Download Ancient Ciphers: Minireview Translation in

Cell, Vol. 89, 1007–1010, June 27, 1997, Copyright 1997 by Cell Press
Ancient Ciphers:
Translation in Archaea
Patrick P. Dennis
Department of Biochemistry
and Molecular Biology
and Canadian Institute for Advanced Research
Program in Evolutionary Biology
University of British Columbia
Vancouver, British Columbia V6T 1Z3
Canada
The process of translation occurs on subcellular particles called ribosomes and involves the decoding of nucleotide sequence information carried on mRNA into
amino acid sequence of protein. The central and universal features of translation, decoding and amino acid
polymerization, utilize RNA-based chemistry that is aided
by the presence or activity of numerous structural and
catalytic proteins. Although translation is sequential
and repetitive, the complexity of the machinery involved and the sophistication at each step in the process
are extraordinary. Our understanding of this process is
greatly enhanced through comparative studies using
organisms from different phylogenetic lineages. Of particular interest are the Archaea because they are the
closest prokaryotic relatives to the nuclear component
of eucaryal cells. As such, they provide a glimpse of
bacterial features mixed with rudimentary features that
have become more highly elaborated within the eucaryal
lineage.
Two instructive examples of this are protein synthesis
initiation and precursor rRNA processing. Archaea frequently utilize Shine-Dalgarno sequences to identify
AUG, GUG, or UUG translation initiation codons. Remarkably, the regiment of initiation factors employed
are clearly homologous to those of eukaryotes that participate in the identification of the AUG initiation codon
using a scanning mechanism. How are these bacterial
and eucaryal features integrated and what special advantage does such a mixed system provide to the Archaea? Similarly, archaeal pre-rRNA processing generally follows the bacterial paradigm. But Archaea also
possess a protein homologous to eucaryal fibrillarin
and, in at least one instance, require a trans-acting RNA
for the endonucleolytic cleavage of pre-rRNA. Is the
trans-acting RNA related to the small nucleolar RNAs
(snoRNAs) of Eucarya, and how wide is its distribution?
Is it also present in bacteria, and does it possibly function in the 59 end maturation of small subunit rRNA?
rRNA Transcription Units and rRNA Processing
Within the Archaea, there is considerable variation in
the number and structure of rRNA transcription units.
The number of units range between one and four per
genome, with multiple units always well separated from
each other within the genome. In the euryarchaeota
branch of Archaea (see figure 1 in Olsen and Woese,
1997 [this issue of Cell]), the units are generally bacteriallike, containing a spacer tRNAAla and a distal tRNACys
gene as well as a cotranscribed 5S gene. In the crenarchaeota, the units are more eucaryal-like, generally containing only 16S and 23S rRNA genes. In this instance,
the 5S gene is relocated and independently transcribed.
Minireview
All operons share the common feature of long inverted
repeats surrounding both the 16S and 23S rRNA genes.
In bacteria, these repeats form helical structures within
the primary transcript that are recognized and cleaved
by the duplex-specific endonuclease, RNaseIII. Although RNaseIII is not an essential activity in E. coli, the
alternate routes for pre-rRNA processing are much less
efficient. These initial cleavages excise pre-16S and pre23S rRNA from this primary transcript. Archaea also
possess a helix-specific endonuclease, but it is unrelated to bacterial RNaseIII (Figure 1). The substrate requirements of this enzyme have been shown to consist
of a helical structure with two 3-base loops on opposite
strands separated by four base pairs (Thompson and
Daniels, 1988). Endonucleolytic cleavage occurs between the second and third bases of the two loops and
thereby releases pre-16S or pre-23S rRNA (Figure 1).
Interestingly, this bulge-helix-bulge (BHB) endonuclease is also employed to excise archaeal introns from the
transcripts of intron-containing tRNA and rRNA genes
(Belfort and Weiner, 1997 [this issue of Cell]). In these
instances, the exon–intron junctions are capable of folding into the requisite bulge-helix-bulge structure.
As more rRNA operon sequences become available, it
is becoming evident that many of the processing helices
contain aberrant bulge-helix-bulge motifs. Without this
motif, how is pre-rRNA processing initiated? The best
studied case is in Sulfolobus acidocaldarius, where the
16S helix contains only a 2-base bulge on the 59 side.
Analysis of in vivo rRNA processing intermediates indicated that little or no cleavage occurs within this region
of the helical stem. Instead, cleavage occurs at two sites
further upstream. An in vitro processing system to study
the two cleavages in the 59 ETS and the cleavage at the
59 ETS-16S junction has been developed. The cleavage activity is sensitive to digestion by micrococcal
nuclease, implying that the responsible endonuclease
contains one or more essential RNA components (Potter
et al., 1995). We previously reported the cDNA cloning
of a U3-like RNA from the processing activity (Potter
et al., 1995). However, we have been unsuccessful in
locating this sequence within the genome of S. acidocaldarius and now have evidence that it arose through
contamination of the Taq DNA polymerase used in the
original amplification protocol. Nevertheless, the endonucleolytic processing activity is sensitive to nuclease
digestion, and we believe that it may contain RNAs related to the snoRNAs that have been implicated in the
processing of pre-rRNA in Eucarya.
The gene encoding a protein related to eucaryal fibrillarin was first identified in Methanococcus vannielii and
has now been found in other Archaea. In Eucarya, fibrillarin is localized to the fibrillar region of the nucleolus
and is essential for pre-rRNA processing and ribosome
subunit biogenesis. Antibodies against fibrillarin coprecipitate a large number of snoRNAs that have been implicated in pre-rRNA processing. Among these, U3, U14,
snR30, and MRP RNA are essential in yeast. The presence of an archaeal fibrillarin-like protein and an RNAcontaining endonuclease activity implicated in pre-rRNA
Cell
1008
Figure 2. Ribosomal Protein L1 Binding Site
Figure 1. Structure of Bulge-Helix-Bulge Motifs
The structure of the BHB motifs are shown for the 16S processing
stem of H. cutirubrum (Hcu) in pre-rRNA, the exon-intron junctions
of Haloferax volcanii (Hvo) in pre-tRNA trp, the 16S processing stem
of S. acidocaldarius (Sac) in pre-rRNA, and the 16S processing stem
of M. jannaschii (Mja) in pre-rRNA. The first two motifs are cleaved
by the BHB endonuclease at the indicated positions (arrows),
whereas the third is not cleaved; cleavage on the fourth has not
been examined.
processing indicates that at least some Archaea may
contain the rudiments of a eucaryotic-like pre-rRNA processing system. The archaeal RNPs may be involved in
or required for one or more of the maturations at the 59
and 39 ends of small and large subunit rRNAs. Perhaps
the elusive 16S rRNA maturase of bacteria also contains
a small RNA component.
r-Protein Genes and Translational Regulation
The complete sequence of the Methanococcus jannaschii genome provides the first clear and comprehensive picture of archaeal organization of ribosomal
protein genes (Bult et al., 1996). A total of 60 genes
exhibiting sequence similarity to known bacterial, archaeal, and eukaryotic ribosomal protein encoding genes
has been identified. In general, the archaeal r-proteins
are more similar in sequence to their eucaryal than to
their bacterial homologs. However, as in bacteria, 53 of
the 60 genes are in 15 separate multigene clusters, the
largest of which contains 16 r-protein genes. Many of
the clusters also contain genes encoding other essential
components such as transcription and translation factors and subunits of RNA polymerase. The nucleotide
spacing between genes within a cluster is generally between 10 and 70 nucleotides or more. Evidence from
other archaeal species indicates that at least some of
the genes within these clusters are cotranscribed.
The most interesting and best studied example is the
L1, L10, L12 gene cluster. Expression of the proteins
from this operon appears to be subject to a bacterial-like
autogenous translational control, whereas the proteins
themselves are more eucaryal in character. In E. coli,
protein L1 is the translational regulator of the L11, L1
bicistronic operon. In the absence of sufficient 23S
rRNA, the L1 protein can bind to the translational operator—a structural mimic of the 23S rRNA binding site that
is present in the 59 leader of the L11, L1 mRNA—and
prevent further translation. In halophilic and methanogenic Archaea, the L11 gene is in a separate transcription unit and the L1 gene is cotranscribed with the L10
The structural mimic of the L1 binding site in 23S rRNA that is
present in the coding region of the M. vannielii L1 gene is shown
on the left. The amino acid sequence encoded by the RNA is also
shown from residue 10 to residue 22. A consensus L1 binding motif
for archaeal (methanogens and halophiles) and bacterial mRNAs is
illustrated on the right with the conserved core nucleotides
indicated.
and L12 genes. In halophiles, the L1 translational operator is located in the unusually long (64 nt) 59 untranslated
leader of L1, L10, L12 mRNA (Figure 2). The situation in
M. vannielii is even more intriguing. Here the site of the
L1 translational operator has been transposed from the
leader to a position between nucleotides 30 and 60 inside the L1 coding region (Hanner et al., 1994). Sitedirected mutagenesis has been used to demonstrate
the importance of this motif in translational regulation.
A nearly identical motif is present at the corresponding
position in the M. jannaschii L1 gene. As one might
expect, the translational operator in Methanococcus
genes lies in a region that is not highly conserved in
other members of the L1 protein family. Further work is
required to determine if there are other r-protein translational operators in Archaea.
Translation Initiation Signals
One of the major challenges in the analysis of raw DNA
sequence data is the identification of functional ORFs.
Generally, identification is based on similarity to known
sequences in the database, although additional features
such as ORF length and codon utilization are also used.
The limit of a coding region is then set according to the
position of in-frame translation initiation and termination
codons. In Archaea, the situation is further complicated
by the frequent use of alternate GUG and UUG initiation
codons. Since N-terminal amino acid sequence is seldom available, identification of the correct in-frame
AUG, GUG, or UUG initiation codon is problematic. How
serious is this problem and what are its consequences?
Based on a partial survey of Methanococcus ORF
assignments, it appears that AUG, GUG, and UUG are
used about 70%, 25%, and 5% of the time, respectively,
as initiation codons. In about half of the cases, the putative initiation codons are preceded at the appropriate
3–10 nt distance by at least a 4-base complementary
match to the pyrimidine-rich sequence (59-AUCACCUCCU-39) at the 39 end of 16S rRNA. In the putative
intergenic regions of closely spaced genes (i.e., in
r-protein gene clusters), AUG, GUG, and UUG sequences are very prevalent but are seldom preceded at
Minireview
1009
the appropriate distance by good Shine-Dalgarno sequences. In the limited instances where the authentic
initiation codon is known with a high degree of certainty
(i.e., from N-terminal protein sequence), a good ShineDalgarno sequence is also present. This raises the distinct possibility (i) that methanogenic Archaea may rely
strongly on a Shine-Dalgarno sequence to recognize
and utilize proper initiation codons, and (ii) that many
of the assignments of putative initiation codons that lack
Shine-Dalgarno sequences may be incorrect. A good
example of this is the proposed M. jannaschii L1 r-protein ORF. The assigned AUG initiation codon is not preceded by any Shine-Dalgarno-like sequence. An inframe AUG triplet that is preceded by a very good
ribosome binding site occurs at codon 14. The protein
generated from this new initiation codon shows high
similarity to the N-terminus of the L1 protein from
M. vannielii.
A similar situation exists with Sulfolobus. The 39 end
of 16S rRNA contains the sequence 59-AUCACCUC-39,
which is complementary to 59-GAGGUGAU-39 sequences
in mRNA. In at least two instances, the L1 and L10
ribosomal protein genes of S. acidocaldarius, it has been
suggested that the Shine-Dalgarno sequence overlaps
with the GUG initiation codon rather than preceding it
by the standard distance of 3 to 10 nt. It is difficult to
imagine how this region of the mRNA could interact
simultaneously with both the 39 end of 16S rRNA and
the anticodon region of the initiator tRNA. More careful
examination revealed the presence of an in-frame GUG
and an in-frame UUG triplet at codon position 4 in the
proposed L1 and L10 reading frames, respectively. It
seems likely that these are the authentic initiation codons and that the normal spacing between the ShineDalgarno sequence and the translation initiation codon
is maintained.
In halophilic Archaea, 59 mRNA leader sequences are
often very short or even absent. In these instances,
translation initiation may occur by a scanning mechanism utilizing the first available AUG, without the aid of
a Shine-Dalgarno interaction. However, within polycistronic mRNAs or RNAs with long 59 leaders, initiation
codons are almost always preceded by sequences with
some complementarity to the 39 end of 16S rRNA.
Clearly, a full understanding of the mechanism used for
selection of translation initiation codons will require a
more detailed and systematic biochemical analysis of
this initiation process.
tRNAs and Amino Acyl tRNA Synthetases
The genome of M. jannaschii contains a full complement
of tRNA but only 16 of the expected 20 amino acyl tRNA
synthetases (Bult et al., 1996). Missing are the enzymes
for asparagine, cysteine, glutamine, and lysine. Generation of Gln tRNAGln and Asn tRNAAsn in halophilic Archaea
is known to involve postcharging conversion of glutamate to glutamine and aspartate to asparagine, respectively (Curnow et al., 1996). It is possible that a similar
pathway might be used to produce Cys tRNACys. The
tRNACys could be charged with serine using seryl tRNA
synthetase and then modified to cysteine. It is more
difficult to imagine how a postcharging conversion could
generate a Lys tRNALys. More likely, a lysyl tRNA synthetase exists but has escaped detection because of low
sequence similarity to other lysyl tRNA synthetases.
Archaea also possess a small number of proteins containing the unusual amino acid selenocysteine. As in
nonarchaeal organisms, there is no selenocysteinyl
tRNA synthetase; selenocysteinyl tRNASec is generated
by first charging the tRNASec with serine using seryl tRNA
synthetase and then converting the attached serine to
selenocysteine with selenocysteine synthetase (Wilting
et al., 1997).
Initiation, Elongation, and Release
Examination of the M. jannaschii genome sequence revealed the presence of eleven genes encoding proteins
implicated in translation initiation, three implicated in
elongation, and one implicated in termination-release
(Bult et al., 1996). Surprisingly, of the 11 putative initiation factor proteins, 10 appear homologous to eucaryal
initiation factors (see Table 1). The first is an eIF1A-like
protein. In Eucarya, eIF1A binds to free 40S ribosomal
subunits and prevents premature reassociation with 60S
subunits. Three other proteins are the a, b, and g subunits of an eIF2-like complex. In Eucarya, eIF2 forms a
ternary complex with met-tRNAinit and GTP. The complex
binds to the 40S subunit to form a 43S preinitiation
complex. Normally, the 43S complex then binds mRNA
via its 7 methyl G cap; binding is mediated by the cap
recognition factor eIF4E and the organizing factor
eIF4G. Archaeal mRNAs do not contain the 59 end 7
methyl G modification and neither eIF4E nor 4G homologs appear to be present in M. jannaschii.
There are three separate genes encoding members
of the eIF4A family. These proteins have ATP-dependent
RNA helicase activity; they function to remove secondary structure in eucaryl mRNA and permit scanning for
the initiation codon by the 43S preinitiation complex. At
the present time, it is uncertain if any or all of the M.
jannaschii helicase proteins are required for protein synthesis initiation, if they are used exclusively for this purpose, or if they are employed to unwind or rearrange
RNA in other cellular processes. Another initiation factor
is homologous to eIF5. In Eucarya, this factor associates
with the eIF2 ternary complex on the surface of the 40S
subunit; when the initiation codon is properly aligned,
it mediates hydrolysis of the GTP bound to eIF2 and
facilitates the ejection of the eIF2·GDP and the initiation
factors from the 40S subunit. This opens the way for
entry of the 60S subunit to form the 80S initiation
complex.
A common feature of eucaryl eIF2 complexes is that
they lack a guanine nucleotide exchange domain and
therefore require an extrinsic guanine nucleotide exchange factor. Not surprisingly, two M. jannaschii genes
encode homologs to the a and d subunits of the guanine
nucleotide exchange factor eIF2B (see Table 1). The d
subunit participates directly in the nucleotide exchange
reaction whereas the a subunit is involved in regulation
of initiation. The M. jannaschii eIF2a subunit contains
a conserved serine at position 51, which in Eucarya can
be phosphorylated by one of a number of different eIF2specific protein kinases. The eIF2Ba subunit acts as a
sensor for phosphoserine at position 51 of EIF2a and,
when present, prevents the guanine nucleotide exchange
reaction. This blocks the formation of eIF2·GTP·mettRNAinit ternary complex and prevents further translation
initiation. Since both the serine at position 51 of eIF2a
Cell
1010
Table 1. Translation Initiation Factors Encoded by the Genome of M. jannaschii
Gene Number
Homolog
Proposed Function
0445
0117
0097
1261
eIF1A
eIF2a subunit
eIF2b subunit
eIF2g subunit
antiassociation, binds to 40S subunit
ternary complex with met-tRNAinit and GTP
1574
1505
0669
eIF4A
eIF4A
eIF4A
ATP-dependent RNA helicase activity
ATP-dependent RNA helicase activity
ATP-dependent RNA helicase activity
1228
eIF5
stimulates GTP hydrolysis on eIF2 when initiation codon is properly aligned
0454
0122
eIF2Ba subunit
eIF2Bd subunit
guanine nucleotide exchange factor for eIF2
0262
IF2 (bacterial)
ternary complex with fmet-tRNAinit and GTP
(bacterial type containing a guanine nucleotide exchange domain)
and the sensing eIF2Ba subunit are present in M. jannaschii, it is possible that this important exchange cycle
is used to regulate protein synthesis initiation, perhaps
in response to environmental stresses such as nutrient
or energy limitation. No eIF2-specific protein kinases
have yet been identified. The three other eucaryal subunits of eIF2B are apparently not present in M. jannaschii.
The last initiation factor gene encodes a protein member of the bacterial IF2 family. In protein synthesis, bacterial IF2 carries out the same function as eIF2. However,
it possesses a guanine nucleotide exchange domain
and therefore does not require a guanine nucleotide
exchange factor for regeneration. Is this IF2-like protein
involved in initiation of protein synthesis in M. jannaschii
or has it been recruited to carry out a separate and
unrelated function? Saccharomyces cerevisiae also encodes a protein of this bacterial IF2 family. The gene is
essential for yeast viability but its function is unknown.
There is no evidence that it is required for protein synthesis. From the above list of factors, it would appear that
Archaea utilize a hybrid system for translation initiation;
the factors are mostly eucaryal-like whereas the basic
mechanism for initiation codon selection in many and
perhaps most instances is bacterial, utilizing an interaction of mRNA with the 39 end of 16S rRNA.
There are three genes encoding elongation factor proteins. The first encodes a eucaryl-like EF1a, the second
a eucaryl-like EF2 and the third, a bacterial-like EFTu.
EF1a forms a ternary complex with aa tRNA and GTP
and is responsible for the codon-dependent deposition
of aa tRNA into the A site of an elongating ribosome.
This protein is expected to require a guanine nucleotide
exchange factor as it lacks its own nucleotide recycling
domain. No candidate for a recycling protein has been
identified in the M. jannaschii genome. The second factor, EF2, carries out GTP-dependent ribosome translocation. This protein possesses a recycling domain and
no exchange protein is required. The third bacterial-like
factor is a highly specialized Tu factor that deposits
selenocysteinyl tRNA into the A site of the ribosome in
the presence of a UGA codon and a selenocysteine
insertion (SECIS) element somewhere on the mRNA
(Wilting et al., 1997). In Methanococcus and eukaryotes,
this element is never in the coding region but rather is
usually located in the 39 untranslated region of the
mRNA. In bacteria, the SECIS element is located immediately 39 to the UGA selenocysteine codon.
A single eucaryal-like release factor capable of recognizing all three translation termination codons has been
identified. In Eucarya, a second protein with GTPase
activity is also required for efficient recognition of termination codons and hydrolysis of the nascent polypeptide from the p site peptidyl tRNA. The homolog of this
auxillary protein has not been identified in M. jannaschii.
In summary, the Archaea appear to utilize an intriguing
mixture of bacterial and rudimentary eucaryl features
in their translation apparatus. Recognizing how these
features interface within Archaea will lead to a more
thorough understanding of the disparate characteristics
of bacterial and eucaryal translation and help to create
a universal or general model for translation, applicable
to all organisms. High on the list of topics demanding
intellectual and detailed biochemical investigations are
pre-rRNA processing, ribosomal subunit assembly, and
translation initiation. The availability of archaeal genome
sequences provides an excellent starting position for
these studies but many challenges remain. Where there
is uncertainty, there are bound to be surprises.
Selected Reading
Belfort, M., and Weiner, A. (1997). Cell, this issue.
Bult, C.J., White, O., Olsen, G.J., Zhou, L., Fleischmann, R.D., Sutton,
G.G., Blake, J.A., FitzGerald, L.M., Clayton, R.A., Gocayne, J.D., et
al. (1996). Science 273, 1058–1072.
Curnow, A.W., Ibba, M., and Söll, D. (1996). Nature 382, 589–590.
Hanner, M., Mayer, C., Köhrer, C., Golderer, G., Gröbner, P., and
Piendl, W. (1994). J. Bacteriol. 176, 409–418.
Olsen, G.J., and Woese, C.R. (1997). Cell, this issue.
Potter, S., Durovic, P., and Dennis, P.P. (1995). Science 268, 1056–
1060.
Thompson, L.B., and Daniels, C.J. (1988). J. Biol. Chem. 263, 17951–
17959.
Wilting, R., Schorling, S., Persson, B.C., and Böck, A. (1997). J. Mol.
Biol. 266, 637–641.