(Received for publication, July 29, 1996)
From the Department of Internal Medicine, Division of
Infectious Diseases, University of Michigan Medical Center, Ann
Arbor, Michigan 48109-0642 and the § Department of
Epidemiology, University of Michigan School of Public Health, Ann
Arbor, Michigan 48109-2029
The product of the Escherichia coli lon gene is the ATP-dependent Lon protease. Lon contributes to the regulation of several important cellular functions, including radiation resistance, cell division, filamentation, capsular polysaccharide production, lysogeny of certain bacteriophages, and proteolytic degradation of certain regulatory and abnormal proteins. Lon homologues are also found in several widely divergent bacteria, as well as in the mitochondria of yeast and humans. E. coli Lon has long been known to bind to DNA, but this interaction has not been further characterized and has generally been assumed to be nonspecific. We now demonstrate that E. coli Lon can bind to a TG-rich DNA promoter element in a sequence-specific manner. This finding is based on the results of experiments employing SouthWestern blotting, protein purification, "shift-shift" electrophoretic mobility shift assays, electrophoretic mobility shift assays using in vitro transcribed and translated Lon, and DNase footprinting. Site-specific DNA binding is likely to be an additional important biochemical characteristic of the multifaceted Lon protease.
The product of the Escherichia coli lon gene (Lon) is an ATP-dependent protease which plays multiple regulatory roles in the organism (reviewed in Refs. 1, 2, 3). Some of these important roles include radiation resistance, cell division, filamentation, production of capsular polysaccharide, lysogeny of certain bacteriophages, and proteolytic degradation of several classes of regulatory and abnormal protein targets. Lon has also been identified as part of the heat shock response in E. coli (4, 5, 6, 7). Lon also has an associated DNA binding activity, and this property was exploited for use in its purification (8). However, the consequences of Lon binding to DNA, previously thought to be nonspecific, are not yet understood.
Purification and cloning of E. coli protease Lon has shown it to be a polypeptide with a predicted molecular mass of 87 kDa that migrates on polyacrylamide gels with an apparent molecular mass of 94 kDa (9, 10, 11). Protein degradation by Lon is dependent on ATP hydrolysis (12, 13). Lon also has a DNA-stimulated ATPase activity (14). The active protease is a tetramer composed of four identical subunits (13, 43). Lon appears to exert its regulatory function by degrading abnormal polypeptides (15, 16) and certain short-lived regulatory proteins such as the cell division inhibitor SulA and RcsA (17, 18).
The lon genes of other Gram-negative and Gram-positive bacteria have also been identified, cloned, and characterized (19, 20, 21, 22, 23, 24, 25, 26). The product of the lon gene in these organisms is very similar to the E. coli Lon and in each organism is also involved in important regulatory processes. In eukaryotes, the lon gene has been cloned from yeast (Saccharomyces cerevisiae) (27) and humans (28). Here they are encoded in the nucleus but localize to the mitochondria (29). Although yeast cells deficient in LON are respiration-deficient and have a nonfunctional mitochondrial genome (30), a role for mutant LON in human disease has not yet been characterized. The yeast and human LON genes are very similar to the E. coli lon gene. In yeast cells lacking the LON gene, complementation with the E. coli counterpart allows maintenance of the integrity of the mitochondrial genome (31). This shows remarkable functional conservation of Lon proteases from prokaryotes to eukaryotes.
In the process of cloning a eukaroytic protein that binds to the peri-ets (pets)1 site of the human immunodeficiency virus type 2 (HIV-2) enhancer (32, 33), we were surprised to note highly specific binding activity in extracts from E. coli. Purification of the binding activity from E. coli gave a polypeptide with an apparent molecular mass of 94 kDa that when subjected to N-terminal sequencing was identified as protease Lon. An antibody recognizing bacterial Lon further shifted the DNA-protein complex found in electrophoretic mobility shift assays (EMSA) using the pets site probe and whole cell bacterial extracts. In vitro transcribed and translated Lon specifically bound to the pets site in EMSA. We further demonstrated preferential binding of Lon to this specific DNA sequence using DNase footprinting. This is the first report that protease Lon can bind to DNA in a sequence-specific manner. This observation suggests that site-specific DNA binding could be another functionally important characteristic of this complex regulatory protein.
E. coli strain Y1089 (Promega) was grown to saturation in LB medium supplemented with 50 µg/ml ampicillin at 37 °C. The bacteria were harvested by centrifugation at 5000 × g for 5 min. The pellet was then resuspended in chromatography buffer A (20 mM Tris-HCl, pH 7.9, 1 mM EDTA, 50 mM KCl, 0.01% Nonidet P-40, 10% glycerol, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) supplemented with 0.5 mg/ml lysozyme and incubated on ice for 15 min. The suspension was sonicated on ice with short bursts until the sample was no longer viscous. The lysate was clarified by centrifugation at 27,000 × g for 60 min at 4 °C. Post-centrifugation, the supernatant was filtered successively though a 1.6-µm glass fiber filter and a 0.45-µm membrane filter prior to column loading.
Purification of the E. coli pets FactorAll procedures were
performed at 4 °C (see Fig. 2). The clarified extract was loaded
onto a Q-Sepharose fast flow column (Pharmacia Biotech Inc.) at a flow
rate of 3 ml/min. Following loading, the column was washed with 10 bed
volumes of buffer A or until A280 reached
baseline. Bound proteins were eluted with a 100 mM KCl stepwise gradient in buffer A. Fractions positive for specific binding
as assayed by EMSA were pooled and dialyzed into buffer A prior to
injection into a heparin-Sepharose column (Pharmacia). Bound proteins
were eluted in a 50 mM to 1 M KCl linear
gradient in buffer A. Positive fractions were pooled and dialyzed into buffer A before purification over a DNA affinity column prepared as
described (34). Proteins bound to the DNA affinity column were washed
off in two steps: a low salt (500 mM KCl) and a high salt
(1 M KCl) wash in buffer A. Eluted proteins were
precipitated with the addition of trichloroacetic acid to 10%
concentration and resolved by SDS-polyacrylamide gel electrophoresis
(PAGE).
Analysis of DNA Binding Proteins
Oligodeoxynucleotides were
synthesized with an Applied Biosystems 300B synthesizer.
Oligodeoxynucleotides were annealed by boiling in 500 mM
NaCl followed by slow cooling before end-labeling with 32P.
DNA was quantitated using A260 = 50 µg/ml
double stranded DNA. After removal of unincorporated
[32P-]ATP using gel filtration through a Sephadex G-50
column, 20,000 cpm of end-labeled oligodeoxynucleotides were used in
each individual reaction. The pets site oligodeoxynucleotide used for
EMSA and for DNA affinity chromatography had the following sequence:
5
-GATCCAGCTATACTTGGTCAGGGCGAATTCTAACTA. The mutant pets
oligodeoxynucleotide had the following sequence: 5
-GATCCAGCTATACTAGATCTGGGCGAATTCTAACTA.
EMSAs were performed as described previously (32). For EMSAs using
purified Lon, 1 µl of the purified Lon fraction (~50 ng as
estimated by Coomassie Blue staining after SDS-PAGE) was used per
individual reaction. Purified Lon (gift of Alvin Markovitz, University
of Chicago) came from a 0.2 M NaCl elution off of a DEAE
column (8). When a sample of this fraction is separated on SDS-PAGE and
stained with Coomassie Blue, more that 95% of the protein is 94 kDa.
The identity of this 94-kDa protein was further confirmed to be Lon by
Western analysis (not shown). For supershift assays, 1 µl of antibody
was added to the binding reaction 10 min prior to loading the gel.
Polyclonal antibody to Lon was prepared as described (8) and provided
by Alvin Markovitz. For Southwestern analysis, proteins were separated
by SDS-PAGE on an 8% gel, electroblotted onto a polyvinylidine
diflouride membrane (Millipore), denatured in buffer A supplemented
with 6 M urea, and renatured in buffer A overnight at
4 °C. An end-labeled, eight-copy pets site oligodeoxynucleotide with
the binding sites all in the same orientation (35) was used as the
probe.
N-terminal protein
sequence analysis was accomplished by automated Edman degradation using
standard procedures on a pulsed liquid phase sequenator (ABI model
473). The 500 mM KCl eluate from the DNA affinity column
was precipitated, separated on an 8% SDS-polyacrylamide gel, and
electroblotted onto a polyvinylidine diflouride membrane. The blot was
stained with Coomassie Brilliant Blue R-250 and the 94-kDa polypeptide,
which corresponded to the band seen in the SouthWestern blotting (see
Fig. 3), was excised and subjected to N-terminal sequence analysis by
the University of Michigan protein structure facility. The identity of
the sequenced protein was determined by sending a FASTA query to Genome
(Eerie-Nimes, France).
In Vitro Transcription and Translation
The plasmid pJMC21
(36, 37), containing the lon gene, was digested with
EcoRI and EcoRV. The fragment encoding
lon was purified with agarose gel electrophoresis and cloned
into pGem7f (Promega) cut with EcoRI and SmaI.
Lon was made in vitro with T7 DNA polymerase using a wheat
germ extract coupled transcription/translation kit (Promega). Inclusion
of [35S]methionine in the transcription/translation mix
labels a 94-kDa band when the in vitro synthesized Lon is
separated using SDS-PAGE.
A 32P-labeled 190-base pair fragment of the HIV-2 enhancer encompassing the pets site was generated using polymerase chain reaction with an end-labeled primer. The polymerase chain reaction product was purified using polyacrylamide gel electrophoresis and sequenced before use. Footprinting was performed as described previously (32).
The transcriptional enhancer of HIV-2
contains an element, termed pets, which mediates enhancer induction in
activated T cells and monocytic cells and binds a very recently
characterized 43-kDa eukaryotic protein (32, 33,
35).2 A similar site is found in the human
T cell leukemia virus type 1 (38). In the process of cloning the
eukaryotic pets-binding protein using a bacteriophage gt11 cDNA
library expressed in E. coli Y1089, we noted striking
binding to the pets element in extracts from the E. coli
themselves. EMSA demonstrated that this binding activity is very
specific (Fig. 1). Whereas the binding could readily be
eliminated by addition of unlabeled pets probe competitor (Fig. 1,
lanes 3 and 4 versus lane 1), even a 1000 molar
excess of a mutated version of the probe, in which the pets site had
been altered, did not significantly affect binding (Fig. 1, lanes
8, 9, and 10). Similarly, the unrelated
B
oglionucleotide failed to eliminate binding activity (Fig. 1,
lanes 5-7).
Purified Bacterial Lon Binds to the pets Site
The avidity and
specificity of the pets binding activity in E. coli Y1089
suggested that the identity of the bacterial protein involved would be
of interest. Therefore, we purified the pets binding activity from
crude bacterial extracts using Q-Sepharose fast flow chromatography,
followed by heparin-Sepharose chromatography and then one round of DNA
affinity chromatography as described under "Experimental
Procedures" (Fig. 2). This highly purified protein
extract was then used to define the size of the polypeptide that binds
to the pets site. Southwestern blotting using a pets site probe
revealed three polypeptides, of molecular masses 25, 30, and slightly
less than 97 kDa, which were detected in the low salt wash after the
DNA affinity column purification (Fig. 3A,
lane 1). The more stringent high salt wash gave only one
band, the 94-kDa polypeptide, when the purified fraction was assayed using SouthWestern blotting (Fig. 3A, lane 2) or
when separated by SDS-PAGE and stained with Coomassie Blue (not shown).
Partial N-terminal sequencing of the purified polypeptides showed that the 94-kDa protein was the Lon protease (Fig. 4).
Therefore, highly purified E. coli Lon was obtained and
assayed by SouthWestern blotting, using a pets site probe. This
experiment clearly demonstrated that highly purified bacterial Lon
could bind to the pets site (Fig. 3B).
Lon Is the Dominant Protein Binding to the pets Site in E. coli Extracts
The above studies suggested that crude E. coli extracts and highly purified bacterial Lon are able to bind
specifically to the pets site. Although a complex was formed between
the pets probe and the E. coli crude cell extract (Fig.
5A, lane 1), no such complex was
seen when the B probe or a probe from the Encaphelon promoter were
used (Fig. 5A, lanes 5 and 6), further
demonstrating that a bacterial protein, presumably Lon, binds to the
pets site but not to several other known eukaryotic enhancer sites. To
further examine whether Lon is indeed the dominant protein recognizing this site in bacterial extracts, shift-shift EMSA ("supershift") experiments were performed. The addition of rabbit preimmune serum (Fig. 5A, lane 2) or antibodies directed against
immunoglobulin (Fig. 5A, lane 3) failed to
supershift the pets/bacterial protein complex. In contrast, polyclonal
antibodies directed against bacterial Lon caused a marked supershift
(Fig. 5A, lane 4). The polyclonal antibody
directed against Lon does not cause any shift by itself when added to
the probe in the absence of additional protein extracts (not shown).
These data further demonstrate the specificity of the Lon-pets
interaction and strengthen the observation made from the purification
and sequencing studies that Lon is the dominant protein in E. coli extracts that recognizes the pets site. To further test the
specificity of Lon binding to the pets site, we performed DNase
footprinting assays and showed that purified Lon is able to protect the
HIV-2 pets site from DNase digestion (Fig. 6, lane
3).
Recombinant Lon Specifically Binds to the pets Site
The above
experiments demonstrated that highly purified bacterial Lon can bind to
the pets site. Although the purified bacterial Lon extracts contained
only one 94-kDa polypeptide when analyzed by SDS-PAGE (not shown), the
possibility still existed that another protein present in the extracts
and obscured by the abundance of Lon might be the actual pets site
binding protein. Therefore, recombinant Lon was prepared by coupled
transcription and translation in wheat germ extract and used to assay
for pets site binding. Although very little binding to the pets site
was seen in unprogrammed wheat-germ extract (Fig. 5B,
lane 1), the extract containing in vitro
transcribed and translated Lon bound avidly to the pets site (Fig.
5B, lane 2). This complex could be
completely eliminated by the addition of the pets oligodeoxynucleotide
(Fig. 5B, lane 3) but was only slightly
diminished by the addition of either the unrelated B
oligodeoxynucleotide or an oligodeoxynucleotide based on an unrelated
region of the E. coli gal operator (Fig. 5B,
lanes 4 and 5).
In the present study, we demonstrate that bacterial protease Lon is a sequence-specific DNA binding protein. Although Lon has been known for some time to bind DNA and in fact was originally purified by exploiting its DNA binding capabilities, recognition of DNA by Lon has previously been felt to be unrelated to specific promoter sequences. We now demonstrate that Lon binds very specifically to a TG-rich sequence found in the HIV-2 enhancer. Although it is clear, of course, that bacterial Lon plays no role in regulating HIV-2 gene expression, this finding strongly suggests that bacterial Lon targets similar regulatory sites in prokaryotic promoters, thus perhaps influencing gene expression (see below). Interestingly, our initial observations concerning the binding of Lon were made in E. coli Y1089, a strain that is felt not to express lon. However, we confirmed the presence of Lon in the strain we used by Western blotting analysis and protein sequencing.
In addition to purification and Southwestern blotting data, we have shown by use of shift-shift EMSA that the complex formed when the pets site probe interacts with E. coli whole cell extracts does indeed contain protease Lon. Further, Lon prepared through in vitro transcription and translation binds specifically to the pets sequence. Highly purified Lon is also capable of protecting the pets site in a DNase footprinting assay. Two other DNA binding proteins copurified with Lon (Fig. 4). These proteins do not appear to have a specific affinity for the pets site, because they were eluted off the DNA affinity column in a low salt wash. Further, in EMSA using recombinant versions of these proteins (L2 ribosomal and S3 ribosomal), either individually or in combination with each other and with Lon, they did not bind to the pets site.2
The functional significance of site-specific DNA binding by the
bacterial protease Lon remains to be determined. However, it is quite
tempting to speculate that this property of Lon may be intimately
involved in its ability to regulate gene expression. In particular, it
has been known for some time that Lon can specifically degrade certain
regulatory proteins at a rate much higher than would be expected in
view of their very low abundance in E. coli. This has
suggested to some investigators that Lon must have a way to be targeted
to these regulatory proteins (1). The demonstration that Lon is a
sequence-specific DNA binding protein suggests a mechanism by which
such targeting could take place. Specifically, Lon could bind to a site
immediately adjacent to where another regulatory protein binds on the
same promoter. Such close physical proximity may lead to the targeted
degradation of the second regulatory protein. In line with this model,
some early studies have suggested that Lon might control the level of
mRNA transcription from the E. coli gal operon (39, 40, 41).
A more recent example is the ability of the Lon protease from
Bacillus subtilis to prevent inappropriate transcription of
genes under the control of the sporulation transcription factor
G, possibly through
G degradation (23).
Although this model of a protease targeting an adjacent binding
regulatory protein has yet to be demonstrated in prokaryotes, it is of
note that a recent report suggests that protease activity can be linked
to the ability of a DNA binding protein to suppress transcription in
eukaryotes (42). A preadipocyte factor (termed AEBP1) has been cloned
that binds to a specific DNA element in the eukaryotic aP2 gene
promoter. AEBP1 can suppress transcription from the aP2 promoter and
also exhibits a carboxypeptidase activity. Interestingly, the
carboxypeptidase domain corresponds to the domain that mediates
transcriptional repression (42). Therefore, it is possible that Lon
affects gene expression in an analogous fashion. Regardless, it appears
likely that sequence-specific DNA binding contributes to the complex
mechanism of action of the Lon protease.
This work is dedicated to the memory of Diane Markovitz.
We thank Phil Andrews of the University of Michigan Protein Core Facility for protein sequencing and Shekelia Taylor for manuscript preparation. We also thank Alvin Markovitz (University of Chicago) for purified Lon, antibodies to Lon, clones, and numerous helpful discussions.