- Diana Schotte,
- Ellen A.M. Lange-Turenhout,
- Dominique J.P.M. Stumpel,
- Ronald W. Stam,
- Jessica G.C.A.M. Buijs-Gladdines,
- Jules P.P. Meijerink,
- Rob Pieters and
- Monique L. Den Boer⇓
- Department of Pediatric Oncology and Hematology, Erasmus MC/Sophia Children's Hospital, Erasmus University Medical Center Rotterdam, the Netherlands
- Correspondence: Monique L. den Boer, PhD, MSc, Erasmus MC-Sophia Children’s Hospital, Dept. of Pediatric, Oncology and Hematology, Room Sp2456 P.O. Box 2060 3000 CB Rotterdam the Netherlands. E-mail:
Background Deregulation of microRNA may contribute to hematopoietic malignancies. MicroRNA-196b (miR-196b) is highly expressed in MLL-rearranged leukemia and has been shown to be activated by MLL and MLL-fusion genes.
Design and Methods In order to determine whether high expression of miR-196b is restricted to MLL-rearranged leukemia, we used quantitative stem-loop reverse transcriptase polymerase chain reaction to measure the expression of this microRNA in 72 selected cases of pediatric acute lymphoblastic leukemia i.e. MLL-rearranged and non-MLL-rearranged precursor B-cell and T-cell acute lymphoblastic leukemias. We also determined the expression of HOXA-genes flanking miR-196 by microarray and real-time quantitative polymerase chain reaction. Furthermore, we used CpG island-arrays to explore the DNA methylation status of miR-196b and HOXA.
Results We demonstrated that high expression of miR-196b is not unique to MLL-rearranged acute lymphoblastic leukemia but also occurs in patients with T-cell acute lymphoblastic leukemia patients carrying CALM-AF10, SET-NUP214 and inversion of chromosome 7. Like MLL-rearrangements, these abnormalities have been functionally linked with up-regulation of HOXA. In correspondence, miR-196b expression in these patients correlated strongly with the levels of HOXA family genes (Spearman’s correlation coefficient ≥ 0.7; P≤0.005). Since miR-196b is encoded on the HOXA cluster, these data suggest co-activation of miR-196b and HOXA genes in acute lymphoblastic leukemia. Up-regulation of miR-196b coincides with reduced DNA methylation at CpG islands in the promoter regions of miR-196b and the entire HOXA cluster in MLL-rearranged cases compared to in cases of non-MLL precursor B-cell acute lymphoblastic leukemia and normal bone marrow (P<0.05), suggesting an epigenetic origin for miR-196b over-expression. Although patients with MLL-rearranged acute lymphoblastic leukemia are highly resistant to prednisolone and L-asparaginase, this resistance was not attributed to miR-196b expression.
Conclusions High expression of miR-196b is not exclusively MLL-driven but can also be found in other types of leukemia with aberrant activation of HOXA genes. Since miR-196b has been shown by others to exert oncogenic activity in bone marrow progenitor cells, the findings of the present study imply a potential role for miR-196b in the underlying biology of all HOXA-activated leukemias.
MicroRNA (miRNA) were discovered to be small non-protein coding RNA molecules that post-transcriptionally regulate the expression of many protein-coding genes by complementary binding to their targeted mRNA.1 Subsequently, the bound mRNA is cleaved or, in the majority of cases, its translation into protein is repressed.2 Despite their name, miRNA play major roles in biological processes such as cell proliferation, differentiation and apoptosis and aberrant activities of miRNA have been found in a variety of malignancies. For example, let-7 has been identified as a tumor suppressor since let-7 down-regulates the expression of the oncoprotein Ras. Consequently, reduced expression of let-7 miRNA was associated with an increased expression of Ras in patients with lung cancer, which may explain their unfavorable prognosis.3,4
Recent studies also showed that aberrantly expressed miRNA contribute to hematopoietic malignancies. For example, over-expression of the oncogenic miR-17–92 polycistron accelerated the formation of lymphomas in an Eμ-Myc transgenic mice model.5 This miR-17–92 cluster is often amplified in human B-cell lymphomas, suggesting a role in lymphomagenesis.6 Enforced expression of miR-155 in Eμ-miR-155 transgenic mice resulted in preleukemic pre-B-cell proliferation followed by mature B-cell leukemia, implying a leukemogenic contribution of miR-155.7 Epigenetic silencing of miR-124a caused up-regulation of its target CDK6 in acute lymphoblastic leukemia (ALL), which may drive proliferation of leukemia cells, and is associated with higher relapse and mortality rates among patients with ALL.8
In general, the 5-year disease-free survival is 85% for children with ALL on contemporary treatment protocols. However, patients with some subtypes such as those with a rearrangement of the Mixed Lineage Leukemia (MLL) gene and BCR-ABL1-positive ALL have a much worse 5-year disease-free survival rate of approximately 50%.9–11 We previously observed aberrant miRNA expression patterns in different genetic subtypes of pediatric ALL.12 One of the miRNA that was most aberrantly expressed in MLL-rearranged cases was miR-196b. The expression level of miR-196b was up-regulated by 500- to 800-fold in the majority of cases of MLL-rearranged precursor B-ALL and in about one-third of selected T-ALL cases compared to precursor B-ALL patients without MLL translocations. In addition, the expression of miR-196b in these cases of leukemia was also higher than in normal bone marrow cells.12 Interestingly, in a recent report it was postulated that miR-196b may be involved in leukemogenesis and that its expression is induced by normal MLL and MLL fusion products such as MLL-AF4.13 The MLL-AF4 fusion is most frequently found in infants with MLL-rearranged leukemia,14 who were included in our previous study.12 MLL normally regulates the expression of the homeobox domain (HOX) gene family which plays an important role in regulating normal hematopoiesis.15 The HOXA cluster genes, and especially HOXA4, HOXA5, HOXA9 and HOXA10, are over-expressed in MLL-rearranged ALL.16–18 However, aberrant expression of HOXA genes is not restricted to MLL-rearranged precursor B-ALL cases and has also been reported to occur in T-ALL patients carrying MLL- or HOXA-rearrangements e.g. inversion of chromosome 7, or fusion products including CALM-AF10 and SET-NUP214.19,20 As miR-196b is mapped between the HOXA9 and HOXA10 genes on chromosome 7p15.2, the level of expression of miR-196b may be linked to HOXA gene transcription, irrespective of the HOXA activating mechanism (MLL fusions or other factors). To test this hypothesis, we measured the expression levels of miR-196b and HOX gene-family members in cases of MLL-rearranged and non-MLL-rearranged precursor B-ALL and T-ALL carrying different genetic abnormalities leading to HOXA gene activation. In addition, since the level of expression of various miRNA may be regulated by gene methylation,21 we investigated the methylation status upstream of the miR-196b locus in MLL-rearranged cases. Since MLL-rearranged ALL cases are often highly resistant to prednisolone and L-asparaginase,22 two drugs that form major components of current ALL treatment, we also investigated whether miR-196b expression levels were linked to responsiveness to these two drugs.
Design and Methods
Leukemic cell samples from children with newly diagnosed ALL were obtained after informed consent from the children’s parents or guardians and approval by the institutional review board. The immunophenotype of the samples was determined by flow cytometry (T-ALL or precursor B-ALL), and the genetic subtype by fluorescence in situ hybridization (FISH) and/or reverse transcriptase polymerase chain reaction (PCR).11,20 In total 12 MLL-rearranged precursor B-ALL i.e. five patients carrying t(4;11), six patients carrying t(11;19) and one positive for t(9;11), 38 non-MLL precursor B-ALL and 22 T-ALL cases were included. The HOXA-linked T-ALL subgroup consisted of selected cases characterized by the fusion genes MLL-AF6 (n=2), CALM-AF10 (n=5) and SET-NUP214 (n=3) as well as one case with an inversion of chromosome 7 [inv(7)(p15q35)]. The HOXA-negative T-ALL group consisted of TAL/LMO-rearranged (n=4), TLX3-rearranged (n=2) and T-ALL cases negative for the above-mentioned abnormalities (n=5). Mononuclear cells were isolated from bone marrow or peripheral blood samples using sucrose density centrifugation.23,24 The percentage of leukemic cells was determined on May-Grünwald-Giemsa (Merck, Darmstadt, Germany) stained cytospins. If the percentage was below 90%, samples were enriched by eliminating non-malignant cells with immunomagnetic beads.23,24
Quantitative stem-loop real-time polymerase chain reaction analysis of miRNA and HOXA expression levels
Total RNA was extracted with TRIzol reagent (Invitrogen, Leek, the Netherlands) according to the manufacturer’s guidelines with minor modifications as described before.25 The 2100 bioanalyzer (Agilent, Amstelveen, the Netherlands) was used to determine the quality of total RNA. All RNA samples had an RNA integrity number of 7.5 or greater. MiR-196b expression was measured by real-time quantitative PCR (RT-qPCR) using a specific stem-loop primer and probe combination designed by Applied Biosystems, USA.26 Endogenous small nucleolar RNA 1 (RNU24) was used as the reference for small RNA-input. The levels of expression of HOXA3, HOXA9 and HOXA10 transcripts were quantified relative to the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) using cDNA synthesized from total RNA, as described previously.20 Primer and probe sequences are presented in Online Supplementary Table S1. All RT-qPCR were performed on an Applied Biosystems 7900HT system. Details on sample preparation, primers, probes and the real-time procedure are given in the Online Supplementary Design and Methods.
Gene expression microarray analysis
Affymetrix U133A and U133 plus 2.0 GeneChips (Santa Clara, CA, USA) were used to determine the expression of all HOXA, HOXB and HOXC-family genes in pediatric ALL cases, according to the manufacturer’s guidelines. Data extraction and normalization procedures of the 22,283 probe sets that both arrays have in common have been extensively described elsewhere.23 The data collected are part of a larger data set that has been deposited in NCBI’s Gene Expression Omnibus (GEO)27 and is accessible via GEO numbers GSE13351 and GSE13425.23
Assessment of methylation status
The methylation status of miR-196b and the HOXA cluster was assessed by the differential methylation hybridization (DMH) procedure using 244K CpG island microarrays (Agilent Technologies, Santa Clara, USA). The microarray labeling and hybridization procedures were performed according to Yan et al.28 as described elsewhere.29 The high-resolution microarrays contain 243,497 60-mer oligonucleotide probes, including numerous CpG island probes related to miRNA. For the present study, the probes containing multiple CpG islands located at chromosome 7p15 in the 5’ promoter region of the miR-196b and HOXA cluster genes were used. A pool of genomic DNA derived from ten healthy individual (five males and five females, Promega Benelux BV, Leiden, the Netherlands) constituted a common reference. Data were extracted using Agilent Feature Extraction 9.5.3 software. Subsequently, the data were normalized and differential methylation assessed in the R and Bioconductor Statistical environment, as described elsewhere.29 Methylation data are presented as ratios of the patient’s signal divided by the common reference signal. For nine MLL-rearranged cases, both miRNA methylation and matching miRNA expression levels were measured. Unprocessed genome-wide DNA methylation data were uploaded in the NCBI Gene Expression Omnibus under the GEO Series accession number GSE18400 as part of a previous study.29
Drug resistance assay
Responsiveness to prednisolone or L-asparaginase was determined by a 4-day in vitro methyl thiazolyl tetrazolium (MTT) drug resistance assay as described elsewhere.30,31 The concentration ranges tested were 0.008–250 μg/mL for prednisolone and 0.003–10 IU/mL for L-asparaginase. The concentration of prednisolone or L-aspariginase that was lethal to 50% of the ALL cells (LC50) was taken as a measure of the cellular drug resistance. LC50 values are known to be predictive of clinical outcome30 and are used to adapt treatment regimens.32,33
Differences in the distribution of variables between groups of patients were analyzed by the Mann-Whitney U test. Correlations between miRNA and mRNA levels were determined using the Spearman’s correlation coefficient (Rs). P values were two-tailed and considered statistically significant when less than 0.05.
The expression of miR-196b was measured in 72 pediatric ALL cases at diagnosis. Figure 1 shows that miR-196b was highly expressed in nine out of 12 MLL-rearranged ALL cases and in 14 out of 22 cases of T-ALL. In particular, all CALM-AF10 (n=5), MLL-AF6 (n=2), SET-NUP214 (n=3) and inv(7) (n=1) positive T-ALL cases showed high levels of expression of miR-196b comparable to the levels found in MLL-rearranged cases (Figure 1). Since these specific chromosomal abnormalities are linked to the activation of HOXA genes19,20 and since miR-196b is mapped between HOXA9 and HOXA10, we quantified the expression of the HOXA9 and HOXA10 transcripts by RT-qPCR. A strong correlation between the level of expression of miR-196b and HOXA9 and HOXA10 expression (Rs ≥ 0.7; P≤0.005) was found in MLL-rearranged precursor B-ALL as well as in cases of T-ALL (Figure 2). Figure 2 and Online Supplementary Figure S1 illustrate that patients with low levels of expression of miR-196b (such as non-MLL precursor B-ALL and the majority of T-ALL cases) also had low expression of HOXA genes. In those cases in which the levels of miR-196b were higher (such as the majority of MLL-rearranged precursor B-ALL and one third of T-ALL cases), HOXA levels were also elevated. The levels of HOXA9 and HOXA10 were also significantly correlated with each other, as shown in Figure 2C and 2F (Rs ≥ 0.90; P<0.0001). These data suggest co-expression of miR-196b with HOXA cluster genes. To determine whether this was only restricted to the two adjacent HOXA genes, we also investigated the expression pattern of other HOX-family genes using a second technique. i.e. the Affymetrix human genome microarray platform. The results confirmed the strong correlation between the level of expression of miR-196b and that of HOXA9 (Rs = 0.8; P≤ 0.0001) and HOXA10 (0.8≤ Rs ≤ 0.9; P≤ 0.0001) and also revealed that miR-196b levels correlated with the expression levels of nearly all other HOXA genes represented on the array platform (0.5<Rs<0.8, P<0.05, Figure 3A for all cases and Online Supplementary Figures S2 and S3 for MLL-rearranged ALL and T-ALL cases, separately).
Whereas miR-196b is encoded within the HOXA cluster, another family member, miR-196a, which differs by only one nucleotide from miR-196b, is encoded by miR-196a-1 located in the HOXB cluster (17q21.32) and miR-196a-2 located in the HOXC cluster (12q13.13). The high homology between miR-196a and miR-196b may hamper the discriminative power of the stem-loop RT-qPCR procedure (and any other quantifying method) for determining solely miR-196b expression levels. However, no significant correlations were found between expression levels of miR-196b and those of members of the HOXB and HOXC cluster, except for HOXB5, HOXB7, HOXB9, and HOXB13 (Figure 3B and 3C). Only one out of two and one out of three probe sets designed for HOXB5 and HOXB7, respectively, showed a significant correlation with miR-196b (0.4 ≤ Rs ≤ 0.6; P<0.05, Figure 3B). These correlations were less significant and less strong than the association observed between expression levels of miR-196b and family members of the HOXA-cluster. This suggests that miR-196b and the HOXA cluster are more likely to be co-transcribed than miR-196b and HOXB or HOXC family genes.
Since the expression of miRNA genes and protein-coding genes can be affected by DNA methylation of promoter regions, we analyzed the methylation state of the 5’ region upstream of miR-196b and the promoter region of all HOXA cluster genes. The methylation of the 5’region of miR-196b (Figure 4A) and of the entire HOXA cluster (Figure 4B) was reduced in MLL-rearranged cases compared to in cases of B-ALL without a MLL-translocation as well as in normal bone marrow (P≤0.01 and P<0.05, respectively), which may explain the increased expression levels of both miR-196b and the HOXA cluster in these cases.
We also investigated whether the level of expression of miR-196b was linked to the sensitivity of the MLL-rearranged ALL and T-ALL patients to prednisolone and L-asparaginase, since resistance to these drugs is indicative of an unfavorable outcome.30,31 Figure 5A and 5C show that the in vitro cytotoxicity (LC50 values) for both drugs did not differ between patients with high and low expression levels of miR-196b. In correspondence, patients who were sensitive, intermediately sensitive or resistant to prednisolone or L-asparaginase did not have significantly different levels of miR-196b (Figure 5B and 5D, P>0.05).
In this study we demonstrated a strong association between the expression levels of miR-196b and genes belonging to the HOXA cluster in pediatric ALL. This co-transcription was not restricted to MLL-rearranged cases, but was also found for T-ALL cases characterized by activation of HOXA genes due to non-MLL mechanisms. Hypomethylation of CpG islands in the 5’ upstream/promoter regions of miR-196b and HOXA cluster genes, as demonstrated in MLL-rearranged cases, may explain the high expression levels of this cluster and embedded miR-196b. In vitro resistance to prednisolone and L-asparaginase could not be explained by differential levels of miR-196b expression.
Popovic et al. reported that the expression of miR-196b is regulated by MLL and MLL fusion products.13 In correspondence, we observed high-level expression in MLL-rearranged cases. However, we demonstrated that high expression of miR-196b is not restricted to MLL-rearranged cases but can also be found in patients with other cytogenetic abnormalities that are known to activate HOXA cluster genes, i.e. CALM-AF10, SET-NUP214 and inv(7)(p15q35). The mechanism by which the HOXA cluster is transcriptionally activated may differ between these patients. It has been demonstrated that CALM-AF10, SET-NUP214 and MLL fusions recruit the DOT1L histone methyltransferase that facilitates gene transcription of the HOXA cluster by dimethylation of histone H3 lysine 79 residues (H3K79). The H3K79 dimethylation possibly allows further epigenetic modification that opens up the entire HOXA locus.20,34–36 We here demonstrate CpG island hypomethylation of the HOXA cluster in MLL-rearranged patients suggesting the existence of additional mechanisms driving HOXA expression. Moreover, inv(7) cases have elevated HOXA10 and HOXA11 expression due to the rearrangement of the T-cell receptor beta locus into this region of the HOXA cluster.36 Taken together these findings indicate that high levels of expression of miR-196b and HOXA cluster genes are not exclusively MLL-driven but can also be due to other routes of HOXA locus activation. It should also be noted that not all MLL-rearranged cases have high expression levels of miR-196b and HOXA cluster genes (Figures 1 and 2). This corresponds with the fact that two distinct subgroups, which are separated based on the expression signature of HOXA cluster genes, have been found in MLL-rearranged ALL.17,37
We demonstrated that the expression of miR-196b was strongly correlated with that of most members of the HOXA cluster (Figure 3) whereas there was a less pronounced correlation with the HOXB and HOXC cluster genes in pediatric ALL cells. HOXA3 microarray-based expression levels did not correlate with miR-196b expression levels. This lack of correlation was due to a less optimal array-probe design for probe set 208604_s_at, exemplified by the fact that microarray and quantitative Taqman-based RT-qPCR data did not correlate for HOXA3 whereas these data were highly correlated for other HOXA cluster genes such as HOXA9 and HOXA10 (Online Supplementary Figure S4). Since we cannot rule out the possibility of non-optimal design for other array-probes, correlations between miR-196b and additional HOX genes may have been missed. However, since the miR-196b gene is positioned between HOXA9 and HOXA10 and is transcribed from the same DNA strand as the HOXA cluster, the high co-expression between the miRNA and HOXA genes suggests co-transcriptional activation. Correspondingly, the expression levels of both miR-196b and HOXA9 are restored upon re-expression of Mll in Mll-deficient mouse embryonic fibroblasts.13 A similar co-activation may explain the strong association for miR-10a (positioned between HOXB4 and HOXB5) as well as miR-196a (encoded between HOXB9 and HOXB13) and the HOXB cluster as observed in acute myeloid leukemia.38,39 Recent studies suggest that miRNA, in general, are often expressed at lower levels in cancer cells than in their normal counterparts.40 In the case of ALL, this phenomenon may be caused by a high frequency of CpG island hypermethylation.21 However, the fact that we observed that miR-196b and HOXA-genes are highly co-transcribed may indicate that this region has reduced DNA methylation. We demonstrated that the level of methylation of the CpG islands in the 5’ region of miR-196b and in the promoter region of the entire HOXA cluster is lower in MLL-rearranged cases than in precursor B-ALL patients without MLL-rearrangements and in healthy individuals. Since MLL-rearranged ALL is characterized by hypermethylation of CpG islands across the genome,29 the hypomethylation of the miR-196b/HOXA region is remarkable. Whether this locus displays a similar methylation status in HOXA-linked T-ALL cases needs to be explored.
Both MLL-rearranged and presumably HOXA-linked CALM-AF10-positive T-ALL patients have a poor clinical outcome.11,41 It has previously been shown that both MLL-rearranged and T-ALL pediatric ALL cases are more resistant to prednisolone and L-asparaginase, as determined by in vitro drug cytotoxicity assays.42 These two drugs are extensively used in the treatment of pediatric ALL and resistance to them is indicative of an unfavorable prognosis.30 However, we did not find evidence that miR-196b contributes to resistance to these drugs since patients with high miR-196b expression were not more resistant to both drugs than patients with low miR-196b expression levels. In contrast to a role in drug responsiveness, miR-196b may have leukemogenic potential since ectopic expression of miR-196b resulted in increased proliferation and reduced differentiation of c-Kit+ bone marrow cells of mice.13
In conclusion, we found that aberrant over-expression of miR-196b is not restricted to MLL-rearranged ALL cases (T-ALL or precursor B-ALL) but also occurs in T-ALL patients with other genetic abnormalities that activate the HOXA gene cluster. This observation is of great importance since miR-196b is known to have oncogenic activity13 and may, therefore, play a role in the biology underlying HOXA-activated precursor B-ALL and T-ALL. The high expression of miR-196b has no effect on the level of cellular responsiveness to prednisolone and L-asparaginase in pediatric ALL. The role of miR-196b in leukemogenesis and survival of HOXA-expressing ALL deserves further studies since targeting miR-196b by ‘antagomirs’ reduced the proliferation capacity of MLL-rearranged bone marrow cells of mice.13
Funding: the work described in this paper was financially supported by the Netherlands Organization for Scientific Research (MLDB, DS), the Sophia Foundation for Medical Research (RWS, RP, DJPMS), the Pediatric Oncology Foundation Rotterdam – Quality of Life Foundation (MLDB, ELT, RP) and the Dutch Cancer Society (grant EMCR 2005-3662, MLdB, RP, DS, JBG, RWS).
The online version of this article has a Supplementary Appendix.
Authorship and Disclosures
The information provided by the authors about contributions from persons listed as authors and in acknowledgments is available with the full text of this paper at www.haematologica.org.
- Received February 4, 2010.
- Revision received April 23, 2010.
- Accepted May 17, 2010.
- Copyright© Ferrata Storti Foundation