Different isoforms of the B-cell mutator activation-induced ... - Nature

Leukemia (2010) 24, 66–73 & 2010 Macmillan Publishers Limited All rights reserved 0887-6924/10 $32.00 www.nature.com/leu

ORIGINAL ARTICLE Different isoforms of the B-cell mutator activation-induced cytidine deaminase are aberrantly expressed in BCR–ABL1-positive acute lymphoblastic leukemia patients I Iacobucci1, A Lonetti1, F Messa2, A Ferrari1, D Cilloni2, S Soverini1, F Paoloni3, F Arruga2, E Ottaviani1, S Chiaretti4, M Messina4, M Vignetti3, C Papayannidis1, A Vitale4, F Pane5, PP Piccaluga1, S Paolini1, G Berton6, A Baruzzi6, G Saglio2, M Baccarani1, R Foa`4 and G Martinelli1 1 Department of Institute of Hematology and Medical Oncology, ‘L and A Sera`gnoli’ S Orsola Malpighi Hospital, University of Bologna, Bologna, Italy; 2Department of Clinical and Biological Science, University of Turin at Orbassano, Turin, Italy; 3Gruppo Italiano Malattie Ematologiche Maligne dell’Adulto (GIMEMA) Data Center, GIMEMA Foundation, Rome, Italy; 4Department of Cellular Biotechnologies and Hematology, ‘La Sapienza’ University, Rome, Italy; 5 CEINGE Biotecnologie Avanzate and Department of Biochemistry and Medical Biotechnology, University of Naples Federico II, Naples, Italy and 6Department of Pathology, Section of General Pathology, University of Verona, Verona, Italy

The main reason for the unfavorable clinical outcome of BCR–ABL1-positive acute lymphoblastic leukemia (ALL) is genetic instability. However, how normal B-cell precursors acquire the genetic changes that lead to transformation has not yet been completely defined. We investigated the expression of the activation-induced cytidine deaminase (AID) and its role in clinical outcome in 61 adult BCR–ABL1-positive ALL patients. AID expression was detected in 36 patients (59%); it correlated with the BCR–ABL1 transcript levels and disappeared after treatment with tyrosine kinase inhibitors. Different AID splice variants were identified: full-length isoform; AIDDE4a, with a 30-bp deletion of exon 4; AIDDE4, with the exon 4 deletion; AIDins3, with the retention of intron 3; AIDDE3-E4 isoform without deaminase activity. AID-FL predominantly showed cytoplasmic localization, as did the AID-DE4a and AID-DE3E4 variants, whereas the C-terminal-truncated AID-DE4 showed a slightly increased nuclear localization pattern. AID expression correlated with a higher number of copy number alterations identified in genome-wide analysis using a single-nucleotide polymorphism array. However, the expression of AID at diagnosis was not associated with a worse prognosis. In conclusion, BCR–ABL1-positive ALL cells aberrantly express different isoforms of AID that may act as mutators outside the immunoglobulin (Ig) gene loci in promoting genetic instability. Leukemia (2010) 24, 66–73; doi:10.1038/leu.2009.197; published online 17 September 2009 Keywords: ALL; AID; BCR–ABL1

Introduction The BCR–ABL1-positive acute lymphoblastic leukemia (ALL) defines the most frequent and prognostically the most unfavorable subtype of ALL.1,2 Recently,3 in genome-wide analyses of DNA copy number abnormalities and loss of heterozygosity events, using single-nucleotide polymorphism (SNP) microarrays, a high frequency of genetic alterations of regulators of B lymphoid development and cell cycle was reported in different subtypes of B-progenitor ALL, including the BCR–ABL1-positive one.4,5 These new findings suggested that additional genomic Correspondence: Professor G Martinelli, Department of Institute of Hematology and Medical Oncology, ‘L and A Sera`gnoli’ S Orsola Malpighi Hospital, University of Bologna, Via Massarenti, Bologna 9-40138, Italy. E-mail: [email protected] Received 1 July 2009; revised 12 August 2009; accepted 14 August 2009; published online 17 September 2009

alterations are frequent events in BCR–ABL1-positive ALL, and are likely central to the pathogenesis and poor outcome of this subtype of leukemia. However, despite the advances made in the last few years in the understanding of the genetics of BCR–ABL1-ALL, how normal B-cell precursor cells acquire the genetic changes that lead to transformation and progression have not been completely defined. Feldhahn et al.6 proposed that the aberrant expression of the B cell-specific mutator enzyme activation-induced cytidine deaminase (AID) in BCR–ABL1-transformed ALL cells may act as a mutator not only within but also outside the Ig gene loci. The AID was identified by the Honjo laboratory in 19997,8 in a search for cDNAs that were specifically expressed in B cells on class-switch recombination (CSR) activation. Soon after that, the generation of AID-deficient mice proved that AID is required not only for CSR but also for antibody somatic hypermutation (SHM) in germinal center (GC) B cells,9 and that the deficiency of AID in humans is associated with an immune deficiency called Hyper-IgM type 2 syndrome.10 Today, there is compelling genetic and biochemical evidence showing that AID initiates SHM and CSR by deaminating cytosines to uracil. The resulting uracil–guanidine mismatches are repaired by error-prone DNA repair mechanisms leading to the introduction of mutations.11,12 Deregulated expression of AID has been reported in GC-derived lymphomas and in leukemias or lymphomas originated from B cells at other stages of differentiation.13,14 AID has also been shown to trigger mutations and chromosomal instability more broadly than previously thought, with numerous tumor-related genes, including Myc, Pim1, Pax5, H2afx, Ocab, Rhoh and Ebf1, now shown as AID targets in normal B cells. The ultimate distribution of mutations is determined by a balance between high-fidelity and error-prone DNA repair.15 Besides its aberrant expression, AID is also alternatively spliced into five mRNA variants in addition to the full-length (FL) form in B-cell malignancies including a subset of B-cell chronic lymphocytic leukemia16 and various types of B-cell lymphomas.14,17 Recently, different AID splice variants were also found in normal GC human B cells and blood memory B cells, and have been shown to display different activities ranging from inactivation of CSR to inactivation or heightened SHM activity.18 As few data are available about AID expression and pre-B ALL, we analyzed AID expression in 61 adult de novo BCR–ABL1positive ALL patients. AID expression was found in 59% of Ph þ ALL and different AID splice variants were identified. To further investigate whether AID expression would be correlated with

AID in BCR–ABL1-positive ALL I Iacobucci et al

67 increased DNA single-strand breaks, we performed a genomewide analysis by 250K NspI and SNP 6.0 arrays (Affymetrix Inc., Santa Clara, CA, USA) and found that ALL patients with expression of wild-type AID have a higher number of copy number alterations (CNAs) compared with patients not expressing AID. Finally, we searched whether recurring CNAs were identified in genes with an established role in leukemogenesis focusing on PAX5, CDKN2A, CDKN2B, BTG1 and EBF1.

Patients and methods

Patients After obtaining informed consent, we collected peripheral blood or bone marrow in Vacutainer tubes with EDTA anticoagulated from 61 patients with BCR–ABL1-positive adult ALL patients. They had a median age of 56 years (range: 18–76) and showed a median blast percentage at diagnosis of 90% (range: 18–99). Diagnosis of all ALL cases was made on the basis of morphological, biochemical and immunological features of the leukemic cells. In addition, the human BCR–ABL1 leukemia cell lines SD1 and BV173 were also included in the analysis. Cell lines were obtained from DMSZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Braunschweig, Germany) and maintained in culture following the DMSZ recommendations.

AID transcript analysis Mononuclear cells were separated by Ficoll-Hypaque density gradient centrifugation, and after RNA extraction, 1 mg was used for cDNA synthesis as previously described.19 To set up a screening for AID transcript variants, cDNA was amplified with a pair of oligonucleotides whose forward primer conjugated with a fluorescent dye (fluorescein, excitation occurs at 494 nm and emission at 521 nm) at its 50 . Primer sequences were: F1: 50 -CTATGGACAGCCTCTTGATGAACC-30 and R1: 50 -CCCAAA GTACGAAATGCGTCTCG-30 . PCR was performed using 1 U of Fast Start (Roche Diagnostics, Mannheim, Germany) DNA polymerase and a final concentration of 1.5 mM MgCl2, on a BIOMETRA Tpersonal thermal cycler (Biometra biomedizinische Analytik GmbH, Goettingen, Germany) set for an initial denaturation at 95 1C for 5 min, for 40–45 cycles with denaturation at 95 1C for 30 s, annealing at 62 1C for 30 s, extension at 72 1C for 45 s, and a final cycle at 72 1C for 10 min and at 60 1C for 45 min to stabilize the fluorescence. In total, 1 ml of each amplicon was added to 9 ml of formaldehyde (Sigma-Aldrich, St Louis, MO, USA) containing 0.2 ml of GeneScan 500 (250) LIZ size standard (Applied Biosystems, Foster City, CA, USA) and loaded on the ABI Prism 3730 DNA Analyzer for automated capillary gel electrophoresis, and the results were plotted with the AbiPrism GeneMapper v3.5 software (Applied Biosystems). The GeneMapper electrophoretograms displayed information about transcript length, peak height and peak area.

Cloning and sequencing analyses Nucleotide sequences of all the observed amplicons were validated by repeating the PCRs with 50 -unmodified primers and cloning the products into pcR2.1-TOPO vectors using the TOPO TA cloning kit and related protocol (Invitrogen, San Diego, CA, USA). TOP10F’ (controllare) strain E. coli cells (Invitrogen) were used as a host for transformation, and colonies containing the recombinant plasmids were screened by PCR with the primer pair for the appropriate amplicon and at the same conditions

described previously. PCR products were purified with the QIAquick PCR purification kit (Qiagen, Hilden, Germany) and directly sequenced by using the ABI PRISM 3730 automated DNA sequencer (Applied Biosystem) and a Big Dye Terminator DNA sequencing kit (Applied Biosystem).

Monitoring of BCR–ABL transcript levels BCR–ABL1 mRNA transcript levels were detected at diagnosis and during follow-up using a standardized real-time quantitative PCR method that was established within the framework of the EU Concerted Action.20 Real-time quantitative PCR was performed on an ABI PRISM 7900 Sequence Detector (Perkin Elmer, Foster City, CA, USA). The quantification principles and procedure using the TaqMan probe (Applied Biosystems) have been previously described.21 All real-time RT-PCR experiments were performed in duplicate.

Western blotting Cells were lysed with sample buffer (2% SDS in 125 mM TrisHCL, pH 6.8). Cell lysates were subjected to SDS-PAGE (polyacrylamide gel electrophoresis) on 12% gels and then transferred to nitrocellulose membranes (Amersham Biosciences, Cologno Monzese, MI, Italy). The blots were incubated for 60 min in the Odyssey blocking buffer before incubation overnight (4 1C) with the polyclonal anti-AID antibody (L7E7 Mouse mAb, Cell Signaling Technology, Danvers, MA, USA). Blotted proteins were detected and quantified using the Odyssey infrared imaging system LI-COR (LI-COR Biosciences, Bad Homburg, Germany).

Subcellular localization studies using confocal laser scanning microscopy The subcellular localization of AID proteins was examined by immunofluorescence and confocal laser scanning microscopy as previously described.22

SNP microarray analysis Genomic DNA was extracted using the DNA Blood Mini Kit (Qiagen, Valencia, CA, USA) from mononuclear cells isolated from peripheral blood or bone marrow aspirate samples by Ficoll gradient centrifugation. DNA was quantified using the Nanodrop Spectrophotometer (Thermo Scientific, Wilmington, DE, USA) and quality was assessed by using the Nanodrop and by agarose gel electrophoresis. Samples were genotyped with GeneChip Human Mapping 250K NspI and Genome-Wide Human SNP 6.0 array microarrays (Affymetrix Inc.) and according to the manufacturer’s instructions. Array image data were analyzed using the Affymetrix GCOS v. 1.4 operating software (Affymetrix Inc.) and Genotyping Console v. 3.0 (Affymetrix Inc.) to derive .cel data files, which were also exported to the Partek Genomic Suite (Partek Inc., Saint Louis, MO, USA) for further data visualization and analysis. All aberrations were calculated with respect to a set of 270 Hapmap normal individuals and a set of samples obtained from acute leukemia cases in remission to reduce the noise of raw copy number data. When available, to exclude inherited copy number variants, a comparison with paired constitutional DNA and with paired remission DNA was performed. Copy number aberrations were also analyzed and eventually confirmed using Genotyping Console 3.0 (Affymetrix Inc.). Leukemia


Statistical analysis The primary study end points were achievement of complete remission (CR), duration of first CR (in terms of disease-free survival (DFS) and of cumulative incidence of relapse), and overall survival. Median follow-up time was estimated by reversing the codes for the censoring indicator in a Kaplan– Meier analysis.23 Differences in the distributions of prognostic factors in subgroups were analyzed by w2 or Fisher’s exact test, and by the Kruskal–Wallis test. Survival was defined as the time from diagnosis to death or date of last follow-up. DFS and relapse incidence were calculated from the time of achieving CR to relapse, death or date of last follow-up. The probabilities of overall survival and DFS were estimated using the Kaplan– Meier method,23 and the probability of cumulative incidence24 of relapse was estimated using the appropriate non-parametric method, considering death in CR as a competing risk. The logrank test was used to compare the treatment effect and risk factor categories, and confidence intervals were estimated (95% CIs) using the Simon and Lee method.25 Cox proportional hazard regression models were used to examine and check for treatment results and the risk factors affecting the time to the event. All tests were two-sided, accepting Pp0.05 as indicating a statistically significant difference. All analyses were performed using the SAS software (SAS Institute, Cary, NC, USA).

Results

Aberrant AID expression in Ph þ ALL depends on BCR–ABL1 kinase activity Studying AID mRNA and protein expression in 61 cases of Ph þ ALL, AID expression was detected in 36 of 61 (59%) patients. To investigate whether the AID expression was correlated to the BCR–ABL1 kinase activity, we simultaneously monitored in vivo the BCR–ABL1 mRNA levels by quantitative PCR and the AID mRNA and protein expression before and after treatment with tyrosine kinase inhibitor (TKI) in eight diagnosis–remission– relapse matched samples. One patient was treated with imatinib and the remaining patients with dasatinib. We found that the AID expression correlated with the BCR–ABL1 mRNA levels: patients were AID þ at diagnosis when the median ratio percentage of BCR–ABL1/ABL was 30.86 (range: 1.30–85.90), but they became negative during treatment with tyrosine kinase inhibitors at the time of remission when they reached undetectable levels of BCR–ABL1 mRNA (median ratio percentage of BCR–ABL1/ABL is less than 0.0001). As soon as the BCR– ABL1 transcript levels increased because of the loss of response, expression of AID was once again detected (Figures 1a and b).

Wb AID

22 kDa

Wb Actin

42 kDa 100 Log10 Ratio BCRABL1/ABL x100

68

10 1 0.1 0.01 0.001 0.0001 0.00001 Diagnosis

Remission

Relapse

Figure 1 Correlation of activation-induced cytidine deaminase (AID) expression with BCR–ABL1 mRNA transcript levels. (a) AID protein expression in patient-derived Ph þ acute lymphoblastic leukemia (ALL) cells before and during treatment with TKI. Anti-actin western blots of the whole cell lysates were performed as controls. BV-173 and SD-1 cell lines were used as AID-positive controls. (b) Monitoring by quantitative PCR of BCR–ABL1 transcript levels in ALL patients before TKI treatment, at the remission and at the time of relapse (both hematological/cytogenetic and molecular). Results were expressed as ratio percentage of BCR–ABL1/ABL.

which led to a C-terminal truncation because of a frameshift. In two patients, we also found a band of 884 bp corresponding to AIDins3, which is characterized by the retention of intron 3. This variant maintains the cytidine domain responsible for the activity of cytidine deamination but lacks the CSR domain because of a frameshift. Overall, 4 of 61 Ph þ ALL patients (7%) expressed the AID DE3-E4 isoform without deaminase activity (deletion of exons 2 and 3) but retaining intact the CSR domain (Figures 2 and 3). Thus, the splicing variants would display an assortment of functional defects when compared with the FL counterpart, and because of this, may constitute a novel mechanism of regulation of CSR and SHM.

AID splice variants have different cellular localization Different AID splice variants are expressed The human AID protein possesses multiple functional domains important for the physiological functions of AID in CSR and SHM, and/or for its potential pathological role in tumorigenesis (Figure 2). Different isoforms of AID were expressed following alternative splicing in BCR–ABL1-positive cells. In 13 of 61 (21%) Ph þ ALL patients, the FL isoform of 592 bp (GeneBank accession number NM_020661) was identified. In 19 of 61 (31%) Ph þ ALL patients, we identified the coexpression of the wild-type isoform and minor bands of various sizes comprising differing domains. In particular, we found a band of 562 bp corresponding to an isoform with a 30-bp deletion of exon 4 (AID-DE4a) and a band of 476 bp corresponding to an isoform characterized by the deletion of the entire exon 4 (AID-DE4), Leukemia

Next, we asked whether the different splicing variants could have altered cellular localization as a result of splicing-induced alterations in the nuclear export signal (NES) in its Cterminus.26,27 Physiologically, AID is a nucleocytoplasmic shuttling protein with a bipartite nuclear localization signal and a NES in its N and C termini, respectively. The amount of AID in the nucleus at the steady state must be tightly controlled to maintain a minimal level, because excessive amounts of nuclear AID may induce unregulated SHM not only in the Ig genes but also in the non-Ig genes.26 As shown in Figure 4, we found that AID-FL showed predominant cytoplasmic localization (panel a), as did the AID-DE4a and AID-DE3E4 variants (panels b and c). This observation was because of the fact that internal in-frame deletions by splicing do not affect either the nuclear localization signal or NES domains, and this is in agreement with previous


69 45

1 SHM/ NLS

AID FL

AID ins3

M 1000 600 500 200 AID ΔE4-E3

AID ΔE4a

AID ΔE4

54

Dimeri zation

94

184

Deaminase activity

1-3 1 e1x

4-52 ex2

ex3

2 e1x

ex2

ex3

3 e1x

ex2

ex3

4 e1x

ex2

52-142

198

Linker

CSR/NES

143-181

184-198

ex4

ex5

S

AID FL

ex5

S

AID ΔE4a

ex4

S

AID ΔE4

ex5

ex5

S

AID ΔE3-E4

intron3 5 e1x

ex2

ex3

S

ex4

ex5 ex5

AID ins3

Figure 2 Alternative splicing of activation-induced cytidine deaminase (AID) mRNA. (a) Bands generated by reverse transcriptase (RT)-PCR using primers derived on exons 1 and 5 and corresponding to the alternatively spliced products of the AID pre-mRNA transcript. PCR products were detected by ethidium bromide staining of 1.5% agarose gel. The left lane is the molecular size marker, Marker VI (Roche); (b) Schematic comparison of the functional domain organization in AID-splicing variants. FL: full-length.

papers indicating that predominant cytoplasmic localization of AID is normal because B cells cannot tolerate AID being predominantly localized in the nucleus.26,28 However, the C-terminal-truncated AID-DE4 showed a slightly increased nuclear localization (panel d) (P ¼ 0.0001).

AID promotes genetic instability Activation-induced cytidine deaminase is an enzyme that physiologically hypermutates Ig genes introducing DNA single-strand breaks. In the absence of effective DNA repair, such DNA single-strand breaks may lead to genetic lesions, such as deletions or chromosomal translocations. To investigate whether AID expression in BCR–ABL1-positive ALL was associated with an increased number of genomic lesions, we performed a genome-wide analysis by SNP arrays (Affymetrix Inc., 250K NspI and SNP 6.0). Interestingly, AID-positive leukemias had a higher number of alterations compared with AID-negative leukemias: cases with AID-positive BCR–ABL1 leukemia had a median CNA of 14 (range: 6–50) compared with 5 CNAs in AID-negative leukemias (range: 2–8, P ¼ 0.02). Genes that were recurrently subject to CNAs were compared for their relative frequency of deletion/amplification events in AIDpositive and -negative leukemias. Recurring copy number abnormalities were identified in genes with an established role in leukemogenesis, such as CDKN2A, CDKN2B, GADD45, PAX5, BTG1, EBF1 and MDS1. In particular, we found that the pattern of AID expression correlated with a high frequency of DNA single-strand breaks within the tumor suppressor genes CDKN2A (Arf) and CDKN2B (INK4B), located at 9p21 (P ¼ 0.03) (Table 1). These findings are in agreement with those reported by Feldhahn et al.,6 showing that AID expression is required to induce DNA single-strand breaks in the CDKN2B gene. Moreover, our results suggest that the increased expression and aberrant activity of AID not only leads to additional genetic lesions but is also associated with poor outcome. To test this possibility, we investigated a potential correlation between outcome and AID expression in Ph þ ALL cells.

AID expression is not correlated with a poor outcome Patient’s characteristics. Of 61 BCR–ABL1-positive ALL patients, 56 (67%) could be evaluated for correlation with

clinical outcome, and for 5 patients the clinical data were not available, as they were lost at follow-up. The characteristics of the patients are shown in Table 2. The median follow-up was 14.8 months (range: 0.4–148.1 months). Initial white blood cells counts ranged between 1.400 and 146.000 per pl (median, 28.100 per pl). No patient had symptomatic central nervous system disease at diagnosis. All cases were evaluable for central review. A total of 46 patients (82%) were enrolled in the GIMEMA clinical trials (3 patients in the GIMEMA LAL 0201-B protocol, 1 patient in LAL2000 and 42 patients in the LAL1205 protocol), whereas 10 patients (18%) were enrolled in institutional protocols. The therapy schemes related to the individual protocols have been described earlier.22,29–32

AID expression and prognosis. Among these patients, AID expression (wild-type isoforms and splicing-derived isoforms containing the cytidine deaminase domain) was detected in 28 cases (50%). The remaining patients (50%) were negative (n ¼ 24) or expressed AID isoforms lacking the cytidine deaminase domain (n ¼ 4). All patients started induction therapy and were consequently evaluable for complete CHR (by intention to treat or ITT). If a patient had lost the final evaluation of the treatment, he/she was evaluated at the last visit available (last observation carry forward or LOCF). All patients (100%) who were enrolled in protocol LAL1205 (Dasatinib frontline) and LAL-0201-B (imatinib and steroids as frontline therapy for elderly patients) obtained a complete hematological response. Therefore, patients evaluable for a correlation between CHR and AID expression were 10 (18%), but among these only 1 had a loss of response, and therefore the association with response was not reachable. The median time of cumulative incidence of relapse of the entire population was 17 months, but we did not find a significant difference between patients who expressed AID and patients who were negative (P ¼ 0.5636) (Supplementary Figure 1SA). AID expression was also not significantly correlated with overall survival (P ¼ 0.8034) and DFS (P ¼ 0.5178) (Supplementary Table 1S and Supplementary Figure 1SB-C). The survival did not also change when different AID variants were correlated with DFS (Supplementary Table 2S). In the Supplementary Table 3S, results from the multivariate analysis were reported. Leukemia


70 40000

200

300

400

500

600

700

800

AID FL (592 bp) 30000 AID ΔE4 (476 bp) AID ΔE4a (562 bp)

20000

10000

AID ΔE3-E4 (205 bp)

AID ins3 (884 bp)

0 AID ΔE3-E4 Ex2

AID ΔE4 Ex5

Ex3

AID ΔE4a Ex3

Ex5

AIDins3 Ex4 del

Ex3

Intron3

Figure 3 (a) Electropherogram of activation-induced cytidine deaminase (AID) PCR products performed using a forward primer conjugated with the fluorescein dye at its 50 . Different AID isoforms were represented in the electropherogram by different peaks. The x axis displays the computed length of the PCR product in base pairs, as determined automatically by the use of an internal lane standard. The y axis represents the peak height in fluorescence units; (b) Pherograms of sequencing of different AID isoforms.

Discussion Physiological AID is required for CSR and SHM in B cells of GC origin by targeting Ig gene loci. However, AID can also target many other endogenous gene loci in normal GC B cells, and the frequency of AID-mediated mutagenesis of non-Ig genes is greater in malignant B cells.13,33 A recent work15 showed that AID has the potential to trigger mutations and chromosomal instability more broadly than previously thought, with numerous tumor-related genes, including Myc, Pim1, Pax5, H2afx, Ocab, Rhoh and Ebf. In many of these genes, there was a clear correspondence between regions targeted by AID and sites of chromosomal deletions and translocations in human B-cell lymphomas.34 The ultimate distribution of mutations is determined by a balance between high-fidelity and error-prone DNA repair. Besides its aberrant expression, AID is also alternatively spliced Leukemia

into four mRNA variants in addition to the FL form in B-cell malignancies,16 including a subset of B-cell chronic lymphocytic leukemia and various types of B-cell lymphomas.14,17 There is also limited evidence that AID alternative splicing may occur in normal B lineage cells.18 As alternative splicing generates isoforms with different exon organization, it significantly impairs AID functional activity. Here, we reported for the first time in BCR–ABL1-positive ALL the expression of different AID isoforms derived from alternative splicing maintaining (FL, AID-DE4a, AID-DE4, AIDins3) and not maintaining (AID DE3-E4) the cytidine deaminase domain. Among 61 de novo adult BCR–ABL1-positive ALL patients, AID mRNA and protein were detected in 36 (59%); their expression correlated with BCR–ABL1 transcript levels and disappeared after treatment with tyrosine kinase inhibitors at the time of remission. Different isoforms of AID were identified: 13 of 61 (21%) patients expressed the FL isoform; 19 of 61 (31%)


71

Figure 4 Subcellular localization of activation-induced cytidine deaminase (AID) variants in leukemic cells from Ph þ acute lymphoblastic leukemia (ALL) patients. In all images, cells were stained with the AID antibody (green) and with propidium iodide (red) to visualize the DNA. In a, b and c, there are confocal images of leukemic cells expressing AID-FL, AID-DE4a and AID-DE3E4 variants, respectively. In d are the confocal images of leukemic cells expressing the C-terminal-truncated AID-DE4. We performed a quantification of the signal intensities and the increased nuclear localization was statistically significant just in AID-deltaE4 (d) (P ¼ 0.0001), whereas the variation that occurred in AID-DE4a and AIDDE3E4 was not significant (P ¼ 0.8 and P ¼ 0.9, respectively) with respect to the AID wild type.

Table 1

Distribution of genomic abnormalities as detected by SNP array analysis in correlation with AID expression

Locus

Gain/loss

1q42.12 3q13.2 4q23-q25 4q13 5q34 7p13 7p21.3 8q24.3 8q24.3 9p13 9p13 9p13 9p21 9p21 9p21 10q24.1 10q26.1 11q12 12q21 17q25

Gain Loss Loss Gain Loss Loss Loss Gain Gain Loss Loss Loss Loss Loss Loss Loss Gain Gain Loss Gain

Start

End

Size (bp)

Functional candidates

223991083 113653130 109210824 74866956 158184727 50391457 13845989 144870494 144945077 36489448 36876065 37106493 21792634 21957750 21992901 97941444 135057610 61648020 91061033 73721871

224233120 113669441 109229104 74934145 158255804 50420046 14119084 144876621 144969537 36717599 36978851 37173936 21855969 21965038 21999312 98021323 135084164 61677211 91063751 73733311

242037 16311 18280 67189 71077 28589 273095 6127 24460 228151 102786 67443 63335 7288 6411 79879 26554 29191 2718 11440

PYCR2 BTLA, CD200 LEF1 CXCL6 EBF1 Contained within IKZF1 ETV1 MAPK15 SCRIB MELK Contained within PAX5 Overlaps with ZCCHC7 Overlaps with MTAP CDKN2A CDKN2B BLNK MTG1 INCENP BTG1 BIRC5

AID-negative (%)

AID-positive (%)

P-value (w2)

0 0 14 0 0 57 14 0 0 14 43 14 14 14 14 0 0 0 0 0

13 25 19 13 13 75 13 19 19 25 25 19 50 63 63 13 19 6 13 19

0.3276 0.1455 0.7949 0.3276 0.3276 0.3918 0.9069 0.2192 0.2192 0.5665 0.3918 0.7949 0.1063 0.0332 0.0332 0.3276 0.2192 0.4988 0.3276 0.2192

Abbreviations: AID, activation-induced cytidine deaminase; SNP, single nucleotide polymorphism. The genes whose alterations are significantly correlated with AID expression are expressed in bold. Leukemia


72 Table 2 Clinical and biological characteristics of 56 adult Ph+ acute lymphoblastic leukemia patients Variable Age (years, median-range) Blast (median-range)

n (%) 53.2 (13.2–73.7) 90 (28–99)

AID expression Wild-type isoforms and splicing-derived isoforms Negative Isoforms lacking deaminase domain

28 (50.0) 24 (41.07) 4 (8.93)

Gender Male Female

29 (51.79) 27 (48.21)

Protocol treatment LAL2000 LAL1205 LAL0201B Other

1 42 3 10

Leukocytes (median-range) Molecular BCR–ABL P210 P190 P210 and P190

(1.79) (75.00) (5.36) (17.86)

21.8 (1.4–146.0) 15 (26.79) 34 (60.71) 7 (12.50)

mutations.36 However, our failure to correlate the AID expression with clinical outcome in vivo suggests that although BCR–ABL1-dependent AID-driven genomic instability may contribute to the development/malignancy of Ph-positive ALL (where numerous genetic aberrations are detectable at diagnosis), it does not contribute to the relapse. As TKIs abrogate AID expression (Figure 1a) and Ph-positive ALLs usually display relatively short-term response to the inhibitors, it is plausible that relapses originate from already existing pre-TKI malignant clones. In this scenario, AID contributes to the induction of Ph-positive ALL, but not to its progression. Furthermore, the failure to correlate AID expression with clinical outcome could be attributed to the heterogeneity of treatments and the short follow-up. It is noteworthy that the majority of patients were treated with dasatinib as frontline therapy, which induced a complete hematological remission in all patients precluding us from performing any correlative analysis. Mouse models are currently under investigation by our group to show whether the expression of AID is responsible in vivo for disease progression. In conclusion, our findings showed that BCR–ABL1-positive ALL cells aberrantly express different isoforms of AID that can act as a mutator outside the Ig gene loci in promoting genetic instability in leukemia cells.

Conflicts of interest

Abbreviation: AID, activation-induced cytidine deaminase.

The authors declare no conflict of interest. coexpressed the wild-type and different AID splice variants with deaminase activity (AIDDE4a, with a 30-bp deletion of exon 4; AIDDE4, with the exon 4 deletion; AIDins3, with the retention of intron 3–4); 4 of 61 (7%) expressed the AIDDE3-E4 isoform without deaminase activity (deletion of exons 2 and 3). Recently, Wu et al.18 showed that the DE4a and DE4 variants express a higher level of SHM activity than FL AID, as the DE4 variant completely lacks the CSR domain, and the 10-amino acid in-frame deletion in DE4a leads to the structural repositioning of the C-terminal domain that may likewise alleviate the inhibitory influence of an intact CSR domain on SHM.35 AID has a nuclear localization signal and a NES, and can thereby dynamically shuttle between the nucleus and cytoplasm. The cellular localization of AID seems to be essential to its function, and excessive amounts of nuclear AID have been shown to induce unregulated SHM not only in the Ig genes but also in the non-Ig genes.26 As alternative splicing may induce alterations in the NES in its C-terminus, we found that AID-FL showed predominant cytoplasmic localization, as did the AID-DE4a and AID-DE3E4 variants, whereas the C-terminal-truncated AID-DE4 showed a slightly increased nuclear localization pattern. To investigate whether AID introduces DNA single-strand breaks, we next performed a genome-wide analysis by SNP array (Human Mapping NspI 250K and Genome-Wide SNP6.0, Affymetrix Inc.). Patients who expressed wild-type AID had a higher number of alterations compared with AID-negative patients (median CNA of 14 versus 4, respectively, Po0.02). In particular, we found that the pattern of AID expression correlated with a high frequency of DNA single-strand breaks within the tumor suppressor genes CDKN2A (ARF) and CDKN2B (INK4B), located at 9p21 (P ¼ 0.03). Mouse modeling of Ph þ ALL revealed that Arf inactivation attenuates responsiveness to targeted BCR–ABL kinase inhibitors, enhances the maintenance of leukemia-initiating cells within the hematopoietic microenvironment and facilitates the emergence of malignant clones that harbor drug-resistant BCR–ABL kinase Leukemia

Acknowledgements This work was supported by AIL, European LeukaemiaNet, AIRC, Fondazione Del Monte di Bologna e Ravenna, FIRB 2006, Ateneo 60% grants and Gimema Onlus Working Party ALL and CML. We specially thank Serena Formica (University of Bologna) and Annalisa Astolfi (University of Bologna) for help in performing SNP arrays, and Barbara Lama (University of Bologna) for collecting clinical data. We also acknowledge all the people who took care of the patients involved in the GIMEMA studies (see Appendix in Supplementary Information). Conception and design II, MB, GM; Provision of study materials or patients, CP, AV, FP; PPP; SP, GP, RF; Collection and assembly of data, AL; AF; SS; FA; SC; MM; FP, MV; Data analysis and interpretation II, DC, FM, AB, EO; Paper writing II; final approval of paper, GM.

References 1 Radich JP. Philadelphia chromosome-positive acute lymphocytic leukemia. Hematol Oncol Clin North Am 2001; 15: 21–36. 2 Faderl S, Jeha S, Kantarjian HM. The biology and therapy of adult acute lymphoblastic leukemia. Cancer 2003; 98: 1337–1354. 3 Mullighan CG, Miller CB, Radtke I, Phillips LA, Dalton J, Ma J et al. BCR-ABL1 lymphoblastic leukaemia is characterized by the deletion of Ikaros. Nature 2008; 453: 110–114. 4 Itoh K, Friel J, Kluge N, Kina T, Kondo-Takaori A, Kawamata S et al. A novel hematopoietic multilineage clone, Myl-D-7, is stromal cell-dependent and supported by an alternative mechanism(s) independent of stem cell factor/c-kit interaction. Blood 1996; 87: 3218–3228. 5 Paulsson K, Cazier JB, Macdougall F, Stevens J, Stasevich I, Vrcelj N et al. Microdeletions are a general feature of adult and adolescent acute lymphoblastic leukemia: Unexpected similarities with pediatric disease. Proc Natl Acad Sci USA 2008; 105: 6708–6713. 6 Feldhahn N, Henke N, Melchior K, Duy C, Soh BN, Klein F et al. Activation-induced cytidine deaminase acts as a mutator in


73 7

8 9

10

11 12 13 14

15 16

17

18

19

20

21

BCR-ABL1-transformed acute lymphoblastic leukemia cells. J Exp Med 2007; 204: 1157–1166. Muramatsu M, Sankaranand VS, Anant S, Sugai M, Kinoshita K, Davidson NO et al. Specific expression of activation-induced cytidine deaminase (AID), a novel member of the RNA-editing deaminase family in germinal center B cells. J Biol Chem 1999; 274: 18470–18476. Honjo T, Kinoshita K, Muramatsu M. Molecular mechanism of class switch recombination: linkage with somatic hypermutation. Annu Rev Immunol 2002; 20: 165–196. Muramatsu M, Kinoshita K, Fagarasan S, Yamada S, Shinkai Y, Honjo T. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell 2000; 102: 553–563. Revy P, Muto T, Levy Y, Geissmann F, Plebani A, Sanal O et al. Activation-induced cytidine deaminase (AID) deficiency causes the autosomal recessive form of the Hyper-IgM syndrome (HIGM2). Cell 2000; 102: 565–575. Peled JU, Kuang FL, Iglesias-Ussel MD, Roa S, Kalis SL, Goodman MF et al. The biochemistry of somatic hypermutation. Annu Rev Immunol 2008; 26: 481–511. Di Noia JM, Neuberger MS. Molecular mechanisms of antibody somatic hypermutation. Annu Rev Biochem 2007; 76: 1–22. Pasqualucci L, Guglielmino R, Houldsworth J, Mohr J, Aoufouchi S, Polakiewicz R et al. Expression of the AID protein in normal and neoplastic B cells. Blood 2004; 104: 3318–3325. Greeve J, Philipsen A, Krause K, Klapper W, Heidorn K, Castle BE et al. Expression of activation-induced cytidine deaminase in human B-cell non-Hodgkin lymphomas. Blood 2003; 101: 3574–3580. Liu M, Duke JL, Richter DJ, Vinuesa CG, Goodnow CC, Kleinstein SH et al. Two levels of protection for the B cell genome during somatic hypermutation. Nature 2008; 451: 841–845. Albesiano E, Messmer BT, Damle RN, Allen SL, Rai KR, Chiorazzi N. Activation-induced cytidine deaminase in chronic lymphocytic leukemia B cells: expression as multiple forms in a dynamic, variably sized fraction of the clone. Blood 2003; 102: 3333–3339. Babbage G, Garand R, Robillard N, Zojer N, Stevenson FK, Sahota SS. Mantle cell lymphoma with t(11;14) and unmutated or mutated VH genes expresses AID and undergoes isotype switch events. Blood 2004; 103: 2795–2798. Wu X, Darce JR, Chang SK, Nowakowski GS, Jelinek DF. Alternative splicing regulates activation-induced cytidine deaminase (AID): implications for suppression of AID mutagenic activity in normal and malignant B cells. Blood 2008; 112: 4675–4682. Palandri F, Iacobucci I, Martinelli G, Amabile M, Poerio A, Testoni N et al. Long-term outcome of complete cytogenetic responders after imatinib 400 mg in late chronic phase, philadelphia-positive chronic myeloid leukemia: the GIMEMA Working Party on CML. J Clin Oncol 2008; 26: 106–111. Esperou H, Boiron JM, Cayuela JM, Blanchet O, Kuentz M, Jouet JP et al. A potential graft-versus-leukemia effect after allogeneic hematopoietic stem cell transplantation for patients with Philadelphia chromosome-positive acute lymphoblastic leukemia: results from the French Bone Marrow Transplantation Society. Bone Marrow Transplant 2003; 31: 909–918. Iacobucci I, Saglio G, Rosti G, Testoni N, Pane F, Amabile M et al. Achieving a major molecular response at the time of a complete cytogenetic response (CCgR) predicts a better duration of CCgR in

22

23 24 25 26

27 28

29 30

31

32

33 34 35 36

imatinib-treated chronic myeloid leukemia patients. Clin Cancer Res 2006; 12: 3037–3042. Iacobucci I, Lonetti A, Messa F, Cilloni D, Arruga F, Ottaviani E et al. Expression of spliced oncogenic Ikaros isoforms in Philadelphia-positive acute lymphoblastic leukemia patients treated with tyrosine kinase inhibitors: implications for a new mechanism of resistance. Blood 2008; 112: 3847–3855. Kaplan E, Meier P. Non parametric estimation from incomplete observation. J Am Stat Assoc 1958; 53: 457–481. Gooley TA, Leisenring W, Crowley J, Storer B. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med 1999; 18: 695–706. Simon R, Lee Y. Nonparametric confidence limits survival probabilities and median survival time. Cancer Treat Res 1982; 66: 37–42. Ito S, Nagaoka H, Shinkura R, Begum N, Muramatsu M, Nakata M et al. Activation-induced cytidine deaminase shuttles between nucleus and cytoplasm like apolipoprotein B mRNA editing catalytic polypeptide 1. Proc Natl Acad Sci USA 2004; 101: 1975–1980. Wu X, Geraldes P, Platt JL, Cascalho M. The double-edged sword of activation-induced cytidine deaminase. J Immunol 2005; 174: 934–941. Marr S, Morales H, Bottaro A, Cooper M, Flajnik M, Robert J. Localization and differential expression of activation-induced cytidine deaminase in the amphibian Xenopus upon antigen stimulation and during early development. J Immunol 2007; 179: 6783–6789. Larijani M, Frieder D, Sonbuchner TM, Bransteitter R, Goodman MF, Bouhassira EE et al. Methylation protects cytidines from AID-mediated deamination. Mol Immunol 2005; 42: 599–604. Vignetti M, Fazi P, Cimino G, Martinelli G, Di Raimondo F, Ferrara F et al. Imatinib plus steroids induces complete remissions and prolonged survival in elderly Philadelphia chromosome-positive patients with acute lymphoblastic leukemia without additional chemotherapy: results of the Gruppo Italiano Malattie Ematologiche dell0 Adulto (GIMEMA) LAL0201-B protocol. Blood 2007; 109: 3676–3678. Foa` R, Vitale A, Meloni G, Guarini A, De Propris S, Elia L et al. Dasatinib as Front-Line Monotherapy for the Induction Treatment of Adult and Elderly Ph+ Acute Lymphoblastic Leukemia (ALL) Patients: Interim Analysis of the GIMEMA Prospective Study LAL1205. Blood 2007; 110: 10a. Foa` R, Vitale A, Guarini A, De Propris MS, Elia L, Cimino G et al. Dasatinib monotherapy as 1st line treatment of Ph acute lymphoblastic leukemia (ALL) patients: update of Gimema LAL1205 study. Haematologica 2008; 93 (s1): 367. Wang CL, Harper RA, Wabl M. Genome-wide somatic hypermutation. Proc Natl Acad Sci USA 2004; 101: 7352–7356. Franco S, Gostissa M, Zha S, Lombard DB, Murphy MM, Zarrin AA et al. H2AX prevents DNA breaks from progressing to chromosome breaks and translocations. Mol Cell 2006; 21: 201–214. Chen KM, Harjes E, Gross PJ, Fahmy A, Lu Y, Shindo K et al. Structure of the DNA deaminase domain of the HIV-1 restriction factor APOBEC3G. Nature 2008; 452: 116–119. Williams RT, Roussel MF, Sherr CJ. Arf gene loss enhances oncogenicity and limits imatinib response in mouse models of Bcr-Abl-induced acute lymphoblastic leukemia. Proc Natl Acad Sci USA 2006; 103: 6688–6693.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia