracellular loops (ECLs) follow a direct pathway and sequentially enter the ER lumen coincident with nascent chain elongation. In contrast, intracellular loops (ICLs) accumulate beneath the ribosome and gain access to cytosol only after the nascent chain is released from the ribosome. These data indicate that the RTC provides a cytosolically restricted environment that simultaneously accommodates multiple hydrophilic peptide regions during early stages of biogenesis and folding.

Experimental Procedures
Plasmid Construction-Cysteine-less rat AQP4 cDNA (AQP4 ⌬6Cys) was constructed from plasmid pSP64.MIWC (34) by replacing the six native cysteine (Cys) residues with alanine (Ala) using PCR with overlap extension (57). Using the same technique, single Cys codons were then reintroduced (Table 1) at amino acid positions indicated in Fig. 1, A and B. The sequence of all PCR amplified and cloned regions of DNA were verified by DNA sequencing.
In Vitro Transcription Translation-Truncated cDNA templates were generated by PCR amplification of single Cys plasmid constructs using a sense oligonucleotide complimentary to bp 2741 of the pSP64 vector (5Ј-CGTAGAGGATCTGGCTA-GCG-3Ј) and antisense oligonucleotides complementary to downstream regions of the AQP4 coding sequence. For each truncation, the last translated codon was converted to valine to minimize spontaneous hydrolysis of the peptidyl-tRNA bond. Full-length AQP4 cDNA was generated using an antisense oligonucleotide complimentary to bp 51 of the pSP64 vector (5Ј-CACAGGAAACAGCTATGACC-3Ј).
PCR products were transcribed in vitro at 40°C for 1 h with SP6 RNA polymerase in the presence of the RNA cap analog diguanosine triphosphate (500 M), as described previously (58). Capped mRNA was then added directly to a rabbit reticulocyte lysate translation reaction supplemented with [ 35 S]methionine and canine pancreatic microsomes (A 280 ϭ 5.0, unless otherwise stated) at 24°C for 1 h, as described elsewhere (58). Where indicated nascent chains were released from the ribosome via incubation with 50 ng/l of RNase for 10 min at room temperature (59).
Mass Tagging with PEG-Mal 5,000 -ER microsomes containing newly synthesized AQP4 polypeptides were isolated from translation reactions following dilution in an equal volume of buffer A (50 mM HEPES-KOH (pH 7.5), 100 mM KCl, 5 mM MgCl 2 ), by pelleting through buffer A containing 0.7 M sucrose supplemented with 1 mM DTT at 175,000 ϫ g at 4°C for 10 min. Pelleted microsomes were then suspended in an equal starting volume of buffer A containing 0.1 M sucrose and 50 M DTT. Ribosome nascent chain complexes (RNCs), prepared in the absence of ER microsomes, were pelleted by layering translation reactions over buffer A containing 0.7 M sucrose supplemented with 1 mM DTT and centrifuging at 386,000 ϫ g at 4°C for 1 h. Pelleted RNCs were suspended in equal starting volume of buffer A in 0.1 M sucrose and 50 M DTT.
RNCs or microsomes prepared as described above were incubated on ice for 1 h with 2 mM methoxypolyethylene glycol maleimide 5,000 Da (PEG-Mal 5,000) (Sigma) after addition of 0.5 M NaCl, 20 M melittin (Sigma), or 2% (w/v) digitonin (Calbiochem, Darmstadt, Germany), where indicated. Aliquots were then quenched with 0.2 M DTT at room temperature for 15 min, solubilized in 4 volumes of Laemmli sample buffer (8% (w/v) SDS and 0.1 M Tris-HCl (pH 8.9)) with 1 M DTT, and run on 10 -18% continuous gradient SDS-PAGE gels. Bands were quantified by phosphorimaging using a Personal FX Phosphor-Imager and QuantityOne software (Bio-Rad). Pegylation effi- where P ϭ band intensity of pegylated protein and U ϭ band intensity of unpegylated protein. To quantify accessibility of nascent ribosome-attached polypeptides, pegylation efficiency was determined from bands corresponding to the intact peptidyl-tRNA species. Pegylation of full-length AQP4 and RNasereleased polypeptides was determined by quantification of bands corresponding to the chain-terminated peptides. AQP4 Functional Studies-Xenopus laevis oocytes were surgically harvested under tricaine anesthesia and digested for 3 h at room temperature with 0.2 units/ml of Blendzyme III (Roche Applied Sciences) in Ca 2ϩ -free MBSH (88 mM NaCl, 1 mM KCl, 24 mM NaHCO 3 , 0.82 mM MgSO 4 , 10 mM HEPES-KOH (pH 7.4)) supplemented with 0.41 mM CaCl 2 . Stage VI oocytes were then collected and injected with 50 nl of RNA transcript and incubated at 17°C in MBSH plus 100 units/ml of penicillin, 100 g/ml of streptomycin sulfate, and 50 g/ml of gentamicin. Osmotic water permeability was measured 48 h after injection by transferring individual oocytes into 24-well plates containing a 10-fold dilution of MBSH in distilled water, and collecting digitized images at 1-s intervals by phase-contrast microscopy using IP-Labs 3.5 Morphometric software (BD Bioimaging). Total oocyte volume at each time point was calculated based on the two-dimensional projected area (average diameter), assuming a spherical shape, as reported previously by us and others (39,60). Osmotic water permeability was calculated from the initial swelling rate (first 30 s) using the following equation, where d(V/V 0 )/dt is the initial rate of swelling, S/V 0 Ϸ 50 cm Ϫ1 , V w ϭ 18 cm 3 mol Ϫ1 , and Osm out Ϫ Osm in ϭ 180 mosmol. Four to seven oocytes were analyzed for each construct, and each experiment was independently repeated 3 to 6 times. Average P f values obtained in a given experiment were normalized to wildtype AQP4 evaluated in the same experiment. Results are presented as mean Ϯ S.E. of 3-6 independent experiments.

Results
The AQP4 water channel is an attractive model PMP with a highly conserved six-membrane spanning topology and well defined tertiary structure (Fig. 1, A and B). Previously we showed that TMs 1, 3, and 5 encode SA activity that initiates polypeptide translocation into the ER lumen, whereas TMs 2, 4, and 6 encode ST activity that cotranslationally terminates translocation and redirects the growing polypeptide into the cytosol (34 -36, 39). These results predict that topology of each TM segment is acquired during synthesis as the RTC dynamically coordinates movement of sequential peptide regions into their appropriate compartment.
Defining Peptide Accessibility by Cysteine Modification-In the current study, native AQP4 Cys residues were substituted with Ala and single Cys residues were systematically introduced at the N and C termini and within each ICL and ECL (Fig. 1, A  and B). Importantly, AQP4 ⌬6Cys retained water channel activity (ϳ60% of wild type), indicating that folding in the membrane is not grossly perturbed (Fig. 1C). To establish baseline accessibility of peptide loops, full-length Cys-substituted constructs were translated in the presence of ER-derived canine rough microsomes (CRMs) and subjected to covalent modification by the aqueous, membrane-impermeable reagent, PEG-Mal 5,000 as described under "Experimental Procedures" (17,25,61). Although the extent of pegylation varied for different sites, residues predicted to reside in the cytosol (Cys-9, Cys-68, Cys-161, and Cys-240) were modified in intact microsomes (0.3-0.7 fraction pegylated) and exhibited minimal or no further increase in pegylation upon microsome permeabilization (Fig. 1, E and F). Lumenal residues (Cys-40s, Cys-121, and Cys-195) exhibited low basal reactivity (0.06 -0.12; Fig. 1D), and a 4 -7-fold increase in pegylation following membrane permeabilization (Fig. 1, E and F). Thus, PEG-Mal modification accurately reports on the topology of in vitro synthesized AQP4 in the ER membrane.
AQP4 Targeting to the ER and RTC Formation-Like most PMPs, AQP4 encodes an N-terminal type II SA that targets the RNC to the ER membrane (25,34,35). To determine the minimum chain length required for targeting, AQP4 mRNA was truncated and translated in the presence of CRMs ( Fig. 2A). Membrane-bound nascent chains were recovered by ultracentrifugation, analyzed by SDS-PAGE (Fig. 2B), and plotted as a function of chain length (Fig. 2C). Half-maximal targeting was observed when the nascent chain reached a length of ϳ65 aa and targeting was essentially complete at a length of 80 aa (Fig.  2C). Membrane binding also became resistant to high salt conditions at a truncation length of ϳ90 amino acids, presumably due to hydrophobic interactions between Sec61 and the SA sequence (25).
RTC Assembly Controls Nascent Chain Access to Cytosol-To understand how the RTC facilitates localization of PMP domains, we determined the accessibility of cytosolic and lumenal loops in truncated AQP4 polypeptides that were cotranslationally captured within intact RTCs. Programmed integration intermediates were synthesized from truncated RNA transcripts to generate uniform populations of nascent polypeptides that remain attached to the ribosome via a covalent peptidyl-tRNA bond (Fig. 3A). Thus, each truncation site provides a static snapshot of the equilibrated nascent chain conformation in the context of native biosynthetic machinery (25,36,(62)(63)(64)(65). Nascent chain accessibility was then determined at defined stages of synthesis, by PEG-Mal modification of Cys residues that were engineered into cytosolic and lumenal peptide loops (25,43,61,66).
When translation was performed in the absence of CRMs (Fig. 3A), all peptide regions examined became accessible to PEG-Mal as the tether from the peptidyl-transferase center (PTC) reached a length of 30 -50 residues, consistent with the predicted span of the ribosome exit tunnel (Fig. 3B). Following ribosome binding to the ER membrane, however, fractional pegylation was reduced 5-10-fold for all sites examined (Fig. 3,  C-I). Surprisingly, the entire polypeptide, except the N terminus, remained inaccessible to cytosolic PEG-Mal throughout translation even when the ribosome had reached the last residue prior to the termination codon, which is ϳ70 residues from the C terminus of TM6 (Fig. 3, C-I). Thus, ribosome binding to the translocon restricts AQP4 cytosolic exposure throughout synthesis of the entire polypeptide, excepting the N terminus.
Lumenal Loops Follow a Prescribed Path through the RTC-We next determined when lumenal loops move into the ER lumen. Because high salt (0.5 M NaCl) disrupts electrostatic interactions between the ribosome and Sec61 (25,67,68), an increase in pegylation following NaCl treatment indicates that the probe site has exited the ribosome tunnel and resides near the ribosome translocon junction (Fig. 4A) (9,17,25). In contrast, an increase in pegylation following membrane permeabilization indicates that the probe resides in the ER lumen (Fig.  4A). To follow the movement of the nascent chain, pegylation profiles were determined for Cys-40, Cys-121, and Cys-195, and plotted as a function of chain length (Fig. 4, B-D). Each peptide loop became sensitive to high salt as the tether length between the Cys probe and the PTC was increased from 30 to 60 aa (Fig. 4, B-D). Upon further elongation, ECLs underwent a coordinated transition characterized by loss of salt sensitivity and increased sensitivity to melittin, which plateaued at levels  NOVEMBER 27, 2015 • VOLUME 290 • NUMBER 48 similar to those observed for full-length AQP4 (Figs. 4, B-D). As expected, melittin did not affect pegylation within the saltsensitive junctional space. Thus, all three ECLs followed an orderly progression into the ER lumen. Movement of each loop occurred sequentially and independently of subsequent loops, whereas the nascent polypeptide remained fully shielded from the cytosol; first by the ribosome exit tunnel, then the ribosome-translocon junction, and finally by the physical barrier of the ER membrane (Fig. 4A).

The Ribosome Simultaneously Shields Multiple ICLs throughout Synthesis-We next examined movement of ICLs.
Our previous studies demonstrated that AQP4 TM1 undergoes headfirst insertion and inversion when expressed as a fusion protein from which downstream TMs have been removed (25). In the presence of downstream TMs, however, it was difficult to distinguish distinct melittin or salt-sensitive phases for the N terminus (Cys-9) during the early stages of synthesis (Fig. 5A). A modest melittin sensitivity was observed upon membrane targeting (Ͻ100 aa length), and gave way to a weak salt sensitivity at nascent chain lengths of ϳ150 aa. As the nascent chain was elongated to ϳ180 aa, pegylation efficiency gradually increased to a level similar to full-length AQP4, suggesting the N terminus had moved into the cytosol.
On other hand, ICL1, ICL2, and the C terminus exhibited salt-sensitive pegylation when the Cys probes were ϳ40 -60 residues from the PTC (Fig. 5, B-D). Residue Cys-68 (in ICL1) also exhibited a transient increase in melittin sensitivity for unclear reasons that may reflect brief exposure of ICL1 to lumenal contents. Salt-sensitive pegylation of ICL1 and ICL2 decreased severalfold as synthesis continued beyond 180 and 230 aa, respectively, and remained below that observed for fulllength AQP4 throughout translation (Fig. 5, B and C). In contrast, the C terminus remained inaccessible to cytosol, but in a salt-sensitive environment.
Ribosomes Shield the Nascent Chain in a Salt-resistant Manner-Ribosome binding to Sec61 is mediated primarily through electrostatic interactions between the exit tunnel and L8 and C-terminal residues of Sec61 (68). To determine whether the persistent cytosolic shielding observed for ICL1 and ICL2 was ribosome independent, microsomes were treated  with RNase prior to incubation with PEG-Mal. RNase digestion restored pegylation for both ICL1 and ICL2 to a level similar to full-length, membrane-integrated AQP4 (Fig. 6, A and B). As expected, RNase treatment had no additional effect on the C terminus pegylation (Fig. 6C). Thus cytosolic regions of the nascent AQP4 polypeptide are continuously shielded from the cytosol by the RTC in at least two biochemically distinguishable environments (Fig. 6D).

Discussion
This study describes cotranslational delivery of a native PMP into its appropriate cellular compartments during repetitive cycles of translocation initiation and termination. By quantifying PEG-Mal accessibility of engineered Cys residues in sequentially truncated integration intermediates, we were able to reconstruct movement of each peptide segment (excepting the N terminus, see above) through the RTC at sequential stages of synthesis. Results show that following ribosome binding to the ER membrane, nascent AQP4 became shielded from the cytosol and remained cytosolically inaccessible throughout its synthesis. Extracellular peptide loops, ECL1, ECL2, and ECL3, passed directly from the ribosome exit tunnel, through the ribosome translocon junction, the translocon pore, and into the ER lumen coincident with nascent chain elongation. Movement along this pathway corresponds to the predicted length of a relatively extended polypeptide tethered to the ribosome peptidyl-transferase center. After membrane targeting, the N terminus was briefly accessible to the lumen and became cytosolically exposed as the nascent chain was lengthened, consistent with previous studies (25). Unexpectedly, intracellular peptide loops, ICL1, ICL2, and the C terminus remained shielded from the cytosol by the RTC until translation was terminated and the nascent polypeptide was released from the ribosome. In addition, these peptide regions were retained in biochemically distinct environments. ICL1 and ICL2 passed from a salt-sensitive to a salt-insensitive location, whereas the C terminus remained salt-sensitive until synthesis was completed. Thus, the assembled RTC simultaneously restricts cytosolic access of at least three distinct peptide loops for a prolonged period of AQP4 synthesis.
These results provide insight into how the cytosolic loops and their associated TMs disengage from the RTC during early stages of helical packing and folding. Previous studies analyzing TM integration by cryo-electron microscopy (16,18), chemical cross-linking, site-specific photocross-linking (36,64), and FRET (56) have suggested two scenarios. One model proposed that TMs rapidly sample the hydrophobic membrane environment upon entering Sec61, and spontaneously partition through the lateral PCC gate into membrane lipids (10 -12, 18, 69) based on the effective tether length and partitioning free energy of the TM segment (49,70). An alternative model suggests that TMs remain adjacent to the translocon and/or its associated proteins, and are released into the bilayer via mechanistically triggered events, i.e. synthesis of downstream TMs or translation termination (28, 33, 36, 52, 54 -56, 71). If TMs integrated rapidly into the lipid bilayer, then the length of the nascent chain tethering the TM segment to the PTC would limit diffusion in the membrane. In that case, cytosolic loops would be expected to sequentially move out from beneath the ribosome as the tether length from the exit tunnel to the C terminus of the TM segment reached ϳ70 Å. Given the length of AQP4 ICLs (10 -30 amino acids), release of TMs into the bilayer would minimally occur for ICL1 and ICL2 at or near truncation positions 166 and 259, respectively. However, neither ICL1, ICL2, nor the C terminus were cytosolically accessible during AQP4 synthesis, suggesting that TMs accumulate within or in very close proximity to the fully assembled RTC during synthesis. This finding is consistent with evidence that  . Salt-insensitive shielding of ICL1 and ICL2 is ribosome dependent. A and B, pegylation of ICL1 (68Cys) and ICL2 (161Cys) at longer truncations following ribosome removal with RNase restores pegylation to a level similar to full-length (FL), membrane-integrated AQP4. C, the C terminus exhibits salt-sensitive pegylation that is not further effected by RNase treatment. D, schematic of salt-insensitive ICL1 and ICL2 shielding and C terminus salt-sensitive shielding by the RTC, which supports a model of TM retention within or near the translocon until translation termination. multiple AQP4 TMs (TM1, -3, -4, and -5) can simultaneously cross-link Sec61␣ (36).
Our results are also consistent with free energies of partitioning (⌬G app ) as calculated by Hessa et al. (70), where TMs with ⌬G app Ͻ 0 were found to promote insertion. The ⌬G app values calculated for AQP4 varied from Ϫ0.3 kcal mol Ϫ1 (TM1 and -3) to greater than ϩ1.0 kcal mol Ϫ1 (TM4 and -5), with an overall average of ϩ0.482 kcal mol Ϫ1 (49). These values are similar to TM helices in PMPs of known three-dimensional structure (49). It has also been proposed that many native PMPs require neighboring TM helices for proper partitioning into the membrane (38,39,49,70). Therefore, TM retention by the RTC may reflect more energetically favorable early protein-protein interactions over less favorable protein-lipid interactions (28,36,56,(71)(72)(73)(74).
Although we are currently unable to determine precisely when AQP4 acquires native tertiary structure, the proposed ribosome shielding likely provides a proteinaceous environment for the nascent polypeptide to sample non-native conformations without the ensuing complications of early cytosolic contact. Taken together, our results lend support to a model in which the RTC may impact the folding landscape by minimizing potential off-pathway folding events (1), in a manner analogous to cytosolic chaperonins such as the Escherichia coli GroEL/ES complex (75) and eukaryotic TRiC (76), and allow nascent proteins to access folding intermediates that would otherwise be energetically unfavorable (i.e. TM boundary adjustment, helical packing, or even TM inversion (25,38,42,77,78)). Because integration intermediates examined here reflect an equilibrated conformation, we cannot assess whether additional kinetic intermediates might exist in cells where translation occurs continuously (79). Examination of different substrates and conditions may be needed to resolve this issue, particularly because TM integration can be influenced by cooperative interactions between adjacent TMs (37,47,80) and protein-protein interactions with the translocon (33,36).
Another unexpected result is that cytosolic loops are retained in an environment that is both ribosome-dependent and salt-insensitive for synthesis of ϳ80 residues. Biochemical and structural investigations have shown that ribosomes bind principally to the cytoplasmic loop 8 and C terminus of Sec61␣ (or homolog SecY) (15,68) and engage ribosomal RNA helices H50-H53-H59 and H6-H24-H50 in the exit vestibule (12,16) via salt-sensitive electrostatic interactions (12,81,82). This raises an interesting question about the nature of interactions responsible for salt resistance and their role in early PMP folding. One possibility is that salt-insensitive pegylation could reflect ribosome effects on PEG-Mal reactivity by altering the pK a of the thiol side chain. Alternately, additional ribosome interactions, perhaps with other translocon components such as Sec62/63 (9), may stabilize the RTC complex in a salt-resistant manner during accumulation of PMP cytosolic peptide loops.
An obvious question therefore is how much cytosolic polypeptide can be accommodated beneath the ribosome in a cytosolically protected state. Similarly, what additional components (if any) participate in delivery of cytosolic domains into the cytosol? Unfortunately, current structural models provide little guidance into these questions, as most studies to date have examined relatively simple proteins in the context of a single Sec61-PCC and ribosome. Our findings suggest that the RTC is a dynamic complex that acts not only as a translocation and integration conduit, but also to influence and orchestrate early PMP folding events. Structural studies of PMP intermediates in assembled RTCs containing native accessory proteins (OST, TRAM, TRAP, Sec62, and Sec63) will likely be needed to fully resolve these issues.
Author Contributions-M. A. P. designed and performed the majority of experiments, obtained the majority of data shown, and contributed to writing the manuscript and figure preparation. A. B. designed and conceived experiments, helped develop the experimental system, and obtained preliminary data needed for the study. P. K. D. initiated the project, performed experiments, and developed the experimental pegylation protocols. J. W. performed the oocyte water channel experiments collected and analyzed this data with L. R. L. R. performed oocyte water channel experiments, supervised J. W., and analyzed and formatted the data and contributed to the conceptual development of the project. Z. Y. designed and synthesized cDNA plasmid constructs and verified sequencing data. W. R. S. conceived the project, supervised all aspects of the project, contributed to writing and figure preparation, and data analysis. All authors analyzed results and approved the final version of the manuscript.