Sequential activation of human signal recognition particle by the ribosome and signal sequence drives efficient protein targeting

Significance

The universally conserved signal recognition particle (SRP) delivers ∼30% of the proteome to the appropriate cellular membrane. How SRP achieves efficient and selective protein targeting in eukaryotes remains elusive. Here, we show that a functional signal sequence on the nascent polypeptide confers a significant kinetic privilege to ribosome-bound SRP during the recruitment of SRP receptor, thereby enabling rapid and selective membrane targeting of SRP-dependent substrates. In addition, single-molecule spectroscopy revealed ribosome- and signal sequence-induced global conformational rearrangements in human SRP, which, together with biochemical analyses, provides a molecular model for substrate-induced activation of mammalian SRP. Analogous mechanisms have been described for replication, transcription, and translation, and may be envisioned for other pathways in which biological fidelity exceeds substrate binding specificity.

Abstract

Signal recognition particle (SRP) is a universally conserved targeting machine that mediates the targeted delivery of ∼30% of the proteome. The molecular mechanism by which eukaryotic SRP achieves efficient and selective protein targeting remains elusive. Here, we describe quantitative analyses of completely reconstituted human SRP (hSRP) and SRP receptor (SR). Enzymatic and fluorescence analyses showed that the ribosome, together with a functional signal sequence on the nascent polypeptide, are required to activate SRP for rapid recruitment of the SR, thereby delivering translating ribosomes to the endoplasmic reticulum. Single-molecule fluorescence spectroscopy combined with cross-complementation analyses reveal a sequential mechanism of activation whereby the ribosome unlocks the hSRP from an autoinhibited state and primes SRP to sample a variety of conformations. The signal sequence further preorganizes the mammalian SRP into the optimal conformation for efficient recruitment of the SR. Finally, the use of a signal sequence to activate SRP for receptor recruitment is a universally conserved feature to enable efficient and selective protein targeting, and the eukaryote-specific components confer upon the mammalian SRP the ability to sense and respond to ribosomes.

Proper localization of nascent proteins is essential for maintaining compartmentalization and protein homeostasis in all cells (1). The universally conserved signal recognition particle (SRP) pathway is responsible for the targeted delivery of ∼30% of the newly synthesized proteome to the eukaryotic endoplasmic reticulum (ER) or the bacterial plasma membrane. SRP recognizes an N-terminal transmembrane domain (TMD) or hydrophobic signal sequence as a nascent protein emerges from the translating ribosome. Through interaction with the SRP receptor (SR), SRP delivers translating ribosomes to the Sec61p (or SecYEG) translocase on the target membrane. Bacteria contain the simplest SRP, comprising the universally conserved SRP54 protein bound to the 4.5S SRP RNA. SRP54 is a multidomain protein that contains an M-domain, which binds the SRP RNA and recognizes signal sequences on the nascent polypeptides, and a special GTPase domain termed the NG-domain, which contacts the ribosome and binds to a homologous NG-domain in SR (termed FtsY in bacteria) (2, 3). Extensive biochemical and biophysical studies demonstrated how an SRP-dependent signal sequence or TMD in a ribosome•nascent chain complex (RNC) regulates the GTP-dependent interaction of SRP with SR and their reciprocal GTPase activation, thereby enabling efficient and specific cotranslational protein targeting in bacteria (4–6).

SRP undergoes an extensive expansion in size and complexity during evolution. The eukaryotic SRP contains a larger 7SL SRP RNA and six protein subunits (SRP9, SRP14, SRP19, SRP54, SRP68, and SRP72). Eukaryotic SR is a heterodimer of SRα and SRβ subunits. SRα contains an NG domain homologous to that in bacterial FtsY and an additional X-domain that binds the cytosolic domain of SRβ. SRβ contains an additional N-terminal TMD that anchors the eukaryotic SR at the ER membrane (3). The complexity of the mammalian SRP has limited in-depth mechanistic analyses, and the mechanism by which the eukaryotic SRP pathway achieves efficient and selective protein targeting remains unclear. Microarray and ribosome-profiling analyses of SRP-associated RNCs in yeast (7, 8) suggested that the eukaryotic SRP can associate with translating ribosomes without or before the emergence of a signal sequence, raising questions as to the timing and specificity of cargo recognition by SRP. On the contrary, proximity-ribosome profiling experiments in yeast and mammalian cells showed that most ER-associated ribosomes targeted by SRP contain TMD targeting signals (9, 10), suggesting that eukaryotic SRP maintains high targeting selectivity. A potential resolution of these observations is that molecular events after ribosome binding govern the selectivity of this pathway. As SRP has a limited time window to complete the targeting reaction before the nascent polypeptide reaches a critical length (11, 12), cargo recognition by SRP must be coupled to efficient delivery to the ER membrane, a process mediated by the direct interaction between the NG-domains of SRP54 and SRα. However, limited information is available on this critical step in the eukaryotic SRP pathway. Previous work showed that an RNC stimulates the GTPase activity of mammalian SRP and SR, presumably when they form a complex (13). Nevertheless, an empty ribosome was also found to stimulate the SRP–SR interaction and their GTPase activity (14), and direct interactions between the mammalian SR and ribosome have been detected (15). These observations raise questions as to whether the recruitment of SR is specific to SRPs bound to RNCs bearing SRP-dependent substrates. The roles and contributions of the ribosome and signal sequence in regulating the membrane-targeting step in the eukaryotic SRP pathway are unresolved, as is the mechanism by which these regulations are exerted.

To address these questions, we reconstituted functional human SRP (hSRP) and SR from recombinant components, which enabled their mechanistic interrogation at high resolution. We found that the ribosome and a functional signal sequence are necessary for the most efficient assembly and reciprocal GTPase activation between mammalian SRP and SR. Single-molecule Förster resonance energy transfer (smFRET) measurements showed that the signal sequence plays a dominant role in inducing a “proximal” conformation of SRP that is optimal for the recruitment of SR, whereas the ribosome unlocks SRP from an autoinhibited mode and allows SRP to sample a variety of conformations. These results, together with cross-complementation analyses, showed that the use of signal sequence to activate SRP for receptor recruitment is a universally conserved feature of SRP pathways, but the mammalian-specific components enable the SRP to also sense and be primed by the ribosome.

Results

Ribosome and Signal Sequence Together Activate the hSRP-hSR GTPase Cycle.

Previous biochemical work on the mammalian SRP pathway (13, 16) has largely relied on native SRP and SR. The low quantity and inability to perturb the system limited in-depth mechanistic investigations. To overcome this barrier, we assembled hSRP from recombinantly purified components using modifications of published procedures (SI Appendix, Fig. S1A) (17–20). Holo-SRP was selectively purified by using DEAE-Sephacel, which effectively removes free 7SL RNA and incompletely assembled SRPs (18) (SI Appendix, Fig. S1 B–D). We also expressed and purified a soluble human SR (hSR) complex comprising full-length hSRα and hSRβΔTM, in which the dispensable N-terminal TMD in SRβ was removed (SI Appendix, Fig. S1 E and F) (21). The reconstituted hSRP and hSRαβΔTM are highly active in mediating the cotranslational targeting and insertion of a model SRP substrate, preprolactin (pPL), into trypsin-digested rough ER microsomes that lack endogenous SRP and SR (22) (SI Appendix, Fig. S1 G and H), with efficiency comparable to that of native SRP (23).

To understand how the GTPases in the mammalian SRP and SR regulate protein targeting, we defined a rigorous kinetic and energetic framework for their GTPase cycle that includes the basal GTPase cycles of free hSRP and hSR (Fig. 1A, upper triangles), their assembly with one another (Fig. 1A, step 4), and GTP hydrolysis in the hSRP•hSR complex (Fig. 1A, step 5). The basal GTPase cycle of hSRP (or hSR) was determined by measuring GTP hydrolysis rates under single-turnover conditions with the enzyme in excess of GTP. The slow observed k_cat (0.00060 s⁻¹ and 0.0033 s⁻¹ for hSRP and hSR, respectively) and k_cat/K_m (3.2 × 10² M⁻¹⋅s⁻¹ and 3.7 × 10² M⁻¹⋅s⁻¹ for hSRP and hSR, respectively) values in the basal GTPase reactions strongly suggest that equilibrium binding of GTP occurs before GTP hydrolysis. Thus, the value of K_m equals K₁ (or K₁′), the equilibrium dissociation constant of GTP for hSRP (or hSR). The binding affinity of GDP for hSRP (or hSR) was determined by using GDP as a competitive inhibitor of the basal GTPase reactions. The interaction and reciprocal activation between hSRP and hSR was initially assessed by measuring the rates of the reciprocally stimulated GTPase reaction by using a small, fixed amount of hSRP and varying concentrations of excess hSR. As validated by independent fluorescence-based measurements of the hSRP–hSR interaction (as detailed later), the value of k_cat/K_m in this reciprocally stimulated GTPase reaction equals k₄, the rate constant for hSRP–hSR assembly, and the value of k_cat reports on the rate constant of GTP hydrolysis from the most stable hSRP•hSR complex that accumulates during GTP turnover.

Fig. 1.

Summary of the individual steps in the GTPase cycles of hSRP and SR. (A) Scheme of the GTPase cycles of hSRP (blue) and hSR (green). Superscripts depict the nucleotide bound to each protein. The triangular cycles (Top) depict the basal GTPase cycles of hSRP54 and hSR, respectively. Binding of GTP and GDP to hSRP (or SR) are characterized by the equilibrium dissociation constants K₁ and K₃ (or K₁′ and K₃′), respectively. Rate constants for GTP hydrolysis from free hSRP and hSR are denoted by k₂ and k₂′, respectively. Complex formation between hSRP and hSR is characterized by the association rate constant k₄ and dissociation rate constant k₋₄. Bound GTPs are hydrolyzed from the ^GTP•hSRP•hSR^•GTP complex, represented collectively by the rate constant k₅, followed by dissociation of the ^GDP•hSRP•hSR^•GDP complex. (B) Summary of the kinetic parameters described in A. Determination of the individual rate and equilibrium constants is described in SI Appendix, Supplemental Methods.

Analogous to their bacterial homologs, hSRP and hSRαβΔTM by themselves displayed weak nucleotide affinities and slow basal GTPase rates (Fig. 1B, K₁, k₂, K₃ and K₁′, k₂′, K₃′, and SI Appendix, Fig. S2 A–C). Bacterial SRP and SR activate the GTPase activities of one another when they form a complex (16, 24). In contrast, this reciprocal GTPase activation was barely detectable when hSRP and hSRαβΔTM were incubated together (Fig. 2A, black circles), even in the presence of the detergent NIKKOL (octaethylene glycol monododecyl ether, C₁₂E₈) that stimulated the GTPase cycle of bacterial SRP and SR (SI Appendix, Fig. S2D) (25). We therefore searched for potential regulators that stimulate the GTPase cycle of mammalian SRP and SR. Given the observation that fusion of a signal peptide to the C terminus of the SRP54 M-domain (26) led to structural reorganization of archaeal SRP54 and its stimulated GTPase reaction with SR (27), we generated mutant hSRP-4A10L in which the C-terminal M-domain of hSRP54 is fused to a model signal sequence, 4A10L (LALALLLLLLALAL; also see SI Appendix, Fig. S3A). In addition, we tested the effect of purified 80S ribosome, which was reported to enhance the stimulated GTPase reaction between canine SRP and SR (13, 14).

Fig. 2.

The ribosome and signal sequence activates the SRP–SR GTPase cycle. (A and B) Representative hSR concentration dependences of the reciprocally stimulated GTPase reaction between SRP and hSRαβΔTM (A) or hSRαΔX (B). Reactions contained 0.2 μM hSRP or hSRP-4A10L, 100 μM GTP, and indicated concentrations of hSR. Purified 80S was present at 0.25 μM where indicated. The lines are fits of the data to SI Appendix, Eq. S2, in SI Appendix, Supplemental Methods. (C and D) Summary of the k_cat (C) and k_cat/K_M (D) values derived from analysis of the data in A and B and their replicates. Data are represented as mean ± SD, with n ≥ 2.

Indeed, the presence of the signal sequence or the 80S ribosome enhanced the reciprocally activated GTPase reaction between hSRP and hSR (Fig. 2A). In contrast, the ribosome and signal sequence provided <50% stimulation for free hSRP (SI Appendix, Fig. S2 E and F), indicating that the observed GTPase stimulations are specific to the hSRP•hSR complex. Quantitative analysis of the GTPase data also revealed modest differences in the effects of the signal sequence and the ribosome. The slope at subsaturating hSR concentrations (i.e., k_cat/K_m), which reports on the assembly between hSRP and hSR (corroborated by fluorescence measurements of the hSRP–hSR interaction as detailed later), is more strongly stimulated by the ribosome (Fig. 2D and SI Appendix, Table S1). The rate constant at saturating protein concentration (i.e., k_cat), which reports on the rate constant of GTP hydrolysis from a stably formed hSRP•hSR complex, is more strongly stimulated by the signal sequence (Fig. 2C and SI Appendix, Table S1). When the ribosome and signal sequence are present, which provides a mimic of the RNC (corroborated by experiments with RNC_4A10L as detailed later), the strongest stimulation was observed for both rate constants (Fig. 2, red, and SI Appendix, Table S1), indicating synergistic actions of both components in activating the GTPase cycle of the mammalian SRP and SR.

A Functional Signal Sequence Confers Kinetic Privilege to hSRP During hSR Recruitment.

To directly monitor complex formation between hSRP and hSR, we developed a FRET assay. A donor dye (Cy3B) was labeled at an engineered cysteine (C47) by using thiol-specific maleimide chemistry in Cys-lite hSRP54, in which all of the solvent-exposed native cysteines were removed (C36T, C136S, and C229A; Fig. 3A). The two remaining native cysteines in hSRP54 were buried and not labeled under our experimental conditions. An acceptor dye (ATTO 647N) was conjugated to the C terminus of hSRα via sortase-mediated ligation (Fig. 3A) (28). Incubation of labeled hSRP and hSR resulted in a significant reduction in donor fluorescence and increase in acceptor fluorescence, and these fluorescence changes can be competed away by unlabeled SR (Fig. 3B), indicating FRET between the dye pair. The mutations and fluorescence labeling did not substantially affect the GTPase activity of hSRP and hSR or their protein targeting activity (SI Appendix, Fig. S4). In the course of these experiments, we found that a simpler hSR construct hSRαΔX, in which the X-domain of SRα and SRβ are removed (SI Appendix, Fig. S1 E and F), displayed SRP–SR assembly, GTPase activation, and preprotein targeting activities that are comparable to or slightly higher than those of hSRαβΔTM (Fig. 2 B–D and SI Appendix, Figs. S1 G and H and S2 B and C). This is consistent with previous observations using canine SRP and SR (14, 29–31) and indicated that hSRαΔX provides a fully functional mimic of hSR for studying the initial assembly between hSRP and hSR. Hence, all subsequent fluorescence measurements of the hSRP–hSR interaction were carried out with hSRαΔX. Finally, to block GTP hydrolysis from the hSRP•hSR complex, which would provide an alternative pathway for complex dissociation [via the less stable^GDP•SRP•SR^•GDP complex (32)], we introduced the R458A mutation in hSR, which disrupts catalytic interactions at the composite active site between hSRP and hSR (SI Appendix, Fig. S5A). As expected, this mutant was catalytically dead in the reciprocally activated GTPase reaction between hSRP and hSR (SI Appendix, Fig. S5B) but effectively competed with WT hSR for interaction with hSRP (SI Appendix, Fig. S5C), indicating that mutant hSRαΔX (R458A) still allows rapid and stable hSRP•hSR complex assembly but specifically blocks GTPase activation in the complex.

Fig. 3.

Ribosome and signal sequence stabilize and accelerate SRP•SR complex formation. (A) The positions of FRET probes are shown on the crystal structure of the NG domain complex between hSRP (blue) and hSRα (green; PDB ID code 5L3Q) (56). (B) Fluorescence emission spectra of 20 nM Cy3B-labeled hSRP-4A10L before (green) and after addition of 400 nM ATTO 647N-labeled hSRαΔX (blue), and of 400 nM ATTO 647N-labeled hSRαΔX alone (red). Addition of 2 μM unlabeled hSRαΔX (black) to a preformed hSRP•SR complex restores the donor fluorescence and reduces acceptor fluorescence, confirming that a large fraction of the observed fluorescence change arises from FRET. The reactions also contained 40 nM 80S to facilitate complex assembly. (C) Representative equilibrium titrations of SRP•SR complex formation using the FRET assay. Titrations used 12.5 nM hSRP or hSRP-4A10L, indicated concentrations of hSR, and 2 mM GTP. Where indicated, 300 nM and 40 nM 80S were used for titrations with hSRP and hSRP-4A10L, respectively. The lines are fits of the data to SI Appendix, Eq. S7, in SI Appendix, Supplemental Methods, and the obtained K_d values are summarized in SI Appendix, Table S2. All measurements were repeated at least twice. (D and E) FRET-based measurements of SRP–SR association kinetics. Reactions contained the same concentrations of all of the factors as in C. The data were fit to SI Appendix, Eq. S5, in SI Appendix, Supplemental Methods, and the obtained k_on values are summarized in E and SI Appendix, Table S2. Error bars denote SD, with n ≥ 2.

By using this FRET assay, we tested how the kinetics and equilibrium of hSRP•hSR complex formation are regulated. Equilibrium titrations showed that the interaction between hSRP and hSR by themselves was weak, with an equilibrium dissociation constant (K_d) in the micromolar range (Fig. 3C and SI Appendix, Table S2). The hSRP•hSR complex was stabilized approximately eightfold by the signal sequence and ∼40-fold by the ribosome (Fig. 3C and SI Appendix, Table S2). With both signal sequence and ribosome present, the equilibrium stability of the complex was 56 nM, similar to that with the ribosome (Fig. 3C and SI Appendix, Table S2). Independent determination of K_d values from the ratio of dissociation and association rate constants (k_off/k_on) yielded similar conclusions (SI Appendix, Fig. S6A and Table S2). On the contrary, kinetic analysis of hSRP•hSR complex assembly showed that this reaction was intrinsically slow, with a k_on of <10² M⁻¹⋅s⁻¹. Assembly was accelerated ∼10²-fold by the signal sequence (Fig. 3 D and E). The ribosome also provided a ∼10³-fold stimulation, and the additional presence of the signal sequence further accelerated hSRP–hSR assembly 20-fold, bringing the k_on value to >10⁶ M⁻¹⋅s⁻¹ (Fig. 3 D and E). These results are consistent with the effect of the ribosome and signal sequence on the k_cat/K_m values measured in the GTPase assay (Fig. 2D and SI Appendix, Table S1). Thus, although the ribosome appears to dictate the equilibrium stability of the hSRP•hSR complex, the signal sequence provides a significant additional stimulation for the kinetics of hSRP•hSR complex formation.

To verify that the combination of ribosome and signal sequence fusion to hSRP54 provides a reasonable mimic for the physiological SRP substrate, we generated stalled RNCs by in vitro translation of a truncated mRNA encoding the first 90 aa of pPL without a stop codon (SI Appendix, Fig. S3A) (33). The signal sequence in pPL was replaced by 4A10L to allow direct comparison with hSRP-4A10L. As a negative control, we inserted two arginines into the pPL signal sequence to generate RNC_2R (SI Appendix, Fig. S3A) (23). In vitro targeting assay confirmed that the 2R mutation disrupted SRP-dependent protein targeting (Fig. 4A and SI Appendix, Fig. S3B). RNC_4A10L and RNC_2R were affinity-purified via a 3xFLAG tag N-terminal to the pPL coding sequence followed by a sucrose gradient to isolate a homogenous population of monosomes bearing the nascent polypeptide. Binding of hSRP to RNC_4A10L and to the 80S ribosome was confirmed by a binding assay based on microscale thermophoresis (MST; SI Appendix, Fig. S3C), and saturating concentrations of the RNC and ribosome with respect to their hSRP binding constants were used for the measurements detailed later.

Fig. 4.

A functional signal sequence on the RNC stimulates the SRP–SR GTPase cycle and accelerates SRP•SR complex formation. (A) Introduction of the 2R mutation in the pPL signal sequence abolished cotranslational targeting by hSRP and hSR. Targeting reactions were carried out as described in SI Appendix, Supplemental Methods, using 50 nM hSRP and the indicated hSR concentrations. (B) Reciprocally stimulated GTP hydrolysis reactions between hSRP and hSR were measured in the presence of 0.3 µM RNC_4A10L (magenta) or RNC_2R (olive). All other reaction conditions are the same as in Fig. 2A. The reactions with hSRP-4A10L and 80S (red) were performed in parallel for direct comparison. The reaction with 80S (green) was taken from Fig. 2B for comparison. The lines are fits of the data to SI Appendix, Eq. S2, in SI Appendix, Supplemental Methods, and the obtained rate constants are summarized in SI Appendix, Table S1. (C) FRET-based measurements of hSRP–SR association kinetics in the presence of RNC_4A10L or RNC_2R, carried out under the same conditions as in Fig. 3C. (D) Summary of the hSRP–SR association rate constants obtained from analysis of the data in C and their replicates. The data with ribosome and SRP-4A10L (red and green bars) are from Fig. 3 and shown for comparison. All error bars denote SD, with n ≥ 2.

Purified RNCs were tested for their ability to stimulate hSRP•hSR complex assembly and subsequent GTPase activation. In the GTPase assay, the reactions with RNC_4A10L behaved more similarly to those of signal sequence-fused hSRP in the presence of the ribosome, whereas the reactions with RNC_2R were more similar to that of hSRP with empty ribosomes (Fig. 4B). In the FRET assay, RNC_4A10L induced rapid hSRP–hSR assembly, with a rate constant similar to that observed with the combination of signal sequence and ribosome, and 10- and 20-fold faster than the assembly rates observed with empty ribosome and RNC_2R, respectively (Fig. 4 C and D and SI Appendix, Table S2). Although RNC_2R presumably binds hSRP more weakly, the RNC_2R-induced stimulation of hSRP–hSR assembly and GTPase activation indicated that binding between hSRP and RNC_2R occurred. Furthermore, the observed GTPase rate constant was not substantially enhanced by increasing the concentration of RNC_2R beyond 300 nM (Fig. 4B and SI Appendix, Fig. S3E), indicating that an RNC_2R•hSRP complex was completely formed under our experimental conditions. Thus, the 20-fold faster hSRP–hSR assembly rates with RNC_4A10L than with RNC_2R cannot be attributed to incomplete binding of hSRP by RNC_2R, and instead reflects the kinetic advantage in SR recruitment provided by a functional signal sequence.

The Ribosome Relieves Autoinhibition in Mammalian SRP.

Compared with the bacterial SRP system, the ribosome plays a much larger role during the receptor recruitment of mammalian SRP (refs. 13 and 14 and this work). To understand the mechanism underlying these differences, we carried out cross-complementation analyses by making a hybrid SRP comprised of hSRP54 bound to the 4.5S SRP RNA from Escherichia coli. GTPase assays revealed surprising similarities and differences between the hybrid and hSRP. First, the hybrid SRP displayed higher intrinsic activity than hSRP in the reciprocally stimulated GTPase reaction with hSR in the absence of external activators (Fig. 5B; cf. black lines in Fig. 5A vs. Fig. 2A). This suggests that the additional components in hSRP inhibit hSRP54 from attaining a conformation conducive to hSR recruitment and GTPase activation. This result also ruled out general defects in the folding or conformation of hSRP54 in the absence of the additional subunits in hSRP. Second, signal sequence fusion to hSRP54 also stimulated the reciprocally stimulated GTPase reaction between the hybrid SRP and hSR, analogous to the hSRP (Fig. 5, blue). This indicates that the ability to sense and respond to a signal sequence is an intrinsic property of hSRP54 and can occur independently of the other SRP protein subunits. Finally, in contrast to the hSRP, the ribosome lost most of its stimulatory effects on the interaction and reciprocal GTPase activation between the hybrid SRP and hSR (Fig. 5, green), indicating that the mammalian-specific components in hSRP are required for the ribosome to exert its stimulatory effects. Together with previous work in bacterial and archaeal SRP (25, 34), the results of these cross-complementation analyses suggest that the signal sequence-induced stimulation of SR recruitment is a universally conserved property of SRP, whereas the ribosome-induced simulation of this event is a mammalian-specific phenomenon.

Fig. 5.

Comparison of the activities of hSRP and hybrid SRP (hSRP54 bound to 4.5S RNA). (A) Reciprocally stimulated GTPase reaction between hybrid SRP and hSR in the presence of the indicated factors. Reactions were measured under the same conditions as in Fig. 2A. The lines are fits of the data to SI Appendix, Eq.S2, in SI Appendix, Supplemental Methods. (B and C) Summary of the k_cat (B) and k_cat/K_M (C) values from analysis of the data in A and their replicates. All data are represented as mean ± SD, with n = 3. The rate constants for hSRP are from Fig. 2 and are shown for comparison.

Signal Sequence Preorganizes hSRP into the Optimal Conformation for hSR Recruitment.

To understand the mechanism(s) by which the signal sequence and ribosome activate the interaction between hSRP and SR, we characterized the global conformational changes of hSRP based on FRET measurements between a donor dye (ATTO 550) labeled at SRP19 (C64) and an acceptor dye (ATTO 647N) labeled at SRP54 (C12). Based on the cryo-EM structure of the native hSRP•RNC complex [Protein Data Bank (PDB) ID code 3JAJ] (33), the distance between the dye pair is ∼44 Å (Fig. 6A). Hence, a high FRET efficiency is expected for this dye pair (Förster radius, 65 Å) if the SRP54 NG-domain is positioned near SRP19, which we term the proximal conformation. FRET was measured at single-molecule resolution (i.e., smFRET) based on fluorescence-aided molecular sorting using alternating laser excitation spectroscopy (ALEX) (35), which optically purifies doubly labeled single SRPs diffusing through a femtoliter-scale observation volume and extracts uncorrected FRET efficiencies (E*) for individual particles (SI Appendix, Fig. S7 A–C) (36). Diffusion of the labeled molecules through the femtoliter-scale observation volume is estimated to take ∼1 ms. Thus, different conformations that exchange on the millisecond or longer timescale can be resolved as discrete populations in a FRET histogram (37).

Fig. 6.

Signal sequence and the ribosome exert different effects on the conformation of hSRP. (A) Approximate positions of fluorescent donor and acceptor dyes on hSRP19 (magenta) and hSRP54 (blue), shown on a cryo-EM structure of the mammalian RNC•SRP complex (PDB ID code 3JAJ) (33). (B–G) smFRET histograms of hSRP in the presence of different ligands. E*, uncorrected FRET efficiency; n, number of bursts used to construct each histogram, obtained from at least five independent measurements; pdf, probability density function. The data were fit with the sum (solid line) of three Gaussian functions (dotted lines), and the dotted red lines denote the peak E* value for each population. (H) Summary of the fraction of SRPs in the low-, medium-, and high-FRET states under the respective conditions.

smFRET measurements showed that hSRP by itself exhibits a FRET distribution that is dominated by a medium-FRET population with a peak E* value of ∼0.5 (Fig. 6B). With signal sequence fused hSRP, the FRET distribution shifted and peaked at a higher E* value (∼0.65; Fig. 6C). The presence of the signal sequence and ribosome further shifted the distribution to higher FRET, with E* peaking at ∼0.7 (Fig. 6D). This predominantly high FRET distribution was also observed with hSRP bound to RNC_4A10L (Fig. 6E), providing additional evidence that the combination of ribosome and signal sequence fusion provides a reasonable mimic of the effects of a signal sequence-bearing RNC. In contrast, empty ribosomes and RNC_2R induced significant conformational heterogeneity in hSRP (Fig. 6 F and G). The FRET distributions of hSRP became broad, and could be accounted for by at least three populations with low (E* value, ∼0.2), medium (E* value, ∼0.5), and high (E* value, ∼0.7) FRET values. Quantitative analyses of the FRET distributions further showed that, whereas free hSRP is dominated by the medium-FRET population (∼80%), hSRP bound to the ribosome and RNC_2R are approximately equally distributed among all three conformations, and the high-FRET population becomes dominant whenever a functional signal sequence is present (Fig. 6H). These results provide strong evidence that a correct cargo displaying an SRP-dependent signal sequence induces SRP into the proximal conformation, and that the signal sequence, rather than the ribosome, plays a dominant role in inducing this conformation.

To exclude possible artifacts due to local environmental perturbations on the photophysics of fluorophores, we repeated these measurements after swapping the position of donor and acceptor dyes in hSRP. smFRET measurements using this swapped dye pair yielded similar signal sequence- and ribosome-induced changes in the FRET distributions of hSRP (SI Appendix, Fig. S7 D–H). In addition, the presence of various interaction partners did not affect the dye photophysics in a way that would alter the FRET distributions (SI Appendix, Fig. S7 I–N). These data strongly suggest that our observed FRET changes can be attributed to the global conformational transitions of hSRP.

In summary, the smFRET data show that hSRP by itself adopts a conformation (or conformations) in which the SRP54-NG domain is positioned away from the proximal site where SRP19 and the SRP54 M-domain are located. In the presence of empty ribosomes or signal-less RNCs, hSRP explores a variety of alternative conformations in which the SRP54 NG-domain can be proximal to or further away from SRP19. In contrast, a functional signal sequence plays a dominant role in inducing the mammalian SRP into the proximal conformation.

Discussion

Efficient and selective targeting of nascent proteins by SRP is essential for the proper functioning of the endomembrane system in eukaryotic cells. Although the mechanism of the simplest bacterial SRP has been deciphered at high resolution (4–6), understanding of the more complex eukaryotic SRP has lagged behind. Previous fluorescence measurements suggested that mammalian SRP exhibits significant binding to empty ribosomes, with a K_d value (∼80 nM) significantly below the in vivo SRP concentration (∼500 nM in mammalian cells) (11, 38). Substantial signal-independent association of eukaryotic SRP with translating ribosomes was also observed in global analyses in yeast cells (7, 8). These observations raise questions as to whether SRP–RNC binding is sufficient to ensure selective cotranslational protein targeting in eukaryotic cells. In the bacterial SRP pathway, substrate selection relies heavily on kinetic discrimination during the recruitment of SR, a step that is more than 10²-fold faster with RNCs bearing SRP-dependent than SRP-independent substrates. Whether this mechanism is conserved in mammalian SRP has been unclear, especially given previous observations that the 80S ribosome by itself can bind mammalian SR and enhance its GTPase activity together with SRP (14, 15). In this work, biochemical and biophysical analyses provide evidence that a functional signal sequence provides a kinetic advantage during SR recruitment in the mammalian SRP pathway, and suggest a molecular model for the mechanism of SRP activation by the ribosome and signal sequence.

Cotranslational protein targeting by SRP kinetically competes with translation elongation, as RNCs lose the competence to be targeted by SRP when the nascent polypeptide exceeds a critical length [∼120 aa from the start of signal sequence or TMD (11, 12)]. Proximity-specific ribosome profiling in yeast further suggested that ER localization of ribosomes was attained immediately after the emergence of signal sequence from the ribosome [∼60 aa from the start of signal sequence (9)]. Considering the elongation rate of 6–10 aa/s in mammalian cells (39, 40), a limited time window of ≤12 s is imposed on the eukaryotic SRP to complete a targeting cycle (Fig. 7A). Although ribosomes and signal-less RNCs also substantially accelerate SRP–SR association, the measurements here suggest that the recruitment of SR at physiological concentrations [∼0.5 µM (38)] would require at least ∼13 s for empty ribosomes and ∼32 s for RNC_2R. The crowded cytosolic environment and competition from other ribosome-associated factors could further delay the bimolecular association between SRP and SR. The additional presence of a signal sequence brings the timescale of SR recruitment to ≤2 s, a timescale sufficient to meet the demands for cotranslational protein targeting (Fig. 7A). In addition, the weaker affinity of SRP for empty ribosomes than signal sequence-bearing RNCs (11) (SI Appendix, Fig. S3C) suggests that SRP dissociates more quickly from the former. Together with the difference in SR recruitment rates, this would allow most of the RNCs bearing SRP-dependent substrates to be successfully delivered to the ER membrane, whereas a larger fraction of signal-less ribosomes could dissociate from SRP before SR is recruited.

Fig. 7.

Comparison of E. coli and hSRP, and model for sequential conformational activation of mammalian SRP during targeting. (A) Comparison of the SRP–SR assembly rate constants between E. coli and hSRP in the absence (black) and presence of the ribosome (green) or RNC (red). The k_on values for E. coli SRP were from ref. 57. The shaded area with increasing red denotes SRP–SR interaction rates that are increasingly targeting-competent. The threshold for targeting competent SRP–SR assembly rates was determined using time windows of 5 s for bacterial SRP and 13 s for mammalian SRP to complete the targeting reaction and an SR concentration of 500 nM. (B) Sequential conformational activation of mammalian SRP by the ribosome and signal sequence for efficient SR recruitment. Free hSRP is in an autoinhibited conformation or an ensemble of autoinhibited conformations. Upon binding to the ribosome and in the absence of a signal sequence, hSRP is unlocked and samples a variety of alternative conformations (“sampling”). The emergence of the signal sequence drives most of the hSRP into the proximal conformation that allows rapid assembly with hSR for efficient targeting to the ER.

The ribosome is much more stimulatory during SR recruitment, whereas the additional kinetic advantage provided by the signal sequence on this event is much smaller, in the mammalian SRP system compared with its bacterial homolog (Fig. 7A, green vs. red). This suggests that additional strategies are used to maintain the fidelity of cotranslational protein translocation in eukaryotes, for example, by using the Sec61p translocase to discriminate against mutant signal sequences (41–43). In addition, the observation that the ribosome alone brings the SR recruitment rate to the threshold of a targeting-competent timescale raises additional possibilities of regulation, wherein other ribosome-interacting factors could drive a nascent protein into or out of the SRP pathway by tuning the relative rates of SR recruitment vs. protein synthesis. This may include ribosome-associated chaperones that alter the conformational landscape of ribosome-bound SRP, such as the nascent polypeptide-associated complex (7, 44, 45), as well as cis-regulatory elements or transinteracting factors that tune translation elongation rates, such as codon usage and arrest sequences at strategic locations on the mRNA (46). Compared with its bacterial homolog, the mammalian SRP may be more poised for diverse mechanisms of regulation by external factors. The same concept and considerations may be extended to evaluate the possibility of preemptive targeting, a model suggested by previous observations that SRP can associate with translating ribosomes before a signal sequence or TMD emerges from the nascent polypeptide exit tunnel (7, 8, 47, 48). An outstanding question regarding this model is whether productive SRP–SR interaction can initiate before the emergence of the targeting signal from the ribosome, and, if so, whether the targeting is specific for SRP substrates. The eukaryote-specific large stimulatory effect of the ribosome during SR recruitment suggests that preemptive targeting might play a more significant role in the mammalian than the bacterial SRP pathway.

Cross-complementation analyses and smFRET measurements of the global conformation of hSRP further suggest a molecular model by which the ribosome and signal sequence activate the hSRP–hSR interaction. The weak affinity and extraordinarily slow association rate of the hSRP–hSR complex indicate that hSRP by itself exists in a conformation inactive for receptor recruitment. The higher activity of the hybrid SRP compared with hSRP further suggests that the eukaryote-specific components in mammalian SRP are responsible, at least in part, for this autoinhibition. smFRET measurements further showed that free hSRP predominately adopts a conformation in which the dye pair between hSRP19 and hSRP54-NG exhibits medium FRET efficiency, indicating that the hSRP54-NG domain is positioned away from the proximal site of hSRP in this autoinhibited state (Fig. 7B). Importantly, a signal sequence-bearing RNC induces hSRP to predominantly adopt the proximal conformation, in which the hSRP54-NG domain is close to hSRP19 (Fig. 7B). Although it is intuitive to envision that the dual interactions of the hSRP54 M-domain with the signal sequence and the hSRP54 NG-domain with the uL23/uL29 ribosomal proteins (33) could induce this conformation, the finding here that signal sequence fusion to hSRP54 is sufficient to induce a near-proximal state of hSRP indicates that signal sequence occupancy in the hSRP M-domain plays a major role in bringing hSRP54-NG near the proximal site. This is in good agreement with crystallographic analyses in archaeal SRP (27) and suggests that binding of a signal sequence in the hSRP54 M-domain induces restructuring of the GM linker to reposition its NG domain.

Here we found a strong correlation between the acquirement of the proximal conformation of hSRP and faster SRP–SR assembly rates (all else held equal), suggesting that this conformation is optimal for SR recruitment. On the contrary, cryo-EM analyses of the bacterial and mammalian RNC•SRP•SR complexes indicate that, in the presence of SR, the density of the SRP54 NG-domain is no longer visible near the ribosomal tunnel exit (49, 50). Instead, the NG-domain complex moves to an alternative docking site near the -distal end of the SRP in both bacterial and mammalian SRP systems (51, 52). These observations strongly suggest that complex formation with SR induces detachment of SRP54-NG from the vicinity of the ribosome exit site in both SRP systems. Nevertheless, the results here provide strong kinetic evidence that initial SRP–SR assembly occurs near the ribosome exit tunnel, predicting an early RNC•SRP•SR intermediate before rearrangement of the NG-domain complex. The structure, dynamics, and interactions of this initial targeting intermediate remain to be defined.

The large stimulatory effect of the ribosome on SRP–SR assembly is a eukaryote-specific phenomenon (5). Previous work showed that the ribosome also binds the mammalian SR and suggested that it could provide a template that brings SRP and SR together for assembly (14, 15). Although this model is highly probable, the loss of ribosome-induced stimulation in the reaction of hybrid SRP with hSR indicates that additional mechanisms are necessary to account for all of the stimulatory effects from the ribosome, and the results here suggested an allosteric mechanism. smFRET measurements showed that, in contrast to the predominantly medium-FRET state observed with free hSRP, the ribosome induces multiple alternative conformations in which hSRP54-NG can be close to or even further away from the hSRP proximal site. In the simplest model, the stimulated recruitment of hSR could arise solely from the subpopulation of ribosome-bound hSRP in the proximal conformation. It is plausible that the low-FRET state observed with 80S-bound hSRP could also recruit hSR, but the correlation between assembly rates and smFRET data strongly suggest that SR recruitment in this state is slower than in the proximal conformation. Together, the results here suggest that the ribosome unlocks hSRP from the autoinhibited state and allows it to dynamically sample more active conformations, thus priming SRP for subsequent receptor recruitment (Fig. 7B, sampling).

In summary, our work provides a model for how a correct cargo activates the mammalian SRP for targeting to the ER membrane (Fig. 7B). In the absence of translating ribosomes, the eukaryote-specific components retain SRP in an autoinhibited conformation whereby its interaction with SR is extremely slow. Binding of the ribosome relieves SRP from this autoinhibited state and enables SRP to sample multiple alternative conformations, including the proximal state conducive to assembly with SR. The direct interaction of the ribosome with SR provides an additional mechanism that can bring SRP and SR together to facilitate their assembly (15). SRPs bound to signalless ribosomes remain largely in the sampling mode, in which SR recruitment, albeit accelerated, does not occur on a biologically relevant timescale. In contrast, the emergence of a signal sequence from the ribosome exit tunnel drives the majority of SRP into the proximal conformation in which SR recruitment can occur rapidly, thus delivering the translating ribosome to the ER membrane. Compared with the bacterial SRP, the use of signal sequence to activate SRP for receptor recruitment is a universally conserved feature of the cotranslational protein targeting pathway; in contrast, the mammalian SRP evolved the unique ability to sense and be primed by the ribosomes, which may poise the pathway for additional layers of regulation.

Materials and Methods

Materials.

All purifications were carried out at 4 °C unless otherwise specified. Recombinant hSRP19, hSRP9/14, hSRP54/hSRP54-4A10L, hSRαβΔTM, and hSRαΔX were expressed from E. coli. hSRP68/72 heterodimer was expressed from Saccharomyces cerevisiae. Proteins were purified through two to three chromatography steps including affinity chromatography, cation exchange, and size-exclusion chromatography. Details of the protein purifications are described in SI Appendix, Supplemental Methods. S7CA, a circularly permutated version of human 7SL SRP RNA for improved SRP assembly, was in vitro transcribed by T7 RNA polymerase as described previously (18). Transcribed RNA was acid phenol-extracted and purified over a denaturing polyacrylamide gel [100 mM Tris, 89 mM boric acid, 1.3 mM EDTA, 7 M urea, and 10% acrylamide (29:1)] (53). RNA extracted from the gel was dialyzed in 20 mM Tris⋅HCl (pH 7.5), flash-frozen in liquid nitrogen, and stored at −80 °C. Large-scale assembly and purification of hSRP from recombinant RNA and proteins were based on modifications of published procedures (18, 54) and are described in detail in SI Appendix, Supplemental Methods.

Engineered single cysteines on hSRP54 and hSRP19 were labeled by using thiol-maleimide chemistry as described in SI Appendix, Supplemental Methods (55). Labeled protein was separated from free dyes on a G25 column (Sigma) as described previously (55). hSRαΔX was conjugated to ATTO 647N-labeled GGGC peptide by using sortase-mediated ligation (28) as described in SI Appendix, Supplemental Methods.

Biochemical Assays.

All proteins except for SRP were centrifuged at 4 °C, 100,000 rpm in a TLA100 rotor for 30 min to remove aggregates before the assay. GTPase reactions were performed in SRP Assay Buffer [50 mM KHEPES, pH 7.5, 150 mM KOAc, 5 mM Mg(OAc)₂, 10% glycerol, 2 mM DTT, and 0.04% NIKKOL] at 25 °C. Reactions were followed and analyzed as described before (24) except that polyethylenimine-modified cellulose thin-layer chromatography was run in 1 M formic acid/0.5 M LiCl. Observed rate constants were determined as described before (24). Steady-state fluorescence measurements to analyze the SRP–SR interaction were carried out on a Fluorolog 3–22 spectrofluorometer (Jobin Yvon) at 25 °C. The binding of SRP to RNC_4A10L or to 80S was measured by MST (Nanotemper) following the manufacturer’s instructions. Details of the determination of the individual rate and equilibrium constants are described in SI Appendix, Supplemental Methods.

smFRET Measurements.

hSRP was diluted to 50–100 pM in SRP Assay Buffer containing 200 µM GTP and 1 μM 80S, 150 nM RNC_4A10L, or 150 nM RNC_2R where indicated. Based on independently determined K_d values (SI Appendix, Fig. S3), these concentrations of RNC and 80S ensure that all observed hSRP complexes were bound with the indicated partner. Samples were placed in a closed chamber made by sandwiching a perforated silicone sheet (Grace Bio-Labs) with two coverslips to prevent potential evaporation during measurements. Data were collected over 30–60 min by using an ALEX-fluorescence-aided molecular sorting setup (35, 36) with two single-photon Avalanche photodiodes (Perkin-Elmer) and 532-nm and 638-nm continuous-wave lasers (Coherent) operating at 150 µW and 70 µW, respectively. Details of single-molecule data analysis are described in SI Appendix, Supplemental Methods.