ABSTRACT
The process of homologous recombination has been documented in bacterial and
eucaryotic organisms. The
Escherichia coli
RecA and
Saccharomyces cerevisiae
Rad51 proteins are the archetypal members of two related families of proteins
that play a central role in this process. Using the PCR process primed by
degenerate oligonucleotides designed to encode regions of the proteins showing
the greatest degree of identity, we examined DNA from three organisms of a
third phylogenetically divergent group, Archaea, for sequences encoding
proteins similar to RecA and Rad51. The archaeans examined were a
hyperthermophilic acidophile,
Sulfolobus sofataricus
(
Sso
); a halophile,
Haloferax volcanii
(
Hvo
); and a hyperthermophilic piezophilic methanogen,
Methanococcus jannaschii
(
Mja
). The PCR generated DNA was used to clone a larger genomic DNA fragment
containing an open reading frame (orf), that we refer to as the
radA
gene, for each of the three archaeans. As shown by amino acid sequence
alignments, percent amino acid identities and phylogenetic analysis, the
putative proteins encoded by all three are related to each other and to both
the RecA and Rad51 families of proteins. The putative RadA proteins are more
similar to the Rad51 family (
~
40% identity at the amino acid level) than to the RecA family (
~
20%). Conserved sequence motifs, putative tertiary structures and phylogenetic
analysis implied by the alignment are discussed. The 5
'
ends of mRNA transcripts to the
Sso radA
were mapped. The levels of
radA
mRNA do not increase after treatment with UV irradiation as do
recA
and
RAD51
transcripts in
E.coli
and
S.cerevisiae
. Hence it is likely that
radA
in this organism is a constitutively expressed gene and we discuss possible
implications of the lack of UV-inducibility.
Homologous recombination is a process whereby two duplexes of DNA interact to
either exchange or rearrange segments of their DNA. Cells use this process for
DNA repair (
1
,
2
), generation of genetic diversity, gene regulation (
3
) and cell cycle-dependent events (
4
). Homologous recombination has been observed in bacteria, eucaryotes and now in
hyperthermophilic archaeans (D. Grogan, personal communication).
Much effort has been put into understanding the process of homologous
recombination at the molecular level. The central step in homologous
recombination is bringing together the two interacting DNA molecules, searching
for the homology and then exchanging the DNA strands between them. In the
bacterium
Escherichia coli,
it is believed a single gene product, called RecA, is responsible for this
activity. Known
in vitro
activities of RecA critical for its biological function include making RecA-ssDNA helical filaments, joint molecule formation between ssDNA and
homologous dsDNA, the ability to promote branch migration and the ability to
hydrolyze ATP (reviewed in
1
,
2
,
5
,
6
).
recA
mutant strains are extremely deficient in homologous recombination, sensitive
to DNA damaging agents (i.e., UV irradiation) and show an inability to process
regulatory and enzymatic proteins used in DNA repair and mutagenesis. Currently
65
recA
genes from different bacteria have been cloned and sequenced (
6
,
7
) (Roca and Cox, personal communication). Phylogenetic analysis of the bacterial
recA
genes has been done (
7
-
9
).
Putative homologs of RecA from eucaryotes have been identified based on amino
acid sequence similarity (
10
). Some of these proteins in
Saccharomyces cerevisiae
are Rad51, Dmc1, Rad57 (
11
) and Rad55 (
12
). The amino acid sequence similarity for RecA is greatest for Rad51 and Dmc1.
Correlations between RecA and Dmc1 have also been predicted based on amino acid
alignments and the X-ray crystal structure of a form of
E.coli
RecA combined with ADP (
13
). Biochemical comparisons between RecA and Rad51 have been discussed elsewhere
(
1
,
2
,
5
,
14
). Physiologically, mutants of
RAD51
,
RAD55
and
RAD57
also appear to be like bacterial
recA
mutants in that they are deficient in types of recombinational DNA repair (
11
,
12
,
15
,
16
). Mutants of
DMC1
are deficient in meiotic recombination, in forming synaptonemal complexes and
in completing meiotic prophase (
4
). Recently Rad51 and Dmc1 proteins have been shown to colocalize in the nucleus
prior to meiotic chromosome synapsis (
17
,
18
).
RAD51
gene expression in
S.cerevisiae
, like
recA
, is inducible at the level of transcription in response to DNA damage (
16
).
Rad51-like proteins have been found in the genomes of other eucaryotes:
Schizosaccharomyces
pombe
(fission yeast) (
19
-
21
),
Homo sapiens
(human) (
20
),
Mus musculus
(mouse) (
20
,
22
),
Gallus gallus
(chicken) (
23
),
Xenopus laevis
(South African clawed frog) [(
24
) GenBank D38488],
Drosophila melanogaster
(fruit fly) (
25
),
Neurospora crassa
(
26
) and
Lycopersicon esculentum
(tomato) (GenBank U22441). Dmc1-like proteins have been found in
H.sapiens
(human) (GenBank D64108),
M.musculus
(mouse) (GenBank D64107),
Candida albicans
(a fungus) (GenBank U39808) and
Lilium longiflorum
(lily) (
27
,
28
).
Genes for synaptase proteins like RecA and Rad51 have not yet been isolated from
the third organismal domain, Archaea. It is of interest, therefore, to see if
organisms from this domain have such genes and, if so, whether the protein they
encode more resembles the RecA or the Rad51 family. Since phylogenetic trees
based upon aminoacyl-tRNA synthetase (
29
), H
+
-ATPase (
30
) and elongation factors genes (
31
) show that the Archaea and Eucarya share a more recent common ancestor than
they do with bacteria, it would be predicted that archaeal RecA-like proteins should more closely resemble the eucaryal Rad51 proteins
than the RecA bacterial proteins. We found putative
recA
and
RAD51
homologues in three archaeans and we tested one of these genes for UV
inducibility.
One
E.coli
strain was used in this work, namely DH5[alpha]. The genotype is
supE44
[Delta]
lacU169
([Phi]
80 lacZ
[Delta]
M15
)
hsdR17 recA1 endA1 gyrA96 thi-1 relA1
.
Sulfolobus
solfataricus
P2, obtained from D. Grogan, was cultured in medium similar to the type used to
grow
Sulfolobus acidocaldarious
(
32
) with the exception that 0.2% sucrose was used instead of 0.2% xylose. Tryptone
(0.2%) was used as the nitrogen source. Standing overnight cultures grown at 75oC were used to innoculate larger cultures. Cultures for DNA and RNA
isolation were grown at 75oC with mild aeration provided by shaking in a rotating water bath.
Sulfolobus solfataricus
cells were grown as described above to an OD
600
of 1.0. Cells of
Methanococcus janaschii
were the generous gift of D. Clark. Cells of each species were pelleted and
resuspended in 1/10 vol 10 mM Tris-HCl pH 8, 1 mM EDTA, 100 mM NaCl. 1/10 vol of a 10% solution of sodium
lauryl sarcosine was added. Then an equal volume of phenol-chloroform-isoamyl alcohol (25:24:1) was added. The mixture was vortexed and
then centrifuged to separate the phases. The aqueous phase was removed and
extracted two more times. The DNA was precipitated by adding 1/10 vol 3 M
sodium acetate pH 7 and 2 1/2 vol of ethanol. After centrifugation, the DNA
pellet was washed once with 70% ethanol, centrifuged and the pellet resuspended
in water. This was then treated with RNase, re-extracted with the phenol-chloroform solution and precipitated with ethanol and 3 M sodium
acetate. The final pellet was washed with 70% ethanol and resuspended in water.
DNA from
Haloferax volcanni
strain WFD11 was the generous gift of M. Dyall-Smith.
PCR primers (5' -> 3') used were: GGWCCWGARTCWTCWGGWAARAC and
ACWGARTCWACWACWATWAC (
Sso
), ACWGAATTTTTTGGWGAATTTGGWTCWGGRAA and CTRAA- WGTWCCYTCWGTRTCWATRTA (
Mja
), and ATCACSGARSTSTTCGGSGARTT and SCGGAASGTSCCYTCSGTGTCGATRTA (
Hvo
) where W = A or T, S = G or C, R = G or A, and Y = C or T.
An MJ Research thermocycler PTC-150 was used for PCR. The reaction mixture contained 50 [mu]l total. DNA (500 [mu]g) was mixed with 250 pmol of each redundant primer mix. An
aliquot of 10 [mu]l of the Buffer `C' from the Invitrogen PCR Optimizer Kit (Cat# K1220-01) was used (5* buffer solution containing 300 mM Tris-HCl, 75 mM (NH
4
)
2
SO
4
, 12.5 mM MgCl
2
at pH 8.5). Two units of
Taq
polymerase purchased from Perkin Elmer was added. The samples were initially
heated to 94oC for 5 min. Then the 5 [mu]l of 2.5 mM solution of the dNTPs was added to final concentration of
0.25 mM. Finally, a layer of mineral oil was added. Then the samples were
heated at 94oC for 1 min before the temperature was reduced to 37oC for 2 min. After this the temperature was increased at a rate of 1o/10 s to a final temperature of 72oC. The samples were then incubated at 72oC for 2 min. This cycle was repeated an additional two
times. The reaction was then cycled for 30 times through 94oC for 2 min; 43oC for 2 min; 72oC for 2 min, where the temperatures were reached as quickly as
possible. Lastly the reactions were incubated at 72oC for 10 min before storage at 4oC.
PCR fragments were first cloned so that they could be analyzed by DNA sequencing
and then were used for a probe in Southern blot experiments. The PCR reaction
mixture was electrophoresed in a 2% agarose gel and a band of the appropriate
size was excised. The DNA was extracted from the gel using Qiaex Gel Extraction
Kit (Cat# 20020). This DNA was cloned using SureClone Kit (Pharmacia). This
mixture was used to transform DH5[alpha] cells to Amp
R
. After genomic fragments were identified by Southern Blot analysis,
Hin
dIII-
Sac
II (
Sso
),
Eco
RV (
Mja
) and
Kpn
I-
Sph
I (
Hvo
) fragments were cloned into pUC118 using standard methods.
Cells were grown as described above to an OD
600
of 0.25-0.35. Cells were centrifuged and resuspended in 1* minimal salts buffer [(in g/l distilled water): K
2
SO
4
, 3.0; NaH
2
PO
4
, 0.5; MgSO
4
.
7H
2
O, 0.3; CaCl
2
.
2H
2
O, 0.1 and the pH adjusted to 3.5 with H
2
SO
4
] (
32
). If the cultures were to be irradiated, this was done at 0.5 J/m
2
/s for 20 s. The cultures were then centrifuged and resuspended in growth medium
and incubated at 75oC with mild aeration for the appropriate time before extraction. Total RNA
was extracted from 10 ml of culture using the Qiagen RNEasy Total RNA Kit (Cat#
T4103).
All the protein sequences were aligned using PILEUP from the GCG Suite of
programs. A comparison region was defined as the minimal amino acid sequence
which all the proteins had in common and was called mCor for minimal common
overlap region. The percent identities in mCor were then calculated using the
number of amino acids in the mCor region as the denominator for each protein.
Adjustments were made at the boundaries of gaps in the sequences as described
in the Results.
Seven different proteins will be compared in this paper: two RecA proteins from
E.coli
and
Bacillus subtilis
, Rad51 and Dmc1 from
S.cerevisiae
and the three new proteins from the archaeans. In the course of describing this
work, all will be referred to as RecA-like proteins. The three archaean proteins and the yeast Rad51 and Dmc1
will also be referred to as Rad51-like proteins.
The
recA
-like genes in the Archaeans were called
radA
. This was not intended to indicate any homology to the
radA
gene of
E.coli
(
33
).
Genomes from three different types of archaeans were screened for
RecA
- or
RAD51
-like genes. Two major groups of
Archaea
,
Crenarchaeota
and
Euryarchaeota
, have been defined by 16S rRNA phylogenies. The
Crenarchaeota
consist mainly of hyperthermophiles. One of these,
S.solfataricus
, is a sulfur-oxidizing acidophilic hyperthermophilic aerobe (
34
). The
Euryarchaeota
consist mainly of two groups, the methanogens and the halophiles.
Methanococcus jannaschii
and
H.volcanii
are respectively members of these two groups of
Euryarchaeota
(
35
,
36
).
Methanococcus jannaschii
was isolated from a submarine hydrothermal vent (
37
). It is a piezophilic, hyperthermophilic
methanogen which grows 5-fold faster at 750 atm of pressure than at 8 atm at 86oC (
38
). It is also a strict anaerobe (
37
) and may have some halophilic tendencies given its habitat.
Haloferax volcanii
grows aerobically in media containing 20% NaCl at temperatures ranging from 28
to 56oC (
39
). A survey of GenBank reveals that genes isolated from
Sulfolobus
and
M.jannaschii
are AT-rich (>60% AT) while the genes from
H.volcanii
are GC-rich (>60% GC).
Initially we synthesized degenerate oligonucleotide PCR primers to encode two
highly conserved regions of the Rad51, Dmc1 and RecA proteins. These correspond
to codons 66-73 (GPESSGKT) and codons 140-146 (VIVVDSV) of
E.coli
RecA protein. These two regions are thought to interact with ATP (
13
,
40
). The GPESSGKT sequence is the Walker `A' motif and is thought to form the
phosphate binding hole. The VIVVDSV sequence is the Walker `B' motif and is
thought to interact with an Mg
2+
ion via a water molecule. The selection of bases for the third position of the
codons in the first set of primers was adjusted for an AT-rich organism to detect a
recA
- or
RAD51
-like gene in
S.solfataricus
(
Sso
). After the sequence of the putative gene from
Sso
, which we called
radA
, was determined, highly conserved regions among the Rad51, Dmc1 and
Sso
RadA proteins which were not conserved in the
Eco
RecA protein could be seen. These sequences were chosen as the basis for PCR
primers to screen the other two archaean genomes. In each case the pairs of
primers were adjusted for the GC content of the organism-
M.jannaschii
(AT-rich) or
H.volcanii
(GC-rich)-by appropriate selection of bases in the third position of each
codon. Both upstream (+) primers of these pairs are similar to those used for
screening the
Sso
genome; they encompassed part of the Walker `A' motif as well as upstream
sequences. The amino acid sequence used to design both downstream (-) primers was YIDTEGTFR. This amino acid sequence was present in
RAD51
,
DMC1
and
Sso radA
. Based on the alignment of Story
et al
., the corresponding region in RecA protein is FIDAEHALD and is thought to be
important for catalysis (
13
,
40
).
The DNA sequences for the three
radA
genes and ~300 bp upstream and downstream of the gene have been deposited in GenBank
with accession numbers U45310 (
Sso
), U45311(
Mja
) and U45312 (
Hvo
). An open reading frame analysis of the sequences in all six potential open
reading frames was carried out (data not shown). With
Sso
,
Hvo
and
Mja
, only one orf is seen which encodes putative proteins similar to RecA and
Rad51. We call that orf the
radA
gene. In the sequences from
Sso
and
Mja
, no other orf that could encode a protein >10 kDa overlaps the
radA
gene in either direction and register. In the
Hvo
sequence, however, a sizeable orf in the direction opposite to that of
radA
was seen. Whether or not this overlapping gene is expressed and whether or not
overlapping orfs are common in the genome of
H.volcanii
remain to be determined.
We assumed that the first ATG in the open reading frame with the homology to
RAD51
to be the start codon of the
radA
gene. In the
Sso
and
Hvo
radA
sequences the next ATG codon is >200 nt distant (data not shown). Using this
codon would eliminate regions similar to
DMC1
and
RAD51
. In the
Mja
sequence, however, there was a cluster of three ATG codons ~100 nt distant from the first (Fig.
1
). Using any one of those would have made the length of the three archaean genes
more similar. Because the sequence around the most downstream ATG has an
adequate RBS (see below), we chose to translate the orf beginning with this ATG
for our analysis.
It is thought that promoters of archaea and eucarya are similar to each other in
that they bind transcriptional and regulatory factors prior to binding RNA
polymerase for transcriptional initiation (
41
). In the bacteria promoters bind directly to RNA polymerase which contains a
component needed for promotor recognition (
42
). A consensus sequence for archaeal promoters has been reached through the
characterization of promoters from several species. This sequence, 5'-T/C T T A T/A A-3', is centered 26 bp upstream of the transcriptional
start site and is called BoxA (
42
,
43
). Interestingly, it is the same AT-rich sequence in all archaean species regardless of their overall GC
content. It has been shown that a purine-rich region just upstream of Box A may also be critical for function of
some promoters (
42
). Archaeal transcription usually initiates with a purine (
42
). Archaeal translational initiation is thought to require ribosomal binding
sites like that of bacteria (
44
,
45
).
The tentative start codon for the
Sso radA
gene is shown in Figure
1
. This was the first ATG in the orf whose putative protein is similar to Rad51
protein. Several nonsense codons block translation from upstream ATG codons.
The sequence of nucleotides at the 3' end of the
Sso
16S rRNA was used to screen for a complementary upstream sequence. Such a
sequence was identified as a potential ribosome binding site (Fig.
1
). This six nucleotide sequence, 5'-GGTGAT-3', is centered 9 nt upstream of the first ATG in the
orf.
Two putative promoters have been located at -52 and -73 nt upstream of the start codon by their similarity to the BoxA
consensus sequence (Fig.
1
). Each also is located downstream of a purine-rich region. We calculated that potential transcriptional start sites
would be found at -24 and -48 respectively by counting 26 bp downstream from the middle of
the Box A sequence and then identifying the nearest purine.
We tested whether or not initiation of transcription occurred
in vivo
from either of these two presumptive promoters by looking for 5' ends of mRNAs that were complementary to the
radA
gene of
S.solfataricus
. The results in Figure
2
(summarized in Fig.
1
) show 5' ends at positions -24, -26 and -45. The bands at -24 and -45 show much greater intensities than
the one at -26. The two extension products corresponding to -24 and -45 show a purine at the terminal position. The 5' end of two major extension products that end at -24 and -45 are both 28 nt downstream of the
two putative BoxA sequences at -52 and -73 respectively. We therefore tentatively conclude that the two
primer extension products at -23 and -45 identify nucleotides where transcriptional initiation occurs.
Either of these two transcription initiation points would provide transcripts
that contain the RBS and have the first ATG codon on the mRNA as the putative
start codon. The other primer extension product of lower intensity terminating
at position -26 may represent an alternative start site for the BoxA sequence centered
at -52 nt or a breakdown product of transcripts initiating at -45 nt.
Using the same criteria and reasoning described for the
Sso radA
gene, we identified a potential ATG start codon for the
Hvo radA
gene (Fig.
1
). Using the 3' end of the cognate 16S rRNA as a visual probe, we identified a
complementary sequence, 5'-AGGcGG-3', centered 14 nt upstream of the ATG as a putative
RBS. There are at least two sequences that have similarity to Box A. They are
located at -27 and -56. The -27 Box A sequence is downstream of a purine-rich region and the -56 sequence is not. The -27 Box A sequence, although having a
better match to the Box A consensus sequence than the -56 sequence and having an upstream purine-rich region, may not be the
in vivo
radA
promoter because transcripts from this promoter would most likely initiate
transcription in between the putative RBS
and start codon. We hypothesize, therefore, that the -56 Box A sequence or another sequence further upstream may constitute the
true promoter for
Hvo
radA.
The assignment of the start codon for the
Mja radA
is complicated by the fact that there are four ATG codons in the 100 nt
upstream region of the orf whose putative protein is similar to Rad51 protein.
Scanning 5-15 nt upstream of each ATG reveals that only one, the fourth (at 1 nt in
Fig.
1
), has a potential RBS sequence 5'-TGGTGA-3' centered 10 nt upstream of the potential start codon.
A potential promoter sequence has been identified at -123 nt (Fig.
1
). This sequence TTTATA is an exact match to the Box A sequence and has a purine-rich sequence immediately upstream. A potential transcriptional start site
would be at -101 nt and allow the fourth ATG codon to be the site for translational
initiation.
Transcriptional initiation of the
recA
and
RAD51
genes is inducible by UV irradiation (
16
). This regulation is mediated in a negative fashion by the LexA protein for
recA
and in a positive fashion for the
RAD51
gene by an upstream promoter enhancer sequence (
16
,
46
). Since archaeal transcriptional regulation is thought to resemble eucaryal
more closely than bacterial transcriptional regulation, the regions upstream of
the three putative promoters were scanned for the presence of a conserved
sequence that could be regulatory in function. No obvious sequence was found.
Nonetheless, since UV-inducible gene expression has been demonstrated by induction of the lytic
cycle of an integrated SSV1 DNA phage of
Sulfolobus
shibatae
(
47
), the inducibility of
radA
gene transcription by UV irradiation for
S.solfataricus
was tested. An increase in the basal level of primer extension products
analyzed above after treatment with UV light would be indicative of an increase
in the amount of mRNA. Using a level of UV irradiation that gave 50% survival
for
S.acidcauldarius
(data not shown), the results shown in Figure
2
reveal that treatment with UV irradiation produced no or very little increase
in the amount of primer extension product seen. We tentatively conclude that
the gene expression of the
radA
gene of
S.solfataricus
, unlike
RAD51
or
recA
, does not increase at the level of transcription after UV irradiation.
Alignment of the three predicted RadA proteins from
S.solfataricus
,
M.jannaschii
and
H.volcanii
with the Rad51 and Dmc1 proteins from
S.cerevisiae
and the RecA proteins from
E.coli
and
B.subtilus
is shown in Figure
3
. Inspection of the alignment reveals a minimal region in which all the
sequences overlap. We call this the mCor. Two subgroups are evident based on
whether or not the proteins have an N- or a C-terminal extension from the mCor region. The first subgroup consists
of the RecA proteins from the two bacteria. These two proteins have C-terminal ends that extend beyond the mCor. The second subgroup consists of
the three archaean proteins and the two yeast proteins. These contain N-terminal extensions beyond the mCor. Analysis of the spaces introduced by
the computer program to align the seven sequences also shows this grouping
among the seven proteins (see below).
From the topology of the tree for 16S rRNA, we expect that the RadA proteins
from
Hvo
and
Mja
, both from the kingdom
Euryarchaeota
should be more similar to each other than to
Sso
RadA protein from the kingdom,
Crenarchaeota
(
35
). This was tested for the alignment in Figure
3
and for two variants by using the PROTPARS program of Phylip Suite 3.57. For
the mCor plus N- and C-terminal sequences, or for the mCor sequences from which ambiguously
aligned amino acids were eliminated as suggested by Eisen (
7
), we found an unrooted tree which corresponds to the 16S rRNA tree (figure not
shown). When we used the mCor regions containing the ambiguously aligned amino
acids, we found an unrooted tree in which
Hvo
and
Mja
are not grouped in a clade like the 16S rRNA tree (figure not shown).
The occurrence of
radA
genes whose putative proteins resemble RecA, Rad51 and Dmc1 proteins in
representatives of both known archaean kingdoms implies that these organisms
may carry out recombination by mechanisms similar to those of bacteria and
eucaryotes. The putative proteins encoded by these
radA
genes are more similar in sequence to predicted Rad51 and Dmc1 proteins than
they are to the two bacterial RecA proteins. This is consistent with the
information accumulated from a wide variety of protein and RNA sequences. It
establishes that in DNA metabolism, RNA metabolism and protein synthesis,
archaeans are closer relatives of the eucaryotes than they are of bacteria.
In our analysis the similarity of proteins depends upon the way they are
aligned. We chose the alignment produced by the PILEUP program but we actually
used two other methods to align the seven proteins in Figure
3
. One was the CLUSTAL3 program. The second was agreement with the alignment
recently proposed by Roca and Cox (personal communication) supplemented by an
alignment of part of the
Sso
RadA sequence supplied by A. Roca (personal communication). Both of the
alignments so produced, however, failed to satisfy our two criteria for
assessing the validity of an alignment: preservation of the continuity of
putative secondary structure elements and consistency with a phylogenetic relationship mimicking that of 16S RNAs (data not shown).
[alpha] and [beta] secondary structure elements have been proposed for
Eco
RecA protein based on its X-ray crystal structure of an ADP-protein complex (
40
,
48
). These are indicated on the alignment of proteins in Figure
3
(
40
,
48
). Only two of those elements ([alpha] helices C and D) are interrupted by gaps inserted by PILEUP to optimize
the alignment. These two helices are on the outside of the molecule in the
region of the protein involved with ATP hydrolysis (
40
,
48
). In the alignment produced by the two other methods, up to eight of the
secondary structure elements were interrupted by gaps. In these alignments we
moved amino acids at the boundaries of the gaps to restore continuity to the
secondary structural elements one by one. After each move a phylogenetic tree
based on the resulting alignment was drawn by PROTPARS. In neither case were we
able to restore continuity to all but two elements without altering the tree
from one that matched the 16S RNA tree to one that did not match. Consequently, we chose the PILEUP alignment to present.
There are nine places in the alignment where there are gaps of two or more amino
acids in some of the sequences (Fig.
4
). These are potential signature sequences. For example, there are two places
(numbered 2 and 4 in Fig.
4
) where the two bacterial sequences lack amino acids contained by the other five
sequences. There are another two places (numbered 7 and 9 in Fig.
4
) where the bacterial sequences have amino acids not possessed by the other
five. Thus, there are potentially four sequence regions which serve to
distinguish a bacterial sequence from that of an archaean or eucaryote.
Scrutiny of the 63 other bacterial RecA protein sequences listed by Roca and
Cox (personal communication) and Eisen (
7
) show that all are similar to
Eco
-RecA and
Bsu
-RecA in these four locations. There are two regions (numbered 1 and 8 in
Fig.
4
) which may serve to distinguish archaean from eucaryote sequences. Two places
(numbered 5 and 6 in Fig.
4
) seem to mark sequences that may serve to distinguish halophilic from other
archaean sequences. We have sequenced the mid-part of the
radA
genes from two other halophile species,
Halobacterium halobium
and
Haloarcula hispanica
, and find sequences very similar to that of
H.volcanii
(unpublished data). Finally, region 3 (in Fig.
4
) may serve as a signature for methanogen sequences or for the
Methanococcus
genus since
Mja
RadA has nine amino acids in this region not contained by the other two
archaean sequences. The mid-parts of three other methanogen
radA
genes have also been sequenced to test this prediction and the data show that
region 3 is a signature sequence for two other
Methanococcus
species but not for a
Methanosarcina
species (unpublished data).
We have explored the possibility that the alignment shown in detail in Figure
3
and diagrammatically in Figure
4
might reveal structural features of the Rad51-like proteins. One way to evaluate this possibility is to mark the three
dimensional structure proposed for the crystalline form of
Eco
RecA complexed with ADP (
40
,
48
) with differences in primary structure shown in the alignment. We have done
this in two ways in Figure
5
. First, there are three regions of the primary structure of
Eco
RecA that are shown as missing from the Rad51-like proteins in Figures
3
and
4
. These `missing' amino acids are shown in red in Figure
5
. Secondly, there are four regions where the Rad51-like proteins are shown in Figures
3
and
4
to have more amino acids in their primary structure than
Eco
RecA. These are marked by highlighting the amino acid on each side of the gaps
in
Eco
RecA primary structure shown in Figures
3
and
4
. We will comment briefly on features of the resulting molecular model in Figure
5
and then evaluate briefly the validity of this technique.
Lastly, this research has been based on the premise that homologous proteins in
different organisms will have similar functions. Others have shown that a
deletion mutation in the
radA
gene in
H.volcanii
leads to a UV-sensitive phenotype (Woods and Dyall-Smith, personal communication). Hence it is likely that the
radA
genes in Archaea will have a role in DNA repair and possibly also in homologous
recombination. We have demonstrated that the
radA
orf in
S.solfataricus
is transcribed and that this transcription is initiated from promoter sequences
that are similar to other promoters identified in Archaeans (Fig.
2
). The outstanding result, however, is that the transcription of
radA
is not UV-inducible (Fig.
2
). This is different from the transcriptional regulation pattern of both
eucaryotic and bacterial
recA
-like genes. There are several possible reasons why we might have observed
this. First, the dose of irradiation may not have been great enough to elicit a
response or the response may be slower than we allowed for in our experiment.
Secondly, since UV-inducible gene expression has been detected in
S.shibatae
(
47
), it is possible that
S.solfataricus
is mutant for this type of regulation. Lastly, it is possible that
S.solfataricus
is constitutive for
radA
expression. If the last possibility is true, and if the common ancestor of all
exant life was both hyperthermophilic and constitutive for primordial
recA
expression (like
S.sofataricus
), then this raises the possibility that regulated UV-inducible
recA
(and
RAD51
) gene expression is a mesophilic adaptation. If so, then the types of UV-inducible DNA repair regulation that now appear in bacteria and eucaryotes
are examples of convergent evolution.
We would like to thank the following people: Mike Dyall-Smith and Wayne Woods for supplying the
H.volcanii
genomic DNA and sharing their data before publication, Doug Clark for growing
and supplying the
M.jannaschii
cells, Dennis Grogan for supplying us with an innoculum of
S.solfataricus
and sharing his results before publication, Aberto I. Roca and Mike Cox for a
copy of their manuscript before publication and Ray Jacobson for instruction
and help on the Midasplus program used to create the diagrams in Figure
5
. This research was supported by grant AI05371 from the National Institutes of
Health.
REFERENCES
Return