Fc-modified HIT-like monoclonal antibody as a novel treatment for sepsis

Abstract

Sepsis is characterized by multiorgan system dysfunction that occurs because of infection. It is associated with high morbidity and mortality and is in need of improved therapeutic interventions. Neutrophils play a crucial role in sepsis, releasing neutrophil extracellular traps (NETs) composed of DNA complexed with histones and toxic antimicrobial proteins that ensnare pathogens, but also damage host tissues. At presentation, patients often have a significant NET burden contributing to the multiorgan damage. Therefore, interventions that inhibit NET release would likely be ineffective at preventing NET-based injury. Treatments that enhance NET degradation may liberate captured bacteria and toxic NET degradation products (NDPs) and likely be of limited therapeutic benefit as well. We propose that interventions that stabilize NETs and sequester NDPs may be protective in sepsis. We showed that platelet factor 4 (PF4), a platelet-associated chemokine, binds and compacts NETs, increasing their resistance to DNase I. We now show that PF4 increases NET-mediated bacterial capture, reduces the release of NDPs, and improves outcome in murine models of sepsis. A monoclonal antibody KKO which binds to PF4-NET complexes, further enhances DNase resistance. However, the Fc portion of this antibody activates the immune response and increases thrombotic risk, negating any protective effects in sepsis. Therefore, we developed an Fc-modified KKO that does not induce these negative outcomes. Treatment with this antibody augmented the effects of PF4, decreasing NDP release and bacterial dissemination and increasing survival in murine sepsis models, supporting a novel NET-targeting approach to improve outcomes in sepsis.

Graphical abstract

Figure 1.

Effects of murine and human PF4 on circulating NDP levels following LPS exposure in WT and cxcl4^−/−mice. (A) WT and cxcl4^−/− mice received LPS (35 mg/kg, IP). Plasma samples were obtained at the indicated time points. Mean plasma cfDNA levels are shown as ± 1 standard deviation (SD). N = 7-14 mice per arm. P values are indicated comparing WT and cxcl4^−/− mice using a Mann-Whitney U test. (B) The same as panel A, except for MPO levels at 3 to 8 hours post-LPS. (C) Histone levels comparing western blot intensity to that of the positive control band. N = 9-10 mice per arm. P values are indicated comparing WT and cxcl4^−/− mice by Mann-Whitney U test. (D) LPS studies comparing WT and cxcl4^−/− mice following implantation of osmotic pumps containing PBS alone or 80 µg of mPF4. Mean ± 1 SD are shown N = 5-8 mice per arm. (E) The same as panel D, except following a single dose of hPF4 (330 µg, IV). N = 5-8 mice per arm. (F) LPS studies comparing plasma levels of MPO-cfDNA complex in cxcl4^−/− mice that received tail vein injections containing PBS alone or 330 µg of hPF4. N = 4-5 mice per arm. P values are indicated comparing WT and cxcl4^−/− mice by Mann-Whitney U test.

Figure 2.

Effect of PF4 on endothelial cells and microbial entrapment by NETs in vitro. Channels lined with TNF-α–stimulated HUVECs were infused with isolated human neutrophils treated with LPS (100 ng/mL) to induce NETosis with and without hPF4 (25 µg/mL). (A) Channels were incubated with the neutrophils for 16 hours, after which the number of residual adherent endothelial cells was counted using ImageJ. (Top) Representative images of remaining attached endothelial cells channels per condition. Size bar and arrows indicate direction of flow. Bottom: mean of endothelial cell counts in 3 ×10 high-powered fields per condition ± 1 SD. Statistical analysis was performed using a Mann-Whitney U test. n = 6 channels per arm. (B) Left shows representative images of NET-lined channels infused with fluorescently labeled S aureus with observed bacterial capture. Size bar and arrows indicating direction of flow are included. NETs in bottom channels compacted with PF4 (100 μg/mL). (C) Representative image of NET-lined channels previously infused with bacteria, following 30-minute infusion of heparin 100 U/mL. (D) The same as panel C, but includes images of NET-lined channels infused with S aureus bioparticles followed by a 5-minute infusion of DNase I (100 U/mL). NETs in bottom channels compacted with PF4 (100 μg/mL). (E) The same as panel C showing image of S aureus–infused channels previously treated with heparin, now following infusion with DNase 1 (100 U/mL). (F) Graph showing number of NET-adherent bacterial bioparticles in channels ± 1 SD following the initial infusion, DNase I (100 U/mL) × 15 minutes, or heparin (100 U/mL) × 30 minutes as indicated. N = 3-15 channels per condition. Analysis performed by a Kruskal-Wallis 1-way ANOVA.

Figure 3.

Binding of DG-KKO to PF4/NET complexes in vitro. (A) Graphs quantifying binding of increasing concentrations of KKO (gray) and DG-KKO (red) to PF4-heparin (PF4-H) using fluorescent plate assay. (B) The same as panel A but using flow cytometry to quantify antibody binding to the platelet surface. (C) Mean ± 1 SD of P-selectin MFI in human whole blood samples incubated with the indicated concentration of antibody, reflecting the degree of platelet activation. N = 3. (D) Mean ± 1 SD of the % decrease in platelets counts in HIT mice injected with 400 µg of the indicated antibody, measured every 12 hours for 3 days. N = 10. (E) Representative confocal images of released NETs as in Figure 2 exposed to no PF4 or 6.5 µg/mL of PF4, labeled with the nucleic acid stain SYOTX green (green) demonstrating change in morphology. The indicated channels were then infused with fluorophore-labeled DG-KKO (white). Size bar and arrows indicating direction of flow are included. Image were obtained at ×10 magnification. (F) Representative widefield images of adherent NETs as in panel A, but in the presence of 100 U/mL DNase I and 6.5 µg/mL PF4 ± 25 µg/mL of DG-KKO. (G) Mean ± 1 SD of the relative area of NETs compacted with PF4 (6.5 µg/mL) alone or with PF4 plus DG-KKO or a polyclonal anti-PF4 antibody control (Ctl) (each, 25 µg/mL) after an infusion of DNase I (100 U/mL, 3 minutes) compared with preinfusion area. N = 7-10 channels per condition. Comparative statistical analysis performed by Kruskal-Wallis 1-way ANOVA.

Figure 4.

DG-KKO treatment in WT, cxcl4^−/−, and hPF4⁺mice undergoing LPS endotoxemia. Mice were injection with LPS (35 mg/kg, IP). Thirty minutes later, they received tail vein injections containing vehicle alone or 5 mg/kg of DG-KKO or DG-TRA isotype control or DG-KKO plus DNase I (20 mg/kg). Six hours following LPS injection, a subset of mice was euthanized and IVC blood samples were collected. (A) Relative to baseline (dashed gray line), 6-hour time point platelet counts. Mean ± 1 SD shown. (B-E) The same as panel A, but for cfDNA concentration, MPO-cfDNA complex fold change relative to negative control, MCP-1 levels as measured by fold change in western blot bandwidth relative to GAPDH control, and MSS. N = 3-10 in each arm. Comparative statistical analysis was performed with a Kruskal-Wallis 1-way ANOVA. (F) Animal survival results for the LPS studies in the cxcl4^−/− mice and hPF4⁺ mice. Results were analyzed with the log-rank test. n = 5 animals per arm.

Figure 5.

DG-KKO treatment in cxcl4^−/−and hPF4⁺mice undergoing CLP injury. All the mice underwent CLP procedure. Immediately following surgery, a subset of mice received an intradermal dose of the antibiotic ceftriaxone (100 mg/kg). Mice also were divided by therapeutic intervention, receiving either vehicle only or 5 mg/kg of TRA, DG-TRA, KKO, or DG-KKO. After 24 hours, platelet counts were quantified. After 48 hours, bacterial CFUs, plasma NDP levels, and survival were measured. Mean ± 1 SD shown. (A) Relative platelet counts 24 hours after CLP injury measured as in Figure 4A. (B) CFUs as measured in the peripheral blood and liver homogenates of animals 48 hours following CLP. Black bars indicate animals that underwent sham surgery. (C-D) The same as Figure 4B-C, respectively, but 48 hours after CLP injury. (A-D) N = 6-10 animals per arm. Statistical analysis performed with a Kruskal-Wallis 1-way ANOVA. (E) Plasma MCP-1 levels in hPF4⁺ mice, measured as in Figure 4D. n = 5. Analysis with a Kruskal-Wallis 1-way ANOVA. (F) Animal survival results for the CLP studies. N = 10. Statistical analysis performed with a log-rank (Mantel-Cox) test, showing a significant increase in survival in hPF4⁺ mice treated with DG-KKO compared with those treated with vehicle alone, KKO, or DG-TRA with and without antibiotics (P < .0001).

Figure 6.

Treatment with hPF4 or hPF4 plus KKO or DG-KKO in WT mice undergoing CLP injury. CLP injuries were done as in Figure 5, but with WT mice that either received vehicle, hPF4 at either 20 or 40 mg/kg, or combination therapy of hPF4 (20 mg/kg) plus either 5 mg/kg KKO or DG-KKO. N = 10 animals per arm. (A) Relative platelet counts as in Figure 5A. (B-C) The same as in Figure 5C-D, respectively. (D) MSS in these studies for up to 48 hours after the injury. Mean ± 1 SD shown. (A-D) Statistical analysis was performed with Sidak’s multiple comparison t test.