Suppression of leukemia development caused by PTEN loss

Abstract

Multiple genetic or molecular alterations are known to be associated with cancer stem cell formation and cancer development. Targeting such alterations, therefore, may lead to cancer prevention. By crossing our previously established phosphatase and tensin homolog (Pten)-null acute T-lymphoblastic leukemia (T-ALL) model onto the recombination-activating gene 1^−/− background, we show that the lack of variable, diversity and joining [V(D)J] recombination completely abolishes the Tcrα/δ-c-myc translocation and T-ALL development, regardless of β-catenin activation. We identify mammalian target of rapamycin (mTOR) as a regulator of β-selection. Rapamycin, an mTOR-specific inhibitor, alters nutrient sensing and blocks T-cell differentiation from CD4⁻CD8⁻ to CD4⁺CD8⁺, the stage where the Tcrα/δ-c-myc translocation occurs. Long-term rapamycin treatment of preleukemic Pten-null mice prevents Tcrα/δ-c-myc translocation and leukemia stem cell (LSC) formation, and it halts T-ALL development. However, rapamycin alone fails to inhibit mTOR signaling in the c-Kit^midCD3⁺Lin⁻ population enriched for LSCs and eliminate these cells. Our results support the idea that preventing LSC formation and selectively targeting LSCs are promising approaches for antileukemia therapies.

T-lymphoblastic leukemia (T-ALL) is a common hematological malignancy that is associated with poor prognosis compared with other ALLs and is often fatal without effective treatment (1). Activating mutations in Notch gene homolog 1 (NOTCH1) are reported in 34–71% of human T-ALL patients (2, 3), whereas deletion or mutations of the phosphatase and tensin homolog (PTEN). tumor suppressor gene have recently been detected in 8–63% of pediatric T-ALL patients (3–6). The mutation status of NOTCH1 and PTEN can divide pediatric T-ALL patients into three groups: (i) those with both NOTCH1 and PTEN mutations, (ii) those with NOTCH1/F-box and WD repeat domain containing 7 (FBXW7) mutations, and (iii) those with only PTEN mutations (3). Interestingly, constitutive activation of the NOTCH signaling pathway is known to down-regulate PTEN expression (5, 7), suggesting that PTEN and its controlled PI3K/v-akt murine thymoma viral oncogene homolog (AKT)/mTOR pathway are critical for the etiology of human T-ALL. Furthermore, PTEN deletion seems to be correlated with poor response to chemotherapy (6) and resistance to pharmacological inhibition of NOTCH1 (5). Therefore, understanding the molecular mechanisms of PTEN-mediated T-ALL pathogenesis and drug resistance is a critical step to improving T-ALL therapeutics.

To investigate the molecular and cellular mechanisms associated with PTEN-controlled T-ALL pathogenesis and therapeutic resistance, we have recently developed a Pten^loxP/loxP;VE-Cadherin-Cre⁺;Rosa26^{loxP-stop-loxP}-LacZ⁺ (Pten null) T-ALL mouse model by conditional deletion of Pten in a subset of fetal liver hematopoietic stem cells (8). The resulting animals develop a transient myeloproliferative disorder followed by T-ALL with 100% penetrance. Besides Pten deletion, at least two subsequent spontaneous alterations, namely β-catenin activation and a Tcrα/δ-c-myc translocation, have been identified, which lead to the transformation of T-progenitor cells to self-renewable leukemia stem cells (LSCs) enriched in the c-Kit^midCD3⁺Lin⁻ subpopulation (8). Our previous work showed that this LSC subpopulation is responsible for initiating T-ALL and repopulating and maintaining leukemia development in SCID recipients on serial transplantation (8). Interestingly, the NOTCH1 pathway is not altered in the Pten-deficient T-ALL model (8), which mimics the genetics of a subset of pediatric T-ALL patients with only PTEN mutations. In this study, we used the Pten null T-ALL mouse model to address three important issues: (i) the genetic alterations and T-cell developmental stage required for LSC formation, (ii) the therapeutic prevention and suppression of T-ALL development, and (iii) the cellular mechanism of therapeutic resistance.

Results

Recombination-Activating Gene 1 Deletion Abolishes Leukemia Development Caused by PTEN Loss.

Nearly 20% of human ALL cases involve translocations between the T-cell receptor (TCR) or Ig loci and various oncogenes. Because the activity of the recombination-activating gene (RAG) endonucleases is required for both TCR and Ig rearrangement, we genetically tested whether depletion of RAG1 activity would prevent chromosomal translocation and LSC formation by crossing the Pten null T-ALL model with Rag1^−/− mice (9). In comparison with age- and genetic background-matched Pten null T-ALL mice (n = 6) (Fig. 1A Center), the lack of RAG1-mediated V(D)J recombination and Tcrβ clonal expansion (Fig. S1 A and B) completely abolished T-ALL development in Pten^loxP/loxP;VE-Cadherin-Cre⁺;Rag1^−/−;Rosa26^{loxP-stop-loxP}-LacZ⁺ (Pten;Rag1 null) mice, as shown by CD45 side-scatter (SSC) flow cytometric analysis (n = 15) (Fig. 1A Right) for leukemic blasts (8, 10, 11) and histopathological analyses (Fig. S2). These results suggest that RAG1 activity is essential for T-ALL development in the Pten null leukemia model.

Fig. 1.

Rag1 deletion completely abolishes leukemia development caused by PTEN loss. (A) No T-ALL or other leukemia was detected in the bone marrow (BM) and thymus of P45–60 Pten;Rag1 null mice compared with Pten null T-ALL mice (n = 15 from 12 independent experiments). (B) Two-color interphase FISH analysis of thymocytes was performed with BAC clone probes (8): green-labeled RG-331N4 and RP23-357G5 on chromosome 14 and red-labeled RP24-194H23 and RP23-55P19 on chromosome 15. Fusion signals are indicated with white arrows. (C) c-myc was not overexpressed in Pten;Rag1 null CD4⁺CD8⁺ thymocytes (P60; percentage ± SD, n = 3 from three independent experiments). (D) Two million Pten;Rag1 null cells failed to develop leukemia in transplanted recipients in contrast to P30–45 Pten null preleukemic cells (representative; 3–5 mo, n = 7 from three separate experiments).

We next determined whether the Tcrα/δ-c-myc translocation or c-myc overexpression could occur without RAG1-dependent V(D)J recombination. Using two-color FISH analysis (8), we did not detect the Tcrα/δ-c-myc translocation in thymocytes from P60 Pten;Rag1 null mice (Fig. 1B), similar to the recently published Pten^loxP/loxP;Rag1^−/−;CD4-Cre⁺ model (12). However, we did not observe trisomy 14 or 15 (Fig. 1B) or activating mutations in Notch1 (Fig. S3B), which are reported in the Pten^loxP/loxP;Rag1^−/−;CD4-Cre⁺ model (12). Our intracellular flow cytometric analysis showed that the level of c-myc expression in Pten;Rag1 null CD4⁺CD8⁺ thymocytes was significantly lower than that in Pten null T-ALL blasts but comparable with WT thymocytes (Fig. 1C). These results suggest that RAG1-mediated V(D)J recombination is required for the Tcrα/δ-c-myc translocation, and Tcrα/δ-c-myc translocation-mediated c-myc overexpression is necessary for LSC formation and T-ALL development in our Pten null leukemia model. Because V-J recombination of the TCRα locus takes place in the CD4⁺CD8⁺ double-positive (DP) stage of thymocyte development (13), our results also identify Pten null DP thymocytes as the population that is susceptible to transformation by the Tcrα/δ-c-myc translocation.

We then examined the relationship between the Tcrα/δ-c-myc translocation and β-catenin activation, the two molecular events that we have identified in the Pten null model that are associated with LSC formation (8). We found that β-catenin activation is not an immediate event after either PTEN loss or Tcrα/δ-c-myc translocation-mediated c-myc overexpression, because β-catenin activation randomly occurs in Pten;Rag1 null thymocytes (Fig. S3A). Importantly, β-catenin activation alone is not sufficient to cause T-ALL in Pten;Rag1 null mice (Fig. S3A), further supporting the notion that β-catenin activation and the Tcrα/δ-c-myc translocation are both involved in LSC formation.

T-ALL was the only leukemia observed in our Pten null model (8). In contrast, both T-ALL and acute myeloid leukemia (AML) were observed in a Pten^loxP/loxP;Mx-1-Cre⁺ model after polyinosinic-polycytidylic acid (pIpC) induction (14, 15). Thus, we asked whether blocking the Tcrα/δ-c-myc translocation would lead to the development of other types of leukemia. Although SCID recipients transplanted with Pten null bone-marrow (BM) and thymic cells consistently developed T-ALL within 2 mo (8) (Fig. 1D Lower), those mice receiving Pten;Rag1 null cells did not give rise to any type of leukemia for up to 5 mo posttransplantation (Fig. 1D Upper). These results confirm the lack of leukemogenic potential of Pten;Rag1 null hematopoietic cells. The absence of myeloid leukemia in both Pten null and Pten;Rag1 null mice suggests that the AML phenotype observed in Pten^loxP/loxP;Mx-1-Cre⁺ mice may result from a higher proportion of hematopoietic stem cells (HSCs) carrying Pten deletion (14, 16), a niche-dependent effect (15, 17), and/or effects of pIpC and pIpC-induced IFN activation on HSCs (18, 19).

PTEN Loss Enables Rag1^−/− T Cells to Differentiate to DP Cells.

Although Pten;Rag1 null mice did not develop leukemia, we consistently observed enlarged thymus (Fig. 2A) (P < 0.01), which often suppressed respiratory function and resulted in premature lethality. The lack of activating Notch1 mutations (Fig. S3B), major chromosomal abnormalities and transplantation success (Fig. 1 B and C) as well as histological appearance associated with the enlarged thymus (Fig. S2A), coupled with no evidence of a leukemic phase or spread beyond thymus (Fig. 1A and Fig. S2B), are most consistent with T-cell hyperplasia. However, we cannot definitively determine the clonality of cells within the enlarged thymus without RAG and antigen receptor rearrangement (Fig. S1B). Therefore, although we strongly favor T-cell hyperplasia, we cannot use simple molecular analysis to definitively rule out the possibility of a less-aggressive thymic malignancy in Pten;Rag1 null mice. The enlarged thymus seen in Pten;Rag1 null mice was somewhat surprising, because loss of RAG activity is known to reduce thymic mass and arrest T-cell differentiation at the CD44⁻CD25⁺CD4⁻CD8⁻ [double negative 3 (DN3)] stage (9) (Fig. 2B).

Fig. 2.

PTEN loss enables Rag1 null thymocytes to bypass β-selection. (A) The mass difference of P30 and P60 Rag1 null and Pten;Rag1 null thymuses was compared by Student t test analysis (n = 4–9 per group; P values are indicated as **P ≤ 0.01 or *P < 0.05). (B) Pten;Rag1 null thymocytes differentiate into DP cells at P30–45 (n = 3–6 per group from five independent experiments; error bars represent SD). (C) FACS-Gal analysis determined that only Pten;Rag1 null thymocytes could bypass β-selection (8, 25). LacZ⁺ cells (green dots) and LacZ⁻ cells (red dots) in the same samples are overlaid. (D) Pten loss restored cell size (FSC) and nutrient potentials (CD98 and CD71) of Rag1 null DN3 cells at P30–45 (percentage ± SD, n = 4; Fig. S4).

Because PI3K/AKT mediates critical regulatory signals (pre-TCR, NOTCH, and IL-7 receptor) required for T-cell proliferation, differentiation, and survival (20, 21), an enlarged thymus may result from overproliferation, enhanced survival of Pten-deficient DN3 thymocytes (22, 23), and/or restoration of T-cell differentiation. To distinguish among these possibilities, we analyzed T-cell progenitors and their differentiated progeny in Pten;Rag1 null thymus (Fig. 2 B and C). A majority of cells in atrophic Rag1^−/− thymus were indeed DN3 T progenitors caused by β-selection blockage (9) (Fig. 2B). In contrast, PTEN loss enabled Rag1^−/− DN3 progenitors to bypass the β-selection checkpoint, as indicated by the acquisition of CD27 (Fig. S4A), a marker expressed on the surface of the T cells that have passed β-selection (24), and the presence of DP subpopulations (Fig. 2B). Our FACS-Gal analysis (flow cytometric analysis with a β-galactosidase fluorescent substrate) (8, 25) further confirmed that PTEN loss is intrinsically required for Rag1^−/− cells to bypass the β-selection checkpoint. All Pten;Rag1 null thymocytes were Pten null and marked by LacZ expression, except for a few arrested LacZ⁻ Rag1 null DN3 thymocytes with intact Pten alleles (Fig. 2C and Fig. S4B). Importantly, Pten;Rag1 null thymocytes remained negative for both Tcrβ and Tcrγδ expression (Fig. S1 C and D). Together, our results suggest that PTEN-controlled signaling pathways can promote T-cell differentiation in the absence of the pre-TCR complex.

Proliferation and differentiation from DN to DP requires extensive energy consumption. Our analysis of forward scatter (FSC) profile, CD98 (a subunit of the large neutral amino acid transporter), and CD71 (transferrin receptor) revealed that PTEN loss restores cell size and nutrient potential of Rag1 null DN3 and DP cells (Fig. 2D and Fig. S4C), which provides support for proliferation and differentiation of rescued Pten;Rag1 null thymocytes. These results suggest that nutrient regulation by PTEN may contribute to the β-selection bypass.

Rapamycin Restores the β-Selection Checkpoint in Pten;Rag1 Null Mice.

AKT activation has been reported to be involved in the regulation of β-selection (20, 21, 26, 27). Among various AKT substrates, mTOR can be activated by AKT, whereas the activated target of rapamycin complex II (TORCII) complex has phosphoinositide-dependent kinase 2 (PDK2) activity and can further activate AKT (28, 29). Therefore, we investigated the role for mTOR in β-selection regulation by treating a cohort of Rag1 null and Pten;Rag1 null mice with rapamycin (4 mg/kg per d), a specific inhibitor of mTOR, for 14 d. Although rapamycin elicited little effect on the Rag1 null thymus (P = 0.053), it dramatically decreased the total number of Pten;Rag1 null thymocytes to values similar to that of the Rag1 null thymus (Fig. 3A) (P < 0.01).

Fig. 3.

Rapamycin restores the β-selection checkpoint in Pten;Rag1 null mice. (A) Rapamycin dramatically reduced total thymocytes in Pten;Rag1 null mice after 14-d treatment (n = 4–5 per group from four independent experiments; P values of t test analysis are indicated as **P ≤ 0.01 or *P < 0.05). (B) Fourteen-day rapamycin treatment restored the β-selection checkpoint in P30–45 Pten;Rag1 null mice (n = 4 per group from four independent experiments; representative FACS plots are in Fig. S4). Error bars represent SD. (C) Rapamycin reduced nutrient potentials (CD98 and CD71) of Pten;Rag1 null DN3 thymocytes after 14-d treatment (percentage ± SD, n = 3; Fig. S5).

Further analysis revealed that rapamycin had a striking effect on T-cell differentiation. In contrast to placebo-treated mice, rapamycin restored the β-selection checkpoint and arrested Pten;Rag1 null thymocytes at the DN3 stage (Fig. 3B and Fig. S5). The immunophenotypic distribution of rapamycin-treated Pten;Rag1 null thymocytes resembled that of Rag1 null thymocytes (Fig. 3B and Fig. S5). Interestingly, inhibition of mTOR by rapamycin dramatically reduced the cell surface expression of CD98 and CD71 on Pten;Rag1 null DN3 thymocytes (Fig. 3C), indicating that PTEN-regulated nutrient sensing during β-selection is mediated by mTOR activity. Taken together, these analyses show that rapamycin can effectively reverse the β-selection bypass of Pten;Rag1 null thymocytes and identify mTOR as a critical signaling mediator in the regulation of β-selection.

Rapamycin Blocks T-Cell Differentiation and Thereby Halts T-ALL Initiation and Development.

Our study of both Pten null T-ALL mice (8) and Pten;Rag1 null mice suggests that, as a result of Tcrα/δ-c-myc translocation and Tcrα-mediated c-myc overexpression (13, 30), Pten null DP cells can be transformed into leukemia-initiating cells. Therefore, any drug that blocks T-cell differentiation from the DN to DP stage can theoretically decrease or even eliminate those translocation-susceptible DP thymocytes, preventing T-ALL occurrence. The potent effect of rapamycin on β-selection and nutrient sensing observed in Pten;Rag1 null mice prompted us to test whether rapamycin could serve as such a preventative agent.

We first determined the effect of rapamycin on thymocyte differentiation in WT and Pten null mice (Fig. 4A, blue bars and Fig. S6A). In accordance with previous reports (31, 32), the total number of WT thymocytes was slightly reduced after 2 d of rapamycin treatment but underwent a striking 10-fold reduction 3 d later. This dramatic downfall was accompanied with a reduction of DP and mature single-positive (SP) cells. However, DN and immature single-positive (ISP) cells accumulated after a transient reduction (Fig. 4A and Fig. S6A), suggesting that rapamycin arrests T-cell differentiation at the DN–ISP stage. This arrest may result from restriction of nutrient uptake (Fig. S6B). Importantly, rapamycin also effectively blocked T-cell differentiation in Pten null preleukemic thymus. Rapamycin-mediated differentiation arrest reduced the total numbers of Pten null thymocytes as well as DP and SP cells but had little effect on the DN and ISP subpopulations (Fig. 4A, red bars).

Fig. 4.

Rapamycin blocks T-cell differentiation to DP thymocytes and thereby, suppresses T-ALL development. (A) Rapamycin dramatically decreased DP cells, which are susceptible to the Tcrα rearrangement and Tcrα/δ-c-myc translocation, and thereby, reduced the chance of translocation and LSC development (n = 4–5 per group from three independent experiments). P values of Student t test analysis are indicated as **P ≤ 0.01 or *P < 0.05, and error bars represent SD (Fig. S5). (B) Long-term rapamycin treatment on preleukemic Pten null mice suppressed T-ALL development, as shown by the Kaplan–Meier survival curves (Left) and a CD45-SSC flow cytometrianalysis of WT and preleukemic mutant mice treated with rapamycin for 80 d (Right). (C) Leukemia occurs in some Pten null mice treated as in B after rapamycin withdrawal. Leukemia mice analyzed for c-myc overexpression (Right) are highlighted with red stars on the Kaplan–Meier survival curves (Left; Fig. S7).

To evaluate a preventative effect of rapamycin on Pten null T-ALL development, we treated cohorts of WT and preleukemic Pten null mice daily with rapamycin (4 mg/kg per d) for 3–4 mo. As a result of this long-term treatment, rapamycin completely blocked T-cell differentiation to DP cells (Fig. S7A) and halted leukemia progression from the preleukemic phase, as indicated by the Kaplan–Meier survival curve and undetectable levels of leukemic blasts in peripheral blood and hematopoietic organs (Fig. 4B). Intracellular flow cytometric analysis showed that the levels of phosphorylated S6 (P-S6), a surrogate marker for mTOR activity, were significantly reduced in rapamycin-treated WT and Pten null CD3⁺ thymocytes as well as in Gr-1⁺Mac-1⁺ myeloid cells (Fig. S7B).

To investigate whether long-term rapamycin treatment is able to cure the disease, we withdrew the drug from those WT and preleukemic mice that had been treated with rapamycin for 3–4 mo and monitored leukemia development for an additional 3 mo. Three treated Pten null mice remained leukemia-free after rapamycin withdrawal (Fig. 4C Left), suggesting that the immunosuppressive activity of rapamycin is sufficient in preventing T-ALL development (14). However, six of nine Pten null mice rapidly became sick and died within 1 mo of drug withdrawal (Fig. 4C Left). We managed to catch two of six sick mice before death, and our detailed analysis on these samples showed molecular and immunophenotypic characteristics of T-ALL, including increased T cell-specific c-myc overexpression and leukemia blasts, similar to untreated Pten null leukemic mice (Fig. 4C Right). These observations lead us to hypothesize that LSCs must have existed in these mice before rapamycin treatment was initiated and contribute to resistance to rapamycin treatment. To test this hypothesis, a cohort of Pten null T-ALL mice (n = 6) was treated with rapamycin at a dose of 4 mg/kg per d. Although this treatment significantly prolonged the animal's lifespan by 40 d, all of the animals eventually relapsed and died of T-ALL (Fig. S7C).

Rapamycin Has Little Effect on Pten Null LSCs.

Although cancer stem cells (CSCs) have been proposed to be responsible for therapeutic resistance in cancer, very few studies have analyzed cancer stem cells after treatment and directly investigated the underlying mechanisms of resistance (e.g., whether a drug can reach and effectively inhibit its target within the cancer stem cells). We have previously characterized the self-renewable LSCs enriched in the c-Kit^midCD3⁺Lin⁻ compartment in Pten null T-ALL mice, which were capable of initiating and maintaining T-ALL in SCID recipients for up to four passages (8). The presence of rapamycin-resistant LSCs in some of our treated animals provided us a unique opportunity to directly assess the role of LSCs in treatment resistance.

We first tested whether rapamycin could reach its intracellular target mTOR in the LSCs. For this, we transplanted 1 × 10⁴ Pten null leukemic cells into NOD-SCID-IL2Rγ^−/− recipients (33) to synchronize disease development (8). Rapamycin treatment was initiated after leukemia establishment and continued for 3 d. c-Kit^midCD3⁺Lin⁻LSCs and c-Kit⁻CD3⁺Lin⁻ leukemia blasts in BM of treated recipients were harvested and measured for the intracellular levels of P-S6. As shown in Fig. 5A and Fig. S7D, rapamycin significantly reduced P-S6 level, as measured by the percentage of P-S6⁺ cells and median of P-S6 fluorescence intensity, in leukemia blasts. In contrast, rapamycin had little effect on the level of P-S6 in the LSC subpopulation. These data suggest that Pten null LSCs and leukemia blasts have remarkably differential responses to rapamycin.

Fig. 5.

Pten null LSCs are resistant to rapamycin treatment. (A) Rapamycin treatment reduced P-S6 levels (as measured by the percentage of P-S6⁺ cells and P-S6 fluorescence median) in c-Kit⁻CD3⁺Lin⁻ T-ALL blasts but not in c-Kit^midCD3⁺Lin⁻ LSCs (three independent experiments with 6,929–49,342 LSCs in placebo-treated samples and 474–64,808 LSCs in rapamycin-treated ones). P-S6 fluorescence medians are normalized to those for CD3⁻ cells as fold changes in corresponding experiments. (B) Rapamycin treatment results in the increased ratio of LSCs to leukemia blasts (three independent experiments). (C) Schematic illustration of the roles of RAG1 and rapamycin in blocking Tcrα/δ-c-myc translocation, LSC formation, T-cell differentiation, and leukemia development. P values of Student t test analysis are indicated as **P ≤ 0.01 or *P < 0.05, and error bars represent SD.

We then compared the ratio of LSCs and blasts in placebo- or rapamycin-treated primary Pten null T-ALL mice. In the presence of rapamycin, leukemia blasts were dramatically decreased, and the ratio of LSCs to blasts was increased by more than fivefold after 7 d of rapamycin treatment (Fig. 5B). This result raises the possibility that the failure of rapamycin to target mTOR within LSCs may cause therapeutic resistance in our model, although the function of these LSCs needs to be further tested by limiting dilution and serial transplantation experiments. Of course, we cannot rule out the possible acquisition of drug-resistant mutations (34) most notably shown by characterization of Gleevec resistance in human chronic myelogenous leukemia (35).

Discussion

PTEN is the second most frequently deleted tumor suppressor in human cancers, and understanding how PTEN deficiency contributes to leukemia development and LSC formation has general relevance to cancer biology. Our previous study (8) showed that the phenotypes associated with the Pten null T-ALL model are consistent with the clinical subtype of pediatric T-ALL that has no NOTCH1 and FBXW7 mutations but PTEN mutations. Our current study shows the essential role of RAG1 activity in the Tcrα-c-myc translocation and the requirement of this event in LSC formation and leukemogenesis in the Pten null T-ALL model. Although Pten deletion and β-catenin activation seem to result in thymic hyperplasia, they are unable to transform T-progenitor cells to LSC and cause T-ALL development in the absence of c-myc overexpression. Interestingly, without RAG activity and the Tcrα/δ-c-myc translocation, PTEN loss alone is not sufficient for leukemia development in other hematopoietic lineages, supporting the notion that phenotypes associated with PTEN loss may be lineage- or developmental stage-dependent.

Understanding PTEN function in the maintenance of genomic integrity is fundamentally important in cancer cell biology. Our study suggests that PTEN may play an important role in regulating nonhomologous end joining, the last step of RAG-dependent V(D)J recombination, and the Tcrα-c-myc translocation observed in the Pten null T-ALL model is fully dependent on RAG1-mediated recombination (12). Therefore, the roles for PTEN in maintenance of genomic integrity and leukemogenesis are clearly different from those of p53, whose loss results in both RAG-dependent and -independent genetic alterations and leukemogenesis (36).

Our results provide experimental evidence for the essential role of PTEN/PI3K/AKT signaling in β-selection and T-cell development (27, 37–39) and identify mTOR as a key molecule downstream of PI3K/AKT signaling in this process. mTOR signaling is likely critical for the metabolic needs of β-selected thymocytes during their extensive expansion and differentiation (27). A recent study suggests an outside-in mechanism of nutrient regulation of mTOR activity and autophagy through bidirectional transport of l-glutamine and l-leucine (40). Our work reveals an inside-out mechanism that regulates nutrient transporter activity and cell survival and differentiation through mTOR activity. This bidirectional control between mTOR activity and nutrient sensing and uptake may play a critical role for cell survival, proliferation, and differentiation.

It is a challenge to cancer biologists to develop approaches that therapeutically target cancer stem cells and prevent their formation. Based on our current study of Pten null and Pten;Rag1 null mice, we reveal a mechanism that the immunosuppressive activity of rapamycin can block differentiation of T-cell progenitors from the DN to DP stage, in which cells are susceptible to the Tcrα/δ-c-myc translocation, and thereby effectively prevent LSC formation and leukemia development in Pten null preleukemic mice. This result is consistent with and provides a mechanistic explanation of data obtained from rapamycin treatment of Pten^loxP/loxP;Mx-1-Cre⁺ mice at the preleukemia stage (14).

Cancer stem cells have been proposed to be responsible for therapeutic resistance. However, very few studies have analyzed cancer stem cells after treatment and directly investigated the underlying mechanisms of resistance. Our studies show that LSCs and leukemic blasts in T-ALL mice show remarkably different responses to rapamycin. Although rapamycin eliminates leukemia blasts, it cannot inhibit mTOR signaling in LSCs and eliminate these disease-initiating cells, favoring the notion that failure to eradicate CSCs is one of the cellular mechanisms responsible for therapeutic resistance (41). In summary, the therapeutic paradigm in preventing LSC formation, together with more effective LSC-targeting strategies, will potentially improve therapeutic interventions against leukemia initiated by LSCs and in general, cancers originated from cancer stem cells.

Materials and Methods

Mice.

Pten^loxP/loxP;VE-Cadherin-Cre⁺;Rosa26^{loxP-stop-loxP}-LacZ⁺ 129/BALB/c mice were previously generated (8) for Pten deletion in fetal hematopoietic stem cells (42–44) and crossed with Rag1^−/− C57BL/6 mice (9) to obtain Pten^loxP/loxP;Rag1^−/−;VE-Cadherin-Cre⁺;Rosa26^{loxP-stop-loxP}-LacZ^+/− mice. Genotyping (Fig. S1A) and animal care were described previously (8).

Repopulating Assay.

Cells were harvested from the BM and thymus of age-matched Rag1 null, Pten;Rag1 null, or Pten null mice. Two million donor cells were injected into sublethally irradiated SCID mice as described previously (8). Leukemia development was monitored monthly by peripheral blood smear, histopathological, CD45-SSC, and FACS-Gal analyses.

Rapamycin Treatment.

Mice were randomized to either placebo (5.2% PEG 400 + 5.2% Tween 80 vehicle) or rapamycin treatment (4 mg/kg per d; LC Laboratories), as previously described (45) and subject to subsequent flow cytometric analysis. i.p. injections of the appropriate substance were given daily for the length of time indicated in the text. For the response of LSCs and leukemia blasts to rapamycin, NOD-SCID-IL2Rγ^−/− recipients were transplanted with 1 × 10⁴ Pten null leukemic cells for 25 d (33) and subject to 3-d rapamycin treatment and a subsequent analysis of P-S6 in BM cells.

Flow Cytometry.

Analysis was performed on a FACSCanto II or LSRII (BD Biosciences) for cells stained with the fluorescence-conjugated antibodies detailed in SI Materials and Methods. Leukemic blasts were analyzed on CD45/SSC plots as described previously (8, 10). Fluorescein di(β-d-galactopyranoside) (FDG) was obtained from Sigma for the FACS-Gal analysis described previously (8, 25). Intracellular flow cytometric analysis with rabbit polyclonal anti–c-myc, monoclonal antiphospho-S6 (61H9; Cell Signaling Technology), mouse monoclonal anti–β-catenin (unphosphorylated, clone 8E7) antibody (Millpore), or control rabbit and mouse IgG (Jackson ImmunoResearch Laboratories) was described previously (8).

FISH.

The method was described previously (8).

Statistical Analysis.

Experimental data were collected and analyzed for the means and SDs indicated in Figs. 1–5. The difference of experimental groups was compared by Student t test analysis with the P values indicated in Figs. 1–5 and Figs. S1–S7. Animal survival was determined by the Kaplan–Meier survival method.