ABSTRACT
The DNA binding domain (DBD) of poly(ADP-ribose) polymerase (PARP) has proved to be a novel, highly sensitive probe
for detecting DNA breaks in intact cells undergoing apoptosis. A recombinant
peptide spanning the DNA binding domain of PARP was expressed, purified and
used to detect DNA strand breaks in fixed cells. Fluorescence microscopy with
this probe followed by detection with anti-PARP antisera initially revealed an increased binding following treatment
of cells with DNA strand-breaking agents (such as
N
-methyl-
N
'
-nitro-
N
-nitrosoguanidine) and, subsequently, using biotinylated PARP DBD, during
the later stages of apoptosis in several cell systems, when internucleosomal
strand breaks became evident. This procedure was found to be at least as
sensitive and required fewer steps to detect DNA strand breaks than those
utilizing Klenow incorporation of biotinylated nucleotides.
Poly(ADP-ribose) polymerase (PARP) is an abundant nuclear protein that is
associated with chromatin. This enzyme covalently attaches to and elongates
homopolymers of poly(ADP-ribose) to a number of nuclear proteins, using NAD, an abundant nucleotide
in eukaryotic nuclei, as substrate. PARP is a zinc finger-containing protein, allowing the enzyme to bind to either double- or single-strand DNA breaks without any apparent sequence preference.
Cell culture systems have demonstrated that PARP is involved in numerous
biological functions, all of which are associated with the breaking and
rejoining of DNA strands (
1
-
6
). The enzyme has an absolute requirement for DNA for activity and is activated
proportionately by the number of strand breaks in DNA. We recently demonstrated
that one of the earliest stages of apoptosis is characterized by activation of
PARP and poly(ADP-ribose) addition to nuclear proteins during the reversible stages of
apoptosis (
7
) and specific proteolysis of PARP has now been closely associated with a later
stage of programed cell death (
8
-
10
). This process occurs in a variety of cell types during organogenesis and
during maturation of the immune system. During apoptosis, clumps of
heterochromatin form adjacent to the nuclear matrix, nuclear fragmentation
occurs and, ultimately, membrane-enclosed apoptotic bodies appear. These changes are accompanied by an
increase in intracellular free Ca
2+
concentration. Increasing amounts of DNA strand breaks also occur during
apoptosis. The first strand breaks that are associated with DNA cleavage at
chromatin loops yield DNA fragment sizes >200 kb and can only be visualized by
pulsed field electrophoresis. This is the stage that corresponds with
activation of PARP (
7
). Later in apoptosis, a specific Ca
2+
/Mg
2+
-dependent nuclease is activated that cleaves DNA in the linker region
between nucleosomes, yielding a characteristic nucleosome ladder when the
chromosomal DNA is analyzed by agarose gel electrophoresis.
Visualization at the level of individual cells allows for the assay of
apoptosis. At the single cell level, the study of apoptosis requires
morphological examination of cells and nuclei, using chromatin- and DNA-specific fluorescent dyes, such as ethidium bromide, bis-benzamide and 4',6-diamidino-2-phenylindole. At the biochemical
level, DNA breaks have been detected
in situ
utilizing the free 3'-OH ends of DNA as a substrate for either terminal transferase or
the Klenow fragment of DNA polymerase I to incorporate biotin or digoxigenin,
which can be subsequently visualized with either visible or fluorescent dyes.
Proteolytic cleavage of PARP was first demonstrated in chemotherapy-induced apoptosis (
11
), where it was shown that PARP was processed into 85 and 24 kDa fragments. The
85 kDa fragment contains the catalytic and automodification domains, while the
24 kDa region consists of the DNA binding domain (DBD) of the enzyme. We
recently explored the significance of PARP cleavage in the osteosarcoma cell
model of apoptosis by examining the various participants in this specific
aspect of programed cell death by immunofluorescence in whole cells (
7
). In doing so, we recognized the potential to utilize the unique aspect of the
PARP DNA binding domain as a direct indicator of DNA strand breaks that occur
during apoptosis, as well as those that occur following DNA damage induced by
alkylation. Accordingly, the experimental validity and general characterization
of this new marker of apoptosis are described in detail below.
Human osteosarcoma cells (American Type Culture Collection no. 11226) were
cultured and apoptosis was induced as described previously (
9
). Burkitt lymphoma cell line BL-30 (
12
) and EBV-induced lymphoblastoid cell line YB-26 were maintained (
13
) and induced to undergo apoptosis (
14
,
15
).
The PARP DBD fusion protein was expressed in
Escherichia coli
(as described in detail in Results) and purified to >95% homogeneity in a
single step by Ni resin column chromatography (Qiagen). Bacterial cell lysate
was loaded onto a Ni-NTA column pre-equilibrated in buffer A (10 mM Tris-HCl, pH 8.0, 1% NP-40, 10 mM 2-mercaptoethanol, 6 M guanidine-HCl). After the column was washed
extensively in buffer A, buffer B (containing 8 M urea instead of guanidine-HCl) and buffer C (buffer B adjusted to pH 6.3), recombinant protein was
eluted with buffer D (buffer B adjusted to pH 5.7). Fractions were then
collected and analyzed by SDS-PAGE. SDS-PAGE revealed the size of the fusion protein to be ~30 kDa, consistent with the predicted molecular mass of the
PARP DBD attached to six histidine residues. PARP DBD protein was subsequently
renatured by dialysis against seven changes of dialysis buffer (50 mM NaCl, 0.5
mM ZnCl
2
and 10 mM MgCl
2
in 50 mM phosphate buffer, pH 7.2) containing decreasing concentrations of urea
(6, 4, 2 and 1 M), followed by decreasing concentrations of NaCl (1 M and 100
mM). Biotin labeling of PARP DBD was performed by incubating 5 [mu]l biotin (long arm)
N
-hydroxysuccinimide ester in dimethylsulfoxide (5 mg/ml) with 250 [mu]l PARP DBD (1 [mu]g/[mu]l) for 2 h at room temperature. The reaction was terminated
with 5 mg glycine and the biotinylated PARP DBD protein was then dialyzed
against 50 mM HEPES, pH 7.2, 20 mM ZnCl
2
, 100 mM NaCl and 70 [mu]l 2-mercaptoethanol.
Antibodies to PARP DBD were derived by immunization of rabbits with a peptide
corresponding to amino acids 25-41 of human PARP. Fixation of cells, immunofluorescence and immunoblot
analysis were performed as previously described (
7
,
16
).
DNA breaks were detected
in situ
using a Klenow fragment-based assay system (TACS1; Trevigen). Cells were fixed and labeled with
biotinylated nucleotides, using streptavidin-conjugated horseradish peroxidase and diaminobenzidine for detection. Cells were
counterstained with methyl green. Brown nuclei were positive for Klenow
labeling. DNA nucleosome ladders were observed by isolation of total genomic
DNA and agarose gel electrophoresis as described previously (
9
).
Cytosolic extracts were prepared from cultured human osteosarcoma cells by
homogenizing phosphate-buffered saline (PBS) washed cell pellets in 10 mM HEPES-KOH, pH 7.4, 2 mM EDTA, 0.1% (w,v) CHAPS, 5 mM dithiothreitol, 1 mM
phenylmethylsulfonyl fluoride, 10 [mu]g/ml
pepstatin A, 20 [mu]g/ml leupeptin, 10 [mu]g/ml
aprotinin (at 1 * 10
8
cells/ml) and collecting the post-100 000
g
supernatant. Assays contained 10 [mu]g protein from the cytosol fractions of osteosarcoma cells derived from days
2 and 8 with purified [
35
S]PARP (~5 * 10
4
c.p.m.), 50 mM PIPES-KOH, 2 mM EDTA, 0.1% (w/v) CHAPS and 5 mM dithiothreitol in a volume of
25 [mu]l. Incubations were performed at 37oC for 1 h and terminated by the addition of 25 [mu]l 2* SDS-PAGE sample buffer containing 4% SDS, 4% [beta]-mercaptoethanol, 10% glycerol, 0.125 M
Tris-HCl, pH 6.8 and 0.02% bromophenol blue. Samples were analyzed by SDS-PAGE and fluorography.
Many of the currently available methods for examining DNA strand breaks
in situ
rely on the ability of exogenous enzymes such as DNA polymerase or terminal
transferase to add labeled dNTPs to the 3'-OH ends of the strand breaks and subsequent detection of the
incorporated nucleotides by immunofluorescence microscopy. We reasoned that the
DBD of PARP might provide a more sensitive probe for DNA strand breaks that
would eliminate the requirement for the often labile enzymes and nucleotide
substrates.
Clone pCD12, containing the full-length cDNA encoding human PARP in an Okayama-Berg vector (
17
), was used as a polymerase chain reaction (PCR) template for construction of a
PARP DBD expression vector. PCR was performed with: (i) a 28 bp primer that
contained a
Bam
HI restriction site upstream (nt 164-180) of PARP cDNA; (ii) a 22 bp primer that contained a
Hin
dIII restriction site downstream (nt 837-854) of PARP cDNA. The PARP cDNA fragment thus amplified encompassed the
region that encodes the two zinc fingers of the enzyme as well as the KKKSKK
nuclear localization signal. Amplification was performed for 21 cycles and the
product was then ligated into the bacterial protein expression vector pQE30
(Qiagen).
The DBD of PARP was subsequently expressed in
E.coli
and purified to >95% homogeneity by affinity chromatography using a Ni-NTA column (Fig.
1
). The PARP DBD fusion protein was recognized on immunoblot analysis by
polyclonal antibodies, obtained subsequently, to this region of PARP (
7
). The double bands of the PARP DBD shown in the Coomassie stained gel in Figure
1
may be due to premature termination of transcription or translation, although
both of these proteins reacted specifically to antibodies to PARP DBD (not
shown). To establish conditions for detecting DNA strand breaks in fixed mouse
cells with the PARP DBD, we first adopted an immunofluorescence approach using
anti-human PARP. The antibody used does not react with the murine PARP (
18
), even though the amino acid sequences of the proteins are >80% identical (
17
,
19
). We therefore incubated mouse 3T3 cells for 30 min in the absence or presence
of 0.4 mM
N
-methyl-
N
'-nitro-
N
-nitrosoguanidine (MNNG) to induce a significant number of DNA breaks,
after which the cells were fixed on slides, incubated at room temperature with
excess purified PARP DBD (25 [mu]g/ml) for 1 h and washed with PBS. DBD bound to DNA strand breaks was then
detected by incubating the slides with the rabbit antibodies which recognize
human PARP DBD, followed by Texas red-conjugated goat antibodies to rabbit immunoglobulin IgG. Whereas no
immunofluorescence was detected in 3T3 cells not incubated with MNNG, marked
immunofluorescence was apparent in cells treated with the alkylating agent
(data not shown). These results indicated the feasibility of using PARP DBD to
detect DNA strand breaks in fixed cells.
To avoid the use of antibodies to detect the PARP DBD bound to DNA strand
breaks, we conjugated the bacterially expressed DBD to biotin so as to allow
detection by reaction with horseradish peroxidase-conjugated streptavidin and enhanced chemiluminescence (ECL; Amersham). We
first tested the modified DBD detection system in two human B cell lines that
are known to undergo apoptosis via endonuclease cleavage of DNA following serum
depletion, unlike normal B cells which become quiescent upon serum withdrawal (
15
). Apoptosis was induced in either a B cell line immortalized with EBV
in vitro
(Fig.
2
) or in Burkitt lymphoma-derived B cells (Fig.
3
) by withdrawal of autocrine growth factor as described (
15
). The occurrence of apoptosis was confirmed by a morphological assay (
20
) using fluorescence microscopy with a mixture of acridine orange and ethidium
bromide (data not shown). The cells were then examined by phase contrast
microscopy and by fluorescence microscopy with biotinylated DBD and horseradish
peroxidase-conjugated streptavidin (Figs
2
and
3
). In virtually all instances, only those cells showing the morphological
characteristics (cell shrinkage and nuclear condensation) of apoptosis were
stained by the biotinylated PARP DBD. The number of stained cells increased
with time after autocrine factor withdrawal.
Satoh and Lindahl (
4
,
5
) recently demonstrated that unmodified PARP binds to a damaged DNA plasmid
in vitro
and inhibits repair in the absence of NAD. It is hypothesized that PARP cycles
between an unmodified form, which blocks DNA strand ends, and a modified form,
which is released from DNA, thereby allowing access of repair enzymes.
Automodification of intact PARP by long chains of branched ADP-ribose polymers has in fact been shown to result in a loss of affinity of
the enzyme for DNA (
22
). We recently tested this model by cycling PARP
in vitro
with bacterially expressed deletion mutants of PARP (
6
). Our data using this
in vitro
assay of DNA showed that those mutants that possess an intact DBD and are
therefore able to bind to single-strand breaks inhibit DNA repair when added to a PARP-depleted HeLa cell extract. However, deletions in the
automodification domain or the NAD binding domain prevented alleviation of the
inhibition exerted by these mutants by NAD. We thus reasoned that the DBD could
be used as a tool to detect DNA damage in intact cells.
The cleavage of PARP into a separate DBD that cannot be automodified also
suggests the possibility that the 24 kDa cleavage product binds irreversibly to
the numerous strand breaks characteristic of the final stages of apoptosis.
This may account for the fact that expression of the DBD in living cells has
been shown to interfere with the DNA repair function of endogenous PARP (
23
,
24
).
We examined the above hypotheses directly by synthesizing a recombinant peptide
spanning this proteolytic fragment of PARP. We tested this new assay by
measuring binding of the recombinant DBD in 3T3 cells treated with MNNG, which
is known to induce strand breaks, as well as in other well-defined apoptotic systems. Immunofluorescence analysis demonstrated an
increased binding of excess biotinylated PARP DBD during the later stages of
apoptosis in osteosarcoma cells. This analysis was easier to perform and was at
least as sensitive as an assay utilizing Klenow incorporation of biotinylated
nucleotides.
In a comparison with a commonly used system based on Klenow incorporation of
biotinylated nucleotides for the detection of DNA strand breaks in fixed cells,
our biotinylated DBD method proved at least as sensitive (compare Figs
5
and
6
). The differential sensitivity of the two assays may relate to several factors,
including increased sensitivity of fluorescence. In addition, Klenow
incorporation of biotinylated nucleotides only occurs at double-stranded DNA 5' overhangs, but not with single-stranded DNA, double-stranded DNA with 3'-OH overhangs or double-stranded DNA with blunt ends. On the
other hand, the PARP DBD binds directly to all single-stranded DNA and double-stranded DNA breaks and requires no enzyme catalysis, indicating
that this a useful and simple tool for detecting apoptotic DNA breaks
in situ
.
This work was supported in part by grant CA13195 from the National Cancer
Institute and by funding from the United States Air Force Office of Scientific
Research through grant AFOSR-89-0053 and the United States Army Medical Research and Development
Command through contract DAMD17-90-C-0053.
REFERENCES
Return