In vivo prostaglandin E₂ treatment alters the bone marrow microenvironment and preferentially expands short-term hematopoietic stem cells

Hematopoietic stem cells (HSCs) must balance self-renewal and differentiation to replenish the entire hematopoietic system and simultaneously survive throughout the life of a person.¹  This essential regulation of stem cells is thought to be determined, at least in part, by the environment, or niche, in which these cells reside.²  We and others first identified osteoblastic cells as a regulatory component in the HSC niche through genetic means,^3,4  and several additional potential cellular components have now been identified.⁵  Thus, microenvironmental signals within the bone marrow, such as in the setting of injury, would be expected to influence HSC behavior both directly as well as indirectly, through their actions on HSC niche cells. One such microenvironmental signal may be prostaglandin E₂ (PGE₂). Prostaglandins are arachidonic acid derivatives produced through cyclooxygenase 1 and 2 (COX-1 and COX-2) activation in a variety of tissues where they are involved in local physiologic and pathologic processes.⁶  PGE₂ can target both stromal and hematopoietic components of the bone marrow. Specifically, PGE₂ or its analogs can stimulate periosteal and trabecular osteoblasts in humans, rats, and mice.^7-14  In addition, PGE₂ has bone-resorbing effects¹⁵  and can either inhibit or stimulate osteoclastic cells, depending on the experimental model used.^16-19  Interestingly, parathyroid hormone (PTH), known to indirectly expand HSCs through its effects on osteoblastic cells in the bone marrow, is one of the principal stimulators of PGE₂ in osteoblastic cells.^20-23 

PGE₂ has long been known to have both inhibitory and stimulatory effects on differentiation of hematopoietic progenitors and on lineage allocation.^24-29  More recently, studies in zebrafish established a role for PGE₂ in HSC regulation in vivo in the setting of kidney marrow recovery after myeloablative injury.³⁰  In addition, ex vivo treatment of murine bone marrow cells (BMCs) with a stabilized form of PGE₂ increased the frequency of long-term (LT) repopulating HSCs.^30,31  However, whether in vivo treatment with PGE₂ expands HSCs in mammals is currently unknown.

Emerging data have begun to demonstrate the therapeutic potential of osteoblastic stimulation to achieve in vivo HSC expansion.^3,32,33  Because PGE₂ and its analogs have been used clinically,^34,35  and given the effects of PGE₂ on bone and on hematopoietic cells, we hypothesized that in vivo treatment with PGE₂ could expand HSCs through its beneficial effects on both osteoblasts and HSCs in the bone marrow. In this study, we modified an in vivo PGE₂ regimen previously used to stimulate osteoblasts and tested its effects on both hematopoietic cells and the skeletal structure of treated mice. We demonstrate that PGE₂ treatment in vivo (1) alters the bone microarchitecture, (2) has a regulatory effect on HSCs in the absence of injury, (3) preferentially expands short-term (ST) HSCs, and (4) does not impact LT-HSC numbers, lineage distribution, or progenitor differentiation. Taken together, our data suggest that PGE₂ treatment (or, potentially, treatment with specific agonists of PGE₂ receptors) could be used to preferentially increase donor ST-HSCs, which are highly proliferative,³⁶  to accelerate hematopoietic recovery after stem cell transplantation, and potentially to expand primitive hematopoietic cells in the absence of injury.

PGE₂ was purchased from Sigma-Aldrich, and synthetic rat PTH (1-34), was purchased from Bachem.

Six- to 10-week-old C57bl/6 male mice were injected intraperitoneally twice daily for 16 days with either 6 mg/kg body weight PGE₂ or vehicle (4% ethanol in molecular grade water). Hindlimbs were harvested 24 hours after the final injection and either used for bone marrow harvest (left hindlimb) or fixed and analyzed by micro-CT and histology, as described³  (right hindlimb). All experiments were approved by the Institutional Animal Care and Use Committee at the University of Rochester School of Medicine.

The bone marrow was flushed from the long bones of the left hindlimb of each mouse. Red blood cells (RBCs) were lysed in 1 mL RBC lysis buffer (156 mM NH₄Cl, 127 μM ethylenediaminetetraacetic acid, 12 mM NaHC₃) for 5 minutes at room temperature. After RBC lysis, the remaining mononuclear cells were suspended in fluorescence-activated cell sorter buffer (phosphate-buffered saline [PBS] containing 3% heat-inactivated fetal bovine serum [HI-FBS]). A total of 10⁷ cells were then stained with appropriate antibodies (supplemental Table 1, available on the Blood website; see the Supplemental Materials link at the top of the online article). The cells were washed, and the data were collected on a LSR-II (BD Biosciences) and analyzed using FlowJo software (Tree Star).

Bone marrow cells flushed from the left hindlimbs were resuspended in α-minimum essential medium (Mediatech) supplemented with 2% HI-FBS at a concentration of 2 × 10⁵ cells/mL; 0.4 mL cell-containing media was added to 4 mL methylcellulose containing M3434 media (Stem Cell Technologies), as previously described.³ 

Before death, 200 μL blood obtained by mandibular venous plexus sampling was run through a CBC-DIFF Veterinary Hematology System (HESKA) to obtain blood cell counts.

Mice expressing the congenic marker CD45.1 were treated with PGE₂ or vehicle as previously described. The bone marrow was flushed from the hindlimbs of these mice and mixed with bone marrow cells from a competitor animal expressing CD45.2 at a ratio of 1:2 or 1:1 (donor/competitor). The cell mixture was resuspended in PBS containing 3% HI-FBS. A total of 7.5 × 10⁵ total cells in 100 μL was injected into the tail veins of CD45.2-expressing recipient animals that had received two 500-cGy doses of radiation from an x-ray source separated by 24 hours (the second dose being given 2 hours before injecting the bone marrow cells). To evaluate engraftment, the peripheral blood of recipient animals was obtained by mandibular venous plexus sampling at the times indicated. The blood was incubated in a 2% solution of 5 × 10⁵ molecular-weight dextran to precipitate the RBCs. The resulting supernatant containing the white blood cells was analyzed by flow cytometric analysis using the hematopoietic markers CD45.1, CD45.2, B220, CD3, and CD11b. For some of the experiments, primary recipients were humanely killed at week 36 after transplantation and the bone marrow was analyzed for lineage⁻ Sca-1⁺ c-kit⁺ (LSK) content and used for secondary transplantation.

BMCs were harvested from the hindlimbs from primary recipients (CD45.2) of vehicle or PGE₂-treated donors (CD45.1) 36 weeks after the primary transplantation. Donor LSKs were enumerated by flow cytometric analysis. BMCs were enumerated and mixed with BMCs from a competitor animal expressing CD45.2 at a ratio of 1:2 (donor/competitor). As for the primary transplantation, the cell mixture was resuspended in PBS containing 3% HI-FBS. A total of 7.5 × 10⁵ total cells in 100 μL were injected into the tail veins of CD45.2-expressing secondary recipient animals that had received two 500-cGy doses of radiation from an x-ray source separated by 24 hours (the second dose being given 2 hours before injecting the bone marrow cells). To evaluate engraftment, the recipient animals were bled by mandibular bleeds at the times indicated. As for the analysis of engraftment after the primary transplantation, the blood was incubated in a 2% solution of 5 × 10⁵ molecular weight dextran to precipitate the RBCs. The resulting supernatant containing the white blood cells was analyzed by flow cytometric analysis using the hematopoietic markers CD45.1, CD45.2, B220, CD3, and CD11b.

The bone marrow from mice treated with PGE₂ or vehicle was flushed and resuspended in PBS containing 3% HI-FBS. Recipient mice that had been lethally irradiated (two 500-cGy doses of radiation from an x-ray source separated by 24 hours) were injected with 6 × 10⁴ cells each. After 8, 10, 11, or 12 days, the animals were humanely killed and the spleens were fixed in Telly solution containing 70% ethanol, 2% formaldehyde, and 5% acetic acid for 24 hours. After fixation, the spleens were transferred to 70% ethanol, photographed, and the colonies counted.

The right hindlimb of each mouse was fixed for 3 days in 10% formalin and then stored in 70% ethanol at 4°C. The limbs were scanned on a Viva CT 40 (Scanco Medical) using a 55-kVp, 145-μA current and a 300-ms integration time. A resolution of 12.5 μm was used. A 2.0875-mm region that was 50 μm above the growth plate in the femur and 50 μm below the growth plate in the tibia was analyzed.

For quantitative assays, treatment groups were reported as mean plus or minus SEM and compared using the Student t test. When multiple comparisons were necessary, analysis of variance with Bonferroni posttest was performed using GraphPad Prism, Version 4.01 for Windows (GraphPad Software). Statistical significance was established at P less than or equal to .05.

PTH increases trabecular bone volume, and PGE₂ treatment has been reported to activate osteoblastic cells, resulting in an anabolic bone effect in rats.⁹  To activate the bone marrow microenvironment, we treated mice in vivo with either PGE₂ or vehicle, modifying a regimen previously found to have bone effects in rats.¹⁴  This regimen (6 mg/kg twice daily for 16 days) was well tolerated without morbidity, mortality, or weight changes in the PGE₂-treated mice (data not shown). In this model of PGE₂ treatment, micro-CT scanning demonstrated significant and consistent changes in trabecular bone microarchitecture, including PGE₂-associated decreased numbers of trabeculae in the metaphyseal areas of femora and tibiae and increased trabecular spacing (Figure 1A-H). Histologic analysis of high-power images from the metaphyseal area of the distal femur also demonstrated disrupted trabeculae with increased trabecular spacing in PGE₂-treated compared with vehicle-treated mice (Figure 1E-F). Neither histologic images (Figure 1K-L) nor micro-CT evaluation (data not shown) demonstrated any significant gains in trabecular bone volume in mice treated with PGE₂ compared with vehicle. These data confirm that PGE₂ significantly alters the bone microarchitecture of treated mice.

To determine whether systemic PGE₂ had hematopoietic effects, we performed peripheral blood cell counts and enumerated bone marrow hematopoietic progenitors using the colony-forming unit (CFU-C) in vitro assay. There was no difference in the hematocrit, the number of platelets, or the total number of white blood cells in the peripheral blood cell counts of PGE₂- and vehicle-treated mice at the end of 16 days of treatment (Figure 2A-C), demonstrating that PGE₂ treatment did not alter lineage allocation or inhibit hematopoietic differentiation. To ensure that systemic PGE₂ did not block differentiation of hematopoietic progenitors, we quantified hematopoietic progenitors using the CFU-C assay. No difference in the number of progenitors was found in the bone marrow comparing PGE₂- and vehicle-treated animals (Figure 2D). Finally, to begin to elucidate effects of systemic PGE₂ on HSCs, we performed flow cytometric analysis of bone marrow to quantify the HSC-enriched LSK subset (Figure 2E-F). We identified a significant increase in the LSK population after in vivo PGE₂ treatment (Figure 2G). This increase in LSK cells was of a similar magnitude as the one induced by PTH.³  The magnitude of the LSK increase was dose dependent (Figure 2H). These data indicate a significant stimulatory effect of systemic PGE₂ on LSK cells but no alteration of hematopoietic differentiation.

The LSK subset of bone marrow cells is enriched for HSCs.^37-41  However, the LSK compartment is heterogeneous and contains at least 3 subpopulations of cells (LT-HSCs, ST-HSCs, and MPPs) with multilineage potential, but with progressively more limited self-renewal. Flow cytometric analysis was next performed to determine whether in vivo PGE₂ treatment preferentially expands specific LSK subsets, identified phenotypically by cell-surface markers. HSC subsets can be discriminated with flow cytometry by their differential expression of the SLAM receptors, CD150 and CD48, where, within the LSK subset, LT-HSCs are CD150⁺ CD48⁻ and ST-HSCs/MPPs are CD150⁻ CD48⁻.^42,43  Using these markers to assess the frequency of the different subpopulations in the bone marrow (Figure 3A), LT-HSCs were not significantly increased with in vivo PGE₂ treatment (Figure 3C), whereas ST-HSCs/MPPs were expanded (Figure 3D).

HSC subpopulations can also be separated within the LSK pool based on expression of Flt3 and Thy1.1, where gain of Flt3 and loss of Thy1.1 correlate with increased proliferation³⁶  and decreased self-renewal capability^44,45  (Figure 4). This phenotypic analysis of bone marrow cells from mice treated with PGE₂ for 16 days demonstrated a significant increase only in the LSK (Figure 4B) and LSK Flt3⁺ Thy1.1^low (Figure 4C) populations, the latter representing mostly MPPs, whereas LSK Flt3⁻ Thy1.1^int (Figure 4E), enriched for LT-HSCs and LSK Flt3⁻ Thy1.1^int (Figure 4D), enriched for ST-HSCs, were not significantly increased (Figure 4C). Taken together, these data support the notion that in vivo PGE₂ treatment increases HSCs that have limited self-renewal capacity.

The in vivo CFU-S assay has been used as an alternative method to quantify subpopulations of progenitor/HSC activities,^30,43,46  with CFU_11-12 being derived from a more primitive progenitor/stem cell population than CFU₈.^47-50  Donor bone marrow cells from mice treated with PGE₂ or vehicle were transplanted into lethally irradiated wild-type recipient mice, and the number of colonies per spleen was quantified at days 8, 10, 11, or 12 after transplantation. Mice transplanted with bone marrow cells from PGE₂-treated donors had significantly more spleen colonies 8 days after transplantation compared with mice receiving cells from vehicle-treated mice (Figure 5). However, at days 10 through 12 after transplantation, there was no difference in the number of colonies between mice receiving marrow from PGE₂ versus vehicle-treated donors (Figure 5).⁴⁸  Thus, these data again suggest that in vivo PGE₂ treatment selectively expands HSCs of limited self-renewal.

To determine whether the observed increase in LSK subsets was the result of an increase in primitive hematopoietic cells with multilineage reconstitution but limited self-renewal, we performed competitive repopulation assays. At 2, 3, and 4 weeks after transplantation, irrespective of the donor/competitor ratio, the peripheral blood of the recipients that received cells from PGE₂-treated donors demonstrated greater reconstitution in the myeloid and B-cell compartments compared with recipients that received cells from vehicle-treated donors as determined by flow cytometric analysis (Figure 6A-B). However, the early increase in engraftment seen in the recipients of PGE₂- compared with vehicle-treated donor cells was subsequently lost. This loss was first evident in the myeloid lineage (Figure 6), beginning 6 weeks after transplantation and persisting one year after transplantation (Figure 6C). The enhanced reconstitution of the B-cell compartment in recipients of PGE₂-treated cells was maintained through 6 and 12 weeks (Figure 6C) but was also lost by 1 year after transplantation (Figure 6C). Similar kinetics were observed for the loss of the greater reconstitution in the T-cell compartment of recipients of the PGE₂-treated cells (Figure 6C). These lineage-dependent differences in duration of PGE₂-dependent engraftment superiority are consistent with the kinetics of ST-HSC and MPP progeny, which directly depend on the half-life of the mature cells in each lineage.⁴³  Noncompetitive transplantation in myeloablated recipients also demonstrated a trend toward increased WBC and platelet counts 3 weeks after transplantation, consistent with a PGE₂-dependent increase in the less quiescent ST-HSCs/MPPs (supplemental Figure 1). There was no increase in hematocrit (supplemental Figure 1), suggesting that in vivo treatment with PGE₂ may not increase erythroid progenitors, as was reported for PGE₂ effects in vitro.^26,29 

Several models have demonstrated that increased proliferation of primitive cell types is associated with depletion of the HSC pool.^51-54  However, in our experimental model, there was no evidence of decreased PGE₂-dependent contribution to engraftment at one year (Figure 6C), suggesting that the accelerated engraftment induced by PGE₂ did not compromise long-term repopulation.

To confirm the transient nature of the increased engraftment of marrow from PGE₂-treated mice, we performed secondary transplantation assays 36 weeks after the primary transplantation (Figure 7A). At this time after transplantation, there was no difference in bone marrow donor-derived LSK between primary recipient mice that received marrow from PGE₂-treated donors and those that received marrow from vehicle-treated donors (Figure 7B). Engraftment in secondary recipients was measured at 10 weeks, when there was no longer any significant PGE₂-dependent increase in lymphomyeloid engraftment (Figure 7C). It is noteworthy that there was no decline in the ability of BMCs derived from PGE₂-treated mice to reconstitute bone marrow after secondary transplantation, again suggesting that the accelerated engraftment induced by PGE₂ did not compromise long-term repopulation (Figure 7C). Taken together, these data demonstrate a transient increase in engraftment of HSCs from mice treated with PGE₂ compared with vehicle, which suggests a PGE₂-dependent increase in HSCs with limited self-renewal capacity.

Microenvironmental bone marrow components have been demonstrated to regulate HSC behavior.⁵  As more of these components are identified, they highlight the complexity of local regulation of HSC cell fate choices and provide additional targets for therapeutic HSC manipulation. Thus, signals in the bone marrow probably influence HSCs both directly as well as through their actions on the HSC niche. Bone marrow injury would be expected to impact HSCs not only directly, but also through its effect on the cellular components of the HSC niche. One of the mediators of injury is PGE₂, known to influence not only hematopoietic cells but also osteoblasts and osteoclasts. Here, we demonstrate in mice that in vivo PGE₂ administration in the absence of injury is able to activate the bone marrow microenvironment and stimulate cells in the HSC pool.

Current technology enables identification and prospective isolation of HSCs using surface markers and flow cytometric analysis and sorting. The LSK subset of bone marrow cells is enriched for HSCs.^37-41  However, the LSK compartment is heterogeneous and contains at least 3 subpopulations of cells with multilineage potential, but with progressively more limited self-renewal. There is significant controversy on the phenotypic markers identifying LT-HSCs versus ST-HSCs and MPPs,^37,40,42,46  as well as to whether cells within this compartment can contribute to all hematopoietic progeny.^37,43  However, the existence of less quiescent HSCs of limited self-renewal and increased proliferative capacity is well documented in both murine and human hematopoiesis.^55,56 

Our competitive repopulation data clearly demonstrate that in vivo PGE₂ treatment preferentially expands HSCs of limited self-renewal because the PGE₂-dependent superior myelogenous and lymphocytic repopulation was lost by 6 and 16 weeks, respectively. Moreover, 36 weeks after primary transplantation, there were no differences in bone marrow LSK cells in primary recipient mice who received bone marrow cells from PGE₂ versus vehicle-treated donors, and secondary transplantation no longer identified significant differences in PGE₂ versus vehicle-dependent engraftment. In addition, we used 2 different strategies to identify these HSC subsets by flow cytometric analysis; and in each case, PGE₂ treatment had a significant and more dramatic effect on ST-HSCs/MPPs.

Another measure of ST-HSC/MPPs activity, CFU-S₈ quantification, also supports the claim that in vivo treatment with PGE₂ preferentially expanded HSCs of limited self-renewal. Taken together, our data strongly suggest that in vivo PGE₂ administration preferentially expands HSCs with limited self-renewal capability. To our knowledge, no other single factor has been reported to preferentially expand ST-HSCs.

The less quiescent subsets of the HSC pool possess higher proliferative capacity³⁶  while being capable of rapidly reconstituting all lineages.⁴³  As such, the expansion of ST-HSC/MPPs could be of great benefit to accelerate recovery after bone marrow injury, as others have suggested.^46,49  However, data in zebrafish and in murine repopulation after ex vivo treatment of murine hematopoietic cells with a stable derivative of PGE₂, 16-16-dimethyl PGE₂, demonstrated increased HSC repopulation ability both in the short and long term.^30,31  Interestingly, this enhanced HSC repopulation ability was seen primarily at 6 and 12 weeks after transplantation, with a less dramatic effect seen at 24 weeks, consistent with our preferential effect of PGE₂ on ST-HSC/MPPs compared with LT-HSCs.³⁰  Certainly, in our model, there does not appear to be a deleterious effect of PGE₂ on LT-HSCs because there were no differences in engraftment after secondary transplantation from PGE₂ versus vehicle-treated donors.

In addition to the fact that in our model bone microenvironmental effects may modulate PGE₂ action on HSCs, our PGE₂ dosing regimen also significantly differed from the experimental regimen by North et al³⁰  and Hoggat et al³¹  because we administered multiple doses of PGE₂ in our in vivo studies and did not use a long-acting PGE₂ analog.

It is possible that the preferential effect of PGE₂ on HSCs of limited self-renewal defined in our experimental model may be the result of the superior quiescence of LT-HSCs compared with ST-HSC/MPPs^57,58  and their relative resistance to injury.⁵⁹  However, in injury, a direct effect of PGE₂ on LT-HSC expansion may be undesirable as it may imperil this essential source of hematopoietic cells. This unfavorable phenomenon has been demonstrated in genetic models in which increased proliferation depletes the HSC pool.^51-54  It is indeed possible that the direct effect of PGE₂ may be targeted to ST-HSCs, with a secondary effect on LT-HSCs, which becomes apparent only when there is sustained stimulation of PGE₂-dependent signaling, as in 16-16-dimethyl PGE₂ treatment. Further studies elucidating PGE₂ receptor expression and regulation within subsets of HSCs may further address this hypothesis.

LT-HSCs are both required and sufficient to ensure lifelong hematopoietic reconstitution after complete myeloablation. However, rapid and efficient hematopoietic recovery is critical after severe insults to overcome life-threatening cytopenias, and others have suggested that such rapid reconstitution could be most efficiently provided by ST-HSCs because LT-HSCs are quiescent and have delayed production of mature progeny.⁴⁶  It is intriguing that PGE₂ is often locally produced in infection and injury, situations in which rapid stimulation of bone marrow reconstitution would be beneficial.

Our findings suggest that PGE₂ can increase ST-HSC/MPPs in the absence of myeloablative injury. This PGE₂ effect could be therefore useful in situations of immunosuppression where there has been no injury to the bone marrow, such as hemolytic disorders or treatment with glucocorticoids.

PGE₂ has been demonstrated to increase bone; however, in mice these effects have only been seen in the periosteum, where negligible amounts of osteoclastic and hematopoietic cells are present.^9,60  The intermittent treatment that increases trabecular bone in rats has not been successfully reproduced in mice. Similarly, we did not measure a net bone gain in the trabecular bone of treated mice. Rather, we demonstrated disruption of the trabecular bone microarchitecture, confirming an effect of PGE₂ in the bone microenvironment. We speculate that the lack of a net bone gain may be the result of simultaneous activation of bone formation and bone resorption, which would be consistent with the trabecular changes we observed. Further studies are needed to define more specifically the effects of PGE₂ in the bone marrow microenvironment. These microenvironmental effects may also explain how our results on preferential ST-HSC/MPPs expansion of PGE₂ in vivo differ from data on HSC treatments ex vivo.^30,31 

These results also suggest that interactions of a ligand with both HSC and niche cells may result in direct and indirect effects that impact HSC behavior. Our data cannot at this stage determine whether direct and/or microenvironmental effects of PGE₂ result in the HSC effects we observed, and we cannot exclude that in vivo PGE₂ may have a small effect on LT-HSCs given the precision of the assays we used. However, we think that our data point to aspects of the response of the bone marrow to injury that should be further explored, namely, the concept that the interaction of HSCs with their microenvironment not only provides regulatory signals but that it also must be altered in situations that disrupt the bone marrow, such as injury, infection and, as some groups are beginning to explore, malignancy.

In addition to the potential therapeutic benefit of using PGE₂ or analogs to specifically expand ST-HSCs, these data raise the possibility that anti-inflammatory drugs that block PGE₂ production may delay or restrict bone marrow recovery after myeloablation. Thus, further studies on the effect of anti-inflammatory drugs in the setting of hematopoietic recovery after chemotherapy and/or stem cell transplantation are warranted.

In contrast to some earlier studies in which PGE₂ had inhibitory effects on hematopoiesis,^24-28  we did not see any inhibition of differentiation of hematopoietic progenitors, as demonstrated by the lack of inhibitory PGE₂ effects on CFU-C, on circulating hematopoietic parameters in PGE₂-treated mice as well as at early time points in primary recipients of noncompetitive transplantation. We suspect that these differences may be the result of the PGE₂ in vivo regimen we used in these experiments.

In conclusion, we have demonstrated that PGE₂ in vivo preferentially expands ST-HSCs without depleting LT-HSCs, a strategy that may be exploited to accelerate recovery of the bone marrow after myeloablative injury. PGE₂ may be a local mediator of the effects of PTH on the bone marrow microenvironment that result in ST-HSC expansion, either by acting directly on ST-HSCs or by regulating other HSC niche components. The addition of PGE₂ or a PGE₂-receptor agonist may therefore potentiate the PTH effect on ST-HSCs. Further studies are now required to elucidate the molecular mechanisms responsible for this novel regulation of ST-HSCs/MPPs by PGE₂.

The authors thank Dr Jim Palis for helpful discussion.

This work was supported by the National Institutes of Health, National Institute of Diabetes, Digestive and Kidney Diseases (K08 DK064381, R01 DK 076876; L.M.C.), the Wilmot Cancer Scholar Foundation (L.M.C.), and the Pew Foundation (L.M.C.). R.L.P is a trainee in the Medical Scientist Training Program (National Institutes of Health T32 GM-07356).

National Institutes of Health

Contribution: B.J.F., R.L.P., J.M.W., B.J.G., and A.J.O.-S. performed experiments; B.J.F., R.L.P., L.M.C., R.J.O., and C.T.J. analyzed experiments; and B.J.F., R.L.P., and L.M.C. designed the research and wrote the manuscript.

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

Correspondence: Laura M. Calvi, Endocrine Division, Department of Medicine, University of Rochester School of Medicine, 601 Elmwood Ave, Box 693, Rochester, NY 14642; e-mail: laura_calvi@urmc.rochester.edu.