Cell intrinsic and extrinsic factors synergize in mice with haploinsufficiency for Tp53, and two human del(5q) genes, Egr1 and Apc

Therapy-related myeloid neoplasms (t-MN) are a late complication of cytotoxic treatment of primary malignant diseases, and include myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML).¹  The incidence of t-MN is rising commensurate with improvements in cancer treatments and increased survival. The prognosis for these patients has not improved substantially in several decades (median survival, 8 months); thus, t-MN is an increasingly challenging clinical problem. Recurring chromosomal abnormalities correlate with prior treatment and clinical features of the disease.^2,3  For example, heterozygous deletions of the long arm of chromosome 5 [del(5q)] are among the most common genetic abnormalities (∼40%) in alkylator-induced t-MN, with a long latency period of 5 to 7 years, trilineage dysplasia, and a poor outcome. A del(5q) also occurs in 10% to 15% of patients with primary MDS, a subset of whom have MDS with an isolated del(5q),⁴  as well as in 15% of patients with AML de novo.

Using molecular mapping approaches, 2 minimally deleted regions have been identified: 5q31.2 in t-MN and MDS/AML, and 5q33.1 in MDS with an isolated del(5q).^1,5,6  Current studies support a haploinsufficiency model, in which loss of a single allele of more than one gene on 5q cooperate in the development of myeloid neoplasms. In fact, a number of genes and several micro RNA sequences located on 5q, including APC, EGR1, CTNNA1, DIAPH1, HNRNPA0, HSPA9, RPS14, and miR-145/146a, have been implicated in the development of myeloid disorders due to a gene dosage effect.^1,7,8  For example, RPS14 and miR-145/146a cooperate in the pathogenesis of MDS with an isolated del(5q).⁶  We previously demonstrated that EGR1 (5q31.2), which is deleted in all t-MN patients with a del(5q) and APC (5q22) and is deleted in >95% of patients, are both expressed at haploinsufficient levels, and inactivating mutations have not been identified in the remaining alleles,^9,10  confirming that these genes are not acting as typical tumor suppressor genes. Moreover, using mouse models, we showed that haploinsufficiency of Egr1 or Apc individually recapitulates some features of human MDS,^8,10  further supporting their role in the pathogeneis of t-MN with a del(5q).

The early growth response 1 gene (EGR1) encodes a transcription factor that is a direct transcriptional regulator of many known tumor suppressor genes, including TP53, CDKN1A, (p21), and PTEN,¹¹  and it has been shown to play a role in maintaining hematopoietic stem cell (HSC) quiescence and retention in the bone marrow (BM) niche.¹²  The adenomatous polyposis coli (APC) gene encodes a tumor suppressor protein best known for its involvement in the WNT signaling cascade via the regulation of β-catenin stability¹³ ; loss of APC acutely activates WNT signaling.^14,15  Activation of WNT signaling in the BM stromal niche plays a role in maintaining the HSC pool throughout life, and WNT signaling in leukemia stem cells is critical for their self-renewal.^16-18  Of note, there is emerging evidence that a large percentage of MDS/AML patients with abnormalities of chromosomes 5 and/or 7 show constitutive activation of canonical WNT signaling in osteoblast stromal precursors.¹⁹  Additional roles for APC include the regulation of mitosis, via control of spindle orientation and chromosome segregation, as well as cell migration.²⁰  In addition to the loss of 5q genes, the results of recent high-throughput sequencing studies have confirmed that loss of TP53 activity, through mutation or loss, is significantly associated with t-MN with a del(5q).^21,22  The well-characterized tumor suppressor gene, TP53, has been shown to play a role in cell cycle arrest, DNA repair, and apoptosis after cellular stress,²³  as well as maintenance of stem cell self-renewal.^24,25 

We and others have previously established a role for reduced expression of EGR1, APC, or TP53 in the pathogenesis of myeloid diseases.^8,10,25  In this report, we modeled the simultaneous reduction in expression of all three genes in mice. We observed an accelerated development of macrocytic anemia in double and triple heterozygous (Apc^del/+) mice that was reminiscent of the early stages of t-MN, suggesting that haploinsufficiency of APC cooperates with loss of EGR1 or TP53. We next examined whether this effect occurred entirely within the hematopoietic compartment (where deletion of 5q genes occur), and demonstrated that alterations within the stromal cell compartment due, in part, to deregulation of the Apc/WNT pathway, were largely responsible for the phenotype. Interestingly, we also demonstrated that exposure of the BM microenvironment to N-ethyl-N-nitrosourea (ENU) accelerates disease development in an Apc haploinsufficient background, raising awareness of the effects that alkylating agent therapy may have on the stromal microenvironment in patients.

All studies were approved by the University of Chicago Institutional Animal Care & Use Commitee. Mice were housed in a fully-Association for Assessment and Accreditation of Laboratory Animal Care-accredited facility. Mx1-Cre⁻Apc^fl/+ (Apc^fl/+) and Mx1-Cre⁺Apc^fl/+ (Apc^del/+) mice were generated by crossing Apc^fl/fl mice²⁶  with Mx1-Cre transgenic mice.²⁷  The efficiency of deletion in hematopoietic cells, after 3 intraperitoneal injections with 10 mg/kg polyinosinic-polycytidylic acid (pI-pC) (GE Healthcare, Pittsburgh, PA) when mice were 2 months old, was verified by polymerase chain reaction (PCR), as previously described.¹⁰  Conditional Apc mice, Egr1 knockout mice (provided by Dr Jeffrey Milbrandt), and Tp53 knockout mice (strain Trp53^tm1Tyi/J, developed in Tyler Jacks’ Laboratory) were all backcrossed onto the C57BL/6 (CD45.2) background. Egr1^+/−, Apc^del/+; Tp53^+/−, Apc^del/+; and Egr1^+/−, Tp53^+/−, Apc^del/+ mice and corresponding Apc^fl/+ controls were generated by crossing Mx1-Cre⁺Apc^fl/fl mice with Egr1^+/− and Tp53^+/− mice. CD45.1/CD45.2 wild-type (WT) mice were generated by crossing C57BL/6 (CD45.2) and B6.SJL (CD45.1). Total BM cells (∼5 × 10⁶ cells, of varying genotype indicated in individual figure legends) were transplanted, via retro-orbital injection, into lethally irradiated (8.6 Gy) Apc^fl/+ or Apc^del/+ mice that had been pretreated, 3 to 4 weeks prior to irradiation, with pIpC to induce deletion. When Apc^del/+ mice were used as donors, BM cells were harvested 3 to 4 weeks after pIpC. Mice were monitored daily and sacrificed either at ×3 months post-pIpC, or when they were moribund and/or severely anemic. ENU was administered at 100 mg/kg either 1 month post-pIpC or 2 weeks post-pIpC in Apc recipient mice used for transplants. Stromal cells were isolated from Apc^fl/+ or Apc^del/+ mice 4 weeks post-pIpC, as described by Soleimani and Nadri.²⁸  For Apc^del/del mice, cells were isolated 1 week post-pIpC.

A complete blood count (CBC) from heart blood was determined with a Hemavet 950 counter (CDC Technologies, Oxford, CT). All organs were recovered, fixed in 10% neutral-buffered formalin, embedded in paraffin, sectioned at 4 to 5 µm, and stained with hematoxylin and eosin for histologic examination by a pathologist (J.A.). Peripheral blood, BM aspirates, and spleen touch preparations were stained with Wright–Giemsa. Images were obtained using an Olympus microscope (Model BX51; Tokyo, Japan), equipped with an Optronics 3CCD 1080p digital camera (Goleta, CA), and processed with Adobe Photoshop (San Jose, CA).

Single-cell suspensions of BM and spleen were stained with fluorochrome-conjugated antibodies specific for CD71, Ter119, Gr-1, Mac-1 (CD11b), CD19, IgM, CD4, CD8, and Annexin V (BD Biosciences, San Jose, CA). Flow cytometry was performed on a FACSCanto or LSRFortessa (BD Biosciences), and data were analyzed with the FlowJo software (Tree Star, Inc., Ashland, OR).

Survival times (time to sacrifice) were estimated by the Kaplan-Meier method and compared between groups via log rank tests. Blood counts were compared using pairwise two-sample Student t tests (Apc^fl/+ vs Apc^del/+; Egr1^+/−, Apc^fl/+ vs Egr1^+/−, Apc^del/+; etc).

We hypothesized that loss of 1 copy of Egr1 may cooperate with Apc haploinsufficiency in the pathogenesis of anemia or other myeloid disorders, and that loss of Tp53 function may further cooperate in disease progression. To test this hypothesis, we first generated mice expressing a single allele of Egr1 and Apc: Egr1^+/−, Apc^del/+, or a single allele of Tp53 and Apc: Tp53^+/−, Apc^del/+, and the corresponding controls in the Apc^fl/+ (Apc WT) background (Figure 1). Loss of Apc is under the control of an interferon-inducible Mx1-Cre promoter, and we induced deletion of a single allele of Apc by the injection of 3 doses of the interferon-inducer pI-pC, when the mice were 2–months-old. Consistent with our previous studies, Apc^del/+ mice developed a fatal macrocytic anemia at 4 to 12 months after deletion of Apc. Interestingly, Egr1^+/−, Apc^del/+ mice developed disease significantly faster with a median survival of 179 vs 255 days (P < .0001); the control Egr1^+/−, Apc^fl/+ mice remain healthy (Figure 1A). In addition, Tp53^+/−, Apc^del/+ mice had a significantly reduced median survival of 144 days vs 255 days (P < .0001) (Figure 1B). To examine whether haploinsufficiency of Egr1 and Apc cooperate with reduced Tp53 function, we generated triple heterozygous mice (Figure 1C). Egr1^+/−, Tp53^+/−, and Apc^del/+ mice died at a significantly faster rate than Apc^del/+ mice, with a median survival of 178 days vs 255 days (P = .0002). However, there was no significant difference in the survival of the double or triple heterozygous mice (P = .66). In the absence of Apc haploinsufficiency, Egr1^+/−, Apc^fl/+; Tp53^+/−, Apc^fl/+; and Egr1^+/−, Tp53^+/−, Apc^fl/+ control mice did not develop a macrocytic anemia, but some died of T lymphomas or soft tissue sarcomas at much later time points (>300 days). In summary, loss of 1 allele of Egr1, or Tp53, or both, equally accelerated the Apc^del/+-induced fatal anemia.

All mice with Apc haploinsufficiency succumbed to a fatal anemia. To determine if there were any distinct characteristics among the 4 different genotypic cohorts, hematologic parameters for Apc^del/+ mice or Apc^del/+ mice crossed with Egr1^+/− or Tp53^+/−, or Egr1^+/− and Tp53^+/− were measured and compared with Apc^fl/+ (WT-Apc) corresponding controls (Figure 2A). Compared with Apc^fl/+ control mice, the red blood cell (RBC) and hemoglobin (Hb) count were significantly decreased, and the mean cell volume (MCV) and red cell distribution width (RDW) were significantly increased in all mouse cohorts with Apc haploinsufficiency. There was also a significantly increased monocyte count in all Apc^del/+ cohorts, consistent with the development of monocytosis. The peripheral blood smears from all Apc haploinsufficient mice showed varying degrees of macrocytosis, anisocytosis, and poikilocytosis (Figure 2B), a common feature in t-MN patients.

In all of the anemic mice, the spleen was enlarged threefold to 10-fold suggestive of extramedullary erythropoiesis. The stages of erythroid development can be defined by the expression of CD71 and Ter119. In all anemic mice with Apc haploinsufficiency, there was an increased proportion of CD71⁺Ter119⁺ erythroblasts and a decreased percentage of CD71⁻Ter119⁺ mature erythroid cells (Figure 3A). Morphologic and fluorescence-activated cell sorter analysis confirmed that this phenotype corresponded to a block at the basophilic erythroblast, or R2 (CD71⁺Ter119⁺) stage (Figures 2B and 3B). Histologic analysis also revealed that there was an effacement of the normal architecture in the spleen, with an expansion of the red pulp (Figure 2B). Effacement of splenic white pulp was reflected by a decrease in the proportion of B and T lymphocytes. Thus, loss of 1 copy of Egr1 and/or Tp53 accelerates the development of the macrocytic anemia in Apc^del/+ mice; however, the phenotype of the mice appears identical at time of sacrifice, with a block at the early erythroblast stage leading to a fatal macrocytic anemia.

To determine why the Egr1^+/−, Apc^del/+ double heterozygous mice developed the fatal anemia at a significantly faster rate, we examined the kinetics of disease development by comparing the phenotype of mice at 3 months post-pIpC treatment (Apc deletion). A comparison of the CBC values shows that early signs of a macrocytic anemia were more pronounced in double heterozygous mice, as the RBC, Hb, and hemocrit counts were lower, and MCV and RDW were slightly increased (Figure 4A). We next compared the stages of erythropoiesis. When Ter119^high cells are analyzed using both the CD71 and forward scatter (FSC) parameters, they consistently resolve into 3 principal populations: Ery.A cells (Ter119^highCD71^highFSC^high) are basophilic erythroblasts, and Ery.B cells (Ter119^highCD71^highFSC^low) are late basophilic and polychromatic erythroblasts, whereas Ery.C cells (Ter119^highCD71^lowFSC^low) are orthochromatic erythroblasts and reticulocytes. We found that at 3 months postApc deletion, the ratio of the Ery.A vs Ery.C population was significantly higher in the double heterozygous mice (Egr1^+/−, Apc^del/+) than in Apc^del/+ and control mice, suggesting that ineffective erythropoiesis develops more rapidly in the double heterozygous mice (Figure 4B). To compensate for the loss of mature erythroid cells, the number of erythroid colony-forming units in the spleen is increased in both the Apc^del/+and Egr1^+/−, Apc^del/+ mice (supplemental Figure 1). In other mouse models (eg, Stat5a/b^−/−), decreased survival of early erythroblasts accounts for the block in differentiation.^29,30  We measured the percentage of Annexin V⁺ cells in the CD71⁺Ter119⁺ (R2) population in the spleen, and consistently observed decreased apoptosis in both the Apc^del/+ and Egr1^+/−, Apc^del/+ mice, suggesting that the block in differentiation was not due to decreased survival of erythroblasts (Figure 4C). A similar phenotype was observed in Tp53^+/−, Apc^del/+ mice (supplemental Figure 2). Although Egr1 and Tp53 expression induces apoptosis in tumor cells,³¹  a reduction in the dosage of Egr1 or Tp53 did not significantly decrease apoptosis in the Apc haploinsufficient background. Thus, the kinetics of disease development appear to be due to a block in early erythroid differentiation that occurs more rapidly in double heterozygous mice, but is not associated with an increase in apoptosis.

Therapy-related myeloid neoplasms with a del(5q) is associated with prior exposure to alkylating agents and mutations of Tp53.^3,21  We previously showed that the loss of 1 copy of the del(5q) gene, Egr1, cooperates with secondary mutations, induced by the alkylating agent ENU, resulting in a myeloproliferative disease with ineffective erythropoiesis in mice.⁸  To examine the interaction of Apc, Egr1, and Tp53, and the impact of cooperating mutations, we treated mice with haploinsufficient levels of Apc, Egr1, and Tp53 with ENU. The ENU equally accelerated the Apc^del/+-induced fatal anemia, regardless of the Egr1 or Tp53 gene dosage (Figure 5A). The median survival in ENU-treated Apc^del/+ mice was 117 days vs 255 days for no ENU treatment (P < .0001). Similarly, the median survival in ENU-treated Egr1^+/−, Apc^del/+ mice was 114 days vs 179 days (P < .0001), and 101 days vs 144 days (P < .0001) in ENU-treated Tp53^+/−, Apc^del/+ mice as compared with mice with no ENU treatment. In addition, ENU treatment of triple heterozygous mice (Egr1^+/−, Tp53^+/−, Apc ^del/+) also accelerated anemia development (median survival was 97 days vs 178 days; P < .0001), and did not appear to alter the phenotype.

To examine why ENU-treated mouse cohorts with Apc haploinsufficiency died of a rapid and fatal anemia, we initially evaluated the hematologic parameters. The RBC and Hb count were significantly decreased and the MCV and RDW were significantly increased in all ENU-treated mouse cohorts with Apc haploinsufficiency compared with Apc^fl/+ (Apc-WT) controls bled at a similar time point (Figure 5B). There was also a significantly increased monocyte count, consistent with the development of monocytosis. There were 2/15 Egr1^+/−, Tp53^+/−, Apc ^del/+, and 1/16 Egr1^+/−, Apc ^del/+ mice that developed T lymphoma in addition to the anemia. All ENU-treated Apc^del/+ cohorts displayed splenomegaly with a decreased percentage of CD71⁻Ter119⁺ mature erythroid cells, and an increased proportion of CD71⁺Ter119⁺ erythroblasts, corresponding to a block at the basophilic erythroblast stage (supplemental Figure 3). These data suggest that all ENU-treated, Apc^del/+ cohorts develop a macrocytic anemia with monocytosis that is similar to non-ENU treated, Apc^del/+ cohorts; however, they develop this phenotype at an accelerated pace, regardless of the loss of Egr1 or Tp53 expression.

Studies with the Apc^min mice have suggested that cell extrinsic factors may contribute to the development of myeloid disease.³²  To determine whether the fatal anemia in Apc^del/+ mice was cell intrinsic, we transplanted WT; Egr1^+/−; Tp53^+/−, Apc^del/+; or Egr1^+/−, Apc^del/+ BM into lethally irradiated Apc^del/+ recipients or Apc^fl/+ control recipients that are WT for Apc. Transplantation of BM cells into Apc^del/+ recipients resulted in a rapid development of anemia about 4 months after transplantation, regardless of the donor cell genotype. In contrast, transplantation into Apc^fl/+ (Apc-WT) or Egr1^+/− recipients (not shown) did not result in disease (Figure 6A and supplemental Figure 4). We confirmed that altered hematopoiesis occurred in donor cells, as Apc^del/+ recipient hematopoietic stem and progenitor cells are not present in recipient mice after transplantation of CD45.1/CD45.2 WT donor BM cells into Apc^del/+ (CD45.2) recipients (Figure 6B). We also showed by using PCR that the splenocytes isolated from anemic recipients were donor-derived and WT for Apc, as anticipated (Figure 6C). Together, this suggests that the fatal macrocytic anemia is induced by an Apc-haploinsufficient BM microenvironment. Supporting this conclusion, we confirmed that stromal cells isolated from Apc^del/+ mice undergo deletion of 1 Apc allele (Figure 6D). Our initial studies support a relationship between Apc dose and β-catenin levels in stroma, whereas, in splenic cells we observed β-catenin protein expression after deletion of both Apc alleles, but not after haploinsufficient loss of Apc (ie, only in Apc^del/del cells), we observed a moderate (∼1.8-fold) and a strong (∼9 fold) increase in β-catenin protein expression in Apc^del/+ and Apc^del/del stromal cells, respectively (Figure 6E).

Similar to Apc^del/+ mice, all Apc^del/+ recipient mice developed a fatal macrocytic anemia, with an arrest of erythroid maturation at the CD71⁺Ter119⁺ (R2) stage (supplemental Figure 5). Lethally irradiated Apc^del/+ recipient mice transplanted with Apc^del/+ BM developed the anemia at a significantly accelerated rate compared with Apc^del/+ mice (median survival of 130 days vs 255 days; P < .0001; comparing Figures 1 and 6), suggesting that alterations of the microenvironment after irradiation accelerated the anemic phenotype. There were 2 exceptions to the macrocytic anemia phenotype: 1 Apc^del/+ recipient that received Tp53^+/− donor cells developed an immature lymphoid leukemia with an infiltration of blasts into the liver and lymph node, and another that received Egr1^+/− donor cells developed a myeloproliferative disorder with an expansion of immature myeloid cells in the spleen (supplemental Figure 6). This suggests that an Apc haploinsufficient microenvironment may be conducive to developing other hematologic diseases, in addition to a fatal macrocytic anemia.

Given that ENU rapidly accelerated the onset of anemia, regardless of Egr1 or Tp53 expression, and that Apc^del/+-induced anemia was accelerated by transplantation into a lethally irradiated Apc haploinsufficient microenvironment, we considered the possibility that alkylating agent therapy and/or radiation may influence early disease progression through alterations to the BM stroma. To examine this possibility more closely, we transplanted CD45.1/CD45.2 WT BM cells into lethally irradiated Apc mice (CD45.2) that had been pretreated with both ENU (to induce mutations) and pIpC (to induce Apc deletion in Apc^del/+, but not Apc^fl/+, Cre^- control mice) (Figure 7). Interestingly, exposing Apc^del/+ recipients to ENU significantly accelerated the development of the fatal anemia. Median survival was 84 days for ENU-treated Apc^del/+ recipients vs 101 days for non-ENU–treated Apc^del/+ recipients (P = .014). In contrast, only 1 in 5 Apc^fl/+ (WT) recipients exposed to ENU developed a T-cell lymphoma (at 145 days) within the timeframe examined (∼224 days). The accelerated rate of disease development in ENU-treated recipient mice was also observed when Tp53^+/− donor BM cells were used (112 days for ENU-treated vs 141 days for non-ENU–treated recipients; P = .011) (supplemental Figure 7). To date, none of the ENU-treated Apc^fl/+ (WT) recipients of Tp53^+/− BM cells have developed disease (at 196 days). Together, these results suggest that exposure of the microenvironment to an alkylating agent increases the pace of disease development, in an Apc haploinsufficient recipient background, and is not influenced by donor cell genotype, under these experimental conditions.

Therapy-related myeloid neoplasia is a complex disease; in ∼40% of cases, the disease involves loss of genetic material from the long arm of 1 chromosome 5 homolog, presumably leading to the disruption of multiple signaling networks. A prevalent hypothesis is that chemotherapy and/or radiation treatment, administered for the primary malignant disease, promotes the acquisition of recurring chromosomal abnormalities, such as del(5q) or −7/del(7q), in hematopoietic stem cells, ultimately leading to acquisition of cooperating mutations, such as TP53 mutations, and resulting in the expansion of clonal populations. In addition to the loss of 5q genes in hematopoietic stem and progenitor cells, emerging evidence in MDS and AML with myelodysplasia-related changes suggests that a permissive BM microenvironment may be required for frank malignancy to develop.³³  For example, in patients with MDS with a del(5q), the beneficial effects of lenalidomide is associated with a substantial improvement of the hematopoietic supporting potential of the BM stroma.³⁴  In this paper, we developed a genetically modified mouse model recapitulating some of the molecular defects that we hypothesize are key to the development of t-MN with a del(5q), namely haploinsufficiency of EGR1, APC, and TP53.

Apc haploinsufficient mice manifest a number of features of the early stages of t-MN, including anemia characterized by anisopoikilocytosis, macrocytosis, and an increased MCV.¹⁰  Herein, we demonstrated that Apc haploinsufficiency in mice, combined with Egr1 and/or Tp53 haploinsufficiency accelerated disease development without altering disease phenotype. These results suggest that loss of 1 copy of Egr1 or Tp53 functioned as an enhancer of the Apc haploinsufficient phenotype, possibly by introducing cross-talk with additional signaling pathways or enhancing a common pathway. Interestingly, the disease was not further exacerbated by the combined heterozygous loss of Egr1 and Tp53, suggesting that the loss of Egr1 or Tp53 may act redundantly, possibly eliciting similar molecular pathways, at least with respect to the anemia phenotype in this mouse model.

Although our genetic crosses of Egr1, Apc, and Tp53 triple heterozygous mice demonstrated cooperation between pathways induced by haploinsufficient expression, they did not address whether cooperation was occurring only within hematopoietic cells, in which loss of 5q genes is known to occur in t-MN patients. We were also interested in the effects on nonhematopoietic cells, as emerging evidence suggests that the BM microenvironment plays a crucial role in the development of myeloid neoplasms. For example, the results of characterization of Apc^min mice have suggested that haploinsufficient loss of Apc may alter the BM microenvironment and contribute to the MDS phenotype.³²  To determine whether the observed cooperation between Egr1 and Apc, or Tp53 and Apc, occurred within hematopoietic or progenitor cells vs BM stromal cells, or resulted from a cooperation between these components of the BM niche, we performed transplantation studies. Regardless of the genotype of the donor cells, the Apc^del/+ recipients developed a fatal macrocytic anemia at a significantly faster rate than did the conditional Apc^del/+ mice (P < .0001). We speculate that irradiation of recipient mice may have been responsible for the faster rate of anemia development, as ionizing radiation induces multiple changes in the BM stromal cell compartment.^35-37  For example, irradiated mesenchymal stromal cells have a high DNA double-strand break repair capacity, allowing them to survive radiation treatment and produce factors, such as tumor necrosis factor-α, transforming growth factor-β, and IL-6, cytokines known to influence the development of MDS and AML.^38,39 

Because MDS patients with a del(5q) have a loss of EGR1 and APC in hematopoietic cells, we also tested whether transplantation of Apc^del/+or Egr^+/−, Apc^del/+ BM cells into WT recipients (Apc^fl/+) caused a deleterious phenotype; however, none of these mice showed defects in erythropoiesis or red blood cell production or other hematopoietic defects (supplementary Figure 5). This suggests that the effects we observed were predominantly due to deregulation of Apc expression in the BM microenvironment. At first glance, the link between our mouse model and t-MN may not be apparent, as there is no strong evidence that del(5q) genes are deleted in mesenchymal stromal cells in patients. However, there is emerging evidence that WNT signaling, of which APC is a major regulator, is disrupted in BM stromal cells from patients with MDS or AML. Kode et al¹⁹  showed that 15 of 53 patients with MDS or AML, chosen at random, showed nuclear localization of β-catenin in stromal osteoblasts. Of note, 12 of these 15 patients (80%) had abnormalities of chromosomes 5 and/or 7, characteristic features of the largest genetic subgroup of t-MN patients. Consistent with the hypothesis that aberrant WNT signaling in the BM microenvironment may play a role in MDS progression, we observed an Apc dose-dependent accumulation of β-catenin in stromal cells isolated from Apc haploinsufficient or null mice (Figure 6E).

A seminal finding in our study points to the potential deleterious effects of cytotoxic therapy, particularly alkylating agents, on the BM microenvironment of patients who develop t-MN. First, we noted a rapid acceleration of the disease phenotype in ENU-treated, conditional Apc haploinsufficient mice, regardless of the germline loss of Egr1 or Tp53 in HSCs and BM stromal cells (Figure 5). Second, transplantation of BM cells from these ENU-treated Apc^del/+ mice engrafted WT recipients, but did not result in disease up to 260 days posttransplant (data not shown). Third, we showed that exposure of the BM microenvironment to ENU increased the pace of macrocytic anemia development in an Apc haploinsufficient, but not a WT, background (Figure 7 and supplementary Figure 7). Together, our data and the literature supports the hypothesis that radiation and/or chemotherapy that t-MN patients receive for their primary malignant disease leads to a deregulation of signaling networks within the BM microenvironment, possibly through a deregulation of the WNT pathway.

In the context of this current mouse model, the impact of Egr1 and Tp53 haploinsufficiency on the hematopoietic cells, stromal cells, or both is more complex. At first glance, the transplantation data supports the hypothesis that loss of Egr1 or Tp53 in the stromal cells is enhancing the effects of Apc haploinsufficiency within the microenvironment. However, it is equally plausible that the effects of radiation, needed for the transplantation process, in itself has particularly strong effects on an Apc haploinsufficient microenvironment that masks a more subtle cooperation between Egr1^+/− or Tp53^+/− in hematopoietic cells with Apc haploinsufficiency in stromal cells. An additional possibility that we cannot rule out at this time, is a cooperation between loss of Egr1, Tp53, and Apc in hematopoietic cells. In this regard, we have found that knock-down of Tp53 (using short hairpin RNA) in Egr1, Apc double heterozygous hematopoietic cells, when transplanted into a WT background, leads to a low frequency of myeloid leukemia (data not shown). In summary, our data point to the importance of considering the effects that radiation and/or alkylating agent therapy has not only on hematopoietic cells, but also on the microenvironment. Moreover, our data raise the possibility that loss of genetic material in hematopoietic cells, combined with the disruption of key signaling pathways within the BM niche, may synergize in the pathogenesis of myeloid disease.

The authors thank Zhijian Qian, Rachel Bergerson, Kenan Onel, James Downing, Scott Lowe, Scott Kogan, Kevin Shannon, David Largaespada, and Richard Larson for helpful discussions, and Elizabeth M. Davis for technical assistance.

This work was supported by National Institutes of Health, National Cancer Institute (P01 CA40046) (M.M.L.B.), and the Integrated Microscopy and Cytometry and Antibody Technologies Core Facilities of the Comprehensive Cancer Center.

Contribution: A.S. designed and supervised research, analyzed data, and wrote the manuscript; A.S., J.W., and A.A.F. performed experiments; J.A. analyzed histology data; T.K. assisted with the statistical analysis and reviewed the manuscript; and M.M.L.B. designed and supervised research and co-wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Angela Stoddart, Section of Hematology/Oncology, University of Chicago, 900 E 57th St, KCBD 7th floor, Chicago, IL 60637; e-mail: astoddar@bsd.uchicago.edu.