Introduction

The chromosomal translocation t(1;19)(q23;p13) with the formation of a chimeric TCF3::PBX1 gene (E2A-PBX1 in older nomenclature) is detected in approximately 3–5% of pediatric and adult acute lymphoblastic leukemia (ALL) cases. Despite its relative rarity, the translocation is still the third most frequent recurrent chromosomal translocation in ALL (after t(9;22)/BCR::ABL1 and t(4;11)/KMT2A::AFF1 in adult ALL and after t(12;21)/ETV6::RUNX1 and t(4;11)/KMT2A::AFF1 in pediatric ALL)1. Affected patients exhibit a characteristic B-cell immunophenotype (CD19+/CD10+/CD33−/CD34−/sIg−, and mostly cyIg+), and gene expression analyses have indicated that TCF3::PBX1-positive patients constitute a separate entity among ALL patients2,3,4. The current WHO classification includes TCF3::PBX1-positive ALL as a distinct subgroup of B-lymphoblastic leukemia5. Historically, TCF3::PBX1-positive leukemia has been associated with a poor prognosis, but this has been overcome by modern therapy regimens and TCF3::PBX1 currently defines a group of ALL patients with a good clinical outcome in childhood ALL16,6,7,8,9, although these patients appear to have an increased risk for CNS involvement at diagnosis10. The prognostic impact of TCF3::PBX1 in ALL in older patients (age > 15 years) is less well defined, and relatively few molecular-based and controversial data have been published in this area11,12,13,14,15. TCF3::PBX1-positive ALL has been found to express the receptor tyrosine kinase ROR1, which may serve as a therapeutic target in the future16,17. Some promising therapeutic in vitro effects have been observed with the SRC inhibitor dasatinib18 and the phosphatidylinositide 3-kinase delta (p110δ) inhibitor idelalisib19. However, no established targeted therapy currently exists for TCF3::PBX1-positive patients, and the assessment of measurable residual disease (MRD) remains the most important tool in therapy stratification and prognostication.

Data on the molecular details of the t(1;19)(q23;p13) translocation in adult ALL are scant. The following work analyzed 49 TCF3::PBX1-positive, clinically well-defined adult cases, identified the chromosomal break sites, and characterized the molecular background of this translocation, thus providing a detailed and extensive molecular overview of this translocation. A method for the easy identification of the breakpoint sites is presented, and the potential utilization of these chromosomal breakpoints for detecting measurable residual disease is demonstrated.

Results

Rationale for use and development of a long range-inverse PCR method

One type of chimeric RNA transcript was predominantly found in TCF3::PBX1-positive patients, showing a fusion of TCF3 exon 16 (reference sequence NG_029953.2) and PBX1 exon 3 (reference sequence NG_028246.2)26. Other transcripts have been described, but they seem to be very rare27. Chromosomal breaks can thus be assumed to occur in the intron 3′ of TCF3 exon 16 (“intron 16”) and in the intron 5′ of PBX1 exon 3 (“intron 2”). The TCF3 reference sequence includes a 3289 bp intron 16 (ncl 1615822–1619110, NC_000019.10, GRCh38.p13 primary assembly). This intron is present in all 41 TCF3 variants listed in the NCBI gene database (updated on 1-Aug-2020). The location of the breakpoint site on chromosome 1 is less clear (8 PBX1 transcript variants with either a 229,182 bp intron 2 (ncl 164563312–164792493) or a 166,397 bp intron 2 (ncl 164626097–164792493). Since the breakpoint region on chromosome 19 appeared to be relatively localized, a long range-inverse PCR (LRI PCR) approach was chosen for the analysis. Commercially available restriction enzymes were screened for those with restriction sites flanking the putative breakpoint region on chromosome 19. Three enzymes were suitable because they had palindromic cutting sites without degenerate nucleotides, produced sticky ends and were frequent cutters: SphI, BamHI and TaqI (Fig. 1A). BamHI had one cutting site 148 bp 5′ of the TCF3 intron end; thus, breakpoints near the intron end could not be detected using this enzyme. The three enzymes provided dense coverage of PBX1 intron 2 with restriction sites (Table S1 and Figure S1).

Figure 1
figure 1

(A) Schematic depiction of the breakpoint region in TCF3 with restriction sites, (B) Examples of long-range inverse PCR products (lanes 2–4: TaqI, lanes 5–12 SphI, lanes 13–14 BamHI, lanes 1 and 15: 1 kb Hyperladder. (C) Examples of long-range multiplex PCR results (with sample numbers).

Full size image

Various PCR primers and PCR conditions were tested for the development of the inverse PCR method. The efficacy of inverse PCR could be optimized because the three enzymes produced a detectable “control” PCR product when testing normal DNA. The final PCR primer locations are depicted in Fig. 1A.

Analysis of patient samples and chromosomal breakage data

Patient samples were first analyzed with SphI, followed by BamHI and TaqI. Twenty-six TCF3::PBX1 breakpoints were identified with SphI, 12 with BamHI and 11 with TaqI. PCR examples are shown in Fig. 1B. Fifty four TCF3::PBX1-RT-PCR-positive samples were analyzed and the genomic TCF3::PBX1 fusion site was identified in 49 cases. The fact that five samples remained uncharacterized on the genomic level may reflect a limitation of the method. Insufficient DNA quality may also have played a role since some samples were more than 15 years old. Of these 49 TCF3::PBX1 sites, the reciprocal PBX1::TCF3 breakpoint was identified in 15 cases. One sample (5741) harbored an inversion with two breaks in PBX1. The break sites displayed a remarkable nonrandom pattern (Fig. 2, Table 1). Fifty-three (83%; 41 TCF3::PBX1, 12 PBX1::TCF3) were located in a narrow 40 bp region in TCF3 intron 16, while 11 (17%; 8 TCF3::PBX1, 3 PBX1::TCF3) occurred outside this region. In PBX1 intron 2 the clustering was less dense but still apparent. Two major break clusters could be delineated in PBX1: one from ~ 220 to 229 kb (near the intron end), and a second from ~ 120 to ~ 155 kb. These two clusters comprised 34 kb (14.8% of the intron) and 88% of all break events (Fig. 2, Table 1). Twenty-four breaks (37%; 16 TCF3::PBX1, 8 PBX1::TCF3) were located in the first cluster, 33 (51%; 27 TCF3::PBX1, 6 PBX1::TCF3) in the second cluster, and 8 (12%; 7 TCF3::PBX1, 1 PBX1::TCF3) outside both clusters.

Figure 2
figure 2

(A) Breakpoint distribution and location of DNA repeats/cRSS sites in TCF3 intron 16, (B) breakpoint distribution in PBX1 intron 2.

Full size image
Table 1 Basic characteristics of patient samples and breakpoint locations.
Full size table

The break sites were analyzed bioinformatically for patterns that could explain the observed distribution. There were no apparent DNA microhomologies at the break sites in PBX1 (Table S2). TCF3 intron 16 included 6 DNA repeats comprising 1699 bp (51.6% of the intron), and PBX1 intron 2 included 342 DNA repeats with 51,699 bp (22.6%, Tables S4, S5). The break cluster in TCF3 intron 16 was located at the 5′ end of a repetitive DNA element (MER20). Eight of the 10 break sites in TCF3 outside the cluster were located in or immediately 3′ of other DNA repetitive elements (L2A, MER20, L1MS, Alujb, AluY; Fig. 2A, Tables 1, S6). For the breaks in PBX1, there was no apparent association with repetitive DNA elements. Eight of the 55 breaks in PBX1 (14.5%) occurred inside repetitive elements (# 3120, 3951, 5741, 5974, 6840, 7236, ML4316, ML11220; Tables 1, S4). Bioinformatical analysis revealed 611 potential cryptic recombination signal sequences (cRSSs) in PBX1 intron 2, covering 20.1 kb (8.8%) of the intron. Five of the 65 breaks in PBX1 (7.7%) occurred in or in the vicinity (± 30 bp) of cRSSs (# 3120, 4641, 7281, ML11220, ML11543; Tables 1, S7).

To complement this analysis, all samples were also investigated for intragenic IKZF1 deletions by PCR. These deletions are found in approximately 20% of B-cell precursor ALL cases and are known to be caused by aberrant VDJ recombinase activity. None of the 49 samples showed an intragenic IKZF1 deletion. This does not exclude a possible role of RAG-mediated secondary aberrations in TCF3::PBX1-rearranged ALL as illustrated by the example of ETV6::RUNX1-positive pediatric ALL286,29.

Chromosomal translocations are occasionally associated with DNA secondary structures, such as inverted repeats with hairpin loops30, and thus, the hotspot region of TCF3 was analyzed with RNAfold. The main break site was located in an open loop that was flanked by regions with relatively strong base pair binding (Fig. S8). The analysis of the 15 cases in which reciprocal PBX1::TCF3 were characterized showed mostly no microhomologies at the break sites, with frequent insertion of nontemplate nucleotides, suggesting a nonhomologous end-joining repair (NHEJ) mechanism31. One sample (4297) showed an insertion from the FGF6 gene on chromosome 12, a gene not previously implicated in the pathogenesis of ALL (Fig. S3).

Development and optimization of a real-time qPCR method

The clustering of chromosomal breaks in a narrow region in TCF3 intron 16 suggested a quantitative PCR method with a common forward primer, a common dual-labeled hybridization probe 5′ of the breakpoint cluster region and a patient-specific reverse primer 3′ of the breakpoint. Several forward primers and dual-labeled probes were first tested on patient samples and control DNA to exclude spurious amplifications. Finally, one combination of a common forward primer and a common dual-labelled probe was selected that was tested on 15 randomly chosen patient samples (Table 2, Fig. 3). In all 15 cases, it was possible to design a reverse PCR primer that yielded data with good sensitivity and specificity. The testing of further samples was not possible because of shortage of sample material. This generic real-time qPCR was designed to quantify breakpoints in the TCF3 hotspot region (~ 80%). Breakpoints outside this region (and likewise the PBX1::TCF3 breakpoints) could theoretically also be used as MRD targets, but in these cases, no generic recipe can be given, and individual patient-specific qPCRs would have to be constructed.

Table 2 Real-time quantitative PCR parameters.
Full size table
Figure 3
figure 3

Examples of real time quantitative PCRs: (A) sample 4316, (B) sample 6255, (C) sample 7601, (D) sample 7979, (E) sample ML11543.

Full size image

Development and optimization of two multiplex long range PCRs

Since breakpoint identification by long range-inverse PCR is an elaborate procedure and since the breakpoints showed clustering in certain regions, efforts were made to simplify the detection procedure. Two multiplex long-range PCRs with a series of PCR oligonucleotides covering the entire breakpoint regions were developed and optimized that allowed the detection of breakpoints in the two breakpoint cluster regions of PBX1. Examples of these multiplex long-range PCRs are shown in Fig. 1C.

Discussion

The translocation t(1;19)(q23;p13) has been described as mostly unbalanced156,32,33,34. This is in accordance with the observation made in this study that in only 14 cases (33%) a reciprocal break site could be characterized.

Clustering of breakpoints

Since the first description of the translocation t(1;19) as a recurrent aberration in ALL in 1984, research has largely focused on cytogenetic aspects of this aberration, and few investigations have been carried out in adult ALL276,346,35. Wiemels et al.36 first systematically investigated the translocation at a molecular level and described 24 cases from various pediatric ALL studies. The median age of the patients was 6.8 years, with only one patient being an adult > 18 years of age. A similar clustering of breakpoints was observed, and the authors speculated that aberrant VDJ recombinase activity might be involved. They identified a reciprocal breakpoint in 5 (21%) cases36.

In this study, no association of t(1;19) chromosomal breaks with repetitive DNA elements was found. While the location of the break cluster in TCF3 intron 16 close to a MER20 element could be coincidental, there was no similar association of the breaks mapping to PBX1 intron 2. Similarly, no direct association with cryptic RSS was observed. None of the 49 patient samples showed an intragenic IKZF1 deletion—an aberration caused by illegitimate VDJ recombination-mediated deletion, and present in approximately 20% of BCR::ABL1-negative B precursor ALLs. Recently, Liu et al. analyzed the TCF3 “fragile zone” and suggested that the initial TCF3 breakage may arise at a CpG site. They found a statistically significant proximity of the activation-induced cytidine deaminase (AID) hotspot motifs WRC and WGCW near the TCF3 breakpoints W = A or T, R = A or G) suggesting AID involvement in the break process 37. This is consistent with the fact that TCF3::PBX1 is predominantly detected in pre-B ALL, which is immunophenotypically the most “mature” entity in B precursor ALL, indicating a relatively late stage of B-cell development.

Real-time qPCR for measurable residual disease detection

Measurable residual disease in ALL is usually assessed by the use of clonally rearranged immunoglobulin (IG) and/or T-cell receptor (TCR) loci for the construction of real-time quantitative PCRs (qPCRs)38. The main advantage of this approach is its universal applicability. Theoretically, it can be applied in any malignant disease of lymphatic origin. However, this method also has some disadvantages. In a significant minority of cases, it is not possible to identify clonal rearrangements, and with the introduction of next generation sequencing techniques it has become apparent that IG/TCR rearrangements are often in fact polyclonal at diagnosis39. IG/TCR-based MRD monitoring is thus often based on only one of several clones, and such an analysis may miss the decisive clone. IG/TCR-based qPCRs frequently show a suboptimal sensitivity (below 10–4), because of the difficulty of constructing a specific PCR against a highly homologous background. In addition, the IG/TCR rearrangements are potentially unstable, and further rearrangements can occur without loss of the malignant cell phenotype, leading to false negative results.

In those cases where chromosomal translocations lead to the expression of a chimeric mRNA transcript, MRD monitoring can also be performed by the relative quantification of this transcript40. However, this approach has been widely discarded in ALL (with the exception of BCR::ABL1), because it only allows a quantification relative to a “housekeeping gene”, assumed to be stably expressed. “Dormant” tumor stem cells with low expression of the oncogene may escape detection by RT-PCR. This is exemplarily illustrated by the observation that in BCR::ABL1-positive ALL, only a limited correlation between BCR::ABL1-mRNA-based and IG/TCR-based MRD levels is found41. Additionally, RNA is relatively unstable and significantly more difficult to handle than DNA.

An alternative approach is targeting the breakpoint sites of chromosomal translocations to detect and monitor MRD by constructing patient-specific qPCR assays. These are stable molecular markers that cannot be lost in the course of disease because they are linked to molecular drivers of the disease. This approach has been exploited in various entities, such as ALL with t(12;21)/ETV6::RUNX1426,43, ALL with 11q23/KMT2A aberrations446,45, ALL or CML with t(9;22)/BCR::ABL1416,46,47 and other hematopoietic malignancies48,49,50. In most of these cases, it could be shown that the break site-specific PCRs were at least as reliable as the IG/TCR-based methods and yielded a superior sensitivity. The main disadvantages of this approach are the technical difficulties posed by the individual characterization of break sites which may in some cases be dispersed over hundreds of kilobases of genomic DNA, precluding this approach for routine clinical studies with the exception of KMT2A-rearranged ALL, where relatively standardized techniques for break site identification are in use51. With the increasing availability of next-generation sequencing techniques and their technical advances (e.g., nanopore sequencing or mate-pair sequencing) and better knowledge of the molecular background these difficulties are likely to be overcome in the future and MRD detection methods based on chromosomal breakpoints will become increasingly important526,53.

Conclusions

The present work characterizes the t(1;19) chromosomal breakpoints of a large number of adult ALL patients from a well-defined study population and is the largest and the first major investigation on this topic in adult ALL. The results provide a representative and relatively unbiased overview of the molecular details of this aberration. Based on the experimental results, a simplified method for the rapid identification of chromosomal breakpoints is proposed and the usefulness of these chromosomal breakpoint data for measurable residual disease detection is demonstrated. While the theoretical advantages of such an MRD approach appear obvious, clinical studies are necessary to validate the TCF3::PBX1 breakpoint fusion as MRD marker in a clinical context. Further testing and comparisons will have to be performed to fully establish TCF3::PBX1 breakpoints as valuable MRD targets.

Methods

Patient samples and ethics statement

Patient samples were collected from residual diagnostic material obtained between 2001 and 2021 in the context of the German Multicenter ALL Therapy Studies (clinicaltrials.gov identifiers: 00199056 and 00198991). Patients gave written informed consent to scientific investigations on study inclusion and the studies were approved by local and central ethics committees, among them an ethics board of the Goethe University, Frankfurt/Main, Germany and the ethics board of the Charité Universitätsmedizin, Berlin, Germany. Our study complied with the principles set forth in the World Medical Association Declaration of Helsinki.

Patient characteristics

Patient clinical details are summarized in Table 1. All patient samples included in this study (31 bone marrow, 17 peripheral blood, one unspecified) had been investigated by flow cytometry and RT-PCR at diagnosis. All patients exhibited a B precursor immunophenotype (CD19+/CD10+/CD33−/CD34−/sIg−). Forty-three (88%) showed a cyIg+ (pre B) and six a cyIg− (common) immunophenotype. All samples were tested negative for BCR::ABL1 and positive for TCF3::PBX1 by RT-PCR. Twenty-seven (55%) of the 49 patients were female and 22 male. The median age was 39.5 years (range 17–77 years).

DNA isolation

DNA was isolated from archived or fresh samples using either the Gentra PureGene method (QIAGEN, Hilden, Germany), the AllPrep DNA/RNA Kit (QIAGEN) or in a few cases the DNA preparation from TRIzol (ThermoFisher Scientific, Darmstadt, Germany) with subsequent DNA purification.

Long range-inverse PCR (LRI PCR)

The LRI PCR methods were developed and optimized for this study. The following restriction enzymes were used: SphI (GCATG|C), BamHI (G|GACC) and TaqI (T|CGA). FastDigest enzymes were used according to the manufacturer’s recommendations (ThermoFisher Scientific, Darmstadt, Germany). The conditions for the long range-inverse PCR were partially adopted from previous work20. Five hundred nanograms of genomic DNA was digested in a 50 µl volume, the reaction mix was inactivated, purified using the MaXtract High Density kit (QIAGEN, Hilden, Germany), ethanol-precipitated and dissolved in a final volume of 30 µl. The entire volume was used in the ligation procedure (50 µl final volume, 5 U T4 ligase, 16 °C overnight). After purification and ethanol precipitation as described above, the ligation mix was dissolved in 30 µl H2O. Five microliters was used in the long-range PCR with the Expand Long Template PCR System kit (Roche, Mannheim, Germany) with buffer 2 and the following cycler program: 95 °C 2 min, 15 cycles (94 °C 30 s, 65 °C 30 s, 68 °C 6 min), 20 cycles (94 °C 30 s, 65 °C 30 s, 68 °C 5 min with 20 s increment/cycle), and 68 °C 10 min, 4 °C. One enzyme-specific reverse (R) PCR primer was combined with a forward (F) primer. The following primer combinations were used: SphI: TCF3-F2/TCF3-R5, BamHI: TCF3-F2/TCF3-R4, TaqI: TCF3-F2/TaqI-R. If a PCR product was visible the PCR was repeated and primer TCF3-F2 replaced by primers TCF3-F7 or TCF3-F6 to try to generate a smaller PCR product for easier sequencing. PCR products of interest were excised from the agarose gel, purified, and analyzed by Sanger sequencing.

Oligonucleotide sequences for the long range-inverse PCR

All oligonucleotides were obtained from tib molbiol (Berlin, Germany). The LRI PCR primer sequences were (5′-3′): TCF3-R4 GAAGGCCTGGGCTACGGAGGGGAACAGCT, TCF3-F2 CTCCCTGACCTGTCTCGGCCTCCCGACT, TCF3-F6 ACCTTGATTCTATCACTCCTAGGCCAGGGCA, TCF3-R5 CACAGGCCTCCATTCATGTCCCTTCCGCA, TaqI-R AGGCCGTGGAGACCCCCGTCGTAGCT. Normal DNA (without t(1;19) translocation) generated “control bands” of the following sizes: TCF3-F2/TCF3-R4 3308 bp, TCF3-F6/TCF3-R4 2131 bp, TCF3-F2/TCF3-R5 5026 bp, TCF3-F6/TCF3-R5 3849 bp, TCF3-F2/TaqI-R 7567 bp, TCF3-F6/TaqI-R 6390 bp.

Sanger sequencing

Apart from the oligonucleotides detailed above several ad hoc designed oligonucleotides were used for Sanger sequencing of individual samples. Technical Sanger sequencing of PCR products was performed by Microsync SeqLab (Göttingen, Germany). Analysis of chromatograms and sequence data assembly was performed at the Charité laboratory in Berlin.

Sequence data

All nucleotide sequence data (91 328 bp) were submitted to the GenBank/ENA/DDBJ database and are available under the accession numbers OK334233-OK334288, ON383218-ON383224, ON809522.

Real-time quantitative PCR

Real-time qPCR was performed on a RotorGene RG-3000 cycler (formerly Corbett Research, now subsidiary of QIAGEN, Hilden, Germany) using the ABgene PCR QPCR Mix (Thermo Fisher Scientific, Darmstadt, Germany). The forward primer TCF3-qF 5′-CAGGCAGACTTTCCAAGTACCTT-3′ was used with the dual-labelled probe TCF3-FAM 5′-6FAM-CTATCACTCCTAGGCCAGGGCATCT-BHQ1-3′ and a patient-specific reverse primer.

Multiplex long range PCR

The two multiplex long range PCRs comprised the following oligonucleotides (100 nM of each oligonucleotide per reaction mix). For breakpoint cluster 1 (5′-3′): TCF3-F7 AGGAGGGTTTCAGGCAGAGGGCGCA, PBX-long1 CCCGGGGTTGTGCTTCCTCCACCCTT, PBX-long2 TGCGCTCTCTCCCTCCCCCTCATCTCT, PBX-long3 ACGTGGTCCTGCGAGGAGCTCTTAGA, PBX-long4 TGCCCATGCAGCAGGTGACAAGGG, and for breakpoint cluster 2 (5′-3′): TCF3-F7, PBX1-long5 ACGAATCAGGCAGCTGTACAGAAAGCA, PBX1-long6 TCGGCCTCACCTAACTGACTTGCAGGT, PBX1-long7 AGCACCATCCTGAAGTTGCTCGGCT, PBX1-long8 TGCGGGAGGCTGGCAACATTGAGTC, PBX1-long9 ACACAGGTGCTACCTCTGCTCTGCCA, PBX1-long10 TCCAGCTACCTCATGGCTCGCTAGA. PCR conditions were the same as for the LRI PCR.

PCR for IKZF1 deletions

The main four intragenic IKZF1 deletion variants Δ2–7, Δ2–8, Δ4–7, Δ4–8 were investigated by four different PCRs, and the intragenic IKZF1 deletion variant Δ2–3 by one single RT-PCR as outlined previously21.

Bioinformatics and software

Genomic repeats were analyzed with RepeatMasker version 4.0.9, RSSSite and the Tandem repeats finder22,23,24. DNA secondary structures were investigated using RNAfold 2.4.1825. Sanger sequence chromatograms were analyzed with 4Peaks (Nucleobytes, Aalsmeer, The Netherlands) and Nucleotide BLAST (blastn) against the GRCh38.13 reference primary assembly human genome.