Etoposide selectively ablates activated T cells to control the immunoregulatory disorder hemophagocytic lymphohistiocytosis

Abstract

Hemophagocytic lymphohistiocytosis (HLH) is an inborn disorder of immune regulation caused by mutations affecting perforin-dependent cytotoxicity. Defects in this pathway impair negative feedback between cytotoxic lymphocytes and APCs, leading to prolonged and pathologic activation of T cells. Etoposide, a widely used chemotherapeutic drug that inhibits topoisomerase II, is the mainstay of treatment for HLH, although its therapeutic mechanism remains unknown. We used a murine model of HLH, involving lymphocytic choriomeningitis virus infection of perforin-deficient mice, to study the activity and mechanism of etoposide for treating HLH and found that it substantially alleviated all symptoms of murine HLH and allowed prolonged survival. This therapeutic effect was relatively unique among chemotherapeutic agents tested, suggesting distinctive effects on the immune response. We found that the therapeutic mechanism of etoposide in this model system involved potent deletion of activated T cells and efficient suppression of inflammatory cytokine production. This effect was remarkably selective; etoposide did not exert a direct anti-inflammatory effect on macrophages or dendritic cells, and it did not cause deletion of quiescent naive or memory T cells. Finally, etoposide's immunomodulatory effects were similar in wild-type and perforin-deficient animals. Thus, etoposide treats HLH by selectively eliminating pathologic, activated T cells and may have usefulness as a novel immune modulator in a broad array of immunopathologic disorders.

Figure 1. Etoposide treatment rescues LCMV-infected prf-/- mice from HLH-like disease

LCMV-infected prf-/- mice were treated with etoposide (ETOP), or carrier by ip injection 5 days post-infection. LCMV-infected wild type mice treated with carrier 5 days post-infection are included for comparison. Mice were monitored longitudinally for survival (A) and disease severity using a clinical scoring system described in the methods section (B). Plasma was assessed by ELISA at day 8 after infection for soluble CD25 (C) and at day 12 for ferritin (D), and fibrinogen (E). Blood was drawn on day 15 day after LCMV infection for assessment of blood counts, including hemoglobin (F), platelet (G), and neutrophil (H) levels. Tissues were obtained 12 days after infection and stained with hematoxylin/ eosin, revealing preservation of normal splenic architecture, decreased liver infiltrates, and improved marrow cellularity with etoposide treatment (markers = 100 microns) (I). Data are presented as mean ± SE and are compiled from 8-23 mice per group assayed across 3 or more experiments. † = all mice in group deceased. *p<0.005,

Figure 2. Only select chemotherapeutic agents have efficacy in treating murine HLH

LCMV-infected prf-/- mice were treated with cyclophosphamide (CPM), etoposide (ETOP), methotrexate (MTX), cisplatin (CDDP), clofarabine (CLO), dexamethasone (DEX), doxorubicin (DOXO), fludarabine (FLU), 5-fluorouracil (5FU), vinblastine (VBL), or carrier control (Prf-/- CON) by IP injection 5 days post-infection. LCMV-infected wild type mice treated with carrier control (WT CON) 5 days post-infection are included for comparison. Mice were monitored serially for disease severity using a clinical scoring system to measure therapeutic effect (A) or lack thereof (B). Blood was drawn on day 15 post-LCMV infection to assess hemoglobin levels (C). Data are the mean ± SE of 3-4 mice per group, representative of 3 or more independent experiments. † = all mice in treatment group deceased, *p<0.004.

Figure 3. Etoposide treatment does not directly decrease macrophage activation or the quality of antigen presentation by dendritic cells in LCMV-infected prf-/- mice

LCMV-infected prf-/- mice were treated with etoposide or carrier 5 days post-infection. LCMV-infected wild type mice treated with carrier are included for comparison. Surface phenotype of splenic macrophages (F4/80+) was assessed two days later by flow cytometry, including CD80 and MHC class II levels (A and B). Dendritic cells (CD11c+/ MHC II+) were magnetically sorted from collagenase treated spleens (pooled, 4 animals/group) seven days after infection and plated with LCMV-specific transgenic CD8+ or CD4+ T cells (P14 or SMARTA) to assess MHC class I and class II restricted presentation of endogenously acquired viral antigens. Antigen presentation to T cells was quantitated by IFN-γ production after overnight culture. No IFN-γ was detected when T cells of irrelevant specificity were cultured with DC’s (not shown). Data are the mean ± SE of 3-4 animals per group, representative of 3 independent experiments. *P<0.05

Figure 4. Etoposide alleviates murine HLH by suppressing IFN-γ levels in LCMV-infected prf-/- mice

LCMV-infected prf-/- mice were treated with carrier or etoposide 5 days post-infection and serum was obtained on days 3, 6, 8, 10, and 13 for assessment of IFN-γ levels by ELISA (A). LCMV-infected prf-/- mice were treated with etoposide either before or after peak IFN-γ production (day 5 or 10) and monitored longitudinally for disease severity using a clinical scoring system (B) or bled on day 15 to assess blood hemoglobin level (C). LCMV-infected wild type mice and prf/ IFN-γ (DKO) deficient mice are included for comparison. LCMV-infected prf-/- mice were treated on day 5 post-infection with carrier or etoposide. On day 5 these mice were also implanted with osmotic pumps (containing either IFN-γ or saline) to ‘add back’ IFN-γ that is suppressed by etoposide treatment. Pumps were calibrated to maintain serum IFN-γ levels between 3 and 6 ng/ml for up to 7 days. Mice were monitored longitudinally using a clinical scoring system (D) and blood was drawn for assessment of hemoglobin 15 days post-LCMV infection (E). Data are the mean ± SE of 3-4 mice per group, representative of 3 or more experiments. †=all mice in treatment group deceased, *p<0.01 when comparing etoposide vs. carrier treated mice, DKO vs. prf-/- mice, treatment at day 5 vs. day 10, or IFN-γ vs. saline infused animals.

Figure 5. Etoposide acts via selective destruction of activated effector T cells in LCMV-infected prf-/- mice

LCMV-infected WT or prf-/- mice were treated with etoposide or carrier 5 days after LCMV infection. Spleens were harvested 8 days post-infection after in vivo brefeldin administration for quantitation of T cells which were producing IFN-γ in vivo (A). In parallel experiments, day 8 spleen cells were stained using MHC-peptide tetramers (Gp33 in the context of D^b) to delineate virus-specific CD8+ T cells. Representative plots of live gated-spleen cells from each group are shown. Percentage shown reflect % of live-gated/CD8+ cells. (B). Absolute numbers of virus-specific (D^b-GP33 tetramer+), naïve (CD44^lo), and quiescent memory CD8+ T cells were quantitated in LCMV-infected, etoposide (or carrier) treated animals (C). Fold change with etoposide treatment was calculated by comparing the total number of each cell population in spleens of carrier and drug treated animals. For assessment of quiescent memory T cells, ovalbumin-specific T cells primed by vaccinia-ova infection >1 month prior to LCMV infection were tracked. Virus-specific CD4+ T cells were enumerated with IA^b-GP61-80 tetramer and compared to naïve CD4+ T cells in LCMV-infected, etoposide (or carrier) treated animals (D). *P<0.01 Data are ± SE, with >8 mice per group, from 3 or more experiments.

Figure 6. Etoposide directly and preferentially induces apoptosis of activated T cells

Spleen cells from LCMV-infected (day 6) prf-/- mice were cultured in the presence of etoposide or carrier for 4 hours, then washed and cultured overnight. Cells were then assessed for permeability (with 7-AAD) and phosphatidylserine exposure (using a recombinant PS-binding protein, see methods). CD8+ cells are shown (A). LCMV-activated (day 6, CD44^hi cells) and naïve (CD44^lo from uninfected mice) CD8+ cells were exposed to a titration of etoposide (or carrier) as in A, and apoptosis induction (%phosphatidylserine+) was measured (B). *P<0.01