| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Biology and Genetics |
Sir Alastair Currie Cancer Research UK Laboratories and Department of Pathology, University of Edinburgh, Edinburgh EH4 2XU, Scotland, United Kingdom
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Mutations in the DNA ligase I gene have been identified in a patient with similar growth retardation, immunodeficiency and photosensitivity as found in BS (4) . A Glu-566 to Lys mutation at the active site inactivated one allele, whereas an Arg-771 to Trp mutation on the second allele resulted in 20-fold reduced enzyme activity (5) . The 46BR cell line, isolated from this patient, showed increased sensitivity to DNA-damaging agents, delayed joining of Okazaki fragments, and elevated sister chromatid exchange, but not to the level seen in BS (6, 7, 8, 9) .
DNA ligase I is involved at the replication fork in the joining of Okazaki fragments during lagging strand DNA replication (10) and is considered the main replicative ligase. A correlation between DNA ligase I enzyme activity and the rate of cell proliferation supports this observation (11) . DNA ligase I has also been implicated in NER (12 , 13) and acts in the long-patch form of base excision repair (reviewed in Ref. 14 ). DNA ligase I has been reported to be an essential gene for mammalian cells growing in vitro (15) .
We knocked out the mouse Lig1 gene to study the function of DNA ligase I and as the first stage in the production of a mouse model for the 46BR patient (16) . As expected, the DNA ligase I mutation was an embryonic lethal, but mutant embryos developed much further than anticipated; they were indistinguishable from their siblings at embryonic development day 10.5 but then became progressively more anemic and died by embryonic development day 16.5. From embryonic development day 11.5 onwards, the liver takes over from the yolk sac as the major site of hematopoiesis. Mutant embryos failed to make this transition, with livers from mutant embryos lacking erythropoietic islands, leading to a severe deficiency of mature enucleate erythrocytes in the circulation. We found that hematopoietic progenitors were present in mutant fetal liver and that they were able to repopulate the hematopoietic system and rescue lethally irradiated recipients, which showed that there was not a qualitative hematopoietic block in Lig1 null cells, but rather that there was a quantitative impediment. We think that other tissues can compensate for the lack of DNA ligase I, presumably through the action of another ligase, but that this compensation fails under the high replicative pressure of fetal erythropoiesis, leading to reduced proliferation, anemia, and death. Using Lig1 null fibroblast cell lines, we found no evidence of any DNA repair deficit associated with DNA ligase I deficiency, but there was a delayed joining of DNA replication intermediates and decreased genome stability (17) . Our work contradicts the suggestion (15) that Lig1 is essential for cell viability.
We have now carried out a further round of gene targeting to introduce the Arg-771-to-Trp mutation found in the 46BR patient into the mouse Lig1 gene. In this report, we describe the production of mice homozygous for the 46BR mutation and use the animals to study the relationship between DNA replication failure, genome instability, and cancer susceptibility.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Two hundred µg of 46BR targeting vector DNA was restricted with EcoRI and electroporated as described (19) into 3 x 107 HM-1 ES cells heterozygous for the Lig1-(no.12) knockout allele (17) . Selection in 6-thioguanine was carried out as described previously (20) , and resistant clones were picked and screened for the 46BR mutation. This was achieved by AciI restriction of a PCR product for exon 23 with primers M5175, Forward, CGCGGATCCGCTGAAGAAGGACTACCTTGA (located at the 5' end of exon 23 with an additional synthetic BamHI restriction site underlined) and M5176, Reverse, CCGGAATTCCGGCTTGCATATAGCCTGAAGCTCTTC (located at the 3' end of exon 23 with an additional synthetic EcoRI restriction site underlined). PCR was carried out as described (21) , in 35 cycles with cycle conditions as follows: 94°C for 1 min, 63°C for 30 s, and 72°C for 30 s.
Clones heterozygous for the 46BR mutation were used for blastocyst injection and germ-line transmission was obtained as described (19) . Germ-line chimeras were outbred with BALB/c and resulting 46BR homozygous lines and a wild-type control line were all maintained on a segregating 129/OlaXBALB/c background. Primary fibroblasts were isolated from mouse embryos on embryonic development days 10.5–13.5 and cultured as described previously (22) .
Hematopoietic Cell Culture.
Single-cell suspensions were obtained from bone marrow, spleen, and thymus by passing tissue progressively through finer needles down to 25-gauge. Cells were plated at 106/ml in 60-mm dishes in RPMI, supplemented with 10% FCS, penicillin, and streptomycin. Proliferating spleenocyte cultures were also supplemented with lipopolysaccharides (25 µg/ml) and interleukin 4 (5 ng/ml). One h after plating, bone marrow and thymocyte cultures were irradiated through the medium with UV (254 nm; 10 Jm-2 for bone marrow, 20 Jm-2 for thymocytes). Proliferating spleenocyte cultures were treated with hydrogen peroxide (600 µM) in medium at 37°C for 30 min before replating into growth medium.
Flow Cytometry.
Cell suspensions from hematopoietic tissues were prepared as described previously (16)
, counted electronically (Beckman Coulter), and diluted to 107/ml in PBS. Cells (106) were stained with the desired fluorochrome-linked antibody and washed twice to remove unbound antibody; 25,000–50,000 cells were acquired using a Becton Dickinson FACScan, and the resulting data were analyzed using Lysis II software. The antisera used were obtained from various suppliers: from Caltag Ltd., antimouse IgM F(ab')2 PE-conjugated affinity-purified goat IgG, antimouse CD4 PE-conjugated rat monoclonal IgG2a (clone CT-CD4), antimouse CD8a FITC-conjugated rat monoclonal IgG2a (clone CT-CD8a), antimouse B220 PE-conjugated rat monoclonal IgG2a (clone RA3-6B2), and antimouse GR1 FITC-conjugated rat monoclonal IgG2b (clone RB6-8C5); from PharMingen, antimouse CD45 PE-conjugated rat monoclonal IgG2b (clone 30F 11.1), antimouse TCR
ß FITC-conjugated hamster monoclonal IgG (clone H57-597), and antimouse Ter-119 FITC-conjugated rat monoclonal IgG2b (clone Ter-119); and from Sigma-Aldrich Company Ltd., antimouse CD3 FITC-conjugated rat monoclonal IgG (clone 53–7.3).
Propidium iodide-stained nuclei (20,000) were assayed on a Coulter EPICS XL to determine DNA content as described previously (22) . Apoptotic cells were detected using the TACS Annexin V-FITC apoptosis detection kit using the manufacturer’s protocol (R&D Systems).
Detection of Micronuclei.
Single-cell suspensions prepared from spleen and thymus were washed twice in PBS and fixed in absolute ethanol at -20°C. Fixed cells were stained with 4,6-diamidino-2-phenylindole (10 µg/ml). Fifteen µl of fixed cell suspension (
106 cells/ml) were mounted on a slide, a coverslip added, and micronuclei were scored under a fluorescence microscope.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The structures of the 3' end of the wild-type Lig1 allele, the knockout allele, the 46BR-targeting vector, and the 46BR allele are shown schematically in Fig. 1
. Homologous recombination between the knockout allele and the targeting vector will result in the incorporation of the 46BR mutation into the Lig1 locus, with selection for targeting events being provided by loss of the Hprt minigene and resistance to 6-thioguanine. Resistant clones were picked and 11 of the 38 that were screened contained the 46BR mutation. This was determined by AciI restriction of a PCR product for exon 23, the site of the 46BR mutation. The AciI recognition site (GCGG) coincides with the position of the 46BR mutation (GTGG) so that the Lig146BR allele is not cleaved. Seven clones were analyzed further by Southern blotting with a Lig1 cDNA probe (16)
and found to be correctly targeted (data not shown). All of the clones had lost hybridizing restriction fragments diagnostic for the knockout allele and contained only fragments expected from the wild-type allele, with no evidence of randomly integrated vector sequences. The seven clones heterozygous for the 46BR mutation were used for blastocyst injection, and germ-line transmission was obtained from three. Founders were used to establish two independent lines homozygous for the 46BR mutation. The AciI polymorphism in exon 23 was used to genotype mice (see Fig. 1
). Automated DNA sequencing of RT-PCR products was used as described previously (17)
to confirm that the 46BR mutation was the only change present in the part of the Lig1 coding region (exons 22–27) involved in the gene-targeting manipulations.
|
for males) and a 20% difference in body weight opened up and was maintained for up to 6 months, by which time the weights of wild-type animals had plateaued. When older animals (350–500 days) were sampled, it was found that the weights of 46BR mice had reached wild-type levels. These data are consistent with the DNA ligation defect associated with the 46BR mutation delaying DNA replication and slowing growth.
|
ß (present on thymocytes with rearrangement of the TCR gene); CD3 (present on T cells and a subset of B cells); CD45 (present on nonerythroid hematopoietic cells; GR1 (present on granulocytes); and Ter-119 (present on erythroid cells). The staining profiles of 46BR mice were indistinguishable from wild-type controls (data not shown). We conclude that there is no evidence in older 46BR mice for the severe quantitative hematopoietic defect that we have seen in Lig1 null embryos (16)
and in lethally irradiated mice rescued by Lig1 null fetal liver cells (17)
.
Erythropoiesis Struggles to Meet Demand in Young 46BR Mice.
A very different picture emerged when the spleens of 3-week-old 46BR mice were examined. Spleens were 2.5-fold larger than in age-matched controls (46BR average spleen weight, 98 ± 15 mg; wild type, 39 ± 12 mg; n = 4 for each genotype) and this difference was highly significant (P = 0.0009 by Student’s t test; compare Fig. 3, A and B
). Histology indicated that the increased size was caused by an expansion of the red pulp, with the hypercellularity resulting from massive extramedullary hematopoiesis (see Fig. 3D
).
|
). This difference is illustrated in Fig. 4
for wild-type and 46BR littermates sampled on consecutive days. In the wild-type spleens, the frequency of S-phase cells varied from 4 to 9% (Fig. 4, A, C, and E)
. In the 18-day-old 46BR spleen, 25% of the cells were in S phase (Fig. 4B)
. This high frequency of S-phase cells was maintained in a 19-day-old littermate, but there was now also a prominent and discrete sub-G1 peak comprising 21% of the cells (Fig. 4D)
. This discrete sub-G1 peak was seen in 4% (3 of 73) of 46BR spleens examined but was not observed in wild-type spleens (0 of 45 examined). The explanation for this sub-G1 peak will be considered later. In the 20-day-old 46BR littermate, the frequency of S-phase cells was 16%, and no sub-G1 peak was apparent (Fig. 4F)
.
|
|
. However, the reticulocyte frequency was significantly elevated in age-matched 46BR mice (Fig. 3F)
.
From 5 weeks of age, spleens from 46BR mice were no longer disproportionately large, and the cell cycle distributions in the majority of animals examined had returned to normal levels, with the frequency of S-phase cells not being significantly different from wild type (Table 1)
. Thus, erythropoiesis seems unable to meet the very high demand for red cells in young mice when other hematopoietic lineages are also being established, but the situation quickly resolves. These data are consistent with the ligation defect associated with the 46BR mutation affecting DNA replication and resemble, albeit in a less extreme way, the erythropoietic crisis observed in Lig1 null embryos (16)
and lethally irradiated mice rescued by Lig1 null fetal liver cells (17)
.
Occasional DNA Replication Failure in Bone Marrow from 46BR Mice.
Bone marrow and thymus from 3-week-old 46BR mice did not show the same marked increase in the frequency of S-phase cells that was observed in spleen. However, a slight, but significant, increase in the frequency of S-phase cells was found when flow cytometry profiles from 3–14-week-old 46BR bone marrows were compared with wild-type controls (Table 1)
. Although the large majority of 46BR bone marrows examined had profiles like the wild-type control sample shown in Fig. 5A
, 7% (3 of 45 examined) had an abnormal profile like the example in Fig. 5B
. In these abnormal bone marrows, the frequency of S-phase cells was 2-fold less than in normal 46BR samples (Table 1)
. This difference between abnormal and normal 46BR bone marrow was significant (P = 2.6 x 10-7), as was the lower frequency of S-phase cells in abnormal 46BR bone marrow compared with wild-type controls (see Table 1
). In the abnormal 46BR bone marrows, there was no longer a discrete G2 peak corresponding to cells that had completed S phase. In the abnormal 46BR sample shown (Fig. 5B)
, only 2% of the cells are in G2 compared with 4% of the wild type (Fig. 5A)
and a smaller fraction of the population (8%) is in S phase compared with the wild type (14%). Whereas in the wild-type and normal 46BR samples, the frequency of cells at different stages of S phase was fairly constant, in the abnormal 46BR bone marrows, the frequency of cells in the later stages of replication (with higher DNA content) dropped sharply, indicating that a substantial population of cells were not completing a round of DNA replication. These abnormal distributions of S-phase and G2 cells were not consistently accompanied by the presence of a sub-G1 population. Similar abnormal distributions were not observed in wild-type bone marrows.
|
revealed a classic G1 arrest with very few cells in S phase (4%) and G2 (1%) and with a large accompanying broad sub-G1 peak (16%). Annexin V and propidium iodide labeling of live cells, sampled at the same time as the propidium iodide-stained nuclear preparations were made, confirmed that the broad sub-G1 peak correlated well with the presence of a discrete apoptotic population (23%) of Annexin V and propidium iodide double-labeled cells (Fig. 5D)
. Similar staining of normal 46BR and wild-type bone marrows did not reveal an increased level of apoptosis in 46BR samples (data not shown). The apoptosis assay has not been possible on abnormal 46BR bone marrows because of the low frequency of occurrence.
The Sub-G1 Peak in Thymus from 46BR Mice Is Not Caused by Apoptosis.
The majority of flow cytometry profiles from 3–14-week-old 46BR thymuses were normal, but there was a small, but significant, increase in the frequency of S-phase cells compared with wild-type controls (Table 1)
. However, 27% (19 of 70) of 46BR thymuses examined had abnormal profiles with a pronounced discrete sub-G1 peak (Fig. 5G)
. In the example shown, 30% of the population is in the sub-G1 peak. The frequency of S-phase cells in these abnormal 46BR thymuses was the same as in normal 46BR thymuses (Table 1)
. In abnormal 46BR thymuses with the most pronounced sub-G1 peaks, there was also a small peak corresponding to double the DNA content of the sub-G1 peak (Fig. 5G
, arrow). These discrete sub-G1 peaks were not observed in wild-type thymuses (0 of 39 examined).
These data are consistent with the ligation defect associated with the 46BR mutation having a detrimental effect on DNA replication. In some bone marrows, the effect is severe, leading to a failure of normal completion of DNA replication. The obvious explanation for the discrete sub-G1 peak in 46BR thymuses, namely, that it was caused by apoptosis triggered by the replication defect, was ruled out by Annexin V and propidium iodide double-labeling of live cells. The frequency of double-labeled apoptotic cells was the same (only 3%) in wild-type (Fig. 5F)
and abnormal 46BR thymus (Fig. 5H)
, despite the presence in the latter of 30% of the population in a discrete sub-G1 peak.
The Sub-G1 Peak Can Be Induced in Wild-Type Hematopoietic Cell Cultures by DNA-damaging Agents.
We have shown that the discrete sub-G1 peak that occurs spontaneously in 46BR thymus is not caused by apoptosis. To understand how it could arise we sought to induce a similar peak in wild-type hematopoietic cell cultures. Wild-type thymocyte cultures were established and either UV-irradiated at 20 Jm-2 or left untreated. Eighteen h later, control cultures showed a broad sub-G1 peak (9% of the population in Fig. 6A
), that correlated well with the frequency of apoptotic cells (14%; Fig. 6B
). As expected, the frequency of apoptotic cells was increased in cultures 18 h after UV-irradiation (20%, Fig. 6D
; 25%, Fig. 6F
). Some irradiated cultures (Fig. 6C)
showed the same broad sub-G1 peak as control cultures, whereas others (Fig. 6E)
displayed, in addition, a discrete sub-G1 peak reminiscent of that seen in abnormal 46BR thymuses in vivo. The equivalent levels of apoptosis in the two types of irradiated culture indicated that it was the broad, rather than the discrete, sub-G1 peak that contained apoptotic cells. The effect of a second DNA-damaging agent, hydrogen peroxide, was studied in proliferating wild-type spleenocyte cultures. Control spleenocyte cultures had a higher frequency of S-phase cells and of the broad sub-G1 peak, diagnostic for apoptosis, than unstimulated control thymocytes (spleenocytes, Fig. 6G
: sub-G1, 17%; S phase, 9%; thymocytes, Fig. 6A
: sub-G1, 9%; S phase, 1%). Treatment with hydrogen peroxide (600 µM for 30 min) did not affect the frequency of S-phase cells, but 24 h after treatment, the frequency of cells in the sub-G1 peak increased, and a discrete peak superimposed on the broad sub-G1 peak of apoptotic cells was visible. This discrete sub-G1 peak became more pronounced 2 days after treatment (Fig. 6H)
. Thus, the characteristic discrete sub-G1 peak, seen spontaneously in 46BR spleens and thymuses in vivo, can be reproduced by treatment of wild-type spleen and thymus cultures in vitro with DNA-damaging agents.
|
). Primary embryonic fibroblasts grow very slowly in culture, and we reasoned that the rate of DNA replication may not be sufficient for the delayed joining of Okazaki fragments by the 46BR protein to be manifested as chromosome breakage or loss (i.e., as micronuclei). For this reason we established an in vivo assay for micronuclei on spleen and thymus. We found that the frequency of micronuclei was higher in wild-type spleen than in wild-type thymus, and that the frequency was higher for both tissues in animals that were 15 months old compared with 7 weeks old. The frequency of micronuclei in 46BR thymus of both ages was 1.5-fold higher than in control tissue, but this difference was not significant at the 5% level. The genotype difference was more pronounced in the spleen, in which micronuclei were twice as frequent in 7-week-old animals and 3-fold more common in 15-month-old animals, with both differences being highly significant.
|
). Five times as many tumors occurred in 46BR mice (17 of 53 animals), corresponding to a 2-year tumor-free survival rate of 68%. In addition to the expected lymphomas and adenocarcinomas, a range of other epithelial tumors, rarely seen in mice, occurred in the 46BR stock (These are listed in the legend to Fig. 7
). Most striking were the two superficial, apparently cutaneously based, tumors with histological features of adnexal differentiation. No further visceral epithelial neoplasia was identified in the animals involved to support the alternative interpretation that these might instead represent cutaneous metastases from visceral carcinoma. These tumors exhibited various proportions of pilar and ductal (eccrine) differentiation. Areas of abrupt pilar-type keratinization alternated with lobules of more basaloid-appearing cells, arranged in true ductules with focal necrosis and showing up to moderate pleomorphism and brisk mitotic activity (Fig. 8)
.
|
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Mice Homozygous for the Arg-771-to-Trp Mutation in the Lig1 Gene as a Model for the 46BR Patient.
Two different DNA ligase I point mutations have been reported in a single patient (4)
with similar growth retardation and immunodeficiency as found in BS, a paradigm for inherited genome instability/cancer predisposition diseases (2)
. This 46BR patient developed lymphoma at age 17 and subsequently died from pneumonia. A Glu-566-to-Lys mutation at the active site inactivated one allele, whereas an Arg-771-to-Trp mutation on the second allele resulted in 20-fold reduced enzyme activity (5)
. We have used double-replacement gene targeting in the Hprt-deficient ES cell line, HM-1, to generate mice homozygous for the Arg-771-to-Trp mutation. We chose this milder mutation because only the Arg-771-to-Trp allele is retained in the 46BR cell line. The Glu-566-to-Lys allele, encoding a completely nonfunctional protein, has been lost; and we reasoned that this mutation, like our Lig1 knockouts, would probably be an embryonic lethal in mice. Mice homozygous for the Arg-771-to-Trp mutation are, thus, not a perfect model for the 46BR patient. We also have the potential to generate mice heterozygous for the 46BR mutation over a knockout allele, differences in the phenotype of these two models (Lig146BR/46BR and Lig146BR/-) might be anticipated.
Slower Growth but Normal Hematopoiesis in Older 46BR Mice.
From 5 weeks onward, 46BR mice grew more slowly than wild-type controls and took longer to attain the same mature body weight. This was consistent with their growth being slowed by the DNA ligase I defect leading to the accumulation of DNA replication intermediates. There was no evidence for a spontaneous hematopoietic defect in the older 46BR mice that might have been predicted from the immunodeficiency of the 46BR patient. All hematopoietic lineages were present in the expected amounts. However, the mice were housed in a clean environment, and their health status was good so that the ability of their immune system to respond to infection was not being tested. A difference in the ability of the hematopoietic systems of 46BR and control mice to respond might become apparent in more challenging circumstances.
Erythropoiesis Is under Stress in Young 46BR Mice.
We have previously identified erythropoiesis in Lig1 null embryos and, after transplantation of Lig1 null fetal liver cells into lethally irradiated recipients, as being critical situations in which Lig1 null cells are unable to meet the particularly high demand for DNA replication (16
, 17)
. The enlargement of the spleen in 3-week-old 46BR mice, the elevated frequency of S-phase cells, and the increased frequency of immature reticulocytes released prematurely into the circulation, all indicate that erythropoiesis is facing a similar but less severe difficulty in meeting replicative demands attributable to the DNA ligase I deficiency. Alternatively, if an as-yet-undefined DNA repair deficiency contributes to the phenotype of 46BR mice, then a higher exposure to the relevant damage in spleen might also explain this result. This situation quickly resolved, presumably because the demand for red cells dropped, or because of the increasing number of hematopoietic cells that can be devoted to erythropoiesis, and by 5 weeks of age, the sizes of the spleens and the level of DNA replication in 46BR mice were indistinguishable from those in wild-type controls.
Occasional Failure of DNA Replication in Hematopoietic Tissue from 46BR Mice.
Apart from the 2-fold increase in the frequency of S-phase cells in the spleens from 3-week-old 46BR mice, thymus and bone marrow samples also showed a smaller but significant increase. We believe that this reflects an increased number of cells in S phase, because replication takes longer to complete because of the ligase deficiency, rather than an increased rate of proliferation. In rare samples from 46BR bone marrow, the frequency distribution of nuclear DNA content was consistent with a failure occurring during DNA replication, again presumably because there was insufficient DNA ligase I activity in the 46BR bone marrow to sustain the required high rate of proliferation. This in vivo frequency distribution in 46BR bone marrow was quite distinct from that seen after the induction of apoptosis by DNA damage in wild-type bone marrow in vitro, which was associated with a classic G1 arrest, and the prevention of initiation of DNA replication as opposed to the apparent failure during DNA replication seen in vivo. Indeed, although we have observed spontaneous prominent sub-G1 peaks in 46BR spleen and thymus in vivo, we have shown clearly that this is not caused by apoptosis. Thus, we have no evidence that apoptosis is acting to remove abnormal hematopoietic cells that accumulate in 46BR mice because of the DNA ligase deficiency.
The Discrete Sub-G1 Peaks in 46BR Spleen and Thymus Could Arise through Accumulation of Single-Strand DNA Breaks.
Although broad sub-G1 peaks represent apoptotic cells, we have clearly shown that the discrete sub-G1 peaks in 46BR spleen and thymus are not caused by apoptosis. How then can they be explained? Aneuploidy is not a plausible alternative because the sub-G1 peaks indicate apparent reductions in DNA content of up to 35% (32% in the 46BR thymus shown in Fig. 5B
). Such cells would be inviable and unable to replicate their DNA in the way that the abnormal cells in 46BR thymus appear able to do, if our interpretation of the small peak at twice the sub-G1 peak value in samples such as that shown in Fig. 5B
is correct. Such discrete sub-G1 peaks can also appear rapidly (overnight after DNA damage in wild-type cultures) and may be relatively persistent in vivo. Although we are unable to sample any single animal twice, the presence of discrete sub-G1 peaks in the thymus of 46BR littermates sampled 2 weeks apart indicates that such peaks may be long lasting. We consider that only one possible explanation remains: that the sub-G1 peak represents an apparent, rather than a real, reduction in DNA content. Such a change could arise from single-strand DNA breaks reducing DNA supercoiling and leading to a reduction in the amount of propidium iodide staining. The amount of propidium iodide fluorescence in flow cytometry is dependent on the method of sample preparation (29)
, and the fluorescence from such intercalating dyes is sensitive to the degree of DNA supercoiling, although, in the case of the related dye ethidium bromide, removing supercoiling actually increases the amount of dye that can bind (30)
. An increased number of single-strand DNA breaks would be predicted in 46BR mice from the accumulation of DNA replication intermediates that we have detected in Lig1 null cells (17)
, or from an, as-yet-undetected DNA repair deficiency. This argument is strengthened by our observation that we can induce equivalent sub-G1 peaks in wild-type cultures in vitro with DNA-damaging agents. Single-strand breaks will be produced during NER of UV-induced damage, whereas hydrogen peroxide causes strand breaks directly, in addition to a range of other lesions. This hypothesis for the origin of the discrete sub-G1 peak in 46BR mice can be tested by measuring single-strand breaks directly by end-labeling or using the alkaline comet assay. In addition to the rate of proliferation in a particular tissue, the other factor that may govern the appearance of the discrete sub-G1 peak in 46BR mice is the level of expression of DNA ligase I and of the other ligases, which may in some circumstances be able to compensate for ligase I deficiency (17)
. If our interpretation is correct, we predict that we could increase the frequency of such events by stressing the hematopoietic system, either by blocking DNA replication and then releasing it to generate a surge of replication or by raising a strong, prolonged polyclonal immune response.
Genomic Instability in 46BR Mice Results from Abnormal DNA Replication.
We have previously observed an increased frequency of genome instability, assayed by micronucleus formation, in primary fibroblasts isolated from Lig1 null embryos (17)
. It is not unreasonable to suppose that strand breaks and replication failure might result in increased genome instability in 46BR mice. Because DNA repair functions were normal in Lig1 null cells, we consider that the genomic instability is more likely to result directly from the accumulation of replication intermediates rather than from an as-yet-undetected DNA repair deficiency, and we favor a similar explanation for the increased levels of instability observed in 46BR mouse tissues in vivo, in which micronucleus formation was 3-fold higher in spleens from 15-month-old 46BR mice than from controls. The failure to observe an increase in genomic instability in 46BR primary embryonic fibroblasts in vitro that is similar to what we reported in Lig1 null cells could be caused by the 46BR allele encoding a protein with some residual ligase I activity. The difference between the in vivo and in vitro results for mouse 46BR cells could arise from the greater constraints on the coordination of DNA replication and mitosis in a tightly regulated process such as hematopoiesis in vivo. Pressure to enter mitosis to a precise schedule in vivo but not in vitro, coupled with a delay in replication arising from the accumulation of replication intermediates, could result in a failure of chromosome segregation or the loss of chromosome fragments, which would both generate micronuclei.
Increased Susceptibility to Cancer in 46BR Mice.
Genome instability is a key event in the development of cancer. The modest but significant increase in instability observed in the spleens of 46BR mice correlates well with the modest increase in the spontaneous tumor incidence observed. Thirty-two % of our 46BR mice developed tumors over a 2-year study, compared with only 6% of controls. Our model for the origin of these cancers is as follows: the accumulation of replication intermediates in cells in which the reduced amounts of DNA ligase I activity associated with the 46BR mutation are unable to meet demands for joining Okazaki fragments will result in increased DNA strand breaks with resulting genome instability. This will increase the probability of loss of tumor suppressor genes, or of translocation-mediated oncogene activation and cancer formation. If this explanation is correct, then we would expect to be able to increase the spontaneous tumor incidence in the hematopoietic system by increasing the replicative pressure on these animals by blocking DNA synthesis and then releasing or by stimulating a strong, prolonged polyclonal immune response.
Occurrence of Cutaneous Adnexal Tumors in 46BR Mice.
We have observed two of these unusual tumors in our current study of 46BR mice and an additional six in an ongoing study. No wild-type control animals in these studies have developed any such adnexal neoplasm. Cutaneous adnexal tumors, especially carcinomas, are rare neoplasms both in humans and in the mouse, and precise classification remains the object of discussion. Although the designations "basosquamous tumor," of both benign and malignant types, have been applied to the mouse, this nomenclature is considered inappropriate for the tumors under discussion in this study, given their histological appearance with convincing evidence of ductular differentiation and the rather abrupt and, frequently, very prominent pilar-type keratinization. Although most commonly recapitulating solely one of the four elements in the primary epithelial germ of ectoderm, the pluripotential primordial cells may exhibit multidirectional differentiation to produce tumors with a mixed adnexal pattern. This would appear to be the case in the tumors sampled in the present study. Little is understood about the etiology and pathogenesis of adnexal tumors. In humans there is, for several of these, notably of eccrine and pilar type, a predilection for sun-exposed skin with increasing age, as with epidermal-based carcinomas. In conjunction with this fact, the observation that basal cell carcinomas may show focal terminal differentiation toward any of the pilo-sebaceous components, has perhaps generated the reductionist notion in some quarters that all adnexal tumors are, in fact, ultimately basal cell carcinomas. Such a theory is not supported by the diversity, at least in humans, in the clinical course of these neoplasms, which certainly warrants their distinction. Few studies have investigated the genetic abnormalities associated with adnexal tumors, and, therefore, their molecular pathogenesis remains to be explored. Loss of heterozygosity studies have demonstrated confinement of loss of heterozygosity to 17p, with allelic loss and accompanying p53 stabilization being demonstrated in the transition from benign poromatous to porocarcinomatous areas within the same tumor in one particular case (31)
. A further study suggests that p53 protein accumulation is not uncommon in adnexal carcinomas but is extremely rare in benign adnexal tumors (32)
.
Comparison with Other Genetically Manipulated Mouse Models of Cancer Susceptibility.
Most existing genetically manipulated mouse models for cancer susceptibility involve changes to key genes that act to prevent or cause cancer, such as inactivation of tumor suppressor genes (reviewed in Ref. 33
), inactivation of DNA repair genes (reviewed in Ref. 1
), or activation of oncogenes such as members of the ras family (reviewed in Ref. 34
). Typically, the resulting increases in cancer susceptibility are dramatic (e.g., for the inactivation of the p53 gene, 74% of animals developed tumors within 6 months; Ref. 35
), reflecting the importance of the manipulated gene to carcinogenesis, but there is often a strong bias in the types of tumor seen. Such genetically modified mice, in which the alteration is present in every cell, do not represent good models for the majority of human cancers that originate from sporadic genetic alterations in individual cells surrounded by normal neighbors. This discrepancy has been addressed in an improved model for sporadic cancer involving activation of a mutant K-ras gene in gene-targeted mice when intrachromosomal recombination occurs randomly in individual somatic cells (36)
. If we are correct in our interpretation of the mechanism for carcinogenesis in 46BR mice, this would represent a very different model of cancer susceptibility, one in which the increased cancer frequency results indirectly from increased genome instability rather than directly from the change in expression of a key gene involved in carcinogenesis. As such, this would more closely model the situation in sporadic human cancers that can arise in a wide range of cell types and tissues. The benefits of such a model would be that, potentially, any type of tumor could be obtained, provided that the cell type involved was undergoing DNA replication. The disadvantage is that the tumor frequency, although higher than in control mice, could never be as high as in models in which specific genes directly involved in carcinogenesis have been manipulated. The diverse range of spontaneous tumors seen in our first survey of 46BR mice, particularly the occurrence of a number of unusual epithelial tumors rarely seen in mice, indicates that mice with point mutations in the DNA ligase I gene will, indeed, prove to be a valuable addition to the existing mouse models for cancer.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by Project Grant SP2095/0104 from Cancer Research UK (to D. W. M.). 
2 To whom requests for reprints should be addressed, at Sir Alastair Currie Cancer Research UK Laboratories, Molecular Medicine Centre, University of Edinburgh, Western General Hospital, Crewe Road, Edinburgh EH4 2XU, Scotland, UK. Phone: 44-(0)131-651-1079; Fax: 44-(0)131-651-1072; E-mail: David.Melton{at}ed.ac.uk
3 The abbreviations used are: BS, Bloom’s syndrome; ES, embryonic stem; Hprt, mouse hypoxanthine phosphoribosyltransferase gene; Lig1, mouse DNA ligase I gene; NER, nucleotide excision repair; PE, phycoerythrin; TCR, T-cell receptor.
Received 6/29/01. Accepted 5/16/02.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-lactalbumin gene with its human counterpart in embryonic stem cells and mice. Mol. Cell. Biol., 14: 1009-1016, 1994.This article has been cited by other articles:
|
|
 |
|
  W. Song, D. S. Levin, J. Varkey, S. Post, V. P. Bermudez, J. Hurwitz, and A. E. Tomkinson A Conserved Physical and Functional Interaction between the Cell Cycle Checkpoint Clamp Loader and DNA Ligase I of Eukaryotes J. Biol. Chem., August 3, 2007; 282(31): 22721 - 22730. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Subramanian, S. Vijayakumar, A. E. Tomkinson, and N. Arnheim Genetic Instability Induced by Overexpression of DNA Ligase I in Budding Yeast Genetics, October 1, 2005; 171(2): 427 - 441. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Wu, J. A. Roth, H. Zhao, S. Luo, Y.-L. Zheng, S. Chiang, and M. R. Spitz Cell Cycle Checkpoints, DNA Damage/Repair, and Lung Cancer Risk Cancer Res., January 1, 2005; 65(1): 349 - 357. [Abstract] [Full Text] [PDF]
|
|
| ||
) and 46BR (
) mice. Mice (47 wild-type and 29 46BR animals) were weighed on 2–3 separate occasions to obtain the data shown. Each data point, the mean weight from two or more animals.
