| Nucleic Acids Research | Pages |
Crosslinking of proteins to DNA in human nuclei using a 60 femtosecond 266 nm laser
Introduction
Materials And Methods
Cells and nuclei preparation
Laser
Irradiation of nuclei
Separation of crosslinked complexes
Analysis of DNA crosslinking to the nuclear protein
Measurement of the amount of crosslinked proteins
Western blot analysis of Ku-70
SDS-PAGE and Coomassie brilliant blue staining
Digestion of high molecular weight DNA with ApaI, EcoRI and micrococcal nuclease
Nuclei digestion with HindIII restriction enzyme
Immunoprecipitation of complexes with H3 antibodies
Results
DNA crosslinking to nuclear proteins
Protein crosslinking to DNA
Protein integrity after irradiation of nuclei with a femtosecond UV laser
Integrity of DNA after irradiation of nuclei
Protein specificity of laser crosslinking
Photocrosslinking of Ku-70 in nuclei
Photocrosslinking of histone H3 to telomeres in nuclei
Discussion
Crosslinking efficiency
DNA integrity
Protein integrity
Crosslinking of specific proteins
Acknowledgements
References
Crosslinking of proteins to DNA in human nuclei using a 60 femtosecond 266 nm laser
Received June 3, 1999; Revised and Accepted July 29, 1999
ABSTRACT We developed appropriate conditions to use a laser with 60 femtosecond pulses, a frequency of 1 KHz and a wavelength of 266 nm to efficiently crosslink proteins to DNA in human nuclei for the purpose of using immunoprecipitation to study the binding of specific proteins to specific sequences of DNA under native conditions. Irradiation of nuclei for 30 min with 1-3 GW/cm2 pulses crosslinked 10-12% of total protein to DNA. The efficiency of crosslinking was dose and protein specific. Histones H1 and H3 were crosslinked by 15 min of irradiation with 20-25% efficiency, at least 10 times more strongly than the other histones, consistent with experiments using conventional UV light. Irradiation for 15 min did not damage proteins, as assayed by SDS-PAGE of Ku-70 and histones. Although the same level of irradiation did not cause double-strand breaks, it did make the DNA partially insensitive to EcoRI restriction enzyme, probably through formation of thymidine dimers. Immunoanalysis of crosslinked nucleoprotein showed that Ku crosslinking to nuclear DNA is detectable only in the presence of breaks in the DNA, and that nucleosomes are bound to a significant fraction of the telomeric repeat (TTAGGG)n.
INTRODUCTION
DNA-protein interactions are critical to the structure and function of genes. They play major roles in coordinating events in DNA repair, replication and gene expression. There are different methods to study DNA-protein interactions in vivo and in vitro. Enzymatic and chemical protection studies have been used to `footprint' nucleoprotein complexes along DNA. Gel shift assays have been used to analyze binding of specific proteins to specific DNA sequences. However, these techniques are best applied to nucleoprotein complexes formed in vitro rather than to the native nucleoproteins. Other techniques such as immunoprecipitation of native nucleoprotein complexes attempt to `capture' the native proteins bound to specific sequences. These methods have the potential to give artificial results, because the proteins are non-covalently bound to DNA and are subject to redistribution during chemical treatment, isolation or immunoprecipitation of the DNA-protein complex. The problem can be overcome by chemical or UV-light-induced crosslinking to `freeze' DNA-protein interactions.
Chemical crosslinking with formaldehyde has been extensively used to prepare samples for electron microscopy, and study the distributions of histones and other proteins along DNA (see for example 1,2). However, because chemical crosslinking usually changes protein charge, creates protein-protein crosslinks and is a multi-step process that produces chemical bridges of significant length, perturbations in structure can result (3). Also, because chemical crosslinking is slow it cannot be used for kinetic studies and might trap thermodynamically unfavorable states.
Photochemical crosslinking of proteins to the nucleic acids has been applied to study a variety of nucleic acid binding proteins (4-7). Irradiation with UV light of wavelength near 260 nm produces a `zero length' covalent bond between contact points of nucleic acid and protein and is complete within a microsecond (7). Therefore, it is believed to produce less perturbation to structure than chemical crosslinking. In addition, 266 nm laser light does not produce protein-protein crosslinks (8). Photocrosslinking with low power UV lamps requires long exposures to obtain even low levels of crosslinking and could cause enough photodamage to the DNA to cause redistribution of proteins (9). Some of these problems can be avoided using high powered UV lasers, which achieve high efficiency of crosslinking, sometimes in a single pulse (6,10).
Crosslinking of nucleoproteins in vitro with high power lasers has been shown to increase the crosslinking yield up to two orders of magnitude greater than with a conventional lamp. For example, irradiation of reconstituted chromatin with a picosecond laser crosslinked 15-20% of histones (8), compared to ~0.1% with a conventional lamp (7). Crosslinking efficiency of reconstituted chromatin in solution increases with light intensity and the number of absorbed photons (8,11). The crosslinking efficiency is also higher for single-stranded than double-stranded DNA (8). It was shown that the crosslinking efficiency of progesterone to single-stranded DNA increases for shorter pulses. For instance, the crosslinking efficiency of progesterone receptor was five times higher using a 200 femtosecond laser than a 100 picosecond laser with the same total energy and 10 times higher peak intensity (11).
Irradiation of reconstituted nucleosomes with a picosecond laser did not produce histone-histone crosslinking or protein damage (8). However, there is evidence from other studies that laser irradiation can damage both proteins and DNA. For instance, irradiation of a Rap1-DNA complex with a nanosecond laser can damage up to 25% of the Rap1 as estimated by protein gel analysis (10). There are no data on protein integrity after irradiation with femtosecond lasers.
UV light can also modify DNA. Conventional UV lamps most commonly produce pyrimidine dimers (reviewed in 12). The formation of pyrimidine dimers is associated with the excitation of the triplet state of the DNA base. Because high power picosecond or femtosecond pulses do not extensively populate the triplet states of the bases, pyrimidine dimers are reduced by a factor of more than 10 depending on irradiation intensity (12). Nevertheless, laser irradiation increases the yield of single-strand and double-strand breaks, and interstrand crosslinking (12). The yield of modifications depends on wavelength, intensity and pulse length of the light as well as experimental conditions (reviewed in 13). For example, using Tris-containing buffer instead of water decreases the yield of double-stranded breaks 4-fold (13). Irradiation of a DNA solution using a 200 femtosecond laser can lower the yield of a PCR reaction 5-fold as compared to irradiation using a picosecond laser (11) at comparable dose.
Laser irradiation of cells or intact nuclei has the potential to crosslink nucleoproteins in their most native state. However, there could be fundamental differences between the results of crosslinking in cells or nuclei compared to crosslinking isolated nucleoproteins, as done in most studies. First, suspensions of cells or nuclei scatter significant amounts of the incident light, reducing the intensity of the light within the samples. Second, the local concentration of DNA in nuclei is very high, leading to an internal filtering effect that could also decrease the effective intensity of the light. Third, the high local concentration of DNA in nuclei might trap free radicals that could increase photochemical degradation of the DNA and proteins.There are not adequate experimental data using high power lasers to crosslink nuclei to determine whether any of these hypothetical problems exist. In one study of nuclei a 260 nm nanosecond laser was able to crosslink 6-12% of proteins (14); however, protein and DNA integrity were not studied (14). Angelov et al. (8) successfully used a 266 nm picosecond laser to crosslink specific histone variants to abundant genomic sequences in nuclei
Our interest is to study the structural organization of human telomere nucleoprotein, using immunoprecipitation to map the positions of specific proteins with respect to the ends of chromosomes. Because telomeres represent only ~0.01% of human DNA, we need to achieve efficient crosslinking under conditions that DNA can still be solubilized and analyzed using restriction digestion, amplification and hybridization, and the proteins are sufficiently intact to be immunoreactive. Because we had access to a laser with higher power and shorter pulses than used in previous studies we needed to evaluate the crosslinking efficiency, and properties of the crosslinked protein and DNA using this laser to determine whether it was appropriate for our studies and whether it might offer specific advantages over UV sources used in earlier studies.
We crosslinked intact human nuclei with a 60 femtosecond 266 nm laser and quantitatively analyzed the DNA and protein crosslinking efficiency as functions of irradiation time. We developed a procedure to quickly isolate crosslinked DNA from equilibrium CsCl gradients. Analysis of the density shift of specific DNA sequences in the CsCl gradients was used to estimate the frequency of protein crosslinking to specific parts of the genome. SDS-PAGE was used to assay for protein degradation and protein-protein crosslinking, as well as to determine the spectrum of crosslinked proteins. We analyzed the integrity of the DNA strands and tested the ability to digest the DNA with restriction enzymes. We have shown that histones H1 and H3 can be efficiently crosslinked to DNA, including telomeric DNA, and that the Ku antigen can be crosslinked to nuclear DNA with double-stranded breaks. We conclude that crosslinking of nuclei with femtosecond lasers is a good technique for analysis of DNA-protein interactions, because it is very efficient and leaves proteins and DNA sufficiently intact for common analytical and preparative procedures.
MATERIALS AND METHODS
Cells and nuclei preparation
Cells were grown and nuclei prepared essentially as according to Makarov et al. (15). Human 293 cells were grown to ~80% confluence at 37°C and 5% CO2 in minimal essential Eagle's medium (in Earle's BSS with 2× essential and non-essential amino acids, vitamins and 10% non-inactivated fetal bovine serum), removed from flasks with trypsin and passaged in fresh medium at a density of 1.3 × 104 cells/cm2. Cells (1-2 × 108) were harvested by centrifuging three times for 10 min at 800 g in 15 ml cold PBS followed by resuspension in PBS (~2 × 108/ml). Nuclei were prepared using centrifugations at 4°C as above: 1-2 × 108 washed cells were centrifuged once in 15 ml of nuclear buffer [60 mM KCl, 15 mM NaCl, 15 mM HEPES pH 7.4, 3 mM MgCl2, 6 µM leupeptin, 1 mM phenylmethylsulfonyl fluoride (PMSF)], once in 1.5 ml nuclear buffer, twice in 15 ml nuclear buffer with 0.1% digitonin. Nuclei were resuspended in 1 ml of nuclear buffer and diluted to 107 nuclei/ml with nuclear buffer prepared with 50% glycerol, and frozen in liquid N2.
Laser
Ultrashort UV pulses were produced using a novel frequency-mixing technique in which 20 femtosecond pulses at the fundamental (800 nm) and the second harmonic (400 nm) wavelengths of an amplified Ti:sapphire laser system were mixed in a gas-filled hollow waveguide (16). Output pulses from the amplifier (20 fs, 600 µJ) (17) were first frequency-doubled in a 300 µm thick beta-barium borate crystal with a conversion efficiency of 20%. After delay and alignment of the polarizations, the blue and near-IR pulses were recombined with a dichroic splitter. The collinear beams were then focused with a curved mirror to the entrance aperture of a 140 µm i.d. capillary held in a cell. The cell had 250 µm sapphire windows, and was filled with argon gas to a pressure that gave optimum conversion efficiency. The 266 nm light was separated using dielectric mirrors, then brought to a beam size of 2-3 mm at the sample. The 1 kHz output pulses were 60 fs in duration in air with peak power of 1.3-3.0 GW/cm2.
Irradiation of nuclei
For all experiments, nuclei were irradiated at 4°C in 1.5 ml polypropylene microcentrifuge tubes containing 100 µl aliquots at 1 mg/ml concentration in Buffer A (15 mM HEPES pH 7.8, 60 mM KCl, 15 mM NaCl, 2 mM MgCl2, 1 mM PMSF, 1 µM leupeptin), which were continuously stirred using a 3 mm magnetic stirring bar and standard lab mixer.
Separation of crosslinked complexes
The irradiated nuclei were washed once in Buffer A and resuspended in Buffer B (10 mM Tris-HCl pH 8, 2 mM EDTA, 1% sarcosyl, 1 mM PMSF, 1 µM leupeptin). For the analysis of DNA crosslinking to protein nuclei were sonicated on ice in 30 s bursts for 2 min so that the average size of DNA was 500 bp, as measured by agarose gel electrophoresis. For other experiments nuclei were sonicated for 0.5-1 min to produce an average size of DNA of 2-3 kb. To separate the crosslinked nucleoprotein from free protein and DNA, 200-500 µg aliquots of sonicated nuclei were mixed with 1.71 g/cm3 CsCl in 10 mM Tris-HCl pH 8, 1 mM EDTA, 1 mM PMSF and loaded into 5.1 ml tubes. Centrifugation was performed in a Beckman L5-65 centrifuge with VTI-80 rotor spun at 60 000 r.p.m. for 3 days at 23°C. Gradient fraction collection was monitored by measuring 254 nm absorbance using an ISCO detection system.
Analysis of DNA crosslinking to the nuclear protein
Nuclei were irradiated for 0, 5, 15 or 30 min, sonicated and centrifuged in CsCl. The CsCl gradients were collected into 60 µl fractions, which were immediately measured in a refractometer to determine the density (18). The fractions were diluted 15 times in water and vacuum-transferred onto a Zeta-probe membrane. The membrane was prehybridized in hybridization solution (37% formamide, 0.25 M NaCl, 0.125 mM NaHPO4, 7% SDS) for 10-15 min at 50°C and then hybridized with a 32P-labeled human 18S rDNA probe (851 bp ApaI fragment from human 18S rRNA plasmid p18S from A. Conconi, Department of Biochemistry and Biophysics, Washington State University, Pullman, WA) in fresh solution overnight at 50°C. The membrane was then washed in: (i) 2× SSC, 0.1% SDS, 15 min, 24°C; (ii) 0.5× SSC, 0.1% SDS, 15 min, 24°C; (iii) 0.1× SSC, 0.1% SDS, 15 min, 24°C; and (iv) 0.1× SSC, 0.1% SDS, 20 min, 50°C. Membranes were exposed overnight and read by a Molecular Dynamics PhosphorImager. The membrane was analyzed using ImageQuant software (Molecular Dynamics). The hybridization signal of the fraction was plotted versus the fraction density. The final plot was made and analyzed by using Igor (WaveMetrics)
Measurement of the amount of crosslinked proteins
Nuclei were irradiated and crosslinked complexes were prepared as above, except that the crosslinked and uncrosslinked DNA was collected in a single fraction. The amount of crosslinked protein was measured using the Pierce colorimetry kit calibrated with the Pierce BSA solution. DNA quantities were determined by measuring absorbance at 260 nm.
Western blot analysis of Ku-70
Nuclei were irradiated for 0, 5, 15 or 30 min. Total nuclear protein or micrococcal nuclease (MNase) digested crosslinked complexes were separated by 6% SDS-PAGE. Colored protein markers (Novex) were used to monitor the progress of the gel run. Proteins were electro-transferred to PVDF membrane overnight at 4°C. The efficiency of blotting was controlled by checking blotting efficiency of the colored marker bands. The monoclonal antibody against amino acids 506-541 of Ku-70 was purchased from NeoMarkers and used at 1:1000 dilution. Western blot analysis was performed using ECL blotting and detection reagents (Amersham). The image was recorded on Kodak X-ray film, scanned using a Star 1 cooled CCD camera (Photometrics) and analyzed using ImageQuant.
SDS-PAGE and Coomassie brilliant blue staining
Laemmli 6% and 15% SDS-PAGE were used to separate proteins for the analysis of Ku-70 by western blotting and histones by Coomassie brilliant blue staining, respectively. The colored marker from Novex was used for calibration.
Digestion of high molecular weight DNA with ApaI, EcoRI and micrococcal nuclease
Nuclei irradiated for different times were digested with Proteinase K overnight and extracted twice with phenol-chloroform. The supernatant was precipitated with ethanol and resuspended in TE. Restriction digestions with EcoRI and ApaI (Boehringer Mannheim) were performed using manufacturer's buffers at 37°C. The reactions were stopped with 20 mM EDTA and DNA was phenol-chloroform purified and precipitated with ethanol. MNase digestion was performed as follows. The total DNA peak after CsCl gradient centrifugation was collected, diluted twice with water, an equal volume of isopropanol was added and the mixture was incubated for 15 min at room temperature. The solution was centrifuged at 14 000 g for 15 min at room temperature. The pellet was dissolved in 4 M guanidinium-HCl and precipitated with 3 vol of ethanol overnight at -20°C. The pellet was washed with 70% ethanol three to four times and solubilized in 10 mM Tris-HCl pH 9, 100 mM NaCl, 1 µM leupeptin at 37°C for 4 h. The sample was divided into two equal parts. One part was digested with 2 U of MNase per 50 µg of DNA in the buffer 10 mM Tris-HCl pH 9, 100 mM NaCl, 2 mM CaCl2, 1 µM leupeptin overnight at 37°C. The non-digested sample was stored at -20°C.
Nuclei digestion with HindIII restriction enzyme
Nuclei in a buffer A were incubated with 20 U of HindIII per µg of nuclei for 2 h at 37°C. Nuclei were washed three times with buffer A to remove restriction enzymes.
Immunoprecipitation of complexes with H3 antibodies
Immunoprecipitation was performed as described by Dimitrov et al. (19) with minor modifications. DNA-protein complexes from irradiated nuclei were prepared as for MNase digestion except that the pellet was resuspended at a concentration of 2 mg/ml in TE, 0.1% SDS, 6 µM leupeptin. Aliquots of 15-20 µg of complexes were incubated with 1 µg of anti-H3 from Maine Biotechnology Services, Inc., in 100 µl of 50 mM HEPES pH 7.5, 1 M NaCl, 0.1% SDS, 1% Triton X-100, 1% Na-deoxycholate, 5 mM EDTA, 0.1% BSA for 4 h at room temperature. Sixty microliters of Protein G-Sepharose and 340 µl of additional buffer were added. Following overnight rotation at 4°C, the suspension was washed three times with 0.5 ml of the same solution, twice with 50 mM HEPES pH 7.5, 0.25 M NaCl, 0.1% SDS, 1% Triton X-100, 1% Na-deoxycholate, 5 mM EDTA, 0.1% BSA and three times with PBS. PBS was replaced with TE + 0.5% sarcosyl and incubated with 200 µg of Proteinase K at 45°C for 2 days. The suspension was spun down at 14 000 g for 10 min. The supernatant was ethanol precipitated. The pellet was dissolved in TE and loaded on Zeta-Probe Plus membrane. The dot blot analysis was performed as follows. The membrane was hybridized with T4 kinase-labeled (TTAGGG)4 and washed as described by Bedoyan et al. (2). The autoradiogram was exposed overnight. The hybridization signal was analyzed using ImageQuant. Signal was linear as determined by analysis of serial dilutions of the naked DNA after laser irradiation for 15 min. The membrane was stripped and hybridized with random prime labeled total human DNA. The membrane was exposed for 2 h and analyzed as above.
RESULTS
DNA crosslinking to nuclear proteins
We irradiated 100 µg of human nuclei in 100 µl aliquots with 60 fs, 1 KHz, 266 nm laser pulses with peak intensity 1.3-3.0 GW/cm2. At this intensity the non-linear effect of the light absorption by water is negligible and the broadening of the pulse does not exceed 10-15% in the top 1 mm layer of water (20). The nuclear concentration was chosen to crosslink large amounts of nuclei in a reasonable time. At this concentration a 20 µm layer of sample absorbed or scattered ~50% of the incident light (data not shown). The samples were continuously stirred, which allowed efficient crosslinking without disrupting the nuclei. Intermittent stirring with a pipette gave much less crosslinking.
After irradiation we sonicated the nuclei to fragment the DNA to a size of ~500 bp, followed by purifying the DNA from free protein by equilibrium CsCl centrifugation in order to increase the solubility of the complexes and decrease artifacts due to immunoprecipitation of non-covalently bound nucleoproteins. Without separation of the DNA from free protein it was extremely difficult to resuspend crosslinked complexes in non-denaturing buffer after precipitation with ethanol.
UV crosslinking of proteins to DNA is dependent on the type of proteins and DNA sequence (7). We first chose to analyze the performance of the 60 femtosecond laser by studying crosslinking to the 18S rRNA genes, because these genes are present as a multi-copy family that binds histone and non-histone proteins (19). Southern dot-blot analysis of CsCl fractions was performed to determine which fractions contain 18S rDNA (Fig. 1). We found two distinct peaks after 30 min of irradiation. The heavier peak has the expected density of double-stranded DNA (1.7 g/cm3) and coincides with the peak from non-irradiated nuclei. The lighter density peak at 1.685 g/cm3 is expected to contain crosslinked nucleoprotein. Because the two peaks are spatially separated by about six fractions, it is possible to quantify the fraction of rDNA that has become crosslinked. From the relative areas under the curves for the shifted and unshifted peaks we estimate that 10-45% of the rDNA fragments were crosslinked to protein during the time course of irradiation. Probing the same CsCl fractions with telomeric DNA, we determined that >50% of the telomeric fragments were crosslinked at 30 min (data not shown).
Figure 1. Dot-blot analysis of CsCl gradient fractions hybridized to the 18S rDNA probe after different irradiation times as functions of density. Times of irradiation: 0 min (filled squares), 5 min (filled circles), 15 min (open circles) and 30 min (open squares).
Protein crosslinking to DNA
First, we analyzed protein crosslinking efficiency as a function of the time of irradiation by directly measuring the amount of protein in the CsCl-purified total DNA material (Materials and Methods). Human nuclei have a protein:DNA mass ratio of 1:1 (21). We measured the amount of protein using the colorimetric assay and the DNA using UV absorption. The ratio of protein to DNA was 2% at 5 min, 9% at 15 min and 12% at 30 min of irradiation, after subtracting the background signal of 0.75% from the 0 min crosslinked control sample (Fig. 2). The crosslinking efficiency begins to saturate at higher doses, qualitatively similar to results of in vitro experiments with femtosecond and picosecond lasers (11).
Figure 2. Protein:DNA ratio of crosslinked nucleoprotein as a function of irradiation time. DNA and protein amounts in the crosslinked complexes isolated from nuclei by CsCl were determined by absorbance at 266 nm and colorimetry assay respectively.
Second, we used the density of the crosslinked 18S rDNA nucleoprotein peak (Fig. 1) to estimate the amount of protein bound to the specific sequences of DNA, assuming that the final hydrated volume of crosslinked nucleoprotein is equal to the sum of hydrated volumes of the free DNA and protein under the same conditions. We used an average density of 1.3 g/cm3 for protein and 1.7 g/cm3 for DNA (22), and an experimental density of 1.685 g/cm3 for the 30 min crosslinked complex, giving the relationship:
1/1.685 = [alpha]/1.3 + [beta]/1.7,
where [alpha] is the weight fraction protein and [beta] is the weight fraction DNA
By using this equation, we estimate that weight ratio of protein to 18S rDNA ([alpha]/[beta]) in the light density peak is 0.03 after 30 min irradiation. Taking into account the fact that 55% of the DNA was not crosslinked, the calculated weight ratio of crosslinked protein to total 18S rDNA is only 0.014. Applying the same assumptions to the 30 min data on telomere nucleoprotein we calculated a protein:DNA weight ratio of 0.06-0.09.
Protein integrity after irradiation of nuclei with a femtosecond UV laser
In order to study protein binding to specific sequences in nuclei, the proteins should remain intact and not undergo protein-protein crosslinking. Irradiation of the RAP1-DNA complex in vitro with a nanosecond UV laser can damage up to 25% of the total protein, as assayed by Coomassie blue staining of SDS polyacrylamide gels (10). On the other hand, using the same assay Stefanovsky et al. (5) did not find any degradation of nucleosomes after irradiation with a picosecond UV laser in vitro. Therefore, it is important for us to characterize the damage of proteins after femtosecond UV laser irradiation in suspensions of human nuclei. Low power UV light can produce protein-protein crosslinking because of the large dispersion of wavelengths (23). Protein degradation would interfere with detection and immunoprecipitation of proteins, and protein-protein crosslinks would make it impossible to distinguish between DNA-binding proteins and proteins in contact with DNA-binding proteins.
We tested the integrity of proteins after femtosecond laser irradiation using Ku antigen and histones. Ku antigen is an abundant protein thought to be bound to DNA in vivo, and is known to photocrosslink to DNA in vitro (24-26). Ku antigen binds to DNA as a heterodimer with subunits of 70 kDa (Ku-70) and 80 kDa (Ku-80). Both proteins have significant amounts of aromatic residues [9 and 8% respectively (27,28)] so they should be sensitive to degradation and protein-protein crosslinking. Histones are the most abundant proteins bound to human DNA and make multiple contacts with each other as well as the DNA.
Human nuclei were irradiated for different times and sonicated. Equal amounts of protein from different time points were separated by SDS-PAGE and western blot analyzed using antibody against Ku-70. The results are shown in Figure 3A. Ku-70 degradation was undetectable after 5 and 15 min of irradiation; however, a small amount of degradation product was detected after 30 min. Dilution experiments indicated that our detection limit for Ku-70 was ~1 ng of total nuclei (data not shown). Because each lane in Figure 3A was loaded with ~4 µg of nuclei, we estimate that 0.025% of Ku-70 was degraded, and less than that percentage of Ku-70 was crosslinked to Ku-80. No band is visible at 150 kDa, the mass of the Ku-70/Ku-80 dimer. Total Coomassie blue staining of proteins separated by SDS-PAGE after irradiation of nuclei is shown in Figure 3B. Analysis of histones bands, marked by arrows, shows that the histones are mainly intact, with no detectable histone-histone crosslinking or degradation. We conclude that femtosecond UV irradiation did not cause significant damage or protein-protein crosslinking of Ku-70 or histones.
Figure 3. Effects of laser irradiation upon proteins in human nuclei. (A) Western blot of Ku-70. Samples of 4 µg of sonicated nuclei irradiated for different times were separated by SDS-PAGE and analyzed by western blotting. (B) Samples of 20 µg of sonicated nuclei irradiated for different times were separated by SDS-PAGE and stained with Coomassie blue.
Integrity of DNA after irradiation of nuclei
Analysis of nucleoproteins often involves restriction enzyme digestion and size separation by electrophoresis. Therefore, it is important to determine whether irradiation of the nuclei modified the DNA to prevent recognition by restriction enzymes or caused strand cleavage.
We analyzed the products of the digestion of 18S rDNA with two restriction enzymes following irradiation of nuclei for different times. The result of digestion with EcoRI is presented in Figure 4A. The digestion of rDNA by EcoRI should produce a 6.1 kb fragment that will hybridize with an 800 bp ApaI fragment of the 18S rRNA gene. The single 6.1 kb long fragment was present after 0 and 5 min of irradiation. However, 20 and 75% of the total signal remained in the loading wells or migrated abnormally slowly after 15 and 30 min of irradiation, respectively, indicating that damage to the DNA was preventing complete cleavage by EcoRI. A weak 3 kb band and increased background hybridization was detected in the 30 min irradiated sample, indicating some non-specific cleavage. The total hybridization in each of the lanes was comparable, indicating that irradiation did not significantly interfere with hybridization, e.g., by extensive interstrand crosslinking.
Figure 4. Southern blot analysis of the restriction digested human rRNA genes. High molecular weight DNA was isolated from nuclei after irradiation for 0, 5, 15 and 30 min. The DNA was digested with restriction enzyme, purified and equal amounts loaded and electrophoresed in 1% agarose. The DNA fragments were electroblotted and hybridized with a random prime labeled 800 bp 18S rDNA fragment. (A) Digestion with EcoRI. (B) Digestion with ApaI. Lane M is a 100 bp ladder marker.
Figure 4B shows the result of digestion of the same DNA samples with ApaI, which should produce a single 800 bp band. A diffuse band at 800 bp was found after irradiation for 0, 5 or 15 min. The band after 30 min of irradiation was perhaps somewhat slower and more diffuse.
One possible explanation for the difference in sensitivity of DNA for digestions with EcoRI and ApaI is that the EcoRI recognition site is AT rich (GAATTC) and digestion can be affected by thymidine dimers. On the other hand, the recognition site of ApaI is GGGCCC and the effects of thymidine dimers could be minimal.
Protein specificity of laser crosslinking
It is very important to know the protein specificity of laser crosslinking. Here, we analyzed the weight spectra of crosslinked proteins after laser irradiation by using standard SDS-PAGE gel and staining with Coomassie blue.
Nuclei were irradiated for 0 and 15 min with the femtosecond laser and sonicated to an average DNA size of 3 kb followed by separation of the nucleoprotein complexes by CsCl gradient centrifugation and identification of the free and crosslinked proteins by SDS-PAGE (Fig. 5). Crosslinked proteins were released by MNase digestion before SDS-PAGE. No free or crosslinked proteins were detected after 0 min of irradiation (lane 1). A single weak band of free protein (band 4) was detected after 15 min of irradiation, without MNase digestion (lane 2). Because the intact DNA was 3 kb in size, crosslinked proteins would not have entered the gel. Thus, the CsCl centrifugation has removed all detectable amounts of protein except small amounts of protein in band 4. After MNase digestion there are four strong protein bands (lane 3, bands 1-4), which co-migrate with histone bands from intact nuclei (lane 4) and represent those abundant proteins that were efficiently crosslinked to the DNA. Band 1 co-migrates with intact histone H3. Band 2 is slightly retarded relative to H3 and could represent H3 still bound to a short fragment of DNA. It was reported that MNase could leave three to five nucleotides crosslinked to the proteins (29). Band 3 co-migrates with the slower species of free H1 and probably represents intact or slightly retarded H1. Band 4 could represent a retarded H1 species and/or the uncrosslinked protein that is seen in lane 2. Band 4 was not apparent in all crosslinked preparations and thus is probably an artifact of isolation rather than a crosslinked protein.
Figure 5. SDS-PAGE of protein from crosslinked nuclei. Lane 1, 80 µg of DNA after irradiation for 0 min and digestion with MNase; lane 2, 80 µg of DNA after irradiation for 15 min; lane 3, 80 µg of DNA after irradiation for 15 min and digestion with MNase; lane 4, 20 µg of total nuclear protein. Arrows on the left show the positions of crosslinked proteins. Numbers indicate the protein band number (discussed in Results). Positions of the histones are shown on the right. All lanes are from the same gel.
Assuming that bands 1 and 2 are H3, and band 3 is the slower form of H1, densitometry analysis of lanes 3 and 4 indicate that the crosslinking efficiencies of histones H3 and H1 are 24.5 and 21% respectively. The crosslinking efficiencies of histones H2A, H2B and H4 are at least 10 times lower. These results are in agreement with the crosslinking efficiencies of different histones using irradiation with conventional UV lamps (23). Quantitation of the gel indicates that histones H3 and H1 comprise ~96% of the crosslinked proteins and that the total amount of crosslinked protein is at least 6% of the original protein in the nuclei. This is consistent with our previous estimation of 10-12% using the colorimetric assay.
Photocrosslinking of Ku-70 in nuclei
Ku antigen is an important protein for repair of DNA. Ku antigen binds to the ends of DNA in vitro, can be crosslinked to DNA in vitro, and associates with chromatin in vivo (25,30,31). We decided to analyze crosslinking of Ku protein to DNA as an example of a non-histone protein that is abundant in vivo.
Nuclei were irradiated for 15 min and the total DNA peak was collected from the CsCl gradient. DNA was completely digested with MNase and the equivalent of 100 µg of nuclei was separated by SDS-PAGE followed by western blot analysis of Ku-70. No Ku-70 was detected in the dot-blots of the crosslinked, MNase-digested DNA, although positive control blots with as little as 1 pg of intact nuclei were detectable (data not shown). A strong signal should have been seen with even 0.1% crosslinking of Ku-70. There could be a several reasons why Ku-70 was not detected: (i) Ku-70 binds to nuclear chromatin differently than to naked DNA; (ii) most of Ku-70 was not bound to DNA in nuclei; or (iii) bound Ku-70 is not laser-crosslinked to DNA in intact nuclei.
We attempted to stimulate Ku-70 crosslinking by creating double-strand breaks in the nuclear DNA. It is known that the yeast homolog of Ku can recognize double-stranded breaks induced by restriction enzyme in vivo (32). We digested our human nuclei with HindIII to an average DNA length of ~20 kb and irradiated the degraded nuclei with the femtosecond laser for 15 min. The total DNA peak after CsCl gradient centrifugation was collected and digested with MNase. Restricted and non-restricted nuclear samples were dotted onto filter and probed with the Ku-70 antibodies (Fig. 6). Different dilutions of naked DNA (row D), total nuclear protein (row N) and crosslinked complexes from non-restricted (row C -R) and restricted nuclei (row C +R) were analyzed by western blotting against Ku-70. While no signal was detected for the non-restricted sample and naked DNA, we were able to detect a weak signal from DNA taken from nuclei digested with HindIII and laser crosslinked. The membrane was calibrated using total nuclear protein row N and the crosslinking of Ku-70 (from row C +R) was determined to be ~0.01% of the total Ku-70 in the nuclei.
Figure 6. Western dot-blot analysis of the crosslinking of the Ku-70 protein. Nuclei were digested with HindIII restriction enzyme and irradiated for 15 min with the femtosecond laser. Nuclei were sonicated to average DNA size of 3 kb, separated on a CsCl gradient, collected and desalted. The DNA pellet was dissolved in MNase digestion buffer and digested using MNase. Different amounts of the restricted/CsCl-separated irradiated nuclear material (100, 50, 20, 10 and 5 µg) were blotted on PVDF membrane (row C +R). The same amounts of unrestricted/CsCl-separated irradiated nuclear material were also blotted (row C -R). Row D is a control with the same amounts of naked DNA digested with MNase. Row N was blotted with 1, 0.5, 0.2, 0.1, 0.05, 0.02 and 0.01 µg of intact, control nuclei. The filter was probed with Ku-70 antibodies as explained in Materials and Methods.
Photocrosslinking of histone H3 to telomeres in nuclei
Immunoprecipitation can be used to detect whether a specific protein is bound to a specific sequence of DNA or RNA (19). To do this under the most stringent conditions, the protein must be crosslinked to the DNA in nuclei, the crosslinked nucleoprotein complex separated from free protein and precipitated using an antibody specific to the protein, and the co-precipitated DNA analyzed by hybridization.
We assayed the presence of histone H3 on human telomeric DNA to test whether immunoprecipitation could be effectively used on nucleoprotein from femtosecond laser crosslinked nuclei. Vertebrate telomeres have the sequence (TTAGGG) repeated thousands of times at both ends of each chromosome. In an earlier study we had digested vertebrate nuclei with structure-sensitive nucleases and detected fragments of the telomeric DNA that were characteristic of a closely spaced array of nucleosomes on telomeres (33,34). This was strong evidence of histone binding to telomeres. The `nucleosome ladders' were very strong in all organisms except human, which has relatively short telomeres. The weak human patterns have been attributed to low nucleosome occupancy (32). However, the diffuse patterns could also be the result of irregular nucleosome spacing or non-histone protein binding to the nucleosomes (35). Because nucleosomes contain histone H3, which is efficiently crosslinked to bulk chromatin (Fig. 5), we expect that anti-H3 antibodies should co-precipitate crosslinked telomere DNA fragments, and that the fraction of telomeric DNA precipitated should indicate whether nucleosomes are a major component of human telomeres.
Nuclei irradiated for 15 min were sonicated to an average DNA size of 2-3 kb, and the crosslinked complexes were CsCl isolated and sonicated further to ~400 bp. The nucleoprotein was immunoprecipitated from 1 M salt using anti-H3 and protein G-Sepharose. The fractions of telomere and bulk DNA precipitated were measured using dot-blot hybridization (Fig. 7). Table 1 shows that 5.2% of the bulk DNA and 4.3% of the telomeric DNA were precipitated, whereas non-specific precipitations were 0.1-0.3%. We conclude that histone H3 is crosslinked to a major fraction of the telomeric DNA. The reduced fraction of telomere relative to bulk DNA is possibly due to a small fraction of non-nucleosomal structure as well.
Figure 7. Dot-blot analysis of immunoprecipitation of 20 µg samples of crosslinked nucleoprotein complexes with anti-H3 antibodies. (A) Hybridization of precipitated material with human telomere-specific probe. (B) Autoradiogram of the same filter as in (A), re-probed with random prime labeled bulk DNA. Row Ab, absence (-) or presence (+) of anti-H3 in the precipitation reaction; row pC, precipitation of crosslinked complexes; row pND, precipitation of naked DNA; row cND, dilutions of control uncrosslinked DNA, with the amounts in µg.
Table 1. Co-precipitation of bulk and telomere nucleoprotein DNA with antibodies against histone H3 (from data in Fig. 7)
| Crosslinked DNA | Anti-H3 | Precipitated (%) | Naked DNA | Anti-H3 | Precipitated (%) |
| Telomere | + | 4.3 | Telomere | + | Not detected |
| Telomere | - | 0.1 | Telomere | - | 0.2 |
| Bulk | + | 5.2 | Bulk | + | 0.2 |
| Bulk | - | 0.2 | Bulk | - | 0.3 |
DISCUSSION
We tested a novel technique to crosslink DNA to proteins within intact nuclei using 60 fs, 1 kHz, 266 nm, 1-3 GW/cm2 laser pulses. The crosslinking efficiency of DNA to protein is ~10-12% and depends on the time of irradiation. Protein degradation and protein-protein crosslinks are minimal or not detectable. DNA is partially sensitive to digestion by restriction enzymes and MNase.
Crosslinking efficiency
The crosslinking efficiency of histones to DNA is reported to be 10-15% for a picosecond laser (8) and 1-9% for a conventional UV source (22,36). We estimate a protein femtosecond laser crosslinking efficiency of 10-12% using a colorimetric assay and 6% by Coomassie blue staining of SDS gels. Both assays are subject to large uncertainties due to protein-specific chemical reactions. Our efficiency estimates based on CsCl buoyant density were 1.4% for the 18S rRNA genes and 7.5% for telomeres, demonstrating the large influence of nucleoprotein structure on the results. Thus, the crosslinking efficiency of the femtosecond laser is comparable to that of other light sources. However, the time required to crosslink the same amount of proteins in nuclei is 15 min for the present femtosecond laser compared to up to 80 min with a UV lamp. Anticipated improvements in the laser will enable similar amounts of energy to be delivered in 8 fs 10 kHz pulses with 0.2-1 min exposures.
DNA integrity
Analysis of crosslinked nucleoprotein can be limited by DNA damage which can affect the ability to identify and study the DNA by enzymatic and electrophoretic means (11,37). Because the extent and nature of DNA damage is a complicated function of the light intensity, time of irradiation and buffer conditions it is impossible to precisely compare damage done under different conditions with different lasers. In one study using a conventional UV source achieving only 0.4% crosslinking, the resulting DNA was insensitive to MNase digestion and acid treatment was required to release crosslinked proteins from DNA (23). In a study with a picosecond laser single- and double-stranded breaks in the DNA were detectable (8). We found that 30 min of femtosecond laser irradiation did not cause noticeable double-strand breaks and did not affect digestion with ApaI; however, it did protect 75% of the 18S rDNA from digestion by EcoRI. The difference in sensitivity is most likely due to the presence of thymidine dimers at the EcoRI recognition site. The irradiated DNA was very sensitive to MNase. Thus, we found conditions under which femtosecond laser irradiation produced ~20 times greater crosslinking and much less damage than conventional UV irradiation.
Protein integrity
Protein integrity is crucial for analysis of crosslinked nucleoprotein. Analysis of nucleoproteins by gel shift assay, SDS-PAGE, mass spectrometry, as well as immunoprecipitation, immunolabeling and immunodetection rely on the physical integrity of the protein. Irradiation with a nanosecond laser was shown to degrade up to 25% of proteins in vitro (10); however, a picosecond laser did not degrade histones in vitro at three to four times higher doses (8). A more recent study of crosslinking nuclei using a nanosecond laser showed that 10% of proteins could be crosslinked in a single pulse; however, damage to the proteins or DNA was not assayed (14). In this paper, we assayed protein damage using western blot analysis and Coomassie blue staining of the Ku-70 subunit of Ku antigen and histones, respectively, and found almost no degradation of either protein even after doses used for extensive crosslinking of bulk proteins and histone H3. We have additionally shown that laser irradiation at 266 nm does not produce protein-protein crosslinks, in contrast to the results using conventional UV light, which can produce extensive protein-protein crosslinking (23), which affects analysis of the crosslinked nucleoproteins. Although histones were expected to be resistant to degradation and crosslinking due to low frequencies of aromatic residues, both Ku subunits were expected to be more sensitive due to much larger percentages of aromatic residues (27,28).
Crosslinking of specific proteins
Histones H3 and H1 were crosslinked by the femtosecond laser greater than 10 times more efficiently than the other histones, consistent with crosslinking results of nuclei using conventional UV lamps (23,36). As a test of the usefulness of the femtosecond laser for understanding the binding of a specific protein to a specific sequence of DNA, we showed that immunoprecipitation of histone H3 co-precipitated telomere DNA that could be analyzed by conventional filter hybridization. This result indicates that nucleosomes comprise the majority of the nucleoprotein structure of human telomeres. As a test of the ability of the femtosecond laser to study the binding of specific proteins to specific DNA conformations, we examined Ku-70 binding to intact and damaged DNA in nuclei. Although Ku is specifically bound to and crosslinked to the ends of DNA in vitro and is known to be involved in signaling cell response to DNA damage, there is no direct evidence that it is bound to intact DNA in nuclei. Our inability to crosslink Ku-70 in intact nuclei might be due to poor efficiency of crosslinking this protein with the femtosecond laser or to low binding of Ku to DNA in nuclei. In a positive control, we showed that Ku-70 could be crosslinked to nuclei after introducing artificial double-stranded breaks by restriction digestion, which suggests that its binding to the chromosomes could be stimulated by DNA damage or that the chemical nature of the Ku-70/DNA binding is different in broken and intact DNA.
ACKNOWLEDGEMENTS
The authors thank A. Conconi for human 18S rRNA plasmid p18S. This work was supported by the American Cancer Society, National Science Foundation, the Department of Energy and the Office of the Vice-President for Research, University of Michigan.
REFERENCES
*To whom correspondence should be addressed at: Biophysics Research Division, 4028A Chemistry, 930 North University, Ann Arbor, MI 48109-1055, USA. Tel: +1 734 647 1826; Fax: +1 734 764 3343; Email: langmore{at}umich.edu
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Barker, M. Weinfeld, J. Zheng, L. Li, and D. Murray Identification of Mammalian Proteins Cross-linked to DNA by Ionizing Radiation J. Biol. Chem., October 7, 2005; 280(40): 33826 - 33838. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||