rBPI₂₁ (Opebacan) Promotes Rapid Trilineage Hematopoietic Recovery in a Murine Model of High-Dose Total Body Irradiation

Abstract

The complexity of providing adequate care after radiation exposure has drawn increasing attention. While most therapeutic development has focused on improving survival at lethal radiation doses, acute hematopoietic syndrome (AHS) occurs at substantially lower exposures. Thus, it is likely that a large proportion of such a radiation-exposed population will manifest AHS of variable degree and that the medical and socioeconomic costs of AHS will accrue. Here, we examined the potential of rBPI₂₁ (opebacan), used without supportive care, to accelerate hematopoietic recovery after radiation where expected survival was substantial (42–75%) at 30 days). rBPI₂₁ administration was associated with accelerated recovery of hematopoietic precursors and normal marrow cellularity, with increases in megakaryocyte numbers particularly marked. This translated into attaining normal trilineage peripheral blood counts 2–3 weeks earlier than controls. Elevations of hematopoietic growth factors observed in plasma and the marrow microenvironment suggest the mechanism is likely multifactorial and not confined to known endotoxin-neutralizing and cytokine down-modulating activities of rBPI₂₁. These observations deserve further exploration in radiation models and other settings where inadequate hematopoiesis is a prominent feature. These experiments also model the potential of therapeutics to limit the allocation of scarce resources after catastrophic exposures as an endpoint independent of lethality mitigation.

Introduction

Perhaps more than Hiroshima, the accidents at Chernobyl and Three Mile Island, natural disaster at Fukushima, and omnipresent threat of nuclear detonation highlight that unexpected radiation exposure is a reality that could affect considerable numbers of individuals. As the scope of the resource and managerial challenges after such an event has been increasingly well-delineated, we understand that radiation mitigation strategies must meet an array of efficacy, toxicity, and implementation criteria to be of general value.^1–11 A number of promising strategies that emphasize improving survival of very severely affected individuals have recently emerged, often demanding the use of extensive personnel and other resource intensive medical management.^{12 and rev. in13,14}

A wide range of partial or total body irradiation (TBI) doses can induce profound hematopoietic suppression without causing lethal damage to other organs. It is therefore unsurprising that most proposed radiation mitigation strategies, including hematopoietic cell transplantation, improve some aspect of radiation-induced acute hematopoietic syndrome (AHS). Indeed, in 2015, filgrastim (recombinant human granulocyte colony stimulating factor, rhG-CSF) and pegfilgrastim (PEGylated rhG-CSF), hematopoietic growth factors used to improve recovery from myelosuppressive cancer therapy including transplantation, became the first agents approved for the indication of improving survival in patients exposed to TBI doses associated with hematopoietic suppression.¹³

Even lower TBI doses associated with AHS commonly cause some transient damage to gastrointestinal mucosa.^15–17 As a result, bacteria (including Gram-negative bacteria bearing endotoxin, also known as lipopolysaccharide or LPS, in their plasma membrane) leak into the systemic circulation.^15–19 In humans, who are particularly endotoxin-sensitive, and in animal model systems, both bacterial sepsis and endotoxemia have marked effects on bone marrow (BM) and peripheral blood (PB) cells, and endotoxin itself can induce myelosuppression and hematopoietic stem cell (HSC) exhaustion.^20–22 Endotoxin sensitivity is regulated by interactive proteins that enhance or downregulate responsiveness. Amongst the most potent is bactericidal/permeability-increasing protein (BPI) whose N-terminus binds endotoxin avidly.²³ This interaction both precludes endotoxin from binding to TLR4, its pro-inflammatory, pro-apoptotic receptor, and potentiates endotoxin clearance. As BPI is abundant in the cytoplasmic granules of polymorphonuclear neutrophils (PMN) and monocytes,^23–26 TBI-induced hematopoietic suppression results in loss of BPI and its endotoxin neutralization capacity.

Opebacan (rBPI₂₁, a 21kD recombinant N-terminal fragment of human BPI) has many advantages as a potential ameliorator of hematopoietic damage caused by radiation. It can be administered subcutaneously (SC), has a long shelf-life and an excellent safety profile in approximately 1200 people of all ages in Phase I-III trials of sepsis and one myeloablative chemoradiotherapy study.^27–32 rBPI₂₁ has considerable anti-infective properties^{25,26,29,32,33} and synergizes with antibiotics against multiply-resistant bacteria,³⁴ potentially of value in radiation disaster settings where concomitant trauma contributes to morbidity and mortality.^35,36 We previously explored the potential of rBPI₂₁ to mitigate radiation-induced lethality in a murine model.³⁷ Thirty-day (D30) mortality after 7 Gy TBI was ≥95%, and neither SC rBPI₂₁ nor oral (PO) enrofloxacin (ENR, a fluoroquinolone, used to model suggested supportive care post-radiation catastrophe) alone was sufficient to improve survival. However, combinatorial rBPI₂₁ and ENR treatment, starting 24 hours after TBI, improved D30 survival from ≤5% to ≥65–90% and markedly mitigated AHS.

At the time of our prior study, there was no known role for rBPI₂₁ in hematopoietic regulation or radiation mitigation. rBPI₂₁ and ENR co-administration limited our ability to understand their individual contributions to hematopoietic recovery. Fluoroquinolones had demonstrated some radiation protection but inadequate mitigation in murine models.^38,39 Moreover, FDA warnings about fluoroquinolone toxicities⁴⁰ and the recent association with gastrointestinal perforation⁴¹ have been concerning. We therefore became interested in studying the role of single-agent rBPI₂₁ in ameliorating AHS and exploring the hypothesis that rBPI₂₁ might participate in stress hematopoietic regulation. An easily deployable therapeutic, with an established safety record, that can ameliorate trilineage AHS independent of effects on lethality should improve recovery from, and decrease the resources necessary to support individuals after, significant (but not uniformly lethal) radiation. Herein, we report results of using SC rBPI₂₁ as sole treatment in this setting.

Materials and Methods

Animal Model and rBPI₂₁ Treatment Regimen

Studies were conducted in accordance with Dana-Farber Cancer Institute (DFCI) Animal Care and Use Committee (OLAW Assurance # A3023-01) approved protocol 13-001. At 12 weeks of age, mice were placed in a Rad Disk rodent isolation irradiation cage (Braintree Scientific) and administered a single dose of either 6.25 Gy or 6.5 Gy TBI at 103 rad/min by a Gammacell 40 Exactor (Best Theratronics, Ontario) cesium source irradiator. Littermates were saved as unirradiated, untreated age-matched controls. Irradiated mice were randomly assigned to 3 groups: 1) no treatment (TBI alone) or, starting 24 hours after TBI, 14 days of twice daily (separated by 6–8 hours) SC injection of either 2) rBPI₂₁ at 40mg/kg/day [2.0mg/ml] (Margaux Biologicals, Inc.) in formulation buffer, composed of citrate buffer saline (5mM citrate, 150mM sodium chloride, pH 5.0) with 0.2% poloxamer 188 and 0.002% polysorbate 80, filtered through a sterile 0.22μm PES vacuum filter, or 3) equal volume of the identical formulation buffer (vehicle control, VEH, from Catalent). Mice were monitored and euthanized as previously reported.³⁷ PB and BM studies performed on D3 (to establish the degree of BM ablation), D15 and D19 (to assess recovery) and D30 (recovery and survival endpoint). In some cases, D10 studies were performed.

Processing and Analysis of Blood

Terminal collection of PB was carried out by cardiac puncture and held in EDTA coated microtainer tubes (Becton-Dickinson). CBCs were performed on a Hemavet 950 FS analyzer (Drew Scientific) by DFCI core technicians. Reticulocyte staining was performed using the Retic-Count kit (Becton-Dickinson) per manufacturer’s instructions. Plasma was collected as previously described.³⁷

Processing and Evaluation of Bones, BM Supernatants and BM Cells

One femur/mouse was fixed in 10% neutral buffered formalin for 24–48 hours at 4°C, placed in Kristensen’s decalcification solution for 48 hours at 4°C and paraffin-embedded. 5 μm sections were stained with H&E. Two observers independently determined BM cellularity/femur and megakaryocyte count/high power field (HPF) using an Olympus BX43 with Olympus SC100 for image capture. BM cellularity was assessed at 40× as the percentage of the BM cavity occupied by nucleated hematopoietic cells and not microenvironmental or adipose constituents. Megakaryocyte counts were calculated as the median value from 10 random HPF/femur at 200× magnification with a 0.40 numerical aperture. All images were taken with CellSens Standard software at 400× magnification with a 0.65 numerical aperture of a 40× objective lens. Contralateral bones were held in complete medium (RPMI 1640 supplemented with 1% HEPES, 1% Pen/Strep, 1% L-glutamine, and 10% heat-inactivated FBS) before being processed. To acquire BM supernatant, the epiphyses of both bones in one hind limb were cut off after dissection and each diaphyseal shaft was flushed with 250 μl of complete medium. Flushed BM cells were pelleted by centrifugation at 300×g for 5 minutes at 4°C. Diluted BM supernatant was stored at −80°C. BM diaphyses were then flushed with an additional total 10 ml of ice-cold complete media to ensure removal of remaining BM cells. The eluate was combined with the first flush pellet and centrifuged at 300×g for 5 minutes at 4°C. The pellet was resuspended in RBC lysis buffer (#R7757 Sigma) per manufacturer’s instructions. Viable BM cell counts/limb were enumerated on a hemocytometer using trypan-blue exclusion.

Quantification of Mediators in Plasma and BM Supernatant

Mediator levels were analyzed by multi-analyte bead-based kit (#MCYTOMAG-70K-PMX, EMD Millipore) per manufacturer’s instructions and read on a Bio-Plex 200 System (Bio-Plex 200 System, Bio-Rad). TPO levels were analyzed by ELISA (Elabscience, #E-EL-M0640) per manufacturers’ directions and read at 450 nm with a microplate reader (SpectraMax M3, Molecular Devices, LLC). To estimate the dilution of BM supernatant in complete medium, we measured the diameter and length of 13 femurs using an Olympus BX43 to calculate the volume of the cavity/femur (median 6.6 mm³, range 2.8–10.8 mm³). We estimated that 50% of the cavity was fluid volume, yielding an estimated volume of 3.3 μL/femur. This estimate was used to correct for the dilution of the BM supernatant protein levels. Reported BM supernatant data were calculated using these assumptions.

BM Cell Flow Cytometry Analysis

Non-specific binding was blocked by incubating in stain buffer (Becton-Dickinson) with 2% normal rat serum at 4°C for 20 minutes. Thereafter, antibody staining was performed by the addition of a pre-mixed FITC-conjugated lineage-specific (Lin) cocktail of antibodies (145-2C11, RB6-8C5, M1/70, RA3-6B2, Ter-119; Biolegend), APC-conjugated c-Kit (2B8; Biolegend), and PE-Cy7-conjugated Sca-1 (D7; Becton-Dickinson) diluted [1:20] in FACS staining buffer (Becton-Dickinson) at 4°C for 25 minutes protected from light. Fluorescent-minus-one samples were used as gating controls. 7-AAD [1:100] was used for viability discrimination. Data was acquired on a Miltenyi MACSQuant and analyzed in FlowJo v.10.0.8 (Treestar). Cells negative for lineage-cocktail markers (Lin⁻) were further designated as Lin⁻/Sca-1⁻/c-Kit⁺ (LK) or Lin⁻/Sca-1⁺/c-Kit⁺ (LSK). Population frequencies were determined on a per limb basis and the absolute number determined using BM cell counts/limb. Because D3 cellularity was very low, BM cells were pooled between animals within respective groups prior to preparation for flow cytometry.

Statistical Analysis

Mantel-Cox log-rank was used to compare survival curves (GraphPad Prism 6). PB parameters and BM cell counts were analyzed using an unpaired parametric t-test with Welch’s correction (GraphPad Prism 6). For CBC values, outliers were removed prior to analysis using a modified Z-score (Excel); equal variance was not assumed. For HSC analyses, we fit separate regression models for the log transformed LSK and LK data. We fit linear models in day, as well as models that treated day as a repeated measure. We included interaction terms to enable us to consider differences between TBI alone and VEH and between TBI doses. We selected our final models for the LSK and LK analyses using the Bayesian Information Criterion (BIC). We confirmed approximate normality of the residuals from the models through histogram plots and applied a Bonferroni correction (separately within the LSK analyses and the LK analyses) for the multiple testing across days. The alpha (α= 0.05) for statistical significance was prospectively identified and not varied for any data analysis. For analyses of histopathology results and cytokine parameters, ordinary one-way ANOVAs with Dunnett’s multiple comparison test were used to compare multiple groups. Where appropriate, n sizes are shown in the figure legends. Where indicated in figures, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

Results

6.25 Gy and 6.5 Gy produce a range of intermediate survivals after single fraction TBI

In our studies examining effects of rBPI₂₁ and ENR co-administration on radiation lethality, a single 7 Gy fraction of TBI produced rapid and profound hematologic suppression and resulted in 95% lethality by D30 in 12 week-old male BALB/c mice.³⁷ To explore whether rBPI₂₁ as a single therapeutic could ameliorate severe AHS in a model of single fraction TBI where survival was possible but not universal, we examined the lethality of TBI at 4 different TBI doses: 6 Gy, 6.25 Gy, 6.5 Gy and 7 Gy. Survival determined by Mantel-Cox log rank was significantly different across the 4 dose groups examined (P < 0.0001, Supplementary Fig. S1). D30 survival of the 6.25 Gy cohorts ranged from 60–75% (mean 68%, total n = 19 from 2 independent experiments). D30 survival of the 6.5 Gy cohorts ranged from 42–58% (mean 50%, total n = 24 from 2 independent experiments). As the survival at these doses was both in the desired intermediate range and sufficient to permit comparisons between untreated and treated cohorts, subsequent experiments to examine hematopoietic effects were performed at both 6.25 and 6.5 Gy.

rBPI₂₁ accelerated trilineage PB count recovery after AHS

White blood cell (WBC) and PMN recovery

Both 6.25 and 6.5 Gy TBI were associated with sustained PB pancytopenia, including >15 days of leukopenia and neutropenia in irradiated controls (Fig. 1). In contrast, by D15 after 6.25 Gy, rBPI₂₁-treated mice had a trend to more (2- to 3-fold) WBC and PMN than either control (Fig. 1A,B). After 6.5 Gy (Fig. 1E,F), that difference was significant for WBC (rBPI₂₁ vs TBI alone P = 0.01 or VEH P = 0.009) and PMN (rBPI₂₁ vs TBI alone P = 0.009 or VEH P = 0.01). While median WBC and PMN counts remained low and virtually unchanged through D19 in both irradiated controls, rBPI₂₁ treatment was associated with higher counts after 6.25 Gy (WBC: rBPI₂₁ vs TBI alone or VEH P =0.0003; PMN: rBPI₂₁ vs TBI alone or VEH P = 0.0001) and 6.5 Gy (WBC: rBPI₂₁ vs TBI alone P = 0.004 or VEH P = 0.003; PMN: rBPI₂₁ vs TBI alone or VEH P = 0.002). Only rBPI₂₁-treated mice had D19 median WBC (6.25 Gy) and PMN (both doses) above the lower limit of normal (LLN). Similar recovery was not observed in irradiated controls until D30, when virtually all groups had median values above the LLN. rBPI₂₁-treated mice also experienced accelerated monocyte recovery (Supplementary Fig. S2).

Figure 1.

Platelet recovery

PLT counts recovered with a similar trajectory (Fig. 1C,G). By D15, rBPI₂₁-treated mice had greater median PLT than irradiated controls (6.25 Gy, P = 0.005 vs either irradiated control; 6.5 Gy, P = 0.04 vs either irradiated control). Between D15-19, rBPI₂₁-treated mice continued to recover PLT counts rapidly, by 3-fold (6.25 Gy) and 5-fold (6.5 Gy), nearing or exceeding the LLN. During this period, counts in irradiated controls rose < 2-fold, remaining well below the LLN. The median D19 PLT count of rBPI₂₁-treated mice was significantly greater than TBI alone or VEH counts (P < 0.0001 for each) at both TBI doses. By D30, virtually all groups had median counts at or exceeding the LLN.

Red cell recovery

Median hematocrit (HCT) universally lay below the LLN by D15 (Supplementary Fig. S2). VEH and rBPI₂₁-treated groups had lower D15 HCT than TBI alone after 6.5 Gy (VEH vs TBI alone P = 0.0006 and rBPI₂₁ vs TBI alone P = 0.004) but not 6.25 Gy, raising the possibility that injection related bleeding may have contributed to falls in HCT in the more compromised 6.5 Gy mice. Bleeding at rBPI₂₁ or VEH injection sites (total 28 injections over 14 days) was observed either immediately or as SC hemorrhage at necropsy and was more frequently noted after 6.5 Gy. At D19, HCT had fallen further in irradiated controls, but had risen in rBPI₂₁-treated mice, producing significant differences (6.25 Gy: rBPI₂₁ vs TBI and vs VEH P < .0001; 6.5 Gy: rBPI₂₁ vs TBI P = 0.0001 and rBPI₂₁ vs VEH P = 0.0015). By D30, all groups had median HCT above or nearing the LLN. Changes in hemoglobin (Hgb) and red blood cell count (RBC) were similar (Supplementary Fig. S2). There was a trend to elevated reticulocyte frequency in rBPI₂₁-treated mice by D15 (Fig. 1D,H). By D19, reticulocytes comprised 18% of circulating RBC in rBPI₂₁-treated mice in comparison to low, unchanged values (< 4%) in irradiated controls (6.25 Gy: rBPI₂₁ vs either TBI alone or VEH P = 0.01; 6.5 Gy: rBPI₂₁ vs TBI alone P = 0.0004 or VEH P = 0.0002).

rBPI₂₁ accelerated the rate of recovery of BM hematopoietic populations

rBPI₂₁ accelerated recovery of BM cell counts

On D3, BM cell counts/hindlimb were profoundly reduced (Fig. 2A,D). There was a trend to greater ablation at 6.5 Gy (P = 0.09). Neither VEH nor rBPI₂₁ impacted the degree of hematopoietic suppression on D3 (6.25 Gy: TBI alone vs VEH P = 0.4 or rBPI₂₁ P = 0.6; 6.5 Gy: TBI alone vs VEH P = 0.4 or rBPI₂₁ P = 0.8). We evaluated the time to normal count recovery by comparing each irradiated treatment group to unirradiated controls. In rBPI₂₁-treated mice, median BM counts increased by 7.3-fold (6.25 Gy) and 1.7-fold (6.5 Gy) between D3-15 and reached or exceeded the LLN by D19 (Fig. 2A,D). In contrast, counts remained ≤ 5% of the LLN between D3-19 in irradiated controls, regardless of TBI dose. While all D30 group medians fell above the LLN, rBPI₂₁-treated group values exceeded the normal median whereas irradiated controls remained 33–53% below.

Figure 2.

We also compared counts between irradiated treatment groups (Fig. 2A,D). After 6.25 Gy, counts were greater in the rBPI₂₁-treated group by D15 (rBPI₂₁ vs TBI alone P = 0.0012 or VEH P = 0.0006), D19 (rBPI₂₁ vs TBI alone P = 0.0002 or VEH P = 0.0003) and D30 (rBPI₂₁ vs TBI alone or VEH P = 0.03). Similarly, counts were greater in the rBPI₂₁-treated group after 6.5 Gy by D15 (rBPI₂₁ vs TBI alone P = 0.017 or VEH P = 0.21), D19 (rBPI₂₁ vs TBI alone P = 0.001 or VEH P = 0.004) and D30 (rBPI₂₁ vs TBI alone P = 0.033 or VEH P = 0.008).

rBPI₂₁ accelerated recovery of LK and LSK counts

We first compared values in each irradiated treatment group to values in normal age-matched mice and characterized the fold increase in LK and LSK populations to illustrate the trajectory each group followed during recovery. Median LK counts in rBPI₂₁-treated mice rose early, increasing 253-fold (6.25 Gy) and 97-fold (6.5 Gy) between D3-15. By D19, approximately 50% of rBPI₂₁-treated mice had exceeded (6.25 Gy) or neared the LLN (6.5 Gy, Fig. 2B,E). By D30 all rBPI₂₁-treated mice in both TBI dose groups had LK counts within the normal range. In contrast, D3-15 fold-increases were much lower (range, none to 26-fold) in irradiated controls. By D19, median values in both irradiated control groups at both TBI doses remained at 4–13% of the LLN. Capacity for delayed recovery of LK in irradiated controls was evident in the 10- to 30-fold increases observed from D19-30. By Day 30, the majority of irradiated control mice had LK counts in the normal range.

LSK recovery was comparable. In rBPI₂₁-treated mice, median LSK counts increased by 4.6-fold (6.25 Gy) and 3.3-fold (6.5 Gy) between D3-15 and the majority (6.25 Gy) or some (6.5 Gy) had LSK counts that exceeded the LLN of by D19 (Fig. 2C,F). By D30, the fraction above the LLN had increased for both TBI doses. In contrast, median LSK counts remained unchanged between D3-15 in both irradiated controls regardless of TBI dose, and by D19 their LSK counts were still 6–18% of the LLN. Despite greater fold increases (ranging 3- to 13-fold) from D19-30, less than 50% of the irradiated controls had LSK counts that neared or exceeded the LLN at D30 after 6.25 Gy and virtually none did so after 6.5 Gy.

We also compared irradiated treatment groups to one another (Table 1). For all modelling approaches, we found non-significant interactions representing differences between TBI alone and VEH and therefore we combined these two groups into a single category referred to as irradiated controls for the subsequent analyses. The linear model for LK estimated a significantly higher slope for rBPI₂₁ versus irradiated controls (difference in slopes of 0.036, SE = 0.006, P < 0.0001). The linear model for LSK also obtained a significantly higher slope for rBPI₂₁ than irradiated controls (difference in slopes of 0.024, SE = 0.005, P < 0.0001). The selected model for LK retained the interactions between day and rBPI₂₁ vs irradiated controls and no terms involving dose. The selected model for LSK retained the interactions between day and rBPI₂₁ vs irradiated controls as well as the interactions between day and dose. The repeated measures models for both LSK and LK obtained significantly higher values for rBPI₂₁ versus irradiated control results (after Bonferroni correction) for D10, 15 and 19. This is also true at D30, though is not significant at the corrected p-value (Table 1). Global P-values (5 degrees of freedom) for each analysis for the comparison between rBPI₂₁ and irradiated controls are both < 0.0001.

Table 1.

rBPI₂₁ accelerated recovery of progenitor and stem cell populations

	Day	Difference in slope of rBPI21 vs combined TBI controls	SE	p-value
LK	3	0.2	0.23	0.383
	10	0.99	0.38	0.011
	15	1.34	0.22	<0.0001
	19	1.28	0.21	<0.0001
	30	0.39	0.19	0.04
LSK	3	0.007	0.19	0.972
	10	1.07	0.31	0.0008
	15	0.65	0.18	0.003
	19	0.97	0.17	<0.0001
	30	0.35	0.15	0.022

Data aggregated from 2–5 independent experiments at each TBI dose, n = 3–9 per group. SE = standard error

rBPI₂₁ accelerated recovery of in situ BM cellularity and megakaryocyte counts

The loss of BM hematopoietic elements at D3 was profound (Fig. 2A,D) at both TBI doses, which precluded meaningful histologic comparisons. Thereafter, rBPI₂₁-associated recovery of BM cellularity was more complete than either irradiated control at each time point and TBI dose (Fig. 3A–D). In contrast, we observed no change in the cellularity of either irradiated control through D19 regardless of TBI dose, and D30 median cellularity remained ≤ 50% of normal. These histopathologic results were consistent with the BM cell counts shown in Fig. 2A,D. rBPI₂₁-treated mice had earlier, more robust megakaryocyte recovery than irradiated controls on each day evaluated at 6.25 Gy, and D19-30 at 6.5 Gy (Fig. 3A–C,E).

Figure 3.

rBPI₂₁ was associated with increased cytokine and hematopoietic growth factor levels

The rapid hematopoietic recovery in rBPI₂₁-treated mice led us to examine early levels (D3) of hematopoietic and inflammatory mediators that could have contributed to these later outcomes. For the majority of mediators assessed, D3 plasma levels in rBPI₂₁-treated mice were significantly elevated in comparison to both unirradiated and irradiated controls (Fig. 4A–H). In some cases, these effects were only observed at 6.5 Gy or in more singular patterns (Supplementary Fig. S3).

Figure 4.

We observed fewer effects in BM supernatant, although the high-volume flush required for sample collection introduced substantial dilution of the extracellular fluid from the cavity which may have impacted assay sensitivity. As described in Methods, we corrected for this dilution to provide some reasonable comparability of concentrations of mediators in the dilute supernatants with that in undiluted plasma. Significant rBPI₂₁-associated elevation of G-CSF, KC, and CCL-2 were observed on D3 (Fig 4I–K). For KC and CCL-2, median BM supernatant levels were higher than in plasma whereas G-CSF levels were comparable.

Because thrombopoietin (TPO) participates in regulating both HSC and megakaryocyte production,^42,43 we determined D3 TPO levels in plasma and BM supernatant (Supplementary Table S1). Levels were not different between the TBI alone and VEH-treated cohorts (at 6.25 Gy P = 0.24, at 6.5 Gy P = 0.94), thus they were combined as irradiated controls for TPO analyses. In comparison to unirradiated controls, plasma levels in irradiated controls rose by 1.6- (6.25 Gy) to 1.5- (6.5 Gy) fold while BM supernatant levels rose 4.5-fold (6.25 Gy) and 9.8-fold (6.5 Gy). The addition of rBPI₂₁-treatment led to further increases in both plasma and supernatant levels, which were significant in most cases.

rBPI₂₁ improved survival of mice irradiated at 6.5 Gy

Although these studies were designed primarily to examine effects of rBPI₂₁ on AHS, a limited number of small survival cohorts were also followed. At 6.25 Gy, rBPI₂₁ did not improve D30 survival (rBPI₂₁ vs TBI P = 0.2 and rBPI₂₁ vs VEH, P = 0.93) However, at 6.5 Gy, which was associated with greater D30 mortality, rBPI₂₁ administration resulted in a near doubling of survival (mean 78%, range 67–92) in comparison to VEH (mean 42%, range 33–50, P = 0.0065 by Mantel-Cox; Supplementary Fig. S4).

Discussion

The numerous, interactive challenges to providing adequate medical care following substantial radiation exposure are well-delineated in manuscripts describing the intensive resource and operational requirements of responding to a catastrophic event.^1–11 Optimal management of individuals with unintended radiation exposure will likely depend upon factors including exposure (dose, type of radiation and body surface involved), extent and type of concomitant trauma, and facilities and resources available for triage and treatment.^1–11,35 As AHS occurs at relatively lower radiation doses than other acute radiation toxicities^10,44–46 it is likely that many individuals within a radiation exposed population will manifest hematologic injury of variable degree and duration. Most candidate, and the approved (Neupogen and Neulasta), radiation mitigators improve post-TBI hematopoietic recovery as well as survival although the effects are not equal for all hematopoietic lineages.^13,14,47,48 In order to obtain FDA designation as a radiation mitigator, the path to approval specifies studies demonstrating a survival advantage in lethal/near-lethal TBI model systems. However, the larger number of individuals will experience more variable partial or TBI exposures less likely to lead to death but highly likely to result in clinically relevant AHS.^1,10,45,49 Streamlined approaches that improve the likelihood and rapidity of trilineage hematopoietic recovery in such populations will decrease the frequency of pancytopenia-related complications and reduce demand on scare resources.

In this report, at TBI doses where D30 survival was 42–75% without intervention, we nonetheless observed profound persistent pancytopenia lasting several weeks. Treatment with single agent SC rBPI₂₁ without supportive care substantially decreased the duration and degree of trilineage hematopoietic impairment in both BM and PB and accelerated return to normal values, often by 2–3 weeks. The rapid rise in short-lived PB cells such as PMN, PLT and reticulocytes as well as the recovery of BM megakaryocytes suggests this was largely due to improved production. Notably, a different N-terminal BPI construct was recently shown to promote HSC, progenitor and myeloid cell recovery post-irradiation in a perfused murine BM-on-a-chip model of AHS (RBC, megakaryocyte and PLT production were not quantitated).⁵⁰ In addition, in a small phase I/II study we conducted in patients receiving myeloablative chemoradiotherapy conditioning for allogeneic hematopoietic cell transplantation, a 72-hour rBPI₂₁ infusion was well-tolerated and associated with considerably more rapid PLT recovery and substantially fewer infections and Grade 3–5 toxicities in the first 35 days post-transplant than that observed in protocol-eligible cotemporaneous controls.⁵¹ In aggregate, these data support the hypothesis that rBPI₂₁ could impact transfusion requirements, pancytopenia-related complications including infections, and associated material, personnel and hospitalization utilization post-irradiation.

Hematopoietic cell transplant patients become endotoxemic and BPI deficient after chemoradiotherapy-based myeloablation.³⁷ Moreover, endotoxin ligation of TLR4 and downstream signaling play a major role in aberrant myeloid differentiation, HSC exhaustion and death during sepsis or experimental endotoxemia.^{21,22,52–54} Together, these observations led us to investigate a potential role for rBPI₂₁-mediated endotoxin neutralization in AHS mitigation. As rBPI₂₁ effects had been attributed almost solely to its ability to neutralize endotoxin and ameliorate inflammatory cytokine release during endotoxemia,^{23–26,30,31,34} we anticipated that rBPI₂₁ administration would be associated with decreased HSC death or damage and decreased levels of inflammatory or hematopoietic mediators that had previously been described as elevated after radiation or during sepsis-related emergency hematopoiesis.^22,55–58 Instead, rBPI₂₁ administration markedly augmented levels of these mediators in plasma and/or BM supernatant. Data from two unrelated systems suggest that rBPI₂₁ may produce this effect at least in part by an endotoxin-independent mechanism. In retinopathy models, rBPI₂₁ was found to act directly on retinal epithelial, pericyte and endothelial cells,^59,60 influencing growth and apoptotic regulation and inhibition of VEGF-induced angiogenesis. In an organ-on-a-chip model of AHS, an N-terminal rBPI construct used to perfuse sterile BM stroma and HSC substantially increased hematopoietic cell proliferation.⁵⁰ These observations support the notion that endotoxin-independent, direct interactions with cells, potentially including endothelium and pericytes in the hematopoietic microenvironment, led to some of the rBPI₂₁ effects observed.

Independent of mechanism, elevation of hematopoietic regulators, has been previously reported in association with AHS mitigators. Notably, the survival benefit associated with CBLB502 (flagellin)-mediated TBI mitigation was correlated with elevation of plasma G-CSF and IL-6 levels.⁵⁵ As blocking either molecule decreased the efficacy of CBLB502, the authors suggested this combination might be a biomarker of an effective AHS mitigation treatment. Furthermore, co-administration of IL-6 and G-CSF has been shown to provide more effective AHS mitigation than either alone.⁶¹ In rBPI₂₁-treated mice, levels of these two mediators were 50–100-fold greater than in VEH-treated mice. Moreover, some mediators that rose in association with rBPI₂₁ administration, including G-CSF, TPO, and IL-6, have been evaluated as possible AHS mitigators in multiple species where they have accelerated single- or multi-lineage recovery.^{12–14,47,48,58,62,63} Other mediators that rose in association with rBPI₂₁ treatment are also known to effect hematopoietic cells. For example, the myeloid growth factor GM-CSF is included as a radiation countermeasure in the Strategic National Stockpile.¹³ IL-9 can potentiate megakaryocytopoiesis in conjunction with other mediators, including TPO,⁶⁴ and has a critical role in resolving inflammation.⁶⁵ CCL2 can reduce HSC cycling, potentially supporting a longer period for repair after TBI and increased HSC survival.⁶⁶ Thus, it is possible that the quantitative and temporal relationship of mediators observed in response to rBPI₂₁ treatment, in conjunction with amelioration of endotoxin-related toxicity, support both hematopoietic recovery and survival through a variety of independent and interactive mechanisms.

Comparisons between therapeutics are always colored by differences in model selection and experimental design and regimens. With these caveats, the effects of rBPI₂₁ on the time to onset, rate and completeness of hematopoietic recovery after AHS reported here exceed or equal those observed with other agents, including rhG-CSF administration for a similar duration in murine models of sub-lethal TBI.⁶¹ Most reports detail greatest effects on PMN recovery, while effects of equivalent magnitude on erythropoiesis and/or thrombopoiesis are less commonly reported.^13,14 While PMN are essential to decreasing infectious risk and maintaining tissue integrity, PLT also play a critical role in wound repair and vascular integrity. PLT reconstitution is also an important endpoint as collection of platelets for transfusion is costly, time intensive, yields a product with limited shelf-life, and administration is associated with high risk of reactions and infection.⁶⁷ However, the goal of the experiments reported here was not to establish the relative merits of rBPI₂₁, but rather to determine the capacity of rBPI₂₁ to rapidly resolve TBI-induced pancytopenia when given as a single agent in the absence of supportive care. This study suggests that there are likely multiple mechanisms, some novel, that contribute to the effects on hematopoiesis observed here and that these deserve further exploration in regard to normal hematopoietic regulation and other stress conditions such as myelosuppressive therapy, sepsis and acquired or congenital marrow failure. Finally, this report emphasizes the need to identify meaningful endpoints other than improved survival to foster development of approaches that limit the allocation of scarce resources to the large number of individuals expected to experience consequential, but not necessarily irreversible, syndromes such as AHS.