Irradiation Selects for p53 Disruption in Hematopoietic Cells
To address the impact of irradiation on selection of cells with dysfunctional p53, we generated mice with mosaic hematopoietic systems, containing a small percentage of cells with disrupted p53 activity and co-expressed GFP (). To do so, we transplanted bone marrow (BM) progenitors transduced with low titer MSCV-ires-GFP retroviruses (MiG) encoding DDp53 (or empty vector controls) into lethally irradiated recipients. DDp53 encodes for the multimerization domain of p53 (amino acids 302-390), and expression of DDp53 leads to potent inhibition of endogenous p53 activity 
. The transplanted animals were allowed to recover for 6 wk, at which point hematopoiesis was restored with relatively normal peripheral leukocyte counts (unpublished data). At this point, roughly 2% of the cells were GFP+
both in myeloid and B-cell lineages (, time 0). Thus, this model creates a context wherein the fate of a small percentage of p53 disrupted hematopoietic progenitors can be monitored in an otherwise WT background and that also eliminates potential effects of p53 deficiency in non-hematopoietic tissues 
X-irradiation selects for hematopoietic progenitors expressing DDp53.
The mice were then sub-lethally X-irradiated (2.5 Gy), and the percentage of transduced (GFP+) cells was monitored over 4 wk in peripheral blood cells. The percentages of DDp53 transduced cells (GFP+) in the non-irradiated controls remained stable (; control groups), indicating that p53 disruption does not provide a substantial advantage during normal steady state hematopoiesis. In contrast, irradiation led to a significant increase in the percentages of DDp53 cells, as average percentages of GFP+ cells increased 5-fold in the myeloid (Mac1+ cells) lineage and 3.5-fold in the B cell (B220+ cells) lineage (; irradiated groups). Examples of flow cytometric profiles are shown in . Given that irradiation had no effect on the expansion of vector transduced cells, inhibition of p53 activity is advantageous to hematopoietic cells upon or after irradiation. An advantage conferred to early progenitors but not mature myeloid cells post-irradiation may account for the delayed rise in the percentages of myeloid cells expressing DDp53 starting at Week 2. As for the MAC1+ and B220+ lineages, irradiation caused increases in the percentages of DDp53 cells in the CD4+ and CD8+ T cell lineages ().
To confirm these findings using a different model of p53 disruption, we created BM chimeric mice containing cells with null genetic disruption of both p53
alleles. For these experiments, the null p53
was bred into a transgenic (Tg) line that expresses GFP in all tissues from the Ubiquitin-C promoter 
. We generated mosaic mice by transplantation of lethally irradiated recipients with WT BM mixed 7
1 with either p53
+/+ or p53
−/− GFP Tg BM. After hematopoiesis was allowed to recover for 6 wk, the mice were sublethally irradiated (2.5 Gy) and competitive hematopoiesis was observed by monitoring peripheral blood over the next 4 wk (). Similar to our results with DDp53 transduced cells, irradiation led to a dramatic selection for p53
−/− cells in the MAC1+
lineages, resulting in a virtual selective sweep of the p53 mutation within the hematopoietic systems of most recipient animals (). No increase in the percentage of p53 null cells, beyond those expected based on initial ratios transplanted, was evident within unirradiated hematopoiesis.
Irradiation selects for p53−/− hematopoietic progenitors.
Irradiation of recipient mice also led to dramatic increases in the percentages of p53−/− cells in the CD4+ and CD8+ T cell lineage (). But in contrast to retrovirally delivered DDp53, we also observed modest selection for p53−/− cells in non-irradiated CD4+ and CD8+ cells, which could either reflect the pre-existence of partially transformed p53−/− cells in the T-cell lineage (consistent with eventual T lymphoma development) or skewed selection for p53−/− T progenitors in the abnormal thymic environment of irradiated recipient mice.
Irradiated recipients of 7
−/− BM developed thymomas and T cell leukemias with high penetrance, while development of malignancies in the unirradiated group was significantly reduced and delayed (). All malignancies that developed with or without irradiation expressed GFP and thus were derived from the GFP Tg p53
−/− donor BM (Figure S2
). As expected, recipients of WT BM did not develop thymomas, with or without irradiation, as a single dose of radiation is not sufficient to induce lymphomas within this time frame.
P53 Disruption Protects Hematopoietic Progenitors from Irradiation-Induced Ablation
To assess whether p53 disruption provides an immediate survival advantage following irradiation, we analyzed hematopoietic tissues at 48 h post-irradiation, focusing on hematopoietic progenitor populations, whose relative survival should determine effects on long-term selection. We used the same experimental design described in , except that a 19
1 mixture of WT:p53
−/− GFP BM was transplanted. At 48 h, cells that were killed by direct irradiation-induced damage should have been cleared, while the extent of new proliferation should still be minimal. Irradiation of these chimeric mice at 2.5 Gy had a substantial negative effect on the hematopoietic system, leading to approximately 4-fold reductions in BM cellularity and 2-fold reductions in spleen weight (Figure S3
). Yet the impact of irradiation was not equivalent for different hematopoietic lineages. B lineage cells were more radiosensitive than myeloid lineage cells: while BM myeloid cells were reduced in number by about 8-fold, the total BM B220+
population was reduced about 30-fold, and BM pro-B and pre-B progenitor populations were reduced 40–60-fold (, S4
, and S5A
Loss of p53 protects from acute irradiation-induced ablation.
Accordingly, irradiation caused substantial increases in the percentages of p53
−/− cells in the pre-B and pro-B populations ( and S5A
), as well as in total B220+
population (Figure S4
). Similar results were obtained at 48 h post-irradiation for recipients of BM where p53 disruption was mediated by expression of DDp53 (unpublished data). Importantly, irradiation reduced the numbers of not only WT
but also p53
−/− pre-B and pro-B cells ( and S5A
). However, the reduction of p53
−/− B progenitor numbers was clearly less extensive. Therefore, p53 disruption provides partial radioprotection. Examples of flow cytometric profiles for detection of GFP expression in myeloid, pre-B, and pro-B cell populations are shown in Figure S6
(double-positive; DP) T cell progenitors in the thymus are known to be very sensitive to irradiation-induced apoptosis 
. Indeed, at 48 h post-irradiation we observed dramatic ablation of the DP population, while single-positive cells remained relatively unaffected (Figure S7
). Similarly to B cell progenitors, disruption of p53 provided a clear radioprotective effect in thymic T-cell progenitors, as irradiation led to increased percentages of p53
−/− cells in the DP population (). Again, this radioprotection was not absolute: while the numbers of WT GFP+
DP cells dropped about 28-fold following irradiation, the numbers of p53
DP cells dropped about 7-fold (). As also observed in experiments shown in , for the T cell lineage, p53 loss may confer some advantage even without irradiation, as p53
−/− percentages increased in DP cells in non-irradiated controls relative to pre-irradiation, and thus the irradiation induced increase in the percentage of GFP+ p53
− DP cells is less evident than in other lineages ().
We next examined the effect of irradiation on hematopoietic stem cells (HSC) by examining the HSC-enriched CD150+
BM compartment 
. In contrast to the lymphoid progenitor pools, CD150+
cell numbers were not affected by irradiation, and we did not observe changes in the percentages of p53
−/− cells (Figures S5B
). Therefore, disruption of p53 does not appear to provide an immediate survival advantage in HSC pools.
P53 Disruption Preserves Clonogenic Capacity for Irradiated Hematopoietic Progenitor and Stem Cells
Protection from immediate irradiation-induced ablation does not necessarily correlate with maintenance of long-term proliferative capacity 
. Therefore, we assessed the impact of p53 disruption on maintenance of clonogenic capacity by progenitor and stem cells. To this end, p53+/+
− mice were irradiated (2.5 Gy) and BM was harvested 48 h later. Note that X-irradiation resulted in a ~5-fold reduction in BM cellularity by 48 h for WT mice but only ~2-fold reduction for p53
− mice (). BM cells were isolated from p53+/+
− mice 48 h post-irradiation, and either plated in methylcellulose cultures for determination of colony forming units in vitro (CFU-GEMM for granulocytic/erythroid/megakaryocyte/macrophage progenitors or CFU-B for B-lymphoid progenitors) or transplanted into lethally irradiated mice for determination of CFU in spleens (CFU-S, derived from early multipotent progenitors) 
. Consistent with the analyses above, irradiation resulted in dramatic reductions in CFU-GEMM, CFU-B, and CFU-S numbers (from 20× to 100×; ), and p53 disruption provided substantial protection from irradiation-induced elimination of these progenitors (numbers were reduced only 2–3-fold).
p53 loss protects progenitors from irradiation-induced loss of clonogenic potential.
To determine the p53-dependent impact of irradiation on numbers of functional HSC, we performed limiting dilution assays. Varying numbers of “test” cells (control or 48 h post-irradiation) were transplanted into lethally irradiated recipient mice, together with a fixed number of competitors to ensure radioprotection. Since irradiation can dramatically reduce the competitive ability of HSC 
, we used competitor BM harvested from previously irradiated donors, in order to ensure that contributions of irradiated “test” HSC are not masked by non-irradiated competitors. In contrast to the lack of irradiation-induced ablation of phenotypically defined HSC (as detected by flow cytometry; Figures S5B
), we observed dramatic reductions in the frequencies of functional WT HSC but not p53
− HSC, following irradiation (). Considering the p53-dependence of the effects of irradiation on BM cellularity (), the loss of p53 confers substantial protection from irradiation-induced reductions of functional HSC numbers per mouse.
In summary, the selective advantage for p53 disruption is evident within 48 h in both long- and short-term progenitor populations, supporting a direct role for p53 in radiation-induced cell death. Moreover, beyond preventing the immediate death of stem and progenitor cells following irradiation, the loss of p53 provides an additional selective advantage through protecting clonogenic capacity.
P53 Disruption Partially Protects Hematopoietic Progenitors from Irradiation-Induced Persistent Reductions of Functional Capacity
Experiments described above demonstrate that loss of p53 protects cells from the acute effects of irradiation by preserving cell survival as measured by phenotypic and functional assays. However, the impact of irradiation is not limited to acute damage. We and others have demonstrated that hematopoietic progenitors suffer from impaired functional capacity long after the acute effects of irradiation have been reversed; i.e., irradiation-induced loss of functional capacity appears to be permanent 
. We therefore asked whether p53 disruption protects hematopoietic progenitors from this long-term reduction of functional capacity. To this end, we irradiated WT and p53
null mice, allowed them to recover for 6 wk, and used BM harvested from these mice to set up competitive transplantation experiments with non-irradiated GFP+
BM cells at 19
1 ratios (p53
−/−:GFP or WT:GFP). By 6 wk post-irradiation, BM cellularity and the numbers of early progenitors are restored 
, and thus these assays measure the impact of p53 disruption on stable reductions of fitness per progenitor caused by irradiation, as opposed to the immediate physical or functional elimination of hematopoietic progenitors.
At 3 wk post-transplantation, the percentages of non-irradiated GFP+ cells had increased well beyond the initial 5% in the transplanted mixture for both the p53−/−:GFP or WT:GFP groups (), reflecting impaired hematopoietic fitness in previously irradiated BM. Still, irradiated p53−/− cells fared substantially better than WT cells against unirradiated GFP+ competitors as assessed both in the myeloid lineage (GR1+) and in total peripheral blood cells (; ~70% of hematopoiesis was still p53−/−). At 12 wk post-irradiation, however, non-irradiated competitors completely took over the myeloid lineage, irrespective of the p53 status of the irradiated donor BM, as the percent GFP+ within the myeloid lineage was indistinguishable from recipients reconstituted with GFP+ BM only (“GFP” groups). Still, p53−/− cells maintained a substantial presence within the B220 lineage, while WT GFP-negative competitors could not be detected.
p53 mutation present at the time of X-irradiation provides a long-term fitness advantage during competitive reconstitution.
Thus, loss of p53 failed to completely prevent irradiation-induced loss of fitness of stem/progenitor cells as compared to the fitness of non-irradiated WT cells. However, more relevant to irradiation-induced selection is the selective advantage of mutant cells relative to similarly irradiated WT cells. Therefore, we directly compared the fitness of previously irradiated p53
− and WT cells, using an experimental design similar to that used in , but this time using 1
1 ratios. As controls, we used 1
1 mixtures of BM isolated from non-irradiated p53
− and WT donors. As shown in , previously irradiated p53
− BM was clearly more competitive when measured against irradiated WT BM, both in myeloid and lymphoid lineages (, red bars). In contrast, the selective advantage for p53 disruption is much less obvious in recipients of non-irradiated BM mixtures (, blue bars). These results indicate that while loss of p53 is unable to completely protect cells from irradiation-induced loss of fitness, p53 deficiency is still capable of endowing a clear competitive advantage relative to irradiated WT cells. We therefore conclude that in addition to protection from irradiation-induced ablation, protection from persistent loss of fitness contributes to selection for p53-deficientclones by irradiation.
Disruption of p53 in BM Cells after Recovery from Irradiation Does Not Confer a Durable Selective Advantage
Experiments described above demonstrate that when disrupted at the time of irradiation, p53 loss provides a strong and sustained selective advantage in hematopoietic stem/progenitor cells. We asked whether disruption of p53 is still selectively advantageous when introduced after the acute effects of irradiation have been resolved. To address this question, we introduced DDp53 or empty vector into BM progenitors harvested from donors that had been irradiated 6 wk prior to the harvest (or control donors) and transplanted the transduced BM into lethally irradiated recipients (). While transduction efficiency was similar to experiments described in , disruption of p53 failed to provide cells with a long-term selective advantage. We consistently observed statistically significant overrepresentation of DDp53 expressing cells in the B-cell lineage at 3 wk post-transplantation (); however, this advantage was no longer apparent at 8 wk post-transplantation. Thus, the observed transient advantage for p53 disruption in the B-cell lineage may reflect an advantage in short-term progenitors that is only evident during the reconstitution phase post-irradiation. Importantly, even at 21 wk post-transplantation, we did not detect increased expansion of DDp53 expressing cells in either non-irradiated or irradiated hematopoiesis, despite the clear presence of a low percentage of GFP+ cells in multiple hematopoietic lineages in most recipients (). The continued presence of GFP+ cells in multiple lineages more than 4 mo post-transplantation indicates that retroviral delivery of DDp53 did occur in long-term HSC, but that DDp53 expression was not adaptive (i.e., advantageous) within irradiated HSC or more committed progenitors.
p53 disruption after recovery from the acute effects of irradiation does not provide a selective advantage.
We therefore conclude that p53 loss does not provide a selective advantage after the acute effects of irradiation are resolved, and thus selection for p53 mutant clones requires that p53 function is defective at the time of irradiation.
Competition from Non-Irradiated Cells Blocks the Selective Expansion of Irradiated p53-DeficientClones
The experiments described above argue that the selective advantage of p53 loss at the time of irradiation is attributable to protection of p53-deficient
progenitors from immediate ablation, loss of clonogenic capacity, and sustained fitness reductions. However, it is possible that selection for p53 deficiency is in part due to additional oncogenic events that are induced by irradiation and whose ability to drive uncontrolled proliferation is permitted by the lack of p53′s critical tumor suppressive function 
. Should this be the case, then one would expect that once a cell has acquired the ability for uncontrolled proliferation, this clone will expand whether or not competing cells were irradiated.
However, experiments presented in argue against this scenario, as unirradiated competitors are capable of effectively outcompeting irradiated p53 null cells. In these experiments, competition was initiated after recovery from the acute effects of irradiation. To determine the effects of non-irradiated competitors on the acute irradiation-dependent selection for p53 loss, we asked whether non-irradiated competitors, added immediately after irradiation, can counter the selective effect of p53 disruption. To this end, we used an experimental design similar to the one presented in . Consistent with results described in , DDp53 expression conferred a clear selective advantage in myeloid, T-, and B-cell lineages. In contrast, the addition of non-irradiated competitors completely prevented this selection ().
Transplantation of non-irradiated competitors reverses the selective advantage conferred by p53 disruption following irradiation.
We performed similar experiments using mice with chimeric p53
− hematopoiesis. As expected, irradiation resulted in strong selection for p53
−/− cells in multiple peripheral blood lineages (). Similar to the results seen with inhibition of p53 by DDp53, non-irradiated competitors potently inhibited this expansion, which is particularly evident in the myeloid and B-cell lineages. As the myeloid lineage is most responsive to changes in HSC pools 
, these data indicate that non-irradiated competitors can reverse selection for p53 disruption within irradiated early progenitor pools. That the effect of competitors on T-lymphoid lineages is more delayed and less dramatic is consistent with the longer half-lives of both mature T cells and their progenitors. We therefore conclude that selection for p53 deficiency by irradiation does not depend on acquisition of additional oncogenic hits in the p53-deficient
We next asked whether inhibiting selection for p53
− progenitors by transplantation of non-irradiated competitors translates into reduced incidence of p53
− thymomas. The cohorts of recipient mice depicted in were monitored for tumor development. As before (), the irradiation of chimeras containing a minor fraction of p53
− hematopoiesis resulted in substantial promotion of p53
−/− thymoma development ( and S2
). Surprisingly, transplantation of non-irradiated competitor BM after sublethal irradiation only modestly delayed p53
−/− thymoma development, and this delay was not statistically significant. Notably, in contrast to the myeloid and B-cell lineages, we observed a substantial delay in the ability of non-irradiated hematopoiesis to displace irradiated p53
− T-lineage cells (). Therefore, a large pool of p53
− T-progenitors might be maintained for a sufficient period of time to enable the occurrence of transforming secondary oncogenic events, thus underlying the failure of the non-irradiated transplant to effectively prevent thymoma development.
Irradiation promotes p53−/− lymphomagenesis.
Alternatively, if the causal link between irradiation and tumorigenesis from p53−/− cells does not involve the selection of radioresistant p53 null cells, and instead completely relies on the mutagenic action of irradiation, then irradiation should enhance development of T-cell lymphomas in mice with non-chimeric p53−/− hematopoietic systems. When essentially all hematopoietic cells are p53−/− mutant, the contribution of selection for p53 null cells toward tumorigenesis should be negated. Thus, we transplanted radio-conditioned recipients with p53−/− BM and allowed their hematopoietic systems to return to equilibrium (at which point almost all hematopoiesis was donor-derived and thus p53−/−). We then split the mice into irradiation and control groups and followed the development of lymphomas after irradiation. Contrary to the predicted mutagenic mechanism of irradiation induced tumorigenesis, irradiation not only failed to enhance the development of T-cell malignancies but actually impeded their development, extending the mean survival of the mice ().
In summary, our results demonstrate that irradiation strongly selects for p53-deficientcells in pools of stem and progenitor cells and that this selection does not rely on acquisition of additional oncogenic mutations. These studies instead indicate that altered selection for p53 loss contributes to the causal links between irradiation, p53 disruption, and tumorigenesis.