IL-1 receptor antagonism reveals a yin-yang relationship between NFκB and interferon signaling in chronic lymphocytic leukemia

Abstract

Nuclear factor kappa B (NFκB) is a pathogenic factor in chronic lymphocytic leukemia (CLL) that is not addressed specifically by current therapies. NFκB is activated by inflammatory factors that stimulate toll-like receptors (TLRs) and receptors for interleukin-1 (IL-1) family members. IL-1 is considered a master regulator of inflammation, and IL-1 receptor signaling is inhibited by the IL-1 receptor antagonist anakinra. These considerations suggested that anakinra might have a role in the treatment of CLL. Consistent with this idea, anakinra inhibited spontaneous and TLR7-mediated activation of the canonical NFκB pathway in CLL cells in vitro. However, CLL cells exhibited only weak signaling responses to IL-1 itself, and anakinra was found to inhibit NFκB along with oxidative stress in an IL-1 receptor-independent manner. Anakinra was then administered with minimal toxicity to 11 previously untreated CLL patients in a phase I dose-escalation trial (NCT04691765). A stereotyped clinical response was observed in all patients. Anakinra lowered blood lymphocytes and lymph node sizes within the first month that were associated with downregulation of NFκB and oxidative stress in the leukemia cells. However, inhibition of NFκB was accompanied by upregulation of type 1 interferon (IFN) signaling, c-MYC-regulated genes and proteins, and loss of the initial clinical response. Anakinra increased IFN signaling and survival of CLL cells in vitro that were, respectively, phenocopied by mitochondrial antioxidants and reversed by IFN receptor blocking antibodies. These observations suggest that anakinra has activity in CLL and may be a useful adjunct for conventional therapies as long as compensatory IFN signaling is blocked at the same time.

Keywords: IL1 receptor antagonist; chronic lymphocytic leukemia; interferon; signal transduction.

Fig. 1.

Effect of anakinra on NFκB in CLL cells in vitro. (A) RNAseq was performed on CLL cells from 3 patients cultured with or without anakinra for 24 h. The enrichment plot indicates downregulation of NFκB-activated genes by anakinra. (B–D) CLL cells from untreated patients were cultured for 24 h with or without resiquimod (S) at 1 μg/ml (B) or resiquimod and IL-2 (2S) at 500 IU/ml (C and D) in the presence and absence of anakinra (K). (B) Phospho-p65 was quantified in lysates collected at the indicated times by immunoblotting and densitometry using β-actin as a loading control. A sample blot is shown above the summary graph, indicating averages and SE for 7 patient samples. The averages and SEM at each time point (S vs S + K) are 0.848±0.111 vs 1.308 ± 0.220; 1.167 ± 0.161 vs 1.177 ± 0.263; 1.586 ± 0.103 vs 1.263 ± 0.125; 1.174 ± 0.174 vs 0.822 ± 0.133; 1.235 ± 0.159 vs 0.702 ± 0.163; 1.087 ± 0.075 vs 0.484 ± 0.105. (C) The lines (N = 10) indicate TNFα in culture supernatants measured by the ELISA for single patient samples. (D) Cell surface CD83 was measured by flow cytometry (N = 6). Examples of histograms are shown, with the numbers representing differences between percentages of cells stained with CD83 antibody (red histogram) or control antibodies (gray histogram). (E and F) p105/50 and p100/52 were measured in CLL cells cultured alone (E) or with TNFα (10 ng/ml) or trimeric CD40L (1 μg/ml) to activate canonical and noncanonical NFκB, respectively (F) with or without anakinra. Examples are shown. Averages and SE for (E) are p105: 0.948 ± 0.146 vs 0.821 ± 0.119, p50: 1.187 ± 0.125 vs 0.947 ± 0.105 (N = 19), p100: 0.6 ± 0.121 vs 0.355 ± 0.132, and p52: 0.404 ± 0.121 vs 0.48 ± 0.096 (N = 16) and for (F) are p105: 1.090 ± 0.203 vs 0.824 ± 0.198, p50: 1.125 ± 0.186 vs 0.713 ± 0.138 (N = 6), p100: 0.934 ± 0.182 vs 0.980 ± 0.121, and p52: 1.458 ± 0.176 vs 1.158 ± 0.186 (N = 8). Each closed circle in the summary graphs indicates results for a single sample. Statistical significance was determined with Student’s t tests and one-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.

Fig. 2.

Localization of anakinra and effect on oxidative stress in CLL cells. (A) IL-1α (1.24 ± 1.006) (N = 3 0), IL-1β (0.997 ± 0.158) (N = 34), and IL-1Ra (98.475±28.467) (N = 28) were measured in supernatants of CLL cells after 48 h. Patient numbers differ because samples were assayed in different batches. Each circle is the result for an individual sample. Averages and SE are shown. (B) CLL cells were stimulated with IL-1β or resiquimod (S) (N = 3). Phospho-p38 was quantified after 30 min by immunoblotting and densitometry with β-actin as a loading control. A representative blot is shown. Individual patient samples in the graph are indicated by dots. Averages and SE are Con: 0.33 ± 0.201, IL-1β: 0.264 ± 0.003, and S: 3.497 ± 0.458. (C) CLL cells precultured with or without neutralizing IL-1 receptor antibodies for 2 h before adding IL-2 and resiquimod (2S) in the presence or absence of anakinra (K) for 48 h were analyzed by confocal microscopy as described in the materials and methods. Representative images are shown, with similar results obtained with 5 different patient samples. (D) IL-1Ra-staining quantified for 5 images per treatment per patient sample is shown in the summary graph, with each dot representing a different image. Averages and SE of mean fluorescence intensities (MFIs) are Con: 91.407 ± 3.589, 2S:169.692 ± 38.945, 2S+K:1263.622 ± 145.864, and 2S+K + anti-IL-1RI: 1741.328 ± 208.275. (E) CLL cells (N = 7) were cultured with or without anakinra (K) for 48 h and stained with DCFH to measure intracellular ROS levels. Representative histograms (Left) and MFIs of DCFH staining for individual patient samples (Right) are shown. Statistical significance was determined using Student’s t tests. *P < 0.05; **P < 0.01; ns, not significant.

Fig. 3.

Effect of anakinra on initial clinical response, NFκB signaling, and oxidative stress in vivo. (A and B) Lymphocyte count (A) and a lymphadenopathy score defined as the sum of the products of bidimensional measurements of the largest palpable nodes plus spleen length in cm (B) were recorded before (C1D1) and 2 or 4 wk (C1D14 or C2D1) after starting anakinra. Results for each patient are shown by the lines in the graphs, using the lowest value at C1D14 or C2D1 (N = 10). C, D. Differentially expressed genes in CLL cells purified from the blood of 6 patients at C1D1 and C1D14 were compared by GSEA. The enrichment plots indicate downregulation of NFκB response genes (C) and upregulation of antioxidant genes (D) in CLL cells exposed to anakinra in vivo. (E) An antioxidant index was obtained from the RNAseq data by using the number of reads mapped per gene per million reads mapped (RPM) for HMOX1, HMOX2, SOD1, PRDX3, and PRDX4. The sum of the value for each gene normalized to the C1D1 value was calculated for each patient. The average sums and SE for all 6 patients were 5 ± 0 at C1D1 and 5.302 ± 0.111 at C1D14. (F) HMOX1 and HMOX2 transcripts in CLL cells cultured with and without anakinra (K) were measured after 24 h by RT-PCR (N = 12). Circles in the graph represent results for individual patients. Averages and SE were 4.334 ± 1.204 vs 5.708 ± 1.452 for HMOX1 and 6.485 ± 1.258 vs 13.993 ± 3.477 for HMOX2. P-values were determined using paired Student’s t tests. *P < 0.05; **P < 0.01; ns, not significant.

Fig. 4.

Effect of anakinra on IFN responses in CLL cells in vivo and in vitro. (A) DEGs in circulating CLL cells from six patients at C1D1 and C1D14 were compared by GSEA. The enrichment plot indicates that anakinra up-regulated IFN-stimulated gene expression. (B) Phospho-STAT2 was measured in protein extracts from cryopreserved CLL cells obtained at C1D1 and C1D14 (N = 5) by immunoblotting and densitometry using β-actin as a loading control. (C and D) CLL cells were cultured with and without anakinra (K). Phospho-STAT2 (exposure time 1 min) was measured after 24 h (N = 11) (C). CXCL10 was measured by the ELISA after 48 h (N = 10) (D). (E–G) CLL cells were cultured with or without anakinra for 24 h and then stimulated with IFNβ. E. OASL transcripts were measured after 4 h (N = 11). Averages and SE are Con: 0.219 ± 0.076, K: 0.386 ± 0.117, IFNβ: 7.41 ± 1.617, K + IFNβ: 10.115 ± 2.041. (F) Phospho-STAT1 and -STAT2 (exposure time 5 s) were measured after 30 min (N = 7). Averages and SE are Con: 0.09 ± 0.032, IFNβ: 1.271 ± 0.197, K: 0.098 ± 0.035, K + IFNβ: 2.348 ± 0.244 for P-STAT1 and Con: 0.063 ± 0.024, IFNβ: 1.161 ± 0.254, K: 0.058 ± 0.018, K + IFNβ: 2.395 ± 0.456 for P-STAT2. (G) LAG3 was measured after 24 h. A representative example is shown in Supplementary Fig. S6. Percentages of LAG3⁺ cells in each condition normalized to the result for IFN-treated cells without anakinra are plotted in the summary graph. Examples of immunoblots are shown, and each line or dot in the graphs represents results for an individual patient. The means and SEMs normalized to IFNβ are Con: 0.251 ± 0.045, K: 0.327 ± 0.117, IFNβ: 1 ± 0, and IFNβ+K: 1.183 ± 0.045. (H and I) CLL cells were cultured in RPMI-1640 in the absence (H) or presence (I) of IFNβ with or without K, anifrolumab (ANI), or both (N = 14). Live (DAPI⁻) cells were measured after 4 d. Examples of contour plots are shown, with numbers indicating % DAPI⁻ cells. Each dot in the summary graphs represents an individual patient. (H) Averages and SE of % live cells are Con: 29.666 ± 1.591, K: 38.369 ± 2.546, ANI: 30.65 ± 1.733, and K + ANI: 31.428 ± 2.104. (I) Averages and SE are Con: 60.839 ± 5.642, K + IFNβ: 78.786 ± 4.617, K + IFNβ+ANI: 51.804 ± 6.57. Statistical analysis was performed with Student’s t tests and one-way ANOVAs. *P < 0.05; **P < 0.01; ****P < 0.0001; ns, not significant.