Preclinical characterisation and development of a novel myelodysplastic syndrome-derived cell line

Myelodysplastic syndrome (MDS) represents a group of heterogeneous haematopoietic disorders characterised by diverse clinical symptoms varying from mild anaemia to multilineage cytopenia.^1,2 Importantly, one-third of MDS patients will exhibit a transformation towards acute myeloid leukaemia,^3,4 increasing the complexity of the disease aetiology. Over the last decade, our understanding of the molecular pathogenesis of MDS was greatly increased by the development of several genetically and epigenetically modified mouse models.^5,6 However, a specific clinically relevant humanised MDS mouse model to evaluate the efficacy of new therapies is lacking.^7–9 Efforts to develop xenograft strategies have also been frustrated by misassignment of MDS cell lines (i.e. ofthe 31 MDS cell lines published to date, only one has been determined to truly represent MDS).,¹⁰ The MDS92 cell line, described by Tohyama et al.,¹¹ has been confirmed to have an MDS phenotype¹⁰ and originated from a bone marrow isolate of a 52-year-old man with refractory anaemia (RA) exhibiting ring sideroblasts. Subsequent investigations confirmed the presence of typical MDS aberrations such as 5q and 17p deletion, monosomy 7, a complex karyotype,^10,12 and clonal heterogeneity evidenced by MDS92’s ability to generate several subblast derivatives, including MDS-L.¹² Following its initial reporting,¹³ the MDS-L cell line has become a standard preclinical tool for MDS therapeutic development,^7–9 culminating in the establishment of a reproducible and widely used preclinical MDS model.⁸ However, a study by Kida et al.¹² recently demonstrated that MDS-L exhibits two major clones, termed MDS-L-2007 and MDS-LGF, where the MDS-LGF cells can proliferate with low addition of IL-3 (1 ng/ml), and MDS-L-2007 requires higher supplementation of IL-3 (100 ng/ml). Given the significant use of the MDS-L cells in MDS preclinical research,^7,14,15 it is critical to delineate and characterise whether both clones, independently or cumulatively, in the xenograft setting contribute to MDS towards development of appropriate preclinical models of MDS.

To evaluate the engraftment potential of the two cell lines, we performed extensive preclinical experiments in the NSG and NSGS mouse strains. Both cell lines were transduced with a dual reporter gene [luciferase and green fluorescent protein (GFP)] permitting longitudinal bioluminescence visualisation of in vivo cell engraftment. Thus, 5 × 10⁶ cells were injected intravenously into NSGS (n = 5) mice (Fig 1A). Clear evidence of engraftment was determined for the proposed MDS-L-2007 clone in all NSGS mice as early as week 4 with a progressive increase in bioluminescence (Fig 1A, B) in haematopoietic organs (Supplementary Figure S1A, B) leading to fatal haematologic disease, with an average latency of 55 ± 3 days (Fig 1C). MDS-L-2007 cells have also engrafted in NSG mice albeit with a longer disease latency (112 ± 6 days, data not shown). In contrast, MDS-LGF cells did not engraft either NSG or NSGS mice (in total n = 89 mice) by any route or precondition regimen tested (Supplementary Table SI). These results suggest that previously published xenografts of MDS-L might represent a specific subclonal engraftment of the MDS-L-2007 clone.⁸ To further characterise the differences between MDS-LGF and MDS-L-2007 clones, we performed immunophenotyping of the two cell lines by mass cytometry (Supplementary Table SII). As shown in Fig 1D, we found major differences in CD7 and HLA-DR expression, confirming Kida et al.’s previous observations. However, our extended panel additionally identified aberrant expression of CD33 (highly expressed on the cell surface of MDS-LGF cells) and CD38 (highly expressed on the cell surface of MDS-L-2007 cells; Fig 1D). Interestingly, low CD38 expression is a characteristic found in most MDS patient CD34⁺ cells.¹⁶ Conversely, CD33 is frequently expressed on MDS cells¹⁷ but its expression has not been found to correlate with clinical cytogenetics, therapy response or survival in MDS. However, as the biological function of CD33 has not been fully elucidated, its potential involvement in disease development cannot be excluded. Subsequently, fluorescence in situ hybridisation (FISH) analysis of chromosome 5q was performed on both cell lines. As shown in Fig 1E, F, the 5q deletion (5q31, EGR1) was observed in 82% of MDS-L-2007 cells with a log2 ratio of −0.321 by using a copy number variation (CNV) array (Supplementary Table SIII), although the log2 ratio in MDS-LGF was −0.187 and, surprisingly, below the probe cut-off (<10%). The presence of this aberration in MDS-L-2007 cells corresponds to results obtained in a previously established animal model using the MDS92 mother cell line.^8,10–12,18 Subsequently, we performed copy number arrays, G-banding and sequencing of 54 genes found to be frequently mutated in myeloid leukaemia. The results (Supplementary Table SIII) indicated that both cell lines present with complex karyotypes and mutations in several genes involved in the development of MDS (NRAS, CEBPA and TP53). The presence of these three mutations corroborates the previous observation by Kida et al.¹² In addition, the MDS-L-2007 line presents an increase in TET2 mutations (50%) (Supplementary Table SIII). TET2 mutation is one of the most common mutations in MDS with del5q. It is associated with the disease’s progression to either high-risk MDS or its transformation to AML.^19,20 Subsequently, we evaluated the cytokine release profile as previous observations have indicated that both cell lines present with contrasting dependencies to IL-3 in vitro. We looked at the expression of 17 cytokines using an ELISA approach. The cytokine profile of the two cell lines was similar for VEGF, Gas 6 and IL-8. However, the concentration of CCL2 was significantly higher for the MDS-L-2007 cells (P < 0.001) than for MDS-LGF cells (Fig 1G). Overall, our observations underline that the two MDS cell lines (MDS-LGF and MDS-L-2007) present with quite drastic differences in major key phenotypes characterising MDS (i.e. del5q and TET2 mutation). Moreover, we have also demonstrated that MDS-L-2007 cells overexpress CCL2 and CD7 in contrast to those of MDS-LGF. In conclusion, we suggest that most of the established MDS models, using the widely used MDS-L cell line, likely originate from the MDS-L-2007 cells.

Fig 1.

Development and characterisation of MDS-L-2007 and MDS-LGF sublines. (A) NSGS immunodeficient mice were injected intravenously with 5 × 10⁶ MDS-L-2007 (n = 5) or MDS-LGF (n = 5) luciferase-expressing cell lines and imaged by bioluminescence weekly. Sample data from one mouse per group are presented. (B) Quantification of total bioluminescent flux from luciferase-expressing cell lines in weeks 1 to 9 post injection. (C) NSG (n = 5) and NSGS (n = 5) animals were injected intravenously with 5 × 10⁶ MDS-L-2007 cells and evaluated for survival based on the Kaplan–Meier method; results of the Mantel–Cox log-rank test are shown. (D) Mass cytometric analysis of MDS-L sublines MDS-L-2007 and MDS-LGF with a panel of several monoclonal antibodies against cell surface markers and the proliferation marker Ki67 with all values. (E) Fluorescence in situ hybridisation (FISH) analysis of chromosome 5q, in n = 279 counted MDS-LGF, and (F) n = 272 MDS-L-2007 cells. 2G2R are cases with normal 5q with two green and two red, and 2G1R, direct loss of heterozygosity (del5q). (G) Cytokine expression of in vitro cultures of MDS-L-2007 and MDS-LGF versus blank controls was determined and data are represented as standard error of the mean (SEM) ± mean, by unpaired Student’s t-test on GraphPad prism version 7 (GraphPad Software, San Diego, CA, USA). Individual groups were compared using an unpaired t-test, *P < 0.05; **P < 0 01; ***P < 0.001.

As we have previously reported the use of a biocompatible 3D scaffold for engraftment of haematopoietic cells,²¹ we were interested to determine if it will still be possible to establish an MDS model from the MDS-LGF cell line. By using a 3 dimensional scaffolds (Department of Chemical Engineering, University of Massachusetts, Amherst, MA, USA), we first show, in vitro, that the scaffold supports and promotes MDS-LGF cells proliferation (Fig 2A, B). Employing these mechanically durable bio-mimetically designed hydrogel scaffolds supplemented with Matrigel® (Corning Life sciences, Tewksburry, MA, USA), growth of the MDS-LGF line (5 × 10⁵ cells expressing GFP and luciferase) implanted subcutaneously in NSG and NSGS mice was monitored weekly in vivo by bioluminescence imaging (Fig 2C). While NSG mice demonstrated only transient engraftment (Fig 2C), NSGS mice exhibited persistent engraftment of MDS-LGF in the scaffold for the duration of the experiment (20 weeks) (Fig 2D). No indication of extra-scaffold dissemination was apparent, in stark contrast to MDS-L-2007 engraftment in the same strain (Fig 1A). Immunohistochemistry and May–Grünwald Giemsa staining of scaffolds and recovered cells demonstrate that the cells present the same morphology as in vitro cultured MDS-LGF cells (Supplemental Figure S2). Finally, mass cytometric analysis of engrafted cells demonstrated a near-identical phenotype of MDS-LGF in vitro with engrafted cells showing enriched expression of the markers CD34, CD38 and CD45 (Fig 2E). To conclude, this series of experiments indicates that MDS-LGF cells can be used to model an indolent high-risk MDS phenotype.

Fig 2.

Development of an MDS xenograft model based on MDS-LGF cells. (A) Growth of MDS-LGF [1 × 10⁵ cells expressing green fluorescent protein (GFP) and luciferase] either suspended in Matrigel® and loaded into hydrogel scaffolds (+) or culture media (−) was monitored and quantified (B) in vitro by bioluminescence imaging at the times indicated. (C) Luciferase-expressing MDS-LGF cells (5 × 10⁶) were loaded into hydrogel scaffolds (n = 16) in Matrigel®, implanted subcutaneously in NSG (n = 8) and NSGS (n = 8) mice and imaged weekly. (D) Quantification of total bioluminescent flux for NSG (n = 8) and NSGS (n = 8) mice implanted subcutaneously with hydrogel scaffolds loaded with 5 × 10⁶ MDS-LGF cells in Matrigel®. (E) Mass cytometric analysis of in vitro cultures and ex vivo-recovered MDS-LGF cells from hydrogel scaffolds. Statistical analysis was performed by using an unpaired Student’s t-test to compare the differences between individual groups, *P value < 0.05; **P value < 0.01; ***P value < 0.001.

The MDS-L cell line has proven to be a crucial preclinical tool for the study of MDS. However, as MDS-L is a composite of two subclones (MDS-LGF and MDS-L-2007), a definitive characterisation and clarification are critical to the field. Our study provides an in-depth characterisation of both the MDS-L-2007 and MDS-LGF subclones and reports the development of a novel indolent high-risk MDS phenotype model using the MDS-LGF cells. This new model complements the aggressive phenotype MDS model using the MDS-L-2007 cells and these models should be of great benefit for the development of future therapeutics for MDS.

Supplementary Material

Fig S1. (A) Flow cytometry analysis of harvested cells from bone marrow and spleen of a mouse engrafted with MDS-L-2007 cells, which presented as histogram and evaluated the expression level of human CD33 and CD34. (B) May-Grünwald and Giemsa staining was performed from in vitro cultures of MDS-L-2007 and linked with bone marrow cytospine, spleen and peripheral blood of a mouse engrafted with MDS-L-2007 cells. (H) Compilation of genetic analysis including karyotyping, copy number analysis was performed using CytoScan HD SNP arrays and Amplicon sequencing results of the two cell lines using the 54-gene TruSight Myeloid Sequencing Panel. Segments with >50 markers (deletions) and >90 markers (duplications) are reported. Regions with mosaic copy number alteration are manually merged based on log2 and b-allelic difference.

Fig S2. Histopathology analysis of MDS-LGF xenografts. (A) May-Grünwald Giemsa staining of scaffolds and recovered cells exemplified the same morphology as in vitro cultured MDS-LGF cells in addition to positive staining for (B) MPO (inset 63x), (C) CD33 and (D) CD34 by immunohistochemistry. Microscopy images captured at 20x or 63x.

Table SI. In vivo studies completed with MDS-LGF cell line by using different injection routs.

Table SII. Antibodies used in Mass Cytometer.

Table SIII. Genetic analysis of MDS-L-2007 and MDS-LGF using karyotyping, G-Banding and True sight myeloid panel.

Table SIV. Number of cytokines used in Luminex assay.