Results

↓32

3.1  Ribosome nascent chain complex purification and reconstitution of SRP-RNC complex

↓33

Since formation of a stable complex was a prerequisite for this study, wheat germ RNCs and canine SRP were used for reconstitution of the targeting complex. This well characterized combination, which led to the discovery of SRP, displays strong elongation arrest activity [15]. Assuming that this activity is a result of equally stable binding of the S-domain and the Alu-domain to the ribosome, this heterologous complex was considered as the most suitable candidate for structure determination.

First programmed ribosomes carrying a functional signal sequence (RNCs) were isolated from an in vitro translation reaction as described in materials and methods section. For the translation a wheat germ in vitro system was used (Ambion), with mRNA encoding for first 90 amino acids of the type-II membrane protein dipeptidylpeptidase B (DPAP-B) from yeast, containing a signal anchor sequence and, in addition, an HA/His tag. Since the mRNA did not have a stop codon translating ribosomes were stalled at the end of the mRNA with the peptidyl-tRNA in the P-site. These ribosomes were affinity-purified by metal affinity chromatography. The programmed ribosomes were eluted under native conditions using imidazol. The enrichment of ribosome nascent chain complexes is shown in fig. 11a. When comparing the specific signal before purification (R) and after purification (E) it is apparent that a several-fold stronger nascent-chain dependent signal with lower amounts of ribosomes indicates at least a 5-fold enrichment. If in the translation system, conservatively estimated, only 20% of the ribosomes were programmed and stably stalled, more than 95% of the ribosomes in the final fraction can be expected to carry a nascent chain.

Stalled RNCs were used for the reconstitution with excess amounts of purified canine SRP. To ensure specific, i.e. signal sequence dependent complex formation, sucrose density gradient centrifugation was performed under high salt conditions (500 mM KOAc) [98], which confirmed high affinity binding of SRP to RNCs with an estimated occupancy varying between 50-90%.

↓34

Figure 11: Purification of RNCs and reconstitution of the RNC-SRP complex.

a, A truncated mRNA coding for the first 90 amino acids of dipeptidylpeptidase B with His-tag was translated in a wheat germ cell-free system and RNCs purified as described in the Methods section. Aliquots of fractions were subjected to SDS-PAGE, blotted onto nitrocellulose and amido black stained (top) and probed using a monoclonal anti-HA antibody (bottom). T, total; S, supernatant; R, crude ribosomes; dR, depleted crude ribosomes after column incubation; W, washes; S, supernatant after elution; E, pellet after elution using imidazol. Note the enrichment of DP90 RNCs and the characteristic pattern of ribosomal proteins in the final fraction (E). b, Purified DP90 RNCs (RNC) were reconstituted with excess SRP (SRP) and, subjected to sucrose density centrifugation at 500 mM KOAc, fractionated and analyzed by SDS-PAGE and Sypro-Orange stain. Brackets indicate contaminating proteins in the gradient. The asterisk marks RNase inhibitor, which partially copurifies with RNCs, but is not present in the reconstituted RNC-SRP complex (compare Top with 80S). High salt resistant RNC-SRP complexes migrate in the 80S fraction indicating signal sequence-dependent SRP binding to RNCs.

3.2 Structure of the signal recognition particle interacting with the elongation arrested ribosome

Cryo-EM and three-dimensional (3D) reconstruction of the targeting complex revealed the typical appearance of an 80S ribosome at 12 Å resolution (7.7 Å according to 3σ criterion, see fig. 12, 13) with two additional densities (fig. 12): firstly, a tRNA is visible in the P-site in the ribosomal intersubunit space. Secondly, a large elongate mass representing SRP stretches from the peptide exit site of the 60S ribosomal subunit (S-domain) into the intersubunit space (Alu-domain), forming a total of 6 connections (C1-C6) with the ribosome (fig. 12, 14).

Figure 12: Cryo-EM map of mammalian SRP bound to 80S RNC at 12.0 Å.

Ribosomal and SRP densities are shown at different contour levels and amplitude correction was done with different B factors. The lower contour level necessary for the SRP density is a result of lower occupancy of SRP as ligand and uncompleted removal from the dataset of RNCs without SRP during sorting. Accordingly, the lower B factor used for amplitude correction of SRP density is also due to lower occupancy of SRP and the lower overall resolution of the SRP density. a, The RNC-SRP map is shown with the separated colour-coded densities. Yellow, 40S small ribosomal subunit; blue, 60S large ribosomal subunit; green, P site tRNA; red, SRP. C1-C6 assigned positions of RNC-SRP connections (see also table 1); h1 and h2 are hinges of the 7S RNA backbone of SRP; St, stalk; SB, stalk base. b, As a but rotated by 70° to the right. c, As a but rotated upward by 90°. d, Same orientation as b but with molecular models for SRP and the 80S RNC in transparent densities.

↓35

Figure 13: Resolution curve: Fourier shell correlation (FSC) curve for the cryo-EM map of the RNC-SRP complex.

FSC = 0.5 indicates 12.0 Å resolution. Some structural information extends to 7.7 Å according to the more lenient 3σ criterion.

The tRNA density reflects the presence of the nascent peptidyl-tRNA containing the signal sequence, which is stalled at the 3’-end of the truncated mRNA and stabilized by cycloheximide (purification described in Methods section). Sorting of the dataset according to the presence of SRP density resulted in a subset of ~70% of particles, in agreement with the occupancy estimated by SDS-PAGE, which were used in the final reconstruction. Sorting of the dataset according to the tRNA presence removed ~1-2% of particles from the dataset showing that the purification of programmed ribosomes was highly efficient.

Figure 14: Cryo-EM map of the SRP-RNC complex.

a, RNC-SRP map shown with separated colour-coded densities, all at the same contour level and before application of Fourier filtering using B-factors. Colour coding and labels as in fig. 12. b, As a but rotated by 70°. Notably, the dynamic expansion segment 27 (ES27) is not visible. Noise between the S-domain density and the ES27 position near L1 (visible only at very low contour levels) indicates that ES27 has no preferred orientation in this complex.

↓36

In order to allow interpretation on a molecular level, crystal structures and molecular models were docked into the electron densities. In the case of the ribosome, the remarkable similarity between the wheat germ RNC and yeast RNC [99] (see fig. 18 and fig. 27) allowed the use of a molecular model generated earlier [100]. Therefore, here was used the yeast nomenclature for the molecular description of ribosomal components (family names in parenthesis). In the case of the SRP, recently solved X-ray structures of SRP fragments were docked into the density as rigid bodies (see Methods).

SRP shows a bent conformation with one of two hinges apparently facilitating a major kink (hinge 1) separating the S- and Alu-domains. Hinge 1 separates the 160 Å long S-domain of SRP near the peptide exit site from a RNA linker connecting Alu-domain in a region close to the subunit interface (spanning a length totalling 120 Å). The RNA at hinge 1 represents a large loop around nucleotides 100/250 and forms an angle of almost 90° (fig. 12b, 15). Hinge 2 is located in a region corresponding to a small loop formed by nucleotides 70 and 275 of 7SL RNA. This hinge facilitates a bend (~ 30°), which leads to an orientation of the 5’ Alu-RNP that is perfect for its entry into the intersubunit space (fig. 12a, 16b).

Figure 15: Molecular model of SRP.

a, Secondary structure of the SRP RNA with protein binding sites and hinges indicated. H1-H8 denote the RNA helices of SRP, following the nomenclature of Weichenrieder et al[40]. In the case of the Alu-domain. Cyan, blue and grey, SRP proteins; red and yellow, 7S RNA; green, signal sequence. b, Molecular model of SRP with density transparent and colour coding as in a. Top view showing SRP as seen from the ribosome. c, As b but rotated upwards. d, As c but rotated left.

↓37

The structure of a large fragment of the mammalian S-domain containing 7SL RNA helices 6-8 and part of helix 5 as well as the SRP19 protein and the SRP54 M-domain [46] was docked into the density identified as the S-domain. The original SRP54 M-domain was replaced with another model [93] differing only in the position of helix 1 and the finger loop, since this model was fitting better into the density. As a signal sequence, an alpha-helical peptide fragment of 16 amino acids was positioned into the corresponding density near the exit site which was not occupied by the crystal structure. As a result, the signal sequence can contact the hydrophobic groove of the SRP54 M-domain and the phosphate backbone of SRP helix 8 RNA with its positively charged N-terminal part [11]. The conserved structure of a prokaryotic SRP54 NG-domain from T. aquaticus bound to the nonhydrolisable GTP analog GMPPNP [94] was docked next. Interestingly, the NG-domain is positioned such that a gap of ~20 Å is separating it from helix 8 of 7S RNA, only connected by the M-domain. In this position, a part of the NG-domain is in too close proximity to the finger loop of the M-domain indicating a more compact conformation of the loop in the case of signal sequence binding. In the crystal structure of M-domain [46,93] the loop is wide open because of high detergent concentrations apparently destabilizing hydrophobic interactions. In support of this model a very recent crystal structure of an archaeal SRP54-RNA complex revealed a similar overall arrangement of M- and NG-domains [53] with the finger loop closed. SRP in the cryo-EM structure is bound to a signal sequence and, therefore, the loop can be expected in the open conformation covering the bound signal sequence, but not extensively open as in the structure of SRP54 M-domain by Kuglstatter et al.

Extra density in the S-domain was interpreted as the SRP68/72 dimer of a hitherto unknown structure. It is located mainly at the junction of helices 5-8; however, additional density is present at the hinge 1 region of helix 5 (fig. 15c) and it belongs most likely to SRP68/72 dimer. This is in accordance with foot printing experiments showing protection of all of these regions of 7S RNA (including nucleotides 100/250) [49]. The fragmented mass may be an indication for a tertiary structure containing thin, extended ‘tentacle’ regions as observed for some ribosomal proteins such as L19e or L22 [101]. In the described position, SRP68/72 can serve as a brace between the core of the S-domain and the dynamic hinge 1, thereby functionally connecting Alu- and S-domain.

The X-ray structure of the mammalian SRP9/14 dimer bound to the 5’ part of the Alu-RNA[40] fits perfectly into the density in the intersubunit space (fig. 16). This model matches the conformation suggested before [40], in which the 5’-RNP, comprising the SRP9/14 dimer and the first 48 nucleotides of the RNA, folds back onto the 3’RNA stem of the Alu-domain. Thus, this back-folding appears indeed to be a necessary assembly step of the Alu-domain.

↓38

In a final step three fragments from a model provided by the SRP-database [95] were used as a ruler to corroborate that the missing part of 7SL RNA can span the distance between the docked Alu- and S-domain fragments.

Taken together, density could be assigned to all known components of the mammalian SRP, consistent with their size and structure, thus, leading to a first molecular model of SRP in the functional context of a ribosomal targeting complex.

An overview of the complete model in the context of the ribosome is illustrated in fig. 12d and fig. 16a-b. The docked fragments easily span the distance between the peptide exit site and the elongation factor binding site. The three connections of the S-domain with the ribosome, found in the immediate vicinity of the peptide exit site, are contributed exclusively by the SRP54 protein. A fourth connection is contributed by 7S RNA and the SRP68/72 dimer. The Alu-domain bridges, in a tight fit, the ~65 Å distance between the large and small ribosomal subunits (fig. 16b) where it interacts with both small and large ribosomal subunit. An empty space in the intersubunit cavity, corresponding to the unoccupied A-site is indicating that the Alu-domain would not interfere with a tRNA bound in this position (fig. 16a-b).

↓39

Figure 16: Molecular model of SRP and the ribosome.

A, SRP with isolated 40S and 60S ribosomal subunits exposing P-site tRNA (green) shown from the 60S and the 40S side, respectively; A and E, position of A- and E-sites; Alu, Alu-domain; head, head of 40S subunit; L1, L1 protuberance; CP, central protuberance. Other labels as in a. B, Cut top view showing the Alu-domain in the intersubunit space, with labels as in e. Same view magnified, with molecular model of Alu-domain colour coded as in fig. 15.

3.2.1  Environment and function of the Alu-domain

The two connections between the Alu-domain and the ribosome are contributed exclusively by the 5’-RNP comprising the first 48 nucleotides of SRP RNA and the SRP9/14 heterodimer (fig. 17). In connection 5, the 5’RNA of SRP is interacting with both RNA and protein of the large ribosomal subunit. The loops L1.2 and L2 as well as the short helix 2 of 7S RNA contact the large ribosomal subunit via the so-called stalk base and, probably, the universally conserved α-sarcin-ricin loop (SRL). The participating components of the stalk base are the N-terminal part of rpL12 (L11p) and the tip of helix 43 of 25S ribosomal RNA (fig. 17b).

Figure 17: SRP Alu-domain interaction with 80S ribosome.

a, SRP density with models showing connection 6 between SRP9/14 and ribosomal 18S RNA (helix 5, 15 and 14) in an orientation similar to fig. 12b. Colour code as in fig. 15; pale yellow, 18S ribosomal RNA; light blue, 25S RNA; orange, 60S proteins. Bottom: as top view but tilted toward the viewer. b, Connection 5 between SRP Alu-RNA (loop 1.2 and helix 2) and the GAC of the 60S ribosomal subunit (rpL12, helix 43 and SRL, sarcin-ricin-loop). Bottom: as top view but tilted toward the viewer.

↓40

In connection 6 (fig. 17a), the only contact of SRP with the small ribosomal subunit is established through the SRP9/14 dimer and ribosomal 18S RNA. Mainly the SRP14 surface participates in contacts with helices 5 and 15 of 18S RNA. SRP9 is in contact with the same helices and, in addition, is close to helix 14. According to biochemical data [102] deletion of the C-terminus of SRP14 abolishes elongation arrest. This functionally essential region and a large loop between the ß2- and ß3-strand of SRP14 are not resolved in the X-ray structure [40], therefore, at the given resolution it is not possible to draw any conclusions regarding their participation in ribosomal contacts.

3.2.2 Environment and function of the S-domain

The S-domain makes altogether four connections with the large ribosomal subunit. The first connection between S-domain and the large ribosomal subunit is formed by the tip of the SRP54 N-domain (fig. 18a). Two loops connecting helices 1 and 2, and helices 3 and 4, which build the four-helix bundle, come into close proximity with rpL25 and rpL35 (corresponding to L23a/L35 in wheat germ and L23p/L29p in E. coli). This contact site is in agreement with previous cross-linking experiments identifying the same proteins as the main proteinaceous ribosomal constituents in immediate vicinity to SRP54 [13]. In similar experiments, the same region of the bacterial SRP54 N-domain has been found in a position adjacent to L23p [103], suggesting that this interaction is evolutionary conserved. In addition to SRP, signal sequences [104] and the chaperone trigger factor [105,106] have been shown to bind to L23p, and the protein-conducting channel (PCC) of the ER (the Sec61 complex) to rpL25/rpL35 [99] (fig. 19c). Thus, the rpL25/35 (L23p/29p) proteins constitute a promiscuous binding site of the ribosome, which facilitates the interaction with multiple factors involved in the different aspects of cotranslational processing (fig. 19c).

The second connection is formed by the N-terminal part of the SRP54 M-domain contacting the helix 59 (expansion segment 24, ES24) of 25S ribosomal RNA (fig. 18b). As in the case of the first connection involving rpL25/35, this contact site of SRP is shared with the Sec61 complex [99]. The signal sequence is also closest to this connection when bound to SRP in the suggested position.

↓41

The third connection engages the C-terminal region of the M-domain, in particular the 7S RNA-binding part of it, which, again like the Sec61 complex, interacts with helix 24 of the 25S ribosomal RNA (fig. 18b). The interaction is established between helix 3 of the SRP M-domain and the tip of helix 24. The M-domain of mammalian SRP54 contains an additional region of about 60 residues at the C-terminus, which is present as density in the EM structure but not present in any of the current high resolution crystal structures. The helix 5 and the C-terminal part of the M-domain may play also a role in this connection. However, in contrast to the Sec61 complex, which contacts the stem of this helix, the SRP binding site is shifted toward the tip of rRNA helix 24.

The exclusive involvement of ribosomal RNA in the connections 2 and 3 is in agreement with the observation that rpL25/35 (L23p/29p) are the only ribosomal proteins cross-linked to SRP54.

Figure 18: SRP S-domain interaction with 80S ribosome.

a, Density with molecular models docked and cut to show connection 1 (C1) between SRP54NG and rpL25/rpL35. SRP colour code as in fig. 15; light blue, 60S ribosomal density and 25S RNA; orange, 60S proteins. b, Connections 2 and 3 (C2, C3) between SRP54M and ribosomal H59 and H24, respectively. c, Density in orientation as b and d, cut to show connections C2 to C4. d, Connection 4 with helix 5 of SRP (and SRP68/72) contacting the 60S subunit. 2850, position of nucleotide 2850 in 25S ribosomal RNA model; H99 and H98, 25S RNA helices; ES39, expansion segment 39.

↓42

Table 1: Contacts between mammalian SRP and the 80S ribosome

Contact

Typea

SRP

RNA/protein position

80S Ribosomeb

RNA/protein positionc

S-domain

60S subunit

C1

p-p

54NG

60-75

15-26

rpL25(L23)

rpL35(L29)

130-35

16-27

C2

p/R-p

p-R

54M/H8

54M

N-term of M

signal sequence

H59[ES24]

1627-34d

C3

p-R

54M

388-99

H24

490-95

C4

R-R/p

p-R/p

H5

68/72

218-28,121-27

H99, 100/101, rpL16

H98/ES39, rpL16

2907-10, 2849-51

Alu-domain

60S subunit

C5

R-p

R-R

R-R

L1.2

L2

H2

35-36

13

09-19

rpL12(L11)

H43

H95[SRL]

65-69

1171

2696-99

40S subunit

C6

p-R

p-R

p-R

14

9

9

74-89

57-75

55-60

h5, h15

h5, h15

h14

55-58, 356-59; 368

55-58, 368

341-44

a) R and p correspond to RNA (R) and protein (p) involved in contacts
b) Yeast nomenclature is used with family name in parenthesis
c) Positions correspond to model based on yeast 80S ribosome[100]
d) Positions correspond to yeast 25S RNA secondary structure (http://www.rna.icmb.utexas.edu)

The fourth connection is the only one involving the 7S RNA of the S-domain. It appears to be split into two entities, one of which (4a) is very close to RNA helix 5 of SRP. However, it is likely that SRP68/72 is involved in this connection (4b) as well, since there is additional density around RNA helix 5 and connection 4. The second connection (4b) is located right next to a ribosomal density that has been identified as expansion segment 39 (ES39) [100], formed by an extension of helix 98 of 25S RNA. This interaction would be then characteristic only for eukaryotic ribosomes, since ES39 is not present in eubacteria and archaea. In addition, rpL16 (L13p) projects a loop into the vicinity of connection 4. The closest ribosomal structures found for connection 4a are RNA helix 99 and a small loop at the junction between helices 100 and 101 of 25S RNA. The corresponding region in the E. coli ribosome, around nucleotide 2828 of 23S RNA (nucleotide 2850 in our model), has been found in vicinity of bacterial SRP 4.5 S RNA by cross-linking [107], again, suggesting a conserved mode of interaction.

Thus, considering the previously mentioned localization of SRP54, the core of the eukaryotic S-domain, including SRP54 and helices 8 and 5, appears to be positioned on the ribosome in a similar overall orientation as its counterpart in prokaryotes.

↓43

Figure 19: SRP S-domain interaction with the 80S ribosome.

a, Model of S-domain covering peptide exit site of 60S subunit. SRP colour code as in fig. 15; light blue, 60S ribosomal subunit, yellow, 40S ribosomal subunit. b, As e with transparent ribosomal density and contour of SRP density to show locations of connections. Asterisk denotes peptide tunnel exit. c, Comparison of 80S RNC-Sec61 complex from Yeast with 80S RNC-SRP complex in same orientation. The magnified area is shown with contour of SRP (red) and Sec61 (black line) and their partially overlapping contact sites, SRP (blue), Sec61 (black line).

3.2.3 Functional states of SRP54

The most conserved part of SRP is the core of S-domain consisting of protein SRP54 (Ffh for Fifty-four-homolog in bacteria) bound to helix 8 (domain IV in bacteria) of SRP RNA [4,26]. Structures of individual domains of SRP54 from various species are known since several years, but not their relative orientation [47]. Only recently, the spatial arrangement of these domains in two different functional states of SRP, in the free state (before binding to the ribosome or SR) and in the ribosome-bound state, became available from X-ray [53] and cryo-electron microscopy (cyro-EM) data, respectively (fig. 20). The X-ray structure of SRP54 of the Crenarchaea Sulfolobus solfataricus was obtained with and without SRP RNA helix 8, resulting in very similar domain arrangement. Check for repetition and adjust (you cannot quote yourself!) Although the crystal packing might have influenced the conformation, it is very likely that the overall domain arrangement indeed represents the conformation of SRP54 in the free state since these structures were obtained in different crystal forms with different crystal packing. At the same time, the differences between them reflect the intrinsic flexibility of SRP core in free state. A cryo-EM structure shows the complete mammalian SRP after binding to the RNC and represents the ribosome-bound state of SRP. Here, the SRP core is part of the active targeting complex after binding both the signal sequence and the large subunit of the ribosome near the peptide exit tunnel. Although the resolution of the cryo-EM map is limited to 12 Å, the accuracy of interpretation can exceed that resolution several fold by docking of molecular models. The comparison of these two structures allows identifying large dynamic changes of SRP core between the free state and the ribosome-bound state during step I of SRP cycle. In addition, X-ray structures of the interacting NG-domains of bacterial SRP and SR provide a glimpse at the docking step (step II) at the target membrane [77,78]. It turns out, that extensive conformational changes within the SRP core take place between the functional states, in particular between the free and the ribosome-bound state.

The X-ray structure of the free SRP core [53] reveals SRP54 as an L-shaped molecule, with SRP54NG as the longer arm of the L which aligns parallel with helix 8 of the SRP RNA (free state) (fig. 20a). Although SRP54NG does not directly contact the RNA, biochemical data [108] and differences within the SRP54 structures with and without RNA indicate that such an interaction is likely to exist. SRP54NG and SRP54M are connected by a flexible linker region that has not been observed in other earlier SRP54 structures. The linker region consists of a conserved 'LGMGD' sequence fingerprint in a loop preceding a long linker helix followed by another loop of variable size. SRP54M itself can be divided into a flexible N-terminal and a rigid C-terminal part. The C-terminal part (MC) binds to helix 8 of SRP RNA as a rigid body and provides a stable platform for the hydrophobic groove which is proposed to bind the signal sequence. The flexible N-terminal part (MN) includes a proline-kinked helix (αM1 and αM1b) followed by the finger loop which in the absence of a signal sequence shields the hydrophobic groove from the aqueous solvent. The comparison with the SRP54M structure from Thermus aquaticus [1] allowed to assign several hinge points in SRP54MN which would be sufficient for anchoring to SRP54MC and for adjusting the hydrophobic groove in order to accommodate the signal sequence [53].

↓44

Figure 20: Structures of the SRP core in the free and ribosome-bound state.

(a) The SRP core in the free state as derived from the X-ray structure [53]. The colour code is: RNA (red ribbon), SRP54NG (yellow), SRP54MN (orange), SRP54MC (dark red). (b) Cryo-EM structure of the mammalian SRP bound to the signal sequence carrying 80S ribosome. The SRP core as part of the S-domain is positioned near the tunnel exit of the large ribosomal subunit. The small 40S and large 60S ribosomal subunits are yellow and light blue, respectively. The SRP density is shown transparent with docked molecular models. Colours of labelled elements are: SRP54NG (turquoise), signal sequence (green), SRP54M (dark blue), RNA helix 8 (red). (c) Close-up view along the membrane surface of the SRP-RNC complex using molecular models. Ribosomal parts around the polypeptide exit tunnel are given in light blue with the exception of the two proteins rpL25 and rpL35 (dark purple and light purple, yeast nomenclature corresponding to L23p/L29p families) contacting SRP54N, and rRNA helix H24 (purple) contacting SRP54M. The nascent chain (yellow) is modelled in the polypeptide exit channel and into the hydrophobic groove of SRP54M which directly resides upon the exit site. The positively charged N-terminus of the signal sequence (red sphere) is in close proximity to the negatively charged SRP RNA next to the tip of helix 8 as predicted previously [11]. Figures have been prepared with programs Iris Explorer, BOBSCRIPT [109] and Raster3D[110] by Klemens Wild.

Compared to the described free state of SRP, the cryo-EM structure of SRP bound to an active 80S ribosome reveals a strikingly different conformation of the SRP core, referred to as open conformation (fig. 20b-c, 21, 22). The conformational transition of SRP core is shown schematically in fig. 21. The SRP core is positioned directly at the tunnel exit site of the large ribosomal subunit and the dynamics within SRP54 upon binding the signal sequence on the ribosome indeed reflect the dynamic potential as inferred by the X-ray structures [53]. Moreover, the observed structural changes are immense and came as a surprise even with the knowledge of the intrinsic flexibility of SRP54. The superposition of the SRP core from both X-ray and EM data is shown in fig. 22. On the ribosome, SRP54NG is rotated by 50° and shifted about 50 Å away from the aligned position with RNA helix 8 and is found at the very tip of the SRP core instead (fig. 21, 22a-b). The two distal loops of SRP54N interact with the two ribosomal proteins rpL25 and rpL35 (corresponding to L23a/L35 in wheat germ and L23p/L29p in E. coli) in agreement with previous cross-link data [13,103,106] (fig. 20c). As the rest of SRP is fixed on the ribosome, the linker region has to accommodate the large conformational changes. One rigid anchor point between the resting and the moving part could be assigned by comparing different X-ray structures of SRP54M [53]. It localizes to the end of the linker region at the N-terminus of helix αM1 of SRP54M and corresponds very likely to a conserved leucine residue (L329 in Sulfolobus solfataricus) which is deeply buried in the hydrophobic core of SRP54M [53]. The significance of this residue is underlined by the finding that its mutation abolishes signal sequence binding [93]. Although it is not resolved at the present resolution, in the open conformation of SRP54 the signal sequence can be accommodated in the hydrophobic groove in an orientation, which positions one end of the signal sequence near the backbone of the SRP RNA helix 8 as proposed earlier [11]. Direct participation of the SRP54NG domain in signal sequence interaction as suggested before [111] appears unlikely in this position.

3.2.3.1 A flexible domain linkage between SRP54M and SRP54NG

When further comparing SRP54 in the free and the signal sequence-bound state on the ribosome, it is apparent that especially two regions forming hydrophobic contacts within SRP54 are of particular significance for its dynamic behaviour: (i) in the free state a peripheral loop of SRP54N establishes a contact with αM1b of SRP54M. This is the only direct contact between SRP54N and the signal sequence binding part of the M-domain. It contributes to the stabilization of the compact conformation in the free state and is lost in the ribosome-bound state of SRP54; (ii) the linker region connecting the SRP54NG and M-domains maintains a contact to helix αM1 in both, the free and the ribosome-bound state, and undergoes the largest conformational change between the two states. The linker region consists of three parts: the conserved 'LGMGD' sequence fingerprint in a loop connecting the G-domain with the linker, an 18 residue long linker helix (termed αML) and a loop of variable size up to the conserved leucine residue. The first loop (with two conserved glycines in the 'LGMGD' motif permitting rotational freedom) has to change its conformation upon ribosome binding while the linker helix αML seems to persist as a rigid body. The linker helix αML seems to persist as a rigid body although its relative orientation changes dramatically (almost 90°) with respect to both SRP54NG and SRP54M (fig. 20, 21b). In the EM density an empty tube, not accounting for any other parts of the separated domains, very likely represents this helix indeed spanning the distance from SRP54NG to SRP54M. The linker helix mainly interacts via hydrophobic interactions with helix αM1 and has already previously been implied to form a kind of greasy slide upon which the interface can be smoothly adjusted [53].

↓45

Figure 21: Dynamics of the SRP core.

Schematic view of the dynamic behaviour of the SRP core (RNA helix 8 and the SRP54 subunit) when switching from the free state (derived from the X-ray structure [53]) to the ribosome-bound state (derived from the cryo-EM structure). The ribosome engagement causes a 50° rotation and a 50 Å shift of the SRP54NG domain as indicated. The linker helix (orange), which connects the SRP54NG with the SRP54M domain, rotates almost 90°. The signal sequence is positioned in the hydrophobic groove of SRP54MN which directly resides upon the exit site and adjusts accordingly: helix αM1a moves towards the groove and helix αM1b and the finger loop are shifted out of the groove and form a lid over the helical signal sequence. The positively charged N-terminus of the signal sequence is in close proximity to the negatively charged SRP RNA next to the tip of helix 8 as predicted previously [11]. The colour code is the same as in fig. 20, except that the linker helix is shown in orange (RNA, red ribbon; SRP54NG, turquoise; SRP54MN, blue; SRP54MC, dark blue; signal sequence, green; linker helix, orange).

Figure 22: Dynamics of the SRP core with secondary structure.

(a) Superposition of the free and the ribosome-bound SRP core. The superposition is based on the rigid parts present in both structures (SRP54MC, and SRP RNA helix 8). The conformation of the free SRP54 as derived from the X-ray structure is shown in red colours, the ribosome-bound SRP54 placed in the cryo-EM data is depicted in blue colours. A putative signal sequence (helix represented as yellow cylinder) is modelled in the hydrophobic groove of SRP54M. The position of the fixed anchor point leucine (L329 in Sulfolobus solfataricus) between the MN and MC domain is given as a magenta sphere. (b) Movement of SRP54NG. The NG domain and the linker region are shown up to L329. The ribosome engagement causes a 50° rotation and a 50 Å shift as indicated. The linker region is highlighted: the conserved 'LGMGD' motif is given as a blue sphere. The linker helix (αML) rotates almost 90°. A flexible loop connects the linker to the anchoring leucine of SRP54M. The correct conformations of the two loops flanking αML remain elusive. (c) Conformational changes in SRP54M during signal sequence binding. The M domains are shown from the anchoring leucine to the C-terminal helix αM5. The model for SRP54M with bound signal sequence is based on a comparison of the M domain with a closed hydrophobic groove as seen in the free SRP54 [53] with an open structure [1] adjusted at several hinge points [53] (coloured spheres) to fit the EM data. Labelling is for SRP54MN with bound signal sequence only. Helix αM1 rotates towards the groove and helix αM1b and the finger loop are shifted out of the groove and form a lid over the helical signal sequence. Figure has been prepared by Klemens Wild.

Helix αM1 seems to undergo a rotation around the already mentioned conserved leucine residue and is part of the signal sequence binding groove (fig. 22c). This allows for the accommodation of a signal sequence in the orientation mentioned before, which positions one end of the signal sequence near the backbone of the SRP RNA helix 8 as proposed earlier [11]. Signal sequence accommodation is facilitated by adjusting the subsequent helix αM1b and the following finger loop. In agreement with the previously proposed model [53] the adjustments necessary to fit the EM data involve a conserved GP/PG tandem motif flanking helix αM1b and the N-terminus of helix αM2, which marks the start of the rigid C-terminal half of SRP54M. Similar to the 'LGMGD' loop, also the second loop, which connects the linker helix αML with SRP54MN, has to change its conformation in order to bring the linker helix in the correct position. This loop may also be involved in signal sequence binding as it is very close to the hydrophobic groove and often contains methionine residues. The linker region ends with a deeply buried conserved leucine residue (L329 in Sulfolobus solfataricus) serving as a rigid anchor point at the N-terminus of helix αM1 of SRP54M [53].

↓46

When the N-domain interaction with SRP54M is lost in the ribosome-bound state, the linker region appears to provide the only physical link between the signal sequence-binding domain (SRP54M) and the GTPase domain (SRP54NG). Therefore, one can easily imagine that the change in the position of the N domain or a conformational change of the linker region is transmitted to the signal sequence binding groove and vice versa.

3.3 Structure of the signal recognition particle receptor interacting with the SRP-RNC complex

Following the protein targeting cycle the SRP-RNC complex docks in the next step to the membrane-bound SRP receptor (SR) in GTP dependent manner. Comparing of cryo-EM structures of RNCs in complex with the translocon [99] and SRP reveals that both, SRP and the translocon, bind at the tunnel exit of the ribosome in a mutually exclusive way. Therefore, the question remains how the transfer of the signal sequence and the entire RNC from SRP to the translocon is facilitated. Hence, our aim was to determine the structure of a SR-SRP-RNC complex.

3.3.1  Reconstitution of SR-SRP-RNC complex

The SR-SRP-RNC complex was reconstituted in vitro in an analogous way to the reconstitution of the SRP-RNC complex. Stalled RNCs were used for the reconstitution with excess amounts of purified canine SRP and human SR. Sucrose density gradient centrifugation under high salt conditions (500 mM KOAc) confirmed specific and high affinity binding of SRP and SR to RNCs with an estimated occupancy between 70-80% (fig. 23). Since the SRP-SR interaction is GTP dependent, in the reconstitution reaction GMP-PNP was used, a nonhydrolisable analogue of GTP. In the absence of GTP analogues no stabile SRP-SR interaction was observed.

↓47

Figure 23: Purified DP90 RNCs were reconstituted with an excess of SR and SRP, subjected to sucrose density gradient centrifugation in the presence of 500 mM KOAc, fractionated and analyzed by SDS-PAGE and Sypro-Orange staining. SR-SRP-RNC complexes migrate in the 80S fraction indicating SR and SRP binding to RNCs. S, supernatant; p, pellet.

3.3.2 Cryo-EM map of SR-SRP-RNC complex

The preliminary cryo-EM map shows the typical appearance of an 80S ribosome at 9.1/6.6 A (0.5/3σ cut-off of FSC curve) with additional density, stretching from the peptide exit site to the intersubunit space, similar to the SRP density observed before (fig. 24). The part of the density, previously described as Alu-domain, appears at higher contour levels and is better resolved than the density around peptide tunnel exit site consisting of SRP S-domain and SR. This is most likely due to higher rigidity of Alu-domain which does not go through conformational changes after the interaction of SRP-RNC complex with SR. The Alu-domain resolution, which is close to that of the ribosome, is indicating high occupancy of the ligands in the final reconstruction. For this, a subset of 75% of the particles was used as a result of sorting according to the presence of ligands. The differences in the appearance of S-domain and Alu-domain densities, which is accompanied by a lower resolution, might reflect the higher flexibility of the S-domain in the newly formed complex after SR interaction.

Figure 24: Cryo-EM maps of mammalian SRP bound to 80S RNC at 12 Å and mammalian SR and SRP bound to 80S RNC at 9.1 Å. Ribosomal and SRP/SR densities are shown at different contour levels and amplitude correction using B-factors was applied only to the ribosome.

a, RNC-SRP map showing the separated colour-coded densities. Yellow, 40S small ribosomal subunit; blue, 60S large ribosomal subunit; green, P site tRNA; red, SRP. C1-C6 assigned positions of RNC-SRP connections (see also table 1); h1 and h2 are hinges of the 7S RNA backbone of SRP; St, stalk; SB, stalk base. b, SR-SRP-RNC map with separated colour coded densities. Green SR-SRP complex. The arrow is indicating delocalized NG domain. c, As a but rotated upward by 90°. d, As b but rotated upward by 90°.

↓48

The comparison of the electron density of SR-SRP-RNC complex with the density of SRP-RNC complex reveals several differences in the S-domain region of SRP. First, the S-domain of SRP is rearranged on the 60S subunit by shifting and tilting as seen in fig. 25. As a result of the shift SRP-SR complex moves slightly away from the peptide exit site and at the same time, as a consequence of the tilt, it comes closer to the ribosome, especially the protein SRP19 as seen in fig. 25b. This rearrangement could be necessary for the interaction with translocon in the next step of protein targeting.

Figure 25: Cryo-EM map of mammalian SR-SRP-RNC complex at 9.1 Å superimposed with SRP density from SRP-RNC complex.

Densities are colour coded as in figure 24. a, Peptide exit site view of the ribosome. The arrow is indicating shift of SR-SRP complex to the right related with SRP. b, View from the back of large subunit. The arrow is indicating a tilt of SR-SRP complex towards the ribosome.

The second main difference between the two densities is the apparent delocalization of both NG-domains. The density for these two components is not visible and it can be seen only as a very weak undefined mass at low contour levels (fig. 24b). Upon SRP-RNC complex interaction with SR in presence of GTP, the SRP54 NG-domain forms together with the SRα NG-domain a heterodimeric complex. Delocalization of these two domains dissociates SRP from binding site C1 which involves ribosomal proteins L23e and L35. These two proteins provide also the binding site for the translocon and, therefore, clearing of the binding site C1 allows spatial access of the translocon.

↓49

The third clear difference compared to the SRP structure is extra density visible as part of S domain around the previously described connection C4 and the RNA kink. This density is present on both sides of SRP RNA and it can be identified as SRα and SRβ. SRα consists of three domains, two of which are visible in the cryo-EM electron density, while the third, NG-domain, is disordered. The density for SRX1-domain, the N-terminal domain of SRα, is well resolved and allows unambiguous docking of the crystal structure [81] (fig. 26). The positively charged SRX2-domain, a long linker domain, stretches parallel along the SRP RNA and connects SRX1-domain with the SRNG-domain as seen in fig. 26. The docking of the SRα results in steric clashes of SRβ with other structures when in complex with SRα as in the crystal strucuture. This leads to the assumption that SRβ may dissociate from SRα and bind to the SRP and the ribosome at different location. The additional density which is not present in the SRP-RNC complex can be actually identified as SRβ and preliminary model can be built.

Figure 26: Cryo-EM map of mammalian SR-SRP-RNC complex at 9.1 Å.

Peptide exit site view of the ribosome. Densities are colour coded as in figure 23. a, SR-SRP complex density with pointed out densities belonging to SR. SRa(X1), X1 domain of SRa, SRa(X2), X2 domain of SRa. b, Transparent density of SR-SRP complex with preliminary models for SRb (yellow) and X1 domain of SRa (red).

3.4 High resolution structure of the ribosome and localization of L30e

The present cryo-EM map of the SRP-RNC complex with its resolution limited to approximately 12 Å leaves several interesting questions unresolved. How does SRP54 bind the signal sequence and how does that correlate with GTP binding to the SRP54 NG-domain? How does the interaction between N- and G-domain change the GTP affinity of the G-domain? To get more insights into these problems, a 3D reconstruction of the SRP-RNC complex at higher resolution was aimed at. Although the resolution of the map improved to 9Å with alpha-helical secondary structure resolved in the ribosomal density, the density of the SRP is not at a sufficiently high resolution yet. Since the resolution of the ribosomal density allows fold recognition, the eukaryotic ribosomal protein L30e was located in the 60S ribosomal subunit.

↓50

L30e binds to the transcript of its own gene to inhibit splicing to mature mRNA [112,113] and to reduce translation [114]. This prevents accumulation of L30e in excess of amounts needed to assemble ribosomes [115]. Like other ribosomal proteins, L30e has been highly conserved through evolution. Yeast L30e is 63% identical to wheat germ (and mammalian) and 33% identical to archaeal L30e. All these L30e proteins have a highly conserved structure lacking major insertions or deletions most likely due to requirements of its interaction with ribosomal RNA and proteins. The functional conservation allows archaeal L30e to bind to the RPL30 transcript of yeast resulting in inhibition of splicing [116].

Recently, several crystal and NMR structures of yeast and archaeal L30e have been solved, however, leaving open the localization of L30e on the ribosome. Based on its interaction with its own mRNA, there have been several attempts to assign the RNA region for L30e binding to the ribosome. As one candidate site of interaction the 25S RNA, region 830-862 (helix 34) was identified [116] in S. cerevisiae on the basis of strong binding of L30e to this helix when isolated. As another candidate, helix 38 of ribosomal 25S RNA was identified as the likely binding site for L30e by satisfying the consensus requirements for RNA kink turns [117]. This characteristic structure is required to bind L30e and, in the case of the large ribosomal subunit of H. marismorturi which lacks L30e, no interaction of this region with any other ribosomal protein has been observed [117]. Thus, the questions remain where L30e is located in the ribosome and what is its essential function there?

Figure 27: Localization of L30e in a 9.5 Å cryo-EM map of the 80S ribosome.

a, Cryo-EM density of the wheat germ 80S ribosome. The separated ribosomal subunits, tRNA and L30e are colour-coded. Yellow, 40S ribosomal subunit; blue, 60S subunit; green, P-site tRNA; magenta, L30e; landmarks: SB, stalk base; CP, central protuberance. b, As a but rotated by 150° to the right. c, L30e as part of isolated 60S subunit density. SRL, sarcin ricine loop. Top insert indicates the orientation of the subunit. d, L30e shown with isolated 40S subunit density. Sh, shoulder; Pt, platform. Bottom insert indicates the orientation of the subunit

↓51

With the high resolution cryo-EM map of the 80S wheat germ ribosome it is now possible to determine the localization of L30e based on its characteristic tertiary fold. The map shows the ribosome at 9.5/6.9 Å resolution (0.5/3sigma cutoff in FSC curve) with α-helical secondary structure of proteins clearly resolved (fig. 27a). This map was generated by improving the resolution of a previous SRP-RNC complex map through extension of the dataset to 52,000 particles and using the best 21,000 particles for a final 3D reconstruction. The ribosome is in the posttranslational state with peptidyl-tRNA bound in the P-site. Since helix 34 of 25S RNA was predicted as one of the binding sites [116] it was localized in the map and the L30e fold was recognized in its vicinity. This localization was then confirmed with the signature search [96] determining L30e’s position in the 60S subunit participating in the interface between large and small ribosomal subunit (fig. 27b-d). The yeast crystal structure was used for the search of L30e (1NMU chain D) [97]. The same crystal structure was also used for homology modeling of the wheat germ L30e (3D-JIGSAW) [118]. When comparing the electron density with the model or the crystal structure it is apparent that different conformations of the region between residues 70 and 86 of L30e exist. On the ribosome, the helix 4 in this region is flipped down which allows contacts of L30e with different ribosomal proteins and RNA. The same region of L30e is already known as highly flexible from different crystal and NMR structures of isolated L30e in which it adopts different conformations [97]. Another difference between the conformation of L30e in the ribosome and in isolation is a minor shift of the two helices next to the N- and the C-terminus towards helix 4 (fig. 28a-d). The homology model was adjusted accordingly in those two regions to fit the EM density (fig. 28c-d). It was completed by positioning missing residues of the N- and the C- terminus in corresponding density of the EM-map. Two residues of the C-terminus were added, where the density indicates an interaction with β-strand 3, and three residues of the N-terminus were added reaching over to the small ribosomal subunit (fig. 28c-d). Due to the limited resolution of the map, these N- and C-terminal residues as well as the loops connecting helix 4 of L30e could not be positioned precisely in an unambiguous manner. The wheat germ ribosomal proteins contacting L30e were homology modeled (3D-JIGSAW) and adjusted to the EM map. Ribosomal RNA in this region was adjusted to the EM map using the H. haloarcula (for 25S) and the T. thermophilus (for 18S) model used before for the yeast 80S ribosome [100].

Figure 28: Molecular model of L30e and its ribosomal environment.

a, Isolated L30e density (contour at ~2.5 sigma) in a “side” view (top) and rotated 90° to the left (bottom). b, As a but shown at lower contour level (~1 sigma). N-term, N-terminus. c, Adjusted homology model of wheat germ L30e shown with transparent EM density. C-term, C-terminus; αh 2-4, α-helix 2-4. d, Crystal structure 1NMU_D shown to illustrate conformational differences. Note the different position of the α-helix 4. e, Molecular environment of L30e in the ribosomal interface. Blue, 25S RNA; orange, L37Ae, magenta, L30e; yellow, 18S RNA; green, S13e. The inset indicates the orientation of the view. f, As e but rotated by 90° upwards. g, Comparison of L30e in a 80S ribosome in the post state and in the eEF2-bound state. The 60S subunit of the wheat germ 80S ribosome map (post state) is shown at 9.5 Å resolution (left) and filtered to 13 Å (center) together with the 60S subunit of the yeast 80S ribosome-eEF2 complex at 11.7 Å (right). Landmarks as in fig. 27. B4 and eB9 indicate the location of intersubunit bridges. Note the different location of the density for the helix 4 of L30e in EF2-ribosome complex.

Located in the ribosomal intersubunit space, L30e makes several connections to both, proteins and RNA, of both subunits. In the 60S subunit, L30e contacts two RNA helices (table 2). The largest and most important contact is formed between aa residues 24-28 and 86-88 of L30e and RNA helix 58 of 25S RNA (nt 1591, 1603-1605, H. marismorturi model). In addition, the tip of the helix 4 of L30e (aa 79-81) comes into close proximity to RNA helix 34 (nt806-808) and likely forms a second contact with 25S RNA (fig. 28e-f). Two more protein-protein contacts exist: the loop between β-strand 2 and α-helix 3 (aa 46–48) of L30e interacts with a loop of the ribosomal protein L37Ae (aa 41-43). Yet another contact of L30e in the 60S subunit involves a unknown ribosomal protein which is part of the previously described cluster II [100].

↓52

Interacting with the 40S subunit, L30e contacts one 18S RNA helix and one ribosomal protein (table 1). The N terminus of L30e contains 3 lysines (aa 1, 3 and 4) and interacts with the helix 22 of 18S RNA (661-662, T. thermophilus model) forming a relatively week connection visible only at lower contour levels (fig. 28e-f). The helix 4 of L30e forms two connections with two helices of the ribosomal protein S13. The tip of the helix 4 (aa 77-81) connects to the C terminal part of S13 (aa 143-145) which is also interacting with the helix 34 of 25S RNA of large ribosomal subunit. The residues 74-76 of the helix 4 interact with a second S13 helix involving aa residues 92-94. Finally, the N and C terminal part of L30 appear to form another weak contact to yet another helix of S13 (aa 77-79). Many of these residues are shown to be evolutionary conserved between archaeal and yeast L30e [117].

Although helix 34 and helix 38 of 25S RNA were predicted as the most likely ribosomal binding sites for L30e, ribosomal RNA helix 58 turns out to be the main binding site. The interaction with helix 34 is significantly weaker (tip of helix 4 of L30e) and it is possible only in this conformation of L30e (see below). In agreement with our findings L30e has been shown to bind to fragments of both helices (34 and 58) in vitro [116]. Moreover, RNA helix 58 indeed meets the conformational requirements suggested previously [117] and binds L30e via a RNA kink turn.

In the determined position L30e appears to be the only 60S constituent of the intersubunit bridge eB9 [99,100] and may contribute also to the bridge B4 formed mainly by 25S RNA helix 34 on the 60S side. It has been shown previously that lack of L30e in S. cerevisiae leads to stalled initiation complexes suggesting a subunit joining defect [119]. A role of L30e in subunit joining or 80S stabilization could therefore be explained by its function in bridge formation. Interestingly, bridge eB9 formed by L30e is dynamic since it is present in the post translational state of the ribosome but absent in the eEF2-bound state [120]. Moreover, when comparing our present EM map in the post state with the map of the 80S ribosome-eEF2 complex at similar resolution, it is evident that the flexible region of L30e involving helix 4 adopts a different conformation in this functional state of the ribosome. Here, the contact to B4 is lost and the conformation of the flexible region is likely to be more similar to the conformation observed in the crystal structure(1NMU) [97]. Hence, it is possible that the conformation of the flexible region of L30e is related to the conformational states of the ribosome and that L30e plays a role in facilitating the ratchet movement of the 80S ribosome during the translation cycle. Therefore, in contrast to the majority of ribosomal proteins which play a role in merely fine tuning the 3D structure of ribosomal RNA, L30e appears to have adopted a direct function in controlling large scale conformational changes of the ribosomal machinery.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
13.12.2006