Identifying a Murine Xenograft Model Relevant to Light-Chain Specific Approaches to Human Ig Light-Chain Diseases

INTRODUCTION: Ig light chain (LC) diseases such as systemic AL amyloidosis (AL) and κ light chain deposition disease (LCDD) cause organ damage and are treated with chemotherapeutic agents targeting the clonal plasma cells producing the LC. Not all patients respond to therapy and responding patients remain at risk of relapse. New treatments are needed. In theory, alternative approaches that directly target LC for reduction include novel enzymatic agents (Infect Immun. 2003;71:2563), agents that retard LC monomerization (eLife. 2015;4:e10935. doi:10.7554), anti-light chain monoclonal antibodies (Blood 2004;104:2416), and small interfering RNAs that impair LC production (Gene Therapy 2016;23:727). In order to study such approaches, murine xenograft models with circulating human LC can provide useful experimental systems even if the models do not recapitulate the organ damage of human disease. Alternative measures of tumor activity would be useful in such models in order to confirm the impact of interventions on LC. To provide an in vivo model, we tested RPMI 8226, ALMC-1, NCI-H929 and JJN3 human myeloma reporter cell lines stably expressing FFL and GFP in NOD scid γ (NSG) mice using different routes of inoculation. We sought an optimal xenograft model that would provide reliable tumor-take, brief latency for circulating LC, rapid short-term increase in LC levels, measurable β2-microglobulin (β2M) levels and ease of administration of multiple injections of investigational agents.

MATERIALS & METHODS: Five-week old female NSG mice were obtained from Jackson Laboratories. RPMI 8226 (λ) and NCI-H929 (κ) human myeloma cell lines were obtained from ATCC (Manassas, VA, USA), and the JJN3 (κ) cell line from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). ALMC-1 cells (IgGλ+λ) were a gift from D Jellinek. All cells were tested for mycoplasma and cultured as directed. The cell lines were infected with a lentiviral construct expressing firefly luciferase and GFP (Biosettia, San Francisco, CA). Cells were introduced via flank injections (5x10⁶ cells) with Matrigel or via intra-peritoneal (IP) injection (10⁷ cells). Mice were regularly observed for health irregularities, were bled via submandibular vein and imaged with luciferin at intervals. Collected sera were tested by ELISA for λ or κ light chains (Bethyl Laboratories, Montgomery, TX) and for β-2 microglobulin (R&D Systems, Minneapolis, MN).

RESULTS: The latency periods for flank plasmacytomas exceeded 3 weeks in all cases. The use of ALMC-1 cells in an IP model also was hampered by a lengthy latency period > 3 weeks. We were able to compare JJN3, RPMI 8226 and H929 cells in IP models. As shown in Table 1,

the NSG JJN3 IP model has a 90% tumor-take and a 5-day LC latency. In this model, IP xenograft CD138+ cells are found in liver (subcapsular) and spleen at sacrifice on day 12. On day 5, the NSG JJN3 IP mice have median serum levels (Q1-Q3) of κ LC and β2M of 2.37μg/mL (1.68-3.32) and 1.56ρg/mL (0.58-5.38), values that strongly correlate (r=0.76, P<<0.01), as do the day 5 κ LC and FLUX values (r=0.88, P<<0.01). That RPMI 8226 cells make negligible amounts of β2M has been previously identified (Transplant Rev 1974;21:53).

CONCLUSIONS: The ALMC-1, RPMI 8226 and H929 IP models did not meet the criteria we required in order to effectively test specific interventions on circulating LC (Table 1). The NSG JJN3 IP model had a short latency period for LC, β2M and FLUX measurements, and had significant correlations among these measurements. The NSG JJN3 model met our criteria. We are currently evaluating soluble BCMA levels in this model also, These measurements will allow investigators to distinguish the impact of interventions on LC specifically, independent of tumor cells. We have used this model successfully to test an RNAi approach to reducing κ LC by knocking down expression of the LC mRNA using siRNA specific for the LC constant region delivered via lipidoid nanoparticle. The feasibility of multiple IP injections of volumes as high as 600uL is also a strength of this model; tail vein injections in the 20g NSG mice are difficult to perform more than a small number of times. The major challenge of this model is the rapid tumor take and growth kinetics of JJN3 cells IP in NSG mice.

View largeDownload slide

View largeDownload slide

Close modal