Regulation of hematopoietic progenitors by estrogens as a basis for new antileukemic strategies

Certain types of hematopoietic cells, particularly lymphocytes, have long been known to be sensitive to regulation by sex hormones. Estrogen receptors (ERs) are expressed by immature and mature lymphoid cells; increased levels of 17β-estradiol reduce B-cell precursors in mice,¹ whereas estrogen deficiency after ovariectomy expands the B-cell compartment. Until recently, the effects of estrogens on hematopoietic stem cells (HSCs) have been much less clear. Indirect effects were expected from the anabolic role of estrogens in bone formation, resulting in remodeling of the bone marrow microenvironment. Estradiol was shown to increase the number of vascular-associated HSCs and progenitor cells (HSPCs), which was attributed to the regulation of stromal cells.² However, whether a direct regulation existed was unknown. In principle, HSCs interacting with vascular niches are exposed to hormones transported by the blood. Reanalysis of published gene expression data revealed very high levels of Esr1 (estrogen receptor α, ERα) in HSCs and early progenitors, with downregulation during differentiation; this was confirmed independently by us and others^3,4 and suggested that HSPCs could be directly responsive to estrogens.

A related and unsolved question was the potential regulation of malignant hematopoietic cells by estrogens. Epidemiologic data show a lower incidence of most hematologic cancers in women than in men, suggesting the influence of sex-related factors. The different prevalence of lymphoid malignancies could be related to the effects of estrogens on lymphopoiesis. It was recently reported that Esr2 (ERβ) activation inhibits T-cell lymphoma growth in mice;⁵ however, it was unclear why a similar pattern was observed for myeloid neoplasms, which frequently originate from mutations in HSCs.

These questions are clinically relevant because estrogen signaling can be targeted pharmacologically with selective ER modulators (SERMs) such as tamoxifen, which is widely used for the prevention and treatment of ER⁺ breast cancer. The action of SERMs is cell type-specific; whereas tamoxifen has antiestrogenic effects on breast cancer cells, we demonstrated that it functions as an ER agonist in HSCs, activating the expression of estrogen-responsive genes.³

Following up on unexpected preliminary data showing that tamoxifen administration to mice expressing inducible CreER^T2 recombinases altered the composition of HSPC subpopulations, we set out to examine the role of estrogen signaling in normal and malignant hematopoiesis. Pharmacologic activation of ER in vivo provoked 2 types of response in normal primitive hematopoietic cells.³

The first response was increased proliferation of quiescent, long-term HSCs. This observation confirmed recent data showing that the frequency of HSC division is higher in female compared to male mice, and is further enhanced by exogenous 17β-estradiol administration or pregnancy.⁴ We observed that tamoxifen had a similar proliferative action on HSCs. These effects occurred without a concurrent change in total bone marrow HSC numbers, suggesting that estrogen-expanded HSCs may rapidly differentiate. In fact, we showed that tamoxifen decreased the expression of genes associated with self-renewal and the ability of HSCs to reconstitute hematopoiesis after stress, accompanied by activation of the Myc transcriptional program, which was previously associated with HSC differentiation.

Secondly, tamoxifen treatment induced apoptosis of short-term HSCs and multipotent progenitors (MPPs), resulting in their rapid elimination from the bone marrow.

Both effects were absent in Esr1^−/− mice and in wild-type mice transplanted with Esr1^−/− bone marrow cells, proving that HSPCs are directly regulated by estrogens via hematopoietically expressed ERα. Surprisingly, despite these detrimental effects on HSPCs and a rapid reduction of marrow cellularity after tamoxifen treatment, hematological parameters were virtually unchanged in healthy mice except for a mild decrease in platelet counts. This showed that tamoxifen, even after continuous administration for several months, did not have appreciable hematological toxicity. On the other hand, its effects on primitive hematopoietic cells—impaired HSC activity and apoptosis of their immediate progeny—seemed to be favorable properties for a potential therapeutic for HSPC-derived malignancies.

We chose to test the therapeutic effect of tamoxifen in a mouse model of chronic myeloproliferative neoplasm (MPN) induced by the Janus kinase 2 (JAK2) V617F mutation (Mx1-Cre;JAK2^V617F mice), which recapitulates the pathogenesis of human polycythemia vera.⁶ Tamoxifen treatment had a dramatic effect on MPN development (Fig. 1) and abolished all phenotypes of the disease including leukocytosis, thrombocytosis, erythrocytosis, splenomegaly, and myelofibrosis/osteosclerosis associated with late-stage disease. Underlying this therapeutic effect was inhibition of JAK2^V617F-mediated HSPC survival and expansion by tamoxifen, which restored normal levels of apoptosis in mutant HSPCs. Despite qualitatively similar effects on normal and mutant HSPCs (depletion of short-term HSCs/MPPs via apoptosis), competitive transplantation assays showed preferential action of tamoxifen on mutant cells, although the basis for this selectivity is currently unclear. Tamoxifen also enhanced the elimination of leukemic cells by chemotherapy in a model of MLL-AF9–induced acute myeloid leukemia (AML). Therefore, our results suggest a therapeutic approach for a set of malignancies that currently lack curative treatment using a safe, readily available drug that is already approved for use in humans.

Among the relevant questions that need to be addressed by future work is the exact mechanism by which tamoxifen induces HSPC apoptosis. Multiple ER-dependent and independent pathways have been implicated in tamoxifen-mediated apoptosis of breast cancer cells.⁷ We have observed that, in vitro, estrogen-induced apoptosis of AML cells was preceded by rapid inhibition of mitochondrial respiration, which was likely transcription-independent³ (direct inhibition of mitochondrial complex I by tamoxifen was previously described in isolated mitochondria⁸). It is likely, however, that the action of tamoxifen in vivo combines metabolic and ER-dependent effects, since the proapoptotic effect of tamoxifen on HSPCs strictly required ERα. Given that tamoxifen cancels the survival advantage provided by JAK2^V617F, it will be interesting to determine whether the drug interferes with Jak/Stat signaling.

An immediate step will be to initiate clinical trials to assess the therapeutic effect of tamoxifen in patients with MPN. Dose optimization may be advantageous, since our analyses in the mouse model showed that a 10-fold reduction in the initial experimental dose did not diminish the therapeutic effect (Fig. 1). The possibility of using ERα-selective agonists to further increase specificity and reduce side effects deserves to be investigated.