Characterization of DbpA, an
Escherichia coli
DEAD box protein with ATP independent RNA unwinding activity
Characterization of DbpA, an Escherichia coli DEAD box protein with ATP independent RNA unwinding activity
Nina
Böddeker
+
,
Katrin
Stade
[sect]
and
François
Franceschi*
Max-Planck-Institut für Molekulare Genetik, AG Ribosomen, Ihnestraße 73, 14195
Berlin
,
Germany
Received October 14, 1996
Revised and Accepted December 2, 1996
ABSTRACT
DbpA is a putative
Escherichia coli
ATP dependent RNA helicase belonging to the family of DEAD box proteins. It
hydrolyzes ATP in the presence of 23S ribosomal RNA and 93 bases in the
peptidyl transferase center of 23S rRNA are sufficient to trigger 100% of the
ATPase activity of DbpA. In the present study we characterized the ATPase and RNA unwinding activities of DbpA in more detail. We report that-in contrast to eIF-4A, the prototype of the DEAD box protein family-the ATPase and the helicase activities of DbpA are not
coupled. Moreover, the RNA unwinding activity of DbpA is not specific for 23S
rRNA, since DbpA is also able to unwind 16S rRNA hybrids. Furthermore, we
determined that the ATPase activity of DbpA is triggered to a significant
extent not only by the 93 bases of the 23S rRNA previously reported but also by
other regions of the 23S rRNA molecule. Since all these regions of 23S rRNA are
either part of the `functional core' of the 50S ribosomal subunit or involved
in the 50S assembly, DbpA may play an important role in the ribosomal assembly
process.
INTRODUCTION
Specific RNA-protein interactions often depend on so called `RNA-chaperones' (
1
), proteins capable of altering RNA secondary structures. These `helper-proteins' presumably destabilize or unwind RNA molecules to give them the
appropriate structure needed for the correct binding of a specific protein.
Members of the DEAD box protein family belong to these `helper-proteins', and are characterized by the `DEAD' motif (Asp-Glu-Ala-Asp) as well as by seven other highly conserved amino
acid motifs (
2
). Proteins of this family hydrolyze ATP only in the presence of RNA. New
members classified as DEAD box proteins based on sequence homologies alone are
generally designated `putative' ATP dependent RNA helicases, and only a few of
those have actually been shown to unwind RNA experimentally. The eukaryotic
Initiation Factor 4A (eIF-4A) (
3
,
4
) is a true ATP dependent RNA helicase and is generally recognized as the
`prototype' of the DEAD box protein family. eIF-4A plays a key role in unwinding secondary structures of mRNA molecules
thus allowing the subsequent binding of the 40S ribosomal subunit to the mRNA
which is a prerequisite for the formation of the pre-initiation complex in protein synthesis (
4
,
5
). Unlike eIF-4A, however, some DEAD box proteins appear not to be ATP dependent. An
example of this is CsdA (also known as deaD;
6
), a cold-shock protein in
Escherichia coli
and a member of the DEAD box protein family, which was recently shown to be
able to destabilize RNA duplexes in the absence of ATP (
7
).
To date, five different genes encoding DEAD box proteins have been identified in
E.coli
(
8
). Among them, DbpA has been designated a `putative' RNA helicase and has been
suggested to play a role in protein biosynthesis and/or ribosome assembly. DbpA
hydrolyzes ATP only in the presence of bacterial 23S rRNA (
9
), and recently it was determined that a region of 93 nucleotides (nt) at the
peptidyl transferase center of 23S rRNA is sufficient to stimulate the ATPase
activity of the protein (
10
). The helicase activity of DbpA, however, and its possible effect on protein
synthesis or ribosomal assembly has not yet been established. Thus, we
characterized DbpA further by determining its ability to destabilize or unwind
RNA-RNA and DNA-RNA hybrids. We report here that DbpA indeed possesses an RNA
unwinding ability. This RNA unwinding or helix destabilizing activity is not
coupled to ATP hydrolysis. By carefully scanning the whole 23S rRNA molecule we
found that, in addition to the previously identified 93 bases at the peptidyl
transferase center, other regions of the 23S rRNA rich in secondary structure
are also able to stimulate the ATPase activity of the protein to significant
levels. Since all these regions of 23S rRNA are either part of the functional
core of the 50S ribosomal subunit or involved in the 50S assembly, we propose a
function of DbpA in the assembly process of this subunit.
MATERIALS AND METHODS
Materials
Enzymes were obtained from Boehringer Mannheim except RNAse H, which was a
generous gift from Dr R. Brimacombe. Radiochemicals were obtained from
Amersham. Oligonucleotides were synthesized from TIB Molbiol (Berlin).
Cloning, overexpression and purification of DbpA
The gene coding for DbpA (GenBank accession No. X52647) was amplified from
E.coli
DNA by PCR, and after confirming the sequence
,
subsequently cloned into a pET11a expression vector (Novagen). BL21(DE3)LysS
cells (
11
) were transformed with the plasmid and grown at 37oC until they reached an OD
560 nm
of 0.4. Overexpression of the protein in a 2 l culture was induced by the
addition of IPTG to a final concentration of 1 mM, and a temperature shift from
37 to 30oC in order to avoid formation of insoluble inclusion bodies. After
induction the culture was grown for 5 h after which cells were pelleted by
centrifugation (3000
g
, 10 min), resuspended in 20 mM HEPES pH 8.0, 10 mM MgCl
2
, 300 mM NaCl and disrupted in a French press. Cell debris was removed by
centrifugation for 30 min at 10 000 r.p.m. in a Ti45 Beckmann rotor. The
supernatant was cleared of ribosomes by another centrifugation step (Ti45
Beckmann rotor, 35 000 r.p.m., 16 h). The clear supernatant was dialyzed
against 20 mM HEPES pH 8.0, 125 mM NH
4
Cl and applied to a 1.5 * 10 cm DEAE-Sepharose column (Pharmacia). The flow-through was collected, precipitated by the addition of 60%
(NH
4
)
2
SO
4
and redissolved in 5 ml 20 mM HEPES pH 8.0, 50 mM NH
4
Cl. After dialysis against the same buffer the protein was further purified on
an 1.8 * 100 cm ACA44 gel filtration column (IBF Biotechniques). Fractions
containing DbpA were collected and dialyzed against 20 mM HEPES pH 8.0, 1.5 M
(NH
4
)
2
SO
4
. The resulting protein solution was finally applied to a FPLC Phenyl-Superose HR 5/5 (1 ml) column (Pharmacia) and eluted using a linear 1.5 to
0 M (NH
4
)
2
SO
4
gradient with a flow rate of 0.5 ml/min. All purification steps were analyzed
on SDS-PAGE (
12
). Purified DbpA was dialyzed against 20 mM HEPES pH 8.0, 50 mM NH
4
Cl and the concentration was determined according to Bradford (
13
). The protein solution was frozen in liquid nitrogen and stored at -80oC.
RNA-binding assay
The binding of DbpA to RNA was determined by the retention of
32
P-labeled RNA on nitrocellulose filters. Reactions contained 20 pmol protein
and 50 pmol RNA unless otherwise stated. Reactions were carried out in binding
buffer (20 mM HEPES pH 8.0, 5 mM Mg-acetate, 50 mM NH
4
Cl, 1 mM DTT) in a final volume of 100 [mu]l. When 200 mM ammonium chloride was present in the assay, the binding of
rRNA was reduced to 20% compared to the binding in the absence of ammonium.
Therefore, 50 mM ammonium chloride was chosen arbitrarily for these assays. The
ammonium dependence of the binding remained consistently the same regardless of
the kind of rRNA (23S, 16S or 5S rRNA) used in the assays. Under the conditions
used, RNA is not retained on nitrocellulose filters unless it is bound to
protein. The reaction mix was incubated at 37oC for 15 min, then stopped by the addition of 2 ml ice-cold wash buffer (same as binding buffer with the addition of 0.1 mM
EDTA) and immediately applied to nitrocellulose filters. The reaction tubes
were rinsed twice with 2 ml ice-cold wash buffer. Filters were dried under an infrared lamp and
radioactivity was determined by liquid scintillation counting.
Cross-linking of DbpA to
32
P-labeled ribosomal RNA and [
[alpha]
-
32
P]ATP
The cross-linking experiments were performed based on Pause
et al
. (
14
). DbpA was cross-linked to either 16S or 23S
32
P-labeled rRNA, or [[alpha]-
32
P]ATP. The amount of DbpA used for all these experiments was 20 pmol. For the
cross-link of DbpA to ATP, 100 pmol [[alpha]-
32
P]ATP (1 [mu]Ci) were incubated with DbpA for 5 min at 37oC in Tris-HCl pH 7.5, 50 mM KCl, 10% glycerol, with or without the
addition of unlabelled tRNA or 23S rRNA. For the cross-link of DbpA to rRNA, 50 pmol (1 [mu]Ci) of 16S or 23S
32
P-labeled rRNA were incubated with DbpA for 5 min at 37oC in 20 mM HEPES-KOH pH 7.5, 10 mM NH
4
Cl, 5 mM Mg-acetate and 1 mM DTT, with or without the addition of unlabelled tRNA
and/or ATP. The reaction mixtures were then placed on ice and UV-irradiated for different times (Fig.
1
), using four 65W lamps (Sylvania G8T5, 330nm) placed 2 cm away from the
samples. After irradiation, the reaction mixtures were incubated with 10 [mu]g RNAse A for 10 min at 37oC. Cross-linked products were analyzed by autoradiography after separation
by 10% PAGE.
Preparation of ribosomal subunits and
in vivo
32P-labeled ribosomal RNA
Ribosomal subunits were prepared according to Rheinberger
et al
. (
15
). Ribosomal RNA was prepared by phenol extraction following a standard protocol
(
1 6
).
Conditions for the preparation of
32
P-labeled ribosomes were as described by Stiege
et al
. (
17
). Cultures of
E.coli
MRE600 (40 ml) were inoculated with 2 ml of an overnight preculture and 5 mCi ortho-[
32
P]-phosphate was added. Cells were grown until the amount of incorporated
radioactivity reached 80% and then harvested by centrifugation. Bacteria were
lysed by sonication and cell extracts were layered onto 10-40% linear sucrose gradients in 10 mM Tris-HCl pH 7.8, 0.3 mM Mg-acetate, 150 mM KCl and then centrifuged at 4oC in a Beckmann SW27 rotor for 16 h at 20 000 r.p.m.
After gradient fractionation, radioactivity of the fractions was determined and
fractions containing 50S and 30S ribosomal subunits, respectively, were pooled and precipitated with 2 vol ethanol.
32
P-labeled rRNA was obtained from the radioactive labeled subunits by phenol
extraction (
16
).
RNAse H digestion of ribosomal RNA
Defined RNA fragments were obtained by digestion with ribonuclease H (
18
). Pairs of DNA oligonucleotides (10-15mers) containing complementary sequences to 23S rRNA were mixed in
equimolar amounts with the target RNA in 15 mM Tris-HCl pH 7.8, 50 mM NH
4
Cl, 1 mM MgCl
2
, 0.1 mM DTT and heated to 55oC for 10 min. A predetermined amount of RNAse H was added and samples were
incubated for a further 30 min at 55oC. The reaction was stopped by the addition of EDTA and SDS to a final
concentration of 1 mM and 0.1%, respectively. RNA fragments were separated on
5% denaturing polyacrylamide gels and cut out under UV light. Extraction from
the gel was achieved by stirring gel pieces overnight at 4oC in 10 mM Tris-HCl pH 7.8, 2 mM EDTA, 0.1 M Na-acetate, 0.1% SDS and 50% of phenol. RNA fragments were
recovered by ethanol precipitation (
16
).
Preparation of RNA/RNA hybrids
Fragments of 16S and 23S rRNA were generated using the `RNAse H method' as
described above. To allow the formation of RNA hybrids, fragments with
complementary sequences were combined in each assay. In order to follow hybrid
formation and subsequent unwinding, one of the two fragments was
32
P-labeled (see preparation of
32
P-labeled ribosomes). For lengths and positions of RNA sequences see Table
2
. Hybridization was achieved as follows: 50 pmol of each RNA fragment were mixed
in a buffer containing 20 mM Tris-HCl pH 7.6, 150 mM KCl and heated to 65oC for 2 min. After slow cooling to 37oC, samples were divided into three parts, one of which served
as a control for successful hybridization and the other two for monitoring
unwinding activity. To each of these samples 30 pmol DbpA was added, either in
TMK buffer (20 mM Tris-HCl pH 7.6, 2.5 mM MgCl
2
, 50 mM KCl) or in TMAK buffer (same as TMK with the addition of 1 mM ATP).
Samples were incubated at 37oC for 20 min, chilled on ice and separated on 7.5% non-denaturing polyacrylamide gels at 4oC. Gels were dried and subjected to autoradiography.
Preparation of DNA/RNA hybrids
DNA oligonucleotides were selected based on their ability to bind to distinct
regions of the 16S or 23S rRNA and were labeled at their 5'-end using [[gamma]-
32
P]ATP and T4 polynucleotide kinase (
16
). 20 pmol labeled oligonucleotide was mixed with 35 pmol
32
P-labeled 16S or 23S rRNA in 20 mM Tris-HCl pH 7.6, 150 mM KCl, heated to 55oC for 15 min and slowly cooled to 37oC. The sample was divided into three aliquots and the
subsequent unwinding reaction was carried out as described above. Samples were
separated on 3% composite acrylamide-agarose gels (
19
) and gels were dried and subjected to autoradiography.
ATPase activity assay
ATPase activity was determined using 15 pmol DbpA and various amounts of 23S
rRNA or 23S rRNA fragments as indicated in the figure legends. The reaction
contained 50 mM Tris-HCl pH 7.5, 5 mM MgCl
2
, 1 mM DTT, 0.1 mM ATP and 0.05 [mu]Ci [[gamma]-
32
P]ATP in a final volume of 50 [mu]l. The reaction mixture was incubated at 37oC for 30 min and stopped by the addition of 300 [mu]l 7% activated charcoal powder (Merck) suspended in 50 mM HCl and 5
mM H
3
PO
4
. Samples were mixed and centrifuged at 10000 r.p.m. for 10 min in a tabletop
centrifuge. Aliquots (250 [mu]l) were taken from the supernatant and the radioactivity was measured using
a liquid scintillation counter (Packard). The background radioactivity was
determined in a similar reaction without protein and subtracted from all other
reactions. Obtained values were converted to pmol [[gamma]-
32
P
i]
released/min.
RESULTS
Binding of rRNA and ATP to DbpA
DbpA binds to various RNAs, but hydrolyzes ATP only in the presence of 23S rRNA,
one of the three rRNAs found in prokaryotic ribosomes (
9
). However, the extent of binding of DbpA to the remaining two rRNA molecules,
the 16S rRNA and the 5S rRNA, has not been reported. Thus, we purified rRNA and
DbpA to homogeneity and tested the extent to which it was able to bind to
isolated 23S, 16S and 5S rRNAs in a filter binding assay. DbpA bound to
radioactive labeled 5S, 16S as well as 23S rRNA molecules (Table
1
). Although binding of DbpA to rRNA was dependent on the ammonium salt
concentration (data not shown), DbpA showed no binding preference for any of
the rRNA species tested under the ionic conditions used. Radioactive labeled
rRNAs could be chased by adding unlabeled competitor rRNA.
5'-radioactively labeled 23S, 16S and 5S rRNA (50 pmol) was incubated
with DbpA (20 pmol) and the extent of binding was determined by filter binding assays (see Materials and Methods). No significant difference could be detected in the
binding of DbpA to the different rRNAs.
.
DNA probes and rRNA fragments used in the helix destabilizing experiments
Hybrid type
23S rRNA
16S rRNA
DNA-rRNA
175-188
760-776
(DNA-probe hybridized
1415-1430
996-1012
to rRNA at position)
1704-1720
1356-1376
2528-2552
1391-1406
rRNA-rRNA
1625-1736
1302-1453
(rRNA fragment pairs forming
1966-2054
1453-1542
doubled stranded hybrids)
2442-2546
1416-1445
2546-2670
1454-1494
2494-2528
2532-2587
DNA-rRNA hybrids, DNA oligonucleotides complementary to the rRNA positions listed above were hybridized to the respective rRNA.
RNA-RNA hybrids, RNA fragments corresponding to the positions shown were generated by RNAse H digestion of either 23S or 16S rRNA. Pairs of
fragments known to hybridize based on the secondary structure model of 16S and 23S rRNA
were used to generate the RNA-RNA hybrids.
DNA-rRNA hybrids as well as the pairs of rRNA fragments shown in bold in this
Table correspond to the fragments shown in Figure 2. Numbers represent the base
number in the 16S or 23S rRNAs.
It has been well established that DbpA hydrolyzes ATP only in the presence of
23S rRNA, but not in the presence of 5S or 16S rRNA (
9
). To better understand the interaction of DbpA with ATP and rRNA, UV cross-linking experiments were performed by independently cross-linking radioactive ATP, 16S rRNA or 23S rRNA to DbpA. Competition
experiments were then carried out where cross-linking to radioactive ATP (Fig.
1
a) and rRNA (Fig.
1
b) were performed in the presence of increasing amounts of competitor, i.e.
unlabeled rRNA (23S rRNA or tRNA) or ATP, respectively. Our results indicated
that DbpA can be cross-linked to ATP, 23S and 16S rRNA and that these cross-links are competitive, since addition of unlabeled 23S rRNA or tRNA
greatly reduced the cross-link of DbpA to [
32
P]ATP (Fig.
1
a) and addition of unlabeled ATP or tRNA equally reduced the cross-link to [
32
P]-23S rRNA (Fig.
1
b).
Since DbpA showed no exclusive binding specificity for 23S rRNA, and cross-links of rRNA and ATP were competitive, we decided to investigate the
helicase activity of DbpA for 23S rRNA, 16S rRNA, as well as other RNA
molecules in the presence and absence of ATP.
Hybrid destabilizing activity of DbpA
Although DbpA has been described as a `putative' RNA helicase due to sequence
homology to other members of the DEAD box protein family (
9
), its helicase activity has never been proven. Thus, we investigated the
ability of DbpA to destabilize DNA-rRNA as well as rRNA-rRNA hybrids.
To test the destabilizing activity of DbpA for DNA-rRNA hybrids, we used radioactively labeled 16S and 23S rRNA in
conjunction with radioactively labeled DNA oligonucleotide probes selected in
base of their ability to hybridize to rRNA. We selected four DNA probes that
independently hybridized to 23S rRNA and four others that hybridized to 16S
rRNA. The DNA probes used in these experiments and the positions to which they
hybridize with rRNA are listed in Table
2
. DNA-rRNA hybrids were shown to be stable by PAGE (Fig.
2
a, lanes 5 and 8; Fig.
2
b, lanes 3, 5 and 7).
When radioactively labeled DNA-rRNA hybrids, DNA probe `A' + 23S rRNA or DNA probe `B' + 23S rRNA were
incubated with DbpA, the protein was capable of destabilizing both hybrids.
This can be seen in Figure
2
a by the appearance of the band corresponding to the DNA probes. Surprisingly,
addition of ATP was not necessary to trigger the destabilizing activity, but
instead seemed to slightly hamper the unwinding reaction as judged from Figure
2
a (compare lane 6 with 7 and lane 9 with 10). By performing the corresponding
experiments with 16S rRNA instead of 23S rRNA, we found that DbpA could
destabilize DNA probes `C', `D' and `E' hybridized to 16S rRNA as well (Fig.
2
b). DNA-16S rRNA hybrids showed the same response to ATP as described above (data
not shown). Thus, the destabilizing activity of DbpA appears to be independent
of the type of ribosomal RNA used, and ATP is not needed for the unwinding
activity.
23S rRNA regions able to stimulate the ATPase activity of DbpA
For mapping regions of 23S rRNA able to stimulate ATPase activity of DbpA, we used a different approach from that described by Nicol and Fuller-Pace (
10
). To screen individual regions of 23S rRNA, 26 different fragments were
generated by RNAse H digestion of native 23S rRNA. Thus, it was possible to
assay the ATPase activity with `native' instead of transcribed 23S rRNA,
avoiding the risk of incorrect folding of 23S rRNA during
in vitro
RNA synthesis by T7 RNA polymerase (
22
). Moreover, RNA fragments produced from native ribosomal RNA still contain all
modified nucleotides which presumably play an important role in the maintenance
of the structure and/or function of rRNA (
23
).
We identified a region of 23S rRNA similar to that reported by Nicol and Fuller-Pace (
10
), that was able to stimulate 100% of the protein's ATPase activity compared to
intact 23S rRNA (Fig.
3
, fragment D, nucleotides 2500-2600). In addition, however, we discovered that four other regions of 23S
rRNA were able to trigger up to 60% of the ATPase activity (Fig.
3
, fragments A, B, C and E). It can be seen in Figure
3
(bottom part) that the regions stimulating ATPase activity are not concentrated
in any particular portion of the molecule but are rather scattered all along
23S rRNA. Although these regions display no sequence consensus, their common
feature appears to be a high extent of stem-loop structures.
DISCUSSION
Proteins containing the conserved DEAD box motifs are assumed to hydrolyze ATP
while unwinding RNA (
25
). However, some DEAD box proteins need the interaction with other proteins in
order to exert their ATPase or helicase activities. For example, eIF-4A needs to interact with eIF-4B in order to display ATPase and helicase activity (reviewed in
26
). The
E.coli
protein RhlB also shows no ATPase activity unless it is part of the so-called `degradosome' (
27
,
28
). Here we show that DbpA-in spite of being a DEAD box protein-possesses a helix destabilizing activity which is ATP independent.
Thus, DEAD box proteins such as DbpA and CsdA (
7
) display ATPase independent RNA helix destabilizing activity, and apparently
need no interaction with other proteins in order to display ATPase or RNA
destabilizing activity. If we consider that five DEAD box proteins have been
found in
E.coli
to date (
29
), and that the two analyzed for helicase activity (i.e. DbpA and CsdA) can
destabilize RNA helices in the absence of ATP, it is clear that ATPase and
helicase activities are not necessarily coupled in all DEAD box proteins.
Koonin and Rudd (
30
) suggested that some proteins that have evolved from the helicase superfamily I
contain only the N-terminal motifs of the DEAD box proteins thought to be responsible for ATP
hydrolysis, and can exist separately from proteins containing only the C-terminal motifs needed for the helicase activity. DbpA has an unusual 70
amino acid C-terminal domain consisting of 25% positively charged residues and 29%
small (Gly/Ala) residues (
24
). Therefore, it is possible that the protein's C-terminal region is responsible for RNA binding without inducing ATP
hydrolysis, whereas the domain that induces the 23S rRNA dependent ATP
hydrolysis is linked to the central DEAD domain (
31
).
Zinc fingers and cysteine rich metal-binding domains are known to interact with nucleic acids (
32
). Interestingly, DbpA possesses a putative metal-binding cysteine-rich sequence motif:
Cys
-Val-Val-Phe-
Cys
-Asn-Thr-Lys-Lys-Asp-
Cys
-Gln-Ala-Val-
Cys
(Cys-X3- Cys-X5-Cys-X3-Cys) between the `SAT' and `ARGXD' DEAD-box conserved motifs. Thus, one could
speculate that DbpA has two different RNA binding domains, one of which is
specific for 23S rRNA probably due to the putative metal-binding motif, and the other is located at the C-terminal domain of the protein which is responsible for the non-specific RNA binding due to the positive charges of its
residues. If this putative metal-binding sequence lies structurally close to the ATPase (or DEAD) motif,
the dependence of the 23S rRNA specific binding and the ATPase activity of DbpA
can be explained.
Moreover, the fact that DbpA can be cross-linked to ATP in the absence of RNA (Fig.
1
a) and vice versa (Fig.
1
b) indicates that the binding of one of the substrates does not require the
binding of the other. In this respect, DbpA is different from eIF-4A where cross-linking to RNA is dependent on ATP hydrolysis and is the driving
force for productive binding to RNA (
14
). It is possible that ATP hydrolysis of DbpA is used to destabilize the
interaction between 23S rRNA and the putative metal-binding motif after RNA-unwinding or destabilization. It has been shown for RNA helicase A
for example, that the ATP consuming step is the release of the RNA (
33
). Moreover, since the binding of DbpA to RNA is rather non-specific (
9
) (Table
1
) and DbpA concentration in the cell is very low (
24
), a rapid turnover of DbpA could be ensured after its binding to 23S rRNA if
ATP was the driving force for binding release.
Frequently `non-ribosomal' proteins have been suggested to play a role in the ribosome
assembly process (
34
,
35
). These proteins could be a possible cause for the dramatic differences
observed between
in vivo
and
in vitro
ribosome assembly time, temperature and ionic conditions (
36
). For example, the chaperone DnaK has been implicated in ribosome assembly in
E.coli
(
37
). In addition, two putative RNA helicases (of the DEAD box protein family) have
been recognized in
E.coli
as gene-dosage dependent suppressors of temperature sensitive mutants in genes of
ribosomal proteins. One of these, SrmB, is a suppressor of a temperature-sensitive mutation in the ribosomal protein L24 (
38
), and the other is encoded by the deaD gene that codes for CsdA, a cold shock
ribosome-associated protein capable of suppressing a temperature sensitive mutation
linked to the ribosomal protein S2 (
6
,
7
). However, to date no DEAD box protein has been identified that has a direct
influence on the
E .coli
ribosome assembly process.
Our analysis of the 23S rRNA fragments triggering the ATPase activity of DbpA
shows that-aside from the region previously identified by Nicol and Fuller-Pace (
10
)-additional regions can stimulate this activity. It is interesting to note
that these regions are spread across the 23S rRNA molecule and show no apparent
consensus sequence. However, all of them are rich in stem-loop structures, and all of these 23S rRNA regions are either part of the
`functional center' of the 50S ribosomal subunit (
39
) or have been found to be related to ribosomal proteins playing a leading role
in the assembly process (
36
). The functional center of the 50S subunit or `core structure' is rich in
ribosomal RNA, contains nearly all of the 23S rRNA modified nucleotides found
so far, and has a complex structural organization (
23
,
39
). As an example, the ribosomal protein L23 cross-links to nucleotides A63 and U138 in helices 6 and 9 of 23S rRNA (
40
,
41
). These helices are part of fragment A shown in Figure
3
. Cross-linking experiments with a photoaffinity-analogue of the antibiotic puromycin identified L23 as a target
protein, and therefore, L23 is hypothesized to be localized in the vicinity of
the peptidyl transferase center of the 50S ribosomal subunit (
42
). Interestingly L24, one of the two 50S assembly initiator proteins (
43
) has also been cross-linked to fragment A, implicating this region as essential during the
early assembly steps.
Cross-links to L11 and `L8' (the pentameric complex formed by L7/L12 and L10)
have been found for fragment B (Fig.
3
). These proteins are an essential part of the GTPase center of the 50S subunit
also known as the `translocation' domain, the place where the elongation
factors have been mapped to (
44
,
45
). Footprint sites to tRNA (A-site) have been also found in helix 43 (Fig.
3
, fragment B) (
46
,
47
).
Fragment C (bases 1585-1603) of Figure
3
which has been identified in our experiments to stimulate the ATPase activity
of DbpA to a significant extent, contains parts of helices 51-54 of 23S rRNA where L23 binds to (
48
). Cross-linking experiments with nascent peptides identified position A1614 in
helix 59a, corresponding to the cross-linking site pX in fragment C (Fig.
3
), as being in close contact with the growing peptide chain when it leaves the
ribosome (
49
). Fragment C contains a modified base in position A1618. L9 has been cross-linked to helix 58 in fragment C (
41
), and antibodies against L9 inhibited protein synthesis almost completely (
50
). L9 has also been cross-linked to L28 (fragment A) and these two proteins (L9 and L28) have been
cross-linked to L2 (
51
,
52
), an essential component of the peptidyl transferase center (
53
,
54
).
Fragment D in Figure
3
belongs to the peptidyl transferase center and contains multiple footprint
sites for tRNA (A-site and P-site) (
46
,
47
) as well as many modified nucleotides (
23
,
55
,
56
).
Fragment E (Fig.
3
) shows a cross-link to ribosomal protein L32, which has been mapped to the vicinity of
the peptidyl transferase center (
57
) as well as to L3, that together with L24 is one of the two ribosomal proteins
that leads the assembly of the 50S subunit (
43
).
In summary, it seems reasonable to assume that all regions of 23S rRNA described
above are highly interrelated within the three-dimensional structure of the 50S ribosomal subunit. Therefore, we propose
that DbpA may play an important role in the assembly of the `functional center'
of the 50S subunit, possibly by assisting the correct folding of these regions.
At early assembly stages, DbpA could interact with the different regions of the
23S rRNA prior to the binding of ribosomal proteins to the 23S rRNA molecule.
At the assembly level, the interaction of DbpA with 23S rRNA could destabilize
certain regions of the 23S rRNA molecule, thus, lowering activation energy
barriers and allowing the binding of the ribosomal proteins under
thermodynamically favorable conditions. Based on all these findings as well as
on preliminary results on the effect of DbpA in
in vitro
assembly of the 50S subunit
(N. Böddeker, C. Glotz. and F. Franceschi, unpublished results), we hypothesize
that DbpA could play a major role in ribosome assembly, specifically in the
assembly process of the `active center' of 50S ribosomal subunits.
ACKNOWLEDGEMENTS
We thank C. Glotz and C. Paulke for valuable discussions and excellent technical
assistance. We also thank Drs K. Nierhaus and R. Brimacombe for helpful
discussions. We appreciate the help of Dr A. Hofmann for critical reading and
very helpful suggestions during the preparation of the manuscript.
18 Brimacombe,R., Greuer,B., Gulle,H., Kosack,M., Mitchell,P., Osswald,M., Stade,K. and Stiege,W. (1990) In Spedding,G. (ed.) Ribosomes and Protein Synthesis: A Practical Approach. IRL Press, Oxford, UK, pp. 131-159.
19 Rickwood,D. and Hames,B.D. (1982) Gel Electrophoresis of Nucleic Acids: A Practical Approach. IRL Press Limited, Oxford, UK.
*
To whom correspondence should be addressed. Tel: +49 30 8300 0774; Fax: +49 30
8413 1774; Email: Franceschi@mpimg-berlin-dahlem.mpg.de
Present addresses:
+
Department of Immunology and Microbiology, UCSF, San Francisco, CA 94143, USA
and
[sect]
Department of Biochemistry and Biophysics, UCSF, San Francisco, CA 94143-0448, US
