An orthotopic syngeneic mouse model of bortezomib‐resistant multiple myeloma

While bortezomib has significant benefits in multiple myeloma (MM) therapy, the disease remains incurable due to the invariable development of bortezomib resistance. This emphasises the need for advanced models for preclinical evaluation of new therapeutic approaches for bortezomib‐resistant MM. Here, we describe the development of an orthotopic syngeneic bortezomib‐resistant MM mouse model based on the most well‐characterised syngeneic MM mouse model derived from spontaneous MM‐forming C57BL/KaLwRij mice. Using bortezomib‐resistant 5TGM1 cells, we report and characterise a robust syngeneic mouse model of bortezomib‐resistant MM that is well suited to the evaluation of new therapeutic approaches for proteasome inhibitor‐resistant MM.

The introduction of proteasome inhibitors (PIs), like bortezomib, carfilzomib and ixazomib, into the treatment of multiple myeloma (MM) patients has significantly extended their survival. 1The anti-MM effects of PIs arise mainly through the induction of an apoptotic endoplasmic reticulum stress response in MM cells, although other mechanisms also contribute, including modulation of NF-κB transcription factor signalling and changes to the bone marrow (BM) microenvironment. 1 MM, however, remains an incurable disease with drug resistance invariably developing.
Resistance to bortezomib, the first-in-class and most commonly used PI, is complex and can involve several mechanisms, including proteasome mutation/upregulation, MM cell de-differentiation and autophagy. 1Next-generation PIs such as carfilzomib can be effective against bortezomib-resistant MM; however, cross-resistance between PIs can exist, 2 suggesting that the development of new PIs may not be a durable solution to overcoming bortezomib resistance.
These findings emphasise the need for advanced models for the preclinical evaluation of new therapeutic approaches against bortezomib-resistant MM.Multiple groups have generated PI-resistant cell lines in vitro and used these to generate subcutaneous flank xenograft models in immunocompromised mice. 3However, MM cell survival and drug resistance are strongly influenced by interactions with the BM microenvironment, 4 meaning that orthotopic and immunocompetent models are likely to better represent the disease.While orthotopic, syngeneic PI-resistant mouse models of MM have been derived from Vk*Myc, NRas/Myc or Bcl-xL/ Myc mice, these transgenic mouse models best represent disease driven by Myc activation, [5][6][7] which, although common, is not observed in around a third of MM patients. 8he most well-characterised syngeneic MM mouse model is that derived from spontaneous MM-forming C57BL/KaLwRij (KaLwRij) mice.The murine 5TGM1 MM cell line derived from these mice can be isografted into young KaLwRij mice, which forms aggressive disease that accurately replicates many facets of human MM, including lytic bone disease and serum paraprotein production, 9 but unlike previously generated Mycdriven models, [5][6][7] does not have a Myc transgene, nor Myc amplification/mutation/rearrangement. 10 To study new approaches to target bortezomib-resistant MM, we previously generated bortezomib-resistant 5TGM1 mouse MM cells by in vitro culture at escalating doses of bortezomib. 2 The resulting 5TGM1.BR cells are approximately 10-fold less sensitive to bortezomib in vitro than the parental (5TGM1.wt)cells. 2 Since the main utility of 5TGM1 cells is their capacity to rapidly engraft into the BM of KaLwRij mice to generate an immunocompetent syngeneic model of MM 2-3 weeks after intravenous administration of cells, we next attempted to engraft 5TGM1.BR cells into KaLwRij mice.However, engraftment was negligible after 3-4 weeks in most KaLwRij mice, even with a fivefold increase in the number of injected cells (Figure 1A).

Manjun
We reasoned that the low engraftment of 5TGM1.BR cells into KaLwRij mice may be due either to loss of homing to the BM or to immune rejection.To examine this, we administered 5TGM1.BR cells intravenously to immunocompromised NSG mice.Here, the 5TGM1.BR cells were able to generate significant disease within the BM (Figure 1B), suggesting that they retain BM-homing ability but may be subject to immune rejection in immunocompetent KaLwRij mice.
We next subcutaneously engrafted 5TGM1.wtand 5TGM1.BR cells into the flanks of both NSG and KaLwRij mice.Both 5TGM1.wtand 5TGM1.BR cells yielded similar-sized tumours in NSG mice (Figure 1C), consistent with the comparable proliferation rates of these cells in vitro (Figure S1A).In contrast, 5TGM1.BR cells formed substantially smaller subcutaneous tumours in KaLwRij mice than 5TGM1.wtcells (Figure 1C; Figure S1B).Analysis of the tumour immune infiltrate showed that while CD8 + T cell infiltration was similar in both tumour types forming in KaLwRij mice (Figure S1C), there were more CD4 + T cells and NK cells in tumours formed from 5TGM1.BR cells compared to 5TGM1.wt cells (Figure 1D).Taken together, these observations suggest that the lack of efficient engraftment of 5TGM1.BR cells into KaLwRij mice may be due to immune rejection.Consistent with this, comparative gene expression analysis 2 showed significant downregulation of multiple pathways associated with the major histocompatibility (MHC) class I antigen presentation pathway in 5TGM1.BR cells (Figure 1E; Figure S1D), which is well known to enhance NK cell-mediated cell killing through 'missing-self' recognition. 11hile 5TGM1.BR cells failed to efficiently engraft into KaLwRij mice, some mice did show engraftment by bioluminescence imaging after an extended period (10 weeks; Figure 1F).However, engraftment was highly variable (Figure S2A) and, therefore, not useful as a model to evaluate new therapeutic approaches for PI-resistant MM.To overcome this, we isolated BM from KaLwRij mice showing engraftment at later times and expanded the cells in vitro.These cells retained bortezomib resistance, albeit less than the original 5TGM1.BR cells (Figure 1G).This modest decrease in bortezomib resistance was observed across cells isolated from multiple engrafted mice (Figure S2B).The cells retained green fluorescent protein expression and luciferase activity (Figure S2C,D) and strong resistance to ixazomib but not carfilzomib (Figure S2E,F).Eliciting carfilzomib resistance and further enhancing bortezomib resistance in these cells will be the subject of future studies.
Next, we attempted to re-engraft these cells isolated from the BM of mice back into KaLwRij mice via intravenous administration.This resulted in quite consistent BM engraftment with disease progression comparable to that observed for 5TGM1.wtcells, with low but detectable disease 3 weeks after administration of cells that progressed markedly in the following week (Figure 1H).While some engraftment variability remained, this aligned with the sex of the recipient mice, with more efficient engraftment in males compared to females (Figure 1H).This effect appeared largely due to a delay in engraftment in female mice, with most female mice eventually showing engraftment 6 weeks after cell administration (Figure 1I).Thus, this process generated engraftable 5TGM1.BR cells, which we termed 5TGM1.BR E cells.
We next examined the effect of bortezomib in the 5TGM1.BR E /KaLwRij mouse model of MM. 5TGM1.BR E cells were engrafted into KaLwRij mice, and the disease was allowed to establish for 3 weeks.Mice were randomised into groups, then treated for 3 weeks with bortezomib (0.5 mg/kg, ip, t.i.w.) or vehicle control.Disease burden was followed by bioluminescence imaging, and the mice were monitored until they reached ethical end-points.Consistent with the in vitro resistance to bortezomib of 5TGM1.BR E cells, bortezomib showed little effect on MM burden, mouse survival or MM-associated osteolytic bone disease in this 5TGM1.BR E /KaLwRij mouse model (Figure 2A-C).This contrasts with the anti-MM effects of bortezomib in KaLwRij mice engrafted with 5TGM1.wtcells (Figure 2D,E).Thus, the 5TGM1.BR E /KaLwRij mouse model appears suitable for the in vivo study of bortezomib-resistant MM.
Next, we assessed the utility of the 5TGM1.BR E /KaLwRij mouse model to evaluate therapeutic approaches to overcome bortezomib resistance in MM.For this, we initially performed RNAseq analysis on the 5TGM1.BR E cells to identify the mechanisms driving bortezomib resistance.Like 5TGM1.BR cells, 2 the 5TGM1.BR E cells retained an Ala79 → Gly mutation in the PSMB5 proteasome subunit, a site known to be mutated in some MM patients, 12 and which we previously demonstrated rendered the proteasome substantially less sensitive to inhibition by bortezomib. 2 Notably, the downregulation of MHC class I genes observed in 5TGM1.BR cells was dampened in 5TGM1.BR E cells (Figure 2F), and the effect was confirmed at the protein level (Figure S3), potentially explaining the lack of immune rejection of these cells.
Further examination of changes in gene expression revealed differences between 5TGM1.BR and 5TGM1.BR E cells (Figure S4), but dramatic loss of Igf2r expression was prominent in both 5TGM1.BR E cells (Figure 2G) and 5TGM1.BR cells. 2 IGF2R is a decoy receptor that dampens insulin-like growth factor (IGF) signalling, and thus, either loss of IGF2R or gain of IGF1R enhances IGF signalling. 13While not previously reported in MYC transgene-driven models of murine MM, [5][6][7] enhanced IGF signalling, through elevated IGF1R expression, has been previously implicated in bortezomib resistance in human MM cell lines, 3 and poor MM patient prognosis. 14Therefore, we investigated this further.Loss of IGF2R protein in 5TGM1.BR and 5TGM1.BR E cells was confirmed by immunoblot analysis, as was enhanced activation of IGF1R effectors AKT and PRAS40 (Figure 2H).Consistent with this, 5TGM1.BR E cells were somewhat more sensitive to the IGF1R/insulin receptor inhibitor, linsitinib, than 5TGM1.wt cells (Figure 2I).Since linsitinib also enhanced the sensitivity of 5TGM1.BR E cells to bortezomib in vitro (Figure 2J), we examined this in vivo using the 5TGM1.BR E /KaLwRij mouse model of MM.While bortezomib or linsitinib alone showed no significant attenuation of MM progression in this model, combining linsitinib and bortezomib caused a small, but significant increase in mouse survival compared to vehicle control or bortezomib alone (Figure 2K).This is consistent with the reported modest, but positive effects of linsitinib in combination with bortezomib and dexamethasone in a small phase I clinical trial in relapsed/refractory MM, 15 and suggests that this mouse model of bortezomib-resistant MM may be suitable for preclinical evaluation of new therapeutic approaches.
In conclusion, we have generated an orthotopic, syngeneic, bortezomib-resistant MM mouse model based on the most well-characterised syngeneic MM isograft mouse model derived from spontaneous MM-forming KaLwRij mice.Our data suggests this model is robust, reproducible and well suited to the evaluation of new therapeutic approaches for PI-resistant MM. .BR E cells) were treated for 3 weeks with bortezomib (0.5 mg/kg, ip, t.i.w.; indicated by the shaded area) or vehicle control.Disease burden was followed by bioluminescence imaging (A), mice monitored until they reached ethical end-points in Kaplan-Meier survival analysis (log-rank test) (B) and osteolytic bone disease assessed by micro-CT imaging of an intact mid-coronal section of the distal femur (C).This was compared to the response to bortezomib of the (bortezomib naive) 5TGM1.wt/KaLwRijmouse model of MM by treating mice with established disease (3 weeks after intravenous administration of 5TGM1.wtcells) with bortezomib for 3 weeks (0.5 mg/kg, ip, t.i.w.; indicated by the shaded area) or vehicle control.Disease burden was followed by bioluminescence imaging (D) (**p < 0.01), and mice were monitored until they reached ethical end-points in Kaplan-Meier survival analysis (log-rank test) (E).Gene expression analysis revealed differential expression of MHC class I related genes in 5TGM1.wt,5TGM1.BR and 5TGM1.BR E cell lines (**p < 0.01, ***p < 0.001, ****p < 0.0001) (F), with the volcano plot of differential gene expression between 5TGM1.wt and 5TGM1.BR E cells shown, indicating loss of Igf2r expression (G).Loss of IGF2R protein and downstream signalling was confirmed by immunoblot analysis of IGF2R, p-AKT and p-PRAS40 in 5TGM1.wt, 5TGM1.BR and 5TGM1.BR E cells (H).5TGM1.wtand 5TGM1.BR E cells were cultured in increasing concentrations of linsitinib for 48 h, and then assessed for cell viability by flow cytometry using Annexin-V and PI staining.Statistical significance was evaluated using a two-way ANOVA (I).5TGM1.BR E cells were cultured with increasing concentrations of bortezomib (BTZ) with and without 15 μM linsitinib (Lin) for 48 h, and cell viability was then assessed by flow cytometry using Annexin-V and PI staining (J).Dotted lines indicate IC 50 values for bortezomib, which linsitinib reduced from 29.9 ± 1 to 17.6 ± 0.8 nM.All the data shown represent the mean ± SD of three independent experiments.Combinational index (CI) values were calculated using CompuSyn, where a CI of <1 indicates synergy ( †). (K) Kaplan-Meier survival analysis (log-rank test) of the 5TGM1.BR E /KaLwRij mouse model after treatment with either vehicle, 0.5 mg/kg (ip, t.i.w.) bortezomib alone, 30 mg/kg (po, t.i.w.) linsitinib or a combination of both for 28 days (indicated by the shaded area).MM, multiple myeloma.

F I G U R E 2
Evaluation of the 5TGM1.BR E /KaLwRij mouse model of bortezomib-resistant MM.To assess in vivo bortezomib resistance in the 5TGM1.BR E /KaLwRij mouse model, mice with established disease (3 weeks after intravenous administration of 5TGM1