Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 May 18.
Published in final edited form as: Chem Biol. 2007 May;14(5):553–563. doi: 10.1016/j.chembiol.2007.04.004

Effects of Photo-chemically Activated Alkylating Agents of the FR900482 Family on Chromatin

Vidya Subramanian 1, Pascal Ducept 2, Robert M Williams 2,3,5, Karolin Luger 1,4,5
PMCID: PMC3097141  NIHMSID: NIHMS288462  PMID: 17524986

SUMMARY

Bioreductive alkylating agents are an important class of clinical antitumor antibiotics that cross-link and mono-alkylate DNA. Here we use a synthetic photochemically activated derivative of FR400482 to investigate the molecular mechanism of this class of drugs in a biologically relevant context. We find that the organization of DNA into nucleosomes effectively protects it against drug-mediated cross-linking, while permitting mono-alkylation. This modification has the potential to form covalent cross-links between chromatin and nuclear proteins. Using in vitro approaches, we found that interstrand cross-linking of free DNA results in a significant decrease in basal and activated transcription. Finally, cross-linked plasmid DNA is inefficiently assembled into chromatin. Our studies suggest new pathways for the clinical effectiveness of this class of reagents.

INTRODUCTION

Many anticancer agents act by impeding rapid proliferation that is characteristic of cancer cells. While some impact cytoskeletal dynamics and cell-signaling pathways, others affect basic cellular processes such as replication and transcription. Compounds that fall into the latter class include interstrand DNA cross-linking / DNA alkylating agents. DNA cross-linking affects DNA-related processes specifically via the formation of covalent bonds between both DNA strands. This is a potent drug-mediated modification associated with severe cytotoxicity [1]. The detailed effects of drug-mediated DNA alkylation on cell proliferation are yet to be determined, since such lesions are likely to be removed and repaired via DNA base-excision repair mechanisms [2, 3].

Bioreductive alkylating agents include families of bifunctional compounds that are capable of forming monoalkylated DNA and interstrand DNA cross-links under highly reducing conditions. The specific targeting of these drugs to cancer cells has been rationalized by their ability to be activated under the hypoxic conditions that prevails in cancer cells as a result of their high replication rate [4]. A representative clinically significant drug is Mitomycin C (MMC, 1; Figure 1) [5]. MMC is the most thoroughly studied member of this class of compounds and has been in clinical use for over thirty years as an antitumor agent [6–8]. FR900482 (2, Figure 1), a natural metabolite obtained along with FR66979 (3, Figure 1A) from Streptomyces sandaensis No. 6897, is structurally and functionally related to MMC [9, 10] but was dropped from clinical development in Phase I studies due to the development of vascular leak syndrome in patients. More recently, a semi-synthetic derivative, FK317, has exhibited promising antitumor activity [10–15] and has advanced from Phase I to Phase II human clinical trials. FK317 (4, Figure 1A) shares structural similarities and properties with FR900482 (2, Figure 1A), but exhibited fewer side effects, in particular with respect to vascular leak syndrome, when compared to FR900482 [11, 16].

Figure 1. Structure of the various mitosene-based anticancerous agents and the basis for their sequence specificity.

Figure 1

Open in a new tab

(A) Structures of Mitomycin C, FR900482 and congeners. Compound 5 is a photo-activated Mitosene progenitor. (B) Geometrical basis for mitosene-based cross-linking specificity.

MMC can be activated via a one-electron or two-electron reduction pathway to form a highly reactive bis-electrophilic mitosene species that is capable of both mono-alkylating and cross-linking DNA. In contrast, FR900482 and FK317 are activated specifically by a two-electron reduction pathway to form a structurally and functionally related bis-electrophilic mitosene species (8; Figure 1) that is likewise capable of mono-alkylating (10) and cross-linking DNA (11). In all three instances, it has been demonstrated that mitosenes preferentially cross-link B-form DNA at 5′CpG3′ steps in the minor groove [17–19]. Mitosene formation appears to be the rate-limiting step in the formation of the final drug-mediated lesion [19]. The solution structure of mono-alkylated DNA indicates that the monoadduct represents an intermediate species in a reaction that finally results in the formation of an interstrand cross-linked product [20–22]. Recently, we reported a new generation of synthetic mitosene progenitors based on FR900482 that can be activated with alternative chemical signals not requiring reductive activation [23]. These agents can potentially obviate the slow generation of the mitosene, and also allow for controlled release of these highly reactive species. In particular, the totally synthetic NVOC-derivative related to FR900482 (5, Figure 1) is activated photochemically and exhibits biochemical reactivity similar to FK317 [23] including the characteristic 5′CpG3′ sequence specificity for interstrand cross-linking.

The striking preference of all mitosene-based drugs for cross-linking 5′CpG3′ steps is thought to be due to the geometrical arrangement of the two electrophilic sites on the mitosene which closely aligns with the distance between the opposing dG residues at a 5′CpG3′ step in B-form DNA (Figure 1B). This requirement has been shown to be less stringent for mono-alkylation [20, 24]. While the sequence specificity for both mono-alkylation and interstrand cross-linking was originally determined on free B-form DNA [25], the interactions of this class of drugs with DNA in the context of nucleosomes has not been studied in detail. This is a significant concern since ~ 80 % of all eukaryotic DNA is organized in nucleosomes. Nucleosomal DNA conformation deviates significantly from that of unassembled B-form DNA and as such, the identification of the covalent interactions of these drugs in an environment more closely resembling that of chromosomal DNA as packaged in the nucleus of living cells is particularly relevant. A previous study revealed a decrease in the levels of MMC-mediated interstrand cross-linking on nucleosomal DNA when compared to free DNA [26], however, MMC-mediated mono-alkylation on linear versus nucleosomal DNA could not be demonstrated conclusively. Furthermore, the effects of DNA cross-linking on transcription and chromatin assembly have not been investigated in detail.

Here we describe the effect of a photochemically activated NVOC-derivative (5; Figure 1) on two distinct free and nucleosomal DNA templates that are both 146 base pairs in length. Apart from base composition, the templates differ in the number and location of 5′CpG3′ steps (Figure 2). We investigated the ability of 5 to cross-link or alkylate free and nucleosomal DNA. We observed a dramatic decrease in cross-linking of nucleosomal DNA and a commensurate increase in mono-alkylation. Compound 5 significantly inhibits in vitro transcription and nucleo-some assembly. Our studies conclude that cross-linking of free DNA mediated by 5 may affect cellular proliferation by targeting two vital processes – chromatin assembly and transcription.

Figure 2. The two DNA fragments differ in the number and location of 5‘CpG3’ dinucleo-tide steps.

Figure 2

Open in a new tab

(A) Sequence of the two DNA fragments under investigation. The asymmetric 5S DNA template contains nine 5′CpG3′steps highlighted (pink), the palindromic alpha-satellite DNA contains a single 5′CpG3′step (green). (B) Representation of target site in a structural context. The left and middle panels represent the two halves of the 5S DNA (5SNCP (R) and (L) respectively), with the dyad indicated as 0. The numbers ± 1–7 represent the superhelical axis location (SHL) of nucleosomal DNA. Since the structure of the 5S NCP is not known, we show a projection of the target sites onto the structure of the α-sat NCP as a model. The right-most panel is a graphical representation of the NVOC target site on α-satellite nucleosomal DNA. The arrows indicate the location of the symmetry dyad.

RESULTS

Nucleosomal DNA is cross-linked inefficiently by compound 5 compared to free DNA

The preference of mitosene-based drugs for 5′CpG3′ dinucleotide steps results from the favorable geometric arrangement of the electrophilic sites of the drug and the target sites on linear B-form DNA (Figure 1). Nucleosomal DNA, although still classified as B-form, is distorted compared to unassembled DNA due to the high degree of bending upon superhelix formation [25]. To test the effect of this distortion on the efficiency of cross-linking, we compared two different 146 base pair DNA sequences in the context of highly positioned nucleosomes; a palindromic DNA fragment derived from human α-satellite DNA and an asymmetric fragment derived from the Xenopus laevis 5S rRNA gene. While the former has been used routinely to generate homogenous nucleosome species mainly for crystallization purposes (reviewed in, [27]) the latter has been used extensively for biochemical analyses. The two sequences differ significantly in the number and location of 5′CpG3′ steps (Figure 2).

Free DNA and assembled nucleosomes (5S and α-sat NCP ) that had been separated from any remaining free DNA by preparative gel electrophoresis were treated with 5 and were analyzed by native PAGE for (Figure 3A & B). Even at a 60-fold molar excess of 5, nucleosomes remained intact and exhibited no release of free DNA. DNA was isolated from nucleosomes, and the degree of cross-linking in free and nucleosomal DNA was analyzed by alkaline agarose electrophoresis (Figure 3C–F). The percentage of cross-links as a function of molar excess of drug is plotted in Figure 3G. As expected from the differences in the number of 5′CpG3′ sites on α-sat DNA compared to 5S DNA (1 versus 9 in each 146 bp DNA fragment, Figure 2), free 5S DNA is cross-linked much more efficiently than free α-sat DNA (Figure 3C versus 3D respectively). In both cases, the level of cross-linking is reduced significantly in nucleosomes (Figure 3E & 3F for 5S NCP and α-sat NCP respectively), indicating that the inherent distortions in nucleosomal DNA and the partial protection by histones inhibit the formation of interstrand cross-links.

Figure 3. Reduced cross-linking mediated by the NVOC-derivative 5 on nucleosomal DNA.

Figure 3

Open in a new tab

Panels (A) and (B) represent NCPs reconstituted with 5S DNA and α-satellite DNA re-spectively, treated with increasing molar ratios of NVOC. The samples were analyzed on a 5% native PAGE gel stained with ethidium bromide to determine any appearance of free DNA due to NVOC-induced NCP dissociation. Lane 1 in panel A and B indicate free DNA while lane 2 in both panels shows NCP controls without the drug. Lane 3 shows NCP treated with UV radiation without the drug, while lanes 4–8 show a 1:1, 1:10, 1:20, 1:40 and 1:60 ratios of NCP to NVOC, in both panels. Panels (C) and (D) show denaturing alkaline agarose gels of 5S and α-sat DNA treated with increasing ratios of NVOC. Lane 1 in both panels shows untreated DNA, while the DNA in lanes 2–8 was treated with 1:1, 1:5, 1:10, 1:20, 1:40, 1:60 and 1:80 molar ratios of NVOC. Panels (E) and (D) show denaturing alkaline agarose gels of 5S and α-sat nucleosomal DNA treated with increasing ratios of NVOC. Lane 1 in both panels represents untreated DNA, while lanes 2–8 represents a 1:1, 1:5, 1:10, 1:20, 1:40, 1:60 and 1:80 ratios of NCP to NVOC. The alkaline agarose gels were stained and visualized using Sybr Gold and the intensity of the cross-linked and uncross-linked bands were determined using the ImageQuant v 5.1 software. (G) Graphical representation of the efficiency of NVOC-mediated cross-linking on DNA/nucleosomal α-satellite and 5S DNA. The efficiency of cross-linking was determined from the ratio of the intensity of the cross-linked band to the sum of the intensities of the cross-linked and uncross-linked bands and values thus obtained were plotted as NVOC-mediated percentage cross-linking as a function of the fold excess of NVOC (over DNA/nucleosomal DNA).

Similar experiments were also performed with the FK317 compound (Figure 1; 4). The rate of FK317-mediated modification was exceedingly slow, requiring as long as 3 months after activation of the drug to obtain detectable levels of cross-linking on free 5S DNA. No detectable levels of cross-links were observed on either 5S NCP, or unassembled and assembled α-sat DNA (data not shown). Since the integrity of nucleosome samples for such extended periods of time was uncertain, further studies with this compound were deemed impractical.

Alkaline agarose gel electrophoresis allows us to distinguish between cross-linked and uncross-linked species, but does not possess the resolution to visualize mono-alkylation. In contrast, denaturing PAGE allows for good resolution of the unmodified single-stranded DNA, mo-noalkylated single-stranded DNA, and cross-linked DNA strands. Figure 4A shows the analysis for free and nucleosomal α-satellite DNA. A cross-linked DNA control, purified from a denaturing gel, was loaded as a control (Figure 4A, lane 2). The gel was scanned, and the percent of cross-linked DNA obtained by this method was compared to the numbers plotted in Fig. 3G. The numbers obtained for a 40-fold molar excess of compound 5 (1.2 and 14 % for nucleosomal and free α-satellite DNA, respectively, compared well to numbers plotted in Fig. 3G (4.5 and 11 %, respectively).

Figure 4. Preferential NVOC-mediated mono-alkylation of nucleosomal DNA compared to linear DNA.

Figure 4

Open in a new tab

Urea-8% polyacrylamide gel (8%DPAGE) analysis was conducted on [γ32-P] ATP labeled nucleosomal and free DNA. (A) DPAGE analysis on NVOC-treated α-sat DNA/nucleosomal DNA. Lane 1: α-sat DNA, lane 2: Cross-linked α-sat DNA control. The two different orientation isomers are illustrated at the side of the gel. Lane 3 and 4: α-sat-DNA treated with a 20 and 40 molar excess respectively of NVOC to DNA. Lane 5: Nucleosomal α-sat-DNA treated with proteinase K; lane 6 and 7: Nucleosomal DNA treated with 20 and 40 molar excess of NVOC respectively. (B) DPAGE analysis of free 5S DNA. Lane 1: 5S DNA. The two bands correspond to the difference in charge and hence migration of the two complimentary strands of 5S DNA. Lanes 2 and 3: 5S DNA treated with 20 and 40 molar excess of NVOC respectively. (C) DPAGE analysis of 5S nucleosomal DNA. Lane 1: Nucleosomal 5S DNA treated with proteinase K, lane 2 and 3: Nucleosomal 5S DNA treated with 20 and 40 molar excess of NVOC respectively. The concentration of DNA used in these reactions was 6.25 μM, and that of the nucleosomes was 12.5 μM.

Even though the α-sat DNA template has only one 5′CpG3′ site (Figure 2), two bands are visible in Fig. 4A upon incubation with compound 5. Although we cannot exclude the possibility that these represent adducts formed at other sites, we surmised that these may represent positional cross-linked isomers. These isomers may be formed due to the difference in orientation of the reactive mitosene residue in the 5′CpG3′ step, which is largely dependent on which of the reactive centers on the mitosene moiety reacts with the exocyclic N2 amine on guanosine [18]. The nucleosomal template has a much higher abundance of the monoadduct than observed for free DNA, demonstrating that nucleosomal DNA is preferentially monoalkylated. Since only a single mono-alkylated species is apparent, we suggest that the sequence preference for 5′CpG3′ may be maintained in the context of the nucleosome even for mono-alkylation by 5.

The same experiment was repeated with 5S DNA which contains nine (as opposed to one) 5′CpG3′ sites. No significant mono-alkylated species were observed on free 5S DNA at a 40-fold molar excess of drug (Figure 4B, lane 3). Instead, we observed a family of multi-alkylated products whose migration rates are slower than the mono-adduct species but faster than the more prominent cross-linked species. Nucleosomal 5S DNA treated with similar amounts of 5 is predominantly multi-alkylated, whereas cross-links are barely visible (Figure 4C, lanes 2 and 3). We conclude that the NVOC-derivative 5 is bifunctional and can cause both interstrand DNA cross-linking as well as alkylation. The outcome apparently depends entirely on the conformational state of the DNA.

The presence of monoalkylated DNA raises the intriguing possibility that histones may become cross-linked to DNA upon treatment with 5. However, preliminary assays in which dissociation of nucleosomes in response to salt was measured showed no difference between untreated and 5-treated nucleosomes, indicating that no protein – DNA cross-links were formed. This was not entirely unexpected since the distances between lysines and DNA bases in the nucleosome structure are too large for effective cross-links. However, more careful analysis is required to exclude the possibility of drug-mediated histone – DNA induced cross-links.

Compound 5-mediated cross-linking of unassembled DNA represses in vitro transcription

Histones are transiently removed from DNA during transcription, often with the help of ATP-dependent chromatin remodeling factors and perhaps also histone chaperones [28, 29]. This generates a window of opportunity for efficient cross-linking of nucleosome-free DNA by mito-sene-based drugs. To investigate the effect of cross-linking on basal and activated transcription, we employed a previously established in vitro system [30, 31] that utilizes a well-characterized plasmid template (p4TxRE/G-less). The plasmid contains 178 5′CpG3′ dinucleotide steps that are distributed quite uniformly along its entire length, with the exception of the 380 bp long G-less cassette. We made use of an in vitro assembly system consisting of purified assembly factors, and that recapitulates many aspects of in vivo chromatin assembly [30].

Transcript levels were monitored under basal and activated conditions in the absence and presence of the co-activators Tax and CREB [32]. To account for variations in the recovery of transcripts, each reaction was normalized against a recovery standard (see experimental procedures). Under both conditions the cross-linked template showed a significant reduction in transcription levels when compared to a control that had not been activated by light. Figure 5B is a quantitative representation of the data shown in Figure 5A. Basal transcript levels from cross-linked DNA template decreased 5 fold compared to untreated template. Activated transcript levels dropped 25 fold when compared to the transcript levels of the inactive-NVOC reaction.

Figure 5. Decreased levels of in vitro transcription from cross-linked DNA templates.

Figure 5

Open in a new tab

(A) In vitro transcription assay indicating a decrease in transcript levels only in the presence of NVOC-mediated cross-linked product. * indicates activated NVOC. Each transcription reaction contains 150ng of the p4TxRE template and 60μg of the nuclear extract- CEM from HTLV-negative human T cells. 100ng of the purified recombinant Tax and 100ng of purified recombinant CREB were added to the reaction where indicated. (B) Quantitation of the transcript levels was done using ImageQuant analysis and the relative transcript levels were normalized to the highest Tax & CREB activation. The shaded bars represent those reactions activated by Tax and CREB. The fold increase in transcription levels due to this activation (compared to the corresponding the basal transcription levels -the empty bars) is indicated above each of these bars.

Compound 5-mediated cross-linking of plasmid DNA reduces the efficiency of chromatin assembly

Mitosene-based drugs are particularly effective in rapidly dividing cancer cells where DNA synthesis and concomitant chromatin assembly are frequent events. We therefore wanted to quantify the effect of DNA cross-linking and alkylation on the efficiency of in vitro chromatin assembly. The same p4TxRE/G-less templates (untreated and treated with compound 5) that were used for the in vitro transcription assay above were employed for this assay.

The plasmid was treated with a 40-fold molar excess of compound 5 prior to assembly to ensure complete saturation of all available sites. Cross-linking was verified by linearizing the plasmid, followed by analysis by alkaline gel electrophoresis (data not shown). Cross-linked plasmid was purified from unreacted DNA and free 5 by gel elution under native conditions, exploiting the fact that cross-linked plasmid exhibits a slower electrophoretic mobility than untreated plasmid (data not shown). Untreated and 5-treated supercoiled plasmids were assembled into chromatin using the purified chromatin remodeling factor ACF1 and the histone chaperone dNAP-1 [33]. The assemblies were assayed by micrococcal nuclease digestion and quantitative agarose gel electrophoresis-multigel analysis.

Micrococcal nuclease (MNase) digestion confirmed that the nucleosomes reconstituted onto untreated DNA template are spaced on average 165–170 bp apart (Figure 6A, lanes 2–5). Interestingly, chromatin assembled on 5-treated plasmid DNA was resistant towards digestion with MNase, whereas no effect was found if the plasmid was treated with light in the absence of 5. A DNA band corresponding to mononucleosomes was visible only at the highest MNase concentrations (Figure 6A lane 15). This could be due to the fact that cross-linked DNA is a poor substrate for most nucleases [33]. This was addressed in a control experiment in which the effect of 5 treatment on the sensitivity of unassembled plasmid DNA towards digestion with MNase was assayed (Figure 6A, right panel). Whereas untreated DNA is digested rapidly and without a trace, cross-linked plasmid DNA exhibited increased resistance towards digestion.

Figure 6. Cross-linking of DNA affects the efficiency of chromatin assembly.

Figure 6

Open in a new tab

(A) Qualitative analysis of the assembled template by micrococcal nuclease digestion. The various templates used in these analyses are indicated above the lanes. The left panel represents the micrococcocal nuclease digestion on chromatin templates at different time points (1, 2, 4 and 8 minutes, respectively). The right panel represents the micrococcal nuclease digestion analyses on untreated and NVOC-treated p4TxRE. Lanes 2, 8 and 14 represent untreated p4TxRE, p4TxRE treated with 20 molar excess NVOC and p4TxRE treated with 40 molar excess NVOC respectively. Lanes 3–6, 9–12 and 15–18 represent the digestion times of 0.5, 1, 2, and 4 minutes. (B) The right and left panels represent the QAGE analyses performed on untreated p4TxRE/assembled p4TxRE and NVOC-treated p4TxRE/ assembly with cross-linked p4TxRE respectively. The numbers over each lane indicates the agarose concentrations. (C) The QAGE results are represented quantitatively in a Ferguson Plot. The y-axis has been set to logarithmic scale. Intact T3 phage was used as a migration control as well as in determination of the unknown μo value for unassembled and assembled DNA. The plot is first used to determine the gel-free mobility (μo′) [35]. The values obtained in these plots are listed in Table 1.

The results shown in Figure 6A raise the possibility that chromatin assembly is inefficient on a cross-linked DNA template. To further investigate this, we used Quantitative Agarose Gel Electrophoresis - Multigel Analysis (QAGE), a method which has been used successfully in the past for the analysis of chromatin to determine parameters such as average surface charge density and particle deformability [34, 35]. More precisely, this method allows us to quantitate the average surface charge density of unassembled DNA and assembled DNA, thus effectively ‘counting’ the number of nucleosomes on any given DNA fragment. The same analysis was also performed with the free DNA sample. A typical example is shown in Figure 6B. Ferguson plots (semi logarithmic plots of migration distance μ versus agarose percentage; Figure 6C) allow for the determination of μo and μo′. The μo′ values for all the species in Figure 6 are listed in Table 1. Cross-linked DNA species and the untreated DNA species do not vary significantly in their μo′ values. This demonstrates that there are no differences in surface charge densities in the free DNA upon treatment with compound 5. This not only validates the technique but also establishes the fact that any change in nucleosome assembly on the 5-treated and untreated plasmid is not associated with charge.

Table 1.

Quantitative analysis of the data presented in Figure 6C. Number of nucleosomes were calculated as described in materials and methods. Experiments were repeated three times.

Samples μo cm2/V/sec (×10−4) Number of Nucleosomes
p4TxRE/G-less −2.27635
Assembled p4TxRE/G-less −1.6764 18.5 + 2
NVOC-treated p4TxRE/G-less −2.47647
Assembled NVOC-treated p4TxRE/G-less −1.9661 14.5 + 1
Open in a new tab

The μo′ values listed in Table 1 were used to calculate the number of nucleosomes assembled on the two templates [34, 35]. This analysis shows that an average of 18.5 nucleosomes is assembled on the untreated p4TxRE/G-less plasmid. On 3245 base pairs, this converts to a repeat length of 175 bp, a number that corresponds well to the repeat length established by MNase (Figure 6A, left panel). On cross-linked DNA, an average of 14.5 nucleosomes was assembled. This 22% reduction in the number of nucleosomes assembled on the 5-treated DNA template when compared to an untreated DNA template is reproducible and indicates that cross-linking indeed has an inhibitory effect on chromatin assembly under physiological conditions.

DISCUSSION

Bioreductive alkylating agents are an important class of anticancer drugs that act through their ability to covalently cross-link two strands of DNA, thus impeding DNA replication of rapidly dividing cancer cells. The synthetic NVOC derivative (5) is a photo-activated compound that generates in high chemical yield a doubly-electrophilic mitosene species that is structurally and mechanistically homologous to the mitosenes generated by bioreductive activation of FK317, MMC and congeners. Here we showed that interstrand cross-linking of nucleosomal DNA by compound 5 is suppressed while at the same time, the extent of mono-alkylation markedly increases. Our finding that cross-linked DNA is inefficiently transcribed and assembled into chromatin suggests additional mechanisms by which mitosene-based drugs display their cytotoxic effect.

It has been shown previously that the efficiency of MMC-induced cross-linking on nucleo-somes was reduced compared to free DNA [26]. In the present study we have significantly extended this finding by demonstrating the preferential formation of mono-alkylated over cross-linked products at the nucleosomal dyad. Nucleosomal DNA is distinct from unassembled DNA in its helical parameters [25]. The exact geometry of each given CpG step depends on its rotational and translational position with respect to the histone octamer. By using nucleosomes assembled on α-satellite DNA, we have the opportunity to directly measure the distance between the reactive centers of the only CpG step that is located near the nucleosomal dyad [36]. The distance measured between the two opposing exocyclic amine residues in the minor groove at this particular location in the nucleosome is 3.95 Å, whereas the distance between the same reactive centers on unassembled B-form DNA is 3.4Å. This 0.5 Å difference is significant in light of the estimated distances between reactive centers in other dinucleotides that are NOT cross-linked on B-form DNA. For example, the distance between the targets sites in a GpG step is ~ 4.2Å and in a GpC step it is around 4.6 Å.

Apparently, nucleosomal DNA is too structurally constrained to allow for mitosene-mediated interstrand cross-linking. Instead, monoalkyation at presumably the same site is observed. What then is the biological significance of mitosene-mediated alkylation of nucleosomal DNA? Alkylation has been considered a less cytotoxic modification compared to interstrand DNA cross-linking as it is easily corrected by the nucleotide base excision repair machinery [2, 3]. However, it remains to be investigated how efficiently this type of damage is repaired in the context of chromatin. Mono-alkylation has also been proposed as a suitable intermediate for the formation of cross-links. Earlier studies have shown that congener FR66979 can cause protein – DNA cross-links via drug-mediated mono-alkylation of DNA and the oncoprotein HMG A1 in vitro [37]. Similar cross-links were observed in vivo upon treatment with FR900482 and FK317 [38]. This opens the interesting hypothesis that mono-alkylation of nucleosomal DNA especially at the nucleosomal dyad could result in the cross-linking of chromatin-associated protein to nucleo-somal DNA, with far-reaching implications on chromatin dynamics and DNA accessibility.

About 80% of all eukaryotic DNA is organized in nucleosomes, leaving only parts of the linker DNA and the occasional highly active promoter histone-free. The processes of replication, transcription and DNA repair require transient exposure of DNA [39, 40]. This may represent a window of opportunity for mitosene-derived drugs to cross-link the two strands of DNA. This modification is likely to be refractory to all processes that require melting of the DNA double helix. Our studies show that basal as well as activated transcription from cross-linked DNA templates is severely impaired. This defect stems from the inability of the polymerase to initiate transcription, since the G-less cassette is free of cross-links.

The potential effect of DNA cross-linking on chromatin assembly has largely been ignored in investigations of the mechanisms of mitosene-based cytotoxicity. Using a defined in vitro assembly system, we found a 22% decrease in nucleosome occupancy on a cross-linked template. The majority of chromatin is assembled during S-phase, where the histones are deposited on the newly replicated DNA by Chromatin Assembly Factor-1 (CAF-1) [41]. Suppression of the p60 subunit of CAF-1 by RNAi results in a slightly reduced efficiency of chromatin assembly [42]. These cells exhibited cell cycle arrest and programmed cell death. Thus, it appears that a relatively small decrease in assembly could be a sufficient signal to induce cell-cycle arrest in the S-phase especially in rapidly dividing cells. Since post-transcriptional chromatin assembly is also a key process for quiescent cells, it is possible that mitosenes may be toxic for non-dividing cells through these mechanisms.

These studies reveal new and hitherto unresolved possible mechanisms for how mitosene-based anti-tumor agents may exert their cytotoxic effect in rapidly proliferating cells. We are continuing to study the molecular interactions of such agents in the context of dynamic processes of chromatin assembly and there remains the possibility for the formation of additional covalencies between the DNA-bound drug and histones as well as other proteins closely associated with the complex processes of replication and transcription.

SIGNIFICANCE

This study has been able to emphasize the differential activity of the photo-triggered NVOC derivative on conformationally distinct unassembled and assembled DNA. The bifunctional activity of compound 5 seems to be dependent on the conformation of the DNA template. We believe that this finding will help provide insight for the design of a specific alkylating agent based on the contours of nucleosomal DNA and hence its effective use as an anti-cancerous agent.

This study has also shown conclusively the effect of compound 5-mediated interstrand cross-linking on transcription and chromatin assembly. We believe that this is the first time the quantitative effect of cross-linking on both these functional processes has been demonstrated. This could be a major step in understanding one of the many levels at which these DNA cross-linking agents act to decrease and disrupt cellular replication and hence behave as effective anti-cancerous drugs.

EXPERIMENTAL PROCEDURES

Drug stock solution

Compound 5 was prepared according to published procedures [23]. A stock solution of 10 mg/ml (15.8 mM) of 5 was prepared in 100% DMSO and placed in a light-safe tube to prevent photo-decay. Appropriate dilutions were made in 100% DMSO.

Nucleosome preparation

The 146bp 5S and α-satellite DNA templates were prepared as published [46] and the corresponding nucleosomes were reconstituted using recombinant full length Xenopus histone octamer via salt dialysis as described earlier [27].

DNA cross-linking and alkaline agarose electrophoresis

The final volume of the cross-linking reaction was 20 μl. The final concentration of free DNA (both 5S and α-satellite) in these reactions was 6.25 μM, and 12.5 μM for nucleosomal DNA. The samples were irradiated in clear Eppendorf tubes at 25 °C for an hour with an ultraviolet lamp (Rayonet) containing 350 nm bulbs [23, 43], and incubated in the dark at 4 °C. Excess drug was removed from free DNA samples by ethanol precipitation. In order to remove excess drug and histones from compound 5-treated nucleosomal samples, the 20 μl reaction mixture was treated with 5.2 μl of 4M NaCl, 4μl of 20 mg/ml glycogen and 60 μl of 100% ethanol. The nucleosomes were then pelleted by spinning the samples at 13.2 K rpm for 30 minutes. Histones were removed by Proteinase K treatment in a volume of 100 μl containing a final concentration of 0.1 mg/ml of Proteinase K (Sigma-Aldrich). The samples were incubated at 37°C for at least an hour followed by phenol chloroform extraction and ethanol precipitation. All samples (both nucleosomal and free DNA) were analyzed by 1.4% alkaline agarose electrophoresis [44]. The gels were run at 75 volts for 50 to 55 minutes and stained with SyBr Gold (Invitrogen, Molecular Probes™). The appropriate bands were quantified using the Image Quant Analysis (version 5.1) software.

Analysis of mono-alkylation

Free and nucleosomal DNA samples were subjected to treatment with compound 5, and DNA was isolated from nucleosomes as described above. 10μg of extracted DNA were radiolabeled at the 5′ end with [γ32-P] ATP using standard protocols in a final volume of 50 μl and in the presence of T4 polynucleotide kinase (NEB). The radiolabeled DNA was then subjected to phenol-chloroform extraction and ethanol precipitation to remove enzyme and excess radiolabel in the reaction mix. The resulting samples were analyzed by 8% PAGE in the presence of urea [43]. The gel was first pre-run at 30 W for about 30 minutes (until the temperature of the gel was monitored to be around 50 °C) and then run at 30 W for 40 minutes in 1XTBE buffer. The gel was dried and exposed to a Phosphorimager screen for 10 to 15 hours. The screen was then scanned and visualized using the Image Quant Analysis (version 5.1) software.

In vitro chromatin assembly

The p4TxRE/G-less plasmid template, native Drosophila core histones, recombinant ACF and dNAP1 were purified as described [30]. The in vitro assemblies were performed as described in [30]. The standard chromatin assembly reaction was 98 μl and contained 2.1 μg of plasmid DNA (0.994 pmoles of 3.2 kp plasmid DNA), 1.505 μg of purified native Drosophila core histones (0.11 nmoles of each histone), 12.012 μg of purified recombinant dNAP1 (0.2145 nmoles of dNAP1 monomer) and 0.36 μg of purified recombinant Drosophila ACF (1.285 pmoles of ACF peptide), in 10 mM HEPES pH 7.6, 50 mM KCl, 5 mM MgCl2, 5% glycerol, 1% PEG, 3 mM ATP, 0.01% NP40, an ATP regenerating system (30 mM phospho-creatinine and 1 μg/ml creatinine phosphokinase) and 140 ng BSA. The reaction was performed at 27 °C for 4–12 hours. Micrococcal nuclease digestion was performed as described [45].

Quantitative Agarose Gel Electrophoresis (QAGE)-Multigel Analysis

To determine the number of assembled nucleosomes, the average surface charge density of unassembled and assembled DNA template had to be quantitated [34]. The mobility of the species (μ) was calculated as a function of the agarose concentration by electrophoresis performed on nine gels each of different concentrations of agarose varying from 0.2–1.0%, embedded in a 1.5% agarose frame and run at 1 V/cm. The values of μo′ (gel free mobility) and μo (μo′ corrected for the electro osmotic effects of the buffer) were estimated as described [35]. In this study the 18-lane template was used to run unassembled DNA with its corresponding assembled species in the same gel frame. Both the frame and running gel were cast in TAE buffer (40 mM Tris acetate, 1 mM EDTA pH 7.8) or E Buffer (40 mM Tris HCl, 0.25 mM EDTA pH 7.8) as indicated and the corresponding EEO (electroendoosmosis) values were used to calculate μo. The assembly reaction containing 2.1 μg of assembled DNA along with 1 μg of intact T3 phage (migration standard) was loaded such that 0.23 μg of assembled DNA was present in each well. The same amount of supercoiled unassembled DNA was loaded along with the T3 control. The gels were run at room temperature (48 volts for 6 hours) with running buffer circulating for the duration of the experiment. With the μo values thus obtained for unassembled and assembled DNA templates, the number of assembled nucleosomes were calculated as reported [34, 35].

The p4TxRE/G-less plasmid was treated with NVOC in a manner similar to that of unassembled DNA as mentioned previously in the same section.

In vitro Transcription Assay

The in vitro transcription assay was carried out in a final reaction volume of 30 μl containing 1 mM acetyl CoA, 150 ng of DNA (treated or untreated as mentioned) and 70 μg of CEM (HTLV-negative human T-cells). For activated transcription reactions the same conditions were used along with 100 ng of Tax and 100 ng of CREB. The reactions were incubated at 30°C for an hour. The samples were then incubated with 250 μM ATP, 250μM CTP, 12 μM UTP and 0.8 μM [α32P] UTP (3,000 Ci/mmol) for 30 min at 30 °C. This allows for transcription through the 380 nt G-less cassette in the p4TxRE/G-less plasmid [31]. The read through transcripts were cleaved after treating these reactions with RNase T1 (100 units/reaction) at 37 °C for 30 minutes. RNase T1 is an endoribonuclease that specifically degrades single-stranded RNA at the G residues. This enzyme is therefore extremely useful for detection of RNA transcript levels from DNA templates containing the G-less cassette. The reaction was terminated by addition of 123 mM NaCl, 0.5% SDS and 2.5 mM EDTA, and the samples were subjected to Proteinase K treatment at 50 °C for 30 minutes. The transcripts were then extracted via precipitation with a final concentration of 0.48 M ammonium acetate and 0.3 mg/ml of carrier tRNA. In this step an appropriate amount of radiolabeled DNA fragment was added as a recovery standard. The precipitated samples were analyzed on a urea-6.5% polya-crylamide gel. The gel was subsequently dried and visualized by Image Quant Analysis (version 5.1). The software was used to quantitate the transcript yields under the various conditions indicated. The transcript levels in each lane were normalized against the recovery standard (internal control). The highest signal (due to activation with Tax and CREB) was assigned the number 1 and all the other normalized transcript levels were then represented as fractions of the highest value.

Acknowledgments

We thank Drs. Konesky, Lu, Nyborg and Stargell for guidance and reagents, and P. N. Dyer for help with histone and DNA preparation. We thank Fujisawa Pharmaceutical Co. (now Astellas) for the generous gift of FK317. This work was financially supported by NIH RO1 GM061909 to KL, and by RO1 CA51875 to RMW. KL is an Investigator with the Howard Hughes Medical Institute.

References

  • 1.Zwelling LA, Anderson T, Kohn KW. DNA-protein and DNA inter-strand cross-linking by cis- and trans-platinum(II) diamminedichloride in L1210 mouse leukemia cells and relation to cytotoxicity. Cancer Res. 1979;39:365–369. [PubMed] [Google Scholar]
  • 2.Lawley PD, Lethbridge JH, Edwards PA, Shooter KV. Inactivation of bacteriophage T7 by mono- and difunctional sulphur mustards in relation to cross-linking and depurination of bacteriophage DNA. J Mol Biol. 1969;39:181–198. doi: 10.1016/0022-2836(69)90341-6. [DOI] [PubMed] [Google Scholar]
  • 3.Sancar A. DNA excision repair. Annu Rev Biochem. 1996;65:43–81. doi: 10.1146/annurev.bi.65.070196.000355. [DOI] [PubMed] [Google Scholar]
  • 4.Sartorelli AC. The role of mitomycin antibiotics in the chemotherapy of solid tumors. Biochem Pharmacol. 1986;35:67–69. doi: 10.1016/0006-2952(86)90559-9. [DOI] [PubMed] [Google Scholar]
  • 5.Iyer VN, Szybalski W. A Molecular Mechanism of Mitomycin Action: Linking of Complementary DNA Strands. Proc Natl Acad Sci U S A. 1963;50:355–362. doi: 10.1073/pnas.50.2.355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Tomasz M. Mitomycin C: small, fast and deadly (but very selective) Chem Biol. 1995;2:575–579. doi: 10.1016/1074-5521(95)90120-5. [DOI] [PubMed] [Google Scholar]
  • 7.Tomasz M, Lipman R, Chowdary D, Pawlak J, Verdine GL, Nakanishi K. Isolation and structure of a covalent cross-link adduct between mitomycin C and DNA. Science. 1987;235:1204–1208. doi: 10.1126/science.3103215. [DOI] [PubMed] [Google Scholar]
  • 8.Pratt WBR, Ensinminger RW, Maybaum WD. The Anti-Cancerous Drugs. 2. New York: Oxford University Press; 1994. [Google Scholar]
  • 9.Uchida I, Takasae S, Kayakiri H, Kiyoto S, Hashimoto M. Structure of FR900482, a novel antitumor antibiotic from Streptomyces. J Am Chem Soc. 1987;109:4108–4109. [Google Scholar]
  • 10.Inami M, Kawamura I, Tsujimoto S, Nishigaki F, Matsumoto S, Naoe Y, Sa-sakawa Y, Matsuo M, Manda T, Goto T. Effects of FK317, a novel anti-cancer agent, on survival of mice bearing B16BL6 melanoma and Lewis lung carcinoma. Cancer Lett. 2002;181:39–45. doi: 10.1016/s0304-3835(02)00044-7. [DOI] [PubMed] [Google Scholar]
  • 11.Naoe Y, Inami M, Matasumoto S, Nishigaki F, Tsujimoto S, Kawamura I, Miyayasu K, Manda T, Shimomura K. Cancer Chemother Pharmacol. 1998;42:31–36. doi: 10.1007/s002800050781. [DOI] [PubMed] [Google Scholar]
  • 12.Naoe Y, Inami M, Kawamura I, Nishigaki F, Tsujimoto S, Matasumoto S, Manda T, Shimomura K. Jpn J Cancer Res. 1998;89:666–672. doi: 10.1111/j.1349-7006.1998.tb03269.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Naoe Y, Inami M, Takagaki S, Matasumoto S, Kawamura I, Nishigaki F, Tsujimoto S, Manda T, Shimomura K. Jpn J Cancer Res. 1998;89:1047–1054. doi: 10.1111/j.1349-7006.1998.tb00495.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Naoe Y, Inami M, Matasumoto S, Takagaki S, Fujiwara T, yamazaki S, Ka-wamura I, Nishigaki F, Tsujimoto S, Manda T, Shimomura K. Jpn J Cancer Res. 1998;89:1306–1317. doi: 10.1111/j.1349-7006.1998.tb00528.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Naoe Y, Kawamura I, Inami M, Matasumoto S, Nishigaki F, Tsujimoto S, Manda T, Shimomura K. Jpn J Cancer Res. 1998;89:1318–1325. doi: 10.1111/j.1349-7006.1998.tb00529.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Beckerbauer L, Tepe JJ, Eastman AR, Mixter FP, Williams RM, Reeves R. Differential Effects of FR900482 and FK317 on Apoptosis, IL-2 Gene Expression, and Induction of Vascular Leak Syndrome. Chemistry and Biology. 2002;9:427–441. doi: 10.1016/s1074-5521(02)00122-9. [DOI] [PubMed] [Google Scholar]
  • 17.Tomasz M, Lipman R, Chowdary D, Pawlak J, Verdine GL, Nakanishi K. Isolation and structure of a covalently cross-link adduct between mitomycin C and DNA. Science. 1987;235:1204–1208. doi: 10.1126/science.3103215. [DOI] [PubMed] [Google Scholar]
  • 18.Williams RM, Rajski RM, Rollins BS. FR900482, a close cousin of mitomycin C that exploits mitosene-based DNA cross-linking. Chemistry and Biology. 1997;4:127–137. doi: 10.1016/s1074-5521(97)90256-8. [DOI] [PubMed] [Google Scholar]
  • 19.Millard JTW, MF, Raucher S, Hopkins PB. Determination of the DNA Cross-linking Sequence Specificity of Reductively Activated Mitomycin C at Single-Nucleotide Resolution-Deoxyguanosine Residues at CpG are Cross-linked Preferentially. J Am Chem Soc. 1990;112:3637–3641. [Google Scholar]
  • 20.Sastry M, Fiala R, Lipman R, Tomasz M, Patel DJ. Solution Structure of the Monoalkylated mitomycin C-DNA complex. J Mol Biol. 1995;247:338–359. doi: 10.1006/jmbi.1994.0143. [DOI] [PubMed] [Google Scholar]
  • 21.Li VKH. Studies on the Bonding Specificity for Mitomycin C-DNA Monoal-kylation Processes. J Am Chem Soc. 1991;113:275–283. [Google Scholar]
  • 22.Kohn HLV, Shiltz P, Tang M. On the Origins of the DNA Sequence Selectivity of Mitomycin Monoalkylation Transformations. J Am Chem Soc. 1992;114:9218–9220. [Google Scholar]
  • 23.Judd TC, Williams RM. Synthesis and DNA cross-linking of a photo-triggered FR900482 mitosene progenitor. Org Lett. 2002;4:3711–3714. doi: 10.1021/ol0266774. [DOI] [PubMed] [Google Scholar]
  • 24.Kumar S, Lipman R, Tomasz M. Recognition of specific DNA sequences By Mitomycin C for Alkylation. Biochemistry. 1992;31:1399–1407. doi: 10.1021/bi00120a016. [DOI] [PubMed] [Google Scholar]
  • 25.Richmond TJ, Davey CA. The structure of DNA in the nucleosome core. Nature. 2003;423:145–150. doi: 10.1038/nature01595. [DOI] [PubMed] [Google Scholar]
  • 26.Millard JT, Spencer RJ, Hopkins PB. Effect of nucleosome structure on DNA interstrand cross-linking reactions. Biochemistry. 1998;37:5211–5219. doi: 10.1021/bi972862r. [DOI] [PubMed] [Google Scholar]
  • 27.Muthurajan UM, Park YJ, Edayathumangalam RS, Suto RK, Chakravarthy S, Dyer PN, Luger K. Structure and dynamics of nucleosomal DNA. Biopolymers. 2003;68:547–556. doi: 10.1002/bip.10317. [DOI] [PubMed] [Google Scholar]
  • 28.Krude T. Chromatin assembly during DNA replication in somatic cells. Eur J Biochem. 1999;263:1–5. doi: 10.1046/j.1432-1327.1999.00508.x. [DOI] [PubMed] [Google Scholar]
  • 29.Tyler JK. Chromatin assembly Cooperation between histone chaperones and ATP-dependent nucleosome remodeling machines. Eur J Biochem. 2002;269:2268–2274. doi: 10.1046/j.1432-1033.2002.02890.x. [DOI] [PubMed] [Google Scholar]
  • 30.Fyodorov DV, Kadonaga JT. Chromatin assembly in vitro with purified recombinant ACF and NAP-1. Methods Enzymol. 2003;371:499–515. doi: 10.1016/S0076-6879(03)71037-4. [DOI] [PubMed] [Google Scholar]
  • 31.Anderson MG, Dynan WS. Quantitative studies of the effect of HTLV-I Tax protein on CREB protein--DNA binding. Nucleic Acids Res. 1994;22:3194–3201. doi: 10.1093/nar/22.15.3194. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Kashanchi F, Duvall JF, Kwok RP, Lundblad JR, Goodman RH, Brady JN. The coactivator CBP stimulates human T-cell lymphotrophic virus type I Tax transactivation in vitro. J Biol Chem. 1998;273:34646–34652. doi: 10.1074/jbc.273.51.34646. [DOI] [PubMed] [Google Scholar]
  • 33.Pan SS, Iracki T, Bachur NR. DNA alkylation by enzyme-activated mitomycin C. Mol Pharmacol. 1986;29:622–628. [PubMed] [Google Scholar]
  • 34.Fletcher TM, Serwer P, Hansen JC. Quantitative analysis of macromolecular conformational changes using agarose gel electrophoresis: application to chromatin folding. Biochemistry. 1994;33:10859–10863. doi: 10.1021/bi00202a002. [DOI] [PubMed] [Google Scholar]
  • 35.Fletcher TM, Krishnan U, Serwer P, Hansen JC. Quantitative aga-rose gel electrophoresis of chromatin: nucleosome-dependent changes in charge, sharp, and deformability at low ionic strength. Biochemistry. 1994;33:2226–2233. doi: 10.1021/bi00174a032. [DOI] [PubMed] [Google Scholar]
  • 36.Luger K, Mader AW, Richmond RK, Sargent DF, Richmond TJ. Crystal structure of the nucleosome core particle at 2.8 A resolution. Nature. 1997;389:251–260. doi: 10.1038/38444. [DOI] [PubMed] [Google Scholar]
  • 37.Rajski SR, Rollin SB, Williams RM. FR66979 Covalently cross-links the Binding Domain of the High-Mobility Group I/Y proteins to DNA. J Am Chem Soc. 1998;120:2192. [Google Scholar]
  • 38.Beckerbauer L, Tepe JJ, Cullison J, Reeves R, Williams RM. FR900482 class of anti-tumor drugs cross-links oncoprotein HMG I/Y to DNA in vivo. Chem Biol. 2000;7:805–812. doi: 10.1016/s1074-5521(00)00028-4. [DOI] [PubMed] [Google Scholar]
  • 39.Krude T. Chromatin Nucleosome assembly during DNA replication. Curr Biol. 1995;5:1232–1234. doi: 10.1016/s0960-9822(95)00245-4. [DOI] [PubMed] [Google Scholar]
  • 40.Narlikar GJ, Fan HY, Kingston RE. Cooperation between complexes that regulate chromatin structure and transcription. Cell. 2002;108:475–487. doi: 10.1016/s0092-8674(02)00654-2. [DOI] [PubMed] [Google Scholar]
  • 41.Kaufman PD, Kobayashi R, Kessler N, Stillman B. The p150 and p60 subunits of chromatin assembly factor I: a molecular link between newly synthesized histones and DNA replication. Cell. 1995;81:1105–1114. doi: 10.1016/s0092-8674(05)80015-7. [DOI] [PubMed] [Google Scholar]
  • 42.Nabatiyan A, Krude T. Silencing of chromatin assembly factor 1 in human cells leads to cell death and loss of chromatin assembly during DNA synthesis. Mol Cell Biol. 2004;24:2853–2862. doi: 10.1128/MCB.24.7.2853-2862.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Sambrook J, Russell DW. Molecular Cloning- A laboratory manual. 3. Vol. 2. New York: Cold Spring Habour Laboratory Press; 2001. [Google Scholar]
  • 44.Cech TR. Alkaline gel electrophoresis of deoxyribonucleic acid photo-reacted with trimethylpsoralen: rapid and sensitive detection of interstrand cross-links. Biochemistry. 1981;20:1431–1437. doi: 10.1021/bi00509a005. [DOI] [PubMed] [Google Scholar]
  • 45.Wongwisansri S, Laybourn PJ. Reconstitution of yeast chromatin using yNap1p. Methods Enzymol. 2004;375:103–117. doi: 10.1016/s0076-6879(03)75007-1. [DOI] [PubMed] [Google Scholar]
  • 46.Dyer PN, Edayathumangalam RS, White CL, Bao Y, Chakravarthy S, Muthurajan UM, Luger K. Reconstitution of nucleosome core particles from recombinant histones and DNA. Methods Enzymol. 2004;375:23–44. doi: 10.1016/s0076-6879(03)75002-2. [DOI] [PubMed] [Google Scholar]

RESOURCES