Initiating and Cancer-Propagating Cells in TEL-AML1-Associated Childhood Leukemia

See allHide authors and affiliations

Science  18 Jan 2008:
Vol. 319, Issue 5861, pp. 336-339
DOI: 10.1126/science.1150648


Understanding cancer pathogenesis requires knowledge of not only the specific contributory genetic mutations but also the cellular framework in which they arise and function. Here we explore the clonal evolution of a form of childhood precursor–B cell acute lymphoblastic leukemia that is characterized by a chromosomal translocation generating a TEL-AML1 fusion gene. We identify a cell compartment in leukemic children that can propagate leukemia when transplanted in mice. By studying a monochorionic twin pair, one preleukemic and one with frank leukemia, we establish the lineal relationship between these “cancer-propagating” cells and the preleukemic cell in which the TEL-AML1 fusion first arises or has functional impact. Analysis of TEL-AML1–transduced cord blood cells suggests that TEL-AML1 functions as a first-hit mutation by endowing this preleukemic cell with altered self-renewal and survival properties.

A popular (16), though not unchallenged (7), hypothesis is that fully transformed human cancer clones are arranged, like their normal tissue counterparts, in a hierarchical fashion and maintained by rare “tumor-propagating cells” (also referred to as cancer stem cells). Less is known about the crucial earliest first-hit events in cancer and the “precancerous” cells in which they occur. We have explored these precancerous cell hierarchies in the context of childhood precursor–B cell (pre-B cell) acute lymphoblastic leukemia (also known as common ALL, cALL) which is frequently associated with a chromosomal translocation creating the TEL-AML1 (ETV6-RUNX1) fusion gene (8). This fusion arises predominantly in utero, producing a persistent but clinically covert preleukemic clone (911) that may convert to frank leukemia with acquisition of additional genetic changes (10, 12).

Most of the leukemic cells in cALL coexpress CD19 and CD10 accompanied by clonal rearrangement of the immunoglobulin heavy chain gene (IgH) indicative of a pre-B cell identity (13). However, the identity of the cALL-propagating cell has been contentious (1417). Interestingly, childhood ALL is associated with a rare population of CD34+CD38–/lowCD19+ cells that is not detectable in normal bone marrow (17, 18). We prospectively isolated these cells from TEL-AML1–positive cALL patients (Fig. 1A, R1) and assessed their capacity to engraft nonobese diabetic, severe combined immunodeficient mice (NOD/SCID mice), an assay of their leukemogenic potential (19). Engraftment was obtained from four patient samples (Fig. 1B and fig. S1) and secondary grafts were obtained from all of these. The CD34+CD38–/lowCD19+ population was reestablished within both the primary (Fig. 1C) and secondary (Fig. 1D) recipients. Interphase fluorescence in situ hybridization (FISH) analysis confirmed the presence of the TEL-AML1 fusion gene in these grafts (Fig. 1E). These data suggest that the CD34+CD38–/lowCD19+ cells can function as self-renewing cancer-propagating cells.

Fig. 1.

Identification of tumorpropagating cells in cALL patients. (A) Representative fluorescence-activated cell sorting (FACS) analysis of leukemic bone marrow mononuclear cells showing the CD34+CD38–/lowCD19+ (R1) population. (B) Representative analysis of bone marrow from a primary murine recipient showing human CD45+CD19+ engraftment. (C) Detailed analysis of engrafted cells from a primary recipient showing the CD34+CD38–/lowCD19+ population. (D) Similar analysis of engrafted cells from bone marrow of a secondary recipient showing reestablishment of the same population. (E) Interphase FISH showing TEL-AML1 expression in a human CD19+ cell from a secondary recipient. TEL signal, green; AML1, red; TEL-AML1 fusion (green-red) indicated by arrowhead. Fusion identified in >97% human cells analyzed from recipients of two separate cALL samples.

How do these cells relate to those in which the TEL-AML1 fusion first arises and initiates leukemogenesis? Analysis of monochorionic twins provides a unique opportunity to investigate the initiating events in this disease because a preleukemic clone, established in utero in one twin may spread to the other twin via their shared placenta (9). Because additional mutations are required for progression to leukemia (9, 10), when one twin presents with leukemia, the other may still carry the ancestral preleukemic clone but may remain clinically normal. We investigated a pair of monochorionic female twins. One twin was diagnosed at age 2 with TEL-AML1–positive pre-B cell ALL, and her leukemic blasts showed additional loss of the uninvolved TEL allele (Fig. 2A). Her bone marrow contained the CD34+CD38–/lowCD19+ cancer-propagating population (Fig. 2B). The other twin was healthy. For ethical reasons, our analyses were strictly limited to peripheral blood (PB) [supporting online material (SOM) text]. We detected TEL-AML1–positive B lineage–affiliated CD19+ cells at low frequency (∼0.1%), but these cells all contained a normal TEL allele (Fig. 2A), consistent with preleukemia status (11).

Fig. 2.

Hematopoiesis in monochorionic t(12;21) twins. (A) Immuno-FISH for TEL-AML1 translocation and CD19 protein in PB mononuclear cells from the leukemic twin, healthy twin, and a normal individual. CD19 indicated by blue staining, TEL by green, AML1 by red, and TEL-AML1 by green-red (filled arrowhead). Remnant of the AML1 locus left by the translocation, small red signal, thin arrow; intact AML1 allele, large red signal; intact second TEL allele, green signal (open arrowhead). (B) Diagnostic bone marrow from the leukemic twin showing a CD34+CD38–/lowCD19+ population, gate R2. (C) CD19 expression in PB CD34+CD38–/low cells from healthy twin (top) and a normal age-matched child. (D) Reverse transcription PCR for TEL-AML1 and β-actin expression in normal cord blood (lanes 1 and 2) and in PB compartments from healthy twin (lanes 3 to 5). Lane 1, cord blood CD34+ (100 cell equivalents). Lane 2, cord blood CD34+ (10 cell equivalents). Lane 3, twin CD34+CD38–/lowCD19+ (8 cell equivalents). Lane 4, twin CD34+CD38+CD19+ (pro–B cells, 25 cell equivalents). Lane 5, CD34+CD38–/lowCD19 (multipotential stem/progenitors, 10 cell equivalents). (E) Genomic PCR for IgH gene rearrangements. DQ52-JH (heavy-chain joining region) (left) and VDJ (right) analyses for control Kasumi cells (C) and CD34+CD38–/lowCD19+ populations from healthy (H) and leukemic (L) twin. Arrows indicate 160 bp (germline) and 80 bp (rearranged). (F) Representative sequence analyses of individual cloned DQ52-JH PCR products from healthy and leukemic twins. Long arrows indicate oligonucleotide primers, short arrow indicates a G-A transversion suggestive of somatic hypermutation. A minority of (3 out of 11) clones from the CD34+CD38–/lowCD19+ cells of the leukemic twin were different from each other and those of her sibling, which may suggest ongoing DJ recombination. (G) CD10 expression in CD34+CD38–/lowCD19+ cells from leukemic (bone marrow) and healthy (PB) twins. (H) CD10 expression in bone marrow CD34+CD38–/lowCD19+ cells from an unrelated cALL patient (representative of five individuals).

We detected a rare population of CD34+ CD38–/lowCD19+ cells at a frequency of 0.002% of total mononuclear cells in the healthy twin, but not in the PB of hematologically normal age-matched children (Fig. 2C, R3). This cell population has been stable over 18 months of serial observation (fig. S2). TEL-AML1 transcripts were observed in as few as 10 cell equivalents of this sorted fraction (Fig. 2D, lane 3). In contrast, we did not detect TEL-AML1 transcripts in similar numbers of progenitor B cells (pro-B cells) from the healthy twin (Fig. 2D, lane 4) consistent with the low proportion of B cells affected in the blood of this hematologically normal child, or in more primitive multipotent stem cell fractions not expressing the B-lineage marker CD19 (Fig. 2D, lane 5).

To investigate the relation between the CD34+CD38–/lowCD19+ populations in the healthy and leukemic twins, we examined IgH gene rearrangements involving the variable (V), diversity (D), and joining (J) loci. Polymerase chain reaction analysis (PCR) revealed DJ but not VDJ recombination events in the healthy twin sample (Fig. 2E). Sequencing of the cloned DJ recombination PCR products from the healthy twin's CD34+CD38–/lowCD19+ cells revealed the presence of only one species, which suggested that these cells represented a clonally expanded population and that the fusion gene arose or had first functional impact in a cell that had already undergone DJ recombination. CD34+CD38–/lowCD19+ cells from the leukemic twin harbor DJ, as well as VDJ, recombination products (Fig. 2E). Sequencing of cloned DJ segments revealed some heterogeneity in recombination products, but crucially, the majority of clones were, with the exception of a single base change, highly related to the DJ segment seen in the CD34+CD38–/lowCD19+ population from the healthy twin (Fig. 2F). This suggests that the CD34+CD38–/lowCD19+ cancer-propagating population in the leukemic twin is a clonal and more differentiated descendant of the CD34+CD38–/lowCD19+ population in the healthy twin. Consistent with this, the majority of CD34+CD38–/lowCD19+ cells in the leukemic twin express common ALL antigen, CD10 (20); those in the healthy twin do not (Fig. 2G). It is noteworthy that a minor component of the CD34+CD38–/lowCD19+ population in leukemic samples was also found to be CD10 (Fig. 2H, R1). The presence of the TEL-AML1 fusion gene in these cells was confirmed by FISH (fig. S3).

We next established a model of TEL-AML1–driven preleukemia by transplanting TEL-AML1–transduced human cord blood cells into NOD/SCID mice (fig. S4). At the level of expansion of B cell progenitors and B-lineage differentiation impedance or arrest, this xenograft model exhibits salient features of the preleukemic phase of pre-B cell ALL (fig. S5) (9, 10). CD34+CD38–/lowCD19+ cells were observed in the TEL-AML1–expressing, but not control, marrow fractions (Fig. 3A and figs. S5 and S6) or in normal cord blood (fig. S7). Xenograft-modeled CD34+CD38–/lowCD19+ cells displayed DJ rearrangements (Fig. 3B) and, like their counterparts in the healthy twin (Fig. 2G), did not express CD10 (fig. S8). Collectively, our data support the notion that TEL-AML1 can, as a single mutation, generate abnormal cells that resemble the TEL-AML1– expressing CD34+CD38–/lowCD19+ cells observed in the healthy twin.

Fig. 3.

First-hit functions of TEL-AML1 in transduced cord blood stem cells. (A) Representative immunophenotype of bone marrow graft from a primary recipient transplanted with TEL-AML1–transduced cells showing a CD34+CD38–/lowCD19+ population in the green fluorescent protein (GFP)+ compartment only. (B) PCR analysis of DJ rearrangements in purified CD34+CD38–/lowCD19+ cells. Reactions for the DXP4, DA4, DK4, DN4, DM1, DLR1, and DQ52 families (lanes 1 to 7) are shown. A germline product is indicated for DQ52 (lane 7). Rearrangements are seen for the DXP4, DN4, and DLR1 families (lanes 1, 4, and 6, red arrows). (C) Photomicrograph (left) and immunophenotype of typical B cell colony after MS-5 stromal cell coculture of xenograft modeled CD34+CD38–/lowCD19+ cells. (D) CD34+CD38–/lowCD19+ cells (1) but not TEL-AML1–expressing (2) or control (3) pro-B cells from primary recipients generated colonies on MS-5 stroma (n = 3, mean ± SD) and (E) these generated secondary colonies on replating—an index of self-renewal capacity. (F) Analysis of secondary recipients of purified modeled CD34+CD38–/lowCD19+ (GFP+) cells (representative of three experiments) showing reconstitution of the original population (which is also CD10) and a more mature population (CD19+CD38+). (G) Enrichmentof CD34+CD38–/lowCD19+ compartment from the bone marrow of a TEL-AML1 primary recipient after culture with or without soluble Fas ligand (sFasL) (cells were gated on CD19 and GFP before analysis for CD34 and CD38). (H) Means (±SD) change in size of CD34+CD38–/lowCD19+ compartments in the bone marrow of TEL-AML1 primary recipients after culture with (shaded) or without (open) camptothecin (1), Fas-L (2), melphalan (3), paclitaxel (4), tumor necrosis factor–α (5), methylprednisolone (6), and etoposide (7) (in each case, n = 5, P < 0.05 by t test).

The xenograft-modeled, TEL-AML1–generated CD34+CD38–/lowCD19+ cells displayed B cell differentiation and self-renewal potential in vitro (Fig. 3, C, D, and E, and fig. S9). To investigate whether the TEL-AML1–generated CD34+CD38–/lowCD19+ population can initiate and maintain a “preleukemic” state in vivo, we prospectively isolated these cells from engrafted primary mice and injected them into the tibiae of secondary NOD/SCID recipients. To our surprise, these cells engrafted and gave rise to more mature B cells (CD38+CD19+), as well as reconstituting a CD34+CD38–/lowCD19+ population (Fig. 3F). In contrast, TEL-AML1–expressing CD34+CD38+CD19+ (pro–B) cells did not engraft. These experiments indicate that the TEL-AML1–generated CD34+CD38–/lowCD19+ population has significant self-renewal potential. Analysis of cell survival characteristics of TEL-AML1–generated CD34+CD38–/lowCD19+ cells indicates that TEL-AML1 markedly enhances resistance to some, but not all, apoptotic stimuli (Fig. 3, G and H). Finally, preliminary microarray analysis indicates that CD34+CD38–/lowCD19+ cells contain a mixture of both pro-B cell and human stem cell–associated gene expression profiles (fig. S10).

What then are the salient features of the CD34+CD38–/lowCD19+ cell observed in the healthy twin and the role of TEL-AML1 in its generation? Its clonal expansion, coupled with its persistence since birth, is highly suggestive of self-renewal. Its clonal relation to more differentiated cell types both in the healthy and in the leukemic twins implies differentiation potential. The balance of evidence thus favors the notion that this cell may itself function as a preleukemic stem cell. This proposal is supported by our xenograft modeling studies, which further suggest that TEL-AML1 may be sufficient to generate this population of preleukemic stem cells.

Our results suggest that a hierarchical structure, which has been demonstrated in frank leukemia (2, 4), is also a feature of “early” or preleukemic populations. Understanding the nature of the preleukemic hierarchy is fundamental to understanding the function of the first-hit mutation and how it predisposes to leukemic transformation. Our studies therefore have implications for disease etiology, and the xenograft model presented may provide a tool for examining the biological role of genetic alterations that cooperate with the TEL-AML1 fusion gene. Our studies may also be relevant to cancer therapy where specific targeting of tumor propagating cells may be desirable. The observation that children in lengthy remission can relapse late with a novel leukemic clone (21), but which nonetheless appears to derive from the identical preleukemic clone that initiated the disease at presentation, suggests that the preleukemic stem cell compartment may persist even when the cells propagating the overt leukemia have been effectively eradicated.

References and Notes

Stay Connected to Science

Navigate This Article