We demonstrate here that, when administered i.p. to adult mice in combination with the cytostatic cyclophosphamide, an inducer of cross-links in the DNA molecule, human exogenous DNA, having reached the nuclear space of liver, thymus, spleen cells, integrates into the mouse genome. The integration of foreign DNA produces change in blood counts and is lethal to the treated mice. It is suggested that the integration mechanism acts through the repair events induced by the formation of covalent interstrand cross-links resulting in double strand breaks during replication fork arrest.

Chemical cytostatic agents whose molecules intercalate between stacked base pairs, thereby causing DNA interstrand cross-links (ICLs) between the complementary strands of DNA, are extensively used in anticancer chemotherapeutics. Cyclophosphamide (CP) has gained wide recognition as an anticancer drug. CP is activated in the liver by four hydroxylation reactions accomplished by the catalysts cytochromes. Antitumor effect of CP is believed to be due to its genotoxic DNA-alkylating phosphoramide mustard (PM). PM forms adducts with purine bases of the DNA molecule, especially with adjacent guanine residues and gives rise to ICLs in the two DNA strands (Yu et al, 1999; De Silva et al, 2000; Karle et al, 2001). The arisen ICLs are efficiently repaired in both the prokaryotic and eukaryotic cells.
In Escherichia coli, ICL repair is accomplished through excision and recombination pathways (Van Houten et al, 1986; Cheng et al, 1988; Sladek et al, 1989). In yeast cells, a similar mechanism involving excision endonucleases provides repair of ICLs in DNA (Jachymczyk et al, 1981; Magana-Schwencke et al, 1982, Prakash et al, 1993; Friedberg et al, 1995). With respect to mammals, humans too, the DNA machinery whereby such defects in the chromosomes are repaired remains unclear. A complex of excision repair factors has been implicated for the ICL repair mechanism. The factors first identified in xeroderma pigmentosum, XP, a hereditary human disease with a defect in the excision repair mechanism, include XPA, RPA, TFIIA, XPC, XPG, XPF-ERCC1 (Mu et al, 2000; Park et al, 1995). In vitro experiments have established certain details by which such DNA molecule damage is repaired. The XPG factor makes excisions at the 3'side (Harrington and Lieber, 1994; O'Donovan et al, 1994; Matsunaga et al, 1995) and the XPF-ERRC factor at the 5'side of the experimentally induced lesion (Bessho et al, 1997a). Kinetic experiments with incision formation provided evidence for the occurrence of 3' prior to 5' incisions (O'Donovan et al, 1994; Mu et al, 1996; Sijbers et al, 1996). An unwound stretch of the DNA molecule of about 30 bp is required for the two factors to be active (Matsunaga et al, 1996; Evans et al, 1997). It has been demonstrated that the XPF-ERCC1 heterodimer possesses additional 3'-5' exonuclease activity and, in the presence of replication protein A (RPA), the enzyme can bypass cross-link between the two DNA strands, thus forming a single-stranded DNA with a dinucleotide adduct (Mu et al, 2000). Experimental data have indicated that the XPA component of the excision complex is requisite for DNA opening (Evans et al, 1997). The total incision reaction is dependent on the ATP presence, it proceeds in the presence of the ATP-dependent TFIIH factor with helicase activity (Schaeffer et al, 1993; Hoeijmakers et al, 1996).
It has been shown (Bessho et al, 1997b; Caldecott and Jeggo, 1991; Collins, 1993; Pastink and Lohman, 1999; Wang et al, 2001) that the replication fork stoppage in the location of ICL induces DSB initiation and starts the repair processes deleting arisen damage. ICL repair in DNA strands implies at least three independent events entailing (i) an unknown system generating specific DSBs arising during replication in proliferating cells; (ii) an excision enzyme complex; (iii) a DSB repair system. In the case of DSBs induced by ionizing radiation, repair process occurs as nonhomologous joining of broken ends or synthesis depending strand annealing, in contrast, homologous recombination with Holiday structure formation is the major pathway for repair of the DSBs induced by cytostatic drugs (De Silva et al, 2000; Mu et al, 2000; Reardon et al, 1991; Saffran et al, 1994; Lambert et al, 2005; Li et al, 1999).
A model for ICL repair in DNA molecule strands has been suggested. According to the model replication is the major inducer of ICL repair (Li et al, 1999; Niedernhofer et al, 2004). Encounter of replication fork with ICLs renders it inactive. This is associated with DSB generation by an unknown mechanism (presumably with the involvement of specific endonuclease Mus81).
Excision factors machinery and transit exonuclease activity of XPF-ERCC1-RPA cleave the ICLs and relieve of torsional strain around it. The break and single strand gap with excision enzyme forming initiate homologous recombination whose first step is the formation of the heteroduplex by the activity of XRCC2 and XRCC3 between the lesioned and any other homologous sequence in the nucleus. The recombination process ends up the last step specific cuts made by excision enzymes, complete removal of the adduct, and repair synthesis on new template. The excisions can be possibly made at the time when the replication fork movement is blocked or later, during elimination of the adduct (De Silva et al, 2000).
Multiple DSBs arising in DNA chemically interacting with an alkalyting agent brings the nuclear synthetic process to a halt and impairs the cell cycle. Ultimately, a self-destruction program is turned on, the cell is either subject to apoptosis or to profound genetic changes. It is precisely such DNA damage that makes cancer cells lethal in anticancer therapy with cytotoxic agent.
In our previous studies (Yakubov et al, 2002, 2003), we have hypothesized the existence of a natural mechanism that may have an influence on the genetic component of the cells of multicellular organisms, using genomic DNA from biological fluid as an external genomic reference. The mechanism implies that the cell surface DNA binding receptors deliver genomic DNA fragments produced by natural apoptosis from the external environment (blood plasma, intertissue fluid, lymph) into the nucleus, that this turnover is continuous. Within the nuclear space, the internalized DNA fragments may be involved in all repair systems described above where the presence of a damage-free homologous sequence is the necessary condition for the process. From a survey of the literary data and in the light of our concept, it follows that, if the blood bed of an organism exposed to a strong cross-linking mutagen contained DNA fragments with homology to the host genomic sequences, the fragments would be used as substrate for homologous recombination in repair of DNA damage induced by cytostatic agent.
With this possibility in mind, a preparation of fragmented human DNA and an alkylating cyclophosphamide (CP) were administered to mice. Molecular genetic analysis of genomic DNA from treated mice revealed human DNA fragments integrated into the host genome.

A. DNA sources and probe pre-treatment
Human DNA was isolated from placenta of healthy women in childbirth in two maternity wards in Novosibirsk. This material had the required sanitary-regulatory documents for the out-patient clinics for serological confirmation that they were negative for HIV, hepatitis B and C, syphilis. The isolated DNA was certified and PCR diagnosed for the HIV, hepatitis B and C causative agents. Again, the results were negative.
Mouse DNA was isolated from liver, spleen, and thymus of CBA mice bred at the animal facility of this Institute.
Full-size genomic DNA was disintegrated hydrodynamically using the sonic disintegrator unit UZDN-37 (Russia) down to fragments of 600-6000 bp for further injections to CBA mice (Figure 1).

B. Experimental design for cyclophosphane treatment
Cyclophosphane (CP) was injected i.v. to CBA male mice (n=20) at 200 mg/kg body weight. One group (n=10) was administered i.p. 1 mg of human DNA 30 min before CP injection, 0.5 mg portion of DNA was administered 30 min after CP injection, this was followed by administration of the same DNA amounts on days 2 and 3. The controls (n=10) received saline.

C. Blood collection for counts in treated mice
To prepare smears, blood was withdrawn from the tail vein of each mouse before CP injection, also 4, 7, 10, 14, 17 days after it. The treated and control groups consisted of 10 mice each. Calculations for the leucogram were standard (Menshikov VV, 1987).

D. PCR amplification
Thermocycling conditions for amplification of human genome specific fragments were as follows: 94 ∫C for 2 min, 1 cycle; (94 ∫C for 30 sec, 72 ∫C for 1 min), 33 cycles; storage at 4 ∫C. The following approach was used to identify the minor fragments of human origin in the treated mice genomes. The minor human templates for the treated mice genomes were first amplified in the presence of a specific primer. Then, the obtained material was amplified in the presence of two specific primers under the indicated thermocyclic conditions in another PCR round. This approach allowed us to detect human DNA fragments in the genomes of treated mice No 1 and No 8. The human genome specific primers used were as follows:
Pr.9 CGAGGCGGGAGGATCACTTGAGCCC (25)
Pr.10 CGGCTCACTGCAGCCTCGACCTCCC (25)
Pr.11 GCGCGCGCCACCACGCCCGGC (21)
Amplification was performed using the Tertsik DNA-technology equipment (Russia).

E. Sequencing PCR fragments
PCR fragments were sequenced at the Interinstitute DNA Sequencing Center, Siberian Department of the Russian Academy of Sciences. Big Dye 3.1 (Applied Biosystems, USA) was used to sequence DNA fragments.

F. Southern-blot-dot hybridization
Southern-blot-dot hybridization was performed as described by Maniatis et al (1982) at 68∫C under stringent conditions. The membrane was washed with 0.1xSSC containing 0.1% SDS The amount of labeled material (spots and hybridization bands cut out from membrane filter) was determined using a 1209 RacBetta counter (Finland).

G. Calculations of the potential number of substitution sites for exogenous extrachromosomal DNA fragments in DNA ICL repair

The mouse genome is ~2.5x109 bp = 2.5x106 kb.
The human genome is 3.3x109 bp = 3.3x106 kb.
The length of Alu-repeat is ~0.3 kb (Jurka, 2004).
The number of Alu-repeats in the human genome is ~10% (Bogerd et al, 2006), or 106 copies of 0.3 kb monomer.
The length of B1 repeat is ~0.15 kb (Kolchanov et al, 1988).
The number of B1 repeats in the mouse genome is ~10%, or ~105 copies of 0.15 kb monomer.
Alu elements occur every 3-4 kb.
B1 repeats occur every 3 kb.
A cross-linking agent at therapeutic doses induces 1000-2000 DNA ICLs per cell (Palom et al, 2000; Warren et al, 2001; Niedernhofer et al, 2004).
Exogenous DNA fragments are 2-30 nucleosome units in size, this makes up ~0.5-6.0 kb (3.0 kb on average).
When 1000-2000 μg of DNA, which corresponds to 50-100 μg of the preparation per 1 g of body weight, is administered to mice (~20 g), the nuclear extrachromosomal space of the proliferating cells can harbor up to 2% (of the haploid genome) of exogenous DNA fragments (Rogachev et al, 2006). Thus, the nuclear space can contain at the same time of the order of 2000 Alu-monomers in DNA fragments of 3.0 kb on average. CP plus a preparation of fragmented human DNA were administered to mice. In the treated mice, CP through its phosphoramide mustard metabolite forms cross-links in the DNA molecule, concomitantly blocking replication fork. In the arisen recombinogenic situation, with stalled replication fork and activated incision enzyme system, homologous sequence for homologous exchange repair all must be available for repair to be sufficient. We suggested that Alu-monomers of human exogenous DNA by virtue of their high homology to the mouse genome B1 structures, would complete homologous exchange repair at the expense of regions homologous to B1 repeats. An expected consequence would be integration of exogenous foreign DNA into the mouse host genome.
B1 repeats are spaced every 3 kb in the genome, the number of cross-links induced in the nucleus being 1000-2000. About 20,000 Alu-monomers distributed among ~3.0 kb fragments are present in the interchromosomal space. This means that the number of sequences potentially capable of substitution exceeds 10-fold that of the available substitution sites.
The probability for homologous substitution of B1 repeat by Alu-monomer is calculated as P = [exchange region length – B1 repeat length]/[average period (distance between B1 repeats)]. This yields [3.0 – 0.15]/3.0 = 0.95, i.e. 95%

H. Estimation of the number of genomes in treated tissue cells containing X-Alu homologous sequences based on absolute counts of labeled material in dots

DNA from treated and control mice in amounts indicated below was dotted onto Hibond N membrane (Amersham) and hybridized to a α-32P labeled X-Alu DNA fragment.
Absolute counts of labeled material in dots (cpm)

To obtain counts for moderate hybridization zones, we chose spots in which the amount of applied DNA did not exceed the resolving capacity of Hibond N membrane. Counts per minute (cpm) were determined in dots of the two treated mice after subtraction of the average specific background, it was at the background level for mouse No 1, some cpms above or equal to the average specific background for mouse No 8. The method was not sensitive enough to accurately define the ratio of the applied DNA quantity to cpms per dot. For this reason and relying on the Southern blotting, we estimated copy number of X-Alu monomers in the genomes of the treated mice. Four X-Alu monomers, one or several copies per haploid genome, were determined for mouse No 8, less than a single copy per haploid genome for mouse No 1.
1. Count for the 0.1 ng X-Alu containing dot is the same as that for the 10 ng human DNA containing a dot. Consequently, the human genome contains ~ 1% of human X-Alu fragments.
2. The dot, which contains 5000 ng of DNA from mouse No 8, is virtually not above the specific background level, its count is 50-fold smaller than that for the dot containing 0.1 ng X-Alu DNA or 10 ng human DNA, i.e. 0.00004% of the human X-Alu fragments are present in the genome of the treated mouse.
3. The human haploid genome contains about 3.3x109 bp DNA, 1% X-Alu about 300 bp long amounts to 105 copies per haploid genome.
4. The mouse genome is of about 2.5x109 bp, a little shorter than the human genome. For simplicity, we consider the genomes of equal lengths in both organisms, if 1% of the haploid genome contained 105copies of X-Alu monomer, it would be present in about 4 copies in the genome of mouse No 8. This means that monomer X-Alu is present in at least 4 (a single or several) copies in the single cell genome. Count is 5-6-fold less for treated mouse No 1 than for mouse No 8, so that copy number is smaller than a single copy per haploid genome. Taking into account the possible clusterisation of X-Alu monomers into a block and the existence of several integration sites, less than 1 cell in the three treated organs of the two mice contains X-Alu monomers.
Thus, several kbs of foreign human DNA homologous to Alu-repeats may be present in the host genomes of the treated mice. At the same time X-Alu repeats are limited by the restriction sites BamHI, HindIII or BamHI-HindIII, as follows from the genomic blot analysis.
The following parameters must be defined to estimate the possible variants of exogenous DNA integration.
The number of cells in organs from which DNA was isolated.
Liver, 1.0 mg, 106 cells. Average liver weight is 2 g. Total cell number is 2x109.
Thymus, 5-6x107 cells.
Spleen, ~3x108 nucleus-containing cells.
Spleen colony, a derivative of blood stem cell, contains 10x6 cells.
Thus, the total cell number is 23.5x108, spleen colonies constitute 0.04% of this number.

A. Integration of foreign DNA into the genome of somatic tissue cells in the adult
Based on our earlier observations, we attempted here to obtain assurance that exogenous human DNA integrates into the mouse genome. Our previous work has shown that exogenous DNA affected the development of erythro- and myelopoiesis progenitors, thereby speeding up the recovery of white and red blood cell progenitors following exposure to the cytostatic agent (Yakubov et al, 2003; Nikolin et al, 2006). The important observations were stimulation of the division of the stem blood cells exposed to cyclophosphan, as well as rescue (viability retention) of a considerable proportion of blood stem cells subjected to intense chemotherapy (Yakubov et al, 2003; Nikolin et al, 2006; Patent application No 2006127134, Russia). Based on previous considerations (Yakubov et al, 2002, 2003) and our research results, it has been suggested that, having entered the nuclei of the stem blood cells (and of any proliferating cells), exogenous DNA becomes involved in repair of DNA damage incurred by cytostatic agent. Based on the data on hemopoiesis stimulation under the effect of both allogenic and xenogenic DNAs (Yakubov et al, 2003; Nikolin et al, 2006), we also suggested that the DNA of taxomically close species can be involved in homologous exchange during repair of cytostatic induced DNA damage and, therefore, integrate into the host genome.
We assume that, in the recombinogenic situation arising when replication fork is arrested, DSB form at the cross-link site, the excision repair system activated is crucial for the DNA integration (both local on the chromosome and extrachromosomal of allogenic and xenogenic origin). In this situation, the presence of exogenous DNA in somatic cell nuclei in animal tissues during injection of cytostatic agent is a requisite for designing experiments.
The major criteria for the events that affected the state of the treated mice was the recovery rate of the hemopoietic function defined by changes in blood counts, their unexpected shift elicited by CP plus exogenous DNA cotreatment, and mouse general status.
The pattern of changes in blood counts (Figure 2) supported the experimental results that demonstrated accelerated recovery of inhibited hemopoiesis under the effect of combined therapy with CP and fragmented human DNA preparation in all the treated mice. However, the unusually great increase in granulocyte number persisted in mice No 1 and 8. It is pertinent to note that neutrophilia may be caused by diseases associated with tissue disintegration or immune system stress (Roitberg and Strutynsky, 1999).
Surprisingly, mice subjected to combined treatment started to succumb on day 6. The last one died on day 56. Mice phenotypically manifested a different sets of symptoms. It appeared likely that the very dramatic consequences were caused by integration of the foreign DNA into the treated mouse genome. It appeared likely also that the beneficial effect of xenogenous exogenous DNA on the inhibited white blood cell progenitors was not due to homologous substitution of the chromatin regions under the impact of cyclophosphan. In fact, integration of foreign DNA is lethal to the host. Regrettably, post mortem studies were not performed because the dramatic events were unpredictable. Liver, spleen and thymus were excised for extraction of DNA and its subsequent analysis from all the dead mice.
The Alu-sequence was chosen to determine whether human DNA integration is possible.

B. Molecular characteristics of integrated sequences homologous to the human Alu-repeats
In choice of the Alu-repeats for analysis, we reasoned as follows. As known, Alu-elements comprise up to 10% of the haploid genome, this corresponds to about 106 copies of 300 bp Alu monomers. The human Alu-repeats are scattered throughout the genome forming clusters, occurring every 3-4 kb on average. The structure of most human Alu-repeats is dimeric. A number of tetrameric sequences and the monomeric repeats in single instances were identified. Structurally similar to Alu-repeats were found to be present in the mouse genome. They were called B1 and are almost always monomers (Britten et al, 1988; Kolchanov et al, 1988). The homology degree between the B1 and Alu-repeats is higher, their consensus sequences being up to 65% identical. The percent can be as high as 90% for the real monomers. The mouse B1 repeats are organized similarly to the Alu-repeats in the genome.
Fragments of exogenous extrachromosomal DNA that represent up to 2% of the haploid genome can accumulate in the nuclear space of the cell (Rogachev et al, 2006), consequently, about 20,000 copies of the Alu monomers can be present in the genomic fragments of about 3 kb on average in the cell nuclei of mice treated with exogenous DNA.
The cytostatic agents at doses used in chemotherapy and research induce of the order of 1000-2000 cross-links per nucleus (Palom et al, 2000; Warren et al, 2001). Like Alu, the B1 repeats occur every 3 kb in the mouse genome, it follows from calculation that the probability of a homologous substitution between two similar repeats is over 95% (see Materials and Methods section). This means that a cross-link is present in virtually each cell in the B1 repeat or in its vicinity, that it can induce exchange with an Alu-sequence of extrachromosomal origin.
In the case of damage that affects genomic repetitive sequences in the genome, numerous potential homologous regions in its most different parts can be recruited for homologous repair of the damage, if the repair requires homologous exchange. When a DSB caused by a stalled replication fork occurs in a homologous region, variants of repair may be envisioned. One is gene conversion whereby DNA is nonreciprocally transferred from donor to host, with the damaged allele being, as a rule, the host. Another variant is pairing of a single strand followed by synthesis and migration of the strand without an associated crossing over (Langston and Symington, 2004; Leung et al, 1997; Bartsch et al, 2000; Nickoloff, 2002). Provided that the extrachromosomal fragment is long enough and that linear sequences, which flank the damage homologous to the fragment ends, are present in the genome, the fragment can completely repair the damaged locus by double reciprocal exchange through the described mechanism (Hastings et al, 1993; Leung et al, 1997; Li et at al, 2001; Yakubov et al, 2003; Langston and Symington, 2004). The central portion of the sequence, which forms by its end regions two heteroduplexes with the homologous chromosome regions, integrates into the host genome without inducing any changes, by just replacing the chromosome region bounded by two repaired end regions. In DSB homologous repair, the nucleus can contain numerous homologous sequences, for example, repeated genomic, in such a case, vicinity to the homologous exchange site and homology degree are the two important factors that affect the choice of the appropriate homologous segment (Schildkraut et al, 2006).
It has been suggested that the large number of Alu-containing fragments in the nuclear space of mouse cells and the high homology between human Alu and mouse B1 repeats would enable them to pair with each other and be involved in homologous pairing.
Thymus, liver and spleen from treated mice were used to isolate DNA in the conducted experiments. Analysis of the X-Alu copy number in the two positive mice established that it was less than a single copy for mouse No 1 and about 4 (a single – a few) copies for mouse No 8 per haploid genome. This meant that only a part (under the assumption that the X-Alu monomers integrated as blocks and/or at a number of genomic sites in the cell for mouse No 8 and without this assumption for mouse No 1) of all the cells of the chosen tissues may contain integrated X-Alu copies. If the cell received several monomer copies, or in the case of X-Alu monomer clusterization, the portion of cells containing foreign sequences integrated into the genome would be still smaller.
Analysis of the strong hybridization patterns made apparent the following characteristics of γ-Alu containing human DNAs that integrated into the genome.
1) The relation between the estimated number of copies and detected bands, as well as the experimental data on the organization of the Alu and B1 repeats in the genome, evidence that in both mice presumably only a part of cells from which DNA was isolated contain homologous X-Alu sequences. The important finding was the integration of several different X-Alu monomer-containing fragments.
2) It appeared unlikely that exogenous DNA fragments integrated into a unique site in the different cells. Accepting multiple integration, the detection of individual fragments that contained sequences homologous to X-Alu may be explained as follows. Integration occurred at different sites of the genomic regions showing homology to the X-Alu sequence. The integrated fragments were of different lengths, not shorter than about 2 kb. The integrated fragments harbored a single X-Alu repeat or a monomer cluster bounded by BamHI, HindIII or BamHI-HindIII restrictases. Digestion with these restrictases liberated the same monomer or cluster of monomers, regardless of their location and number in the host genome. The fragment was of ~2.2 kb for mouse No 1, fragments were about 2.0, 2.2, 3.1, 4.0, 6.0 kb for mouse No 8. It is inexplicable why only the fragments containing homologous X-Alu sequences integrated into the genomes of treated mice. However, the difference in the number and distribution of the X-Alu repeats among the genomic fragments of the two treated mice evidence for individual specificity in integration in the given experimental conditions.
3) The size of the integrated fragments containing X-Alu cluster was in the 2 – 6 kb range. This meant that a long homologous region was required for integration in the form of gene conversion or that a mechanism of double reciprocal exchange of paired end regions of the fragment.
4) The rest of the succumbed mice, as well as those of the other experimental series (a single injection of CP and/or human DNA, 30 mice), did not contain integrated X-Alu sequences. However, their death during the long experiment suggested that fragments of exogenous human DNA also integrated into the genomes of these mice whose sequences represented different parts of the human genome that discriminated homologous regions in the genomes of the treated mice.
There exists another possibility, integration in several brain marrow stem blood cells whose survived offspring gave rise to individual colonies in spleen. Precisely these colonies consisted of cloned cells into which individual fragments integrated. It was expected that the fragment would integrate at a single site of the genome defined by a parental blood stem cell. Estimates of copy number for X-Alu monomers in the genomes of treated mice in the case of their possible clusterization of X-Alu monomers are consistent with our vision of the events. The estimates were not less than a single – a few copies per haploid genome; cell number in the spleen colony was estimated as 0.04% of the total number of treated cells, with the total number of spleen colonies being up to 40.

C. Events possibly due to exogenous DNA integration into the genome in treatment with CP plus exogenous DNA preparation
The question was: How transgenesis might have occurred in the tissue cells of adult mice given the cross-link inducer CP in combination with fragments of exogenous extracellular foreign DNA? Numerous recent studies have shown that the appearance of ICLs in proliferating cells is associated with arrest of the replication fork encountering steric hindrance. This event enhances repair with the result that an extremely recombinogenic structure forms at the cross-link site. The consequences are manifold: DSB formation in the newly synthesized strand in the immediate vicinity to the cross-link site, replication fork arrest, the generation of a single strand region from 70 bp (Mu et al, 2000) to 700 bp (Sinden and Cole, 1978) through the activity of the XRCC1-XPF-RPA complex. The two structures form sequentially at the individual ISL site. There presumably exists a time when they are present together. A number of models for repair of arisen DNA damage have been described, they all incorporate excision exonuclease activity of the XRCC1-XPF-RPA complex followed by homologous recombination (De Silva et al, 2000; Niedernhofer et al, 2004). In the case of generation of such a structure, foreign extrachromosomal DNA has presumably the opportunity for integration into the recombination site by the same mechanism (Kucherlapati et al, 1984) or repair via homologous recombination with the sister chromatid (Niedernhofer et al, 2004). Two DNA damages are obviously required for repair. In the conceived version of events, any one of the known repair mechanisms of the DNA strand is possibly feasible (Thomas et al, 1986; Van Houten et al, 1986; Hastings et al, 1993; Rouet et al, 1994; Jasin, 1996; Liang et al, 1998; Leung et al, 1997; Bartsch et al, 2000; Li et at al, 2001; Nickoloff, 2002; Yakubov et al, 2003; Langston and Symington, 2004). However, the ultimate necessary condition is repair by homologous exchange between the damaged region and the homologous sequence internalized within the nuclear space.
It may be assumed that ICL occurred between two B1 repeats. Our search for homologous sequence for repair of the ICL break induced by the stalled replication fork was successful: the fragment containing Alu-repeat or its clusters within the 2 – 6 kb fragments seen in the genomic blot paired with its end monomers to the B1 sequences. Further, the intermediate is resolved with integration of the entire fragment into the host genome by the mechanism as previously described (Li et al, 2001). This is the reason why the regions of human DNA integrated into the mouse genome are rather long.
Retroposition of integrated into the mouse genome Alu-repeats, which was induced by the genotoxic stressor etoposide, has been observed; by blocking topoisomerase II, etoposide resulted in DSBs and global activation of all repair mechanisms (Hagan et al, 2003).
We do not exclude the possible triggering of the retrotransposition mechanism for fragments with the X-Alu sequences and the promoter region for RNA polymerase III, moreover that the described transposition phenomenon has been observed for a mouse-human system virtually identical to ours, thereby indicating that there exists an enzymic system in mouse cell capable of transposing human Alu-sequences.
However, the fact of cluster retransposition of Alu-repeats has not been established as yet, moreover using extrachromosomal fragments as a template. Single copies of the repeats integrate into the target site by the reverse transposition mechanism, involving RNA synthesis, reverse RNA synthesis, transposition of newly synthesized DNA copies to the target site. First, a copy of Alu-element within the host genome is requisite for the process; second, site-specific integration requires specific endonuclease-reverse transcriptase, a DSB-inducing enzyme that directly accomplishes integration and repair of the DNA Alu-copy. To our knowledge, there has been no reference to the presence of this enzyme during events associated with ICL repair in the available literature. Finally, genomic hybridization failed to reveal distinct hybridization band during site-specific Alu retransposition into the nuclear space of the treated mice and also in the case of contamination of the samples with human DNA. This was because the integrated Alu-repeat did not contain sites for the chosen restrictases statically distributed throughout the genome. Hence, a united restriction fragment could not appear in random transposition of different cells into different genomic restriction fragments.
In summary, then, we believe that in the current experiments a CP-mediated in vivo transgenic somatic transformation by fragments of exogenous foreign DNA occurred in adult mice.
D. Envisioned therapeutic application of the revealed DNA integration and of its consequences
The factual evidence of the integration of exogenous DNA fragments into the eukaryotic genomes in a recombinogenic situation generated by a cytostatic cross-linking anticancer agent is thought-provoking. There appear to be unprecedented opportunities to develop a new approach to treatment of cancer and variety diseases in genetically damaged cells (Patent application No 2006127134, Russia).
The repair system of cell promptly becomes active in response to the multiple ICLs induced by anticancer drugs. The choice of the cytostatic is tailored in such way that each cell receives a cytostatic shock (about 2 000 ICLs per cell) deadly to the cancer and other proliferating cells. Cells die presumably because
i) they cannot cope to form at the same time sufficient amounts of repair complexes so that a part of the incurred damage remains unheeded, triggering apoptosis;
ii) when a homologous chromosome or a sister chromatid serves as substrate for homologous recombination of a homologous chromosome, multiple ICLs pose steric hindrance to simultaneous synapsis of several damaged sites with homologous regions. This is another obstacle to complete repair of all the arisen ICLs with inevitable apoptotic cell death;
iii) the last, not the least important, cause of cell death is that the genetic mode of the cell remains the same as before ICL induction, when a homolog or a sister chromatid is used for homologous recombination repair, i.e., if the cytostatic diadduct arose in the region of the gene whose mutation caused cell malignization, repair using endogenous cell substrate would not bring about beneficial genetic change in the mutated gene. If the genome harbored oncomutations, it would continue to do so despite this repair.
The only way to get rid of such oncomutations is to kill the cancer cells and, with one stroke, all the other proliferating cells. Cytostatic drugs do this job well.
It is suggested that fragments of exogenous allogenic DNA would be involved in repair culmination when all the cross-linked DNA strands are restored. A favorable circumstance is that up to 2% of the haploid genome of extracellular DNA can be present in the nuclear interchromosomal space as fragments of size commensurate with that of DNA fragments resulting from normal apoptosis (Rogachev et al, 2006). The extremely recombinogenic situation engendered by ICL induction and replication fork stalling makes the recombination machineries of the cell highly active. If DNA fragments homologous to the repaired site and serving as substrate for homologous recombination were present at that time in the nucleus, recombination would surely take place, not with the defective endogenous homolog, but with the mutation-free exogenous therapeutic DNA. The newly integrated sequence would become fixed as a result of successive cell divisions, and the oncomutation would be resolved.
The artificially created high recombinational activity of the cell is the pivot of the proposed antitumor therapy based on coadministration of exogenous DNA and cytostatic cyclophosphan. Recombination occurs at those sites where ICLs were induced, owing to the chemical nature of the inductors, ICLs can arise at any site of the genome, regardless of its organizational level and conformational DNA-protein association. ICL emergence is a purely static chemical event, so that complete replacement of all the genomic ICLs is dependent only on the number of ICLs induced at oncomutation sites, this number being in turn dependent on the number of oncomutations, chemotherapies, and the patient's luck, whether an ICL will happen to be located in an oncomutation-containing site or not.
Exogenous DNA similarly affects healthy cells exposed to a cyclophosphan. In such a case, by involvement in repair in healthy cells, fragments of exogenous therapeutic DNA rescue them from apoptosis, promoting the retention of cell populations in various tissues (blood stem cells, epithelium), facilitating chemotherapy, and allowing to design subsequent cycles of treatment with cytostatic. Thus, we demonstrate a strategy effective at controlling cancer.

The authors thank Alexandra Kokoza-Bogacheva and Anna Fadeeva for their technical help in preparing the paper and translation.
The work was funded by LLC Panagen.