[1] 17w Excision of amplified viral DNA at palindromic sequences from the adenovirus type 12-transformed hamster cell line T637

[1] 241w In the DNA of the adenovirus type 12 (Adl2)-transformed hamster cell line T637 -20-22 viral DNA molecules per cell are covalently linked to cellular DNA. Spontaneously arising morphological revertants of T637 cells have lost the bulk of the viral DNA. We have been able to mimic the excision event of viral DNA, as it occurs during reversion, by autoincubation of isolated nuclei from T637 cells. The same Adl2 DNA sequences, which had been deleted in morphological rever- tants, proved highly sensitive to endogenous nucleases in isolated nuclei of T637 cells. Viral DNA sequences, which persisted in the revertants, are resistant to endogenous nucleases in isolated T637 nuclei. All attempts to clone the nuclease-sensitive sites of Adl2 DNA in cell line T637 have so far failed. After denaturation and renaturation of T637 DNA followed by treatment with S1 nuclease, large fold-back structures of DNA have been found. These snap-back struc- tures were derived from precisely those viral DNA restriction fragments which were uncloneable. The fragments containing palindromic sequences were both highly sensitive to endo- genous nucleases in isolated T637 nuclei and were absent from the DNA of all revertant cell lines. Moreover, the palindromic sequences are susceptible to the phage T4-speciflic endonuclease VII which specifically attacks cruciform struc- tures in DNA. The peculiar structures at the termini of inte- grated Adl2 DNA molecules are highly sensitive to endo- genous nucleases in isolated nuclei. These nucleases may be related to the reversion event.

[1] 170w Cell line T637 was generated by transformation of BHK21 cells with human adenovirus type 12 (Ad12) (Strohl et al., 1970). The viral DNA molecules are probably inserted in clusters and were not arranged in a true tandem fashion (Stabel et al., 1980). Blot hybridization experiments showed that at least the terminal 143 nucleotide pairs of Adl2 DNA were over-represented relative to the total number of Ad12 genomes persisting in T637 cells (Stabel et al., 1980). Using different prokaryotic vector systems, all attempts so far to clone junction fragments between Adl2 DNA and DNA from hamster cell line T637 failed, while internal viral DNA fragments of cell line T637, or junction sites from other Ad 12-transformed lines or Ad 12-induced tumors could readi- ly be cloned (Deuring et al., 1981; Stabel and Doerfler, 1982). Thus, the idea arose that in cell line T637 peculiar structural arrangements of Adl2 DNA, perhaps palindromes, might ex- ist at the Ad2-cell DNA junction sites, or at the sites of linkage between adjacent viral DNA molecules.

[2] 149w Spontaneous morphological revertants of cell line T637 had lost all or the bulk of the integrated Adl2 DNA se- quences (Groneberg et al., 1978; Eick et al., 1980). In some of the revertant cell lines, only the right-hand end of the Adl2 genome, or one intact and a fraction of a second viral genome, persisted. These persisting viral genomes were co- linear with the authentic viral DNA, and there was no evidence for rearranged termini. It appeared that the original viral genome, that had been integrated intact and co-linear, eventually persisted in the revertants, whereas amplified genome equivalents were prone to rearrangements and dele- tions (Eick and Doerfler, 1982). Loss of previously integrated Adl2 genomes proved particularly common in cell lines recently established from Adl2-induced tumors (Kuhlmann et al., 1982). Thus, unknown post-integrational events must be responsible for the permanent, stable fixation of viral DNA (Kuhlmann and Doerfler, 1983).

[3] 76w We have now compared the mode of excision observed in spontaneously arising revertants with that generated ex- perimentally by incubating isolated nuclei of T637 cells. We present evidence that the Ad12 sequences which are lost from cell line T637 in generating the revertant line TR12, are also sensitive to autodigestion of T637 nuclei, presumably due to the action of endogenous nucleases recognizing specific DNA arrangements, perhaps palindromic sequences at sites located in between adjacent viral genomes.

[1] 13w Patterns of AdJ2 DNA integration in cell line T637 and in morphological revertants

[2] 208w We have continued to investigate morphological revertants of the Ad12-transformed cell line T637 in order to analyze mechanisms of insertion and excision of viral (foreign) DNA in mammalian cells. Upon cleavage of the DNA from cell line T637 with the restriction endonuclease PstI, three off-size bands arose which hybridized to Adl2 DNA. These off-size bands were designated a, b and c (Figure 1) and were due to the terminal PstI fragments of integrated Adl2 DNA. There was possibly a fourth off-size band migrating between PstI fragments C and D. In these fragments, terminal DNA of an integrated Adl2 DNA molecule was linked to DNA of another Adl2 DNA molecule or to cellular DNA. In T637, the terminal fragments of Adl2 DNA were absent from the authentic positions in virion DNA (Figure 1) and were dis- placed to higher mol. wt. positions (a, b, c). All of the inter- nal PstI fragments of Adl2 DNA were represented in the DNA of T637 cells and co-migrated with the authentic virion DNA fragments. Fragments D, E, G, H, P, I and M ap- peared under-represented, probably because most of the in- tegrated Adl2 genomes in line T637 had an inversion and partial deletion at the right terminus (Eick and Doerfler, 1982).

[3] 172w It was interesting to observe that in the revertant lines G12, TR12 and TR 1l, the PstI off-size bands a and b (Figure 1) were deleted, whereas off-size band c remained. Apparently, viral DNA was excised at certain specific junction sites (cor- responding to bands a and b), while the junction site cor- responding to fragment c was unaltered. The internal viral Fig. 1. Patterns of persistence of Adl2 DNA in T637 cells and of four morphological revertant lines G12, TR12, TR II and TR2. The DNA of the lines or of Ad 12 as indicated was cleaved with the restriction endonuclease Pstl. The DNA fragments were separated on a 1 % agarose slab gel and transferred to nitrocellulose filters by the Southern blotting technique. Ad12-specific sequences were detected by hybridization to Adl2 DNA which was 32P-labeled by nick-translation. Adl2 (20 or 20x in Figures 2 and 5) indicates that an amount of Ad 12 DNA was used which was equivalent to 20 genome equivalents relative to 10 Itg of cellular DNA.

[4] 37w That was the amount of cellular DNA used per slot. Adl2 marker DNA was mixed with 10 /tg of B3 (BHK21) DNA as carrier. Off-size bands a, b and c in line T637 are indicated by arrows.

[5] 261w DNA fragments all persisted in amounts equivalent to one copy and a fraction of a second one per diploid genome, ex- cept in revertant line TR2 in which only the right 60-700/o of one viral DNA molecule persisted (Figure 1). We concluded that in cell lines G12, TR12 and TRi 1, junction sites between Adl2 DNA molecules or Adl2 DNA and cellular DNA pre- sent in line T637 had been deleted together with the bulk of the integrated Adl2 DNA molecules. One site of junction ob- viously remained intact. and T637 nuclei subjected to autodigestion. Nuclei of T637 cells were prepared as follows. Cells were washed three times with phosphatebuffered saline and were scratched off the plate with a rubber policeman. Cell membranes were lysed in 20 mM NaCl, 10 mM MgCI2, 10 mM Tris- HCI, pH 7.5, and 0.5Wo NP-40. After an incubation for 5 min at 0°C, the nuclei were spun down and washed three times in the same buffer without NP40. The nuclei were incubated in the washing buffer for 1 h at 370C. The reaction was stopped by extracting the total intranuclear DNA by the SDS-proteinase K-phenol method. DNA from T637 nuclei and from sources as indicated was cleaved with the PstI restriction endonuclease. DNA fragments were separated on a 1 o agarose slab gel, blotted and hybridized to 32P-labeled Adl2 DNA. Details have been described, see also legend to Figure 1. Off-size bands of T637 cells were designated a, b, and c. The arrow pointing to c refers to that band in T637 and TR12 DNAs.

[6] 264w We attempted, mainly by in situ nuclease digestion, to mimic the excision event of Ad12 DNA sequences as this event might have occurred in T637 cells and might eventually have led to the generation of revertant cells. In the procedure finally adopted, isolated nuclei of T637 cells were auto- digested as described (see Materials and methods). Subse- quently, the DNA was extracted and analyzed by blot hybridization using Adl2 DNA, T637 DNA and DNA from the revertant line TR12. Following autodigestion of T637 nuclei the Adl2 off-size bands that had also been lost in the Fig. 3. Fold-back DNA at the sites of junction between viral and cellular DNA in cell line T637. T637 DNA was digested with restriction endonuclease HindlIl (1), BamHl (2) or with both enzymes (3). After digestion, the DNA was denatured in a 100°C oil-bath for 5 min and subsequently chilled on ice (denatured). The fold-back reaction of palindromic sequences was carried out for 15 min at room temperature. The DNA was then ethanol precipitated. The DNA was dissolved in SI buffer and incubated with 1000 U of SI nuclease for 30 min at room temperature under conditions described in Materials and methods (denatured + SI). Thus, '(denatured + S1)' refers to DNA preparations that were denatured, briefly renatured, and S1 treated. HindIll, BamHl and EcoRI maps of Adl2 DNA are also presented. revertant line TR12, (i.e. bands a and b) were obliterated (a) or nearly obliterated (b). Off-size band c was not noticeably altered. The internal fragments of integrated Ad 12 DNA pro- ved insensitive to autodigestion (Figure 2).

[7] 80w Obviously, the junction sites corresponding to bands a and b in cell line T637 were particularly prone to digestion by endogenous nuclear nucleases, and these sites were deleted in the revertants investigated. Thus, it was conceivable that the DNA sequences at junction sites a and b exhibited special conformations and/or lacked protection by proteins bound to DNA in chromatin. More complicated explanations were also possible. Evidence for inverted repeats at the a and b sites ofjunction in cell line T637

[8] 282w One possible configuration of DNA at the a and b sites of junction was that of palindromic sequences. This possibility was investigated by denaturing and briefly renaturing T637 DNA that had been cleaved with restriction endonuclease HindIll or BamHI or HindIll and BamHI, and subsequently treating the DNA with SI nuclease. The DNA fragments were then separated by electrophoresis on agarose gels, blotted and identified by hybridization to Adl2 DNA (Figure 3a) or to the terminal fragments EcoRIC (Figure 3b) or BamHIC of Adl2 DNA (Figure 3c). In T637 cells, the bulk of the integrated Adl2 genomes carry an inversion at the right terminus such that the originally internal BamHIC fragment is transposed to the terminal position (Eick and Doerfler, 1982). It was ap- parent from the data presented in Figure 3 that fold-back structures at the site of junction existed. Upon cleavage with HindIII or BamHI, one distinct band of viral DNA was created by denaturation and brief renaturation followed by SI treatment (Figure 3a). Denaturation alone did not yield the same bands. The HindIII or BamHI generated bands were of unequal size which was explainable by the different location of restriction sites. Cleavage with both restriction en- donucleases probably generated very small fragments that could not be detected on the gel (Figure 3a). The Adl2 DNA fragment persisting upon HindIll cleavage, fold-back reac- tion and SI treatment was localized to the left terminal EcoRIC fragment (Figure 3b). In both the HindIII and BamHI generated fragments, the inverted right terminus of integrated viral genomes was represented (Figure 3c). In none of these experiments did denaturation by itself generate the same, or similar fragments compared with the snap-back reaction (Figure 3a -c).

[9] 66w If fold-back structures really existed they should be eliminated by denaturation and SI cleavage of the DNA. The experimental results verified that prediction as shown for T637 DNA that had been cleaved with HindIII and sub- sequently treated as shown in Figure 4. The denaturation and brief renaturation followed by SI treatment of DNA from the revertant line TR12 did not give rise to fold-back structures.

[10] 49w Obviously, no such structures persisted in the revertant cell line. The data presented here, provide evidence for the ex- """ &.4. so HindMIxAdl2 EcoRIC Fig. 4. The SI-resistant DNA fragment became SI-sensitive after a second denaturation of the DNA. Experimental details were described in the legend to Figure 3.

[11] 14w istence of fold-back structures in between adjacent integrated viral genomes in cell line T637.

[12] 548w The DNA sequences at junction sites which gave rise to foldback structures were the same as those that could be eliminated by autodigestion of T637 nuclei Figure 5 correlates the data obtained by autodigestion of T637 nuclei and by denaturation and brief renaturation followed by SI nuclease treatment of T637 DNA. The DNA of T637 cells was cleaved with the restriction endonuclease PstI, and the DNA was subsequently denatured, briefly renatured and treated with SI nuclease. Blot hybridization us- ing the BamHIC fragment of Adl2 DNA revealed two distinct bands, a and b (Figure 5), representing fold-back structures at the right-hand junction sites. Ad12-specific DNA bands a and b were not detectable in the DNA of T637 DNA and of DNA from autodigested nuclei upon denaturation and brief renaturation of the DNA. Experimental details were described in the text or in the legends to Figures I -3. In the experiment presented here, the DNA was cleaved with restriction endonuclease Pstl. The 32P-labeled BamHIC fragment of Adi2 DNA was used as probe. cells or nuclei, nor were they created by simple denaturation of T637 cellular or nuclear DNA (Figure 5, lanes 2-5 from the left). When T637 nuclei were autodigested as described above (Figure 2), and when the resistant DNA was also denatured, briefly renatured and treated with SI nuclease, the DNA sequences corresponding to fold-back structures a and b were completely (a) or partly (b) eliminated by endogenous nucleases. The fold-back structure equivalent to band b prov- ed somewhat more resistant to autodigestion, and thus might differ in structure from band a. These data show that the Ad 12 DNA sequences at the right terminus of integrated viral DNA that carried palindromic sequences (Figure 3) were also sensitive to endogenous nucleases in T637 nuclei. The palindromic sequences appear to occur in between adjacent viral quences under physiological conditions might assume cruciform configurations only part of the time. The following experiment was designed to test this hypothesis. T637 DNA was cleaved with the restriction endonuclease PstI (Figure 6, track 1), with PstI and then with endonuclease VII (Figure 6, track 2), or with the same enzyme combination in reverse order (Figure 6, track 3). Subsequently, the fragments were analyzed by blot hybridization using the BamHIC fragment of Adl2 DNA as probe. As mentioned above, this fragment assumes the right terminal position in most of the Adl2 DNA copies integrated in line T637. The results show that the junc- tion fragment (arrows in Figure 6) was at least partially sen- sitive to cleavage by endonuclease VII and that distinct lower mol. wt. fragments were generated in the process. Obviously, the order in which the two enzymes were allowed to react was of importance. The reaction was much more extensive when endonuclease VII reacted first (Figure 6, track 3), although the band created when PstI was allowed to cleave first (Figure 6, track 2), had the same size as one of the newly arising bands in track 3 (endonuclease VII followed by PstI). The reason for the dependence on the order of the two enzyme reactions might be sought in the cruciform configuration itself. If it was destroyed by PstI cleavage, endonuclease VII, which is thought to attack at the center of the cruciform structure, might then lack the appropriate substrate.

[13] 171w The results summarized here provided evidence for the presence of cruciform structures due to palindromes at the junction sites in cell line T637. As a note of caution, we hasten to add that sensitivity of DNA sequences to endonuclease VII does not prove the occurrence of cruciform structures since other peculiar DNA conformations might also serve as substrates for this enzyme. Fig. 6. DNA sequences which are homologous to Adl2 restriction fragment BamHIC are recognized by endonuclease VII, an enzyme which cuts cruciform DNA. Experimental details were described in the text and in the legends to Figures I -3. genomes integrated in the DNA of cell line T637. These palindromic sequences were less well protected from endogenous nuclease activities than internal viral DNA segments. As shown above, these sequences were also identical to those ex- cised in the process of generating revertants of cell line T637 by excision of integrated viral DNA. Thus, the designations a and b in Figures 2 and 5 refer to the same viral DNA se- quences.

[14] 72w Sensitivity of palindromic sequences to endonuclease VII digestion Bacteriophage T4 endonuclease VII is involved in packaging T4 DNA into phage heads (Kemper and Garabett, 1981). It specifically recognizes cruciform structures in DNA (Mizuuchi et al., 1982). The palindromic structures at the junction sites a and b described above in T637 DNA might, therefore, at least in part, be recognized by this endonuclease. Only partial sensitivity was to be expected as palindromic se-

[1] 77w Since it proved impossible to clone DNA sequences at the sites of linkage in T637 cells between viral and cellular or viral and viral DNA, we have determined the structures of these sites by more indirect ways. Fortunately, it was possible to correlate junction sites in cell line T637, which were un- protected and hence sensitive to autodigestion by endogenous nucleases, with the sites of junctions that had been lost in the morphological revertants of these cells.

[2] 194w Taken together, the data we report here imply an unusual sequence arrangement at certain viral-cell junctions or at junctions between adjacent viral DNA molecules in cell line T637. These sequences are accessible to endogenous nuclear nucleases and appear less well protected, probably by chromatin, than internal sequences of the integrated viral genomes. This lack of protection also explains that viral genomes are occasionally excised at these crucial sites, an event that generates morphological revertants of the cell line. The same sites may be characterized by the presence of in- verted repeats or palindromic sequences that can lead to the establishment of cruciform structures. It will be interesting to investigate to what extent the establishment of palindromic sequences generated as a consequence of the insertion event directs insertion and bestows upon it elements of specificity. In studying a large number of linkage sites in adenovirus- transformed cells and in adenovirus-induced tumors we have detected stemmed loop structures at the junction site in at least one cell line, CLAC3 (Deuring et al., 1981). Much fur- ther work will, however, be required before we can describe at a molecular level the events leading to the insertion and/or

[3] 7w excision of foreign DNA in mammalian cells.

[1] 30w Human adenovirus type 12 (Adl2), Huie strain, was grown on KB cells in monolayer or suspension cultures. Virions and viral DNA were purified as described elsewhere (Sutter et al., 1978).

[2] 111w Cell line BHK21 (subline B3) and its Adl2-transformed derivative T637 (Strohl et al., 1970) were propagated as described (Sutter et al., 1978). The revertant cell lines G12, TR12, TR1 1 and TR2 spontaneously arising from cell line T637, were characterized in detail previously (Groneberg et al., 1978; Eick et al., 1980). Revertant line G12 was isolated independently of all the other revertant lines. The revertant lines G12, TR12 and TR1 1 contained one intact, co-linearly arranged viral genome and part of a second one. Cell line TR2 car- ried about half a genome equivalent per diploid genome. The patterns of viral DNA integration in these revertants proved to be quite stable.

[3] 33w Labeling of viral DNA and of cloned viral DNA fragments Adl2 DNA and the specific cloned Adl2 DNA fragments (Vogel et al., 1981) were labeled with 32P by nick-translation (Rigby et al., 1977).

[4] 108w The procedures employed in this report were detailed previously (Sutter et al., 1978; Stabel et al., 1980; Eick et al., 1980): extraction of cellular DNA (Sutter et al., 1978), restriction enzyme cleavage, Southern blotting (Southern, 1975), DNA-DNA hybridization (Wahl et al., 1979), and autoradiography. Nuclei of T637 cells These were prepared after lysis of the cytoplasmic membranes by in- cubating the cells in 0.01 M Tris-HCl, pH 7.5, 0.02 M NaCl, 0.01 M MgCl2 and 0.507o of the non-ionic detergent Nonidet P40 (NP-40) for 5 min at 0°C. Subsequently, the nuclei were separated by low speed centrifugation and washed three times in the same buffer devoid of NP-40.

[5] 173w For autodigestion experiments, these nuclei were incubated in 0.01 M Tris- HCI, pH 7.5, 0.02 M NaCl, 0.01 M MgCl2 for 1 h at 37°C. The reaction was stopped by adding 1/10 of a volume of 10% SDS and 200 mg of proteinase K per ml of the reaction mixture and incubation at 37°C for 2 h. The DNA was then extracted by standard procedures (Sutter et al., 1978). Denaturation and renaturation of cellular DNA, treatment with SI nuclease DNA was denatured by heating the solution in a 100°C oil-bath for 5 min and subsequent quenching in ice. Renaturation of presumptive palindromic sequences was effected at room temperature for 15 min. Subsequently, DNA was ethanol precipitated, redissolved in SI buffer, 0.25 M NaCl, 0.03 M potassium acetate, pH 4.5, 0.001 M ZnSO4, 5% glycerol, and incubated with 1000 units of the nuclease S1 for 30 min at room temperature. The entire pro- cedure of denaturation, brief renaturation and SI treatment is sometimes referred to as denatured + SI (cf., Figures 3, 4, 5).

[6] 11w Treatment of DNA preparations with endonuclease VII from T4-infected Escherichia coli

[7] 47w Endonuclease VII was isolated from bacteriophage T4-infected E. coli cells as described elsewhere (Kemper and Garabett, 1981). DNA from T637 cells was incubated in 0.05 M Tris-HCI, pH 8.6, 0.01 M MgCl2, 0.01 M 2-mercaptoethanol, and 1 mg/ml of bovine serum albumin for 15 min at 370C.