Several lines of evidence indicate that eukaryotic DNA is arranged in highly supercoiled domains or loops, and that the repeating loops are constrained by attachment to a nuclear skeletal structure termed the nuclear matrix. Active genes are transcribed at the nuclear matrix and during replication the loops are reeled through fixed matrix-associated replication complexes. We have investigated whether the repair of DNA damage also occurs in the nuclear matrix compartment. Biochemical analysis of confluent normal human fibroblasts, ultraviolet (u.v.)-irradiated with 30 J m−2 and post-u. v. incubated in the presence of hydroxyurea, did not show any evidence for the occurrence of repair synthesis at the nuclear matrix either 30 min or 13 h after irradiation. Autoradiographic visualization of repair events in single DNA halo-matrix structures confirmed the biochemical observations. At a biologically more relevant dose of 5 J m−2 repair synthesis seems to initiate at the nuclear matrix, although only part of the total repair could be localized there. In u.v.-irradiated (30 J m−2) normal human fibroblasts post-u.v. incubated in the presence of hydroxyurea and arabinosylcytosine for 2h, multiple single-stranded regions are generated in a DNA loop as a result of the inhibition of the excision repair process. Different biochemical approaches revealed that most of the single-stranded regions are clustered, indicating that the repair process itself is non-random or that domains in the chromatin are repaired at different rates. Preferential repair of certain domains in the chromatin was shown to occur in xeroderma pigmentosum cells of complementation group C (XP-C). In XP-C cells these domains are localized near the attachment sites of DNA loops at the nuclear matrix. In contrast, xeroderma pigmentosum cells of complementation group D as well as Syrian hamster embryonic cells with limited excision-repair capacities, revealed a random distribution of repair events in DNA loops. The preferential repair of matrix-associated DNA in XP-C cells may be related partly to repair of transcriptionally active DNA and this may account for the ability of XP-C cells, in contrast to XP- D cells, to recover u.v.-inhibited synthesis of DNA and RNA.

Models for the spatial and temporal organization of DNA replication and segregation of the duplicated DNA during mitosis imply the existence of a scaffoldlike structure to which the DNA is attached (Comings, 1968; Dingman, 1974). Such a structure would provide binding sites for the origins of the tandemly arranged replicons.

Evidence for the binding of chromosomal DNA to a scaffold-like nuclear structure has been obtained from nuclei extracted with high concentrations of salt (Berezney & Coffey, 1977). This structure, termed nuclear matrix, appears to be a ubiquitous structure since it has been isolated from a wide variety of eukaryotic organisms. Different approaches indicate that the DNA is attached to the nuclear matrix via multiple supercoiled loops (Cook & Brazell, 1975; Mullenders et al. 1983a; Vogelstein et al. 1980) and that the anchorage points of the loops are organized non- randomly, transcriptionally active genes being in close proximity to the nuclear matrix (Mirkovitch & Laemmli, 1984; Small et al. 1985). Direct sequencing of matrix-associated DNA provides evidence for the involvement of specific sequences in the binding to the nuclear matrix (Goldberg et al. 1983).

The loop organization appears to be of relevance to DNA replication and transcription. A close relationship between average loop size and replicon size has been shown to exist (Buongiorno-Nardelli et al. 1982) and recent results suggest that the attachment sites of the loops at the nuclear matrix contain replication origins (Aelen et al. 1983). After initiation the DNA is reeled through fixed matrix- associated replication complexes (Dijkwel et al. 1979; Pardoll et al. 1980). Besides replication the nuclear matrix may also have a key function in transcription and RNA processing (van Eekelen et al. 1981). It is tempting to speculate that the association of regulatory elements with the matrix brings about functional compartmentalization of the nucleus.

Whether the matrix compartment also contains major binding sites for enzymes involved in repair synthesis is not known. We have investigated this question using similar methodology to that reported previously for the study of DNA replication (Dijkwel et al. 1979). Moreover, considering the non-random organization of DNA sequences in DNA loops, we have investigated whether different regions within DNA loops are repaired at different rates.

To study the extent of involvement of the nuclear matrix in DNA repair we have introduced damage into DNA by ultraviolet (u.v.) irradiation of the cells. For such a study u.v. irradiation as the source of DNA damage has a number of advantages. Since the distribution of u.v.-induced lesions within the genome is random (Rahn & Stafford, 1974), difficulties in interpretation of the data resulting from initial nonrandom distribution of damage are avoided. Moreover, the detection of excision repair sites by radioactive labelling is relatively simple and the effects of inhibitors on u.v.-induced repair synthesis have been investigated intensively. We have studied the localization of repair events in DNA loops by two approaches: a biochemical approach that involves progressively detaching DNA from the nuclear matrix with DNase I, and an autoradiographic approach that directly visualizes the distribution of repair events in DNA loops at the single cell level using the DNA halo-matrix technique (Vogelstein et al. 1980).

DNase digestion studies

In order to detect repair synthesis by radioactive labelling of u.v.-irradiated cells, it is necessary to reduce the incorporation of label by replicative synthesis to a sufficiently low level. In our studies this was achieved by using confluent human fibroblasts post-u.v. incubated in the presence of hydroxyurea (HU) or HU and arabinosylcytosine (araC). Under these conditions the label incorporated in unirradiated cells amounted to about 8% of the label incorporated in irradiated cells. However, since the rate of DNA synthesis after u.v. irradiation is reduced, the relative incorporation of label by replicative synthesis is expected to be even less in irradiated cells.

DNA attached to the nuclear matrix can be isolated as a rapidly sedimenting complex from nuclei, treated with 2M-NaCl (nuclear lysate) and subsequently centrifuged in neutral sucrose gradients (Wanka et al. 1977; Fig. 1A). These complexes contain almost all the nuclear DNA arranged in supercoiled loops (Mullenders et al. 1983a). The relative positions of repair events (and replication forks) in DNA loops can be examined by cleaving the DNA with increasing concentrations of DNase I, since the probability of a DNA fragment being released from a loop by breaks will decrease the closer it is situated to a region bound on the nuclear matrix. DNase digestion of nuclear lysates prepared from unirradiated exponentially growing or confluent cells pulse-labelled for 10 min (Fig. IB) indicates that nascent DNA is attached to the nuclear matrix by binding sites close to the replication forks (Dijkwel et al. 1979). The fact that the matrix is still enriched for nascent DNA after 30 or 60 min in the presence of inhibitors (Fig. 1C) is consistent with a reduction in DNA chain growth (Radford et al. 1982); under non-inhibitory conditions, the 1H/1C ratio decreases as the pulse length increases (Dijkwel et al. 1979). Since the attachment of replicating DNA to the nuclear matrix was also observed after u.v. irradiation, it is clear that the conditions used to reduce the replicative synthesis in favour of detection of repair synthesis do not alter the spatial organization of DNA replication at the nuclear matrix.

Fig. 1.

Association of replicating DNA with the nuclear matrix (from Mullenders et al. 19836). A. Exponentially growing 1C-prelabelled cells were pulse-labelled for 10 min with [1H]thymidine (25, μCiml−1) and nuclei were lysed in 2M-NaCl (nuclear lysate). Samples of the nuclear lysate were incubated with 0 (AJ, 2 (A2), 4 (A3) and 6 (A4) μg ml−1 DNase I and analysed in neutral sucrose gradients. Direction of centrifugation in this and other figures was from right to left. The numbers in the panels represent the ratios of the % of 1H and 1C ctsmin−1 in the DNA-nuclear matrix complex. B,C. Exponentially growing cells (○, ▵) or confluent cells (•) pulse-labelled with [1H]thymidine for 10 min without inhibitors (B) or 30 min in the presence of 10 mM-HU (C) or 60 min in the presence of 10 mM-HU and 0-1 mM-araC (▵; C). B and C, as well as Figs 3, 6, 8, show the 1H/1C ratio of the DNA—nuclear matrix complex versus the relative amount of prelabelled DNA remaining associated with the nuclear matrix.

Fig. 1.

Association of replicating DNA with the nuclear matrix (from Mullenders et al. 19836). A. Exponentially growing 1C-prelabelled cells were pulse-labelled for 10 min with [1H]thymidine (25, μCiml−1) and nuclei were lysed in 2M-NaCl (nuclear lysate). Samples of the nuclear lysate were incubated with 0 (AJ, 2 (A2), 4 (A3) and 6 (A4) μg ml−1 DNase I and analysed in neutral sucrose gradients. Direction of centrifugation in this and other figures was from right to left. The numbers in the panels represent the ratios of the % of 1H and 1C ctsmin−1 in the DNA-nuclear matrix complex. B,C. Exponentially growing cells (○, ▵) or confluent cells (•) pulse-labelled with [1H]thymidine for 10 min without inhibitors (B) or 30 min in the presence of 10 mM-HU (C) or 60 min in the presence of 10 mM-HU and 0-1 mM-araC (▵; C). B and C, as well as Figs 3, 6, 8, show the 1H/1C ratio of the DNA—nuclear matrix complex versus the relative amount of prelabelled DNA remaining associated with the nuclear matrix.

Measurement of repair synthesis was performed by short pulse-labelling of the cells at different times following u.v. irradiation at 30 J m−2. In the presence of HU (or HU and araC) no preferential association of repair-labelled DNA with the nuclear matrix during either early repair (directly after irradiation) or late repair (13 h after irradiation) was observed (Fig. 2). Although these data suggest that repair synthesis after 30 J m−2 does not proceed via repair enzymes associated with the nuclear matrix, two aspects of the methodology employed have to be considered: the length of the pulse and the presence of HU. It is important, for this conclusion, to point out that pulses of 4–10 min are suitable for detecting matrix-associated repair. The time required to repair u.v.-induced dimers is about 3 min without inhibitor and about 15 min in the presence of HU (Erixon & Ahnstrôm, 1979). Using the enzyme Balli nuclease we could show (see Fig. 6 and text below) that under our experimental conditions about 75 % of the repair patches made during a 10 min pulse, in the presence of HU, were not completed.

Fig. 2.

Distribution of repair-labelled DNA in DNA—nuclear matrix complexes after 30 J in−2 u.v. irradiation (from Mullenders et al. 1983b). Prelabelled confluent cells were u.v.-irradiated and pulse-labelled with [1H]thymidine (50 μCiml−1) in the presence of HU. Nuclear lysates were incubated with increasing concentrations of DNase I: A, 10 min pulse, directly after irradiation; B, 10 min pulse, directly after irradiation in the presence of HU (•) or HU and araC (▵); C, irradiation, 10min incubation, 10min pulse; D, irradiation, 20min incubation, 10min pulse; E, irradiation, 20min pulse; F, irradiation, 13 h incubation, 10 (•) or 20 (○) min pulse.

Fig. 2.

Distribution of repair-labelled DNA in DNA—nuclear matrix complexes after 30 J in−2 u.v. irradiation (from Mullenders et al. 1983b). Prelabelled confluent cells were u.v.-irradiated and pulse-labelled with [1H]thymidine (50 μCiml−1) in the presence of HU. Nuclear lysates were incubated with increasing concentrations of DNase I: A, 10 min pulse, directly after irradiation; B, 10 min pulse, directly after irradiation in the presence of HU (•) or HU and araC (▵); C, irradiation, 10min incubation, 10min pulse; D, irradiation, 20min incubation, 10min pulse; E, irradiation, 20min pulse; F, irradiation, 13 h incubation, 10 (•) or 20 (○) min pulse.

Fig. 6.

Analysis of the structure of repair-labelled DNA. DNA-nuclear matrix complexes were isolated from u.v.-irradiated (30 J m−2) cells post-u.v. incubated in the presence of [1H] thymidine and HU (A) or HU and araC (B) for 10 min or 2h. Samples of DNA-nuclear matrix preparations were digested with Bal31 nuclease and acid- precipitable radioactivity was determined; 10min incubation (○), 2h incubation (•).

Fig. 6.

Analysis of the structure of repair-labelled DNA. DNA-nuclear matrix complexes were isolated from u.v.-irradiated (30 J m−2) cells post-u.v. incubated in the presence of [1H] thymidine and HU (A) or HU and araC (B) for 10 min or 2h. Samples of DNA-nuclear matrix preparations were digested with Bal31 nuclease and acid- precipitable radioactivity was determined; 10min incubation (○), 2h incubation (•).

It is unlikely that HU, which delays repair, may cause incomplete repair patches to detach from the matrix. The ligation of repair patches that is going on in the presence of HU is incompatible with a model in which the total repair process, i.e. from incision to final sealing by ligase, is localized at the nuclear matrix. Also the results of the autoradiographic analysis (see below) performed without HU show no evidence for a repair process occurring at the nuclear matrix.

A number of studies (Miller & Chinault, 1982; Dresler & Lieberman, 1983) suggest that the types of DNA polymerase involved in the repair of damage, not only depend on the kind of damage but also on the dose of damaging agent. On the basis of incorporation of [1H]thymidine in the presence of inhibitors, Dresler & Lieberman (1983) proposed that after low levels of u.v. damage, repair is largely carried out by polymerase β, while at high doses of polymerase tris the enzyme involved. However, at both low and high u.v. doses, under conditions of polymerase α inhibition, polymerase β may be responsible for the remaining repair (Smith & Okumoto, 1984). Since the location of repair processes in the nucleus may depend on the particular enzymes involved, we have investigated the repair process after a dose of 5Jm−-2. Fig. 3 shows that when a short pulse was given directly after u.v. irradiation, repair-labelled DNA was enriched at the nuclear matrix. The preferential labelling of matrix-associated DNA was dependent on the length of the time between irradiation and pulse-labelling. Compared to a 10 min pulse given directly after irradiation a 10 min pulse given 2h after irradiation results in a less preferential labelling of matrix-associated DNA. An obvious explanation of the data could be that at 5 J m−2 a dynamic repair process is operating that is localized at the nuclear matrix. However, considering the average repair time for a dimer in the presence of HU to be 10 min, an only two- to threefold enrichment is inconsistent with repair of lesions at the matrix. This may imply that matrix-associated repair is not inhibited by HU, or is inhibited to a lesser extent, resulting in a much shorter repair time (i.e. 3 min; Erixon & Ahnstrom, 1979) and that it may represent a fraction of repair, insensitive to inhibitors (Smith & Okumoto, 1984; Mullenders et al. 1985). If one assumes that the time required to repair a dimer by matrix-associated repair enzymes does not vary during the post-u.v. period, the distribution of repair patches after a short pulse is also not expected to vary at different times after u.v. However, we observed a variation in the distribution of repair label, which may be due to several factors. First, the time required to repair a dimer may be dependent on the period of post- u.v. incubation. Second, there may be two processes operating simultaneously: a matrix-localized repair process, and another that is not restricted to this substructure and is inducible. The relative contribution of the presumed inducible process to the total repair will increase during recovery and may increase also with increasing u.v. dose. The occurrence of such a process would account for the variation in the distribution of repair events at 5 J m−2, and for the absence of any sign of a matrix- associated process at 30 J m−2. There are some experimental data that are in favour of the hypothesis. The generation of DNA breaks in the presence of HU and araC is maximal 30–60 min after u.v. irradiation and this maximal number of breaks is dose-dependent (Erixon & Ahnstrom, 1979). An alternative explanation is that the two- to threefold enrichment is not a reflection of a dynamic repair process operating at the matrix, but merely a reflection of preferential repair of matrix-associated DNA sequences.

Fig. 3.

Distribution of repair-labelled DNA in DNA-nuclear matrix complexes after 5Jm−2 u.v. irradiation. Confluent 1C-prelabelled human fibroblasts were u.v.- irradiated and directly pulse-labelled with [1H]thymidine (50 μCi ml−1) for 5 min (○) or for 10 min (•) or for 2h (▫) in the presence of HU. In another set of experiments cells were u.v.-irradiated and incubated for 30min (▿) or 2h (▾) before starting 10min pulse-labelling. Nuclear lysates were incubated with increasing concentrations of DNase I.

Fig. 3.

Distribution of repair-labelled DNA in DNA-nuclear matrix complexes after 5Jm−2 u.v. irradiation. Confluent 1C-prelabelled human fibroblasts were u.v.- irradiated and directly pulse-labelled with [1H]thymidine (50 μCi ml−1) for 5 min (○) or for 10 min (•) or for 2h (▫) in the presence of HU. In another set of experiments cells were u.v.-irradiated and incubated for 30min (▿) or 2h (▾) before starting 10min pulse-labelling. Nuclear lysates were incubated with increasing concentrations of DNase I.

Autoradiographic analysis of repair events in DNA loops

To investigate the distribution of repair events at the single cell level we employed the DNA halo-matrix method (Vogelstein et al. 1980). Cells growing on coverslips are extracted with non-ionic detergent and high salt concentrations, including ethidium bromide in the final step. After relaxation of the DNA, the nuclear matrix surrounded by a halo of DNA can be seen in the fluorescence microscope (Fig. 4A). It is important to note that, in contrast to the biochemical experiments, the autoradiographic analyses are performed with exponentially growing cells in the absence of inhibitors. In our preparations about 18 % of the grains in prelabelled cells were associated with the nuclear matrix. In S-phase cells about 75 % of the label was overlying the nuclear matrix region after 6 min labelling. Following 6 min pulse-labelling directly after u.v. irradiation (30 J m−2), grains were predominantly found in the halo region in non-S-phase cells (Fig. 4B). On average, about 33·9% and 23·8% of the grains were recovered from the nuclear matrix after 6min and 10 min pulse-labelling, respectively. Although an enrichment of grains at the matrix was observed after a very short pulse (4–6 min), we did not see a very distinct preferential labelling of the nuclear matrix region compatible with the initiation of repair at the nuclear matrix as observed in the initiation of DNA replication. Thus the autoradiographic analysis confirms the biochemical observations and suggests that at a dose of 30 J m−2 initiation of repair occurs in the DNA loops without requiring attachment of damaged sites to the nuclear matrix.

Fig. 4.

Localization of u.v.-induced repair patches in DNA halo-nuclear matrix structures. A. Fluorescence micrograph of a DNA halo-matrix structure, prepared from normal human cells. B. Autoradiogram of a DNA halo-matrix structure prepared from cells pulse-labelled for 6min directly after u.v. irradiation (30 J m−2).

Fig. 4.

Localization of u.v.-induced repair patches in DNA halo-nuclear matrix structures. A. Fluorescence micrograph of a DNA halo-matrix structure, prepared from normal human cells. B. Autoradiogram of a DNA halo-matrix structure prepared from cells pulse-labelled for 6min directly after u.v. irradiation (30 J m−2).

A similar approach has been used by McCready & Cook (1984) to investigate the initiation of u.v.-induced repair synthesis in HeLa cells. They reported a preferential labelling (2·5 or 5 min pulse) of matrix-associated DNA after high doses of u.v. light. However, as stated by the authors, the preferential labelling could be due to abortive replication or to repair of transcriptionally active DNA. At least in the experiments in which we pulse-labelled the u.v.-irradiated cells for 10 min, we could distinguish the S-phase cells by prelabelling cells with [1H]thymidine 2h prior to irradiation. In experiments with shorter pulses prelabelling was omitted because it increased the background of grains too much.

As the data suggest that at least at 30 J m−2 u.v.-induced repair synthesis does not occur at the nuclear matrix, we have investigated whether the repair occurs in a coordinated non-random way within single DNA loops, and whether domains in DNA loops are repaired at different rates. Evidence for non-random repair in normal human cells has been recently reported by Cohn & Lieberman (1984) and in hamster cells by Bohr et al. (1985).

Biochemical experiments were performed with confluent human cells post-u.v. incubated for 2h in the presence of HU or HU and araC. Nuclear lysates were incubated with nuclease Si and DNase I to elucidate the distribution of repair events in DNA loops. To visualize the repair events in DNA loops, DNA halo-matrix structures were prepared and processed for autoradiography.

Distribution of repair events in normal human cells

When DNA—nuclear matrix complexes were prepared from confluent human fibroblasts post-u.v. incubated for 2h in the presence of HU or HU and araC, and digested with DNase I, the 1H/1C ratio remained at about 1, suggesting a random distribution of repair events in the total population of DNA loops (Fig. 8A). Identical results were found in exponentially growing HeLa cells (Mullenders et al. 1986). An alternative way of investigating the distribution of repair events is to use nuclease Si. Using this enzyme, single-strand breaks accumulating in u.v.-irradiated cells in the presence of inhibitors can be converted to double-strand breaks and, consequently, when two or more breaks are present per DNA loop, DNA will be released from the matrix. Fig. 5 shows that in the presence of HU or HU and araC about 70% and 80%, respectively, of the DNA was released from the matrix by nuclease Si, representing 3·5 or 5 single-strand breaks per DNA loop, assuming a random distribution. Thus it is obvious that the frequency of breaks is not limited to one at a time in each DNA loop as proposed by Collins et al. (1984). As shown in Fig. 5, nuclease Si released repair-labelled [1H]DNA made in the presence of HU and araC to a greater extent than prelabelled DNA. This was also the case in the presence of HU, but only during the first 20min of post-u.v. incubation. The preferential release of repair-labelled DNA is related to the extent of ligation of repair patches, which was determined by digestion with Bal31 nuclease. Owing to its mode of action, labelled repair patches near single-stranded regions, i.e. unligated patches, are rendered acid-soluble much more rapidly than 1C-prelabelled DNA (Smith & Okumoto, 1984; Mullenders et al. 1985). To compare digestion of repair-labelled [1H]DNA to prelabelled [1C]DNA we have plotted the relative ratio as 1H and 1C radioactivity versus the time of incubation (Fig. 6). As shown in Fig. 6, 90 % of the repair patches made in the presence of HU and araC were unligated regardless of the duration of post-u.v. incubation, whereas in the presence of HU only during a 10 min pulse was the major part of the repair unligated.

Fig. 5.

Preferential release of repair-labelled DNA from the nuclear matrix by nuclease Si (from Mullenders et al. 1985). Confluent 1C-prelabelled cells were u.v.-irradiated (30 J m−2) and incubated in the presence of [1H]thymidine and HU and araC (A,B) or HU (C) for various periods of time. Nuclear lysates were treated with and without nuclease Si and analysed in sucrose gradients. A,C- Normal human fibroblasts. B. Xeroderma pigmentosum cells of complementation group A. 1C with (• – – – •) and without (•_____•) nuclease Si; 1H with (▵ – – –▵) and without (▵ —▵) nuclease Si. The relative amount of DNA at the matrix is plotted versus the time of post-u.v. incubation.

Fig. 5.

Preferential release of repair-labelled DNA from the nuclear matrix by nuclease Si (from Mullenders et al. 1985). Confluent 1C-prelabelled cells were u.v.-irradiated (30 J m−2) and incubated in the presence of [1H]thymidine and HU and araC (A,B) or HU (C) for various periods of time. Nuclear lysates were treated with and without nuclease Si and analysed in sucrose gradients. A,C- Normal human fibroblasts. B. Xeroderma pigmentosum cells of complementation group A. 1C with (• – – – •) and without (•_____•) nuclease Si; 1H with (▵ – – –▵) and without (▵ —▵) nuclease Si. The relative amount of DNA at the matrix is plotted versus the time of post-u.v. incubation.

Fig. 8.

Distribution of repair-labelled DNA in normal and in xeroderma pigmentosum cells (from Mullenders et al. 1986). 1C-prelabelled confluent cells were u.v. irradiated (30 J m−2) and incubated for 2h in the presence of [1H]thymidine and araC and/or HU. Samples of the nuclear lysates were treated with DNase I and analysed in sucrose gradients. A. Normal human cells. B. XP-C cells. XP21RO (○), XP6RO (•), XP1TE (V), XP8CA (▾). C. XP-D cells: XP3NE (▿), XP7BE (○).

Fig. 8.

Distribution of repair-labelled DNA in normal and in xeroderma pigmentosum cells (from Mullenders et al. 1986). 1C-prelabelled confluent cells were u.v. irradiated (30 J m−2) and incubated for 2h in the presence of [1H]thymidine and araC and/or HU. Samples of the nuclear lysates were treated with DNase I and analysed in sucrose gradients. A. Normal human cells. B. XP-C cells. XP21RO (○), XP6RO (•), XP1TE (V), XP8CA (▾). C. XP-D cells: XP3NE (▿), XP7BE (○).

Theoretically, a random distribution of repair patches within DNA loops is inconsistent with a preferential release of repair-labelled DNA from the matrix by nuclease S1. To obtain more insight into the localization of repair events, the size distribution of matrix-associated DNA, as well as loop-DNA obtained after nuclease Si digestion of nuclear lysates, was analysed in neutral sucrose gradients. As shown in Fig. 7, the 1C profile was displaced relative to the 1H profile in matrix-associated as well as loop-DNA. This is due to the different distribution of label: 1C label was uniformly distributed, while the 1H label was located at the end of molecules, owing to incomplete repair patches and cleavage by nuclease Si (Cleaver, 1983). It is obvious that the 1C profiles of matrix-associated and loop-DNA were very similar, which suggests that the incision events were distributed randomly along the DNA loops. Comparison of the 1H profiles reveals that although there was a considerable similarity, loop-DNA was composed of smaller repair-labelled DNA fragments than matrix-associated DNA. These data suggest that although the majority of incision events are distributed randomly in DNA loops, there must also exist clusters of incision events. The probability of detecting these clusters in the 1C profile in sucrose gradients will depend on the relative amount of DNA constituting the clusters. From the similarity in 1C profiles of matrix-associated and loop-DNA (Fig. 7), it is obvious that this amount must be small. However, owing to 1H-labelled repair patches at the end of the DNA fragments after nuclease Si digestion, small amounts of DNA will be visible in the 1H profiles. The appearance of clustered repair events may indicate that at 30 J m−2 two different repair systems are operating simultaneously: a processive system resulting in clustered repair events, and a system operating by random collisions. Alternatively, non-random distribution of repair events could occur when domains of DNA are repaired at different rates, as a result of variation in chromatin organization rendering some regions more accessible to repair enzymes than others.

Fig. 7.

Analysis of the size distribution of matrix-associated and loop-DNA after nuclease S, digestion. Nuclear lysates prepared from 1C-prelabelled u.v.-irradiated (30 J m−2) cells post-u.v. incubated for 2h in the presence of HU and araC and [1H]thymidine were digested with nuclease Si. Matrix-associated DNA was separated from nuclease Sj detached DNA by neutral sucrose gradient centrifugation. Both fractions were dialysed, incubated with proteinase K and again analysed in a neutral sucrose gradient. A. Matrix-associated DNA. C. Loop-associated DNA. B. 1C-labelled DNA: loop-DNA (▾), matrix-associated DNA (▪). D. 1H-labelled DNA: loop-DNA (▿), matrix-associated DNA (▫). The 1H/1C ratio of loop-DNA and matrix-associated DNA was 2·13 and 1·09, respectively. The matrix-associated DNA represented about 20 % of the total DNA.

Fig. 7.

Analysis of the size distribution of matrix-associated and loop-DNA after nuclease S, digestion. Nuclear lysates prepared from 1C-prelabelled u.v.-irradiated (30 J m−2) cells post-u.v. incubated for 2h in the presence of HU and araC and [1H]thymidine were digested with nuclease Si. Matrix-associated DNA was separated from nuclease Sj detached DNA by neutral sucrose gradient centrifugation. Both fractions were dialysed, incubated with proteinase K and again analysed in a neutral sucrose gradient. A. Matrix-associated DNA. C. Loop-associated DNA. B. 1C-labelled DNA: loop-DNA (▾), matrix-associated DNA (▪). D. 1H-labelled DNA: loop-DNA (▿), matrix-associated DNA (▫). The 1H/1C ratio of loop-DNA and matrix-associated DNA was 2·13 and 1·09, respectively. The matrix-associated DNA represented about 20 % of the total DNA.

Distribution of repair events in xeroderma pigmentosum cells

A recent study by Mansbridge & Hanawalt (1983) has indicated that the residual level of excision repair in xeroderma pigmentosum cells of complementation group C (XP-C) occurs in localized domains of the chromatin. Since nothing was known about the localization of these domains in the chromatin, we have investigated the distribution of repair events in DNA loops using biochemical and autoradiographic approaches.

DNase I digestion of DNA-nuclear matrix complexes prepared from four different XP-C cell lines post-u.v. incubated for 2h revealed that repair patches were preferentially situated near the attachment sites of DNA loops (Fig. 8B). In two xeroderma pigmentosum cell lines belonging to complementation group D (XP-D), which have a repair capacity comparable to XP-C (relative level of Unscheduled DNA Synthesis (UDS): 25% in XP-D and 18% in XP-C), no preferential localization of repair label at the nuclear matrix was observed (Fig. 8C). The enrichment of repair patches at the nuclear matrix in XP-C cells was independent of the duration of post-u.v. incubation (Mullenders et al. 1984), which suggests that the non-uniform distribution must be due to repair of particular DNA segments situated near the nuclear matrix.

Autoradiographic analysis of DNA halo-matrix structures was performed to check the biochemical observations by direct visualization of repair events in DNA loops. In XP-C cells post-u.v. incubated for 2h in the presence of [1H]thymidine, grains were preferentially overlying the nuclear matrix: 62·5 % of the grains were found in the matrix region compared to 18·9% in normal cells (Fig. 9). In XP-D cells the distribution of grains resembled the normal cells: 22·4 % of the grains were found in the nuclear matrix region. Since DNA halo-matrix structures are prepared from exponentially growing cells not treated with inhibitors, it is clear that the nonuniform repair in XP-C cells is not dependent on the presence of inhibitors or on physiological conditions, i.e. confluent or proliferating state. Nuclease S1 digestion of nuclear lysates prepared after 2h post-u.v. incubation in the presence of HU and araC, resulted in a detachment of 30% of the 1C-labelled DNA from the nuclear matrix compared to 80 % in normal human cells.

Fig. 9.

Autoradiograms of DNA halo—matrix structures prepared from normal human and XP-C cells. Cells were u.v. irradiated (30 J m−2) and incubated with [1H]thymidine. A. Normal human cells incubated for 10min. B. XP-C cells incubated for 2h.

Fig. 9.

Autoradiograms of DNA halo—matrix structures prepared from normal human and XP-C cells. Cells were u.v. irradiated (30 J m−2) and incubated with [1H]thymidine. A. Normal human cells incubated for 10min. B. XP-C cells incubated for 2h.

Sedimentation analysis of the size distribution of DNA fragments released from the nuclear matrix by nuclease Si revealed striking differences between XP-C and normal cells (Fig. 10). Owing to endlabelling of DNA fragments, the 1C profile is displaced relative to the 1H profile. In normal human cells the molecular weight (Mr) of nuclease-Si-released 1C-labelled DNA fragments was about 2×101. In XP-C cells the enzyme detached DNA fragments of about 101Mr, which is the average loop size in human fibroblasts (Mullenders et al. 1984, 1985). In contrast to lysates from normal human cells, the 1H profile of XP-C lysates showed a non-homogeneous distribution. The high molecular weight part is slightly shifted relative to the 1C profile, due to endlabelling. In addition the 1H profile shows a second peak consisting of low molecular weight DNA fragments. We have interpreted this non- homogeneous distribution of 1H-labelled DNA in XP-C cells as a consequence of preferential repair of DNA segments near the attachment sites of the DNA loops. The assumption is that in DNA regions near the nuclear matrix DNA repair events are very close to each other. Nuclease S1 will then release loop-sized 1C-labelled DNA as well as small 1H-labelled fragments. Comparison of the Mr of the small DNA fragments (about 5×101) and the interdimer distance of 30Jm-2 (2×101) suggests that repair events in the domains subject to repair do indeed occur very close to each other. Considering the accuracy of the Mr determinations at the very top of the gradient, the repair events may in fact be localized at the interdimer distance.

Fig. 10.

Size distribution of nuclease-Si-released DNA with and without proteolytic digestion (from Mullenders et al. 1986). 1C-prelabelled normal human cells as well as XP-C cells were u.v. irradiated (30 J m−2) and incubated for 10 min (normal cells) or 2h (XP-C cells) in the presence of HU and araC, and [1H]thymidine. Samples of nuclear lysates were incubated with nuclease Si, and finally with and without proteinase K. A. XP-C, nuclease Sp B. Normal cells, nuclease Si- C. XP-C cells, nuclease Si and proteinase K. D. Normal cells, nuclease S] and proteinase K. % cts min−1 of 1H and 1C at the sucrose cushion: A, 40·6% 1H, 52% 1C; B, 15·9% 1H, 19·2% 1C; C, 8·1 % 1H, 26·3% 1C, D, 1·0% 1H, 1·6% 1C.

Fig. 10.

Size distribution of nuclease-Si-released DNA with and without proteolytic digestion (from Mullenders et al. 1986). 1C-prelabelled normal human cells as well as XP-C cells were u.v. irradiated (30 J m−2) and incubated for 10 min (normal cells) or 2h (XP-C cells) in the presence of HU and araC, and [1H]thymidine. Samples of nuclear lysates were incubated with nuclease Si, and finally with and without proteinase K. A. XP-C, nuclease Sp B. Normal cells, nuclease Si- C. XP-C cells, nuclease Si and proteinase K. D. Normal cells, nuclease S] and proteinase K. % cts min−1 of 1H and 1C at the sucrose cushion: A, 40·6% 1H, 52% 1C; B, 15·9% 1H, 19·2% 1C; C, 8·1 % 1H, 26·3% 1C, D, 1·0% 1H, 1·6% 1C.

Since in XP-C cells about 70 % of the DNA was still attached to the nuclear matrix upon nuclease Si digestion (Fig. 10A), we were interested in the distribution of repair events in this DNA. One way to examine this is to analyse the size of this matrix-associated DNA. The matrix-associated DNA can be converted to free DNA molecules by proteolytic digestion of the nuclear matrix (Wanka et al. 1977). Thus, we made a comparison of 1H and 1C profiles of nuclease-Si-treated nuclear lysates with and without a final proteolytic digestion (Fig. 10). No differences were seen in normal cells post-u.v. incubated for 10 min in the presence of HU and araC (Fig. 10B,D). In XP-C cells two main differences were observed. First, even after proteolytic digestion about 25 % of the DNA was collected on the sucrose shelf and consisted of very long molecules. Second, 1H and 1C profiles shifted towards positions of higher molecular weight, concomitant with a reduction in the relative amount of small molecules. This sedimentation analysis suggests a rather complex distribution of repair events in XP-C cells. The DNA fragments initially released by nuclease S1 from the DNA-nuclear matrix complex, originate from loops with repair events at both attachment sites. This distribution occurs in about 30% of the loops. The DNA recovered from the sucrose shelf after the combined nuclease S1/proteolytic digestion obviously consisted of regions comprising multiple loops, which are excluded from the repair process during the first 2h after u.v. irradiation. The presence of DNA fragments of somewhat larger size than an average loop suggests that small clusters of loops with repair events at only one attachment site are also present. This DNA will not be detached from the matrix by nuclease Si, since single-stranded gaps at both attachment sites are required to release DNA. An alternative explanation could be that a subset of cells is not carrying out repair at all. However, autoradiographic analyses of DNA halo-matrix structures as well as data published by Mansbridge & Hanawalt (1983) revealed no such inhomogeneity of repair between cells.

The non-uniform distribution of repair events in XP-C raises the question of what the characteristics are of DNA that is subject to repair. Actively transcribed genes remain close to the nuclear matrix by attachment at the 5’ end of the genes (Small et al. 1985; Mirkovitch & Laemmli, 1984; Cook et al. 1982). From these anchorage points active genes are thought to extend into DNA loops. A three- to fourfold enrichment of repair events at the nuclear matrix in XP-C cells indicates that the domains subject to repair must extend beyond the attachment sites and may therefore cover the transcribed regions, at least partly (Mullenders et al. 1984). This hypothesis is supported by a study of Mayne & Lehmann (1982), who reported a partial recovery of u.v.-inhibited RNA synthesis in XP-C cells, but not in XP-D cells. They also showed that u.v.-inhibited DNA synthesis could recover appreciably in XP-C cells, but not in XP-D cells. Since inhibition of DNA replication after u.v. irradiation is due to inhibition of chain elongation, the matrix-associated replication forks may reside in the domains in XP-C cells subject to repair. Although the domains in XP-C cells are repaired to the same extent or even to a greater extent than the genome in normal cells (Mansbridge & Hanawalt, 1983; Mullenders et al. 1984) it is obvious that XP-C cells are much more sensitive to u.v. irradiation than normal cells. A number of reasons may account for this. Not all active regions may be covered by the domains subject to repair or be repaired to the same extent. Clusters of DNA loops that are not repaired at all may also include active regions. Furthermore, a large gene may extend beyond the domains, resulting in efficient repair of only parts of a gene. The factors that regulate the restricted repair in XP-C cells and factors that are necessary to repair damage outside the domains are not known. However, it seems unlikely that the repair of restricted areas in the genome is simply mediated by the accessibility of the chromatin. Exposure of confluent XP-C cells to sodium butyrate, which converts the chromatin into an ‘active’ configuration (Simpson, 1978), and stimulates the u.v.-induced repair synthesis twofold, does not influence the non-random distribution of repair events in XP-C cells (Mullenders et al. 1986).

Distribution of repair events in rodent cells

One of the striking properties of rodent cells is the high rate of survival after u.v. irradiation in spite of the very limited capacity to perform excision repair. Since recovery after u.v. irradiation seems to be directly related to repair capacity (Chan & Little, 1979; Simons, 1979), possible explanation for the apparent discrepancy may be the occurrence of preferential repair in domains covering active genes. Indeed, in established Chinese hamster ovary cells (CHO), Bohr et al. (1985) have shown that active DNA was repaired to a much greater extent than the total genome. On the other hand, Nairn et al. (1985) did not find evidence for preferential repair of active DNA in CHO cells. We have investigated the distribution of repair events in Syrian hamster primary cells. For biochemical experiments these cells have the advantage of growing to confluency. DNase I digestions of nuclear lysates prepared from cells post-u.v. incubated for 2h did not result in any evidence for non-random repair in Syrian hamster cells. This was confirmed by an alternative approach, in which the distribution of bacteriophage T4 endonuclease-sensitive sites was analysed. In contrast to XP-C cells, there was no indication of a non-random distribution of repaired sites (Mullenders et al. 1986).

Owing to increasing knowledge of the structure and organization of chromatin, it is possible to investigate the role of chromatin structure in cellular processes. The involvement of scaffold-like structures such as the nuclear matrix in DNA replication provides the possibility of functional compartmentalization of this process with respect to regulating factors as well as unwinding of parental DNA molecules. Excision repair induced by high doses of u.v. (30 J m−2) appears to proceed without participation of the nuclear matrix and this may provide an efficient and rapid way of restoring damaged DNA at any position in DNA loops. However, at low and biologically more relevant doses (5 J m−2) at least part of the total excision repair may initiate at the nuclear matrix. It is tempting to speculate that matrix-associated repair may be the important repair pathway at low u.v. doses and be related to that part of excision repair that is resistant to inhibitors of polymerase α. In fact we cannot rule out the possibility that the distribution of repair events at high doses may be due partly to abnormal repair activity.

Comparison of recovery of RNA synthesis and removal of lesions after u.v. irradiation suggests the existence of a process of preferential repair of damage in transcriptionally active chromatin. From the spatial distribution of repair events in DNA loops there is evidence for preferential repair of damage in transcriptionally active DNA in xeroderma pigmentosum cells of complementation group C, consistent with the appreciable recovery of RNA synthesis in these cells. However, using this methodology, there is no evidence for preferential repair of transcriptionally active chromatin in normal human and hamster primary cells.

We thank A. C. van Kesteren and C. J. M. Bussmann for excellent technical assistance.

This work was supported by the Association of the University of Leiden with Euratom, contract no. BIO-E-407-81-NL.

Aelen
,
J. M. A.
,
Opstelten
,
R. J. G.
&
Wanka
,
F.
(
1983
).
Organization of DNA replication in Physarum polycephalum. Attachment of the origins of replicons and replication forks to the unclear matrix
.
Nucl. Acids Res.
11
,
1181
1195
.
Berezney
,
R.
&
Coffey
,
D. S.
(
1977
).
Nuclear matrix. Isolation and characterization of a framework structure from rat liver nuclei
.
J. Cell Biol.
73
,
616
637
.
Bohr
,
V. A.
,
Smith
,
C. A.
,
Okumoto
,
D. S.
&
Hanawalt
,
P. C.
(
1985
).
DNA repair in an active gene: removal of pyrimidine dimers from the DHFR gene is much more efficient than in the genome overall
.
Cell
40
,
359
369
.
Buongiorno-Nardelli
,
M.
,
Micheli
,
G.
,
Carri
,
M. T.
&
Marilly
,
M.
(
1982
).
A relation between replicon size and supercoiled loop domains in the eukaryotic genome
.
Nature, Land.
298
,
100
102
.
Chan
,
G. L.
&
Little
,
J. B.
(
1979
).
Resistance of plateau phase cell, human normal, and xeroderma pigmentosum fibroblasts to the cytotoxic effects of ultraviolet light
.
Mutât. Res.
63
,
401
412
.
Cleaver
,
J. E.
(
1983
).
Structure of repaired sites in human DNA synthesized in the presence of inhibitors of DNA polymerases alpha and beta in human fibroblasts
.
Biochim. biophys. Acta
739
,
301
311
.
Cohn
,
S. M.
&
Lieberman
,
M. W.
(
1984
).
The distribution of DNA excision repair sites in human diploid fibroblasts following ultraviolet irradiation., biol. Chem.
259
,
12463
12469
.
Collins
,
A. R. S.
,
Downes
,
C. S.
&
Johnson
,
R. T.
(
1984
).
An integrated view of inhibited repair. InDNA Repair and its Inhibition
(ed.
A. R. S.
Collins
, C. S. Downes &
R. T.
Johnson
).
Nucl. Acids Symp. Series, no
.
13
, pp.
1
-
11
. Oxford, Washington, DC: IRL Press.
Comings
,
D. E.
(
1968
).
The rationale for an ordered arrangement of chromatin in the interphase nucleus
.
Am. J. hum. Genet.
20
,
550
560
.
Cook
,
P. R.
&
Brazell
,
I. A.
(
1975
).
Supercoils in human DNA
.
J. Cell Sci.
19
,
261
279
.
Cook
,
P. R.
,
Lang
,
J.
,
Hayday
,
A.
,
Lania
,
L.
,
Fried
,
M.
,
Chiswell
,
D.
&
Wyke
,
J. A.
(
1982
).
Active viral genes in transformed cells lie close to the nuclear cage
.
EMBOJ.
1
,
447
452
.
Dijkwel
,
P. A.
,
Mullenders
,
L. H. F.
&
Wanka
,
F.
(
1979
).
Analysis of the attachment of replicating DNA to a nuclear matrix in mammalian interphase nuclei
.
Nucl. Acids Res.
6
,
219
230
.
Dingman
,
C. W.
(
1974
).
Bidirectional chromosome replication: some topological considerations
.
J. theor. Biol.
43
,
187
195
.
Dresler
,
S. L.
&
Lieberman
,
M. W.
(
1983
).
Identification of DNA polymerases involved in DNA excision repair in diploid human fibroblasts. Ji biol. Chem.
258
,
9990
9994
.
Erixon
,
K.
&
AhnstrOm
,
G.
(
1979
).
Single-strand breaks in DNA during repair of UV-induced damage in normal human and xeroderma pigmentosum cells as determined by alkaline unwinding and hydroxylapatite chromatography
.
Mutât. Res.
59
,
257
271
.
Goldberg
,
G. L
,
Collier
,
I.
&
Cassel
,
A.
(
1983
).
Specific DNA sequences associated with the nuclear matrix in synchronized mouse 3T3 cells
.
Proc. natn. Acad. Sci. U.S A.
80
,
6887
—6891.
Mansbridge
,
J. N.
&
Hanawalt
,
P. C.
(
1983
). Domain-limited repair of DNA in ultraviolet irradiated fibroblasts from xeroderma pigmentosum complementation group C. In
Cellular Responses to DNA Damage
(ed. E. C. Friedberg &
B. R.
Bridges
), pp.
195
207
.
New York
:
Alan Liss
.
Mayne
,
L. V.
&
Lehmann
,
A. R.
(
1982
).
Failure of RNA synthesis to recover after UV irradiation: An early defect in cells from individuals with Cockayne’s syndrome and xeroderma pigmentosum
.
Cancer Res.
42
,
1473
1478
.
McCready
,
S. J.
&
Cook
,
P. R.
(
1984
).
Lesions induced in DNA by ultraviolet light are repaired at the nuclear cage. J’
.
Cell Sci.
39
,
53
62
.
Miller
,
M. R.
&
Chinault
,
D. N.
(
1982
).
The roles of DNA polymerases a and f in DNA repair synthesis induced in hamster and human cells by different DNA damaging agents
.
J. biol. Chem.
257
,
10204
10209
.
Mirkovitch
,
J.
&
Laemmli
,
U. K.
(
1984
).
Organization of the higher-order chromatin loop: Specific attachment sites on nuclear scaffold
.
Cell
39
,
233
232
.
Mullenders
,
L. H. F.
,
van Kesteren
,
A. C.
,
Bussmann
,
C. J. M.
, van
Zeeland
,
A. A.
&
Natarajan
,
A. T.
(
1984
).
Preferential repair of nuclear matrix associated DNA in xeroderma pigmentosum complementation group C
.
Mutât. Res.
141
,
75
82
.
Mullenders
,
L. H. F.
,
van Kesteren
,
A. C.
,
Bussmann
,
C. J. M.
, van
Zeeland
,
A. A.
&
Natarajan
,
A. T.
(
1986
).
Distribution of UV-induced repair events in higher-order chromatin loops in human and hamster fibroblasts
.
Carcinogenesis
7
,
995
1002
.
Mullenders
,
L. H. F.
,
van Kesteren
,
A. C.
, van
Zeeland
,
A. A.
&
Natarajan
,
A. T.
(
1985
).
Analysis of the structure and spatial distribution of ultraviolet-induced DNA repair patches in human cells made in the presence of inhibitors of replicative synthesis
.
Biochim. biophys. Acta
826
,
38
48
.
Mullenders
,
L. H. F.
, van
Zeeland
,
A. A.
&
Natarajan
,
A. T.
(
1983a
).
Comparison of DNA loop size and supercoiled domain size in human cells
.
Mutât. Res.
112
,
245
252
.
Mullenders
,
L. H. F.
, van
Zeeland
,
A. A.
&
Natarajan
,
A. T.
(
1983b
).
Analysis of the distribution of DNA repair patches in the DNA-nuclear matrix complex from human cells
.
Biochim. biophys. Acta
740
,
428
435
.
Nairn
,
S. R.
,
Mitchell
,
D. L.
&
Clarkson
,
J. M.
(
1985
).
Role of chromatin structure in the repair of DNA photoadducts in mammalian cells
.
Int.J. Radiat. Biol.
47
,
181
189
.
Pardoll
,
D. M.
,
Vogelstein
,
B.
&
Coffey
,
D. S.
(
1980
).
A fixed site of DNA replication in eukaryotic cells
.
Cell
19
,
527
536
.
Radford
,
I. A.
,
Martin
,
R. F.
&
Finch
,
L. R.
(
1982
).
Effects of hydroxyurea on DNA synthesis in mouse L-cells
.
Biochim. biophys. Acta
696
,
145
153
.
Rahn
,
R. O.
&
Stafford
,
R. S.
(
1974
).
Measurements of defects in ultraviolet-irradiated DNA by the kinetic formaldehyde method
.
Nature, Lond.
248
,
52
54
.
Small
,
D.
,
Nelkin
,
B.
&
Vogelstein
,
B.
(
1985
).
The association of transcribed genes with the nuclear matrix of Drosophila cells during heat shock
.
Nucl. Acids Res.
13
,
2413
—2431.
Simons
,
J. W. I. M.
(
1979
).
Development of a liquid-holding technique for the study of DNA- repair in human diploid fibroblasts
.
Mutât. Res.
59
,
273
283
.
Simpson
,
R. T.
(
1978
).
Structure of chromatin containing extensively acetylated H3 and H4
.
Cell
13
,
691
699
.
Smith
,
C. A.
&
Okumoto
,
D. S.
(
1984
).
Nature of DNA repair synthesis resistant to inhibitors of polymerase alpha in human cells
.
Biochemistry
23
,
1383
1391
.
van
Eekelen
,
E. A. G.
&
VAN Venrooy
,
W. J.
(
1981
).
HnRNA and its attachment to a nuclear protein matrix
.
J. Cell Biol.
88
,
554
562
.
Vogelstein
,
B.
,
Pardoll
,
D. M.
&
Coffey
,
D. S.
(
1980
).
Supercoiled loops and eukaryotic DNA replication
.
Cell
22
,
79
85
.
Wanka
,
F.
,
Mullenders
,
L. H. F.
,
Bekers
,
A. G. M.
,
Pennings
,
L. J.
,
Aelen
,
J. M. A.
&
Eygensteyn
,
J.
(
1977
).
Association of nuclear DNA with a rapidly sedimenting structure
.
Biochem. biophys. Res. Commun.
74
,
739
747
.