© 2021. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 RESOURCE ARTICLE Extramedullary multiple myeloma patient-derived orthotopic xenograft with a highly altered genome: combined molecular and therapeutic studies Lourdes Farre1,‡, Gabriela Sanz2, Neus Ruiz-Xivillé 3,4, Manuel Castro de Moura4, Juan Francisco Martin-Tejera1, Samuel Gonçalves-Ribeiro1, Maria Martinez-Iniesta1, Monica Calaf1, Jose Luis Mosquera5, José Ignacio Martı́n-Subero6,7, Isabel Granada3,4, Manel Esteller4,7,8,9, Eva Domingo-Domenech2, Fina Climent8,10, Alberto Villanueva1,11, * and Anna Sureda2,* ABSTRACT progression in this case. Exome sequencing showed 24 high impact Extramedullary multiple myeloma (EMM) has an overall survival single nucleotide variants and 180 indels. From the genes involved, of 6 months and occurs in 20% of multiple myeloma (MM) patients. only TP53 was previously described as a driver in MM. A rather Genetic and epigenetic mechanisms involved in EMM and the balanced proportion of hyper/hypomethylated sites different to therapeutic role of new agents for MM are not well established. previously reported widespread hypomethylation in MM was also Besides, well-characterized preclinical models for EMM are not observed. Treatment with lenalidomide, dexamethasone and available. Herein, a patient-derived orthotopic xenograft (PDOX) carfilzomib showed a tumor weight reduction of 90% versus non- was generated from a patient with an aggressive EMM to treated tumors, whereas treatment with the anti-CD38 antibody study in-depth genetic and epigenetic events, and drug responses daratumumab showed a reduction of 46%. The generation of PDOX related to extramedullary disease. A fresh punch of an extramedullary from a small EMM biopsy allowed us to investigate in depth the cutaneous lesion was orthotopically implanted in NOD.Cg- molecular events associated with extramedullary disease in PrkdcscidIl2rgtm1Wjl/SzJ(NSG) mouse. The PDOX mimicked combination with drug testing. histologic and phenotypic features of the tumor of the patient. KEY WORDS: Extramedullary multiple myeloma, Patient derived Cytogenetic studies revealed a hyperploid genome with multiple orthoxenograft, Genetic alterations, Epigenetic alterations, genetic poor-prognosis alterations. Copy number alterations (CNAs) Therapeutic responses were detected in all chromosomes. The IGH translocation t(14;16)(q32;q23)IGH/MAF was already observed at the medullary INTRODUCTION stage and a new one, t(10;14)( p?11-12;q32), was observed only with Multiple myeloma (MM) accounts for 13% of hematological extramedullary disease and could be eventually related to EMM malignancies (Becker, 2011; Howlader et al., 2009). It is characterized by clonal proliferation of neoplastic plasma cells 1 within the bone marrow (Bladé et al., 2011). Survival of patients Group of Chemoresistance and Predictive Factors, Subprogram Against Cancer Therapeutic Resistance, Catalan Institute of Oncology, Oncobell Program, Bellvitge with MM has increased significantly in the last two decades due Biomedical Research Institute (IDIBELL), 08908 L’Hospitalet del Llobregat, to the use of high-dose therapy followed by autologous stem cell Barcelona, Spain. 2Department of Clinical Hematology, Catalan Institute of transplantation and the emergence of multiple novel therapies Oncology, Bellvitge Biomedical Research Institute, 08908 L’Hospitalet del Llobregat Barcelona, Spain. 3Hematological Laboratory, Germans Trias i Pujol (Anderson et al., 2012). However, over time, MM patients often can Hospital, Catalan Institute of Oncology, 08916 Badalona, Barcelona, Spain. 4 exhibit worse and less durable responses to the different lines Cancer and Leukemia Epigenetics and Biology and Experimental and Clinical Hematology Programs, Josep Carreras Leukaemia Research Institute, 08916 of treatment, and even develop extramedullary disease in 10-30% Badalona, Barcelona, Spain. 5IDIBELL Bioinformatic Unit – Bellvitge Biomedical of cases (Bladé et al., 2011; Varga et al., 2015; Bhutani et al., 2020). Research Institute (IDIBELL), 08908 L’Hospitalet del Llobregat, Barcelona, Spain. In extramedullary multiple myeloma (EMM), myeloma cells 6 Biomedical Epigenomics Group, Institut d’investigacions Biomèdiques August Pi I become independent of bone marrow microenvironment, infiltrate Disease Models & Mechanisms Sunyer (IDIBAPS), University of Barcelona, 08036 Barcelona, Spain. 7Institució Catalana de Recerca i Estudis Avançats, 08010 Barcelona, Spain. 8Centro de other organs and/or circulate freely in the blood (Bladé et al., 2011). Investigació n Biomé dica en Red de Cancer, Carlos III Institute of Health, 28029 The most common sites for extramedullary disease were skin, soft Madrid, Spain. 9Physiological Sciences Department, School of Medicine and Health Sciences, University of Barcelona, 08036 Barcelona, Spain. 10Department tissues and liver (Usmani et al., 2012). The outlook for patients with of Pathology, Hospital Universitari de Bellvitge – Bellvitge Biomedical Research EMM is poor, with survival no longer than 3 years (Bhutani et al., Institute, 08907 L’Hospitalet de Llobregat, Barcelona, Spain. 11Xenopat S.L., Business Bioincubator, Bellvitge Health Science Campus, 08907 L’Hospitalet de 2020; Bladé et al., 2011). EMM management is challenging due Llobregat, Barcelona, Spain. to the lack of a current rationale favoring a specific therapeutic *These authors contributed equally to this work class among those available (Touzeau and Moreau, 2016). The ‡ Author for correspondence (lfarre@iconcologia.net) current treatment strategy is tailored to patient age and fitness. The molecular mechanisms involved in EMM are not well known. L.F., 0000-0002-3168-1940; M.E., 0000-0003-4490-6093 Furthermore, relevant in vivo preclinical models for EMM are not This is an Open Access article distributed under the terms of the Creative Commons Attribution available. These difficult in-depth studies combined molecular and License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, therapeutic approaches for extramedullary disease. Patient-derived distribution and reproduction in any medium provided that the original work is properly attributed. orthotopic xenografts (PDOX), in which a human tumor biopsy is Handling Editor: Elaine R. Mardis implanted in immunodeficient mice in the same organ as the tumor Received 7 December 2020; Accepted 5 May 2021 is grown in the patient, are the most advanced in vivo tumor 1 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 preclinical models, as they reproduce better patient tumor behavior thickening and thoracic lymphadenopathies were also observed. The and therapeutic responses than other models, such as subcutaneous patient died five months later after showing no response to two other xenografts or models based on the injection of cell lines (Byrne et al., lines of treatment that consisted of the salvage chemotherapy regime 2017). Herein, we aimed to generate a PDOX derived from an D-PACE (dexamethasone, cisplatin, Adriamycin, cyclophosphamide extramedullary cutaneous lesion of a resistant and rapidly progressive and etoposide) in November 2016, and the combination of EMM patient with very poor outcomes to exhaustively investigate the methotrexate and cytarabine, which started in December 2016. genetic and epigenetic events and therapeutic responses related to Details of the patient clinical course and treatment schemes are extramedullary disease. described in Materials and Methods. A fresh small punch biopsy of an extramedullary skin lesion RESULTS was obtained at the beginning of the third line of treatment to generate EMM patient and PDOX model generation the EMM-derived PDOX model (EMM PDOX). The model was The PDOX was derived from an EMM cutaneous lesion of a 62- successfully generated by orthotopic cutaneous surgical implantation year-old female patient who was diagnosed with IgG kappa MM, in one NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG) mouse. The 1 mm3 stage IIIA (Durie–Salmon classification), in July 2016. During tumor fragment implanted in the mouse reached a volume of 1 cm3 diagnosis, the patient’s bone marrow aspirate demonstrated a 44% 30 days after implantation, showing a rapid growth rate. The infiltration by plasma cells with CD38 positive and CD56 and macroscopic aspect of the xenograft tumor is shown in Fig. 1A. CD119 negative expression. She rapidly progressed to EMM with The mouse was sacrificed, and the tumor was re-implanted into two the presence of multiple cutaneous lesions after the first line of other animals to expand and to perpetuate the model, and then treatment consisted of a combination of bortezomib, thalidomide processed in order to exhaustively investigate this aggressive EMM and dexamethasone (VTD). At extramedullary disease, multiple case by way of histologic, phenotypic, genetic and epigenetic studies. cutaneous lesions were positive for dermal infiltration by plasma cell EMM PDOX histology was similar to its biopsy precursor, showing neoplasm (Kappa positive, CD138 and MUM-1 positive; Ki67 groups of large cells with poorly differentiated morphology, broad positivity of 90%). Massive bone marrow infiltration by plasmoblasts, cytoplasm and nuclei with prominent nucleoli and multiple mitotic progressive left pleural effusion that was also specific and pleural figures (Fig. 1B). EMM PDOX cells showed a Ki67 positivity of 90% Disease Models & Mechanisms Fig. 1. Characterization of EMM PDOX. (A) Macroscopic appearance of PDOX. (B) H&E staining of the patient’s extramedullary lesion (PT, upper panel) and xenografted tumor (PDOX, lower panel), showing histological similarities between the original tumor sample and the originated EMM PDOX (at 400× magnification). (C) EMM PDOX cells showed a Ki67 positivity of 90% and CD38 expression, as observed in the patient’s precursor tumor. (D) Paired microsatellite genotyping of PDOX and the patient’s lesion from which it was derived. The genetic match between the generated PDOX and the patient’s tumor was assessed by microsatellite genotyping. The microsatellites used for tumor genotyping were (1) D5S299, (2) D5S346, (3) D3S1612, (4) D5S82 and (5) D3S3564. For each microsatellite, the upper panel corresponds to the patient’s biopsy and the lower panel corresponds to PDOX. 2 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 and were positive for CD38 expression, as observed in EMM patient heterogeneity (LOH) (Fig. 4A). Trisomy of chromosomes 3, 5, 7, 9, lesions (Fig. 1C). The genetic matching between EMM PDOX and 15, 17 and 21, and tetrasomy of 11 and 19 were observed. The most the patient’s biopsy was also investigated, and was demonstrated by frequent CNAs were extensively intrachromosomic amplifications. microsatellite genotyping using six different microsatellite markers as In chromosomes 1, 5, 7, 9, 11, 12 and 19, gains of four or more shown in Fig. 1D. copies were detected. Among them, chromosome 7 was notable An EMM cell line was also derived from tumor xenograft, and with 10 to 18 copies in the 7p12 region, 20 to 28 copies in the 7p15 was grown in vitro in a suspension culture. Most of the cells formed region and 16 copies in the 7p21 region, as well as chromosome 19 cell clumps and showed a doubling time of 41±6 h (mean±s.d.). The with 7 copies in 19p13 (Fig. 4B,C; Table S1). CNAs affected an EMM cell line was maintained until passage 10, and after each exceptionally large number of genes that are annotated in Table S1. passage an aliquot was cryopreserved. During each passage, clumps Among others, we underlined the following alterations previously were brought into a single-cell suspension and the cells associated with poor prognosis: losses in 1p32.3 affecting the FAF1 spontaneously formed clumps again. The EMM cell line also grew and CDKN2C genes; gains in 1q21 affecting the CKS1B gene; LOH and generated new plasmocytomas when injected in NSG mice. A of chromosome 13; and LOH in the short arm of chromosome 17 total of 1×106 cells soaked in Matrigel were intradermally injected in (13.3 mb), which includes the TP53 gene. In chromosome 7 two back flanks of NSG mice (n=6 injections in three NSG mice), ( p15.3), a region of 1210.71 kb, with a gain of 28 copies, affected generating tumor masses in all cases (Fig. 2A). Plasmocytomas 17 genes, including the gene encoding the cytokine IL6. reached a mean volume of 280±34 mm3 (mean±s.e.m.) 30 days Besides cytogenetic abnormalities, 712 single nucleotide post-injection (Fig. 2B,C). Cell line-derived plasmocytomas showed variations (SNVs) (24 of high, 202 of moderate, 409 of low impact a similar histology to the original PDOX tumor and the patient’s and 77 modifier), 114 deletions and 66 insertions were found by tumor, as well as similar Ki67 expression (Fig. 2D,E). whole-exome sequencing in the EMM PDOX (Fig. 5A). These alterations were distributed in all chromosomes, with a high number Genetic and epigenetic analysis of EMM PDOX of variants in chromosome 19 but fewer in chromosomes 9, 13, 18 Karyotype analysis of EMM PDOX identified a hyperploid and 21 (Fig. 5B). In our analysis, indel length was limited to 20 bp, genome. Fluorescence in situ hybridization (FISH) analysis for with the most frequent being those with 1 to 3 nucleotides for both IGH rearrangements was performed in the EMM PDOX and insertions and deletions (Fig. 5C). Genes affected by high impact additionally in the remaining patient’s bone marrow aspirate SNVs are annotated in Table S2. Among them, we highlight genes obtained at MM diagnosis, already at medullary stage implicated in cell cycle regulation (TP53, cdc7, CDC25A, TATD1 and (Fig. 3A,B). Both samples showed positivity for the t(14;16)(q32; PDE5), gene transcription (HDAC4), cell-to-cell adhesion (CTNNA3) q23)IGH/MAF rearrangement but were negative for the and metabolic processes (AMD1 and GAPDH). Additional analysis t(4;14)( p16;q32)IGH/FGFR3 rearrangement. The EMM PDOX by type/transcript biotype percentages and transversion/transition also showed the t(10;14)( p?11-12;q32) IGH translocation, distributions are shown in Fig. 5D-F. identified at G banding resolution, that was not observed in the Whole DNA methylome of the EMM PDOX was also evaluated bone marrow plasma cells at medullar stage (Fig. 3C,D). and compared with a published dataset of 101 newly diagnosed Genetic imbalance analysis with a Cytoscan 750K array in the and chemotherapy-naïve MM patients and normal plasma cells EMM PDOX showed a highly disturbed genome with multiple (NPCs) (Agirre et al., 2015) (Fig. 6A). As compared to NPCs, we CNAs in all chromosomes, including gains, losses and/or loss of identified a total of 138,648 differentially methylated CpGs in the Fig. 2. Derived EMM cell line generated plasmocytomas when re-implanted in NSG mice. (A) Representative image of a plasmocytoma derived from EMM cell line by intradermal injection of 1×106 cells soaked in Matrigel in the flank of an NSG mouse. (B) Tumor growth rate of plasmocytomas generated after the intradermic injection of 1×106 cells of the EMM cell line (n=6). (C) Tumors reached a mean volume of 280±34 mm3 at 30 days post-injection. (D) H&E staining of the cell line-derived plasmocytomas showing histological similarities with the PDOX and the patient’s biopsy (at 400× magnification). (E) Cell line-derived plasmocytomas showed a high Ki67 expression as observed in the EMM PDOX (at 400× magnification). Data are mean±s.e.m. Disease Models & Mechanisms 3 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 Fig. 3. Fluorescence in situ hybridization analysis of the immunoglobulin heavy chain gene locus (IGH). (A) IGH rearrangement was observed in only one chromosome at diagnosis before the occurrence of extramedullary disease (Break Apart Probe). (B) FISH analysis of t(14;16) IGH-MAF rearrangement of EMM PDOX showing the translocation in the two chromosomes. (C,D) Micrograph of G-band metaphase chromosomes of the PDOX (C) and FISH (D) analysis of IGH rearrangements, identifying chromosomes 10 (region p) and 16 (region q) implicated in IGH t(14;16)(q32;q23) and t(10;14)( p? 11-12;q32) translocations. The later translocation was not detected at MM diagnosis. All images are at 1000× magnification. EMM PDOX, including 55.3% hypomethylated and 44.7% dexamethasone, which resulted in an 81% tumor weight reduction hypermethylated (Fig. 6B). We then compared the methylome compared with control tumors (P=0.0030). Daratumumab had a of the EMM PDOX with the methylomes of MM patients significant but only partial response, showing a 46% tumor weight and identified 112,551 differentially methylated CpGs in the reduction compared with control tumors (P=0.043). No signs of EMM PDOX, including 10.4% hypomethylated and 89.6% animal toxicity were observed with any of these combinations. In hypermethylated, thus observing that EMM PDOX was more the residual masses, we evaluated: (1) histological changes by hypermethylated than the MM cases (Fig. 6C). Methylation- Hematoxylin and Eosin (H&E) staining; (2) cellular proliferation by regulated genes in EMM PDOX in comparison to MM cases are Ki67 staining (IHQ); (3) the presence of apoptotic cells by TUNEL annotated in Table S3. Distributions of hyper/hypomethylated sites assay; (4) the expression of CD38 by immunohistochemistry and (5) by genome functional regions are shown in Fig. S1. Even a the molecular regulation of cell cycle, cell death and endoplasmic significant number of CpGs that were differentially methylated reticulum stress pathways by western blotting. between EMM PDOX and NPCs, the percentage of total In lenalidomide plus dexamethasone-treated tumors and in hypermethylated and hypomethylated CpGs, and their distribution those that additionally received carfilzomib, TUNEL assay showed by genome functional regions, were quite similar. The a similar number of apoptotic cells compared with control tumors, differentially methylated CpGs between EMM PDOX and MM and a decrease of Ki67 expression was observed (Fig. 7D). Western cases were distributed more frequently in body regions for both blot analysis showed a decrease of cyclin D1, cyclin E and hypermethylated and hypomethylated CpGs. Bcl-XL expression, and FAK phosphorylation. In lenalidomide and dexamethasone-treated tumors, a decrease of eIF2α phosphorylation Therapy responses in EMM PDOX in comparison to control tumors was detected, which was not Owing to the lack of response to high-dose chemotherapy regimens observed in tumors also treated with carfilzomib. The addition of observed in the patient, we took advantage of the generated EMM carfilzomib to the lenalidomide and dexamethasone treatment was PDOX to test new available therapies for MM not used with the associated with a decrease in AKT expression (Fig. 7E). patient from whom the PDOX derived, and which have not been The specificity and effectiveness of daratumumab treatment Disease Models & Mechanisms explored much for extramedullary disease control. Tumor was were confirmed by a decrease of CD38 expression detected by expanded by orthotopical implantation in 40 additional NSG mice. IHQ and western blot in residual tumor masses, observed only We included the anti-CD38 antibody daratumumab in monotherapy for this treatment (Fig. 7D,E). Daratumumab response was (in this case EMM cells showed CD38 expression) and the associated with an increase of apoptosis (evaluated by TUNEL combination of lenalidomide and dexamethasone with or without assay) compared to control tumors but similar Ki67 expression the proteosome inhibitor carfilzomib, as these schemes are (Fig. 7D). In Daratumumab-treated tumors, an increase of p53 and recommended for the treatment of MM in patients who have cyclin D1 and B1 expression, and a decrease of AKT, cyclin E and received at least one previous therapy. A significant reduction in Bcl-XL expression and FAK and eIF2α phosphorylation were both tumor volume and weight was observed in all treated groups observed (Fig. 7E). compared with control group (Fig. 7A-C). The highest antitumoral effect was achieved with the combination of lenalidomide, DISCUSSION dexamethasome and carfilzomib, with a tumor weight reduction The generation of an EMM PDOX starting from a small fresh punch of 90% compared with tumors treated with vehicle (control tumors, of an extramedullary cutaneous lesion allowed us to investigate in P=0.0015), followed by the combination of lenalidomide and depth the molecular alterations in combination with the therapeutic 4 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 Fig. 4. Cytogenetic abnormalities observed in EMM PDOX. (A) Diagram representing the cytogenetic abnormalities observed in EMM PDOX by single- nucleotide polymorphism genotyping microarray performed using a CytoScan 750 K Array. CNAs are extensively distributed along all chromosomes and included gains, losses and LOH. (B,C) Presence of extensive intrachromosomal CNAs. Details of CNA analysis of the short arm of chromosome 7 (B) showing a highly disturbed genome with gains of up to 28 copies concomitant with extensive genetic losses and LOH and detail of CNA in the short arm of chromosome 19 (C) showing a gain region of 12.8 mb with seven copies. responses of an aggressive case of MM that progressed rapidly to translocation that could eventually be related to disease progression extramedullary disease. PDOX have been demonstrated to be a to extramedullary disease in this case. To our knowledge, this powerful tool for expanding patient tumor tissue while maintaining translocation has not been previously reported in MM (Bhutani et al., their histological and genetic features, supporting the detection of new 2020; Shah et al., 2018; Walker et al., 2018) but has been in six cases genetic and epigenetic events associated with extramedullary disease of chronic lymphocytic leukemia and in one case of nodal marginal and allowing parallel and rational evaluation of in vivo drug response. zone B-cell lymphoma (Martín-Subero et al., 2007; Rouhigharabaei Disease Models & Mechanisms The molecular mechanisms involved in EMM are not well et al., 2013). known. In the case shown here, a highly disturbed genome with Even trisomy in odd chromosomes is commonly associated with IGH translocations, wide chromosomal imbalances, SNVs and a good prognosis in MM. This case also presented a lot of a different methylation pattern to MM cases were detected. previously reported poor prognostic cytogenetic abnormalities that Chromosomal translocations involving IGH genes have a central could contribute to the observed poor outcome, such as the loss in role in MM, and their incidence increases with each successive 1p32.3 and gain in 1q21 (An et al., 2014; Avet-Loiseau et al., 2009; stage of disease (Bergsagel and Kuehl, 2018; Walker et al., 2018). Besse et al., 2016; Boyd et al., 2011; Hebraud et al., 2015; Shah This aggressive case already presented the t(14;16)(q32;q23)IGH/ et al., 2018; Walker et al., 2010). The LOH observed in the short arm MAF translocation in bone marrow tumoral plasma cells at diagnosis, of chromosome 17(13.3 mb) was indicative of a deletion of this which, mainly in combination with negative expression of CD56, as region that includes the TP53 gene. This poor prognosis alteration observed here, has been associated with an unfavorable outcome in was not detected in the patient at the MM diagnosis and may also be MM (Bhutani et al., 2020; Narita et al., 2015; Shah et al., 2018; related to extramedullary disease in this case. Monosomy of Walker et al., 2018). Furthermore, extramedullary tumor in the chromosome 13 and deletion 1p13.2-34.2 have been reported in PDOX additionally presented the t(10;14) (p?11-12;q32) another multidrug-resistant EMM patient (Egan et al., 2013), 5 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 Fig. 5. Variants in EMM PDOX detected by whole-exome sequencing. (A) Total number of SNVs classified as high, moderate and low impact and modifier, and insertions and deletions. (B) Variant counts per chromosome in terms of high-impact SNVs, insertions and deletions (indels). (C) Counts of indels by length. In the analysis, indels were considered as deletions or insertions of up to 20 bp. (D,E) Percentage of SNVs per impact by type (D) and by transcript biotype (E). All 712 SNVs were included in the analysis. (F) Relative concentrations of point mutation types detected. To obtain the COSMIC signatures contribution, it was assumed that all SNVs detected corresponded to somatic mutations. suggesting that these events may have a role or be frequent in newly diagnosed and chemotherapy-naïve MM patients (Agirre extramedullary disease. In some chromosomes, gains of more than et al., 2015). Indeed, the ratio of hyper/hypomethylated sites in five copies were detected. A region of 1210.71 kb in 7p15 with a this EMM case was more similar to NPCs than to MM, although gain of 28 copies contained the gene encoding for the IL6 cytokine methylation in this EMM case and in NPCs affected different that has been reported as a crucial factor for the proliferation and CpGs. A higher degree of hypermethylation has also been observed survival of myeloma cells (Rosean et al., 2014). in plasma cell leukemia in comparison to MM, and has been related to Among genes with high-impact point mutations observed in this independence from the bone marrow microenvironment (Walker case, only TP53 was described as a driver in MM (Walker et al., et al., 2011). It could be interesting to further investigate whether 2018). Mutations in the other genes have not been previously DNA hypermethylation plays a role in the progression of MM to reported in MM or EMM (Bhutani et al., 2020; Morgan et al., 2012; more aggressive forms, in contrast with the hypomethylation process Rasche et al., 2017; Walker et al., 2018) but some of them have been observed in the transition from NPC to MM. Differently implicated in MM pathogenesis. For instance, HDAC4 is implicated hypermethylated CpGs between this EMM case and MM cases in MM cell survival and migration, and panobinostat, a pan-HDAC were preferably distributed within gene bodies and intergenic regions. Disease Models & Mechanisms inhibitor, in combination with a proteasome inhibitor and The role of new agents in the treatment success of EMM is dexamethasone, has improved survival in relapsing and refractory still poorly understood. Several clinical reports demonstrate MM patients (Imai et al., 2019). An inhibitor of PDE5, tadalafil, has effectiveness of bortezomib (Patriarca et al., 2005; Rosiñol et al., been demonstrated to have a dramatic and durable anti-myeloma 2006), as well of lenalidomide and pomalidomide (Ito et al., immune and clinical response in a patient with end-stage relapsed/ 2013; Nakazato et al., 2012). The data in EMM with incorporation of refractory MM (Noonan et al., 2014). An inhibitor of Cdc7, the carfilzomib are not readily available. There are limited data regarding PHA-767,491, has been reported to block proliferation and to the efficacy of daratumumab, either alone or in combination, in induce apoptosis of MM cell lines (Coyne et al., 2009). It has also EMM, although information coming from retrospective analyses been described that STK4 has a role in MM cell survival, and STK4 does not seem to be very optimistic, with modest (Jullien et al., 2019) inhibitors may represent novel therapeutic options for patients with or low (Pick et al., 2018) responses to daratumumab. Moreover, MM (Cottini et al., 2014). expression levels of CD38 do not seem to be a predictive factor of Regarding methylation, the rather balanced proportion of response, as suggested by the responses found in our murine model. hyper/hypomethylated sites observed in this EMM case contrasts Our preclinical results may suggest the potential value of the with the previously reported widespread hypomethylation in combined therapies in EMM even with a highly disturbed genome. 6 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 Fig. 6. Whole-genome methylation analysis of EMM PDOX. (A) Unsupervised hierarchical clustering analysis of EMM PDOX, MM, NPCs from tonsil (tPC) or NPCs from bone marrow (bmPC). The 5000 CpG sites with the most variable methylation values were used. (B) Differentially methylated CpGs in EMM PDOX compared to NPCs and (C) newly diagnosed and chemotherapy- naïve MM patients (Agirre et al., 2015). Compared with NPCs, we identified 138,645 differentially methylated CpGs, including 55.3% hypomethylated and 44.7% hypermethylated. As compared with MM, we identified 112,551 differentially methylated CpGs, including 10.4% hypomethylated and 89.6% hypermethylated. Epigenetic data were obtained using an Infinium MethylationEPIC BeadChip array (Illumina) for EMM PDOX (from four different fragments of the tumor) and using a 450 K methylation array (Illumina) for MM and NPCs (Agirre et al., 2015). To compare the results with the published 450 K methylation arrays, only the common probes with EPIC were taken into account, and average methylation values were used for each group. Relevant and well-characterized preclinical models that simulate MATERIALS AND METHODS the patient’s disease are critical tools for advancing understanding of Patient data the pathogenesis of EMM and in the preclinical discovery of new The patient from whom the PDOX was derived was a 62-year-old female, therapy strategies. Available preclinical models of MM mainly with a past medical history of toxic nodular goiter and bilateral breast include myeloma models in immunocompetent mice (Radl et al., cancer in 2003, who was diagnosed with IgG kappa MM, stage IIIA (Durie– Salmon classification, International Score System of 2) in July 2016. The 1988; Fowler et al., 2009), xenograft models in severe combined patient was started on VTD [1.3 mg/m2 bortezomib (days 1, 4, 8 and 11)], immunodeficiency (SCID) and non-obese diabetic (NOD)/SCID thalidomide up to 150 mg daily and 40 mg dexamethasone (days 1-4 and 9- mice, in which human MM cell lines or cells derived from MM 12) for 4 cycles, with the objective of consolidating the response with patients are injected (Chauhan et al., 2006; Neri et al., 2008), Disease Models & Mechanisms autologous stem cell transplantation. After disease progression, she was and genetically modified models (Lwin et al., 2016). However, started on D-PACE [40 mg dexamethasone, 10 mg/m2 cisplatin, 10 mg/m2 preclinical models for EMM are scarce and unique. PDOX models adriamycin, 400 mg/m2 cyclophosphamide and 40 mg/m2 etoposide (days have been shown to mimic the same pattern of pharmacological 1-3), ×2 cycles] in November 2016, after disease progression with the response as tumor in the patient (Byrne et al., 2017; Hidalgo et al., reappearance of disseminated skin lesions. The patient began a third line 2014; Piulats et al., 2018; Capellá et al., 1999; Vidal et al., consisting of 1000 mg/m2 methotrexate (day 1) and 1000 mg/m2 (days 2-3) 2012), aspects that do not generally occur in xenograft models based in December 2016. Unfortunately, the clinical condition of the patient rapidly deteriorated and she died. The patient gave written consent to on subcutaneous implantation whereby the tumor loses the tumor participate in the study. The Ethics Committee of the Bellvitge Hospital microenvironment, or based on the injection of cell lines that lose the approved the study protocol, and the animal experimental design was heterogeneity of the tumor. The generation of EMM PDOX reported approved by the IDIBELL animal facility committee. This study was here also facilitated the creation of a novel EMM cell line capable of conducted according to the principles expressed in the Declaration of forming new plasmacytomas when injected in mice that will allow Helsinki. future complemental functional studies. As is the case with in vivo preclinical models, cell lines for EMM are scarce. Additionally, the PDOX generation approach presented here could also potentially support a The PDOX model was generated by orthotopic cutaneous surgical personalized treatment strategy for EMM in a clinical setting. implantation of the fresh small punch biopsy of an infiltrative 7 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 Fig. 7. In vivo therapeutic assays using EMM PDOX. EMM xenografts were expanded orthotopically to 40 additional NSG mice. After evidence of homogeneous tumor growth, animals were randomly distributed in four groups (n=10) and treated with saline (control), with daratumumab (DARA) in monotherapy, with the combination of lenalidomide (LENA) and dexamethasone (DEXA), or the combination of lenalidomide, dexamethasone and carfilzomib (CARFIL) for 15 days. No signs of animal toxicity were observed with any of these treatments. (A) Evaluation of tumor volumes during tumor treatment. (B) Tumor weights at the end of the treatment at mice sacrifice for the different experimental groups. (C) Representative diagram of tumors dissected at sacrifice for the different experimental groups. (D) H&E staining, IHQ evaluation of Ki67 and CD38, and TUNEL assay (at 400× magnification). A decrease of CD38 expression was observed only in daratumumab-treated tumors. A decrease of Ki67 was observed after the treatment with lenalidomide-dexamethasone with or without carfilzomib, without a significant increase of apoptotic cells. A pro-apoptotic effect was observed in daratumumab tumors with a significant increase of apoptotic cells evaluated by TUNEL assay with respect to the control group without significant Ki67 expression change. (E) Western blot analysis of CD38 and the relevant proteins implicated in cell cycle control and apoptosis. A decrease of CD38 expression was observed also by western blot only in daratumumab-treated tumors. Data are mean±s.e.m. Disease Models & Mechanisms extramedullary skin lesion in one 6-week-old female NOD.Cg- tumor mass was fragmented and processed for in-depth histologic, PrkdcscidIl2rgtm1Wjl/SzJ (NSG) mouse. Briefly, a cutaneous back shave immunophenotypic, genetic and epigenetic studies. biopsy of normal mouse skin was performed by removing a rhomboid area of ∼6 mm2 (4×3 mm), and then the small fragment of tumor tissue Histologic and immunohistochemistry analyses was anchored with a prolene 7.0 suture to the edge of the skin, fixing the For histologic analyses, a tumor fragment was fixed in formalin for 24 h, tumor at four anchored sites coinciding with each corner. To improve the embedded in paraffin and stained with H&E. The histology of tumor PDOX implantation process, the anchored skin tumor was protected by making a was compared with the histology of extramedullary cutaneous lesion from skin fold that was fixed with three points with 5.0 silk sutures (Vizoso et al., which the PDOX was derived by a haematopathologist. For IHQ analysis, 3- 2015). Tumor growth and animal weight were monitored once a week. The µm slices of paraffin-embedded tissues were used. Primary antibodies animal was sacrificed when tumor volume reached 1 cm3. Upon tumor used were monoclonal antibodies for Ki67 (RM-9106-51, Sigma-Aldrich) growth, three fragments of ∼20 mm3 volume were re-implanted and CD38 (NB-22-8022, Neobiotech), and were used at a 1:200 dilution. orthotopically in three other NSG mice in order to expand the tumor. Retrieval was performed with citrate buffer ( pH 6.0) using a pressure Two fragments of similar volume were cryopreserved, and the remaining cooker. Reactions were visualized using the EnVision anti-mouse 8 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 antibody system, and developed using the DAB-Plus Kit (Dako, following the manufacturer’s instructions. Quality control of raw reads Copenhagen, Denmark). Slides were counterstained with Harry’s was performed using FastQC (version 0.11.7) (Wingett and Andrews, 2018). modified Hematoxylin. Raw reads were trimmed using Timmomatic software (version 0.38) (Bolger et al., 2014). Trimmed reads were aligned to the human genome Microsatellite genotyping (version hg38 from the UCSC) using the BWA-MEM algorithm (www.arxiv. We compared the amplification pattern of D5S299, D5S346, D3S1612, org/abs/1303.3997) from BWA (version 0.7.17) (Li and Durbin, 2010) D5S82 and D3S3564 genomic microsatellites between the generated tumor with default parameters. Resulting BAM files were sorted and indexed xenograft and the patient’s skin biopsy from which it was derived. DNA using Sambamba (version 0.6.8) (Tarasov et al., 2015). Duplicate reads were was extracted from a fresh fragment of the tumor xenograft ( passage 1) identified and removed from sorted BAM files using Picard Tools (version using the phenol/chloroform/isoamyl method, and from five slides of 2.18.16) (http://broadinstitute.github.io/picard). For the variant calling, ten mycras of a formalin-fixed fragment of the patient’s skin lesion using filtering and annotation, SNVs and short indels were called with the a DNA QiAamp FFPE tissue kit (Qiagen), following the manufacturer’s Genome Analysis Toolkit (GATK) (version 3.8-1-0) with default parameters instructions. First, the microsatellites were amplified by PCR using (McKenna et al., 2010). SNVs and indels were separately filtered using labeled primers. The primer sequences were as follows: D3S1612 UP-5′- the recommended variant quality score recalibration according to GATK (YAKYE)GCTCTCCTCAGTGGAAAATT-3′ and DW-5′-ATGTAGAA- best practice recommendations (Depristo et al., 2011). Filtered called GAGGATGATCTCC-3′; D3S3564 UP-5′-(FAM)AGCTAAACACAGTC- variants were annotated with dbSNP 146 rsIDs, the 1000 Genomes Phase TAACTGCAT-3′ and DW 5′-CCCACAGAGTGATAGGGA-3′; D5S299 1 high-confidence SPNs release, and the Mills and 1000G gold standard UP-5′-(ATTO550)GTAAGCAGGACAA GATGACAG-3′ and DW indels. Prediction of effects was performed using snpEff (Cingolani et al., 5′-GCTATTCTCTCAGGATCTTG-3′; D5S82 UP-5′-(FAM)CCCAATT- 2012). GTATAGATTTTAGAAGTC-3′ and DW 5′-ATCAGAGTATCAGAAT- TTCT-3′; and D5S346 UP-5′-(FAM)ACTCACTCTAGTGATAAAT- Methylation analysis CGGG-3′ and DW 5′-AGCAGATAAGACAGTATTACTAGTT-3′. All Genome-wide DNA methylation analysis of EMM xenograft tumor was microsatellites were amplified in the same conditions but in separated performed using an Infinium Methylation EPIC BeadChip kit (Illumina), reactions, starting with 50 ng of DNA and with 5 µl of Mega Mix double following the manufacturer’s recommendations. Four different tumor (Microzone) and 10 µM of each labeled primer in a final volume of 10 µl. fragments were included ( passage 1) to cover possible tumor Amplifications were carried out using an initial denaturing step at 96°C for heterogeneity. DNA was extracted using a phenol/chloroform/isoamyl 5 min, 35 cycles of denaturing at 96°C, annealing at 55°C (D5S299/ D5S82/ protocol. The quality of the DNA was checked by electrophoresis using a D5S346) or 58° (D3S1612/D3S3564), and extension steps at 72°C (30 s 1% agarose and 1/100,000 SyberSafe gel, and quantified using a Nanodrop- each), and a final extension step of 72°C for 3 min. The PCR products were 1000 spectrophotometer. For bisulfite conversion, we started with 600 ng of first visualized in a 1.5% agarose gel and then diluted 1/100 in a buffer each sample, and a EZ DNA methylation kit (Zymo Research) was used containing HI-DI formamide and Genescan 500 LIZ standard marker for according to the manufacturer’s recommendations for the Illumina Infinium their visualization and analysis in a capillary sequencer (ABIPRISM assays. The effectiveness of the bisulfite conversion was checked for three 3130XL, Applied Biosystems) using GeneScan Analysis (Thermo Fisher controls that were converted simultaneously with the samples. Bisulfite- Scientific). converted DNA (4 μl) was used for hybridization on Infinium Methylation Epic BeadChip, following the Illumina Infinium HD Methylation protocol. Analysis of genomic imbalances The array was evaluated using the Illumina HiScan SQ scanner. The Cytogenetic analysis was performed with DNA from tumor xenograft intensities of the images were extracted using the minfi 1.30 R package and ( passage 1) using a CytoScan 750K Array (Affymetrix, Thermo Fisher normalized using the ssNoob method. The Illumina control probes and the Scientific) with genome-wide coverage (750,000). The sample was snp-containing probes were removed from subsequent analysis. The processed in the GeneChip System 3000 platform from Affymetrix methylation score for each CpG was represented as a beta value according (Thermo Fisher Scientific). The Chromosome Analysis Suite program to the fluorescent intensity ratio. Beta values may take any value between 0 was used for the analysis (v.3.2, Thermo Fisher Scientific) with version 33.1 (non-methylated) and 1 (completely methylated). Differential methylation that uses the UCSC hg19 version of the human genome for the annotations analysis was performed using the MEAL v1.14 R library and CpGs with of genes. The alterations were analyzed with a minimum of 25 affected an absolute beta value difference greater than 0.2 and a false discovery markers and an average resolution of 110 kb. rate lower than 0.05 were selected. For comparison, a published DNA methylome dataset of 101 newly diagnosed and chemotherapy-naïve MM patients and NPCs obtained using HumanMethylation450 Beadchip Karyotype and FISH analysis (Illumina) were included (Agirre et al., 2015). In that dataset, NPCs PDOX tumor cells were dissociated using collagenase, and then cultured in were composed from three normal bone marrows ( pools from four RPMI medium (GIBCO, Thermo Fisher Scientific) with 20% fetal bovine patients each) and eight non-tumoral tonsils. In order to compare the serum (FBS; GIBCO, Thermo Fisher Scientific), insulin (2.5 UI/ml) and 1× results with the published 450K methylation microarrays, only the common streptomycin and penicillin (GIBCO, Thermo Fisher Scientific) for 4 days. Disease Models & Mechanisms probes with EPIC were taken into account, and average methylation Conventional cytogenetic analysis was performed on cultured cells and values were used for each group. The selected CpGs were annotated with the chromosome G-banding was carried out using standard techniques. UCSC hg19 genome version and the categories TSS 1500, TSS 200, 5′ Interphase FISH was performed on fixed cells from bone marrow at UTR, first exon, gene body, 3′ UTR and intergenic, related to their gene diagnosis and from dissociated and cultured tumor cells with the following location. probes: XL IGH Break Apart Probe; XL t(4;14) FGFR3/IGH Translocation/ Dual Fusion Probe; and XL t(14;16) IGH/MAF Translocation/Dual Fusion In vivo therapeutic assays Probe (MetaSystems). For in vivo therapeutic assays, tumor was expanded by orthotopic implantation in 40 additional 6-week-old female NSG mice, and when Whole-exome sequencing tumor volumes raised 400-500 mm3 homogeneous size, animals were Genomic DNA from the xenograft tumor was fragmented to an average size of distributed randomly to four experimental groups (n=10) and treated for 150 bp and subjected to DNA library creation using established Illumina 15 days with: (1) group 1 – control (saline intraperitoneally five times a week); paired-end protocols. Adapter-ligated libraries were amplified and indexed (2) group 2 – daratumumab (three doses of 20 mg/kg intraperitoneally on via PCR. The Agilent SureSelect All Exon V6 library preparation kit (without day 1, 7 and 13); (3) group 3 – lenalidomide (1 mg/kg, oral) plus UTR, Agilent) was used and sequence targets were captured and amplified in dexamethasone (15 mg/kg, intraperitoneally), four consecutive doses of accordance with the manufacturer’s recommendations. Enriched libraries both with an interval of 2 days between each cycle (total of twelve doses); and were subjected to 150 base paired-end sequencing (Illumina Novaseq System) (4) group 4 – carfilzomib (6 mg/kg, i.p., two consecutive doses at days 1-2, 9 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 6-7 and 12-13) plus lenalidomide plus dexamethasone (same regime as IDIBELL for mouse care and maintenance. We also thank the staff of the Affymetrix described in group 3). During treatment, tumor growth and animal weight Microarrays Platform of the Josep Carreras Leukemia Research Institute. were monitored twice a week. Tumor volume was estimated by measuring tumor width (W, mm) and length (L, mm) with a caliper, and using the Competing interests formula v=(π×L×W2)/6. At the end of treatment, animals were sacrificed A.V. is a co-founder and shareholder of Xenopat S.L. (Barcelona, Spain). and tumors were harvested, weighed and processed for histological, A.S. receives honoraria from Janssen and Celgene and participates on their Advisory Boards. No potential conflicts of interest were disclosed by the other immunohistochemical and molecular studies. authors. Western blot analysis Author contributions Protein extraction was performed using RIPA buffer [50 mM Tris-HCl ( pH Conceptualization: L.F., N.R.-X., I.G., M.E., E.D., F.C., A.V., A.S.; Methodology: L.F., 7.4), 150 mM sodium chloride, 1 mM ethylenediaminetetraacetic acid, 1% G.S., N.R.-X., J.F.M.-T., S.G.-R., M.M.-I., M.C., J.L.M., J.I.M.-S., F.C.; Software: NP-40, 1% sodium deoxycholic acid and 0.1% sodium dodecylsulfate] with M.C.d.M., J.L.M.; Validation: L.F., G.S., J.F.M.-T., M.M.-I.; Investigation: G.S., N.R.- the addition of protease and phosphatase inhibitor cocktails (Roche). Cyclin X., M.C.d.M., J.F.M.-T., S.G.-R., M.M.-I., M.C., J.L.M., J.I.M.-S., I.G., E.D.-D., F.C.; D (sc-8396), Cyclin E (sc-247), Bcl-XL (sc-8392) and tubulin (sc-5286) Resources: L.F., I.G.; Data curation: M.C.d.M., S.G.-R., J.L.M., J.I.M.-S.; Writing - antibodies were purchased from Santa Cruz Biotechnology and were used at original draft: L.F., A.V., A.S.; Writing - review & editing: G.S., N.R.-X., M.C.d.M., J.F.M.-T., S.G.-R., M.M.-I., M.C., J.L.M., J.I.M.-S., I.G., M.E., E.D.-D., F.C., A.V., a 1:1000 dilution. p-FAK (8556, used at a 1:500 dilution), p53 (2524, used A.S.; Visualization: F.C.; Supervision: A.V., A.S.; Funding acquisition: L.F., M.E., at a 1:2000 dilution), AKT (4691, used at a 1:1000 dilution) and P-elf2alfa A.V., A.S. (3398, used at a 1:1000 dilution) antibodies were obtained from Cell Signaling Technology. The CD38 (NB-22-6755) antibody was purchased Funding from Neobiotech and was used at a 1:1000 dilution. The TUNEL assay This study was supported by grants from the Spanish Ministry of Economy and (ApopTag Peroxidase In Situ Apoptosis kit, Millipore) was performed Competitiveness – Instituto de Salud Carlos III FEDER (PI16/01,898 and PI19/ following the manufacturer’s recommendations. 01320), the Fundació n hna and the Fundació n Cientı́fica Asociació n Espanola ̃ Contra el Cá ncer (IDEAS098VILL-IDEAS16), projects to A.V. This project also Derivation of EMM cell line received funding from the European Union Horizon 2020 research and innovation program under the Marie Sklodowska Marie Curie grant agreement number 799850 To derive the EMM cell line, a xenograft tumor fragment of 125 mm3 from (to L.F.). The IDIBELL Bioinformatics Unit was supported by the Instituto de Salud passage 2 was first excised manually using two sterile scalpels with small Carlos III, through the grant ‘Contratos de té cnicos bioinformá ticos de apoyo a la fragments in a sterile 100 mm culture dish with 8 ml of Dulbecco’s modified investigació n en los IIS 2018’ (CA18/00045), and the European Social Fund (‘ESF – Eagle medium (DMEM) medium containing 15.000 IU of IV collagenase. Investing in your future’). The suspension of fragments was then transferred to a 50 ml conical centrifugation tube and digested in the collagenase solution for 1 h at 37°C. Data availability During this period, the tube was swirled vigorously every 15 min. The cell Data from exome sequencing were submitted to the EMBL-EBI European suspension was then disaggregated by pipetting, filtered through a 100 μm Nucleotide Archive public repository under accession numbers PRJEB45604 cell strainer (Corning, Merck) and centrifugated at 180 g for 10 min. The cell (ERP129739) for the project and ERX5685456 for the data. Data from the pellet was rinsed with PBS and counted in a chamber slide using an automated methylation array and CGH array were submitted to the EMBL-EBI ArrayExpress public repository under accession numbers E-MTAB-10654 and E-MTAB-10630, cell counter (Countess II, Thermo Fisher Scientific). Cells were re-centrifuged respectively. in the same conditions and resuspended in DMEM containing 5% glutamine (DMEM Glutamax, GIBCO, Thermo Fisher Scientific) and complemented References with 20% FBS, 1% streptomycin and penicillin, 10 μg/ml of insulin and Agirre, X., Castellano, G., Pascual, M., Heath, S., Kulis, M., Segura, V., 30 μg/ml of transferrin (all obtained from Thermo Fisher Scientific) to a Bergmann, A., Esteve, A., Merkel, A., Raineri, E. et al. (2015). Whole- concentration of 2.5×104 cell/ml in a T-25 flask. Cells were cultured at 37°C epigenome analysis in multiple myeloma reveals DNA hypermethylation of B cell- in a CO2 incubator. To perform different passes, cells were brought into a specific enhancers. Genome Res. 25, 478-487. doi:10.1101/gr.180240.114 single-cell suspension by pipetting. Then cells were counted in a chamber An, G., Xu, Y., Shi, L., Zhong, S., Deng, S., Xie, Z., Sui, W., Zhan, F. and Qiu, L. slide using an automated cell counter. Then, cells were rinsed with PBS (2014). Chromosome 1q21 gains confer inferior outcomes in multiple myeloma (Thermo Fisher Scientific) by centrifugation at 180 g for 10 min and re- treated with bortezomib but copy number variation and percentage of plasma cells involved have no additional prognostic value. Haematologica 99, 353-359. doi:10. suspended in a 50:50 mixture of new and reconstituted medium (from the 3324/haematol.2013.088211 previous passage), and subcultured by dilution. Cells were cryopreserved Anderson, K. C. (2012). The 39th David A. Karnofsky Lecture: bench-to-bedside in FBS with 10% DMSO. All protocols were performed using sterile translation of targeted therapies in multiple myeloma. J. Clin. Oncol. 30, 445-452. techniques in a Class II Type A2 laminar flow hood. To test whether the doi:10.1200/JCO.2011.37.8919 EMM-derived cell line could generate new plasmocytomas, 1×106 cells Avet-Loiseau, H., Li, C., Magrangeas, F., Gouraud, W., Charbonnel, C., soaked in 50 µl of Matrigel (BD Biosciences) were intradermally injected in Harousseau, J.-L., Attal, M., Marit, G., Mathiot, C., Facon, T. et al. (2009). two flanks of three NSG mice (total of 6 injections). Tumor growth was Prognostic significance of copy-number alterations in multiple myeloma. J. Clin. Disease Models & Mechanisms monitored every 5 days post-injection by measuring tumor width (W, mm) Oncol. 27, 4585-4590. doi:10.1200/JCO.2008.20.6136 and length (L, mm) with a caliper until day 30 post-injection when mice were Becker, N. (2011). Epidemiology of multiple myeloma. Recent Results Cancer Res. 183, 25-35. doi:10.1007/978-3-540-85772-3_2 sacrificed. Tumor volume was estimated using the same formula indicated Bergsagel, P. L. and Kuehl, W. M. (2018). Detailing the genomic landscape of above. At sacrifice, tumors were harvested and processed for histological and myeloma. Blood 132, 554-555. doi:10.1182/blood-2018-06-857508 IHQ studies. Besse, L., Sedlarikova, L., Greslikova, H., Kupska, R., Almasi, M., Penka, M., Jelinek, T., Pour, L., Adam, Z., Kuglik, P. et al. (2016). Cytogenetics in multiple Statistical analysis myeloma patients progressing into extramedullary disease. Eur. J. Haematol. 97, Statistical analysis was performed in WES, and methylation studies were 93-100. doi:10.1111/ejh.12688 described in the corresponding sections. Tumor weights were expressed in Bhutani, M., Foureau, D. M., Atrash, S., Voorhees, P. M. and Usmani, S. Z. (2020). Extramedullary multiple myeloma. Leukemia 34, 1-20. doi:10.1038/ mg and tumor volumes in mm3, and represented as mean±s.e.m. Tumor s41375-019-0660-0 volumes, as well as final tumor weight, were compared between each Bladé , J., Ferná ndez de Larrea, C., Rosiñol, L., Cibeira, M. T., Jimé nez, R. and treatment group and control group using a parametric Student’s t-test and Powles, R. (2011). Soft-tissue plasmacytomas in multiple myeloma: incidence, GraphPad Prism v7 Software. mechanisms of extramedullary spread, and treatment approach. J. Clin. Oncol. 29, 3805-3812. doi:10.1200/JCO.2011.34.9290 Acknowledgements Bolger, A. M., Lohse, M. and Usadel, B. (2014). Trimmomatic: a flexible trimmer We thank the Centres de Recerca de Catalunya program/Generalitat de Catalunya for Illumina sequence data. Bioinformatics 30, 2114-2120. doi:10.1093/ for their institutional support. We thank the staff of the Animal Core Facility of bioinformatics/btu170 10 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 Boyd, K. D., Ross, F. M., Walker, B. A., Wardell, C. P., Tapper, W. J., Chiecchio, Nakazato, T., Mihara, A., Ito, C., Sanada, Y. and Aisa, Y. (2012). Lenalidomide is L., Dagrada, G., Konn, Z. J., Gregory, W. M., Jackson, G. H. et al. (2011). active for extramedullary disease in refractory multiple myeloma. Ann. Hematol. Mapping of chromosome 1p deletions in myeloma identifies FAM46C at 1p12 and 91, 473-474. doi:10.1007/s00277-011-1272-6 CDKN2C at 1p32.3 as being genes in regions associated with adverse survival. Narita, T., Inagaki, A., Kobayashi, T., Kuroda, Y., Fukushima, T., Nezu, M., Clin. Cancer Res. 17, 7776-7784. doi:10.1158/1078-0432.CCR-11-1791 Fuchida, S., Sakai, H., Sekiguchi, N., Sugiura, I. et al. (2015). T(14;16)-positive Byrne, A. T., Alfé rez, D. G., Amant, F., Annibali, D., Arribas, J., Biankin, A. V., multiple myeloma shows negativity for CD56 expression and unfavorable Bruna, A., Budinská , E., Caldas, C., Chang, D. K. et al. (2017). Interrogating outcome even in the era of novel drugs. Blood Cancer J. 5, e285. doi:10.1038/ open issues in cancer precision medicine with patient-derived xenografts. Nat. bcj.2015.6 Rev. Cancer 17, 254-268. doi:10.1038/nrc.2016.140 Neri, P., Tagliaferri, P., Di Martino, M. T., Calimeri, T., Amodio, N., Bulotta, A., Capellá , G., Farré , L., Villanueva, A., Reyes, G., Garcı́a, C., Tarafa, G. and Lluı́s, Ventura, M., Eramo, P. O., Viscomi, C., Arbitrio, M. et al. (2008). In vivo F. (1999). Orthotopic models of human pancreatic cancer. anti-myeloma activity and modulation of gene expression profile induced by Chauhan, D., Hideshima, T. and Anderson, K. C. (2006). A novel proteasome valproic acid, a histone deacetylase inhibitor. Br. J. Haematol. 143, 520-531. inhibitor NPI-0052 as an anticancer therapy. Br. J. Cancer 95, 961-965. doi:10. doi:10.1111/j.1365-2141.2008.07387.x 1038/sj.bjc.6603406 Noonan, K. A., Ghosh, N., Rudraraju, L., Bui, M. and Borrello, I. (2014). Targeting Cingolani, P., Platts, A., Wang, L. L., Coon, M., Nguyen, T., Wang, L., Land, S. J., immune suppression with PDE5 inhibition in end-stage multiple myeloma. Cancer Lu, X. and Ruden, D. M. (2012). A program for annotating and predicting the Immunol. Res. 2, 725-731. doi:10.1158/2326-6066.CIR-13-0213 effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Patriarca, F., Prosdocimo, S., Tomadini, V., Vasciaveo, A., Bruno, B. and Fanin, Drosophila melanogaster strain w1118; iso-2; iso-3. Fly (Austin). 6, 80-92. doi:10. R. (2005). Efficacy of bortezomib therapy for extramedullary relapse of myeloma 4161/fly.19695 after autologous and non-myeloablative allogeneic transplantation. Cottini, F., Hideshima, T., Buhrlage, S., Gray, N. S., Richardson, P. G., Tonon, G. Haematologica 90, 278-279. and Anderson, K. C. (2014). Serine/threonine kinase STK4 is a novel target in Pick, M., Vainstein, V., Goldschmidt, N., Lavie, D., Libster, D., Gural, A., myeloma. Blood 124, 645-645. doi:10.1182/blood.V124.21.645.645 Grisariu, S., Avni, B., Ben Yehuda, D. and Gatt, M. E. (2018). Daratumumab Coyne, M. R. E., Naughton, S., Hayden, P. J., Montagnoli, A., O′Dwyer, M. E. and resistance is frequent in advanced-stage multiple myeloma patients irrespective of Santocanale, C. (2009). Targeting Cdc7 kinase in multiple myeloma. Blood 114, CD38 expression and is related to dismal prognosis. Eur. J. Haematol. 100, 3847-3847. doi:10.1182/blood.V114.22.3847.3847 494-501. doi:10.1111/ejh.13046 Depristo, M. A., Banks, E., Poplin, R., Garimella, K. V., Maguire, J. R., Hartl, C., Piulats, J. M., Vidal, A., Garcı́a-Rodrı́guez, F. J., Muñoz, C., Nadal, M., Philippakis, A. A., del Angel, G., Rivas, M. A., Hanna, M. et al. (2011). A Moutinho, C., Martı́nez-Iniesta, M., Mora, J., Figueras, A., Guinó , E. et al. framework for variation discovery and genotyping using next-generation DNA (2018). Orthoxenografts of testicular germ cell tumors demonstrate genomic sequencing data. Nat. Genet. 43, 491-501. doi:10.1038/ng.806 changes associated with cisplatin resistance and identify PDMP as a resensitizing Egan, J. B., Kortuem, K. M., Kurdoglu, A., Izatt, T., Aldrich, J., Reiman, R., agent. Clin. Cancer Res. 24, 3755-3766. doi:10.1158/1078-0432.CCR-17-1898 Phillips, L., Baker, A., Shi, C. X., Schmidt, J. et al. (2013). Extramedullary Radl, J., Croese, J. W., Zurcher, C., Van Den Enden-Vieveen, M. H. M. and myeloma whole genome sequencing reveals novel mutations in Cereblon, De Leeuw, A. M. (1988). Animal model of human disease. Multiple myeloma. Am. J. Pathol. 132, 593-597. proteasome subunit G2 and the glucocorticoid receptor in multi drug resistant Rasche, L., Chavan, S. S., Stephens, O. W., Patel, P. H., Tytarenko, R., disease. Br. J. Haematol. 161, 748-751. doi:10.1111/bjh.12291 Ashby, C., Bauer, M., Stein, C., Deshpande, S., Wardell, C. et al. (2017). Fowler, J. A., Mundy, G. R., Lwin, S. T., Lynch, C. C. and Edwards, C. M. (2009). Spatial genomic heterogeneity in multiple myeloma revealed by multi-region A murine model of myeloma that allows genetic manipulation of the host sequencing. Nat. Commun. 8, 268. doi:10.1038/s41467-017-00296-y microenvironment. DMM Dis. Model. Mech. 2, 604-611. doi:10.1242/dmm. Rosean, T. R., Tompkins, V. S., Tricot, G., Holman, C. J., Olivier, A. K., Zhan, F. 003160 and Janz, S. (2014). Preclinical validation of interleukin 6 as a therapeutic target in Hebraud, B., Magrangeas, F., Cleynen, A., Lauwers-Cances, V., Chretien, M.-L., multiple myeloma. Immunol. Res. 59, 188-202. doi:10.1007/s12026-014-8528-x Hulin, C., Leleu, X., Yon, E., Marit, G., Karlin, L. et al. (2015). Role of additional Rosiñol, L., Cibeira, M. T., Uriburu, C., Yantorno, S., Salamero, O., Bladé , J. and chromosomal changes in the prognostic value of t(4;14) and del(17p) in multiple Montserrat, E. (2006). Bortezomib: An effective agent in extramedullary disease myeloma: the IFM experience. Blood 125, 2095-2100. doi:10.1182/blood-2014- in multiple myeloma. Eur. J. Haematol. 76, 405-408. doi:10.1111/j.0902-4441. 07-587964 2005.t01-1-EJH2462.x Hidalgo, M., Amant, F., Biankin, A. V., Budinská , E., Byrne, A. T., Caldas, C., Rouhigharabaei, L., Ferreiro, J. F., Put, N., Michaux, L., Tousseyn, T., Clarke, R. B., de Jong, S., Jonkers, J., Mælandsmo, G. M. et al. (2014). Patient- Lefebvre, C., Gardiner, A., De Kelver, W., Demuynck, H., Verschuere, J. derived xenograft models: an emerging platform for translational cancer research. et al. (2013). BMI1, the polycomb-group gene, is recurrently targeted by genomic Cancer Discov. 4, 998-1013. doi:10.1158/2159-8290.CD-14-0001 rearrangements in progressive B-cell leukemia/lymphoma. Genes. Howlader, N., Ries, L. A., Stinchcomb, D. G. and Edwards, B. K. (2009). The Chromosomes Cancer 52, 928-944. doi:10.1002/gcc.22088 impact of underreported veterans affairs data on national cancer statistics: Shah, V., Sherborne, A. L., Walker, B. A., Johnson, D. C., Boyle, E. M., Ellis, S., analysis using population-based SEER registries. JNCI J. Natl. Cancer Inst. 101, Begum, D. B., Proszek, P. Z., Jones, J. R., Pawlyn, C. et al. (2018). Prediction of 533-536. doi:10.1093/jnci/djn517 outcome in newly diagnosed myeloma: A meta-analysis of the molecular profiles Imai, Y., Hirano, M., Kobayashi, M., Futami, M. and Tojo, A. (2019). HDAC of 1905 trial patients. Leukemia 32, 102-110. doi:10.1038/leu.2017.179 inhibitors exert anti-myeloma effects through multiple modes of action. Cancers Tarasov, A., Vilella, A. J., Cuppen, E., Nijman, I. J. and Prins, P. (2015). (Basel) 11, 475. doi:10.3390/cancers11040475 Sambamba: fast processing of NGS alignment formats. Bioinformatics 31, Ito, C., Aisa, Y., Mihara, A. and Nakazato, T. (2013). Lenalidomide is effective for 2032-2034. doi:10.1093/bioinformatics/btv098 the treatment of bortezomib-resistant extramedullary disease in patients with Touzeau, C. and Moreau, P. (2016). How I treat extramedullary myeloma. Blood multiple myeloma: report of 2 cases. Clin. Lymphoma, Myeloma Leuk. 13, 83-85. 127, 971-976. doi:10.1182/blood-2015-07-635383 doi:10.1016/j.clml.2012.08.005 Usmani, S. Z., Heuck, C., Mitchell, A., Szymonifka, J., Nair, B., Hoering, A., Jullien, M., Trudel, S., Tessoulin, B., Mahé , B., Dubruille, V., Blin, N., Gastinne, Alsayed, Y., Waheed, S., Haider, S., Restrepo, A. et al. (2012). Extramedullary Disease Models & Mechanisms T., Bonnet, A., Lok, A., Lebourgeois, A. et al. (2019). Single-agent disease portends poor prognosis in multiple myeloma and is over-represented in daratumumab in very advanced relapsed and refractory multiple myeloma high-risk disease even in the era of novel agents. Haematologica 97, 1761-1767. patients: a real-life single-center retrospective study. Ann. Hematol. 98, doi:10.3324/haematol.2012.065698 1435-1440. doi:10.1007/s00277-019-03655-5 Varga, C., Xie, W., Laubach, J., Ghobrial, I. M., O’Donnell, E. K., Weinstock, M., Li, H. and Durbin, R. (2010). Fast and accurate long-read alignment with Burrows- Paba-Prada, C., Warren, D., Maglio, M. E., Schlossman, R. et al. (2015). Wheeler transform. Bioinformatics 26, 589-595. doi:10.1093/bioinformatics/ Development of extramedullary myeloma in the era of novel agents: no evidence btp698 of increased risk with lenalidomide-bortezomib combinations. Br. J. Haematol. Lwin, S. T., Edwards, C. M. and Silbermann, R. (2016). Preclinical animal models 169, 843-850. doi:10.1111/bjh.13382 of multiple myeloma. Bonekey Rep. 5, 772. doi:10.1038/bonekey.2015.142 Vidal, A., Munoz, ̃ C., Guillen, M., Moreto, J., Puertas, S., Martinez-Iniesta, M., Martı́n-Subero, J. I., Ibbotson, R., Klapper, W., Michaux, L., Callet-Bauchu, E., Figueras, A., Padulles, L., Garcia-Rodriguez, F. J., Berdiel-Acer, M. et al. Berger, F., Calasanz, M. J., De Wolf-Peeters, C., Dyer, M. J., Felman, P. et al. (2012). Lurbinectedin (PM01183), a new DNA minor groove binder, inhibits (2007). A comprehensive genetic and histopathologic analysis identifies two growth of orthotopic primary Graft of cisplatin- resistant epithelial ovarian cancer. subgroups of B-cell malignancies carrying a t(14;19)(q32;q13) or variant BCL3- Clin. Cancer Res. 18, 5399-5411. doi:10.1158/1078-0432.CCR-12-1513 translocation. Leukemia 21, 1532-1544. doi:10.1038/sj.leu.2404695 Vizoso, M., Ferreira, H. J., Lopez-Serra, P., Carmona, F. J., Martı́nez-Cardú s, A., McKenna, A., Hanna, M., Banks, E., Sivachenko, A., Cibulskis, K., Kernytsky, Girotti, M. R., Villanueva, A., Guil, S., Moutinho, C., Liz, J. et al. (2015). A., Garimella, K., Altshuler, D., Gabriel, S., Daly, M. et al. (2010). The genome Epigenetic activation of a cryptic TBC1D16 transcript enhances melanoma analysis toolkit: a MapReduce framework for analyzing next-generation DNA progression by targeting EGFR. Nat. Med. 21, 741-750. doi:10.1038/nm.3863 sequencing data. Genome Res. 20, 1297-1303. doi:10.1101/gr.107524.110 Walker, B. A., Leone, P. E., Chiecchio, L., Dickens, N. J., Jenner, M. W., Boyd, Morgan, G. J., Walker, B. A. and Davies, F. E. (2012). The genetic architecture of K. D., Johnson, D. C., Gonzalez, D., Dagrada, G. P., Protheroe, R. K. M. et al. multiple myeloma. Nat. Rev. Cancer 12, 335-348. doi:10.1038/nrc3257 (2010). A compendium of myeloma-associated chromosomal copy number 11 RESOURCE ARTICLE Disease Models & Mechanisms (2021) 14, dmm048223. doi:10.1242/dmm.048223 abnormalities and their prognostic value. Blood 116, e56-e65. doi:10.1182/blood- Walker, B. A., Mavrommatis, K., Wardell, C. P., Cody Ashby, T., Bauer, M., 2010-04-279596 Davies, F. E., Rosenthal, A., Wang, H., Qu, P., Hoering, A. et al. (2018). Walker, B. A., Wardell, C. P., Chiecchio, L., Smith, E. M., Boyd, K. D., Neri, A., Identification of novel mutational drivers reveals oncogene dependencies in multiple myeloma. Blood 132, 587-597. doi:10.1182/blood-2018-03-840132 Davies, F. E., Ross, F. M. and Morgan, G. J. (2011). Aberrant global methylation Wingett, S. W. and Andrews, S. (2018). FastQ Screen: a tool for multi-genome patterns affect the molecular pathogenesis and prognosis of multiple myeloma. mapping and quality control. F1000Research 7, 1338. doi:10.12688/ Blood 117, 553-562. doi:10.1182/blood-2010-04-279539 f1000research.15931.1 Disease Models & Mechanisms 12