(T1330-01-03) Harnessing Dendritic Cell Migration as a Predictor of Subcutaneous Immunogenicity Risk in a Mechanism-Based Screening Tool

Purpose: The safety and efficacy of therapeutic proteins are undermined by immunogenicity or unwanted immune response driven by anti-drug antibodies (ADA). In particular, proteins injected subcutaneously can suffer from enhanced immunogenic potential compared to intravenous delivery. However, there is a lack of immunogenicity risk assessment methods designed for the prediction of immunogenic potential by subcutaneous (SC) delivery. SC injection introduces a high concentration of therapeutic protein into the skin where there are many skin-resident dendritic cells continuously surveying for antigens. Following capture of protein antigens, migration of these skin-resident dendritic cells from the injection site to draining lymph nodes drives cognate T-cell activation and subsequent humoral responses (Fig. 1) [1-4]. The ability of a therapeutic protein to induce dendritic cell (DC) migration is proposed to predict its immunogenic potential following SC injection. Here, DC migratory potential was investigated as a novel marker for immunogenicity screening.

Methods: Healthy HLA-typed PBMCs were purchased from Cytologics LLC and six donors were included to account for inter-individual variability in immunogenicity. Classical CD14⁺CD16^- monocytes were isolated from PBMCs by negative selection on a Miltenyi magnetic column. Monocytes were differentiated into HLA-DR⁺CD11c⁺DC-SIGN⁺ monocyte-derived DCs (moDC) by culturing with IL-4 and GM-CSF. Immature moDC were harvested on day five and plated with therapeutic proteins in media overnight. Control treatment groups were unstimulated/media, lipopolysaccharide (LPS), and keyhole limpet hemocyanin (KLH). moDC were stained for live/dead, HLA-DR, CD11c, CXCR4, and CD40, and then acquired on the BD LSRFortessa. The frequency of CXCR4⁺ moDC in each treatment group was determined in FlowJo (Fig. 2A). In the Transwell migration assay, immature moDC were plated in the upper chambers of 96-well Transwell plates while the lower chambers contained two chemokines in media. Then across the chambers, a low or high concentration gradient of therapeutic protein was established. Cells that migrated into the lower chambers were stained and counted by flow cytometry. Statistical analysis was performed in GraphPad Prism.

Results: Dendritic cell migratory potential was captured by two in vitro readouts: CXCR4⁺ frequency of moDC and Transwell migration index. First, the change in CXCR4⁺ frequency over the unstimulated treatment group was calculated as ‘Stimulation Index’ for each therapeutic protein. Monoclonal antibodies ATR-107 and HuA33 have high immunogenic potential, and in a population of six healthy donors, ATR-107 and HuA33 strongly upregulated CXCR4 on moDCs with an average stimulation index of 2.71 and 3.31, respectively (Fig. 2B). Also, the co-stimulatory marker CD40 was found to be concurrently upregulated with CXCR4 on moDC (Fig. 2C). Then, in the Transwell migration assay, the migration index was calculated as the fold change in percent migrated moDC toward therapeutic protein and chemokines compared to percent migrated moDC toward chemokines alone. A concentration gradient of ATR-107 and HuA33 induced a migration index of 3-fold and 5.5-fold, respectively (Fig. 2D). However, low to moderate immunogenic potential was predicted by a migration index at or below 2.0. Interestingly, for anti-HER2 IgG, which is co-formulated with hyaluronidase (rHuPH20) in an FDA-approved product, the migration index was slightly increased by the presence of rHuPH20 (Fig. 2E). Readouts of dendritic cell migratory potential were found to be protein-concentration dependent (Fig 3A). Finally, a strong positive correlation was established between the combined mean of assay readouts and the clinical ADA incidence (Fig. 3B).

Conclusion: The combined readouts of this in vitro screening tool captured dendritic cell migratory potential and correlated with immunogenic risk for multiple therapeutic proteins. Most notably, HuA33 and ATR-107, which demonstrated high incidence of immunogenicity (73-76%) in early clinical trials [5, 6], strongly upregulated all markers and induced migration of moDC along a concentration gradient. Results indicate that CXCR4 is indeed a mechanism-based marker to include in preclinical risk assessment. Furthermore, the Transwell migration assay has a flexible format allowing the testing of dendritic cell migration under many conditions. The migration index readout would be useful in immunogenicity risk assessment of formulation changes and novel modalities, such as gene therapy or nucleic acids.

References: 1. Jarvi, N.L. and S.V. Balu-Iyer, Immunogenicity Challenges Associated with Subcutaneous Delivery of Therapeutic Proteins. BioDrugs, 2021. 35(2): p. 125-146.
2. Kabashima, K., et al., CXCL12-CXCR4 engagement is required for migration of cutaneous dendritic cells. Am J Pathol, 2007. 171(4): p. 1249-57.
3. Ohl, L., et al., CCR7 governs skin dendritic cell migration under inflammatory and steady-state conditions. Immunity, 2004. 21(2): p. 279-88.
4. Worbs, T., S.I. Hammerschmidt, and R. Förster, Dendritic cell migration in health and disease. Nature Reviews Immunology, 2017. 17(1): p. 30-48.
5. Hua, F., et al., Anti-IL21 receptor monoclonal antibody (ATR-107): Safety, pharmacokinetics, and pharmacodynamic evaluation in healthy volunteers: A phase I, first-in-human study. The Journal of Clinical Pharmacology, 2014. 54(1): p. 14-22.
6. Welt, S., et al., Phase I study of anticolon cancer humanized antibody A33. Clin Cancer Res, 2003. 9(4): p. 1338-46.


Figure 1. Following subcutaneous injection of therapeutic protein, the immunogenic response is driven by skin-derived dendritic cell migration into the injection site and toward the draining lymph node. At which point migratory dendritic cells will induce strong cognate T-cell activation.


Figure 2. CXCR4 induction on human monocyte-derived dendritic cells by therapeutic protein correlates with its immunogenicity potential. (A) Flow cytometry gating strategy for live CD11c+ moDC. (B) CXCR4 and (C) CD40 stimulation index values in six healthy donors for eight protein antigens (mean ± SEM). Each dot represents one donor; treatments were tested in triplicate for each donor. Unpaired student’s t-test. (D) Migration index toward a low protein concentration gradient for six protein antigens in multiple donors (mean ± SEM). Migration index equals the percent migrated DC for each treatment group divided by the average percent migrated DC for the unstimulated group. Treatments were tested in triplicate for each donor. Ordinary one-way ANOVA with Tukey’s multiple comparisons test. (E) Migration index toward a high protein concentration gradient for three protein antigens in two donors. Unpaired student’s t-test. * p < 0.05, ** p < 0.01, *** p < 0.001.


Figure 3. In vitro readouts from the screening tool are protein-concentration dependent and correlate positively with clinical immunogenicity incidence. (A) (Left y-axis) Frequency of CXCR4+ moDC as a function of KLH concentration for one donor (mean ± SEM). (Right y-axis) Transwell migration index as a function of KLH concentration in the lower chamber for two donors (mean ± SEM). Treatments were tested in triplicate for each donor. (B) In vitro assay readouts versus the clinical ADA incidence for each therapeutic protein tested, labeled A-H. The y-axis represents the mean of three assay readouts in all donors (1-CXCR4 stimulation index, 2-CD40 stimulation index, and 3-Transwell migration index). Error bars are mean ± SEM. The x-axis represents the highest reported clinical ADA incidence in US package inserts or literature. The best-fit line is shown for A-F (circles) with the 95% confidence interval. G and H (triangles) follow the data trend but fall outside of this 95% confidence interval. A – Anti-IL6R IgG, B – Bispecific IgG4, C – Anti-CD20 IgG, D – Anti-HER2 IgG, E – ATR-107, F – KLH, G – Anti-TNF IgG, and H – HuA33.

Acknowledgements: Funding was provided by the UB Center for Protein Therapeutics. The authors would like to acknowledge the UB Optical Imaging and Analysis Facility for use of the flow cytometer.