Dominant negative mutator mutations in the
mutL
gene of
Escherichia coli
Dominant negative mutator mutations in the mutL gene of Escherichia coli
Alexander
Aronshtam
and
M. G.
Marinus*
Department of Pharmacology and Molecular Toxicology, University of Massachusetts
Medical School, 55 Lake Avenue,
Worcester
, MA 01655,
USA
Received April 9, 1996
;
Revised and Accepted May 14, 1996
ABSTRACT
The
mutL
gene product is part of the
dam
-directed mismatch repair system of
Escherichia coli
but has no known enzymatic function. It forms a complex on heteroduplex DNA
with the mismatch recognition MutS protein and with MutH, which has latent
endonuclease activity. An N-terminal hexahistidine-tagged MutL was constructed which was active
in vivo
. As a first step to determine the functional domains of MutL, we have isolated
72 hydroxylamine-induced plasmid-borne mutations which impart a dominant-negative phenotype to the wild-type strain for increased spontaneous mutagenesis. None
of the mutations complement a
mutL
deletion mutant, indicating that the mutant proteins by themselves are
inactive. All the dominant mutations but one could be complemented by the wild-type
mutL
at about the same gene dosage. DNA sequencing indicated that the mutations
affected 22 amino acid residues located between positions 16 and 549 of the 615
amino acid protein. In the N-terminal half of the protein, 12 out of 15 amino acid replacements occur at positions conserved in various eukaryotic MutL homologs. All but
one of the sequence changes affecting the C-terminal end of the protein are nonsense mutations.
INTRODUCTION
The MutL protein is part of the
dam
-directed MutHLS mismatch repair system which rectifies replication errors in newly synthesized daughter strands of DNA (
1
-
3
). The MutHLS system also acts on heteroduplex DNA recombination intermediates,
which may result in drastic changes in expected recombination frequencies (
1
). The MutHLS proteins initiate the repair process. MutS binds to the mismatch
in DNA, followed by recruitment of MutL. Subsequent binding of MutH to this
complex activates its latent endonuclease activity which cleaves the newly synthesized strand, followed by excision repair (
2
,
3
).
The
mutL
gene of
Escherichia coli
was identified by the mutator phenotype of mutant alleles (
4
), which produce predominantly AT -> GC and GC -> AT transitions (
5
-
8
). The inability of
mutL
mutants to correct base pair mismatches (
9
) or small loops (
10
) in heteroduplex DNA established a role for the MutL protein in
dam
-directed mismatch repair (reviewed in
2
,
3
). The transient undermethylation at Dam recognition sites (-GATC-) of newly synthesized DNA chains directs repair to this strand (
11
).
Other known phenotypes of
mutL
mutants are consistent with its role in mismatch repair. These include: (i)
extragenic suppression of
recA dam
bacteria (
12
,
13
); (ii) hypermutability to 5-bromouracil (
14
), 9-aminoacridine (
15
) and alkylating agents (
16
); (iii) resistance to the cytotoxic effects of certain alkylating agents of
dam mutL
bacteria compared with
dam
strains (
17
,
18
); (iv) increased homologous (
19
,
20
) and homeologous recombination (
21
); (v) increased excision from the chromosome of certain transposable elements (
22
); (vi) augmenting VSP (very short patch) repair (
23
); (vii) stability of nucleotide di- and tri-repeat sequences (
24
,
25
); (viii) stability of larger chromosomal directly repeated sequences (
26
).
The MutL protein (68 kDa) is a homodimer in solution and has no known catalytic
function in mismatch repair. It increases the size of the heteroduplex DNA
region protected by MutS from ~20 to ~100 bp (
27
). The MutL-MutS protein complex bound to DNA is required for activation of the
endonuclease activity of MutH at the nearest hemimethylated -GATC- site (
28
). The role of MutL may be to stabilize and/or correctly position the MutS and
MutH proteins. An additional role for MutL in the excision process is also
possible.
Homologs of the
E.coli
MutL protein have been found in
Salmonella typhimurium
,
Streptococcus pneumoniae
,
Saccharomyces cerevisiae
, mouse and human (
29
). Multiple homologs are found in eukaryotic organisms, which in certain cases
form heterodimers (
3
,
30
). The region of greatest amino acid sequence similarity appears to be at the N-terminal end of the protein. In humans, mutation in the MutL homologs has
been associated with non-polyposis colon cancer (
31
,
32
).
Little is known about the domain organization of MutL. In the absence of any
structural information, we have used a new approach to identify functional
domains which exploits the properties of dominant-negative mutations. Such mutations are used because: (i) the mutant
proteins are expected to be relatively stable; (ii) the mutations should map
throughout the gene, rather than be localized to one region (due to the
multifunctional nature of MutL). This approach has been used successfully by Wu
and Marinus (
20
; unpublished data) to identify mutations affecting all known enzymatic
activities of the multifunctional MutS protein. A similar strategy should allow
identification of the functional domains of MutL which interact with MutS and
MutH.
MATERIALS AND METHODS
Bacterial strains and plasmids
Escherichia coli
K-12 strains and the plasmids used are listed in Table
1
. GM4250 (
mutL
::Km [Delta]
Stu
) was isolated as a kanamycin-resistant recombinant after P1
vir
transduction of CC106 with phage grown on strain NU747 (
33
). [lambda] DE3 derivatives of strains CC106 and GM4250 were constructed as
described by the suppliers (Novagen). A His-tagged
mutL
+
derivative of pET15b, pMQ378, was constructed by PCR amplification of the gene
on pNU127 (
33
) using primers MM155 (5'-cgggatccgatatcactcatctttcag-3') and MM156 (5'-ccaaactaaggacgacatatgccaattcaggtc-3') and Taq DNA polymerase in
the buffer supplied by the manufacturer (Promega) with 25 mM MgCl
2
for 35 cycles. Cycle settings were 92, 52 and 72oC for 2 min each. The amplified fragment was purified using a QiaQuick
(Qiagen) column, digested with
Nde
I and
Bam
HI and inserted into the corresponding sites of pET15b, to yield pMQ378. DNA sequence analysis indicated an A -> G base change at position 225 compared with the GenBank Z11831 sequence (
34
) and presumably represents an error amplified during the PCR. The sequence
change does not alter the coding for His75.
Media
L broth contained 10 g/l Bacto tryptone, 5 g/l Bacto yeast extract, 10 g/l NaCl
and 2 ml/l 1 M NaOH, solidified when required with 16 g/l agar (Difco). Brain
heart infusion broth (Difco) was used at 20 g/l. The minimal medium was that of
Davis and Mingioli (
35
). MacConkey agar containing lactose was purchased from Difco. Ampicillin,
chloramphenicol, kanamycin and rifampicin (all from Sigma) were added at 100,
20, 40 and 100 [mu]g/ml respectively.
Mutagenesis of plasmid DNA and isolation of mutants
Plasmid DNA (1-2 [mu]g) was exposed to 400 mM hydroxylamine (Kodak) in 50 mM sodium
phosphate buffer, pH 6.0, at 70oC for 60 min, precipitated with ethanol and resuspended in 10 mM Tris, 1 mM
EDTA buffer, pH 8.0. Strain CC106 was transformed to chloramphenicol resistance
with portions of treated plasmid DNA on either MacConkey lactose medium or L
agar. After overnight incubation, the colonies were replica-plated from L agar to L agar + rifampicin and incubated for 1-2 days at 37oC. The colonies on MacConkey agar were scored for papillation
after 2-3 days incubation at 37oC. Plasmid DNA was extracted from rifampicin-resistant or papillating colonies and used to transform strain
CC106. From plasmids retaining the dominant phenotype the promoter-proximal
Ava
I-
Apa
LI and promoter-distal
Apa
LI-
Bam
HI fragments were purified and used to replace the corresponding fragment of the wild-type plasmid. The replacement plasmid showing the dominant phenotype was used in all further work.
Genetic complementation
Each CC106 derivative containing a plasmid conferring a dominant-negative
mutL
phenotype was transformed with a pBR322 derivative containing the wild-type
mutL
,
mutH
or
mutS
gene. The extent of papillation on MacConkey agar was measured after 2-3 days incubation at 37oC. Alternatively, the frequency to rifampicin resistance was measured by plating aliquots of a
saturated broth culture on L agar containing rifampicin and incubating overnight at 37oC.
ara thi
[Delta]
(gpt-lac)5
/F-
lacI
q
lacZ proAB
+
40
GM4250
As CC106 but
mutL
[Delta] (Kan
Stu
)
GM5861
As CC106 but ([lambda]DE3)
GM5862
As GM4250 but ([lambda]DE3)
Plasmids
p613
Amp
R
; synthetic glutamic acid inserting UAG suppressor gene in pGF1B1
39
pACYC184
Tet
R
Cam
R
41
pET15b
Amp
R
; N-terminal fusion to hexahistidine and thrombin cleavage peptide
vector (Novagen)
pMQ350
Amp
R
;
Bam
HI (
mutL
) fragment from pNU127 into
Bam
HI site of pZ150
42
pMQ378
Amp
R
; as pET15b but
mutL
+
on an
Nde
I-
Bam
HI fragment insert obtained
from pNU127 by PCR
pMQ393
Cam
R
; pACYC184 with a
Bgl
II-
Hin
dIII fragment from pMQ378 (includes
the T7 promoter) replacing the
Bam
HI-
Hin
dIII region
pMQ396
Cam
R
; pACYC184 with a replacement fragment from pMQ378 (does not
include the T7 promoter; Fig. 1c)
pNU127
Tet
R
Cam
R
;
mutL
+
on a 6.5 kb
Pst
I fragment in pBR325
33
pSTL113
Amp
R
, Tet
S
; pBR322 with mismatched duplication in the
tet
gene
26
DNA sequencing
Double-stranded plasmid DNA was sequenced by the dideoxy chain termination
technique (
36
) with Sequenase 2.0 (US Biochemical Co.) using a series of synthetic
oligonucleotides of 15-17 residues corresponding to various sites in the
mutL
gene as primers. Either the
Nde
I-
Apa
LI or the
Apa
LI-
Bam
HI region was completely sequenced for each dominant mutant.
Purification of His-tagged MutL, antisera preparation and Western blotting
His-tagged MutL was isolated as recommended by the manufacturer (Novagen). Two milliliters of a saturated culture of the plasmid-containing strain was inoculated into 1 l of L broth and grown to an optical density (A
600
) of 0.8, IPTG (Sigma) was added to 0.5 mM and the culture incubated for an additional 90 min. The cells were harvested by
centrifugation, resuspended in 20 ml binding buffer (5 mM imidazole, 0.5 M
NaCl, 20 mM Tris-HCl, pH 7.9) and sonicated for 5 min using 1 s pulses at 1 s intervals.
After centrifugation in a Ti70 rotor for 30 min at 39 000 r.p.m. at 4oC in a Beckman L5-50 ultracentrifuge, the supernatant was loaded on a 2.5 ml Ni
2+
resin column. The column was washed with 60 mM imidazole and MutL eluted with 500 mM imidazole and frozen at -70oC.
Rabbit polyclonal antiserum to His-tagged MutL was prepared by BAbCO (Richmond, CA). Polyclonal antiserum was pre-adsorbed using a cell extract made from the
mutL
deletion strain GM4250. The strain was grown at 37oC in L broth to an optical density (A
600
) of 1.2, centrifuged and resuspended in a 1/10th volume of 100 mM sodium
phosphate, pH 7.4. A portion of 500 [mu]l was sonicated on ice for 60 s using 1 s pulses at 1 s intervals, followed
by centrifugation at 10 000
g
for 5 min. Fifty microliters of antiserum were mixed with the supernatant and
incubated at room temperature with gentle shaking for 2 h. After diluting 1000-fold, the solution was used for Western blotting. Cells were grown to an
optical density (A
600
) of 1.0 and 200 [mu]l portions were centrifuged and resuspended in 10 [mu]l water, diluted with an equal volume of 2* loading buffer [135 mM Tris-HCl, pH 6.8, 20% glycerol, 4% SDS (Gibco BRL), 0.0025%
bromophenol blue, 10% 2-mercaptoethanol], boiled for 10 min and loaded on SDS-8% polyacrylamide gels in a Hoeffer SE 250 slab gel electrophoresis
unit (
22
). After electrophoresis, the proteins were transferred to Immobilon-P membranes (Millipore) using a semi-dry electroblotter (Owl Scientific). MutL was detected using chemiluminescence according to the manufacturer's (Tropix Inc.) instructions. Bradford protein assay kits and
standards were obtained from BioRad Laboratories.
Stability of tandem repeats
CC106 cells containing plasmid-borne dominant mutations were transformed to ampicillin resistance with
plasmid pSTL113 (
26
) and the transformants were purified by streaking for single colonies. Logarithmic phase cultures (~10
8
cells/ml) were prepared and portions plated on L ampicillin agar and L ampicillin plus tetracycline
agar to determine the number of viable cells and cells with
tet
gene rearrangments respectively (
26
).
RESULTS AND DISCUSSION
Isolation of mutants
In order to facilitate processing of the anticipated mutant MutL proteins we
constructed plasmid pMQ378, a pET15b derivative containing the
mutL
gene which produces MutL with a hexahistidine tag at the N-terminal end (Fig.
1
). The results in Table
2
show that this plasmid complements the
mutL
defect of strain GM4250 (
mutL
[Delta]), indicating that the extra amino acids at the N-terminal end do not interfere with MutL function
in vivo
. A similar result has been reported by Feng and Winkler (
37
) while this work was in progress.
Expression levels of mutant proteins
We tested the assumption that the dominant-negative MutL proteins are relatively stable
in vivo
by measuring the level of these proteins in extracts of late log phase cells by immunoblotting. Figure
2
a shows that the full-length mutant proteins show variable steady-state levels compared to wild-type, but not less than 2- to 3-fold. Figure
2
b indicates that truncated nonsense fragments can be detected by immunoblotting
at the approximate expected molecular weight positions deduced from the DNA
sequence. The difference in MutL level in wild-type cells (lanes 5-7) compared with those with pMQ350 (lanes 1-4) is also shown. Diluting the plasmid-derived extract (Fig.
2
a) shows that the MutL level is increased ~25-fold in cells bearing plasmid pMQ396 compared to a single copy wild-type strain. A faint cross-reacting band is present at the same amount in all
samples in Figure
2
b, suggesting that it is not a MutL degradation product.
Instability of tandemly repeated DNA sequences
Tandemly repeated DNA sequences are unstable in
E.coli
(
26
). Lovett and Feschenko (
26
) have examined the frequency of deletion between two 101 bp repeats in the
tetracycline (
tet
) resistance gene of pBR322. One of the repeats contains four silent mutations
in the
tet
gene, leading to 4% heterology between the repeated sequences. Tetracycline-resistant colonies which have resolved the duplication by deletion arise
at high frequency by a mechanism that does not involve
recA
-dependent homologous recombination. Instead, the frequency to tetracycline
resistance is increased in mutants defective in
dam
-directed mismatch repair and in a
dam
mutant strain, suggesting that these genetic realignments occur during
replication by the resolution of heteroduplex DNA.
We have tested the effect of representative dominant
mutL
mutations on the stability of tandem repeats using the system described by Lovett and Feschenko (
26
). The data in Table
3
confirm that a >30-fold increase occurs in an otherwise isogenic
mutL
[Delta] strain compared with wild-type. The various dominant mutations alter the frequency to
tetracycline resistance over a spectrum of values ranging from wild-type to the
mutL
control value. One mutant allele (
mutL712)
gave a value four times higher than the
mutL
control, but this is probably a statistical outlier due to the low number of
samples (five) used. The spectrum of values in Table
3
indicates that the dominant mutations are heterogeneous with respect to duplication resolution. The correlation between duplication resolution and spontaneous mutation is
variable, depending on the mutation (Table
2
). The correlation is high, for example, for both
mutL711
and
712
(at about the
mutL
[Delta] level) and
715
and
725
(at about the wild-type level). For
mutL720
, however, the correlation is poor, with a high level of spontaneous mutation
but a moderately low duplication resolution frequency. Similarly,
mutL726
shows a low correlation with a low spontaneous mutation frequency but a
moderately high duplication resolution frequency. At present the implication of
these results is unclear, given the qualitative measurements and the absence of
biochemical characterization of the mutant proteins.
Identification of mutational sites
To facilitate localizing the mutations, we took advantage of the centrally
located
Apa
LI site (bp 903) in the
mutL
gene (1848 bp) to replace the promoter-proximal or promoter-distal fragments of the wild-type plasmid with those from the mutant plasmid. The majority
(75%) of the mutations were located in the promoter-proximal fragment.
.
Duplication and spontaneous mutation frequencies in the presence of mutant and
wild-type
mutL
alleles
mutL
allele
Duplication frequency
Spontaneous mutation frequency
Wild-type
0.8
0
[Delta]
(Stu)
30
3+
702
11
2+
703
10
2+
704
17
2+
705
nd
3+
706
5
1+
707
nd
1+
708
nd
1+
709
nd
2+
710
19
2+
711
30
3+
712
120
3+
713
10
3+
714
9
3+
715
1
1+
716
7
1+
717
nd
2+
718
12
3+
719
2
1+
720
7
3+
721
3
1+
722
nd
1+
723
19
1+
724
23
1+
725
0.9
1+
726
14
1+
Plasmid pSTL113 (a pBR322 derivative) was introduced into wild-type cells containing each dominant
mutL
mutation or the wild-type allele on a pACYC184 backbone. Logarithmic phase cultures (~10
8
cells/ml) of the strains were plated on media containing ampicillin plus
tetracycline and ampicillin. The ratio of the viable counts on these plates (*10
-4
) is given for the wild-type and mutant alleles. The last column gives a qualitative estimate of
the strength of dominance, where 0 denotes a wild-type phenotype (based on the level of rifampicin resistance) and 3+ a
mutL
-
phenotype.
nd, not done.
DNA sequencing was carried out on either the promoter-proximal or promoter-distal regions of
mutL
. The locations of the 72 mutations together with the predicted amino acid
substitutions are shown in Table
4
. All but one of the mutations are GC -> AT base changes consistent with the mutagenic action of hydroxylamine. The
mutations occur throughout the gene, affecting amino acids A16-Q549 of the 615 amino acid MutL protein. A mutational hotspot is present
at base 317, since the same mutation was recovered 23 times. Most of the
mutations affect amino acids in the N-terminal end of the protein, which we define as residues 1-308. Of the 15 amino acid replacements, 12 occur at identical
positions in various eukaryotic MutL homologs (bold type amino acids in Table
4
) and the remaining three (S106, S112 and G238) are conserved in bacterial
species. This suggests that the results obtained with our experimental approach
for the
E.coli
MutL protein will also be applicable to bacterial and eukaryotic homologs.
The nucleotide sequence is taken from GenBank Z11831 (
30
). Base 1 is the A of the ATG codon. All mutations except
mutL723
were GC -> AT transitions. A frameshift mutation due to deletion of a G was
identified in
mutL723
.
MutL724
contains two mutations at nt 1201 and 1591. The fourth column is the number of times the mutation was present in the collection. Amino acids in bold are conserved in eukaryotic homologs.
All but one (
mutL720
) of the mutations affecting the C-terminal end of the protein (residues 309-615) are nonsense mutations.
MutL722, 725
and
726
are
amber
mutations, while
mutL721
and
724
are
ochre
mutations. The
amber
mutations were suppressed in the presence of plasmid p613, which encodes a
synthetic glutamic acid-inserting tRNA (
39
), forming mostly full-length protein as measured by immunoblotting (data not shown). The
presence of p613 also converted the phenotype of the
amber
mutants from MutL
-
to MutL
+
in strain GM4250 (
mutL
[Delta]), indicating that substituting glutamic acid for Gln332, Gln519 and
Gln549 does not appreciably affect enzyme function
in vivo
. The truncated nonsense fragments will be useful in future biochemical analysis
to delineate the functional domains of the protein.
MutL723
is a frameshift mutation due to loss of a G residue at bp 1198 which leads to a
protein fragment with the correct sequence of amino acids to codon 399 and out
of frame amino acids until the
ochre
termination codon at position 417. Two mutations were identified in
mutL724
; one at bp 1201, which generates an
ochre
mutation, and a second at bp 1591, which must be inconsequential since a MutL
fragment with the anticipated molecular weight is produced in cells with this mutation (data not shown).
Complementation in a
mutL
strain
Complementation by wild-type
mut
plasmids
A dominant-negative phenotype for MutL in a wild-type background could occur, for example, through the formation of
inactive heteromultimers composed of wild-type and mutant MutL monomers. Alternatively, the defective mutant MutL
could sequester a vital component of the repair complex (e.g. MutS or MutH).
When the mutant protein is expressed from a multicopy plasmid and the wild-type from a single chromosomal copy gene, all of the wild type protein is
expected to be in the form of heteromultimers or deprived of an essential
component on the sequestration model. Increasing the gene dosage of the wild-type, however, should allow for the formation of sufficient wild-type MutHLS complex to alter the phenotype from mutant to wild-type. We have transformed wild-type cells containing each of 25 plasmid-borne dominant mutants (representing all mutated sites)
with a compatible (pBR322 origin) plasmid bearing the wild-type
mutL
gene. In all but one case, the phenotype was converted from mutant to wild-type (Table
5
), consistent with either the heteromultimer or sequestration hypotheses. The level of complementation was at about the wild-type level for all mutants with the exception of
mutL707
, which showed a marginal but consistent decrease relative to the control
mutL
-
level. The
mutL705
mutant, which was not complemented by wild-type, might represent a protein with a higher affinity for its target site
than wild-type.
.
Complementation of
mutL
dominant mutations by
mutL
,
mutS
and
mutH
mutL
allele
Complementation by
mutL
+
mutS
+
mutH
+
mutL
[Delta]
3+
-
-
702
2+
-
-
703
2+
-
-
704
2+
-
-
705
-
-
-
706
3+
1+
2+
707
1+
-
1+
708
3+
-
1+
709
3+
-
-
710
3+
-
-
711
3+
-
-
712
2+
-
-
713
2+
-
1+
714
2+
-
-
715
3+
1+
3+
716
3+
-
3+
717
3+
1+
3+
718
3+
-
2+
719
3+
-
2+
721
3+
2+
2+
722
3+
2+
2+
723
3+
2+
2+
724
3+
-
2+
720
3+
-
2+
725
3+
1+
1+
726
3+
1+
1+
Complementation was measured qualitatively by the level of rifampicin resistance in cell cultures containing compatible wild-type and mutant plasmids. Not all the mutations in the collection were
tested. A wild-type level is indicated by 3+, while - is the
mutL
-
level.
In addition to testing complementation with wild-type
mutL
, we have also used multicopy
mutS
+
and
mutH
+
plasmids. The levels of complementation were variable, ranging from 1+ (weak)
to 3+ (strong) depending on the mutant (Table
5
). In general, the
mutH
+
plasmid complemented the
mutL
mutations in the gene coding for the C-terminal but not the N-terminal end of the protein (Table
5
and Fig.
3
), suggesting a functional difference between the two parts of the protein.
Weak complementation was also observed in eight mutant strains with
mutS
+
, which in all instances occurred with mutations also complemented by
mutH
+
(Table
5
and Fig.
3
). At present we do not know the mechanistic basis for the complementation by
mutH
+
or
mutS
+
. For those mutations complemented by
mutH
+
but not
mutS
+
, the mutant MutL proteins may act to sequester the limited amount of MutH in an
inactive complex. Supplying either MutH or MutL in excess
in trans
should then restore the normal phenotype, as observed. We have no satisfying
explanation for those
mutL
mutations complemented by any of the
mut
genes. Perhaps the currently unexplained requirement for the hydrolysis of two
ATP molecules to form MutS-MutL and an active MutHLS ternary complex respectively is affected by
these MutL mutations, even though only the MutS protein has ATPase activity.
Whatever the inhibitory mechanism is, it suggests these classes of mutants will
be interesting to characterize at the biochemical level.
CONCLUSIONS
The dominant-negative mutations we have isolated map throughout the gene and are heterogeneous with respect to their phenotypic traits.
The mutant proteins are relatively stable. The histidine-tagged mutant and wild-type proteins can be rapidly purified (data not shown), allowing for
the development of new assays to measure MutL-MutS and MutL-MutH interactions on heteroduplex DNA and how these are impaired by the mutant proteins and to identify the functional domains of the MutL
protein.
ACKNOWLEDGEMENTS
We thank S.Lovett, J.H.Miller and M.Winkler for gifts of strains and plasmids,
F.Fandryer for technical assistance and Te-hui Wu and Romas Vaisvilla for suggestions and comments. This work was
supported by grant MCB-9302889 from the National Science Foundation.
REFERENCES
1 Meselson,M. (1988) In Low,K.B. (ed.), Recombination of the Genetic Material. Academic Press, New York, NY.
