Abstract
Background/Aim: Mixed phenotype acute leukemia (MPAL) is a rare hematologic malignancy in which the leukemic cells cannot be assigned to any specific lineage. The lack of well-defined, pathogenetically relevant diagnostic criteria makes the clinical handling of MPAL patients challenging. We herein report the genetic findings in bone marrow cells from two pediatric MPAL patients. Patients and Methods: Bone marrow cells were examined using G-banding, array comparative genomic hybridization, RNA sequencing, reverse transcription polymerase chain reaction, Sanger sequencing, and fluorescence in situ hybridization. Results: In the first patient, the genetic analyses revealed structural aberrations of chromosomal bands 8p11, 10p11, 11q21, and 17p11, the chimeras MLLT10::PICALM and PICALM::MLLT10, and imbalances (gains/losses) on chromosomes 2, 4, 8, 13, and 21. A submicroscopic deletion in 21q was also found including the RUNX1 locus. In the second patient, there were structural aberrations of chromosome bands 1p32, 8p11, 12p13, 20p13, and 20q11, the chimeras ETV6::LEXM and NCOA6::ETV6, and imbalances on chromosomes 2, 8, 11, 12, 16, 19, X, and Y. Conclusion: The leukemic cells from both MPAL patients carried chromosome aberrations resulting in fusion genes as well as genomic imbalances resulting in gain and losses of many gene loci. The detected fusion genes probably represent the main leukemogenic events, although the gains and losses are also likely to play a role in leukemogenesis.
- Pediatric mixed phenotype acute leukemias
- fusion gene
- PICALM
- MLLT10
- LEXM
- ETV6
- NCOA6
- PICALM::MLLT10
- MLLT10::PICALM
- ETV6::LEXM
- NCOA6::ETV6
Mixed phenotype acute leukemia (MPAL) refers to an acute leukemia that cannot be definitely assigned to any specific lineage. Based on morphology and flow cytometry, MPALs may be biphenotypic (co-expression of antigens of more than one lineage but in a single blast population) or bilineal (the bone marrow is characterized by the coexistence of two or more different lineages making up the blast population) (1-6). MPAL is rare, accounting for less than 5% of acute leukemia cases giving an incidence of 0.35 cases/1,000,000 person-years (7-11). Mixing genotypic and phenotypic diagnostic criteria, the disease has been divided into three subgroups based on the presence of the rearrangement t(9;22)(q34.1;q11.2)/BCR::ABL1, aberrations of chromosomal band 11q23 leading to rearrangement of the KMT2A gene, and the leukemic cells’ immunophenotype, i.e., their combinations of myeloid, B-, and T-lineage markers (7, 9-11). The majority of the latter subgroup show features of the B/myeloid subtype, which accounts for 59% of MPAL cases, followed by the subtypes T/myeloid (35%), B/T (4%), and trilineage leukemia (2%) (10, 11). MPALs have genetic and epigenetic alterations, which are a mixture of those seen in both acute myeloid and lymphoblastic leukemia (12-15). The phenotypic and genetic heterogeneity of MPAL, the rarity of the disease, and the absence of well-defined, pathogenetically relevant diagnostic criteria, all contribute to the clinical fact that diagnosing and treating MPAL patients is most challenging (2, 3, 6, 11, 16, 17). Using the database of the Surveillance, Epidemiology, and End Results registry (SEER), Shi and Munker (8) found MPAL to have the worst prognosis among acute leukemias in adults. In a study including both children and adult MPAL patients, Matutes et al. (18) showed that median survival of children was longer than that of adults (139 months versus 11 months).
In the present study, we present the genetic alterations of two pediatric MPAL patients and report two novel, presumably leukemogenic, fusion genes.
Patients and Methods
Ethics statement. The study was approved by the regional ethics committee (Regional komité for medisinsk forskningsetikk Sør-Øst, Norge, http://helseforskning.etikkom.no). Written informed consent was obtained from the patients’ parents. The ethics committee’s approval included a review of the consent procedure. All patient information has been de-identified.
Patients. Patient 1 was a seven and a half-year-old girl diagnosed with MPAL type T/myeloid. The patient was admitted due to enlarged lymph nodes and bruising. Lab analysis showed WBC 312×109/l, Hb 8.2 g/dl, and thrombocytes 105×109/l. Flow cytometry showed that 73% of the bone marrow cells were T-lymphoblasts, but additionally there was a population of neoplastic cells with myeloid markers. The girl was treated according to the ALLTogether1 protocol (19) [ALLTogether1–A Treatment study protocol of the ALLTogether Consortium for children and young adults (0-45 years of age) with newly diagnosed acute lymphoblastic leukaemia (ALL)], induction B and intermediate-risk high arm. At evaluation on day 29, the minimal residual disease (MRD) was <0.05%. Further evaluation showed negative MRD and the rest of the treatment was without larger complications.
Patient 2 was a three-year-old boy diagnosed with MPAL. The boy was admitted due to fever and pancytopenia. Lab analysis showed WBC of 36×109/l, Hb 8.6 g/dl, and thrombocytes 75×109/l. Flow cytometry showed that 70% of the bone marrow cells were T-lymphoblasts, but that a large proportion of these cells also expressed myeloid markers. The boy was treated according to the ALLTogether1 protocol (19) [ALLTogether1–A Treatment study protocol of the ALLTogether Consortium for children and young adults (0-45 years of age) with newly diagnosed ALL], induction B and intermediate-risk high arm. At evaluation on day 29, the minimal residual disease (MRD) was <0.4%. Further evaluation showed negative MRD and the rest of the treatment was uneventful.
G-banding and karyotyping. Bone marrow cells obtained at diagnosis were cytogenetically investigated (20, 21). Chromosome preparations were made from metaphase cells of a 24-h culture, G-banded using Leishman’s stain (Sigma-Aldrich, St. Louis, MO, USA), and karyotyped according to the guidelines of the international system for human cytogenomic nomenclature (2020) (22).
DNA and RNA isolation. Genomic DNA and total RNA were extracted from the patients’ bone marrow samples at diagnosis. DNA was extracted using the Maxwell 16 Instrument System and Maxwell 16 Cell DNA Purification Kit (Promega, Madison, WI, USA) and the concentration was measured using a Quantus fluorometer (Promega). Total RNA was extracted using the miRNeasy Mini Kit and QiaCube automated purification system according to the manufacturer’s instructions (Qiagen, Hilden, Germany), and the concentration was measured with the QIAxpert microfluidic UV/VIS spectrophotometer (Qiagen). The Agilent 2100 bioanalyzer and the DV200 index, which evaluates the percentage of RNA that is longer than 200 nucleotides, were used to assess RNA integrity (23).
Array comparative genomic hybridization (aCGH) analysis. aCGH was performed using the CytoSure array products (Oxford Gene Technology, Begbroke, Oxfordshire, UK) following the company’s protocols (24). The reference DNA was Promega’s human genomic female and male DNA for patients 1 and 2, respectively (Promega). The slides (CytoSure Cancer +SNP array) were scanned in an Agilent SureScan Dx microarray scanner using Agilent Feature Extraction Software (version 12.1.1.1). Data were analyzed using the CytoSure Interpret analysis software (version 4.11.36). Annotations are based on human genome build 19.
RNA sequencing. High-throughput paired-end RNA-sequencing was performed at the Genomics Core Facility, Norwegian Radium Hospital, Oslo University Hospital. The software FusionCatcher was used to find fusion transcripts (25).
RT-PCR and Sanger sequencing analyses. cDNA was synthesized from one μg of total RNA using the iScript Advanced cDNA Synthesis Kit for RT-qPCR according to the manufacturer’s instructions (Bio-Rad, Hercules, CA, USA). cDNA corresponding to 20 ng total RNA was used as template in subsequent PCR assays. The quality of the cDNA synthesis was assessed by amplification of a cDNA fragment of the ABL protooncogene 1, non-receptor tyrosine kinase (ABL1) gene using the primer combination ABL1-91F1/ABL1-404R1 (26).
To confirm the presence of fusion transcripts (see below), the BigDye Direct Cycle Sequencing Kit was used for both PCR and cycle (Sanger) sequencing following the manufacturers’ recommendations (ThermoFisher Scientific, Waltham, MA, USA). The primers are listed in Table I. The forward primers had the M13 forward primer sequence TGTAAAACGACGGCCAGT at their 5′-end whereas the reverse primers had the M13 reverse primer sequence CAGGAAACAG CTATGACC at their 5′-end. The primer combinations were MLLT10-458F1/PICALM-2309R1, PICALM-2144F1/MLLT10-569R1, ETV6-541F1/LEXM-1207R1, and NCOA6-3174F1/ETV6-744R1. Sequence analyses were performed on the Applied Biosystems SeqStudio Genetic Analyzer system (ThermoFisher Scientific). The basic local alignment search tool (BLAST) software was used for computer analysis of sequence data (27). The reference sequences were: NM_007166.4 for the phosphatidylinositol binding clathrin assembly protein (PICALM) gene, NM_004641.4 for the MLLT10 histone lysine methyltransferase DOT1L cofactor (MLLT10) gene, NM_001987.5 for the ETS variant transcription factor 6 (ETV6) gene, NM_014071.5 for the nuclear receptor coactivator 6 (NCOA6) gene, and NM_152607.3 for the lymphocyte expansion molecule (LEXM) gene. The BLAT alignment tool and the human genome browser at UCSC were used to map the sequences on the Human GRCh37/hg19 assembly (28, 29).
Designation, sequence (5′->3′), and position in reference sequences of the forward (F1) primers and the reverse (R1) primers used for polymerase chain reaction (PCR) amplification and Sanger sequencing analyses.
Fluorescence in situ hybridization (FISH) analysis. On the basis of RNA sequencing, RT-PCR, and Sanger sequencing findings (see below), additional FISH analyses were performed on metaphase spreads and interphase nuclei of bone marrow cells from both patients using the CytoCell custom PICALM::MLLT10 dual fusion probe for patient 1 and CytoCell custom ETV6 break apart probe for patient 2.
Results
Cytogenetic and aCGH analyses. G-banding analysis of short-term cultured cells from the patients’ leukemic bone marrows revealed chromosome aberrations whose identity could only be partly established (Figure 1). In patient 1, the karyotype was 46,XX, add(8)(p11),add(10)(p11),del(11)(q21),add(17)(p11)[5]/88~109,del(11)(q21),inc[cp4]/46,XX[2] (Figure 1A). In patient 2, the karyotype was 48,XY,+Y,+Y,der(1)t(1;12)(p32;p13),−2,add(8)(p11),add(12)(p13),der(20)t(1;20)(p32;p13)del(20)(q11),+r[7]/49,XY,+Y,+Y,der(1),+8,add(8)x2,add(12),−16,der(20),+mar[5]/46,XY[2] (Figure 1B).
G-banding analysis of the bone marrow cells from the two patients diagnosed with mixed phenotype acute leukemia. A) Karyogram from patient 1 showing the leukemic clone with the karyotype 46,XX,add(8)(p11),add(10)(p11),del(11)(q21),add(17)(p11). B) Karyogram from patient 2 showing the leukemic clone with the karyotype 48,XY,+Y,+Y,der(1)t(1;12)(p32;p13),−2,add(8)(p11),add(12)(p13),der(20)t(1;20)(p32;p13)del(20)(q11),+r. Arrows indicate abnormal chromosomes.
By aCGH analysis, gains and losses of several regions, and obviously affecting many genes, were detected in the leukemic cells of both patients (Table II and Table III). For the leukemia of patient 1, the main gains were of material from chromosome arms 2p and 13q whereas the main losses were from 4q, 8p, and 21q. The latter loss included the RUNX1 locus in 21q22. aCGH also showed the presence of a small clone with losses from on 12q, 17p, and Xq (Table II). In patient 2, gains were found on 2p, 2q, 8p, 8q, 12q, 19q, Xp, Xq, and Y. Losses were seen on 2q, 8p, 11q14, and 16q (Table III).
Results of array comparative genomic hybridization analysis of patient 1 with pediatric mixed phenotype acute leukemia.
Results of array comparative genomic hybridization analysis of patient 2 with pediatric mixed phenotype acute leukemia.
RNA sequencing, RT-PCR, and Sanger sequencing analyses. In patient 1, analysis of the bone marrow RNA sequencing data using the FusionCatcher software detected 11 MLLT10::PICALM and three PICALM::MLLT10 chimeric transcripts (Table IV). RT-PCR together with Sanger sequencing of the PCR products confirmed the presence of the chimeric transcripts in which exon 3 of MLLT10 had fused to exon 20 of PICALM (MLLT10::PICALM), as well as that in which exon 19 of PICALM was fused to exon 4 of MLLT10 (PICALM::MLLT10) (Figure 2A). No other fusion transcripts were examined. In patient 2, the analysis of RNA sequencing data detected two ETV6::LEXM and six NCOA6::ETV6 chimeric transcripts (Table V). RT-PCR together with Sanger sequencing of the PCR products confirmed the presence of chimeric transcripts in which exon 2 of ETV6 fused to exon 9 of LEXM (ETV6::LEXM) and exon 10 of NCOA6 fused to exon 3 of ETV6 (NCOA6::ETV6) (Figure 2B). No other fusion transcripts were examined.
The MLLT10::PICALM and PICALM::MLLT10 fusion transcripts detected in patient 1 after analysis of RNA sequencing data with FusionCatcher.
Sanger sequencing of the bone marrow cells from the two patients diagnosed with mixed phenotype acute leukemia. (A) Partial chromatograms from patient 1 showing the chimeric transcripts in which exon 19 of PICALM fused to exon 4 of MLLT10 (PICALM::MLLT10) and exon 3 of MLLT10 fused to exon 20 of PICALM (MLLT10::PICALM). (B) Partial chromatograms from patient 2 showing the chimeric transcripts in which exon 2 of ETV6 fused to exon 9 of LEXM (ETV6::LEXM), and exon 10 of NCOA6 fused to exon 3 of ETV6 (NCOA6::ETV6).
The ETV6::LEXM and NCOA6::ETV6 fusion transcripts detected in patient 2 after analysis of RNA sequencing data with FusionCatcher.
Fluorescence in situ hybridization (FISH) analyses. FISH analysis of bone marrow cells from patient 1 detected two fusion signals, corresponding to the PICALM::MLLT10 and MLLT10::PICALM fusion genes, in 93 out of 100 examined interphase nuclei. On metaphase spreads, fusion signals were seen on del(11)(q21) and add(10)(p11) (Figure 3A). FISH analysis of bone marrow cells from patient 2 showed a split signal for the ETV6 locus in 187 out of 200 interphase nuclei. FISH analysis of metaphase spreads showed that the red (distal) signal of that probe had moved to the der(1), whereas the proximal (green) signal remained on the add(12) (Figure 3B).
Fluorescence in situ hybridization (FISH) of the bone marrow cells from the two patients diagnosed with mixed phenotype acute leukemia. (A) FISH on a metaphase spread and two interphase nuclei from patient 1 using CytoCell custom PICALM::MLLT10 dual fusion probe showing two green/red fusion signals, corresponding to PICALM::MLLT10 and MLLT10::PICALM fusion, in addition to normal green (PICALM) and red (MLLT10) signals. (B) FISH on a metaphase spread and two interphase nuclei from patient 2 using CytoCell custom ETV6 break apart-probe showing splitting of the probe. On the metaphase spread, the red (distal) signal of the probe hybridized to der(1), the proximal (green) signal of the probe hybridized on the add(12), and the yellow (green/red) signal, corresponding to the intact ETV6 locus, hybridized to the normal chromosome 12. On the interphase nuclei, the results of the hybridization were a red signal, a green signal, and a yellow (green/red) signal.
Discussion
PICALM::MLLT10 and its reciprocal MLLT10::PICALM fusion gene, both of which were found in the first patient with MPAL, are known but rare leukemogenic chimeras generated by the translocation t(10;11)(p12;q14) (30-32). Mitelman’s database of chromosome aberrations and gene fusions in cancer (last updated on April 27, 2023) contains only 28 cases of acute leukemia with a molecularly identified PICALM:: MLLT10, and only 115 out of 35464 (less than 1%) cases of acute leukemia (myeloid, lymphocytic/lymphoblastic, bilineage/biphenotypic, and undifferentiated) with a cytogenetically detected t(10;11)(p13-12;q13-21) (33). Patient 1 had abnormal chromosomes 10 and 11 in a complex karyotype, designated as add(10)(p11) and del(11)(q21) based on G-banding analysis alone. FISH examination of abnormal spreads using a PICALM::MLLT10 dual fusion probe revealed fusion signals on both add(10)(p11) and del(11)(q21). Thus, it seems that these two abnormal chromosomes were part of a cryptic, possibly complex, 10p12;11q14-rearrangement which resulted in the formation of PICALM::MLLT10 as well as the reciprocal MLLT10::PICALM.
Both in adult patients and children with MPAL, PICALM::MLLT10 is known to be associated with a T/myeloid or B/T subtype (12, 13, 31, 32, 34, 35). Patient 1 had the T/myeloid phenotype.
The PICALM gene (also known as CALM) is ubiquitously expressed and codes for a clathrin assembly protein, which is involved in endocytosis and regulates transferrin uptake in erythrocytes (30, 36, 37). MLLT10 encodes a transcription factor, which interacts with many proteins but especially DOT1L (38). The MLLT10 protein has, at its amino-terminus, two PHD finger domains that are collectively referred to as leukemia-associated protein (LAP) finger(s) (38). Disruption of LAP in MLLT10 fusion proteins leads to malignant transformation (39). Towards the carboxyl-terminus, the MLLT10 protein has an evolutionarily conserved octapeptide motif (EQLLERQW) and a partially conserved leucine zipper. These are collectively known as the OM-LZ domain that interacts with several proteins, such as DOT1L, IKAROS, and GAS41 (40-43). MLLT10 fusion proteins were reported as both necessary and sufficient in leukemogenesis (40-43). The PICALM::MLLT10 fusion gene, which codes for a chimeric protein containing the clathrin-binding domain of PICALM and the OM-LZ domain of MLLT10, was found to induce leukemia in mice (39, 42, 44).
aCGH revealed loss of the RUNX1 locus (Table II). RUNX1 deletions or truncations were reported in AML (45-48) as well as in MPAL (15). In a recent study, pediatric AMLs carrying deleted or mutated RUNX1 had poor survival (48).
The ETV6::LEXM and NCOA6::ETV6 fusion genes found in the second pediatric MPAL case are, to the best of our knowledge, novel chimeras. On the cytogenetic resolution level, they were caused by the der(1)t(1;12)(p32;p13) and add(12)(p13) chromosomal aberrations as seen by G-banding supported by relevant FISH analysis of interphase nuclei and metaphase spreads. The red (distal) signal of the commercial ETV6 break-apart FISH probe hybridized to der(1)t(1;12) (p32;p13). The 1p32 band is where the LEXM gene is located. Both ETV6 and LEXM are transcribed from telomere to centromere. Thus, the G-banding and FISH data, the location and transcriptional orientation of the genes all indicate that formation of the ETV6::LEXM chimera took place on the der(1)t(1;12). On the other hand, the proximal (green) moiety of the ETV6 break-apart FISH probe hybridized to add(12), indicating that the NCOA6::ETV6 chimera was generated on that chromosome. NCOA6 maps to band 20q11 and is transcribed from telomere to centromere. Thus, what was by G-banding seen as an add(12) could be a der(12)t (12;20)(p13;q11) chromosome. If so, the abnormalities der(1)t(1;12)(p32;p13), add(12)(p13), and der(20)t(1;20) (p32;p13) del(20)(q11), all detected by G-banding analysis, could be the chromosomal rearrangements behind the generation of the fusion genes ETV6::LEXM and NCOA6::ETV6.
The simultaneous generation of two chimeras in which exons 1 and 2 of the ETV6 gene are involved as the 5′-end fusion partner and exons 3 to 8 as a 3′-end fusion partner, is rare (49, 50). In a patient with acute myeloid leukemia carrying a complex cytogenetic rearrangement involving chromosome bands 5q13, 12p13, 22q11, and 3q12, exons 1 and 2 of ETV6 were fused to an antisense sequence of FCHO2 (reported as LOC115548) as the 5′-end fusion partner generating a 5′-ETV6::FCHO2-3′ chimera on the derivative chromosome 5. Exons 3 to 8 of ETV6 were fused with exon 1 of MN1 as a 3′-end fusion partner generating a 5′-MN1::ETV6-3′ chimera on the derivative chromosome 12 (49). In a pediatric acute lymphoblastic leukemia of prenatal origin, insertion of a segment from chromosomal band 9q34 into the ETV6 locus was described generating 5′-ETV6::ABL1-3′, 5′-AIF1L-ETV6-3′, and 5′-ABL1-AIF1L-3′ chimeras (50).
Rearrangement of the ETV6 locus is well-known (51-55) in both hematological malignancies and solid tumors. This promiscuous gene may function pathogenetically as both a 5′- and 3′-end partner in fusions with a number of loci (51-55). The chimeras ETV6::FGFR2 and ETV6::NCOA2 code for fusion oncoproteins reported in T/myeloid mixed-phenotype leukemia (56, 57). In the ETV6::FGFR2 chimeric transcript, the fusion is between exon 4 of ETV6 (NM_001987) and exon 5 of FGFR2 (NM_022970) (56). In the ETV6::NCOA2, fusion occurs between exon 4 of ETV6 and exon 15 of NCOA2 or exon 5 of ETV6 and exon 14 of NCOA2 (57). Thus, both above-mentioned ETV6 fusions contain the N-terminal ETV6 pointed domain (PNT, also referred to as helix-loop-helix) known to be involved in protein-protein interactions (58).
In the ETV6::LEXM chimera, exon 2 of ETV6 was fused out-of-frame with exon 9 of LEXM (Table V). It might code for a 135 amino acid peptide consisting of the first 54 amino acids of ETV6 and 81 amino acids encoded by LEXM exons that are out-of-frame because of the fusion. Chromosomal rearrangements resulting in out-of-frame ETV6 chimeric transcripts have also previously been reported and their role in neoplasia investigated (59). In all published cases carrying out-of-frame ETV6 chimeric transcripts, the fusion occurred after exon 1 or 2 of ETV6 (59).
In the NCOA6::ETV6 chimera, exon 10 of NCOA6 fused in-frame to exon 3 of ETV6. Based on the NCOA2 reference sequences NM_014071.5/NP_054790.2 and the ETV6 reference sequences NM_001987.5/NP_001978.1, the chimeric NCOA6::ETV6 transcript would code for a 1,369 amino acid peptide consisting of the N-terminal part of NCOA6 (amino acids 1-972) and amino acids 55-452 from the ETV6 protein. The N-terminal part of NCOA6 contains many functional regions, such as an activation domain, a nucleic acid-binding region, an LXXLL motif, the Med15 subunit of Mediator complex, and binding regions for various proteins, such as trimethylguanosine synthase 1 (TGS1 also known as NCOA6IP), CREB binding protein (CREBBP), and TBP/GTF2A (60). The ETV6 protein between amino acids 55 and 452 contains the N-terminal ETV6 pointed domain and the erythroblast transformation specific domain (also known as ETS-DNA binding domain) (51-53, 55). The ETV6 pointed domain is involved in hetero- and homodimerization of ETV6 with other proteins, whereas the ETS-DNA binding domain interacts with an approximately 10-bp-long DNA sequence, which contains a GGAA/T central core domain and is found in a plethora of gene promoters and enhancers (51-53, 55). Thus, the NCOA6::ETV6 is predicted to be a leukemogenic ETV6-chimeric transcription factor.
In conclusion, we used G-banding, aCGH, RNA sequencing, RT-PCR/Sanger sequencing, and FISH methodologies to identify the genetic aberrations of bone marrow cells in two children diagnosed with MPAL. Chromosomal rearrangements resulting in fusion genes and gains and losses of chromosomal areas were identified. The first patient had acquired the chimeras MLLT10::PICALM and PICALM::MLLT10 in the leukemic bone marrow cells, whereas the second carried ETV6::LEXM and NCOA6::ETV6 chimeras, and we assume that these fusion genes are the main leukemogenic drivers. However, the gain and losses of many genes may have played a role in leukemogenesis.
Footnotes
Conflicts of Interest
The Authors declare that they have no potential conflicts of interest in relation to this study.
Authors’ Contributions
IP designed and supervised the research, performed molecular genetic experiments, bioinformatics analysis, and wrote the manuscript. KA performed FISH, G-banding and karyotyping, molecular genetic experiments and interpreted the data. IMRJ made clinical evaluations and treated the patients. MRT evaluated aCGH, FISH, and cytogenetic data. FM evaluated the cytogenetic data. SH assisted with study design and writing of the manuscript. All Authors read and approved of the final manuscript.
- Received October 11, 2023.
- Revision received November 20, 2023.
- Accepted November 21, 2023.
- Copyright © 2024, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).