FcRn-silencing of IL-12Fc prevents toxicity of local IL-12 therapy and prolongs survival in experimental glioblastoma

Abstract

Glioblastoma remains a challenging indication for immunotherapy: the blood-brain barrier hampers accessibility for systemic treatments and the immunosuppressive microenvironment impedes immune attack. Intratumoral therapy with the proinflammatory cytokine interleukin-12 (IL-12) can revert immunosuppression but leakage into the circulation causes treatment-limiting toxicity. Here we engineer an IL-12Fc fusion cytokine with reduced binding to the neonatal Fc receptor FcRn. FcRn-silenced IL-12Fc avoids FcRn-mediated brain export, thus exhibits prolonged brain retention and reduced blood levels, which prevents toxicity. In murine glioblastoma, FcRn-silenced IL-12Fc induces more durable responses with negligible systemic cytokine exposure and boosts the efficacy of radio- and chemotherapy. It triggers anti-tumor responses independently of peripheral T cell influx or lymphopenia and leads to inflammatory polarization of the tumor microenvironment in patient-derived glioblastoma explants. FcRn-silencing of IL-12Fc may unlock the full potential of IL-12 for brain cancer therapy and could be further applied to containing the activity of other therapeutics targeting neurological diseases.

Fig. 1. Bivalent single-chain Fc-fusion reduces plasma prevalence of IL-12, but increased FcRn abundance in glioblastoma (GBM) may counteract tissue retention.

A Cartoon of IL-12 (Protein Data Bank entry 1F45) and predicted structure of IL-12Fc. B Size exclusion chromatogram (normalized absorbance at 280 nm; A₂₈₀) and C SDS-PAGE of IL-12Fc protein in native (lane 1) and reducing conditions (lane 2). Lanes cropped from a single image. Data representative of three independent experiments. D–F 1 μg of rIL-12 or IL-12Fc was administered into the right striatum of hFcRn Tg32 mice via convection-enhanced delivery (CED). IL-12 brain levels in the ipsilateral brain hemisphere (E) and in plasma (F) 6 h after injection. Unpaired two-tailed t-test. Mean ± SD. IL-12Fc WT: n = 3 (green), rIL-12: n = 3 animals (black). G FcRn protein abundance in human brain tissue of non-related disease vs glioma according to the dataset from ref. . Unpaired two-tailed t-test. Mean ± SD. Control: n = 5; glioma: n = 6 patients. H Relative expression of Fcgrt in GL-261 tumor-bearing hemisphere (day 21 post-implantation) and corresponding healthy contralateral hemisphere plotted as fold change of Fcgrt to reference gene Hprt (2^−ΔCt). n = 10 animals/group. Paired two-tailed t-test. Mean ± SD. Source data are provided as a Source Data file.

Fig. 2. Silencing FcRn binding increases tissue retention and reduces peripheral exposure of IL-12Fc.

A Binding of modified IL-12Fc to FcRn at pH = 6.0 was determined by surface plasmon resonance (SPR) with hFcRn in the immobilized phase. Signal response (RU) over time for IL-12Fc WT (left) or IL-12Fc NHQ (right). Curves relate to concentrations (nM, top to bottom): IL-12Fc WT: 365 – 122 – 41 – 14 – 5; IL-12Fc NHQ: 3645 – 1215 – 405 – 135 – 45. B Dissociation constant KD (nM) of IL-12Fc WT (IHH) and modified variants as well as control antibodies, calculated from SPR equilibrium analysis (Fig. S2C). C IL-12 levels in the ipsilateral hemisphere 6 h after intraparenchymal administration of 1 μg of the indicated IL-12Fc variants via CED into hFcRn Tg32 mice. Mean ± SD. Outlier removal based on Grubbs’ test. One-way ANOVA with Tukey’s multiple comparison test. IL-12Fc WT: n = 9, IL-12Fc IAQ: n = 8, IL-12Fc AAA: n = 9, IL-12Fc NHQ: n = 9 animals. D IL-12 levels in the ipsilateral hemisphere 6 h after intraparenchymal administration of 1 μg of the indicated IL-12Fc variants via CED into hFcRn Tg32 mice treated intravenously (i.v.) with 100 μg of the FcRn inhibitor MST-HN Fc. Mean ± SD. Unpaired two-tailed t-test of groups IL-12Fc NHQ to IL-12Fc WT group with (p = 0.0177) and without FcRn inhibitor (non-significant, p = 0.8663). IL-12Fc WT: n = 5, IL-12Fc NHQ: n = 5, IL-12Fc WT + MST-HN Fc: n = 8, IL-12Fc NHQ + MST-HN Fc: n = 8 animals. E Extended pharmacokinetic (PK) analysis. Indicated IL-12 variants were administered to hFcRn Tg32 mice (n = 7–12 animals/timepoint/group; detailed group sizes in Table S2) and analyzed as in C). Mean concentration of IL-12 in brain (top) and plasma (bottom) across the next 7 days. Outlier removal based on Grubbs’ test. Separate curves with individual data points in Fig. S2I, J. One-way ANOVA with Dunnett’s test IL-12Fc NHQ vs IL-12Fc WT and rIL-12. Statistically significant difference only to rIL-12. IL-12Fc WT: green; IL-12Fc NHQ: orange; rIL-12: black. F Overall exposure (area under curve, AUC) analysis for brain (top) and plasma (bottom) in E) until day 3. Mean ± SEM. Unpaired two-tailed t-test IL-12Fc NHQ vs IL-12Fc WT and rIL-12 corrected for multiple testing using Bonferroni correction. Source data are provided as a Source Data file.

Fig. 3. Increased therapeutic index of IL-12Fc NHQ in murine GBM.

A GL-261:luc cells were injected into the right striatum of WT animals. At day 20, animals were allocated to treatment cohorts with comparable tumor load as assessed by in vivo bioluminescent imaging (BLI), and 1 µg of IL-12 variants was administered intratumoral (i.t.) via CED the next day. Blood samples were taken on day 21 before CED and 6 h after, then on day 22, 24, and 28 (before CED and 6 h after), as well as on days 29, 31, 35, and 49. B Representative frontal overview of a treated GL-261:luc tumor on day 21. Animals were sacrificed 5 min after i.t. administration of mIL-12hFc NHQ as described in A. Nuclear stain (left) and mIL-12hFc NHQ infusate (right). Scale bar = 0.5 mm. C Kaplan–Meyer curves of mice described in A. Log-rank (Mantel–Cox) test. Data pooled from three independent experiments. Control (gray): n = 16, rmIL-12 (black): n = 14, mIL-12hFc WT (green): n = 18, mIL-12hFc NHQ (orange): n = 18 animals. D BLI traces (average radiance) of individual animals described in A, treated with rmIL-12, mIL-12hFc WT, mIL-12hFc NHQ, or buffer control. Dashed vertical lines indicate CED. E–J Systemic cytokine levels of individual animals treated as described in A, C: IL-12 (E–G) and IFNγ (H–J) assessed by multiplex bead assay. E, H Plotted plasma concentration by days post-CED. Control: n = 16, rmIL-12: n = 14, mIL-12hFc WT: n = 18, mIL-12hFc NHQ: n = 18 animals. Data pooled from C, D. Dashed horizontal lines depict the change in the axis scale between the plots for rmIL-12 and other experimental cohorts. Peak values of plasma cytokine levels (F) IL-12 (6 h) and (I) IFNγ (24 h; 72 h for rmIL-12 group). Mean ± SD. Control: n = 10, mIL-12hFc WT: n = 10, mIL-12hFc NHQ: n = 10; rmIL-12: n = 8 animals. One-way ANOVA with Dunnett’s test mIL-12hFc NHQ to mIL-12hFc WT, rmIL-12, and control groups. Overall exposure (AUC) of G IL-12 or J IFNγ, expressed as fold change to control group. Mean ± SD. One-way ANOVA with Dunnett’s test mIL-12hFc NHQ to mIL-12hFc WT, rmIL-12, and control groups. Data pooled from two (E, F, H, I) or three (G, J) independent experiments from C. TNFα and IFNγ positive effector CD4⁺ T (K), Tregs (L), and CD8⁺ T cells (M) in the tumor-bearing hemisphere on day 28, as analyzed by flow cytometry. Diameters of charts correspond to the fold change in cell counts compared to the control group. Gating strategy depicted in Fig. S5A. Data pooled from two independent experiments. Source data are provided as a Source Data file.

Fig. 4. C3H/HeJ IFNγ-sensitive mice tolerate high intraparenchymal doses of mIL-12hFc NHQ without systemic toxicity.

mIL-12hFc NHQ or rmIL-12 were intraparenchymally administered via CED at two dose levels (1 or 5 µg) on days 0 and 7. A Weight of individual animals in each cohort, normalized to starting weight. B Spleen weight. Animals that reached withdrawal criteria before the planned endpoint are not included. Unpaired two-tailed t-test of rmIL-12 vs mIL-12hFc NHQ groups at each dose level. Mean ± SD. Control: n = 4, rmIL-12 5 μg: n = 2, rmIL-12 1 μg: n = 4, mIL-12hFc NHQ 5 μg: n = 4, mIL-12hFc NHQ 5 μg: n = 4 animals. Systemic IL-12 (C, D), IFNγ (E, F), and CCL-2 (G, H) concentrations of individual animals assessed by multiplex bead assay in plasma. Dotted vertical lines indicate CED. Dashed horizontal lines depict the change in the axis scale between the plots for rmIL-12 5 µg and other experimental cohorts. Overall exposure (AUC) was analyzed by an unpaired two-tailed t-test of rmIL-12 vs mIL-12hFc NHQ groups at each dose level. Mean ± SD. Control: n = 4, rmIL-12 5 μg: n = 4, rmIL-12 1 μg: n = 4, mIL-12hFc NHQ 5 μg: n = 4, mIL-12hFc NHQ 5 μg: n = 4 animals. I Histopathological scores of animals in A, B as assessed by a board-certified veterinarian pathologist and normalized to the average of the control group. BALT bronchus-associated lymphoid tissue. Source data are provided as a Source Data file.

Fig. 5. FcRn-silenced IL-12Fc synergizes with chemotherapy and RT, even in models of aggressive treatment-resistant glioma.

A–G GL-261 brain tumor-bearing C57BL/6 mice were treated with mIL-12hFc NHQ as in Fig. 3 and with five daily doses of temozolomide (TMZ) between days 23 and 27, either alone or in combination. B Kaplan–Meyer curves for the experiment described in A treated with buffer control, TMZ, mIL-12hFc NHQ, or mIL-12hFc NHQ + TMZ. Log-rank (Mantel–Cox) test. Data pooled from three independent experiments. Control (gray): n = 16, TMZ (black): n = 16, mIL-12hFc NHQ (orange): n = 19, mIL-12hFc NHQ + TMZ (green): n = 18 animals. C–G: same legend as in B. C Flow cytometric quantification of blood CD4⁺ T cells, CD8⁺ T cells, NK cells, and B cells on day 28. Mean ± SD. One-way ANOVA with Dunnett’s test vs control. Representative data from one of the experiments contributing to B. D–F Tumor-infiltrating lymphocytes (TILs), analyzed by flow cytometry at day 32, 4 days after the second CED. Gating strategy depicted in Fig. S7B. One-way ANOVA with Dunnett’s test to the control group. Mean ± SD. D Total number of CD4⁺ T effector cells and CD8⁺ T cells, as well as E percentage of CD69⁺ PD-1⁺ among total CD8⁺ T cells, F gMFI of Ki67 on CD4⁺ T effector cells, G gMFI of MHCII on microglia and myeloid infiltrates. H–L SB28 cells were injected into the right striatum of WT animals. On day 4, animals were allocated to treatment cohorts with comparable tumor load as assessed by BLI, and 1 µg of mIL-12hFc NHQ was administered intratumorally by CED. Mice received two doses of 10 Gy radiation (RT) targeted to the tumor location on day 7 and day 9. I Frontal overview over treated SB28 tumor (day 5 post-inoculation), 5 min after i.t. administration of mIL-12hFc NHQ via CED. Nuclear stain (left), SB28 tumor (middle), and mIL-12hFc NHQ shown (right). Scale bar: 0.5 mm. J Kaplan–Meyer curves of cohorts described in H treated with buffer control, RT, mIL-12hFc NHQ, or mIL-12hFc NHQ + RT. Log-rank (Mantel–Cox) test compared to controls. Data pooled from three independent experiments. Control (gray): n = 9, RT (black): n = 9, mIL-12hFc NHQ (orange): n = 9, mIL-12hFc NHQ + RT (green): n = 9 animals. K, L TILs were analyzed by flow cytometry at day 16, 4 days after the second CED. One-way ANOVA with Dunnett’s test vs control. Gating as shown in Fig. S9C. Same legend as in (J). Total number of K CD4⁺ T effector cells, CD8⁺ T cells, CD4⁺ PD-1⁺ and CD8⁺ PD-1⁺ T cells as well as L gMFI of MHCII on microglia and myeloid infiltrates in the tumor-bearing hemisphere. Mean ± SD. Source data are provided as a Source Data file.

Fig. 6. Proinflammatory conditioning of FcRn-silenced IL-12Fc-treated patient-derived GBM explants with intact tumor microenvironment.

A Scheme of an in vitro blood-brain barrier (BBB) model formed by a brain microvascular endothelial cell (BMEC) monolayer in a transwell insert, derived from human induced pluripotent stem cells (iPSCs). B Transport rates of IL-12 variants across a BMEC monolayer in basolateral (brain) to apical (blood) direction. One-way ANOVA with Dunnett’s test to the IL-12Fc NHQ group. Outlier removal based on Grubbs’ test. rIL-12: n = 6, IL-12Fc WT: n = 6, IL-12Fc NHQ: n = 5 experiment repetitions. Whiskers represent the minimum to maximum, middle line represents the mean. C Tumor explants from nine patients with newly diagnosed GBM were trimmed, divided into parallel perfusion bioreactor cultures, and treated with 100 ng/mL IL-12Fc NHQ or control medium for up to 7 days. Supernatants were analyzed using multiplex Proximity Extension Assay (PEA Olink Target 96 IO). D Differentially expressed proteins in supernatants of bioreactors treated with IL-12Fc NHQ compared to control medium. Values with a false discovery rate (FDR) < 0.05 (horizontal red line) are considered significant. The vertical red line marks a fold change of 2. E Barcode plot of the hallmark IFNγ gene set. The vertical bars in the lower part of the plot mark the t-statistic of those genes which are in the gene set against the background expression of the remaining genes in the panel. F Sign plots of normalized protein expression (NPX) at day 7, comparing IL-12Fc NHQ vs control-treated bioreactors. The middle line marks the sample median, the hinges correspond to the first and third quartiles, and the whiskers extend up to a further ±1.5 times the length of said inter-quartile range. All protein level p values are FDR-adjusted and are derived from moderated t-statistics. n = 9 patients/group. Source data are provided as a Source Data file.