LIM domain only-2 (LMO2) induces T-cell leukemia by two distinct pathways

Abstract

The LMO2 oncogene is deregulated in the majority of human T-cell leukemia cases and in most gene therapy-induced T-cell leukemias. We made transgenic mice with enforced expression of Lmo2 in T-cells by the CD2 promoter/enhancer. These transgenic mice developed highly penetrant T-ALL by two distinct patterns of gene expression: one in which there was concordant activation of Lyl1, Hhex, and Mycn or alternatively, with Notch1 target gene activation. Most strikingly, this gene expression clustering was conserved in human Early T-cell Precursor ALL (ETP-ALL), where LMO2, HHEX, LYL1, and MYCN were most highly expressed. We discovered that HHEX is a direct transcriptional target of LMO2 consistent with its concordant gene expression. Furthermore, conditional inactivation of Hhex in CD2-Lmo2 transgenic mice markedly attenuated T-ALL development, demonstrating that Hhex is a crucial mediator of Lmo2's oncogenic function. The CD2-Lmo2 transgenic mice offer mechanistic insight into concordant oncogene expression and provide a model for the highly treatment-resistant ETP-ALL subtype.

Figure 1. CD2-Lmo2 transgenic mice develop T-ALL with high penetrance, long latency and upregulation of Lyl1.

(A) schematic shows structure of CD2-Lmo2 transgene used in pronuclear injection. Lower panel shows a Southern blot of tail genomic DNA digested with EcoRI which cuts once within the transgene and probed with Lmo2 cDNA. Red asterisk shows transgene band in equal intensity to other bands which represent the endogenous Lmo2 exons. (B) photo of CD2-Lmo2 transgenic mouse sacrificed due to T-ALL onset. T, thymus; H, hepatomegaly; L, lymphadenopathy; S, splenomegaly. (C) Survival analysis of CD2-Lmo2 transgenic mice (n = 24). (D) PCR analysis of Jβ2 T-cell receptor showed rearrangement in most T-ALL genomic DNA; germline configuration is shown by the presence of only a 2 kb band in the lane. (E) Lmo2 and Lyl1 mRNA were quantified by qRT-PCR on whole RNA isolated from 24 CD2-Lmo2 T-ALLs and shown relative to levels in normal thymus. Values shown are fold increase over normal thymus.

Figure 2. LYL1 is part of an LMO2-associated protein complex in T-ALL cells that binds tandem E boxes.

(A) shows oligos used in EMSA. First oligo has two E boxes separated by 11 bp; second oligo has a scrambled second E box. (B) Oligos were end-labeled with ³²P-ATP and used to bind nuclear extract from LOUCY cells, a T-ALL line that expresses LMO2 and LYL1. The gel shows migration of two different complexes, boxed in red, a slow and fast complex. (C) The slow complex was close to the origin and was resolved with an overnight run on 4% polyacrylamide gel. Blue arrows show the macromolecular complex that is competed with cold oligo and shifted in the presence of antibodies to LMO2, LDB1, E47, and LYL1. The red arrows show altered mobility of protein complexes. (D) shows the faster migrating complex (blue arrow) that is competed away by cold oligo and supershifted with antibody against LYL1 (red arrow).

Figure 3. Hhex expression defines two subtypes of CD2-Lmo2 transgenic T-ALLs.

(A) We analyzed the Retroviral Tagged Cancer Gene Database (RTCGD) for T-ALLs with integrations at Lmo2 and at Hhex. These two genes were never mutated in the same tumor but we observed overexpression of Hhex in all the Lmo2-clonal T-ALLs . The red set (left) of genes were all insertionally mutated concordant with Hhex whereas the blue set (right) of genes was concordant with Lmo2. The genes shown in the overlap, Mef2c, Sox4, Mycn, Irs2, and Ccnd3, and Tcfe2a, were concordant with both Hhex and Lmo2-induced T-ALLs. (B) Bar graph shows qRT-PCR analysis for Hhex transcripts in whole RNA isolated from CD2-Lmo2 T-ALLs. All values are expressed as fold increase above normal thymus and are shown from highest to lowest relative expression. Tumor names appear on x axis. The y-axis was placed between those T-ALLs that expressed Hhex and those that did not. Bar graph shows similar quantification of Mycn and Myc. Here, the T-ALLs are arranged from highest to lowest relative expression of Myc and values are shown as fold increase over normal thymus. The dark gray bars show Myc and light gray show Mycn mRNAs. (C) We divided CD2-Lmo2 transgenic T-ALLs into two classes, those with high Hhex expression (left of y axis in panel A) and those with negative or low Hhex expression (right of y axis in panel A) and plotted the relative expression values for the genes shown. The mean values of the genes are shown by the horizontal bars. The mean expression for each gene was compared by two tailed Mann-Whitney U-test generating the P values shown.

Figure 4. LMO2 and HHEX are co-expressed in ETP-ALL.

(A) The heat map shows a supervised clustering result based on LMO2 (left panel) and HHEX (right panel) expression. The colors denote z scores ranging from −4 to 4. The gene expression dataset of 78 pediatric T-ALLs was analyzed by Limma based on a discrete cutoff of median LMO2 and HHEX expression. The black arrows show 12 Early T-cell Precursor ALL cases. LYL1, HHEX, MEF2C, and MYCN genes were concordantly expressed with LMO2 and HHEX. These genes were expressed highest in the ETP-ALL cases. A full version of the heat maps are in file S1. (B) Survival analysis was performed on LMO2-high versus LMO2-low T-ALL and on HHEX-high v. HHEX-low T-ALLs. The curves for LMO2 classes in the left panel were not statistically significant (P = .19) but the curves for HHEX classes were significantly different (P = .03) by Log rank tests.

Figure 5. T-ALL expression of HHEX requires its promoter and enhancer.

(A) Schematic shows the structure of the mouse Hhex gene with 4 exons (coding sequences shown in blue). We analyzed ChIP-seq data from anti-Ldb1 that showed occupancy within intron 1 (green bar graph). The RefSeq mRNAs for Hhex are shown below the sequencing tag counts. The bar graph under this shows the conservation across multiple mammalian species. The red box denotes an enhancer that was previously functionally characterized as specifying blood-specific expression of Hhex. (B) We cloned the enhancer shown in panel A 5′ to the SV40 promoter and transfected it into LOUCY, K562, and Jurkat cell lines along with pCMV-Renilla luciferase vector. Luciferase values were normalized to Renilla to correct for transfection efficiency and expressed as fold over Renilla alone. (C) We constructed a luciferase expression vector with 1 kb of the Hhex promoter, exon 1, intron 1, and replaced exon 2 with the luciferase cDNA. We transfected this reporter (pG-Hhex-luc) into K562 and LOUCY cells. Clones 1–6 have 50–100 bp deletions 5′ to 3′ within the enhancer region. Luciferase expression was normalized to Renilla for transfection efficiency and expressed as fold above that seen in Renilla alone.

Figure 6. HHEX is a direct transcriptional target of an LMO2 protein complex.

(A) We analyzed the occupancy of LMO2 and its binding partners at the HHEX enhancer, promoter, and exon 4 (nonspecific) in LOUCY cells by ChIP. Bar graph shows a summary of 4 independent experiments using antibodies against LMO2, LDB1, LYL1, GATA3, and E47 for ChIP followed by qPCR analysis relative to IgG control. (B) Nonsilencing shRNA or specific shRNA directed against LMO2 was transfected into LOUCY cells followed by sorting for GFP⁺ cells. Cells were counted and analyzed by propidium iodide flow cytometry for viability. Y-axis denotes viable GFP-expressing cells. (C) LOUCY cells transfected with nonsilencing (NS) or LMO2 shRNAs were harvested 24 h after transfection and whole cell lysate prepared. Protein was separated by SDS-PAGE followed by transfer to nitrocellulose and Western blot with monoclonal anti-LMO2. Quantification was performed by infrared-dye conjugated secondary antibodies which showed specific shRNA knocked down LMO2 protein to 51% of the level present in nonsilencing shRNA. (D) We harvested mRNA from LMO2 knockdown LOUCY cells and analyzed MYCN and HHEX expression by qRT-PCR. Bar graphs show quantity of MYCN or HHEX expressed relative to LOUCY transfected with non-silencing shRNA. The graph shows a summary of three independent knockdown experiments. The mean values were compared by Mann-Whitney U-test generating the P values shown.

Figure 7. Conditional inactivation of Hhex prolongs latency of Lmo2-induced T-ALL.

(A) Genomic DNAs from bone marrow (lanes 2, 5), spleen (3, 6), and thymus (lanes 4, 7) were analyzed using primers specific for floxed Hhex or deleted Hhex (cKO); lane 1 is H₂O; bands are indicated by blue arrows. Tissues from two representative mice are shown: Hhex^lox/lox and Hhex^lox/lox; Vav-iCre (designated Hhex cKO). (B) Single cell suspension was prepared from Hhex^lox/lox or Hhex cKO thymi and analyzed by flow cytometry for CD4 (x-axis) and CD8 (y-axis) antigen expression. Numbers show the percentage of cells in each quadrant. (C) Graph shows survival analysis of Lmo2/Hhex cKO and Lmo2/Hhex^lox/lox mice (n = 16 per group). The median survival of Lmo2/Hhex^lox/lox mice was 306 days whereas for Lmo2/Hhex cKO, the median was not yet attained. P value represents the comparison of survival times by Log-rank test. (D) Genomic DNA was prepared from thymi involved with T-ALL of Lmo2/Hhex^lox/lox (n = 3, lanes 2–4) and Lmo2/Hhex cKO mice (n = 2, lanes 5, 6); lane 7 shows positive controls for PCR of deleted (cKO) and floxed Hhex (blue arrows).