The DEAD-box RNA helicase Ded1 from yeast is associated with the signal recognition particle (SRP), and its enzymatic activity is regulated by SRP21

The DEAD-box RNA helicase Ded1 is an essential yeast protein involved in translation initiation. It belongs to the DDX3 subfamily of proteins implicated in developmental and cell-cycle regulation. In vitro, the purified Ded1 protein is an ATP-dependent RNA binding protein and an RNA-dependent ATPase, but it lacks RNA substrate specificity and enzymatic regulation. Here we demonstrate by yeast genetics, in situ localization and in vitro biochemical approaches that Ded1 is associated with, and regulated by, the signal recognition particle (SRP), which is a universally conserved ribonucleoprotein complex required for the co-translational translocation of polypeptides into the endoplasmic reticulum lumen and membrane. Ded1 is physically associated with SRP components in vivo and in vitro. Ded1 is genetically linked with SRP proteins. Finally, the enzymatic activity of Ded1 is inhibited by SRP21 with SCR1 RNA. We propose a model where Ded1 actively participates in the translocation of proteins during translation. Our results open a new comprehension of the cellular role of Ded1 during translation.


INTRODUCTION 48
The DEAD-box family of RNA helicases are ubiquitous proteins found in all three kingdoms 49 of life, and they are implicated in all processes involving RNA, from transcription, splicing, 50 ribosomal biogenesis, RNA export, translation to RNA decay [reviewed by (1-3)]. They 51 belong to the DExD/H superfamily 2 (SF2) of putative RNA and DNA helicases that contain 52 catalytic cores consisting of two, linked, RecA-like domains containing conserved motifs 53 associated with ligand binding and NTPase activity, where the majority of the proteins are 54 ATPases. In addition, they often contain highly variable amino-and carboxyl-terminal 55 domains [reviewed by (4,5)]. The DEAD-box proteins are ATP-dependent RNA binding 56 proteins and RNA-dependent ATPases that have been shown to remodel RNA and 57 performed RT-PCR on the samples using oligonucleotides specific for the three RNA. Both 158 SCR1 and PGK1 RNAs were amplified much more in the fractions pulled down with  specific IgG than in the control fractions that were pulled down with pre-immune IgG ( Figure  160 2A). In contrast, RPL20B mRNA was weakly amplified in both cases. Hence, Ded1 161 associated with SCR1 RNA in vivo. 162 It was possible that Ded1 interacted with the SCR1 RNA independently of the SRP 163 proteins. To test this, we did Ded1-IgG pull-down experiments, SDS-PAGE separation and 164 Western blot analyses of the recovered proteins. However, we were only able to obtain 165 antibodies against Sec65 [generously provided by Martin R. Pool;(44)]. (We later made IgG 166 against SRP21.) Hence, we cloned all the SRP genes with amino-terminal HA tags, and we 167 did Ded1-IgG pull-downs with strains independently expressing each tagged protein ( Figure  168 2B). We recovered a significant amount of SRP14, SRP54, Sec65 and SRP101. Likewise, we 169 digested the Ded1-IgG-bound complexes with RNase A prior to elution to determine if these 170 complexes depended on SCR1 RNA; in all cases, the signals for the SRP proteins were 171 reduced, but the signals were still more than for the pre-immune-IgG control ( Figure 2B). 172 Oddly, we detected little HA-tagged SRP21 even though it was prominent in our previous 173 mass spectrometry analyses [Tables 1 and 2;(13)]. Moreover, we detected little HA-SRP21 in 174 yeast extracts even though it is of similar size and has similar expression levels as SRP14 175 ( Figure 2B; Table 3). It was possible that the HA tag increased the proteolytic degradation of 176 the protein during extraction or that the HA tag by itself was proteolytically removed. Thus, 177 these results showed that Ded1 interacted with complexes containing the SRP proteins, and 178 that these complexes were stabilized by SCR1 RNA ( Figure 2B). Notably, the bound 179 complexes also contained the SR receptor SRP101, which binds with SRP complexes 180 associated with the ER during translation (26,35). We did not detect SRP102, but it contains 181 an integral membrane domain that could limit its recovery. Thus, Ded1 physically associated 182 with the SRP complex in vivo. 183

Ded1 cosedimented with SRP factors 184
In our previous work, we found that Ded1 migrated at a position corresponding to ~26S on 185 sucrose gradients, but the Ded1-containing fractions were only partially resolved from the 186 protein peak at the top of the gradient (13). Hence, we modified the previous sucrose gradient 187 conditions to better resolve the different complexes. In both cases, we used 5 mM MgCl 2 188 because Ded1 was found to dissociate from higher molecular weight complexes at higher 189 Mg 2+ concentrations and sediment within the protein peak at the very top of gradient. We did 190 sucrose gradients of yeast strains independently expressing each  The polysome profile ( Figure 3A) showed well resolved ribosome peaks that 193 corresponded to the expected distribution of ribosomal RNAs ( Figure 3B). Northern blot 194 analyses with a SCR1-specific 32 P-labeled probe showed that the vast majority of the SCR1 195 RNA migrated as a narrow peak centered at fraction 4 (arrow), but RNA was detected It is known that constitutive over-expression of the SRP proteins can cause them to 205 accumulate in the nucleus, which may affect their distribution on the sucrose gradients (45). 206 Consequently, we did sucrose gradients of the endogenously-expressed Sec65 and probed the 207 membranes with Sec65-specific IgG; it showed a similar distribution as the HA-tagged 208 protein, but it was present as a doublet ( Figure 3E). Likewise, we made SRP21-specific IgG 209 to detect endogenous SRP21 in the gradients ( Figure 3F). The vast majority of the protein was 210 stuck in the well of the gel or migrated only a short distance into the gel that indicated that 211 SRP21 formed large, partially insoluble aggregates. We obtained similar results with 212 recombinant SRP21 in the presence of RNA (see below section: SRP21 did not block Ded1 213 binding to SCR1). Nevertheless, the majority of both Sec65 and SRP21 sedimented at the 214 position corresponding to the bulk of the SCR1 RNA in the gradients. 215 The data were consistent with Ded1 interacting with the SRP complex. However, the 216 vast majority of the material was not associated with translating ribosomes, although there 217 was a smaller peak of Ded1 and most of the SRP proteins in fraction 12 that corresponded to 218 the 80S complex (arrow). These results indicated either that most of Ded1 and the SRP 219 complex were not actively involved in translation or that the complexes were not stably 220 associated with the ribosomes under the conditions used. Thus, it was unclear as to the 221 functional role of the Ded1-SRP interactions. 222

Ded1 was genetically linked to SRP proteins 223
We next tested to see if there was a genetic link between Ded1 and the SRP proteins as we 224 previously demonstrated for the nuclear and cytoplasmic cap-associated proteins (13). We 225 used the same cold-sensitive mutant F162C of Ded1 in the ded1::HIS3 deletion strain and 226 over-expressed the SRP proteins and SCR1 RNA from the pMW292 and pM299 plasmids 227 (44,46). Liquid cultures were serially diluted and spotted on 5-FOA plates that were incubated 228 at 18°C, 30°C and 36°C ( Figure 4). The results showed a slight enhancement of growth at 229 18°C that was consistent with a genetic interaction between Ded1 and the SRP complex, but 230 the signal was too weak to demonstrate a clear link. The weak multicopy suppression was not 231 unexpected because Ded1 is implicated in the expression of multiple mRNAs that are not 232 associated with SRP complexes (14,43); expression of these mRNAs would be insensitive to 233 the over-expressed SRP factors. Thus, it was possible that the SRP complex would be more 234 sensitive to the level of Ded1 expression than vice versa. 235 Previous work has shown that loss of any SRP component leads to a slow-growth 236 phenotype (47-49), although yeast cells are eventually able to adapt to this loss (50,51). Thus, 237 we obtained yeast strains with the DED1, SRP14,SRP21,SEC65,SRP68,SRP72 and SRP101 238 genes under the control of a tetracycline-regulated promoter that could be suppressed with 239 doxycycline (52). Unfortunately, SRP54 under the tetracycline promoter was not available. 240 Cultures of the different strains were grown in liquid culture, serially diluted and spotted on 241 agar plates in the presence or absence of 10 µg/ml of doxycycline. All the strains except 242 SRP101 showed reduced growth with the constitutive expression of the proteins, which was 243 most apparent at 18°C ( Figure 5). In the presence of doxycycline, all the strains show strongly 244 reduced growth except SRP21, Sec65 and SRP101, which showed a slight reduction. Western 245 blot analysis of liquid cultures of TET-SRP21 and TET-SEC65 grown for up to 24 h in the 246 presence of 10 µg/ml of doxycycline showed no diminution in protein level when probed with 247 SRP21-IgG or Sec65-IgG, respectively (data not shown). This indicated that either the TET 248 promoter was not completely shut down with doxycycline in these strains or that the proteins 249 were particularly stable. Interestingly, TET-SRP21 actually grew slightly better in the 250 presence of doxycyline, which further indicated that constitutive expression of the proteins 251 was detrimental ( Figure 5). Constitutive and overexpression of Ded1 was previously shown to 252 inhibit cell growth (53,54). 253 We transformed the different TET strains with a plasmid containing DED1 under 254 control of the very strong GPD promoter and compared it with cells transformed with the 255 empty plasmid and with wildtype yeast cells (55). We likewise transformed the cells with a 256 plasmid expressing the mutant Ded1-F162C protein that had reduced ATP binding and 257 enzymatic activity (46). As expected, the yeast stains TET-SRP21 and TET-SEC65, which 258 continued to express the SRP proteins, showed reduced growth on the plates due to the 259 inhibitory effects of the overexpressed Ded1 ( Figure 6). In contrast, SRP14, SRP68 and 260 SRP72 showed enhanced growth despite the inhibitory effects of Ded1 ( Figure 6). The Ded1-261 F162C mutant showed little or no stimulatory effect, which indicated that the enzymatic 262 function of Ded1 was important for the enhanced growth. Thus, high expression of Ded1 263 partially suppressed the slow-growth phenotype of strains depleted for SRP proteins, and this 264 result established a genetic link between Ded1 and the SRP complex. 265

Ded1 was in cellular foci associated with the endoplasmic reticulum 266
The next question we asked was whether Ded1 co-localized with the ER as would be 267 expected if it was associated with SRP-ribosome complexes that were translating mRNAs 268 encoding polypeptides translocated into the ER. However, we and others have shown that 269 Ded1 has a diffuse location within the cytoplasm under normal growth conditions (13,53,56). 270 Nevertheless, some of the Ded1 protein is sequestered with translation-inactive mRNAs in 271 cellular foci when the translation conditions are altered (53,57,58). We reasoned that if 272 polypeptide import into the ER was transiently blocked then Ded1 would form foci associated 273 with the ER. We used temperature-sensitive (ts) mutants of the Sec61 and Sec62 proteins that 274 form the translocon pore in the ER for the import of SRP-dependent polypeptides during 275 translation (59,60). At the non-permissive temperature, these mutants block or disrupt the 276 Sec61 channel and ER-associated translation is terminated. As a marker, we used the 277 integrated red-fluorescent-tagged amino-terminal domain of Kar2 fused to the HDEL ER 278 retention signal (YIPlac204TKC-DsRed-Express2-HDEL; Addgene, Watertown, MA) in the 279 two sec strains; Kar2 is an ATPase that functions as a protein chaperone for refolding proteins 280 within the lumen of the ER [(61) and reference therein], and consequently the Kar2 chimera 281 serves as a marker of the ER lumen. We also used an ATPase-inactive Ded1-E307Q mutant 282 (Ded1-DQAD) that has a high propensity to form cellular foci with sequestered mRNAs that 283 are no longer undergoing translation. 284 We first looked at the distribution of proteins under permissive conditions ( Figure  285 7A). The distribution of Kar2-RFP around the nuclear envelope (central cisternal ER), as 286 interconnected tubules (tubular ER) and as a cortical halo inside the plasma membrane of the 287 cell wall (PM-associated ER) was consistent with the locations of the ER in yeast; actively 288 translating ribosomes are associated with all these ERs (62). Ded1-DQAD-GFP was 289 uniformly distributed in the cytoplasm, largely excluded from the nucleus, and it formed 290 occasional foci that were distributed at various positions in the cytoplasm ( Figure 7A). Both 291 the sec61-ts and sec62-ts strains showed equivalent phenotypes. The intensity of the 292 fluorescence signals of both Kar2-RFP and Ded1-DQAD-GFP was highly variable between 293 cells, which probably reflected different levels of protein expression between cells. 294 At 37°C, Kar2-RFP showed a similar cellular distribution as at 24°C for both sec61-ts 295 and sec62-ts mutants, although it showed an increased frequency of aggregates with the ER 296 ( Figure 7B). In contrast, Ded1-DQAD-GFP showed a pronounced increase in the number of 297 foci that were highly variable in size ( Figure 7B). Many of these foci were closely associated 298 with Kar2-RFP, particularly as a chain of foci on the cytoplasmic side of the ER around the 299 plasma membrane, where the PM-associated ER was expected to be located, and as a chain of 300 foci corresponding to tubular ER (arrowheads, Figure 7B). Both the sec62-ts and sec61-ts 301 strains showed similar phenotypes. In some cases, the Kar2-RFP aggregates and Ded1-302 DQAD-GFP foci were near each other. Thus, Ded1 was associated with mRNAs that were no 303 longer undergoing translation in close proximity to the ER at the non-permissive temperature. 304 This result was consistent with previous work that showed that Ded1 is recovered with 305 membrane-associated ribosomal-protein complexes (63). 306 We next asked if the SRP proteins showed similar properties. We used GFP-tagged 307 SRP14 and SRP21 proteins that were expressed off the chromosome (GFP bank, Thermo 308 Fisher Scientific, Waltham, MA) and the plasmid-encoded Ded1-DQAD-mCh mutant. The 309 SRP14-GFP showed a weak but uniform signal in all the cells, where the protein was 310 concentrated on the ER ( Figure 7C). SRP21-GFP showed a similar profile (data not shown). 311 In contrast, the plasmid-expressed Ded1-DQAD-mCh showed highly variable expression. We 312 used cells grown under wildtype conditions or depleted for glucose, which promoted the 313 formation of cellular foci. However, depending on the cellular growth we obtained a 314 significant number of foci associated with the ER even under wildtype growth (arrowheads, 315 Figure 7C). Thus, Ded1 was in close proximity to both the ER and the SRP proteins in the 316

cell. 317
Overexpressed SRP proteins accumulated in the nucleus and nucleolus 318 The biogenesis and metabolic pathway of the SRP RNP is complex, and it involves a large 319 number of different steps [reviewed by (32,64,65)]. In yeast, the SRP proteins SRP14, SRP21, 320 SRP68 and SRP72 are assembled on the SCR1 RNA probably in the nucleolus. Sec65 is in 321 the nucleus, but there is some ambiguity about whether it accumulates in the nucleolus as 322 well, although the equivalent mammalian SRP19 protein is found there (32,45,66). The 323 partially assembled SRP complex is then exported to the cytoplasm through the XpoI/CrmI 324 nuclear pore complex whereupon it binds with SRP54, which subsequently associates with 325 the signal sequence of the partially translated polypeptide and causes the SRP to assemble on 326 the 80S ribosomes. We previously showed that Ded1 actively shuttles between the nucleus 327 and cytoplasm using the XpoI and Mex67 nuclear pores (13). Thus, it was possible that Ded1 328 associated very early with the SRP complex within the nucleus, and that it was important for 329 the biogenesis or export of the complex. 330 The XpoI nuclear pore is known to export multiple cargoes, including ribosomal 331 subunits, certain small nuclear RNAs, some viral RNAs and the assembled SRP complex 332 (45,66-68). We used a yeast strain with a mutant xpoI allele that is sensitive to the bacterial 333 toxin leptomycin b from Streptomyces to test whether Ded1 was involved in the XpoI-334 dependent export of the SRP (69); this strain contains a single mutation (XpoI-T539C) that 335 makes the yeast protein sensitive to the drug (70). We previously showed that both the Mex67 336 and XpoI nuclear pore complexes must be disrupted to see a significant accumulation of Ded1 337 in the nucleus (13). We transformed this strain with plasmids expressing Ded1-mCh and with 338 plasmids expressing either SRP14-GFP or SRP21-GFP, and we determined the locations of 339 the tagged proteins. 340 The plasmid-encoded SRP14-GFP had highly variable expression between cells, but it 341 showed a strong nuclear location that was often concentrated in crescent-shaped regions even 342 in the absence of leptomycin b ( Figure 7D, Figure 8A & 8B). In contrast, SRP21-GFP 343 showed a diffuse location throughout the nucleus, which indicated that the overexpressed 344 protein was not able to assemble or accumulate in the nucleolus ( Figure 8C & 8D). However, 345 it occasionally formed nuclear foci (arrowheads, insert Figure 8C). The expression of the 346 plasmid-encoded Ded1-DQAD-mCh was likewise highly variable, and it was largely 347 excluded from the nucleus even in the presence of leptomycin b ( Figure 7D, Figure 8). 348 However, in some instances with leptomycin b, where Ded1-DQAD-mCh was lightly 349 expressed, we found that the protein accumulated in crescent-shaped regions with SRP14-350 GFP (arrowheads, insert Figure 7D). The Ded1-DQAD mutant binds RNA with a high 351 affinity in the presence of ATP but it can not hydrolyze the ATP to recycle the complex. 352 Thus, Ded1 could co-localize with the SRP complex in the nucleolus but only under 353 conditions where Ded1 export was blocked. The absence of Ded1-DQAD in the nucleus and 354 crescents in the absence of leptomycin argued that Ded1 was not needed for SRP assembly 355 and export, but the data could not rule out this possibility. 356

Ded1 physically interacted with SRP factors 357
Our data indicated that Ded1 could bind SCR1, associate with the SRP complex and co-358 localize with the ER. Moreover, we obtained a genetic link between Ded1 and SRP proteins. 359 However, it was unclear whether Ded1 physically interacted with the SRP proteins or 360 indirectly through the SCR1 RNA ( Figure 1). The metazoan SRP14 and SRP9 are known to 361 bind 7SL in the Alu domain, SRP54 and SRP19 bind helices 6 and 8,and SRP68 and SRP72 362 bind around the junction between helices 5e, 5f, 6, 7 and 8 (27,35). However, yeast SRP14 is 363 thought to bind the Alu domain as a homodimer and the role of SRP21 to date is largely 364 speculative, although it is considered a structural homolog of SRP9 (36,38). Yeast Sec65 365 serves a similar role as SRP19, but it is considerably larger (47,49,71). The SRP bound on the 366 80S ribosomes shows an extended structure where the S-domain interacts with the exit 367 channel containing the signal peptide and the Alu domain interacts with the entry region of 368 the mRNA (37,39,40). However, SCR1 contains two hinge regions ( Figure 1); it was possible 369 that the SCR1 RNA was folded upon itself in its free form and that this brought the different 370 regions of the SRP in close proximity. Thus Ded1 might interact with multiple different SRP 371

proteins. 372
To test this, we subcloned the genes encoding the different proteins into pET22 and 373 pET19 plasmids and then purified the recombinant proteins expressed in Escherichia coli on 374 nickel-agarose columns. We then incubated the purified individual proteins or combination 375 therein with purified Ded1, recovered the complexes with Ded1-IgG-Protein-A-Sepharose 376 beads and separated the recovered proteins by SDS-PAGE ( Figure 9). 377 The results showed that Ded1 formed stable interactions with SRP14 and Sec65. 378 Unfortunately, SRP68, SRP72 and SRP101 migrated at positions on the SDS-PAGE that 379 overlapped with Ded1; hence we were not able to obtain unambiguous results, but they 380 appeared to have little or no affinity for Ded1. SRP54 did not stably associate with Ded1 by 381 itself. In contrast, SRP21 was not consistently recovered with Ded1, which indicated that it 382 formed weak interactions with the protein (Figure 9). However, SRP21 was consistently 383 recovered with Ded1 in the presence of SRP14, which was consistent with SRP14 and SRP21 384 forming a stable heterocomplex as previously proposed (38). Likewise, SRP54 was recovered 385 with Ded1 in the presence of Sec65; this was consistent with the two proteins being in close 386 proximity on the S domain of SCR1 (47,49,71). Moreover, all four SRP proteins were 387 recovered with Ded1 when incubated together. 388 The weak interactions between SRP21 and Ded1 were primarily through the amino-389 terminal domain because deleting the 73 carboxyl-terminal amino acids (SRP21∆73) did not 390 eliminate this affinity ( Figure 9B). Finally, we recovered Ded1 in pull-down experiments with 391 SRP21-specific IgG ( Figure 9B). Thus, Ded1 was capable of forming protein-protein 392 interactions with the SRP proteins in the absence of SCR1 RNA. Nevertheless, the presence 393 of SCR1 RNA enhanced the recovery of all the proteins (see section: Ded1 associated with 394

SRP21 inhibited the SCR1-dependent ATPase activity of Ded1 396
Ded1 is an RNA-dependent ATPase. We previously showed that the nuclear and cytoplasmic 397 cap-associated factors would stimulate the ATPase activity of Ded1 in the presence of RNA 398 (13). We wondered whether the SRP proteins would alter the enzymatic activity of Ded1 as 399 well and whether it would be preferential for the SCR1 RNA, which would be the authentic 400 substrate for the assembly of the SRP proteins. We tested this with an in vitro, T7-401 polymerase-transcribed SCR1 RNA that was equivalent to the endogenous SCR1 except that 402 the 5' terminal nucleotide was replaced with a guanosine to facilitate transcription. As a 403 control we used a fragment of the actin pre-mRNA precursor containing short exon sequences 404 and the entire intron. In addition, the actin transcript was of similar size to SCR1 (605 nts and 405 552 nts, respectively). 406 Many of the SRP proteins at nearly a 30-fold excess over Ded1 inhibited the RNA-407 dependent ATPase activity of Ded1 somewhat for both actin and SCR1, although SRP54 408 seemed to enhance the activity slightly, especially with actin ( Figure 10A). This was not 409 unexpected because the SRP proteins were largely basic and positively charged under the 410 reaction conditions (pH 7.5; Table 3); the proteins would be expected to nonspecifically 411 associate with the RNAs and thereby reduce the effective concentration of the RNAs 412 accessible to Ded1. SRP21 showed a much stronger inhibition, especially with SCR1, but it 413 was the most basic (pK i = 11.14) of the SRP proteins. To further elucidate the nature of the 414 inhibition, we compared the ATPase activity of Ded1 with SCR1 and actin RNAs with 415 SRP21. SRP21 is considered the structural homolog of SRP9, and in yeast it probably forms a 416 complex with a homodimer of SRP14 that binds the Alu domain of SCR1 (36,38). Thus we 417 tested to see if SRP14 would enhance the inhibitory effects of SRP21. 418 Equimolar concentrations of both the actin precursor and SCR1 stimulated the ATPase 419 activity of Ded1, but the stimulation was not equivalent. Moreover, there was variability in 420 the stimulatory effects with different RNA preparations, which probably reflected variability 421 in the folding of the RNAs during preparation. Indeed, others have shown that the smaller 422 human 7SL RNA is difficult to recover as a homogeneous structure in vitro (41). Thus, to 423 facilitate comparisons we normalized the activity relative to that of Ded1 with the RNA alone 424 and used the same RNA preparations for comparisons. In addition, we used 8.5-fold less of 425 the SRP proteins to emphasize the differences. 426 SRP14 may have slightly inhibited Ded1 with SCR1 but it had little affect with actin 427 ( Figure 10B). In contrast, SRP21 inhibited the ATPase activity of Ded1 with SCR1 by about 428 75% but only by 25% with actin. This indicated that there was some nonspecific inhibition, 429 but that the strongest inhibition was obtained with the authentic substrate of SRP21. Addition 430 of SRP14 reduced the activity of Ded1 in the presence of SRP21 by an additional 8% for 431 SCR1 but showed no additional reduction with actin. Thus, SRP14 enhanced the SRP21-432 dependent inhibition of Ded1 with SCR1 RNA. Addition of the other SRP proteins to SCR1 433 further reduced the activity by about 4%. None of the purified SRP proteins showed any 434 intrinsic ATPase activity in the absence of Ded1, and none of the SRP proteins affected the 435 ATPase activity of Ded1 in the absence of RNA ( Figure 10B and data not shown). 436 The data indicated that SRP21 in the presence of its authentic substrate was the most 437 effective at inhibiting Ded1. Thus, it was likely that SRP21 bound to the Alu domain of SCR1 438 formed the most effective inhibitory structure. We tested this with deletions of the S domain 439 (SCR1∆S1) and Alu domain (SCR1∆Alu). Previous work has shown that the folding of 440 mammalian 7SL RNA is difficult, and it needs a temperature step for refolding involving 441 slow cooling in the presence of monovalent cations; moreover, the assembly of the SRP 442 proteins is complicated (41). The yeast SCR1 RNA is about two-fold larger with a number of 443 additional hairpins, so we anticipated difficulties in obtaining a functional homogenous 444 structure ( Figure 1). Thus, we assayed various permutations of pre-incubating the RNA with 445 the various proteins prior to adding the ATP, but they all yielded similar results. Ded1 was 446 significantly less active with both SCR1∆S1 and SCR1∆Alu then full-length SCR1 at 447 equimolar concentrations of RNA, but the RNAs were 77% and 58%, respectively, of the size 448 of SCR1 ( Figure 11A). This may account for the reduced ATPase activity, but it was possible 449 that Ded1 was activated by specific structures within the SCR1 RNA that were absent or 450 misfolded in the deletions. SRP21 inhibited the ATPase activity for SCR1 and to a lesser 451 extent actin, but it had little inhibitory affect on the SCR1 deletions, which was consistent 452 with a structure-dependent inhibition. We also tested a carboxyl-terminal deletion of SRP21 453 (SRP21∆73) that lacked the amino acids that did not correspond to those of mammalian SRP9 454 (38); it showed significantly less inhibition, which was consistent with it playing a role in the 455 SRP21 interactions with SCR1 and Ded1 ( Figure 11A). 456 The ATPase activity of Ded1 is stimulated by various RNAs containing single-457 stranded regions, but it is most activated by poly(A)-containing RNAs (72). We repeated the 458 ATPase assays with purified yeast RNA. We needed to use 0.12-0.14 µg/µl of yeast RNA to 459 obtain similar levels of activation of Ded1 as 23 nM of SCR1 (0.0039 µg/µl) or actin (0.0044 460 µg/µl; Figure 11B). Thus, yeast RNA was ~30-fold less effective at stimulating the activity. 461 SRP21 and SRP14 showed no inhibitory effects, and they may have actually stimulated the 462 ATPase activity of Ded1 somewhat, perhaps by acting as RNA chaperones to increase the 463 accessibility of the RNA to Ded1. However, the yeast RNA was a heterogenous mix that may 464 have had both activating and inhibitory RNAs. Thus, we repeated these experiments with 465 yeast tRNAs and poly(A) RNA at 0.12 µg/µl ( Figure 11C). Both RNAs needed ~30-fold 466 higher concentration than for SCR1 to stimulate the ATPase activity of Ded1 to similar levels, 467 but SRP21 had little affect on the activities. Thus, Ded1 and SRP21 preferentially bind RNAs 468 with certain sequences or structural features. 469

SRP21 did not block Ded1 binding to SCR1 470
The previous results indicated that Ded1 was either blocked from binding the SCR1 RNA or 471 that its ATPase activity was inhibited by protein-protein contacts with SRP21 bound on the 472 RNA. To test this, we did electrophoretic mobility shift assays (EMSA) with the different 473 proteins and RNAs. Our previous work showed strong, concentration-dependent binding of 474 Ded1 to short oligonucleotides in the presence of AMP-PNP with a K 1/2 of ~40 nM and weak 475 binding in the presence of ADP or in the absence of a nucleotide (73). We repeated these 476 experiments with the longer RNAs, but we separated the products on agarose gels containing 477 ethidium bromide. Similar results were obtained when the gels were run in the absence of 478 ethidium bromide, which was then soaked into the gels after electrophoresis (data not shown). 479 A 5-to 10-fold excess of Ded1 was able to displace the majority of both SCR1 and 480 actin ( Figure 12A). SCR1 typically migrated as a distinct band but actin often showed more 481 heterogeneity, which probably reflected more profound conformational heterogeneity. This 482 varied somewhat between RNA preparations. In contrast, SRP21 preferentially bound SCR1 483 RNA over actin ( Figure 12B). Moreover, it seemed to form large molecular-weight 484 aggregates that only partially migrated into the gels. Deleting the carboxyl-terminal sequences 485 of SRP21 (SRP21∆73) largely eliminated the binding affinity, indicating that these sequences 486 were either important for binding or for maintaining the correct conformation of the protein 487 ( Figure 12C). 488 We next asked what affect SRP21 would have on Ded1 binding. The results showed 489 that SRP21 had little affect on Ded1 binding, but the retarded bands tended to migrate as 490 higher molecular-weight complexes in the presence of SRP21 for SCR1 ( Figure 13A). We 491 previously showed that the carboxyl-terminal domains of DEAD-box proteins, including 492 Ded1, are important for high affinity binding to RNAs (54). Consistent with this, deleting 78 493 amino acids from the carboxyl terminus of Ded1 largely eliminated RNA binding ( Figure  494 13B). However, addition of SRP21 had little affect on Ded1 binding even though a small 495 amount of material was sequestered near the origin of the gel ( Figure 13C). Instead, Ded1 496 seemed to actually reduce the binding affinity of SRP21 for the RNA, and this was true for 497 the carboxyl-terminal deletion of Ded1 as well ( Figure 13C). Thus, although SRP21 498 modulated the ATPase activity of Ded1, Ded1 seemed to modulate SRP21 binding, perhaps 499 through interactions with the amino-terminal domain of Ded1 or the RecA-like core. 500 Finally, we asked what structural features of SCR1 were recognized by the proteins. 501 These experiments were more ambiguous because the binding site of SRP21 is largely 502 unknown and because there was no guarantee that the RNAs deletions would fold into the 503 anticipated conformations. The results showed that both Ded1 and SRP21 were able to bind 504 the SCR1 RNAs deleted for the Alu and S domains ( Figure 14). Thus, SRP21 probably 505 recognized multiple features of the SCR1 RNA. Moreover, the strong binding of Ded1 to the 506 deleted SCR1 RNAs did not correlate with the reduced ATPase activities of Ded1 with these 507 RNAs ( Figure 11A). Therefore, either Ded1 could bind the RNAs in a nonproductive form or 508 the ATPase activity of Ded1 was modulated by the different structures. 509

DISCUSSION 510
Our experiments show that Ded1 is an SRP-associated factor. It physically interacts with the 511 SCR1 RNA and many of the SRP proteins both in vitro and in vivo. It is genetically linked to 512 these proteins, and it co-sediments with the SRP factors in sucrose gradients. The RNA-513 dependent ATPase activity of Ded1 is inhibited by SRP21 and this inhibition is much more 514 pronounced in the presence of SCR1 RNA, the authentic substrate of SRP21. Although there 515 is probably conformational heterogeneity of the RNAs, SRP21 preferentially binds SCR1 516 RNA over actin RNA, which indicates that it contains or forms the necessary elements for 517 high-affinity SRP21 binding. Likewise, the ATPase activity of Ded1 is preferentially 518 activated by SCR1 and actin RNAs over an equivalent concentration whole yeast RNA, tRNA 519 or poly(A) RNA, which indicates that it is recognizing specific features or structures of these 520 RNAs. The nature of these features or structures is unclear. Finally, Ded1 co-localizes in 521 cellular foci with the ER-associated mRNAs, and it occasionally co-localizes with SRP 522 proteins in the nucleolus. 523 The role of SRP21 to date is largely unclear. It is considered as the structural homolog 524 of metazoan SRP9, which forms a heterodimer with SRP14 on the Alu domains of 7SL RNA, 525 even though there is little or no sequence homology (38). The amino-terminal residues of 526 SRP21 are capable of forming similar structural features as SRP9, but it is over 80% bigger; 527 the carboxyl-terminal sequences are thought to compensate for the abbreviated Alu domain of 528 yeast SCR1, which lacks the characteristic hairpins H3 and H4 (38). Moreover, yeast SCR1 is 529 about 75% bigger than metazoan 7SL, and it contains additional structures between the Alu 530 and S domains (33,34). SRP21 may be needed to stabilize or form the correct conformation of 531 the SCR1 RNA, and thus it may need to recognize multiple structures or features of the RNA. 532 Consistent with this, SRP21 binds full-length SCR1 RNA and the deletions in vitro with 533 similar affinity. In contrast, it has weak affinity for the actin RNA. The carboxyl-terminal 534 sequences of SRP21 are important for this affinity. This is consistent with SRP21 forming a 535 complex with the SRP14 homodimer as previously proposed (38). 536 SRP21 inhibits the RNA-dependent ATPase activity of Ded1, but it is much more 537 effective in the presence of SCR1 RNA than actin RNA. In contrast, Ded1 binds SCR1 and 538 actin RNAs with similar affinities and it is activated to similar extents. Under these 539 circumstances, one would expect SRP21 to reduce Ded1 binding to SCR1 but not to actin 540 because SRP21 would reduce the number of potential binding sites for Ded1 on SCR1. But 541 this is not the case, and if anything SRP21 seems to enhance Ded1 binding to SCR1 slightly. 542 The inhibition is due to protein-protein contacts, but SRP21 is less stably associated with 543 Ded1 in the absence of SRP14 or SCR1 RNA. Thus, SRP21 probably forms a specific 544 inhibitory structure with Ded1 in the presence of SCR1 RNA. This is consistent with a 545 functional regulation of the ATPase activity of Ded1 in the context of the SRP complex. 546 Ded1 is an ATP-dependent RNA binding protein, and it is capable of forming long-547 lived complexes with RNA in the presence of a nonhydrolyzable analog of ATP in vitro (8). 548 Ded1 is considered a translation-initiation factor [(14,15) and references therein], but 549 crosslinking studies on DDX3 show most of the interactions on the open reading frames of a 550 subset of the mRNAs (74). We obtained similar results with Ded1 (data not shown). Thus, 551 Ded1 remains associated with the ribosomes during translation elongation, which can be seen 552 in polysome profiles as well [this work and (13)]. Ded1 likewise is found with membrane-553 associated ribosomes (63). Thus, Ded1 may play important roles in translation elongation as 554 well as in initiation-including membrane-associated translation. 555 Ded1 is associated with 90S ribosomal precursors, which may indicate a role of Ded1 556 in SRP assembly in the nucleolus (75). We do see occasional co-localization of the Ded1-557 DQAD mutant with overexpressed SRP14 in crescent-shaped structures in the nucleus that 558 are consistent with this possibility, but SRP21 has a diffuse location within the nucleus, and it 559 is never seen concentrated in the crescent-shaped structures. Thus, it may associate with the 560 SRP complex outside the nucleolus or it may be transiently located within the nucleolus, as 561 has been proposed for Sec65 (45). Thus, we can not rule out a role for Ded1 in the biogenesis 562 of the SRP complex in the nucleus that is regulated by SRP21. Under these circumstances, 563 Ded1 may associate early with the SRP complex and remain attached even when the complex 564 binds the 80S ribosomes. This would provide a possible mechanism by which ER-specific 565 mRNAs are selected for translocation on the ER. Interestingly, DDX3 crosslinks to 7SL RNA 566 as well (74). Thus, although metazoans lack a clear equivalent to SRP21, DDX3 may also be 567 intimately connected to SRP-dependent translation. 568 On the basis of these observations, we propose the following model for the role of 569 Ded1 (and DDX3-like proteins) in membrane-associated translation. Ded1 interacts with cap-570 associated factors and with Pab2 bound on the 3' poly(A) tail of the mRNA (not shown). The 571 3' UTR is considered important for SRP-dependent targeting of mRNAs [reviewed by (76)]; 572 but the SRP is not known to directly interact with the mRNA, and it may interact through 573 another RNA binding protein (77). Ded1 (and DDX3) could serve this role as it interacts with 574 both 5' and 3' components of the mRNA [(13,16) and references therein]. Ded1 remains 575 attached to the mRNA during scanning by the 43S ribosomes, assembly of the 48S ribosomes 576 and eventual formation of the 80S ribosomes at the AUG start codon ( Figure 15A). This is 577 consistent with crosslinking experiments of ribosomal RNA that show Ded1 near the mRNA 578 entry channel (43). The RNA-dependent ATPase activity of Ded1 is uninhibited, and it is able 579 to translocate on the mRNA with the ribosomes through rapid cycling between the "open" and 580 "closed" conformations, and it may further stabilize the ribosome-mRNA complex during 581 scanning, assembly and translation (14) In the next step, the signal peptide binds in the hydrophobic groove of the GTPase 591 SRP54 that subsequently causes conformational changes of the SRP and its interactions with 592 the ribosome [(40,78,80) and references therein]. SRP14 bound on the Alu domain of SCR1 593 binds at the GTPase center located at the 40S-60S interface and thereby transiently blocks the 594 GTPase elongation factor eEF2 from binding (37,39,40). At the same time, SRP21 binds to 595 Ded1 and inactivates its ATPase activity. This results in Ded1 maintaining a closed 596 conformation that has high affinity for the RNA but that also crimps the RNA bound on RecA 597 domain 1; this prevents both Ded1 and the ribosomes from sliding on the mRNA by 598 effectively clamping the mRNA ( Figure 15B). Another factor than SRP9 may play this role in 599 metazoans. The SRP complex undergoes conformational changes during this time, and Ded1 600 may also bind the SCR1 RNA through its carboxyl terminus to facilitate the subsequent 601 interactions of SRP14 with the ribosomes. 602 The absence of DDX3-like RNA helicases in in vitro reconstituted systems might 603 explain why this pausing is often short or absent (81). Ded1 may also stabilize the paused 604 ribosomes to prevent premature termination or frameshifting. The paused ribosomes then 605 associate with the peripheral-membrane GTPase SRP101 and the integral-membrane protein 606 SRP102 that form the SRP receptor (SR) complex ( Figure 15C). Once associated with the 607 Sec61 translocon, the SRP complex undergoes further conformational changes and is either 608 released from the ribosome or assumes an inactivated form on the ribosomes. The ATPase 609 activity of Ded1 is restored and translation can resume ( Figure 15D). 610 Ribosome pausing events are important for other translational events in addition to 611 SRP-dependent protein translocation. For example, pausing is associated with co-translational 612 protein folding, protein targeting, mRNA and protein quality control, and with co-613 translational mRNA decay [reviewed by (82)]. Ded1 (and DDX3) may be intimately 614 associated with these events by a similar mechanism but with other associated factors besides 615 the SRP proteins. Likewise, ribosome pausing is associated with frameshifting events 616 Finally, we note that the bacterial polypeptide-translocase SecA is a superfamily 2 624 "RNA helicase" that has a RecA-like core structure that is very similar to the DEAD-box 625 proteins (89). It is intimately associated with the SecYEG translocon, and it uses ATP to drive 626 post-translational polypeptides through the pore into the periplasm [reviewed by (26) The sequence encoding the SCR1 RNA was amplified with an oligonucleotide 664 containing a 5' BamHI restriction site and the T7 promoter, and an oligonucleotide containing The SRP genes (SRP14, SRP21, SRP54, SRP68, SRP72, SEC65) were PCR amplified 680 off the pMW295 or pMW299 plasmids with oligonucleotides containing 5' SpeI and NdeI 681 sites and 3' XhoI sites, and they were cloned into the SpeI and XhoI sites of the 2HA_p424 682 plasmid containing an ADH promoter, two HA tags and a CYC1 terminator (92). SRP14 and 683 SRP21 also were cloned into the GFP_p413 plasmid. Except for SRP68 and SEC65, all 684 constructs were subcloned into the NdeI and XhoI sites of pET22b. SRP68 and SEC65 were 685 re-amplified by PCR with oligonucleotides containing XhoI and BamHI sites and cloned into 686 the equivalent sites of pET19b. SRP101 and SRP102 were amplified off purified 687 chromosomal DNA using oligonucleotides with BamHI and XhoI sites and cloned into the 688 equivalent sites of 2HA_p424. The constructs were subcloned into the NdeI and XhoI sites of 689 pET22b. SRP14∆29Cter and SRP21∆73Cter were PCR amplified with the HA tag and cloned 690 into the XbaI and XhoI sites of p413 (55). SRP14∆29Cter and SRP21∆73Cter were 691 subsequently subcloned into the NdeI and XhoI sites of pET22b. 692 The GFP and MCHERRY plasmids were constructed by amplifying genes off the 693 pYM27-EGFP-KanMX4 and pFA6a-mCherry-NatNT2 plasmids, respectively. The PCR 694 products were digested with XhoI and SalI, gel purified, and cloned into the equivalent sites 695 of the yeast plasmids p415, p416 and p413. The DED1, ded1-F162C and ded1-DQAD 696 plasmids were as previously described (13). Theses genes were subcloned into the SpeI and 697 XhoI sites of GFP-p415, p414 and MCHERRY_p416. The KAR2-RFP_YIPlac204 was a gift 698 from Benjamin Glick. 699

Yeast strains and manipulations 700
Manipulations of yeast, including media preparation, growth conditions, transformation, and 701 5-FOA selection, were done according to standard procedures (94). The strains used in this 702 study are listed in Supplementary Table 3. 703 The yeast GFP clone collection was purchased from Life Technologies (Ref 95702;704 Carlsbad, CA). The sec61-ts and sec62-ts yeast strains were a generous gift from Ron 705 Deshaies (59) Table 2. 717 The full-length T7-SCR1 was PCR amplified off the pMW299 plasmid (44) with the 718 SCR1_up2 oligonucleotide containing the T7 promoter and SCR1-low oligonucleotides. The 719 PCR product was digested with BamHI and XhoI, gel purified and cloned into the equivalent 720 sites of pUC18. The T7-SCR1∆Alu was constructed with the SCR1_dAlu_up oligonucleotide 721 containing the T7 promoter and the SCR1_dAlu_low oligonucleotide. The PCR product was 722 digested with BamHI and HindIII, gel purified and cloned into the equivalent sites of pUC18. 723 The T7-SCR1∆S1 construct was made as fusion PCRs with two sets of oligonucleotides. The 724 pUC18_5' oligonucleotide was used with SCR1_dS1_low and pUC18_3' was used with 725 SCR1_dS1_up. The two gel-purified PCR fragments were combined and PCR amplified with 726 oligonucleotides pUC18_5' and pUC18_3'. The PCR product was digested with BamHI and 727 HindIII, gel purified and cloned into the equivalent sites of pUC18. 728

Northern blot probes 729
The oligonucleotides used are listed in Supplementary Table 1

In situ localization 742
To analyze the location of Ded1 relative to the ER or SRP proteins, we first used the green 743 fluorescent protein (GFP)-tagged ded1-DQAD plasmid that was transformed into sec61-ts and 744 sec62-ts mutant strains with the integrated KAR2-RFP plasmid (Supplementary Tables 2 and  745 3). Cells were grown in SD-LEU to an OD 600 of ~0.9-1.0 (logarithmic phase) at 24°C and 746 then shifted to 37°C for 15 min. We subsequently used mCherry-tagged ded1-DQAD plasmid 747 that was transformed into GFP-tagged SRP14 and SRP21 expressed from the chromosome 748 (Supplementary Tables 2 and 3). Cells were grown in SD-LEU to an OD 600 of 0.95 at 30°C. 749 Finally, GFP-tagged SRP14 or SRP21 and mCherry-tagged ded1-DQAD plasmids were 750 transformed in the xpo1-T539C strain (70). Cells were grown in minimal medium lacking 751 histidine and uracil (SD-HIS-URA) to an OD 600 of 0.4 (early logarithmic phase) at 30°C, and 752 then they were split into two parts: one-half was resuspended in SD-HIS-URA and the other 753 in SD-HIS-URA supplemented with ~200 nM leptomycin for 1 h. 754 Inc, Colorado Springs, CO). 780

Recombinant protein expression and purification 781
Recombinant Ded1-His was expressed from the pET22b plasmid (Novagen) and purified as 782 previously described (46). SRP His-tagged proteins were transformed into the Rosetta (DE3) 783 E. coli strain. Cultures containing 500 ml of cells at OD 600 of 0.5 were induced with 0.5 mM 784 isopropyl-1-thio-ß-D-galactopyranoside (IPTG) for 1 h at 37°C for SRP14, Sec65 and 785 SRP101; 16 h at 16°C for SRP68 and SRP72; and for 2 h at 30°C for SRP54 and SRP21.  Table 3 and the purified proteins shown in 803 Supplementary Figure 1. 804

Immunoglobulin G-protein A Sepharose-bead pull-down experiments 805
The G50 yeast strain was transformed individually with 2HA_p424 plasmids containing two 806 amino-terminal HA tags and the genes for SRP14, SRP21, SRP54, SEC65, SRP68, SRP72 and 807 SRP101. Cells were grown to an OD 600 of 0.8-1 in minimal medium lacking tryptophan (SD-808 4°C with mixing. The pH of the eluted proteins was then adjusted to pH ~7 with NaOH. 829

Other pull down experiments 830
Protein A-Sepharose beads were prepared as described above. Ded1 or SRP21 IgG were 831 crosslinked to beads with 0.2% glutaraldehyde as described previously (95)  The protein fractions were concentrated by adding 150 µg/ml, final, of sodium 874 deoxycholate and then the solutions were made 15% in TCA, centrifuged in an Eppendorf 875 5415R, the recovered pellets were washed twice with cold acetone, and then the protein 876 pellets were dried. The recovered material was resuspended in SDS loading buffer, separated 877 by electrophoresis on a 12% SDS-PAGE gel and electrophoretically transferred to 878 nitrocellulose membranes. The Western blots were undertaken as described above. 879 The RNA fractions were made 0.3 M in potassium acetate, extracted with an equal 880 volume of water-saturated phenol, extracted twice with an equal volume of chloroform-881 isoamyl alcohol (24:1), and then ethanol precipitated overnight at -20°C. The RNA was 882 recovered by centrifugation, dried, resuspended in 1X RNA loading buffer (Thermo 883 Scientific), and then electrophoretically separated on a 6% polyacrylamide gel containing 7 M 884 urea and ethidium bromide (to reveal 18S and 23S rRNAs). The RNA were subsequently 885 electrophoretically transferred to Amersham Hybond-N+ nylon membranes (GE Healthcare) 886 and probed with a 32 P-labeled DNA oligonucleotide specific for SCR1 (Supplementary Table  887 1). The image was visualized with a Typhoon FLA9500 phosphoimager (GE Healthcare). 888

In vitro RNA-dependent ATPase activities 913
The ATPase assays were based on a colorimetric assay using molybdate-Malachite green as 914 previously described (46). We typically used 23 nM of SCR1 or actin RNA and 0.14 µg/µl of 915 whole yeast RNA (Roche) that was purified on a DEAE-Sepharose column to remove 916 inhibitors. For the latter RNA, fractions from an elution with increasing concentrations of 917 NaCl were assayed with purified Ded1, and the most active fractions were combined, 918  .32 a Data was collected and analyzed as previously described (13). In brief, 0.5 ml fractions were collected starting from the top of the gradient and subjected to nano-liquid-chromatography electron-spray mass spectrometry analysis b ENO2, SSA2 and FBA1 are reference proteins; they showed the strongest signals in the fraction at the top of gradient. c Fractions correspond to those shown in Supplemental Figure S7 of Senissar et al. (13). d Mascot probably-based scoring. e Spectral counting; the same peptide is fragmented up to six times over a mean elution time of 30 seconds. f Percentage of the protein sequence covered. g Mean error in ppm. 6, 7 ---a Data was collected and analyzed as previously described (13). Equivalent fractions as shown in Table 1 were subjected to IgG-Ded1 pull-downs with protein-A Sepharose beads and subjected to mass spectrometry analysis. b ENO2, SSA2 and FBA1 are reference proteins; they showed the strongest signals in Table 1. c Fractions correspond to those shown in Supplemental Figure S7 of Senissar et al. (13). d Mascot probably-based scoring. e Number of peptide fragments recovered. f Percentage of the protein sequence covered. g Mean error in ppm.   and then revealed with anti-HA IgG. Input, 40 µg (~10%) of the yeast extract was directly 1269 loaded on the gel. Ded1-IgG, Ded1-specific IgG was used to pull-down Ded1 associated 1270 proteins. IgG+RNase, complexes bound to Ded1-IgG-protein A beads were digested with 1271

1250
RNase A (1mg/ml) prior to washing and elution. 1272  peptone, dextrose) rich-medium agar plates, except for TET-SRP101 and BY4742 that were 1296 plated on SD medium agar plates, in the presence (+DOX) or absence (-DOX) of 10 µg/ml of 1297 doxycycline. The G50 and BY4742 strains show wildtype growth. Plates were incubated for 2 1298 days at 30°C and 36°C, and for 4 days at 18°C for the YPD plates, and for 4 days and 7 days, 1299 respectively, for the SD plates. 1300        (33,34). Yeast and other fungal SRP RNAs are unusual in that they are much larger than in other organisms, and they lack the characteristic structure consisting of hairpins 3 and 4 of the Alu domain. Yeast has the additional hairpins 9, 10, 11 and 12 that are poorly characterized and that have other proposed secondary structures. Conserved sequence motifs and tertiary interactions are shown in gray.

Input
Ded1-IgG IgG+RNase Ded1-IgG Pre-IgG Figure 2. Ded1-IgG pull-downs of yeast extracts. Ded1-specific IgG (Ded1-IgG) or IgG from pre-immune serum (Pre-IgG) were used to recover the associated factors. Input, a fraction of the yeast extract used in the pull-down experiments was directly loaded onto the gel or RT-PCR amplified. (A) Purified RNA from yeast extracts (~20% of input) or from IgG pull-downs was reverse transcribed and PCR amplified for 25 cycles with gene-specific oligonucleotides. The resulting products were electrophoretically separated on a 2% agarose gel containing ethidium bromide, and the products visualized with a Gel Doc XR+ (Bio-RAD). (B) Western blot analysis of HA-tagged SRP proteins. Proteins were electrophoretically separated on a 12% SDS-PAGE, transferred to nitrocellulose membranes, and then revealed with anti-HA IgG. Input, 40 µg (~10%) of the yeast extract was directly loaded on the gel. Ded1-IgG, Ded1-specific IgG was used to pull-down Ded1 associated proteins. IgG+RNase, complexes bound to Ded1-IgG-protein A beads were digested with RNase A (1mg/ml) prior to washing and elution.   Figure 5. Phenotypes of proteins expressed with tetracycline promoters. Liquid cultures of the indicated strains were serially diluted by a factor of 10 and plated on YPD (yeast extract, peptone, dextrose) rich-medium agar plates, except for TET-SRP101 and BY4742 that were plated on SD medium agar plates, in the presence (+DOX) or absence (-DOX) of 10 µg/ml of doxycycline. The G50 and BY4742 strains show wildtype growth. Plates were incubated for 2 days at 30°C and 36°C, and for 4 days at 18°C for the YPD plates, and for 4 days and 7 days, respectively, for the SD plates.

TET-SRP14
-DOX +DOX 4 days, 36°C TET-SRP21 Figure 6. Ded1 multicopy suppression of SRP protein depletions. Cells of the indicated strains with the TET promoter were grown in SD-LEU medium, serially diluted by a factor of 10 and spotted on SD-LEU agar plates with (+DOX) or without (-DOX) 10 µg/ml of doxycycline. Cultures were grown 4 days at 36°C. p415, empty LEU plasmid; GPD-DED1, Ded1 in p415 with the high expression GPD promoter and CYC1 terminator; GPD-F162C, a Ded1 mutant with reduced ATP binding and enzymatic activity (46). BY4742, a wildtype yeast strain showing unimpeded growth. The phenotypes were most apparent at 36°C, but similar effects were obtained at 30°C.   Reaction were done as in A but with 23 nM of the SRP proteins except SRP14, which was used at 46 nM to form the homodimer, and 23 nM RNAs. The reaction velocities were normalized relative to the activity of Ded1 in the presence of the RNA (SCR1 or actin) alone. +SRP-All, Ded1 was incubated with SRP14, SRP21, SRP54, Sec65, SRP68, and SRP72; Ded1-GAT, a Ded1 P-loop mutant that lacks ATPase activity; +no RNA, Ded1 was incubated in the absence of an RNA substrate with the SRP proteins. The mean and standard deviations are shown for two independent experiments in panel A and for three in panel B. The lower error bars were deleted for clarity.  Figure 11. The RNA-dependent effects of SRP21 on the ATPase activity of Ded1. (A) Ded1 was pre-incubated with the RNAs at 30°C for 30 min. SRP21 or SRP21∆73 were then added at 200 nM with 1 mM ATP, and the ATPase velocity was measured over 40 min. The mean and standard deviations are shown for two independent experiments. (B) Ded1 at 7 nM was incubated with 23 nM SCR1, 23 nM actin or 0.14 µg/µl yeast RNA and with 1 mM ATP. SRP21 was used at 23 nM and SRP14 at 46 nM (to form homodimer). The reaction velocities were measured over 40 min at 30°C. The mean and standard deviations are shown for three independent measurements are shown for SCR1 and actin and for two independent measurements for yeast RNA. (C) Reactions were done as in B. Ded1 at 7 nM was incubated with 23 nM SCR1 (equivalent to 0.0039 µg/µl) or with 0.12 µg/µl of tRNA or poly(A). The SRP21 was used at 200 nM. The mean and standard deviations are shown for three independent measurements. The lower error bars were deleted for clarity. associates with the mRNA during translation initiation and remains attached to the mRNA in front of the ribosomes. It consists of RecA-like domains 1 (d1) and 2 (d2), an amino-terminal domain (N) and a carboxyl-terminal domain (C). The RNA-dependent ATPase activity of Ded1 is unaltered, and it is often in the "open" conformation with weak affinity for the RNA; it is able to translocate with the ribosomes during translation. (B) The SRP (shown in blue) associates with ribosomes translating mRNAs (or undergoes conformational changes in the case of pre-bound SRP) when the signal peptide leaves the exit channel and obtains a certain length. Ded1 may help in assembling and stabilizing the complex. Conformational changes of the SRP causes SRP14 to block the entry channel and prevent the eEF2 elongation factor (EF) from binding the ribosomes, which pauses elongation. Ded1 may bind part of the Alu domain of SCR1, shown in magenta, during these conformational changes to promote SRP14 binding to the ribosomes. At the same time, SRP21 inhibits the ATPase activity of Ded1, which forms the "closed" conformation with high affinity for the RNA. This ATP-bound form of Ded1 kinks the RNA (red triangle) on domain 1 and locks Ded1 on the RNA. This prevents the ribosomes from frame shifting (sliding) on the RNA and perhaps stabilizes the ribosome-mRNA complex to prevent premature termination of translation. (C) The paused mRNA-ribosome complex associates with SRP receptor (SR) factors SRP101 and subsequently SRP102, which brings the mRNA-ribosome complex to the Sec61 ER translocon. (D) The SRP complex dissociates from the ribosomes, the ATPase activity of Ded1 is restored and translation continues. Note that this model also applies to the SRP-dependent import of polypeptides with internal transmembrane domains, and it does not preclude the possibility that multiple Ded1 molecules are involved, that the SRP associates multiple times with the ribosomes during elongation or that the SRP-associated ribosomes remain on the ER over multiple rounds of translation.