New therapeutic approaches are needed to treat leukemia effectively. Dietary restriction regimens, including fasting, have been considered for the prevention and treatment of certain solid tumor types. However, whether and how dietary restriction affects hematopoietic malignancies is unknown. Here we report that fasting alone robustly inhibits the initiation and reverses the leukemic progression of both B cell and T cell acute lymphoblastic leukemia (B-ALL and T-ALL, respectively), but not acute myeloid leukemia (AML), in mouse models of these tumors. Mechanistically, we found that attenuated leptin-receptor (LEPR) expression is essential for the development and maintenance of ALL, and that fasting inhibits ALL development by upregulation of LEPR and its downstream signaling through the protein PR/SET domain 1 (PRDM1). The expression of LEPR signaling-related genes correlated with the prognosis of pediatric patients with pre-B-ALL, and fasting effectively inhibited B-ALL growth in a human xenograft model. Our results indicate that the effects of fasting on tumor growth are cancer-type dependent, and they suggest new avenues for the development of treatment strategies for leukemia.
Dietary restriction, including fasting, delays aging and has pro-longevity effects in a wide range of organisms, and so has been considered for cancer prevention and the treatment of certain solid tumor types1,2,3,4,5,6. Fasting can promote hematopoietic stem cell–based regeneration and reverse immunosuppression7,8,9, and has been reported to promote the anti-cancer effects of chemotherapy5,10. However, the responsiveness of hematopoietic malignancies to dietary restriction, including fasting, remains unknown.
AML is the most common form of adult acute leukemia, whereas ALL is the most common form of cancer in children; ALL also occurs in adults11,12,13. Although treatment of pediatric ALL is highly effective, a sizeable number of patients are nonresponders who succumb to this disease. The outcome of ALL in adults is substantially worse than for pediatric ALL, with a 5-year survival rate of approximately 40%12. Additionally, some types of ALL have a much poorer prognosis than others12. New therapeutic targets and approaches need to be identified to treat these leukemias more effectively. Here we investigated whether and how fasting regulates the development of B-ALL, T-ALL and AML.
Fasting selectively inhibits the development of ALL but not AML
To extend our previous work on the extrinsic and metabolic regulation of hematopoietic stem cells and cancer development14,15,16,17,18,19,20,21,22, we studied the effects of fasting on leukemia development. Mice from several retrovirus transplantation acute leukemia models, including the N-Myc B-ALL model23, the activated Notch1 T-ALL model24 and the MLL-AF9 AML model25,26, were placed on various dietary regimens. Strikingly, a regimen consisting of six cycles of 1 d of fasting, followed by 1 d of feeding, implemented 2 d after transplantation (Fig. 1a) completely inhibited B-ALL development. The fasted mice had 32.85 ± 5.16, 11.31 ± 5.42 and 0.48 ± 0.12% of leukemic GFP+ cells in peripheral blood (PB) at 3, 5 and 7 weeks post-transplantation, respectively, as compared to 49.52 ± 5.75, 56.27 ± 9.36 and 67.68 ± 8.39% of GFP+ cells of the control mice (Fig. 1b,c). Concordantly, the percentages of leukemic cells in the bone marrow (BM) and spleen (SP) and the numbers of white blood cells (WBCs) in PB were also dramatically lower in the fasted mice at 7 weeks post-transplantation (Fig. 1c,d).
Next, we measured the distribution of B lymphoblastic cells and myeloid cells in the GFP+ compartment of the B-ALL mice. Control mice with B-ALL had 65–80% of B220+ cells (pan B lineage marker) and 0.5–2% of Mac-1+ cells (myeloid lineage marker) in GFP+ fractions of PB, BM and SP (Fig. 1e,f), indicative of fully developed B-ALL. By contrast, there were only 19–28% of B220+ cells and 5–12% of Mac-1+ in GFP+ fractions in fasted mice (Fig. 1e,f), consistent with loss of the B-ALL phenotype. There was also a higher percentage of differentiated B220+IgM+ cells in the GFP+ compartment in fasted mice than in fed mice (Fig. 1g). These results suggest that fasting blocks B-ALL development by decreasing malignant B cell propagation and enhancing B cell differentiation.
In the fasted mice, the infiltration of B-ALL cells into the spleen and lymph nodes, as assessed by organ size, was also significantly decreased, such that the size of the spleen and lymph nodes in fasted mice was similar to that in normal mice (Fig. 1h). Fed, but not fasted, leukemic mice showed tissue destruction in the spleen and liver (Supplementary Fig. 1a). All fed mice with B-ALL died within 59 d after transplantation. In a striking contrast, 75% of their fasted counterparts survived for more than 120 d without any leukemia symptoms (Fig. 1i and Supplementary Fig. 1b-d). The small fraction of GFP+ cells in these surviving mice behaved similarly to normal cells; secondary transplantation of these cells did not result in leukemic symptoms (Supplementary Fig. 2). We conclude that fasting dramatically inhibits B-ALL development.
In the Notch1 T-ALL model, six cycles of fasting resulted in a dramatic decrease of T-ALL development (Fig. 1j–m and Supplementary Fig. 3a). Fasting lowered the percentage of leukemic GFP+ in PB, BM and SP over time (Fig. 1j), decreased WBC counts in the PB and maintained normal spleen and liver structure (Fig. 1k and Supplementary Fig. 3a). Fasted mice had a lower percentage of CD3+ cells (T lineage marker) and a higher percentage of myeloid Mac-1+ cells in the GFP+ compartment, as compared to fed mice (Fig. 1l), and fasting significantly prolonged the survival of the T-ALL mice (Fig. 1m). Thus, this fasting regimen can block the development of T-ALL as well as B-ALL.
In striking contrast to the B-ALL and T-ALL models, fasted mice with MLL-AF9 AML did not show reductions in the percentages of leukemic YFP+ cells, Mac-1+ cells in the YFP+ compartment or Mac-1+Kit+ AML progenitor cells, or in the colony-forming activity of YFP+ BM cells (Fig. 1n–p and Supplementary Fig. 3b). Fasting significantly decreased the percentage of B220+ B lineage cells in the YFP+ compartment (Fig. 1o) and shortened the overall survival time of mice with AML (Fig. 1q). Similarly, fasting did not alter leukemia development in another AML mouse model, driven by the AML1-Eto9a oncogene27 (Supplementary Fig. 4). Thus, fasting inhibits the development of ALL but not AML in mice.
Fasting inhibits ALL development at both early and late stages
We sought to understand whether fasting blocks ALL at tumor initiation or later during leukemia development. First, we tested the effects of varying the numbers of 1-d-fasting–1-d-feeding (1F) fasting cycles, started at day 2 after transplantation, on B-ALL development (Fig. 2a). Every increase of a fasting cycle for up to six cycles resulted in an improvement in leukemia inhibition; a four-cycle fasting regimen essentially eliminated B-ALL. With increased numbers of fasting cycles, we observed augmented decreases in the percentage of leukemic GFP+ cells in PB, BM and SP, in the numbers of WBCs in PB and in leukemia burden rates, together with an increase in B-ALL mouse overall survival (Fig. 2b–e). We also tested another fasting strategy with one to four cycles of 2-d-fasting–2-d-feeding cycles (2F) (Fig. 2a). We found that two cycles of 2F significantly reduced the development of B-ALL, and that three or four 2F cycles had equivalent effects to six cycles of 1F with regard to complete inhibition of B-ALL development, such that 80% of the mice survived longer than 120 d (Fig. 2f–i). Clearly, fasting at the tumor initiation stage effectively inhibits B-ALL development, and the level of inhibition correlates with the number of cycles of fasting.
Next, we tested the effect of fasting at a middle-to-late stage of B-ALL development. We began the fasting regimens—either two and four cycles of 1F fasting (hereafter 2-1F/4-1F) or two and three cycles of 2F fasting (hereafter 2-2F/3-2F) —when the percentage of leukemic GFP+ cells in PB of primary recipients had reached ∼60% (Fig. 2a). Whereas the percentage of GFP+ cells in PB rose over time in control mice, this percentage decreased dramatically at 2 and 3 weeks after the initiation of fasting in the 4-1F and 2-2F/3-2F groups, as did the percentage of GFP+ cells in BM and SP at 3 weeks post-fast (Fig. 2j). Analysis of WBC counts also showed that the fasting regimen reversed B-ALL progression (Fig. 2k). The fasted mice in the 4-1F and 3-2F groups had decreased leukemia burden rates as compared to control mice (Fig. 2l) and lived longer than did the control mice (Fig. 2m); whereas all control B-ALL mice were dead by day 58 after transplantation, more than 60% of the fasted mice survived more than 120 d. These results indicate that fasting at the middle-to-late stage of B-ALL is capable of reversing leukemia development.
To further evaluate how fasting impacts N-Myc-induced B-ALL development at a late stage of the disease, we tested the effects of the 2-IF/4-1F or 2-2F/3-2F fasting protocols in mice that received secondary transplantations (Fig. 2a). We found that the 4-1F or 3-2F fasting protocols inhibited B-ALL development in more than 40% of B-ALL mice at this very late stage. At 7 weeks after transplantation, the percentage of leukemic GFP+ cells in PB, BM and SP, as well as WBC counts in PB, decreased significantly in fasted mice as compared to the fed controls (Fig. 2n,o). Moreover, the fasted mice had a lower percentage of B220+ cells and a higher percentage of Mac-1+ cells in the GFP+ compartment than did controls (Supplementary Fig. 5), and had decreased leukemia burden rates and increased life spans (Fig. 2p,q), indicating that fasting inhibits B-ALL development in mice that have undergone secondary transplantations. We performed a similar set of experiments in the Notch1-induced T-ALL model and obtained similar results (Supplementary Fig. 6). Together, these results indicate that fasting blocks the development of B-ALL and T-ALL at both the tumor initiation stage and later stages of the disease.
Fasting induces rapid ALL differentiation and loss and upregulation of LEPR
To study the kinetics of fasting effects on ALL development, we fasted mice with B-ALL at the time at which leukemia had progressed to the mid-to-late-stage (∼60% GFP+ cells in PB). A single 48-h fast induced a significant reduction in the percentage of GFP+ cells in PB, BM, SP and LV at day 2 and day 5 after the initiation of fasting (Fig. 3a,b). In fasted mice, the percentage of B220 cells decreased greatly in the GFP+ leukemic cell compartment, but not in the GFP− nonleukemic cell compartment (Fig. 3c and Supplementary Fig. 7a). GFP+ leukemic cells had lower rates of apoptosis and proliferation than GFP− cells in control mice; fasting increased apoptosis and proliferation rates in GFP+ cells but decreased both in GFP− cells (Fig. 3c and Supplementary Fig. 7b), indicating that ALL cells and nonleukemic cells respond differently to fasting treatment. Fasting did not have effects on levels of the senescence marker SA-β-gal in GFP+ cells (Supplementary Fig. 7c). With respect to ALL cell differentiation, GFP+ cells in fasted mice were more differentiated than those in control mice, with significantly increased expression of B cell maturation markers, including surface IgM, Ig κ and Ig λ, and reduced expression of immature B cell markers including terminal deoxynucleotidyl transferase (TDT) and surface CD43 (Fig. 3d). With respect to effects on metabolism, 48-h fasting (2F), 24-h fasting (1F) and two cycles of 24-h fasting (2-1F) showed similar effects, including a decrease in circulating glucose and insulin levels, lower levels of insulin-like growth factor (IGF)-1 and leptin in both PB and BM, and an increase in insulin-like growth factor binding protein 1 (IGFBP1) levels in both PB and BM. In many cases, these effects recovered to pre-fasting levels more slowly in the 2F and 2-1F conditions than in the 1F condition, an observation that is consistent with the severity of GFP+ cell reduction in these regimens (Supplementary Fig. 7d–g). Fasting did not alter the homing properties of transplanted B-ALL cells, nor did fasting alter myeloid or lymphoid lineage distributions in the BM of wild-type mice (Supplementary Fig. 8). Taken together, these results indicate that fasting induces rapid differentiation and loss of ALL cells.
To study the underlying mechanisms, we collected GFP+B220+ B-ALL cells from control mice and from 1-d or 2-d fasted mice by flow cytometry for RNA-seq analysis. Consistent with the concept that fasting promotes B-ALL differentiation, the RNA-seq data indicated that the B cell terminal-differentiation program was initiated rapidly by fasting. The levels of mRNA or activity (as inferred by upstream regulator analysis in Ingenuity Pathway Analysis (IPA) tools) of most of the signature transcription factors of B cell terminal differentiation, including the key factors Prdm1, Xbp1 and Irf4 (refs. 28,29), were upregulated by fasting (Fig. 3e and Supplementary Table 1). Pathway analysis showed that the pathways of cytokine–cytokine receptor interaction, as well as the JAK–STAT pathway that acts downstream of cytokine receptors, were altered significantly; in particular, STAT3, a key player in the JAK–STAT pathway, was inferred to be highly activated by fasting (Fig. 3f and Supplementary Tables 2 and 3). Consistent with the inhibitory effects of fasting on N-Myc B-ALL and Notch1 T-ALL phenotypes, the inferred activities of N-Myc and Notch1, as well as the inferred activities of the oncogenes Myc and Bcl3, were reduced profoundly by fasting (Fig. 3f and Supplementary Table 3). Moreover, fasting reduced the protein levels of N-Myc in both B-ALL and T-ALL GFP+ cells as compared to the fed condition (Supplementary Fig. 9). These results suggest that upregulation of cytokine-receptor signaling might be responsible for the inhibitory effects of fasting on ALL.
Consistent with this idea, we found that the expression levels of many cytokine receptors were upregulated in GFP+B220+ B-ALL cells from fasted mice (Fig. 3g). Among the cytokine receptors showing increased expression, we selected the leptin receptor (LEPR) as a candidate for mediating the antileukemic effects of fasting, on the basis of two criteria: its strong upregulation in cells from fasted mice and the negative log2 hazard ratio of its expression to survival rates of human patients with pre-B-ALL (that is, a positive correlation between LEPR expression level and overall survival) (Fig. 3g and Supplementary Table 4). Surface expression of LEPR on GFP+ cells from fasted mice increased from day 1 to day 2 after fasting and was maintained thereafter (Fig. 3h). According to mRNA analysis, the long isoform of LEPR, Ob-Rb, was more highly expressed than the short isoform, Ob-Ra (Fig. 3i), and the level of Ob-Rb mRNA in leukemic cells was increased by fasting (Fig. 3j). Moreover, the levels of phospho-STAT3, the major effector of LEPR30,31, were increased in leukemic cells by fasting (Fig. 3k). The fasting-induced increase in LEPR expression was more pronounced in leukemic GFP+ B220+ cells than in GFP+ B220− cells, and these relative effects were reversed in nonleukemic GFP− cell populations (Fig. 3l). Ob-Rb was also the more highly expressed isoform in T-ALL and AML cells, and the ratio of Ob-Rb/Ob-Ra expression in these cells was not affected by fasting (Supplementary Fig. 10a,b). Moreover, fasting induced rapid upregulation of surface LEPR levels in Notch-transformed T-ALL cells but not in MLL-AF9-transformed AML cells (Supplementary Fig. 10c). Notably, surface LEPR expression was lower on nonfasted B-ALL and T-ALL cells than on normal B and T lineage cells, respectively, whereas surface LEPR expression on AML cells was higher or equal to those on normal myeloid-lineage cells (Fig. 3m). Fasting did not alter the expression of surface LEPR on normal myeloid and lymphoid cells in nontumor bearing mice (Supplementary Fig. 10d), indicating that the effect of fasting on LEPR expression is specific to ALL cells. Taken together, these results demonstrate that fasting induces upregulation of LEPR and downstream signaling in ALL but not AML cells, which is consistent with the effects of fasting on leukemia development in these mouse models.
Fasting is known to reduce the level of plasma leptin32, and adipocytokines have been proposed to affect tumor growth through local and micro-environmental regulation33,34. We found that fasting decreased the levels of not only circulating leptin, but also leptin in BM (Supplementary Fig. 7e,f). After fasting, the rate at which the leptin level returned to the pre-fasting level was slower in BM than in PB, and this effect was more pronounced in the 2F or 2-1F regimens than in the 1F regimen. The rate at which leptin levels recovered after fasting was concordant with the clearance rate of ALL cells (Supplementary Fig. 7e–g), which suggests that leptin might have roles in ALL inhibition at both the systemic and local levels. Moreover, the administration of leptin to B-ALL mice reversed fasting-induced upregulation of LEPR (Fig. 4a), which suggests that LEPR upregulation upon fasting occurs in response to leptin downregulation. Notably, the effect of fasting on decreasing the percentage of leukemic GFP+ cells was partially but significantly rescued by leptin administration (Fig. 4a), suggesting that LEPR upregulation in ALL cells accounts for the antileukemic effects of fasting.
Attenuated LEPR signaling is essential for ALL development
To study the role of LEPR in the effects of fasting on leukemia development, we first used LEPR-deficient mice (hereafter Leprdb/db mice) to investigate whether attenuation of LEPR signaling is essential for ALL development35,36. Using these mice, we previously demonstrated that LEPR signaling promotes breast cancer development and metastasis through suppression of mitochondrial respiration18,19. We transplanted N-Myc-infected Lepr+/+ and Leprdb/db BM Lin− cells into wild-type recipient mice. As compared to mice transplanted with Lepr+/+ cells, mice transplanted with Leprdb/db cells showed a significantly faster rise in the percentage of GFP+ B-ALL cells and showed an increased percentage of GFP+ cells in the PB, BM and SP, as well as increased WBC counts (Fig. 4b,c). Moreover, within the GFP+ compartment, B220+ Leprdb/db cells arose more rapidly than B220+Lepr+/+ cells, and Leprdb/db cells showed a higher percentage of B220+ CD43+ precursors but a lower percentage of B220+IgM+ differentiated cells than B220+ Lepr+/+ cells (Fig. 4d,e). These results suggest that LEPR deficiency blocks the differentiation of B-ALL cells. Furthermore, Leprdb/db B-ALL mice had a significantly lower overall survival rate than control mice (Fig. 4f).
We also examined the effects of LEPR deficiency on T-ALL and AML development. Similarly to the situation in B-ALL, transplantation of Leprdb/db T-ALL cells led to faster leukemic progression as compared to transplantation of Lepr+/+ T-ALL cells (Supplementary Fig. 11). By contrast, transplantation of Leprdb/db AML cells led to slightly slower leukemic progression than did the transplantation of Lepr+/+ AML cells (Supplementary Fig. 12). These results suggest that LEPR negatively regulates ALL development but not AML development, which is consistent with the pattern of LEPR expression in ALL and AML cells.
Next, we used leptin-deficient (Lepob/ob) mice19,37,38,39 or wild-type mice as the recipients of leukemic cell transplantation to test the effect of LEPR-ligand deficiency on B-ALL development. B-ALL developed faster in Lepob/ob recipient mice than in wild-type mice (Supplementary Fig. 13a,b); moreover, the percentage of B220+IgM+ cells in the GFP+ compartment was substantially lower in Lepob/ob than in wild-type recipients, especially in the early stages of disease (Supplementary Fig. 13c), which suggests that B-ALL cells in Lepob/ob mice are less differentiated than those in wild-type mice. Lepob/ob B-ALL mice did not survive as long as wild-type mice (Supplementary Fig. 13d). The exacerbation of leukemia development in Lepob/ob B-ALL mice was reversed by leptin administration (Supplementary Fig. 13a–d). By contrast, AML development was modestly delayed in Lepob/ob recipient mice as compared to that in wild-type mice (Supplementary Fig. 14). Thus, the loss of leptin–LEPR signaling because of either LEPR deficiency in donor cells or leptin deficiency in recipient mice accelerates ALL development, consistent with the notion that attenuated LEPR expression is essential for ALL development.
LEPR is required for fasting-induced inhibition of ALL development
To determine whether LEPR is required for fasting-induced inhibition of ALL development, we transplanted Lepr+/+ and Leprdb/db B220+ B-ALL cells into WT recipient mice; once ∼60% of cells in the PB were GFP+, the mice were fasted for 48 h (Fig. 4g). In mice transplanted with Lepr+/+ cells, fasting led to a drop in the percentage of GFP+ cells in PB from 71.77 ± 6.15% pre-fasting to 1.16 ± 0.69% at day 2, followed by a rise to 2.93 ± 0.80% at day 3 and 63.47 ± 11.52% at day 5 (Fig. 4h). However, the same fasting regimen assigned to Leprdb/db B-ALL mice resulted in a significantly attenuated decrease in the percentage of GFP+ cells in PB at days 2 and 3 (Fig. 4h). Moreover, whereas essentially all Lepr+/+ B-ALL cells had differentiated into IgM+ cells at days 2 and 3 after the initiation of fasting, Leprdb/db B-ALL cells did not undergo differentiation (Fig. 4i). These results indicate that a lack of LEPR in B-ALL cells confers resistance to fasting.
In parallel, we carried out a fasting regimen consisting of two cycles of 2-d fasting/2-d feeding at day 2 after transplantation (Fig. 4g). Similarly to the results with one 48-h fast, LEPR deficiency in ALL cells resulted in the development of resistance by mouse recipients to the effects of two cycles of fasting (Fig. 4j–m). Mice transplanted with Leprdb/db T-ALL cells were also resistant to the antileukemic effects of both fasting regimens (Supplementary Fig. 15). Collectively, our results indicate that LEPR deficiency abrogates the fasting-induced block of ALL development, consistent with the idea that LEPR upregulation on ALL cells is responsible for the inhibitory effects of fasting on ALL.
LEPR induces ALL differentiation through PRDM1
To study how LEPR upregulation might inhibit ALL, we overexpressed LEPR in B-ALL cells. LEPR overexpression significantly elevated the expression of surface IgM and light chains λ and κ, and significantly decreased TdT expression, indicating that LEPR induces B-ALL differentiation (Fig. 5a). Among the key transcription factors required to activate B cell terminal differentiation28,29, LEPR overexpression significantly upregulated the mRNA and protein levels of XBP1, and especially, PRDM1 (Fig. 5b,c and Supplementary Fig. 16a), although no changes in their mRNA levels were observed in B220+ BM cells of ob/ob or db/db mice (Supplementary Fig. 16b). PRDM1 expression can be activated by STAT3 and STAT5 (ref. 28), which are major effectors of LEPR signaling30,31. In addition to its role in B cell terminal differentiation, PRDM1 is also a crucial factor for the differentiation of terminal effector T cells, and the gene encoding PRDM1 is a tumor-suppressor gene in lymphoid malignancies28,40,41. Knockdown of Prdm1 reversed Lepr induced B-ALL differentiation in vitro (Fig. 5d and Supplementary Fig. 16c), which suggests that PRDM1 mediates the effect of LEPR on the promotion of ALL cell differentiation.
Next, we overexpressed LEPR (the Ob-Rb isoform) in mouse B-ALL cells (BM cells of mice transplanted with N-Myc-expressing fetal liver cells), and followed up with secondary transplantation. LEPR overexpression significantly inhibited B-ALL development, as evidenced by decreases in the percentage of GFP+ cells and WBC counts, increased B-ALL cell differentiation and a prolongation of mouse survival as compared to the control vector (Fig. 5e–h). By contrast, LEPR overexpression had little effect on MLL-AF9-driven AML development (Supplementary Fig. 17). Similarly to LEPR overexpression, PRDM1 overexpression in mouse B-ALL cells significantly inhibited B-ALL development (Fig. 5i–l). Additionally, we co-expressed shRNAs targeting Prdm1 with Lepr in mouse B-ALL cells, and found that Prdm1 knockdown blocked the ability of Lepr overexpression to inhibit B-ALL (Fig. 5m–p). This rescue experiment confirms that PRDM1 has an important role downstream of LEPR in regulating B-ALL development.
We also tested the effects of LEPR overexpression when co-infected with N-Myc in primary fetal liver cells. In this setting, LEPR overexpression abrogated B-ALL development (Supplementary Fig. 18a and b). As a control, mice transplanted with LEPR-infected normal fetal liver cells showed only modest changes in myeloid and lymphoid cell percentages (Supplementary Fig. 18c). Little variation in the distribution of these lineages was observed between wild-type and LEPR-deficient (db/db) mice under either fed or fasted conditions (Supplementary Fig. 18d), and the distribution of these lineages was not substantially affected in the nonleukemic GFP− cell compartment of mice transplanted with either Lepr+/+ or Leprdb/db B-ALL or T-ALL cells (Supplementary Fig. 18e).
Fasting and LEPR signaling inhibit human ALL development
To study the importance of LEPR in human leukemia development, we first performed an in silico analysis of LEPR mRNA expression in more than 2,000 human leukemia and normal BM samples (GSE13159). LEPR was expressed at significantly lower levels in all types of lymphoid leukemia cells tested as compared to normal BM samples, whereas LEPR was expressed at similar levels in normal and myeloid-leukemia cells (Fig. 6a). Consistent with these findings, flow cytometry analysis showed that human B-ALL and T-ALL cells, but not AML cells, expressed significantly less cell surface LEPR than did their normal cell counterparts (Fig. 6b). Moreover, pediatric patients with pre-B-ALL with higher LEPR mRNA expression tended to survive longer than those with lower levels (although this was not a significant effect) (Children's Oncology Group clinical trial P9906), whereas there was no such correlation in patients with AML (The Cancer Genome Atlas) (Fig. 6c). These results suggest a conserved role of attenuated LEPR expression in supporting of human ALL development.
LEPR sensitivity can be affected by LEPR gene polymorphism42,43, which would not be detected in microarray experiments. To further study the role of LEPR signaling in human leukemia, we expanded the survival analysis to all known LEPR-signaling-related genes44. We found that the expression patterns of 29 out of 118 LEPR-signaling-related genes (24.6%) were significantly correlated with pediatric pre-B-ALL overall or event-free survival; this correlation was significantly higher than that for LEPR-signaling-unrelated genes (2,846 out of 20,933, 13.6%) (Fig. 6d). By contrast, there was no significant difference between the correlation of LEPR-signaling-related and LEPR-signaling-unrelated genes with the overall survival of patients with AML (Fig. 6d). Furthermore, among the 29 genes whose expression was significantly correlated with the outcome of patients with pre-B-ALL, most of the genes that negatively regulate or are negatively regulated by LEPR signaling showed a correlation with poor outcome, whereas genes that positively regulate or are positively regulated by LEPR signaling showed a correlation with better outcome (Fig. 6e and Supplementary Table 5). These results indicate that LEPR signaling is positively correlated with survival in human pre-B-ALL but not in AML.
Next, we determined the effect of fasting on human ALL development in vivo by using a xenograft model. After showing that fasting inhibited N-Myc-driven B-ALL development in sub-lethally irradiated scid recipient mice (Supplementary Fig. 19), we tested the effects of fasting in sub-lethally irradiated scid mice bearing xenografts of human pre-B-ALL cell line NALM-6. Fasting increased LEPR expression on NALM-6 cells as compared to that in the fed condition (Fig. 6f). All fed control mice displayed paraplegia and died of disseminated leukemia by day 35. However, in mice that were fasted with a regimen consisting of three cycles of 2-d fasting/2-d feeding, initiated on day 2 after the cells were injected, the incidence of paraplegia was markedly reduced, the percentage of hCD19+NALM-6 cells was significantly decreased in BM and SP and overall survival was significantly improved as compared to controls (Fig. 6g–i). These data indicate that fasting effectively inhibits the development of human ALL in the xenograft mouse model.
Dietary restriction is associated with positive therapeutic outcome in both animal models and humans, and it selectively suppresses certain solid tumor types, especially when combined with chemotherapy1,2,3,4,5,6. Here, by using mouse models of N-Myc B-ALL, Notch-1 T-ALL, MLL-AF9 AML and AML1-Eto9a AML, as well as a xenograft mouse model of human leukemia, we demonstrated that fasting alone greatly inhibits the development of ALL but not AML, and we identified the underlying mechanism responsible for the differing response to fasting treatment. Our results indicate that the effects of fasting on leukemia development are cancer-type dependent and reveal the importance of attenuated LEPR signaling in ALL development and maintenance.
In most human populations, there is a strong association between obesity and ALL leukemia, but the underlying mechanisms remain poorly understood45,46,47. Patients with obesity and ALL have worse outcomes46,48 than patients with ALL but without obesity, and patients with ALL have a high risk of obesity after treatment42,49. In addition, diet-induced obesity can accelerate the development of ALL in mouse models50. Obesity is characterized by leptin resistance that is usually marked by elevated leptin levels but attenuated LEPR signaling51. Here we demonstrated the importance of attenuated LEPR expression in ALL development and showed that LEPR signaling correlates with pre-B-ALL patient outcome. Our study suggests a possible link between obesity, LEPR signaling and ALL development. Adipocytes protect ALL cells from chemotherapy and promote relapse46,52. We suggest that adipocytes can provide a niche-like environment for ALL cells in which there are high leptin levels, which leads to the suppression of LEPR expression and thereby the inhibition of ALL cell differentiation.
A key finding of our study is that fasting leads to increased LEPR signaling and subsequent inhibition of ALL development, but further study of the underlying mechanisms is needed. In particular, the mechanism by which fasting induces upregulation of LEPR expression and signaling is not yet clear. We speculate that a reduction in systemic leptin levels by fasting leads to a compensatory elevation of LEPR expression on leukemia cells. A similar induction of LEPR mRNA and LEPR protein levels by fasting has been well documented in the nervous system53. The very low level of LEPR in nonfasted ALL cells suggests that LEPR expression levels are the limiting factor in leptin LEPR signaling, such that fasting-induced elevation of LEPR levels would increase signaling, despite a drop in circulating leptin levels. This situation might be specific to ALL cells that express very low levels of LEPR, as compared to other types of cells, including AML cells that express relatively high levels of LEPR.
It will be important to determine whether ALL cells can become resistant to the effects of fasting, which might provide more insights into leukemia pathogenesis. In addition, although both N-Myc B-ALL and Notch-1 T-ALL models might depend on Myc for the leukemia phenotype54, the MLL-AF9- and AML1-Eto9a- AML models do not. This raises an interesting possibility that fasting might affect leukemia development based on specific oncogenes, such as Myc. Future studies are warranted to define more precisely whether differences in the cell of origin or in oncogene drivers contribute to the differential responses of ALL and AML to fasting. Moreover, aside from LEPR signaling, other systemic and leukemic cell effects will undoubtedly be involved in the inhibitory effects of fasting on ALL. For example, the growth hormone (GH)–GH receptor (GHR)–IGF-1 axis has an important role in the treatment of certain solid tumors by a combination of fasting and chemotherapy5,10. Notably, the GH–GHR pathway shares common JAK–STAT downstream targets with the leptin–LEPR pathway5. Whether these pathways work separately or cooperatively in mediating the effects of fasting on different tumor should be further investigated.
Although the therapeutic strategy of forcing malignant cells to terminally differentiate was proposed several decades ago55, differentiation therapy has so far been used successfully only in a subtype of AML, acute promyelocytic leukemia (APL)56,57. Our data show that fasting can effectively induce B-ALL differentiation at both the early and later stages of disease, which suggests the potential of fasting for differentiation therapy in patients with de novo and relapsed ALL. Our study also provides a platform for identifying new targets for ALL treatment.
C57BL/6 mice were purchased from the University of Texas Southwestern Medical Center animal-breeding core facility. Leptin-deficient mice (B6.Cg-Lepob/J), LEPR-deficient mice (B6.BKS(D)-LEPRdb/J), and BALB scid mice were purchased from the Jackson Laboratory. Mice were maintained at the University of Texas Southwestern Medical Center animal facility and at Binzhou Medical University. All animal experiments were performed with the approval of the UT Southwestern Committee on Animal Care and the approval of Binzhou Medical University Committee on Animal Care. For each experiment, the same sex of mice 6–8 weeks of age were used and randomly allocated to each group. The minimum number of mice in each group was calculated by using the 'cpower' function in R/Hmisc package.
293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS at 37 °C in 5% CO2. Primary mouse B-ALL cells were cultured in IMDM supplemented with 10 ng/ml recombinant mouse Il-7 (Peprotech) on OP9 stromal cells. Primary mouse AML cells were cultured in serum-free stemspan (Stemcell techonologies) supplemented with 10 ng/ml SCF (Peprotech). The human leukemia cell line NALM-6 (DSMZ, ACC128) was cultured in RPMI-1640 supplemented with 10% FBS at 37 °C in 5% CO2. NALM-6 cells display a CD3−/CD19+ phenotype by antibody staining and are able to induce B-ALL when xenografted into scid mice58,59. All cell lines were routinely tested using a mycoplasma-contamination kit (R&D). Primary human AML samples were obtained from the tissue bank at the University of Texas Southwestern Medical Center. Informed consent was obtained under a protocol reviewed and approved by the Institutional Review Board at the University of Texas Southwestern Medical Center. Samples were frozen in FBS (FBS) with 10% DMSO and stored in liquid nitrogen.
Virus construction and infection.
For retrovirus packaging, the retroviral constructs MSCV-Notch1-IRES-GFP (T-ALL)24, MSCV-MLL-AF9-IRES-YFP (AML)25,26, pMXs-N-Myc-IRES-GFP (B-ALL)23, MSCV-LEPR, MSCV-LEPR-DsRed, or MSCV-PRDM1-IRES-DsRed were mixed with PCL-ECO (2:1), followed by transfection into 293T cells using Lipofectamine 2000 (Invitrogen). For lentivirus packaging, pLentiLox3.7-based shRNA constructs were mixed with psPAX2 and pMD2.G (Addgene) at a ratio of 4:3:1 and transfected into 293T cells using Lipofectamine 2000 (Invitrogen). Virus-containing supernatant was collected 48–72 h post-transfection and used for spin infection as described previously16,17. Briefly, the virus supernatant was collected and filtered through a 0.45-mm filter. For infection of Lin− cells from BM or fetal liver, Lin− cells were suspended in virus supernatant supplemented with 4 μg/ml polybrene at 106/ml and placed in six-well plate for spin infection (900g, 90 min at 37 °C). For infection of mouse B-ALL and AML cells, the virus was concentrated by centricon centrifugal filters (EMD Millipore), and loaded onto retronectin (50 μg/ml, Clontech) coated six-well petri plates by centrifugation (2,000g, 90 min, 37 °C). After removing the supernatant, 2 × 106 cells per well were added for centrifugation at 600g for 30 min.
Leukemia transplantation and fasting.
Lin− cells were isolated from the fetal liver or bone marrow of wild-type mice or littermate LEPR-deficient or WT mice, and infected with oncogene-IRES-GFP(YFP) expressing retrovirus. Infected mouse Lin− cells (300,000) were transplanted into lethally irradiated (900 cGy) C57BL/6 mice or leptin-deficient mice (6–8 weeks old) by retro-orbital injection. For the second transplantation step, we used FACS to isolate GFP+ or YFP+ BM cells from primary recipient mice and transplanted 3,000 cells (AML) or 10,000 cells (B- or T-ALL) together with 3 × 105 normal BM cells into lethally irradiated recipients.
According to our approved protocols, one to six 1-d fasting/1-d feeding cycles or one to four 2-d fasting/2-d feeding cycles were performed at day 2 after transplantation or at mid-stage of leukemia development (around 3–4 weeks after transplantation). For in vivo leptin administration during 48-h fasting, mouse recombinant leptin (R&D system, 5 μg/15 g body weight) or control PBS was retro-orbitally injected into mice daily during fasting.
Human AML xenograft and fasting.
Xenografting was performed essentially as we have previously described20,60. Briefly, adult scid mice (6–8 weeks old) were sub-lethally irradiated with 250 cGy total body irradiation before transplantation. Either 5 × 106 pre-B-ALL NALM-6 cells were used per mouse for retro-orbital injection. At 3–6 weeks after transplantation, the PB, BM, spleen and liver were assessed for engraftment. For fasting, mice were fasted with up to four cycles of 1-d fasting/1-d feeding.
We performed the procedure as we have previously described16,17. After transplantation, we monitored the survival, examined the size and histological properties of BM, spleen, and liver and analyzed the numbers and infiltration of leukemia cells in PB, BM, spleen, and liver. WBC counts in PB were measured using a Hemavet 950FS. We also characterized different populations of leukemia cells using flow cytometry. Leukemia characterization was performed by investigators blinded to the experimental groups. Moribund leukemic mice were euthanized and the time was recorded as the time of death. Mice that died for reasons unrelated to leukemia within 10 d after transplantation were excluded from the data analysis.
For flow cytometry analyses of mouse PB, BM and SP, leukemia cells were stained with anti-Mac-1-APC (rat, M1/70, 1:200 dilution), anti-Gr-1-PE (rat, RB6-8C5, 1:200 dilution), anti-CD3-APC (hamster, 145-2C11, 1:200 dilution), anti-B220-PE (rat, RA3-6B2, 1:200 dilution), anti-Kit-PE (rat, 2B8, 1:200 dilution) monoclonal antibodies (BD Pharmingen), anti-CD4 (rat, RM4-5, Biolegend), anti-CD8, anti-CD43 (rat, S7, 1:200 dulution, BD Pharmingen), anti-CD19 (rat, 1D3, 1:200 dilution, eBioscience), anti-IgM (rat, II/41, 1:200 dilution, BD Pharmingen), anti-IgD (rat, 11-26c, 1:200 dilution, eBioscience), anti-Igk (rat, 187.1, 1:200 dilution, BD Pharmingen), anti-Igl (rat, R26-46, 1:200 dilution, BD Pharmingen), or anti-LEPR-biotin (goat polyclonal antibody, 1:50 dilution, R&D system) followed by streptavidin-APC or streptavidin-PE-Cy5.5 (1:200 dilution, BD Pharmingen). Human patient samples were stained with anti-human CD19-PE (mouse, HIB19, 1:50 dilution, eBioscience), anti-human CD3-PE (mouse, HIT3a, 1:50 dilution eBioscience), anti-human CD33 (mouse, HIM3-4, 1:20 dilution, eBioscience), anti-human CD45 (mouse, HI30, 1:50 dilution, BD Pharmingen), anti-human CD34 (mouse, 4H11, 1:50 dilution, eBioscience), anti-human CD4 (mouse, OKT4, 1:50 dilution, eBioscience), anti-human CD8α (mouse, 1:20 dilution, R&D system), anti-human CD40 (mouse, 5C3, 1:20 dilution, BioLegend), anti-human IgM (mouse, SA-DA4, 1:20 dilution, eBioscience), anti-human IgD (mouse, IA6-2, 1:20 dilution, eBioscience), or anti-human LEPR (mouse, 52263, 1:50 dilution) monoclonal antibodies (BD Pharmingen). Intracellular staining of Ki67 (rat SolA15, 1:200 dilution, eBioscience) and PRDM1 (rat, 6D3, 1:50 dilution, BD pharmingen) or pSTAT3 (mouse, LUVNKLA, 1:20 dilution, eBioscience) was performed using the Foxp3/Transcription factor staining set (eBioscience) or Fixation/Methanol protocol (eBioscience) provided by the manufacturers. For analysis of human hematopoietic engraftment in scid mice, we followed our previously published protocol20,60 and used anti-human CD19 (BD Pharmingen) and anti-human CD45-PE (mouse, HI30, BD Pharmingen, 1:50 dilution) to quantify total human AML engraftment.
Mouse AML cells were diluted to the indicated concentration in IMDM with 2% FBS and were then seeded into methylcellulose medium M3534 (StemCell Technologies) for myeloid colony formation analysis, as described previously16,17.
The MSCV-IRES-DsRed plasmid was constructed by replacing GFP with DsRed in the MSCV-IRES-GFP plasmid (Addgene). Mouse LEPR and PRDM1, obtained by PCR from mouse bone marrow cDNA, were used to construct the MSCV-LEPR, MSCV-LEPR-IRES-DsRed and MSCV-PRDM1-IRES-DsRed plasmids. pLentiLoxp3.7 (Addgene) was modified by replacing the GFP reporter with DsRed, and used to construct PRDM1 shRNA expressing lentiviral vectors. The target sequences for shPRDM1-1 and shPRDM1-2 were 5′ GGGTGGTTCGGTCCTGTATTCT 3′ and 5′ ATGTACTCCTTACTTACCACG 3′, respectively.
Total RNA was extracted using RNAeasy kit (QIAGEN) and reverse transcribed into cDNA using SuperScript III Reverse Transcriptase (Invitrogen) according to the protocol provided. Real-time PCR was performed with the primers listed in Supplementary Table 6 using SYBR Green Master Mix (Bio-Rad). mRNA levels were normalized to the level of GAPDH RNA transcripts present in the same sample.
Cells were lysed in Laemmli sample buffer (Sigma-Aldrich) supplemented with protease inhibitor cocktail (Roche Diagnostics). Samples were separated on SDS–PAGE gels (Bio-Rad) and transferred on nitrocellulose membranes (Bio-Rad) for protein detection as described16,17. Primary antibodies including anti-PRDM1 (eBioscience, 6D3, 1:200), anti-XBP-1s (Biolegend, 9D11A43, 1:500), anti-N-Myc (Santa Cruz, sc-53993, 1:500) and anti-Myc (Santa Cruz, sc-40, 1:500), as well as horseradish peroxidase (HRP) conjugated secondary antibodies (R&D system, HAF007, 1:1,000, and HAF005, 1:1,000) and chemi-luminescent substrate (Invitrogen), were used.
RNA was purified from sorted cells with Qiagen RNeasy Mini Kit and then reverse-transcribed with SuperScript III Reverse Transcriptase (Invitrogen) according to the manufacturer's instructions. RNA-seq was performed at the UTSW Genomics and Microarray Core Facility. The cDNA was sonicated using a Covaris S2 ultrasonicator, and libraries were prepared with the KAPA High Throughput Library Preparation Kit. Samples were end-repaired and the 3′ ends were adenylated and barcoded with multiplex adapters. PCR-amplified libraries were purified with AmpureXP beads and validated on the Agilent 2100 Bioanalyzer. Before being normalized and pooled, samples were quantified by Qubit (Invitrogen) and then run on an Illumina Hiseq 2500 instrument using PE100 SBS v3 reagents to generate 51-bp single-end reads. Before mapping, reads were trimmed to remove low-quality regions in the ends. Trimmed reads were mapped to the mouse genome (mm10) using TopHat v2.0.1227 with the UCSC iGenomes GTF file from Illumina.
Methods for data normalization and analysis are based on the use of “internal standards”61 that characterize some aspects of the system's behavior, such as technical variability, as presented elsewhere62,63. Genes with log2 (fold change) > 2, P < 0.05 and RPKM > 0.5 were deemed to be significantly differentially expressed between the two conditions, and used for pathway analysis and upstream transcription factor analysis. Pathway analysis was conducted using the Consensus-Path-DB database (http://cpdb.molgen.mpg.de/CPDB)64. Upstream transcription-factor analysis was conducted using QIAGEN's Ingenuity tool (http://www.ingenuity.com/)65. Gene heat maps were clustered by hierarchical clustering (hclust function in R, http://www.r-project.org).
In silico analyses and survival analysis.
Microarray data of human leukemia patients were obtained from the GEO data set accession number GSE13159. LEPR expression levels were calculated by averaging the values from probe sets that recognize all LEPR isoforms (211354_s_at, 211355_x_at, 211356_x_at). For survival analyses, AML and ALL data were obtained from the National Cancer Institute TCGA AML database (https://tcga-data.nci.nih.gov/tcga) and National Cancer Institute TARGET Data Matrix of the Children's Oncology Group (COG) Clinical Trial P9906 (ftp://caftpd.nci.nih.gov/pub/dcc_target/ALL/Phase_I/Discovery/clinical/), respectively. Patients were divided into two groups at the 50th percentile according the level of a probe set or a gene (average of multiple probe sets of a gene); for LEPR, we used the probe sets recognizing all LEPR isoforms (211354_s_at, 211355_x_at, 211356_x_at). Kaplan–Meier survival analysis was used to estimate overall survival and a log–rank test was used to compare survival differences between patient groups using R software.
Statistical analysis was done using Microsoft Excel, Prism 5.0 (Graphpad) and R software. Data were analyzed by Student's t-test or χ-squared test and were considered to be statistically significant if P < 0.05. A variance similarity test (f-test) was performed before the t-test. All t-tests and χ-squared tests were two-tailed unless otherwise indicated. The survival rates of the two groups were analyzed using a log–rank test and were considered to be statistically significant if P < 0.05.
RNA-seq data have been deposited in Gene Expression Omnibus under accession number GSE89622.
Gene Expression Omnibus
Gene Expression Omnibus
We would like to acknowledge the support by grants from the US National Institutes of Health 1R01CA172268 (C.C.Z.), Leukemia & Lymphoma Society Awards 1024-14 (C.C.Z.) and TRP-6024-14 (C.C.Z.), CPRIT RP140402 (C.C.Z.), 1R01HL089966 (L.J.H.), R01-DK55758 (P.E.S.) and RP140412 (P.E.S.).
Supplementary Figures 1–19 and Supplementary Tables 1–6