Patient Cells and Animal Research

Cryopreserved primary leukemia cells from pediatric BPL patients as well as BPL xenograft cells isolated from spleen specimens of xenografted NOD/SCID mice were used in the described experiments (File S1). The IRB (CCI) at Children’s Hospital Los Angeles (CHLA) (Human Subject Assurance Number: FWA0001914) determined that the use of leukemic cells in our research did not meet the definition of human subject research per 45 CFR 46.102 (d and f) since it does not include identifiable private information. The research was approved by the CHLA IRB/CCI. The IRB approved project numbers were CCI-09-00304 (CCI Review Date 12/21/2009, Approval Date: 12/29/09) for cryopreserved cells and CCI-10-00141 (CCI Review Date 7/27/2010, Approval Date 7/27/2010) for freshly obtained primary leukemia cells. We also used an NOD/SCID mouse model of human B-precursor ALL [2]. NOD/SCID mice (NOD.CB17-Prkdc^scid/J; 4-6 weeks of age at the time of purchase, female) were obtained from the Jackson Laboratory (Sacramento, CA). The research was conducted according to Institutional Animal Care and Use Committee (IACUC) Protocol 280-09, that was approved by the IACUC of CHLA on 11-24-2009 and its 3-year rewrite application 280-12 that was approved on 7-10-2012. All animal care procedures conformed to the Guide for the Care and Use of Laboratory Animals (National Research Council, National Academy Press, Washington DC 1996, USA).

Bioinformatics and Statistical Analysis of Gene Expression Profiles

The publically available archived GSE32311 database [8] was used to compare gene expression changes in control thymocytes from IKZF1 wildtype mice (GSM800500, GSM800501, GSM800502) vs. IK-deficient thymocytes from IKZF1 null mice (GSM800503, GSM800504, GSM800505) with the same genetic background of (C57BL/6 x129S4/SvJae). The Gene Pattern software (http://www.broadinstitute.org/cancer/software/genepattern) was utilized to extract expression values for human lymphocyte precursors from the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database [9]. We compiled the archived gene expression profiling (GEP) data on primary leukemia specimens from 1486 pediatric acute lymphoblastic leukemia (ALL) patients across 7 independent studies (GSE3912, N=113; GSE18497, N=82; GSE4698, N=60; GSE7440, N=99; GSE13159, N=750; GSE11877, N=207; GSE12995, N=175) that combined datasets from U133A and U133 Plus 2 genechips using common IKZF1 probe sets to focus our analysis on 45 validated IK target genes as well as 20 IK-regulated lymphoid-priming genes [10] (File S1; Table S1 and Table S2 in File S1). In addition, we compiled 6 archived gene expression profiling datasets that measured expression from B-precursor ALL childhood patients hybridized to Human Genome U133 Plus 2.0 Array containing the 10 IKZF1 probe sets for determination of exon specific expression (GSE11877, N=207; GSE13159 N=823; GSE13351 N=107; GSE18497, N=82, GSE28460, N=98; GSE7440, N=99; Total, N=1416) (File S1). Further, we performed gene set enrichment analyses (GSEA) in which rank ordered differences in standard deviation units for BCR-ABL⁺ BPL samples (N=20) compared to other samples (N=155) in the Mullighan study (GSE12995) and Ph⁺ BPL (N=122) versus Ph- BPL (N=237) in the “The Microarray Innovations in Leukemia” (MILE) study (GSE13159) were processed for enrichment of lymphoid priming genes (18 genes represented on the gene chips) and IK target genes (39 genes represented in the Mullighan study, 45 genes represented in the MILE study) using a supervised approach implemented in GSEA v2.08 (Broad institute) (File S1). These ranked ordered genes were screened for enrichment of gene sets in 13321 genes (22283 transcripts) for the Mullighan study and 20606 genes (54613 transcripts) for the MILE study using weighted Kolmogorov-Smirnov statistics implemented in GSEA (GSEA v2.08 (Broad Institute).

Standard Biochemical Assays, PCR, Confocal Imaging, and Transfection Methods

Confocal Laser Scanning Microscopy, Western blot analyses, and electrophoretic mobility shift assays (EMSA) were performed as per previously described standard procedures (Data Supplement). We also performed PCR and real-time quantitative PCR on genomic DNA samples using standard procedures (File S1). The primer sets used for amplifying and sequencing Ikaros/IKZF1 exons 4, 5, 6 and 7 and their exon-intron junctions are listed in Table S3 in File S1.

SCID Mouse Xenograft Model of Human BPL

We used an NOD/SCID mouse model of human BPL to study IK expression levels in in vivo clonogenic ALL xenograft cells. The derivation and immunophenotypic features of the xenograft cells were recently published [11].

Examination of Primary Leukemia Cells from Ph⁺ High-Risk BPL Patients for Structural and Functional Integrity of IKZF1 Gene using Genomic PCR, RT-PCR, Western Blot Analysis, and EMSA

In an effort to identify homozygous IKZF1 deletions, we performed exon-specific genomic PCR with DNA sequencing on purified genomic DNA samples from 3 pediatric patients with Ph⁺ high risk BPL. As shown in Figure 1A1-A4, no homozygous deletions involving Exon 4, Exon 5, Exon 6 or Exon-7 were found in any of the 3 cases. This absence of genomic deletions in the E4-E7 segment encoding the DNA binding domain of IK prompted us to further validate our findings using Western blot analyses with an IK-specific antibody. Leukemic cells from all 3 Ph⁺ BPL cases expressed a ~57-kDa immunoreactive protein corresponding in size to full-length IK1 (Figure 1A5). We next sought to determine if the detected IK1 protein in Ph⁺ BPL cells enters the nucleus and binds to DNA in a sequence-specific manner. As shown in Figure 1B, Ph⁺ BPL cells from each of the 4 cases (3 cases from Panel A plus an additional Ph⁺ BPL case) examined showed nuclear expression of DNA binding IK isoform IK1 (57 kDa) as well as at least 2 of the 3 IK-activating regulatory proteins SYK [10], Ku70 [12] and BTK [13]. Likewise, nuclear expression of IK and IK-activating regulatory proteins was confirmed for 4 of 4 Ph^- ALL cases (Figure 1B). In EMSA tests, the nuclear protein extracts of Ph⁺ BPL cells caused selective gel retardation of the IK-BS1 test probe containing a high-affinity IK1 binding site (Figure 1C). The presence of native IK in these in retarded DNA-binding protein complexes was confirmed in supershift assays using an IK-specific monoclonal antibody (Figure 1D). These findings provided compelling evidence that primary leukemia cells from Ph⁺ (as well as Ph^-) pediatric high-risk BPL patients express a fully functional IK protein, thereby confirming and extending the results of the genomic PCR and Western blot assays.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Pediatric Ph⁺ BPL is not Characterized by IK-deficiency.

[A1-A4] Genomic PCR. Analysis of the Human Ikaros/IKFZ1 Gene Exons E4-E7 in High-Risk BPL. We performed exon-specific IKZF1 PCR with DNA sequencing on purified genomic DNA samples from 3 pediatric patients with Ph⁺ high-risk BPL. Exons E4-E7 and their intron-exon junctions were PCR amplified using the PCR primers in Table S3 in File S1. Depicted are agarose gels documenting that normal size PCR products (E4: 577-bp, E5: 646-bp, E6: 530-bp, E7:785-bp) for each of the 4 IKZF1 exons were obtained in each of the 3 patients, providing strong evidence against the existence of homozygous deletions of the entire IKZF1 locus or within. IKZF1 exons E4-E7. [A5] Western Blot Analysis of Ikaros Expression in Ph⁺ High-Risk. BPL. Depicted is an anti-IK Western blot of whole cell lysates of primary leukemia cells from 3. Ph⁺ BPL patients. Similar to RAJI cells and primary cells from an MLL-AF4⁺ ALL case that were included as controls, Ph⁺ ALL cells from each of the 3 cases expressed an intact 57 kDa IK1 protein. [B] Nuclear Expression of Intact Ikaros Protein and Its Regulators in Ph⁺ and Ph^- High-Risk BPL. Depicted are Western blots of nuclear protein extracts (NE) from primary leukemic cells of 8 high-risk pediatric BPL patients, including 4 Ph⁺ ALL cases. NE were examined for presence of IK (B1), Ku70 (B2), SYK (B3) and BTK (B4). See text for discussion. [C & D] Assessment of Sequence-Specific DNA Binding Function of Nuclear Ikaros in Ph⁺ and Ph^- High-Risk BPL. Electrophorectic mobility shift assays (EMSA) were performed on nuclear extracts from BPL cells using the Thermo Scientific LightShift Chemiluminescent EMSA Kit and the IK-BS1 test probe containing a high-affinity IK1 binding site and the IK-BS5 negative control probe that has a single base pair (G>A) substitution at position 3 within the core consensus and does not bind IK [10], [12], [13]. The nuclear protein extracts from each of the 4 Ph⁺ and 4 Ph^- BPL cases showed significant gel retardation of the IK1-specific IK-BS1 probe (but not the control IK-BS5 probe) (C). Supershift assays were performed in 3 Ph⁺ and 4 Ph^- BPL cases with an anti-IK monoclonal antibody (2 µg/sample) to confirm the presence of IK in the retarded DNA-binding protein complexes, as previously described [12] (D). Positions of the retarded and supershifted bands are indicated with arrowheads. In each case, the retarded complex of the IK-BS1 probe was supershifted with the anti-IK antibody.

https://doi.org/10.1371/journal.pone.0080732.g001

Examination of Leukemia Cells from Patients with Newly Diagnosed Pediatric High Risk BPL, Including Its Ph⁺ Subset, for IKZF1 Deletions by Real-time Quantitative PCR Analysis of the Ikaros/IKZF1 Gene

It has been reported that the most common IKZF1 microdeletions in Ph⁺ BPL occur between Exons 4-7 (30%) or Exons 2-7 (15%) [14]. In particular, deletions of Exons 4-7 has been proposed as the molecular cause for the generation of IK6, a non-DNA binding truncated IK isoform that lacks all of the DNA binding domains of IK1. Exons 4 and 5 contain 3 of the 4 zinc fingers and their absence is a characteristic feature of the non-DNA binding IK isoforms IK4, IK5, IK6, IK7, IK8, and IK9. We therefore used real-time qPCR to examine side-by-side genomic DNA from normal hematopoietic cells in normal bone marrow specimens (N=2) and primary leukemic cells from 32 pediatric patients with high-risk BPL, including 4 Ph⁺ ALL patients, for evidence of intragenic deletions of IKZF1 involving Exon 4 or Exon 5. IKZF1 Exon 7 was also amplified as an internal reference control. No Ph⁺ or Ph^- high-risk BPL case was identified by interpatient comparisons to have a significantly higher “outlier” Ct value suggestive of an IKZF1 deletion (Figure 2, Panels A-C). Likewise, none of the leukemic samples had a significantly higher raw (Figure 2, Panels A-C) or normalized (Figure 2, Panels D1 & D2) Ct value higher than the median Ct value of the normal samples. Representative IKZF1 Exon 4-specific (and Exon 7-specific) qPCR amplification plots are depicted in Figure 2, Panels E-H. These results demonstrate that IKZF1 deletions are not consistent abnormalities in pediatric high-risk BPL and they are not signature characteristics of pediatric Ph⁺ BPL, as previously reported [3].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Evaluation of Primary Leukemic Cells from High-Risk Pediatric BPL Patients for IKZF1 Deletions Involving Exons 4 or 5 Using Real-time Quantitative PCR.

[A] Depicted are the Box- and Whisker plots of the mean Ct values from 2 independent technical replicates of exon-specific IKZF1 qPCR performed on genomic DNA of primary leukemic cells from 32 BPL patients and 2 normal control bone marrow specimens. In each PCR reaction, the input of the template was standardized by using 105 ng of genomic DNA. In addition, a reference primer pair for Exon 7 was included for normalization of the qPCR data for Exon 4 and Exon 5. There were no outliers for Ct values for IKZF1 Exons 4, 5 or 7. No BPL case was identified by interpatient comparisons or comparisons between leukemic and normal samples that had a significantly higher Ct value suggestive of an IKZF1 deletion. [B] The Q2 (Median), Q1 (25^th Percentile/Lower quartile) and Q3 (75^th Percentile/Upper quartile) values for the box plots shown in [A]. [C] Bar graphs of Ct data from exon-specific IKZF1 qPCR for 32 BPL patients averaged across 2 independent experiments and then compared with Ct data for normal hematopoietic cells from 2 normal bone marrow specimens averaged from 3 independent experiments in which they were included as controls for side-by-side comparison with leukemic specimens from BPL patients (2 experiments) or BPL xenograft cells (one experiment). The inset shows the mean±SE values. No statistically significant differences were observed between normal vs. BPL specimens as documented by the P-values shown. [D] Bar graphs of Ct data from [C] for Exon 4-specific qPCR and Exon 5-specific qPCR after normalization by referencing the raw data to the Ct values from Exon 7-specific qPCR. The data normalization employed the method reported by Weksberg et al [15] and Moody et al [16] (File S1). [E-H] Depicted are representative cycle number vs. Log (∆Rn) (fluorescence signal) amplification plots obtained using the IKZF1 Exon 4-specific test primers and IKZF1 Exon 7-specific reference primers for the normal hematopoietic cells and randomly picked 5 BPL cases.

https://doi.org/10.1371/journal.pone.0080732.g002

Examination of Leukemia Cells from Patients with Newly Diagnosed Pediatric High Risk BPL, Including Its Ph⁺ Subset, for IKZF1 Deletions by Gene Expression Analysis Using Multiple IKZF1 Probesets

In order to gain further insights into the incidence and biological significance of IKZF1 deletions, we examined the expression levels of IKZF1 transcripts in primary leukemic cells from 327 pediatric Ph^- BPL patients in side by side comparison with 123 pediatric Ph⁺ BPL cases and 74 normal bone marrow specimens obtained from the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database GSE13159 and GSE13351. BLAT analysis on IKZF1 target sequences deposited in Affymetrix NetAffx™ Analysis Center (http://www.affymetrix.com/analysis/index.affx) mapped the 5 probesets to exons 1 to 4 from the Affymetrix Human Genome U133 Plus 2.0 Arrays used in the analysis onto specific IKZF1 exons visualized using the UCSC genome browser (http://genome.ucsc.edu/cgi-bin/hgBlat?command=start). Examination of the Affymetrix probeset coverage in relation to most common IKZF1 microdeletions observed in Ph⁺ BPL cases showed that all of these deletions can be detected by multiple IKZF1 probesets. The most common microdeletion occurs between Exons 4-7 (30%) [14] that would be detected by 1565816_at and 1565818_s_at followed by Exons 2-7 (15%) [14] that would be detected by 1565816_at, 1565818_s_at, 220704_at, 1557632_at). The Mixed Model ANOVA explained 84.2% of the variation in the gene expression data across all 5 IKZF1 transcripts covering Exons 1-4 (31% variation from the random factor for individual cases and 69% from the fixed factors (significant effects of Probeset (F_4,2084 = 1828, P<0.0001), Diagnosis (F_2,521 = 10.9, P<0.0001) and the Interaction term (F_8,2084 = 9.1, P<0.0001) were observed for these fixed factors)). Our analysis demonstrated no changes in expression that would be expected from homozygous or heterozygous deletions of IKZF1 in primary leukemic cells (Figure 3A). In particular, the probesets 1565816_at (specific for Exons 1-4) and 1565818_s_at (specific for Exon 4 only) did not detect any significantly reduced expression levels in Ph⁺ cases comparisons versus Normal or Ph^- cases (Figure 3B) controlling for repeated measures taken from individual cases, and were highly correlated across samples (Figure 3C). Contrary to expectation, both probesets for Exon 3 exhibited significant increases in expression for Ph⁺ cases compared to Ph^- cases (Figure 3B). Taken together, this expression data do not show reductions in levels for Ph⁺ cases that would have suggested (1) homozygous or heterozygous whole IKZF1 locus deletions or (2) intragenic IKZF1 deletions between Exons 4-7 or Exons 2-7.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Exon-specific Detection of IKZF1 Transcripts in Normal Hematopoietic Cells and Primary Leukemic Cells from Pediatric Ph⁺ vs. Ph^- BPL Patients.

IKZF1 transcript levels, as measured by 5 IKZF1 probe sets specific for IKZF1 Exons 1-4, were compared between primary leukemic cells from 123 pediatric Ph⁺ BPL patients, 327 pediatric Ph^- BPL patients and non-leukemic hematopoietic cells from 74 normal bone marrow specimens analyzed in 2 independent studies (viz.: GSE13159 and GSE13351). Depicted in (A) are bar charts of RMA-normalized transcript levels for the 5 IKZF1 probe sets, including 2 probe sets exhibiting Exon 4 specificity to test for reduction in signal due to an intragenic IKZF1 deletion involving a region within Exons 4-7 or Exons 2-7. No significant reduction in expression levels were observed for these 2 Exon 4 probesets (B) and significant increases in expression were observed for both comparisons of Ph⁺ with Normal and Ph^- samples for the two Exon 3 probesets. Heat map depicts up- and down-regulated transcripts ranging from red to green respectively for RMA-normalized IKZF1 transcript levels of leukemic cells mean centered to normal hematopoietic cells and clustered according to the average distance metric (C).

https://doi.org/10.1371/journal.pone.0080732.g003

Constitutive Function of Ikaros in Leukemia Cells from Patients with Newly Diagnosed Pediatric High Risk BPL, Including Its Ph⁺ Subset

Transcript levels for validated IK target genes with documented IK binding sites that are upregulated by IK and downregulated in IK-deficient mice were highly correlated forming a subcluster in the hierarchical cluster representation (Figures S1-S3 in File S1). Forty-two transcripts representing 29 genes out of the 45 validated IK target genes harboring IK binding sites were up-regulated in 390 “high IKZF1 expression” cases compared to 407 “low IKZF1 expression” cases (Figure S1 in File S1). Expression profiles for IKZF1 and these 42 IK target gene transcripts were highly correlated forming a subcluster in the hierarchical cluster representation. We examined the transcript levels and transcription factor function of native IK in primary ALL blast cells in diagnostic leukemic bone marrow samples from 576 B-lineage ALL/BPL and 174 T-lineage ALL patients. 21 of 42 (50%) transcripts representing 14 of 29 (48%) validated human IK target genes were expressed at markedly higher levels in leukemic bone marrow specimens from B-lineage ALL/BPL patients, including most of the patients with BCR-ABL ⁺ /Ph⁺ ALL (Figure S2 in File S1; MANOVA, F_1,748 = 168.7, P<0.0001). Similar results were obtained on diagnostic samples from two independent studies on 59 B-lineage ALL/BPL and 17 T-lineage ALL patients who subsequently experienced a relapse (Figure S3 in File S1; MANOVA, Sum Contrast, F_1,74 = 9.32, P=0.0032).

The transcripts of the IK target genes were not down-regulated in primary leukemia specimens of 122 BCR-ABL⁺ pediatric BPL patients vs. normal hematopoietic cells in 74 healthy bone marrow specimens reported in the “Microarray Innovations in Leukemia” (MILE) study (Figure S4 in File S1). A multivariate analysis of IK target gene expression levels showed that the multivariate mean for primary cells from BCR-ABL⁺ ALL patients was not significantly different from that of normal bone marrow cells (MANOVA, Sum Contrast, F_{1, 194} = 0.768, P=0.382). Thus, when compared to normal bone marrow specimens, the transcription factor function of IK, as measured by transcript expression levels of IK target genes, was not diminished in the leukemic bone marrow specimens from BCR-ABL⁺ pediatric BPL patients.

We next compared the transcript levels for IK target genes in primary ALL cells from 155 BCR-ABL^- and 20 BCR-ABL⁺ patients with high risk BPL. Of the 42 transcripts representing the IK target genes, only 3 transcripts were down regulated in BCR-ABL⁺ ALL cells (Figure 4, Table S1 in File S1). GSEA showed no evidence for down regulation of IK target genes in the BCR-ABL⁺ subset (Figure 4B.1). IK has been shown to direct lymphoid priming during the earliest stages of lymphocyte ontogeny by upregulating the expression of specific genes [17-19]. Of the 20 IK-regulated lymphoid priming genes, only two were down regulated in Ph⁺ leukemia cells (Figure S5 in File S1). A multivariate analysis of lymphoid priming gene expression levels showed that the multivariate mean for primary cells from Ph⁺ BPL patients was significantly higher than that of primary leukemia cells from Ph^- patients (MANOVA, Sum Contrast, F_{1, 173} = 7.04, P=0.0087). GSEA showed no evidence for down regulation of IK-regulated lymphoid priming genes in the Ph⁺ subset (Figure 4B.2). A multivariate analysis of IK target gene expression levels in cytogenetically distinct subsets of BPL showed that the multivariate mean for the Ph⁺ subset was significantly higher than all other groups combined (MANOVA, Sum Contrast, F_1,173 = 11.29, P=0.001) (Figure 4C). For validation of these findings, we next performed a bioinformatics analysis of the publically available international database of the MILE Study Group on gene expression profiles of primary leukemia cells from 122 Ph⁺ and 237 Ph^- pediatric BPL patients (Figure 5). Of the 42 IK target transcripts representing 29 genes, only 7 transcripts representing 5 genes were significantly down regulated (Figure 5A, Table S2 in File S1). Of the 20 IK-regulated lymphoid priming genes, only 3 were down regulated in Ph⁺ leukemia cells (Figure S6 in File S1). GSEA showed evidence for up-regulation (not down-regulation) of IK target genes (P<0.001) (Figure 5B1) and IK-regulated lymphoid priming genes (P=0.013) (Figure 5B2) in the Ph⁺ subset. A multivariate analysis of IK target gene expression levels in cytogenetically distinct subsets of BPL showed that the multivariate mean for the Ph⁺ subset was significantly higher than all other B-precursor ALL groups combined (MANOVA, Sum Contrast, F_1,748 = 168.7, P < 0.0001) (Figure 5C). Taken together, these gene expression profiling data demonstrate that Ph⁺ ALL is not characterized by a functional deficiency of IK, as measured by expression levels of validated IK target genes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Transcript Levels of Ikaros Target Genes in Primary Leukemia Cells from Pediatric Ph^- and Ph⁺ BPL Patients.

Expression levels of IK target genes were compared for primary leukemic cells from 155 pediatric Ph^-/BCR-ABL^- BPL patients and 20 Ph ⁺ /BCR-ABL ⁺ /BPL patients on the Mullighan study (GSE12995). Transcript signal values were obtained from hybridization onto the Affymetrix Human Genome U133A genechip arrays. Heat map depicts up and down regulated transcripts ranging from red to green respectively for mean centered log₁₀ transformed expression values and clustered according to average distance metric (A). Rank ordered difference in standard deviation units for Ph⁺ samples (N=20) compared to other samples (N=155) in the Mullighan study (GSE12995) were processed for enrichment of IK target genes (B1) and IK-regulated lymphoid priming genes (B2) using a supervised approach implemented in GSEA2.08 (Broad institute). Enrichment scores for calculated for the ranked members of the gene sets and normalized to the gene set size (NES) for which the P-value was calculated using 1000 permutations of the pre-ranked gene list and the FDR corrected for testing 2 gene sets. There was a significant enrichment of IK target genes (NES = 1.43, P = 0.046) for Ph⁺ patients that included the leading edge subset comprised of TSPAN13, GSN, MDFIC, ITGA4, TREML2, RNF125, IQGAP2, LAMC1, TES, DHRS3, S100A10, IL12RB1, ADD3, GRAMD3, ATRNL1, RUNX2, MCOLN3, ATP1B1, and CALCRL. Likewise, there was a significant enrichment of lymphoid priming genes (NES = 1.51, P = 0.039) for Ph⁺ patients that included the leading edge subset comprised of IGJ, CNN3, CD52, DNTT, CSF1R, SATB1, LTB, PTGER2, MEF2C, RUNX2 (Figure 1B.2). There was a significant increase in the multivariate mean for 45 transcripts in the Ph⁺ subset of specimens compared to the pooled mean of the other subsets (MANOVA, F_1,173 = 11.29, P=0.001). The mean level of expression for each transcript in each BPL subset is illustrated in the heat map organized using a two-way hierarchical clustering method (average distance metric) to group expression profiles of transcripts and specimen subsets (C).

https://doi.org/10.1371/journal.pone.0080732.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Transcript Levels of Ikaros Target Genes in Primary Leukemia Cells from Pediatric Ph^-/BCR-ABL^- and Ph ⁺ /BCR-ABL⁺ BPL Patients on the MILE Study.

Expression levels of IK target genes were compared between primary leukemic cells (GSE13159) from 122 pediatric BPL patients with t(9;22) translocation (Ph ⁺ /BCR-ABL⁺) and 237 pediatric BPL patients without t(9;22) translocation (Ph^-/BCR-ABL^-). Heat map depicts up and down regulated transcripts ranging from red to green respectively for standardized expression values and clustered according to average distance metric (A). 25 transcripts representing 16 IK target genes were expressed at significantly higher levels in leukemia cells from Ph⁺ patients, no significant differences were observed for 10 transcripts representing 9 IK target genes and 7 transcripts representing 5 IK target genes were significantly down regulated (Table S2 in File S1). Hierarchical cluster analysis identified MDFIC (0.69 SD units, P = 8.8 x 10^-12), DHRS3 (0.6 SD units, P = 3.7 x 10^-11), GSN (0.49 SD units, P = 3.4 x 10^-10), ITGA4 (0.53 SD units, P = 2.6 x 10^-9) and TSPAN13 (0.2 SD units, P = 1.2 x 10^-7) as the most significantly up-regulated genes in the 122 Ph⁺ patient samples. Rank ordered difference in standard deviation units for Ph⁺ samples compared to Ph^- samples for enrichment of IK target genes (B1) as well as lymphoid priming genes (B2) using a supervised approach implemented in GSEA2.08 (Broad institute). Enrichment scores were calculated for the ranked members of the gene sets and normalized to the gene set size (NES) for which the p-value was calculated using 1000 permutations of the pre-ranked gene list and the FDR corrected for testing 2 gene sets. There was a significant enrichment of IK target genes (NES = 2.04, P < 0.001) and lymphoid priming genes (NES = 1.62, P = 0.013) that included the leading edge subsets MDFIC, EIF4E3, DHRS3, ITGA4, PTK2, GSN, S100A10, GRAMD3, F13A1, CALCRL, ATP1B1, LAMC1, TES, ADD3, TCTEX1D1, ATRNL1, IL12RB1, RNF125, MCOLN3, RUNX2, PDLIM2, TSGA10, TREML2 for IK targets and IGJ, CNN3, CSF1R, PTGER2, LTB, DNTT, CD52, MEF2C, and RUNX2 for lymphoid priming genes. There was a significant increase in the multivariate mean for 45 transcripts in the Ph⁺ subset of specimens (MANOVA, F_1,357 = 30.65, P<0.0001). The mean level of expression for each transcript in each BPL subset is illustrated in the heat map organized using a two-way hierarchical clustering method (average distance metric) to group expression profiles of transcripts and BPL subsets (C).

https://doi.org/10.1371/journal.pone.0080732.g005

Our findings are different from the results reported by Iacobucci et al. for adult Ph⁺ BPL patients [20]. Specifically, Iacobucci et al. reported that a group of IK-binding genes represented by 158 transcripts are downregulated in adult Ph⁺ BPL cells with heterozygous IKZF1 deletions. Counterintuitively, IKZF1 itself was not on the Iacobucci list of genes with reduced transcript levels in adult Ph⁺ BPL patients [20]. We examined 5 IKZF1 probe sets specific for exons 1-4 and the 158 probeset Iacobucci signature for IKZF1 deletions (as listed in Table S2 of reference [20]) for their co-regulation in pediatric BPL. Pairwise correlations and clustering of the correlation r-values revealed that the IKZF1 probesets formed a distinct module compared to the Iacobucci gene signature (Figure S7A in File S1). Likewise, the probe sets for the Iacobucci gene signature were correlated with each other and 92% of the correlations were positive with P<0.05. However, the probesets for the Iacobucci genes did not exhibit the expected positive correlations with the IKZF1 probesets interrogating exons 1-4 and 42% of the correlations were negative with P<0.05 while 54% of the correlations did not show a statistical significance. This modular expression profile was also observed with the genes identified to contain IKZF1 binding sites in mice (Figure S7B in File S1) as well as the most significantly down regulated Iacobucci gene set (Figure 6A). We next cross-referenced the 158 probeset Iacobucci signature with gene expression changes in wild type vs. IKZF1 knockout mice and the presence of IKZF1 binding sites from the archived CHiPseq data (GSM803110) [8]. None of the 29 genes from the Iacobucci gene set that exhibited IKZF1 binding sites were down regulated in IKZF1 knockout (“null”) mice [8]. A Gene Set Enrichment Analysis (GSEA) showed no significant enrichment of the downregulated transcripts of the Iacobucci probeset in primary leukemic specimens from Ph⁺ pediatric BPL patients compared to normal bone marrow specimens (Figure 6B). In fact, Ph⁺ patients exhibited a significant over-representation of upregulated probesets (P = 2.8 x 10^-3) that included the leading edge probesets comprised of AEBP1, AFF3, BACH2, BCL2, BDH2, BLK, CD22, DBN1, DKFZP761P0423, ERGIC1, FADS3, FAIM, FLJ32997, FMNL2, HS6ST2, KIAA0802, LOC90925, MAG, MARCKSL1, MGC3032, MSH6, NULL, P4HA2, PCDH9, QRSL1, SMARCC1, TBC1D4, TCL1A, ZHX2). These results demonstrate that the genes reported as a downregulated geneset for adult Ph⁺ BPL patients by Iacobucci are not downregulated in pediatric Ph⁺ BPL patients. The lack of a correlation between the expression levels of these genes and IK expression levels in adult Ph⁺ BPL patients, pediatric Ph⁺ BPL patients, or mice suggest that they are not likely to be genes selectively regulated by IK. The lack of a strong correlation of the IKZF1 probesets with the Iacobucci probeset led us to examine other potential transcription factors that could modulate the expression of these IK-binding but not IK-regulated genes. Specifically, we interrogated the Iacobucci gene signature with the Molecular Signatures Database (http://www.broadinstitute.org/gsea/msigdb/index.jsp) for over representation of transcription factor binding motifs. Significantly over-represented motifs included the binding motif for several transcription factors, including (1) MYC (“CACGTG”; ALDH18A1, ALDH6A1, CAMK2D, CBX5, CLTC, EEF1E1, FADS3, GCSH, LMNB1, PFDN6, PLCG2, SORD, SUPT16H, ZHX2; FDR q-value=0.000423) (2), OCT1 (“MKVATTTGCATATT”; BCL2, BTK, CDCA7, YWHAB, ZFP36L1, ZHX2; FDR q-value = 0.0196) (3), SREBF1 (“TCANNTGAY”; AHCYL1, ALDH6A1, CAMK2D, CAPN3, EIF2S1, SUPT16H, VPS35; FDR q-value = 0.026), and (4) E2F1 (“TTTSGCGS”; ALDH6A1, CDCA7, FLI1, GINS3, MSH2; FDR q-value = 0.026).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Gene Expression Signature for Adult IKZF1 Deletions Reported for Adult Ph⁺ BPL is not Found in Pediatric Ph⁺ BPL.

[A] Correlation cluster forms distinct modules for expression of exon specific probe sets for IKZF1 and significantly down regulated probe sets for adult IKZF1 deletions in primary leukemic cells from pediatric BPL patients. Probeset correlations were clustered for the most significant down regulated probe sets in the Iacobucci study (P<0.01) and IKZF1 probesets specific for Exons 1-4 in pediatric leukemic samples. [B] Rank ordered difference in log₂ transformed RMA expression values between primary leukemic bone marrow specimens from Ph⁺ BPL patients (N=123) and normal bone marrow specimens (N=74) were processed for enrichment of 158 Iacobucci down regulated probe sets using a supervised approach implemented in GSEA2.08 (Broad institute). Significance was assessed using weighted Kolmogorov-Smirnov statistics. Contrary to expectation, the gene signature for probe sets down regulated in IKZF1 deleted adult Ph⁺ BPL specimens were significantly over-represented in probe sets up regulated in Ph⁺ pediatric BPL cases (P-value = 2.8 x 10^-3; hits shown by black lines on the graph and the heat map depicting blue color to indicate down regulation and red color to indicate up regulation of probesets in Ph⁺ BPL cases versus Normal).

https://doi.org/10.1371/journal.pone.0080732.g006

Constitutive Function of Ikaros in Relapse Clones and Xenograft Cells from Pediatric BPL Patients

Gene expression profiles of primary leukemia cells in diagnostic bone marrow specimens of BPL patients who subsequently experienced an early relapse (N=40; time to relapse <36 months) were not suggestive of diminished IK activity when compared to primary leukemia cells from patients who experience a late relapse (N=19; time to relapse ≥36 months) (Figure S8 in File S1). There was a significant increase in the multivariate mean for 45 transcripts (3 transcripts for IKZF1 and 42 transcripts for 29 IK target genes) comparing Early to Late Relapse subsets (Average fold increase = 1.3; MANOVA, F1,57 = 7.11, P=0.01). This finding suggested that high (rather than low) blast cell IK activity may be a biomarker for subpopulations of patients who are at high risk for treatment failure and early relapse.

Gene expression profiling of primary leukemic cells from matched pair relapse vs. diagnostic bone marrow specimens of 59 patients with BPL who relapsed showed similar expression levels for IK target genes (Figure S9 and Table S4 in File S1) (4 transcripts for 4 genes exhibited significant up-regulation of IK targets KIF23, ANKRD6, ATRNL1 and LAMC1 with no significant differences for the remaining IK targets). Since development of aggressive leukemia in mice with heterozygous deletions of the IK exons encoding the zinc fingers required for sequence-specific DNA binding occurs only with a concomitant loss of heterozygosity between 3 and 6 months after birth [19], we sought to determine if leukemic blast cells from aggressive early relapse patients are characterized by homozygous deletions of IKZF1 exons 4-7 involved in reported heterozygous IK deletions in high-risk BPL. To this end, we performed exon-specific PCR with DNA sequencing on purified genomic DNA samples from 1^st relapse bone marrow specimens of 9 BPL patients who experienced an early relapse. As shown in Figure S9B in File S1, no IKZF1 deletions involving E4-E7 were found in any of the 9 cases. Further, no mutations were detected in IKZF1 E4-E7 or adjacent intronic segments. We next used Western blot analysis with an IK-specific antibody to study IK protein expression. Leukemic cells from 9 of 9 relapsed BPL cases studied expressed a ~57-kDa immunoreactive protein corresponding in size to full-length IK1 (Figure S9C in File S1). These results do not suggest that IK deficiency contributes to progression of leukemia or emergence of resistant relapse clones in pediatric BPL.

We also examined the expression of IK in very aggressive in vivo clonogenic human leukemic lymphocyte precursor cells isolated from spleens of xenografted NOD/SCID mice that developed overt leukemia after inoculation with primary leukemic cells from 7 patients with B-precursor ALL. We recently reported that these cells are characterized co-expression of multiple ALL-associated human lymphoid differentiation antigens, including high levels of CD10, CD19, and CD34 that have been reported as markers of putative leukemic stem cells in B-precursor ALL [11]. Xenograft cells from all 7 cases expressed a ~57-kDa immunoreactive protein corresponding in size to full-length IK1 (Figure S10 in File S1). Smaller IK proteins of <47-kDa size (i.e., size of full-length IK2 and IK3) possibly corresponding to dominant negative IK isoforms were also detectable in 2 of the 7 cases. IK expression in ALL xenograft cells from 4 separate BPL cases was also confirmed by using an intracellular flow cytometry method (Figure S10 in File S1, Panel B). Despite the non-specific staining caused by the secondary antibody alone with this assay platform that employs an aqueous buffered solution containing formaldehyde and proprietary ingredients (BD Phospholow Fix Buffer 1, Cat. #: 557870, BD Biosciences, San Jose, CA) for permeabilization of cell membrane, in each of these 4 cases (Figure S10 in File S1, Panels B1-B4), IK expression was confirmed by a marked increase in mean-channel immunofluorescence when the anti-IK antibody was included in the staining procedure (Figure S10 in File S1, Panel B5). Evaluation of the xenograft cells for IK expression at the single cell level using high-resolution confocal microscopy confirmed nuclear and cytoplasmic expression of IK (Figure S10C in File S1). We also used real-time qPCR to examine side-by-side genomic DNA from normal hematopoietic cells in normal bone marrow specimens (N=2) and ALL xenograft cells (N=3) for evidence of intragenic deletions of IKZF1 involving Exon 4. No xenograft case was found to have a significantly higher Ct value than normal controls suggestive of an IKZF1 deletion (Figure S10D in File S1).