Wide Variability in the Time Required for Immunotoxins to Kill B Lineage Acute Lymphoblastic Leukemia Cells: Implications for Trial Design

Abstract

Purpose: Recombinant immunotoxins (rITs) targeting CD22 are highly active in hairy cell leukemia, but less so in acute lymphoblastic leukemia (ALL). This study aims to understand the variable activity of an rIT against ALL toward improving responses in clinical application.

Experimental design: We determined in vitro activity of rITs by WST-8 assays and the time needed to kill ALL cell lines and patient-derived ALL blasts by flow cytometry. The findings were translated into two systemic ALL xenograft models. Differences in time needed to kill KOPN-8 cells for distinct rITs were addressed biochemically.

Results: In vitro activity (IC₅₀) of anti-CD22 rIT varied 210-fold from 0.02 to 4.6 ng/mL. Activity also varied greatly depending on the time ALL cells were exposed to immunotoxin from < 30 minutes to > 4 days. For KOPN-8, the difference in exposure time was related to intracellular rIT processing. We showed in newly developed ALL xenograft models, where immunotoxins have a short half-life, that the needed exposure time in vitro predicted the responses in vivo By replacing bolus dose with small doses at frequent intervals or with continuous infusion, responses were substantially improved. We confirmed exposure time variability on patient-derived ALL samples and showed a correlation between exposure time needed to reach maximal cytotoxicity in vitro and their clinical response.

Conclusions: The exposure time needed for rITs targeting CD22 to kill ALL cells varies widely. Our results suggest that ALL patients would have a better response rate to anti-CD22 immunotoxins if treated by continuous infusion rather than by bolus injections. Clin Cancer Res; 22(19); 4913-22. ©2016 AACR.

Figure 1

LMB-11 with poor response in vivo. A model of HA22 (A) which consists of a dsF_v, domain II, and wild-type domain III and LMB-11 (B), consisting of an F_ab, a furin cleavable linker instead of domain II and the B-cell epitope diminished domain III with mutated amino acids shown as spheres. C. NSG mice were injected with KOPN-8 on day 1. Vehicle treated mice were euthanized on days 8, 12, 16, and 20 and analyzed for KOPN-8 BM infiltration. The treatment group received five doses of 2.5 mg/kg LMB-11 IV on days 8, 10, 12, 14, and 16. Mice were euthanized two days after each time point. Values are averages of three mice, errors as SD, p-value determined by two-way ANOVA. D. KOPN-8 cells were treated in vitro for 1 hour with indicated concentrations of LMB-11-Alexa647, not internalized rIT was washed away with 0.2 M Glycine pH 2.5 and MFI determined by flow cytometry. Absolute molecule numbers were generated using Alexa647-beads. Mice were injected with 2.5 mg/kg LMB-11-Alexa647, euthanized 1 hour after injection and analyzed for Alexa647-signal intensity. Each value shown is an average from three independent animals, error as SEM. *Dashed line indicates an average of 20,500 molecules internalized by the KOPN-8 cells in murine BM, correlating to 200 ng/ml LMB-11 in vitro. ^†Dotted line marks in vitro IC₅₀ of LMB-11 (0.8 ng/ml) correlating to 220 LMB-11 molecules.

Figure 2

Cell- and rIT-specific exposure time needed to induce cell death. Cells were treated with 2.8 nM rIT for indicated times, washed, and re-plated in medium without rIT. Seventy-two hours after assay initiation, cells were washed and stained with Annexin-V PE and 7-AAD. Percentage Annexin-V and 7AAD negative cells are plotted over exposure time. The bar graphs show a representative of at least three similar results.

Figure 3

In vitro determined exposure time predicts in vivo response. A. KOPN-8 cells were left untreated or treated with 2.8 nM LMB-11 for indicated times, washed twice with PBS and five million cells of each set were injected per mouse. Mice were euthanized at >10% weight loss. B. KOPN-8 cells were injected in NSG mice on day 1. Untreated mice were euthanized on day 8. Five mice per group were treated with 100 μg of LMB-11 (5 mg/kg) given either as one single dose on day 8, two consecutive doses of 50 μg every 3 hours on day 8, four consecutive doses of 25 μg given every 3 hours on day 8, or four consecutive doses of 8.3 μg every 3 hours on days 8, 9, and 10. All treated mice were euthanized on day 11 and %-KOPN-8 BM infiltration was determined by flow cytometry. C. KOPN-8 cells were injected IV on day 1. Some mice were euthanized on Day 8 to define tumor burden at treatment initiation or treated three times with either vehicle, 2.5 mg/kg LMB-11, or 0.4 mg/kg HA22 on days 8, 10, and 12. All mice were euthanized on day 14 and BM was analyzed for KOPN-8 infiltration by flow. Shown is the mean percentage of remaining KOPN-8 infiltration. D. Mice were injected with KOPN-8 on Day 1 and with rITs on days 8, 10, and 12 similar to A. Animals were euthanized at >10% weight loss. P-values were determined by log-rank test. E. Mice were inoculated with 2 million HAL-01 cells on day 1. On day 8, mice were either treated IV with vehicle, 3 x QOD with 2.5 mg/kg LMB-11, or 0.4 mg/kg HA22, or mice received an ALZET osmotic pump implant IP which injected 0.5 μg/hour HA22 into the peritoneal cavity for 7 days. B, C, and E, bars indicate mean, symbols show individual mice, error is shown in SD, significance levels for two group comparisons in B, C, and E were determined by unpaired t-test as ns = not significant, * p<0.05, ** p<0.1, *** p<0.001, **** p<0.0001.

Figure 4

Differences in exposure time for KOPN-8 are related to intracellular rIT processing. A. Cartoon describing rIT pathway, as modified from (33). rIT is internalized after receptor binding, furin separates the antibody fragment from PE, which then traverses the ER via retrograde trafficking, enters the cytosol where it ADP-ribosylates EF2 (ADPr-EF2), which arrests protein synthesis. This results in a fall of MCL1 followed by a destabilization of mitochondria and activation of intrinsic apoptosis. B. KOPN-8 cells were incubated with Alexa647-labeled rIT for indicated times. Surface bound molecules were washed off in 0.2 M Glycine pH 2.5 and signal intensity for internalized molecules determined by flow. Each time point represents biological triplicates. Shown is the average of three independent experiments. P-value for difference was determined by 2-way ANOVA. C. One million KOPN-8 cells were treated with 200 ng/ml LMB-11 or equimolar HA22 for indicated times or left untreated. Cells were lysed after the indicated time, separated by SDS-PAGE under reducing conditions, blotted, probed for with polyclonal rabbit anti-PE, and detected with anti-rabbit-HRP. Expected height for whole (wh) HA22-Vh-PE38 (51 kDa), furin-cleaved (cl) HA22 fragment (34 kDa) or whole (50 kDa) and cleaved (25 kDa) LMB-11-Vh-PE24 are indicated. Shown is a representative of four independent experiments with similar resutls. D. One million cells of each cell line were lysed, similar volumes separated by SDS-PAGE, and analyzed by western blot. Shown is a representative blot and average densitometry values from four independent experiments. Error shown as SD. Values were background subtracted (net value) and normalized to actin loading control (net furin/net actin). E. KOPN-8 cells were incubated with 2.8 nM of the respective rIT for indicated times, then ³H-leucine was added, incubated for an additional hour and analyzed for ³H-leucine incorporation. Counts per minute were normalized to untreated cells after background subtraction. Values are derived from two independent experiments each in triplicate. P-value for difference was determined by 2-way ANOVA. F. KOPN-8 cells were treated with 2.8 nM of the indicated rIT various times after which cells were lysed and cell lysates were treated with NAD-biotin and rIT in excess in a cell free reaction to label the remaining unmodified EF2 with biotinylated ADP-ribose, which was then probed by Streptavidin in a western blot (shown as ADP-EF2 signal). The same lysates were used in western blots for MCL1, PARP, and GAPDH. Shown are representatives of at least three biological repeats.

Figure 5

Clinical response for HA22 treated patients correlates with in vitro rIT exposure time needed to kill ALL blasts. Primary blast samples from patients before HA22 treatment were expanded in NSG mice. These cells were plated on OP9 cells and exposed to 2.8 nM LMB-11 (A) or HA22 (B) for the indicated times. The clinical response to HA22 is indicated below each respective patient in B as CR (complete response), PR (partial response), HA (hematological activity), SD (stable disease), or PD (progressive disease). Cells were analyzed 3 days after assay initiation by flow cytometry. After staining with human CD19 to determine the ALL population cells were analyzed for cell death as determined by Annexin V-PE/7-AAD. The living ALL was >50% in all untreated samples and viability was normalized within each sample as 100% to the highest viability. Shown is a representative set of at least three biological replicates with similar results.