Divergent leukaemia subclones as cellular models for testing vulnerabilities associated with gains in chromosomes 7, 8 or 18

Haematopoietic malignancies are frequently characterized by karyotypic abnormalities. The development of targeted drugs has been pioneered with compounds against gene products of fusion genes caused by chromosomal translocations. While polysomies are equally frequent as translocations, for many of them we are lacking therapeutic approaches aimed at synthetic lethality. Here, we report two new cell lines, named MBU-7 and MBU-8, that differ in complete trisomy of chromosome18, a partial trisomy of chromosome 7 and a tetrasomy of the p-arm of chromosome 8, but otherwise share the same mutational pattern and complex karyotype. Both cell lines are divergent clones of U-937 cells and have the morphology and immunoprofile of monocytic cells. The distinct karyotypic differences between MBU-7 and MBU-8 are associated with a difference in the specific response to nucleoside analogues. Taken together, we propose the MBU-7 and MBU-8 cell lines described here as suitable in vitro models for screening and testing vulnerabilities that are associated with the disease-relevant polysomies of chromosome 7, 8 and 18.

www.nature.com/scientificreports/ in myelodysplastic syndrome that is a selection criteria for lenalidomide treatment with a high rate of favourable outcomes 8 . Another example is monosomy 7 in AML, which is associated with a poor prognosis and thus requires aggressive chemotherapy and hematopoietic stem cell transplantation 9,10 . A better understanding of how cytogenetic changes relate to drug responses is needed to improve patient stratification and greater therapy success. While clinical trials routinely assess cytogenetic profiles using conventional methods, in vitro models are needed to functionally test associations between cytogenetic alterations and responses to tool compounds and drugs in the pre-clinical phase.
Matched patient-derived cell lines representing different states of aggressiveness of a disease are powerful tools. This includes the B16 melanoma cell line model 11 and engineered cell lines such as the MBU series for HRAS-driven squamous cell carcinoma 12 . Furthermore, many leukaemia-and lymphoma-derived cell lines have been established harbouring various different cytogenetic abnormalities or particular gene mutations 13 . Genetic drift is often seen in cell lines during long-term cultures. Indeed, karyotypic changes in leukaemia and lymphoma cell lines can be seen as a result of increasing passage numbers 14 . To some extent, these changes reflect the same events of clonal evolution occurring in patients during disease progression 15 . Characterizing the functional effects of these chromosomal changes would help identify vulnerabilities associated with specific acquired genetic abnormalities. This has been exemplified in a recent study performed in 27 strains of the frequently used breast cancer cell line MCF7, which display a multitude of genetic variations and considerably different drug responses 16 . And thus, well-characterized cell lines enable screening for differences in drug sensitivity associated with clinically relevant genetic alterations.
Here, we report the characterization of two divergent leukaemia sub-clones derived from U-937 cells. The two cell lines differed in partial and complete trisomies of chromosomes 7 and 18, respectively, as well as tetrasomy 8p. These cytogenetic differences were associated with a differential drug response suggesting that this pair of cell lines can serve as a model to specifically probe for vulnerabilities associated with the clinically relevant polysomies of chromosomes 7, 8 and 18.

Results
Isolation of divergent cell clones that differ in trisomies 7, 8 and 18. While growing leukaemia cells in suspension over periods of weeks, we noticed a change in the behaviour of the bulk population that was due to the presence and expansion of cells with unexpected characteristics. In order to characterize these cells, we first isolated and expanded single cell clones. We began to characterize two cell lines derived from the expansion of these single cell clones by conventional karyotyping (Fig. 1a). Both cell lines displayed a complex and overall highly similar karyotype. Shared alterations included der(1)t(1;5)(p22;q31.1), der(3)t(1;3)(q21;q26), der(5)t(1;5)(p22;q23.3), add(6)(p25), der(10)t(10;11)(p12;q14)/PICALM-MLLT10, del(11)(q22), der(13)t(1;13) (p32;p11.2), del(22)(q13.2q13.3) and unspecific rearrangements involving chromosomes 2q37, 3q12, 6p25, 12p12, and 16p11.1, as well as complete trisomies of chromosomes 20 and 21 (Table 1). Clear differences were the presence of trisomy 18 and a gain in material of chromosome 8 in one of the cell lines, which for this reason we termed MBU-8. We termed the other cell line MBU-7 as it had a complete trisomy 7, while in MBU-8 part of the q-arm of a third chromosome 7 was lacking. These major differences in the amount of genetic material from chromosomes 7, 8 and 18 has further been confirmed using CytoScan 750 K array-based technology (Fig. 1b  and Supplementary Table S1). Interestingly, the array-based method indicated that the additional fragment of chromosome 8 contained two fused p-arms making it a partial tetrasomy 8. The array further provided additional resolution and helped to improve the karyotype formula, thus substantiating that these cell lines possess an overall complex karyotype (Supplementary Fig. S1 and Supplementary Data S1).
Fluorescence in situ hybridization (FISH) confirmed these differences. While both cell lines showed three copies of the centromeric region of chromosome 7 and the locus 7q22, the locus 7q36 containing the EZH2 gene was only present twice in MBU-8 (Fig. 1c). In line with two additional copies of 8p fused at the peri-centromeric region, we detected a third signal for the centromeric region of chromosome 8 in MBU-8 but only two copies of the 8q24 portion of chromosome 8 containing the MYC gene locus (Fig. 1d). Finally, we confirmed the trisomy 18 in MBU-8 using a FISH probe targeting the locus of BCL2 on 18q21 (Fig. 1e). To examine the cytogenetic stability of MBU-7 and MBU-8 over an extended period of time, we repeated the FISH analysis after 18-20 passages and could not observe any significant differences in the chromosomes assessed (Fig. 1f). This analysis showed that both cell lines are sufficiently stable during a time window that makes them suitable for comparative experimental studies.
In summary, we isolated two cell lines with highly complex karyotypes that, at first glance, did not resemble any other previously described cell line. Importantly, the two cell lines differed in the number of copies of chromosomes 7, 8 and 18. MBU-7 and MBU-8 are divergent subclones of U-937. As an alternative approach to identify the origin of these new cell lines, we analysed a panel of short tandem repeats (STR). Profiling of STRs is a widely used forensic technique to identify human individuals by unique signatures based on the allelic number of 16 highly conserved microsatellite sequences 17 . This method has found an important application in the identification of human-derived cell lines whose profiles are collected and stored in the publicly available Cellosaurus database from the Swiss Institute of Bioinformatics 18 . The profiles of MBU-7 and MBU-8 almost perfectly matched the U-937 cell line (Table 2). We confirmed the almost perfect match by repeating the STR analysis with U-937 cells and by visually comparing the profiles. For instance, the short tandem repeat D13S317 was present in 10 and 12 copies in MBU-7, MBU-8 and U-937 (Fig. 2a). Similarly, the repeat TH01 showed 6 and 9.3 copies in all three cell lines (Fig. 2a).  www.nature.com/scientificreports/ U-937 is a human haematopoietic monoblastoid cell line originally established from a 37-year old male patient with diffuse histiocytic lymphoma 19 , which is a rare, very aggressive type of non-Hodgkin's lymphoma. Besides their popular use as a differentiation model, U-937 cells are well-studied for the translocation t(10;11)(p12;q14) resulting in a PICALM-MLLT10 fusion gene 20 . Gene fusions are a common characteristic of both myeloid and lymphoid leukemias and occur due to chromosomal translocations 21 . The resulting fusion genes frequently act as disease-driving oncogenes and have been investigated as part of diagnostic protocols across different leukemias due to their disease subtype specificity 22,23 . As shown in Fig. 2b, we could detect PICALM-MLLT10, expression in MBU-7, MBU-8 and U-937, but not in the secondary acute myeloid leukaemia (AML) cell line SKK-1 that we included as a control. Conversely, SKK-1, but not MBU-7, MBU-8 or U-937 cells expressed the ETV6-NTRK3 gene fusion (Fig. 2c) that is caused by a t(12;15)(p13;q25) translocation 24 .
To define the mutation profile of MBU-7 and MBU-8, we sequenced a previously described panel of 32 genes related to myeloid disorders 25,26 . Three pathogenic variants of WT1, TP53 and PTPN11 were observed in both subclones with similar variant allele frequency (Table 3). U-937 have been reported to express mutant p53 27 , do not express detectable levels of WT1 28 and harbour a PTPN11 mutation (G20R) 29 .
Taken together, these results suggest that MBU-7 and MBU-8 are subclones of U-937 cells. Several groups have reported different karyotypes for U-937, but reports are consistent in that U-937 cells are close to triploid 30,31 . Comparing these karyotypes with the karyotypes of MBU-7 and MBU-8 (Table 1) suggests that the two cell lines have diverged by acquiring additional cytogenetic alterations, such as multiple losses of genetic material. This includes the loss of the Y chromosome, a striking mosaicism given the male origin of the cell line. Overall, the loss of genetic material was more pronounced in MBU-7 cells than in MBU-8.   19 . In both MBU-7 and MBU-8, we observed a similar cell morphology resembling monocytic cells (Fig. 3a). Within both populations, cells varied in size and maintained the relationship in size between nucleus and cytoplasma. The nuclei had an irregular contour and lax chromatin with the presence of one or more evident nucleoli. Some cells displayed nuclear herniations, which are abnormal protrusions of the nucleus. The extended cytoplasm was basophilic and, in a subset of cells, presented granulation and vacuoles. Interestingly, in addition a majority of cells with the above described characteristics, both cell lines contained some particularly large cells that were multinucleated and had hyperbasophilic cytoplasm (Fig. 3a). These cells were more frequently seen in MBU-7 than MBU-8, which could be confirmed by measuring the forward and side scatter (FSC and SSC) in flow cytometry (Fig. 3b).  www.nature.com/scientificreports/ were negative for myeloperoxidase, which is a marker for granulopoietic cells (data not shown). In terms of markers for the monocytic-macrophage lineage, both cell lines were positive for alpha-naphthyl butyrate esterase and negative for alpha-naphthyl acetate esterase (data not shown). We next analysed the immunophenotypes and found that MBU-7 and MBU-8 have very similar surface marker expression (Fig. 3c). Specifically, they were both positive for the hematopoietic markers CD45, CD99, CD38 and CD44, as well as for the myeloid-monocytic markers CD11b, CD13 and CD15. In terms of stem cell markers, MBU-7 and MBU-8 were largely negative for CD34, HLA-DR, CD117 and CD123. Furthermore, virtually all cells expressed monocytic differentiation markers such as CD64, CD4, CD33 and CD56.
In summary, the immunoprofile and morphology of MBU-7 and MBU-8 cells associates them with the monocytic lineage. This is in line with them being derivatives of U-937 cells that, despite having originated from

MBU-7 and MBU-8 differ in drug response. Finally, we decided to test the response of MBU-7 and
MBU-8 cell lines to a panel of commonly used drugs. Specifically, we have chosen the nucleoside analogues azacitidine, decitabine, and cytarabine. Cytarabine is a DNA toxin and used as a standard chemotherapy drug in AML, acute lymphoid leukaemia (ALL) and non-Hodgkin lymphoma 33 . Azacitidine and decitabine can further act as DNA hypomethylating agents and are the treatment of choice for intermediate-to high-risk MDS patients not eligible for allogenic bone marrow transplantation 34 and for elderly unfit AML patients 35 . We also included venetoclax, an inhibitor of the apoptosis regulator Bcl2 36 , which is currently being approved for the treatment of an increasing number of haematopoietic malignancies (https:// www. fda. gov/ drugs/ resou rces-infor mation-appro ved-drugs/ fda-appro ves-venet oclax-cll-and-sll). Specifically, we have determined the cell viability of MBU-7 and MBU-8 after four days of treatment. MBU-7 cells were significantly more sensitive towards the treatment with all three nucleoside analogues than MBU-8 (Fig. 4a). The biggest difference in response was observed with 0.125 μM decitabine and 0.25 μM cytarabine. Both cell lines were less sensitive to azacitidine but a pronounced difference was observed around 10 μM. In contrast, both MBU-7 and MBU-8 responded in a similar manner to venetoclax suggesting that we are not observing a general propensity of MBU-7 cells to druginduced apoptosis (Fig. 4b).
Taken together, the cytogenetic differences of MBU-7 and MBU-8 are associated with a differential response to a subset of drugs commonly used for the treatment of haematological diseases.

Discussion
In our study, we characterized the two cell lines MBU-7 and MBU-8 that we isolated as clones from a suspension culture. The profile of small tandem repeats identified both cell lines as likely derivatives of U-937 cells. This was consistent with a shared mutational profile and a morphology and immunoprofile of monocytic cells. Although the karyotypes of MBU-7 and MBU-8 were different to U-937, they contained some characteristic features shared with U-937 cells such as the translocation t(10;11)(p12;q14) resulting in a PICALM-MLLT10 fusion gene 20 . Chromosomal instability and loss of heterozygosity has been previously attributed to U-937 37 . Indeed, subclones of U-937 with differences in both small tandem repeats and chromosomal aberrations have been identified following long periods in culture 38,39 . This can explain the divergence in karyotypes of MBU-7 and MBU-8 when compared to U-937. More specifically, U-937 cells have previously been described to have a near triploid karyotype 31 , while MBU-7 and MBU-8 only share trisomies of chromosomes 3, 20 and 21 but differ in complete or partial polysomies of 7, 8 and 18. U-937 cells are derived from a male donor 19 . However, Y chromosome is known to be frequently lost during cell culture 38 . Occasionally this leaves U-937 cells with two X-chromosomes 31 , which we also observed in MBU-7 and MBU-8. Taken together, our results suggest that MBU-7 and MBU-8 are clonal cell lines with divergent karyotypes that spontaneously derived over time from U-937 cells.  www.nature.com/scientificreports/ MBU-7 and MBU-8 have highly similar characteristics but differ in trisomies 7 and 18 and tetrasomy 8p in the context of a shared complex karyotype. Gains of material of chromosome 8 are frequent in myeloid malignancies. Trisomy 8 is the most frequent cytogenetic alteration in AML occurring in 10 to 15% of patients 40,41 . This includes MDS, where trisomy 8 represents the sole genetic aberration in 11% of cases 42 . Patients with trisomy of chromosome 8 are included in the intermediate-risk group and their prognosis thought to be defined by accompanying aberrations 42 . Trisomy 8 is frequently associated with mutations in driver genes including RUNX1, ASXL1, DNMT3A 43 , TET2 44 , and IDH1 and IDH2 45 . AML with tetrasomy of chromosome 8 has poor prognosis 46 . To our knowledge, tetrasomies of only the p-arm of chromosome 8 have not yet been reported in myeloid diseases. However, it needs to be pointed out that the partial tetrasomy 8 can easily be mistaken to be trisomy 8. Indeed, this would have been our interpretation of conventional karyotyping and FISH data in the absence of quantitative data from the CytoScan array. Importantly, the q-arm carries several cancer-relevant genes such as FGFR1, KAT6A and PCM1 (Supplementary Table S1).
In conclusion, MBU-7 and MBU-8 differ in discrete gains of genetic material of chromosomes 7, 8, and 18. The clinical significance of these differences remains to be determined in future studies.
Abnormal and additional chromosomes may have consequences for disease outcome due to changes in gene expression. Compared to MBU-7, MBU-8 had an additional copy of chromosome 18, two additional copies of the p-arm of chromosome 8 but lacked part of a third copy of chromosome 7. We wondered whether these differences in chromosomes would be associated with differences in the response to drugs. Interestingly, our results showed a clear difference between MBU-7 and MBU-8 in the response towards the nucleoside analogues azacitidine, decitabine and cytarabine, with MBU-7 being significantly more sensitive. In this respect, it is interesting to note that 20% of MDS patients that relapsed after decitabine-treatment had acquired additional cytogenetic abnormalities including trisomy 8 47 . Taken together, these observations might indicate a possible link between gains in chromosome 8 and the response of cells to decitabine and related drugs.
Overexpression of the BCL2 gene on chromosome 18 contributes to MDS by blocking cell death 48 . Even though MBU-7 and MBU-8 differed in the copy number of the BCL2 gene, they were equally sensitive to the Bcl2-inhibitor venetoclax. From this observation we draw two conclusions: First, MBU-7 cells do not have a general hypersensitivity to treatments that induce cell death. Second, gene copy number of drug targets is not a reliable predictor of drug response. Potential synthetic lethality needs to be assessed on the functional level. The combination of well-characterized cell lines with powerful genetic loss of function screens has the potential to identify novel vulnerabilities associated with distinct genetic and cytogenetic abnormalities in close to unbiased manner 49 .
We propose the here described cell lines MBU-7 and MBU-8 as suitable models for screening for vulnerabilities that are associated with the disease-relevant gains in chromosome 7, 8 and 18.

Methods
Cell lines and cell culture. The human cell lines U-937 and SKK-1 were obtained from the Leibniz-Institute DSMZ-German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). We obtained SKK-1 cells as part of a collaboration with Prof. Hans Drexler that included their detailed cytogenetic and genetic characterization 50 . MBU-7 and MBU-8 have been isolated as single-cell clones and are described here in great detail. Both cell lines will be deposited at DSMZ. All cell lines were cultured in RPMI 1640 (ThermoFisher Scientific, Waltham, MA) supplemented with 10%, 1% Pen/Strep, and 1% l-glutamine at 37 °C in 5% CO 2 . All cell lines were routinely analysed for the presence of mycoplasma.
Identifier test. Cells were identified by comparing their short tandem repeat (STR) profiles to their corresponding profiles on the searchable cell line database, Cellosaurus (web.expasy.org/cellosaurus). Genomic DNA was extracted from collected cells using PureLink Genomic DNA Mini Kit (ThermoFisher Scientific) and processed by the Genomics facility (IGTP) using the CLA IdentiFiler Plus PCR Amplification kit (ThermoFisher Scientific) according to the manufacturer's instructions. Following the PCR, capillary electrophoresis was performed with the amplified microsatellite loci fragments and the allelic ladder, GeneScan 500 LIZ Size Standard, using the Applied Biosystems 3130xl Genetic Analyzer to discriminate alleles that differ by single nucleotides. GeneMapper software (ThermoFisher Scientific) was then used to call alleles according to size and quality based on the selected allelic ladder.