Peptide-guided delivery improves the therapeutic efficacy and safety of glucocorticoid drugs for treating acute lung injury

Abstract

Acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) are life-threatening conditions with excessive inflammation in the lung. Glucocorticoids had been widely used for ALI/ARDS, but their clinical benefit remains unclear. Here, we tackled the problem by conjugating prednisolone (PSL) with a targeting peptide termed CRV. Systemically administered CRV selectively homes to the inflamed lung of a murine ALI model, but not healthy organs or the lung of healthy mice. The expression of the CRV receptor, retinoid X receptor β, was elevated in the lung of ALI mice and patients with interstitial lung diseases, which may be the basis of CRV targeting. We then covalently conjugated PSL and CRV with a reactive oxygen species (ROS)-responsive linker in the middle. While being intact in blood, the ROS linker was cleaved intracellularly to release PSL for action. In vitro, CRV-PSL showed an anti-inflammatory effect similar to that of PSL. In vivo, CRV conjugation increased the amount of PSL in the inflamed lung but reduced its accumulation in healthy organs. Accordingly, CRV-PSL significantly reduced lung injury and immune-related side effects elsewhere. Taken together, our peptide-based strategy for targeted delivery of glucocorticoids for ALI may have great potential for clinical translation.

Graphical abstract

A macrophage-targeting peptide was found to selectively home to the inflamed lung, but not healthy organs, in the murine acute lung injury model. Peptide functionalization improves the in vivo biodistribution of a glucocorticoid drug, prednisolone, and thus enhances its therapeutic efficacy and safety when treating acute lung injury.

Introduction

Acute lung injury (ALI) is a life-threatening condition characterized by excessive and uncontrolled systemic inflammatory responses accompanied by extensive inflammatory cell infiltration, disruption of the alveolar epithelial-endothelial capillary barrier, and destruction of alveolar structure, all of which finally lead to respiratory failure.¹ ALI and its more severe form, acute respiratory distress syndrome (ARDS), are regarded as a major health threat in the respiratory tract, which accounts for about 40% of mortality in intensive care units worldwide.^1,2 With potent anti-inflammatory activities, glucocorticoid drugs were considered as an obvious choice for ALI treatment and had been tested in multiple clinical trials over the past several decades.^3,4,5 However, these studies failed to show a clear clinical benefit of glucocorticoid drugs for ALI/ARDS treatment due to their immune-related side effects caused by nonspecific drug accumulation in healthy organs after systemic administration.⁶ A technology to deliver these drugs more specifically to the disease site and to confine or even increase their immunosuppressive activity in the disease site while reducing the side effects in healthy organs is highly desirable.

Here, we tackled this problem by covalently conjugating glucocorticoid drugs with a peptide that selectively targets the lung tissue under inflammatory conditions. Via an in vitro phage screen on a macrophage cell line, a cyclic peptide (CRVLRSGSC, termed CRV, from the first three residues) was identified, in which two terminal cysteines render the peptide cyclic by forming a disulfide bond.⁷ Using a murine lung infection model by bacteria, we found that CRV selectively homes to the infected lung and predominantly colocalizes with macrophages, but not healthy organs, upon intravenous injection.⁷ CRV was able to improve the delivery of porous silicon nanoparticles (pSiNPs) and achieve a higher efficacy of pSiNP-drug complex to attenuate infection-induced acute inflammation. We have identified retinoid X receptor β (RXRB) as the CRV receptor, whose cell-surface presence is limited to a subset of macrophages in the tumor tissue.⁸ It remains to be further clarified in regard to CRV affinity to other types of myeloid cells under various pathological conditions.

Macrophages and neutrophils account for the majority of the immune infiltrates in ALI.^9,10 We speculated that RXRB expression may change with the inflammatory status in the lung and that CRV may facilitate the drug delivery to the inflamed lung. Although CRV-pSiNP formulation has shown promising effects, a significant accumulation of nanoparticles (NPs) was still found in liver and spleen due to the large sizes of NPs.⁷ Therefore, we directly conjugated CRV with prednisolone (PSL) in this study. To release the drug, a reactive oxygen species (ROS)-responsive linker was used between CRV and PSL. ROS is mainly found intracellularly and is excessively generated under inflammatory conditions.^11,12 This design aimed to ensure that CRV-PSL is stable in the circulation while PSL is only cleaved from CRV after entering the cells in response to intracellular ROS. In this study, we explored the specificity of CRV to recognize the inflamed lung in a murine ALI model and tested whether our CRV-PSL conjugate exhibits a stronger therapeutic efficacy and safety profile than free PSL.

Results

CRV targets myeloid cells in inflamed lung due to elevated RXRB expression

In this study, we primarily used an endotoxin-induced ALI model by intratracheal injection of lipopolysaccharide (LPS), which is one of commonly used murine models to mimic the inflammatory responses of ALI.¹³ We first tested the overall RXRB expression in this ALI model, and CRV homing upon systemic administration. RXRB expression increased in the lung tissue upon LPS stimulation (Figure 1A). At 4 days after ALI induction, fluorescein amidite (FAM)-CRV was injected intravenously into the mice for 1-h treatment before the lung and control organs were collected. Immunohistochemistry (IHC) and immunofluorescence (IF) studies showed that CRV accumulates in the inflamed lung at a significantly higher level than control peptides, including GGS, scrambled CRV, and FAM-Cys (CRV replaced with a single cysteine), but there was little accumulation of these peptides in the healthy lung (Figures 1A, 1B, and S1). In healthy organs, these peptides exhibited little accumulation with or without LPS stimulation (Figure S1). CRV predominantly colocalized with CD11b-positive immune cells in the inflamed lung but not in healthy tissues (Figure S1). When given intravenously, an anti-RXRB antibody also showed lung-specific accumulation in ALI mice but not in healthy mice or healthy organs of ALI mice (Figures 1C and S2).

Figure 1.

CRV selectively targets the myeloid cells in the inflamed lung due to the elevated RXRB expression

(A) Representative immunohistochemistry (IHC) images of targeting RXRB by CRV peptide in the lung of ALI mice. Control (PBS) or ALI (LPS-treated) mice were intravenously injected with FAM-labeled CRV or a control peptide (GGS) for 1 h homing, respectively. The lungs were collected for IHC analysis on whole-cell RXRB expression (left) and peptide amount (detected by anti-FITC antibody, right). (B) Quantification of FITC expression for FAM-labeled peptides in the lungs by IHC staining. (C) Representative immunofluorescence (IF) images of colocalization of rabbit anti-RXRB antibody with CD11b in PBS controls (top) and acute lung injury (ALI) mice (bottom). The lungs were collected for IF staining for the indicated markers after 6 h in vivo homing by intravenous injection of 25 μg of rabbit polyclonal RXRB antibodies in 100 μL of PBS in mice. The mice receiving intratracheal injection of PBS served as healthy controls. All the animal experiments were performed in three mice per group. (D) Relative percentage of RXRB-positive cells in different tissues of ALI mice. The indicated tissues were collected from the mice with ALI, dissociated into live single cells, and stained with different antibodies before fixation and flow-cytometry analysis. The healthy mice served as the controls. The inflammatory cells were stained with surface CD45, CD64, and RXRB. The percentage of RXRB⁺ cells (CD45⁺CD64⁺ cells) was normalized to that of healthy control as fold changes and plotted on the y axis. All the animal experiments were performed in 5–8 mice per group and repeated three times. (E) Fractions of indicated cells in total leukocytes of ALI lung (left) and percentage of RXRB⁺ portions in each immune subset of ALI lung after 24 h and 4 days of LPS injection (right). The myeloid cells were selected by plotting CD45⁺ and CD11b⁺ cells. The neutrophils were identified as Ly6G⁺ cells. The macrophages were further selected from non-neutrophils (Ly6G⁻) with F4/80⁺ cells. The healthy mice served as the controls. All the animal experiments were performed in 5–8 mice per group and repeated three times. (F) Preferential binding of FAM-CRV to immune cells by flow cytometry. In vivo binding study of FAM-CRV to cells was performed by intravenous injection of FAM-CRV or FAM-GGS control peptide in ALI mice for 1 h homing. The isolated cells from ALI lung were stained with surface CD45, Ly6G, and F4/80 for different immune cells, respectively. The bound CRV or GGS on the cell surface was stained by anti-FITC antibody for FAM. Left: mean fluorescent intensity (MFI) of FAM⁺ cells in each immune subset. Right: flow-cytometry histogram of CRV binding to the immune cells. The fluorescence intensity for CRV or GGS binding per cells was plotted in the histograms. The cells stained with secondary antibody only served as negative controls. (G) Representative IHC images of RXRB, CD64, and CD11b staining in the lung from human subjects with interstitial lung diseases (ILD). (H) Quantification of RXRB in IHC sections from lungs of human subjects with ILD. The samples from human subjects were performed in four normal lung tissues and four from patients with ILD, respectively. FAM, carboxyfluorescein; FITC, fluorescein isothiocyanate; A.U., arbitrary units; MFI, mean fluorescence intensity; Neg, negative. Data shown are mean ± SD. Student’s t test was performed for statistical analysis. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus control or healthy group. Scale bar, 100 μm.

To further understand the basis of CRV targeting, we investigated the cell-surface RXRB expression in the immune cells of healthy and ALI lungs, as surface RXRB determines the CRV binding to cells.⁸ Single cells of healthy and ALI lungs were first sorted with CD45 for leukocytes, followed by CD64 to gate out myeloid infiltrates, mainly macrophages and neutrophils.^14,15 The percentages of these inflammatory cells were significantly increased in the ALI lungs compared with healthy ones, agreeing with previous reports (Figure S3A).¹⁶ Like our previous findings, a very small fraction of white blood cells or splenic cells of healthy or ALI mice was expressing RXRB on the surface,⁸ while the percentage of RXRB-positive leukocytes was significantly increased in the ALI lung compared with the healthy one (Figure S3B). Accordingly, RXRB-positive portion of myeloid infiltrates (CD64-positive) also significantly increased in the ALI lung compared with healthy ones, but not for their counterparts in blood or spleen (Figure 1D). It was already known that the neutrophil recruitment occurs after 3–6 h of LPS treatment and reaches peak on day 3–4 after stimulation together with macrophage infiltration.^16,17 To have a kinetic view, we investigated the immune subsets (neutrophils [Ly6G⁺], macrophages [Ly6G-F4/80⁺], and lymphocytes) in the lung after 24 h and 4 days of LPS injection. Agreeing with previous reports,¹⁸ the fraction of myeloid infiltrates significantly increased in the lung at either onset or 4 days after ALI (Figure 1E). The surface expression of RXRB-positive cells in neutrophils and macrophages was significantly elevated in the ALI lung, but not for lymphocytes (Figures 1E, S3C, and S3D). We further isolated the cells from ALI lung to discriminate the interstitial macrophages (IM; CD11b⁺F4/80⁺CD11C⁻) and alveolar macrophages (AM; CD11b⁻F4/80⁺CD11C⁺).^19,20 Both types of macrophages have elevated RXRB expression, suggesting the possibility of CRV targeting (Figure S3E). To corroborate the above results, we detected the content of CRV in different immune subsets of ALI lungs after intravenous injection. Macrophages showed the highest CRV signal per cell, indicating a preferential binding or internalization ability (Figure 1F). CRV exhibited a specific binding (higher than GGS) to neutrophils and macrophages, but not lymphocytes (Figure 1F). Upon CRV injection, CRV binds more AM than IM (Figures S3F and S3G). Overall, these results suggest that CRV mainly targets the myeloid infiltrates in the ALI lung likely due to the elevated level of cell-surface RXRB.

Besides the murine model, we also acquired a lung tissue specimen from patients with interstitial lung disease (ILD). Compared with subjects with healthy lungs, the level of RXRB was much higher in the lungs of ILD patients, as was the overall amount of myeloid infiltrates (Figures 1G, 1H, S4A, and S4B). We also found that RXRB colocalized well with CD64-positive cells (Figure S4C).

Synthesis and in vitro characterization of CRV-drug conjugate

Next, we investigated whether CRV can improve the delivery of glucocorticoid drugs to the inflamed lung. PSL was used as a model glucocorticoid drug to covalently conjugate with CRV as schemed in Figure 2A (hereafter CRV-PSL). The experimental conditions of conjugation reactions are described in more detail in materials and methods and shown in Figure S5A. We also replaced CRV with a single cysteine, termed Cys-PSL, as a control conjugate without targeting peptide (Figure S5B). An ROS-responsive linker, 2,2′-thiodiacetic acid, was used between PSL and CRV (Figure 2A). The chemical structures of both PSL conjugates were verified by ¹H nuclear magnetic resonance (NMR) (Figures S5C and S5D).

Figure 2.

Synthesis and in vitro characterization of CRV-prednisolone conjugate with reactive oxygen species responsive linker

(A) Chemical structure of CRV-prednisolone (CRV-PSL) conjugate showing the drug payload PSL, reactive oxygen species (ROS)-responsive linker, PEG2000, maleimide, cysteine, Ahx linker, FAM, and CRV. PSL, prednisolone; PEG2000, polyethylene glycol 2000; Cys, cysteine; Ahx, 6-aminohexanoic acid; FAM, fluorescein amidites. (B) Chromatograms by HPLC showing absorption peaks of CRV-PSL after incubation with mouse plasma with inflammation for 0 h and 24 h, respectively. The plasma was obtained from the mice receiving intraperitoneal injection of 0.3 mg kg⁻¹ LPS for 4 h. (C–F) Gene expression of inflammatory cytokines in LPS-stimulated RAW 264.7 macrophages by CRV-PSL conjugate. CRV-PSL or free PSL was added to the cells 1 h before LPS stimulation (pre-PSL) or 1 h after LPS stimulation (post-PSL). The effect of CRV-PSL on inflammatory cytokines in LPS-stimulated RAW cells was examined after 18 h of drug treatment. The cells receiving PBS alone served as the control group. mRNA expression of inflammatory markers of IL-1β (C), iNOS (D), IL-6 (E), and MCP-1 (F) was quantified and normalized by that of PBS control group (y axis). The experiment was repeated three times and the data shown are mean ± SEM. Student’s t test was performed for statistical analysis. ∗∗p < 0.01, ∗∗∗p < 0.001 versus PBS control; ^###p < 0.001 versus LPS group. (G–H) Protein expression of IL-1β and iNOS in LPS-stimulated RAW 264.7 macrophages by CRV-PSL treatment. GAPDH and tubulin, respectively served as internal controls of the target proteins. The experiment was repeated three times and the data shown are mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. There is no statistical difference between any other groups. A.U., arbitrary units; IL-1β, interleukin 1β; MCP-1, monocyte chemoattractant protein-1; iNOS, inducible nitric oxide synthase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

The ROS-responsive linker is expected to be cleaved under oxidative conditions in ALI,²¹ at which time PSL will be freed from the conjugated peptide (Figure S6A),²² To validate this point, CRV-PSL was incubated with 10 mM H₂O₂ at 37°C for 5 min, 24 h, and 72 h, respectively. The amount of CRV-PSL and PSL was determined by high-performance liquid chromatography (HPLC). The cleavage of CRV-PSL into PSL was seen after 5-min incubation with H₂O₂ and became more evident after longer times of incubation (Figure S6B). Moreover, H₂O₂ induced the release of free PSL from ROS-responsive conjugates in a dose-dependent manner (Figure S6B). The majority of CRV-PSL remained uncleaved in the mouse plasma at least for 24 h (Figure 2B). Overall, these results suggested that the ROS-responsive conjugation is stable in the circulation and allows the release of free PSL under oxidative conditions.

It was previously reported that polyethylene glycol (PEG) conjugation renders drugs, including corticosteroids, inactive until cleaved, after which drugs can become active and subject to metabolism.^23,24 Therefore, our first and most important task of in vitro validation was to verify whether CRV-PSL has immunosuppressive activity equivalent to that of PSL. LPS-induced inflammatory responses have been shown to generate ROS in different cell types including RAW 264.7 macrophage cell line.^25,26 Thus, as a proof-of-principle demonstration, we used RAW 264.7 cells to validate the intracellular cleavage and anti-inflammatory activity of CRV-PSL. CRV-PSL or PSL was added to the cells 1 h before LPS stimulation to exclude the influence of LPS on glucocorticoid receptor expression and functions (pre-PSL).²⁷ We also added the drugs 1 h after LPS stimulation to mimic the drug treatment after the onset of inflammation in a clinical situation (post-PSL).²⁸ At 18 h after drug treatment, LPS-induced inflammatory responses were measured based on the expression of proinflammatory markers including interleukin-1β (IL-1β), IL-6, monocyte chemoattractant protein-1 (MCP-1), and inducible nitric oxide synthase (iNOS); and anti-inflammatory/resolution markers such as transforming growth factor β1 (TGF-β1) and collagen type I α1 (ColI-α1). Under both pre- and post-PSL conditions, CRV-PSL exhibited similar activity as PSL to suppress LPS-induced inflammatory responses (Figures 2C–2F, S7A, and S7B). Consistently, CRV-PSL exhibited an inhibitory effect similar to that of PSL on LPS-induced inflammation as evaluated by the expression of IL-1β and iNOS at the protein levels (Figures 2G and 2H). Similar results were also found in human THP-1 macrophages (Figures S7C–S7E). Additionally, we also detected PSL metabolites inside cells, which is another indicator of PSL release. After cellular uptake, we detected several PSL metabolites, such as 20-α-OH-PSL and 20-β-OH-PSL, using mass spectrometry as previously reported.²⁹ CRV-PSL was able to generate 20-α-OH-PSL and 20-β-OH-PSL in a level similar to that of free PSL (Figure S6C). Together, the above results suggest that CRV-PSL can be processed to release free PSL in cells under oxidative stress and, more importantly, possess immunosuppressive activity similar to that of PSL.

Pharmacokinetics and biodistribution of CRV-PSL conjugate in vivo

We first evaluated the plasma pharmacokinetics of CRV-PSL in healthy mice. PSL, Cys-PSL, and CRV-PSL were injected intravenously at a dose of 1 mg kg⁻¹ (equivalent PSL weight), and their plasma half-life (T_1/2) and area under the plasma concentration curve (AUC) were calculated using a noncompartmental model (Table 1 and Figure S8). CRV and Cys conjugates showed a higher plasma concentration than PSL alone as early as 1 h after injection and remained until at least 48 h (Figure S8). Thus, both PSL conjugates showed a prolonged T_1/2 and about 1-fold increase of AUC compared with PSL alone (Table 1).

Table 1.

Pharmacokinetic parameters of PSL after intravenous administration of free PSL, Cys-PSL, and CRV-PSL to mice (1 mg kg⁻¹ equivalent dose of PSL per mouse)

Pharmacokinetic parameters	Free PSL	Cys-PSL	CRV-PSL
T1/2 (h)	3.78 ± 0.50	4.47 ± 0.22	6.13 ± 2.59
AUC0-t	41,279.66 ± 5,527.07	82,414.35 ± 3,511.63∗	82,880.57 ± 3,611.63∗

Values are expressed as mean ± SD (n = 3), ∗p < 0.001 relative to free PSL. T_1/2, half-life; AUC_0-t, area under the concentration-time curve.

Next, we investigated the in vivo biodistribution of PSL conjugates in ALI mice. At 1 h after intravenous injection, IF staining revealed that CRV-PSL accumulation in the lung of ALI mice was significantly higher than that of Cys-PSL, while there was little accumulation of both conjugates in the lung of healthy mice (Figure 3A). In the ALI lung, CRV-PSL colocalized well with F4/80-positive macrophages (Figure 3A). The accumulation of CRV-PSL was similar to that of Cys-PSL in healthy organs of both healthy and ALI mice (Figure S9).

Figure 3.

Tissue biodistribution and targeting of CRV-PSL conjugate in vivo

(A) Representative IF images of the colocalization of FAM-labeled peptide with F4/80 in the lung of PBS controls and ALI mice. The lungs were collected for IF staining for the indicated markers after 1 h of intravenous injection of FAM-Cys-PSL and FAM-CRV-PSL, with a 1 mg kg⁻¹ equivalent dose of PSL, respectively. (B) Tissue distribution of free PSL in ALI mice by liquid chromatography coupled to tandem mass spectrometry (LC/MS). After 4 days of intratracheal LPS challenge, the mice were treated with free PSL, Cys-PSL, and CRV-PSL, respectively. The dose was equivalent of 1 mg kg⁻¹ of PSL in weight. The tissues were collected after 2 h of treatment. All the animal experiments were performed in 3–4 mice per group. Data are presented as mean ± SD. Statistical significance was calculated using Student’s t test. ∗p < 0.05, ∗∗p < 0.01. There is no statistical difference between any other groups. ND, not determined. Scale bar, 150 μm.

In the above studies, we detected and quantified the conjugates as one entity. However, it is free PSL that is the active component in inducing therapeutic efficacy and side effects. Therefore, we used mass spectrometry to directly quantify the amount of PSL in various tissues of ALI mice. ALI mice were intravenously injected with PSL, Cys-PSL, and CRV-PSL at a dose of 1 mg kg⁻¹ in PSL weight, respectively, and the tissues were collected at 2 h after injection. The CRV-PSL group exhibited a higher amount of PSL in the lung than PSL or Cys-PSL groups (Figure 3B). In the meantime, the liver accumulation of PSL in Cys-PSL and CRV-PSL groups was lower than that of the PSL group (Figure 3B). There was little difference among the three groups in other organs (no detectable PSL signal in the heart) (Figure 3B). Collectively, these results demonstrated that CRV conjugation increases the PSL accumulation in the ALI lung, whereas it reduces it or has no impact in healthy organs.

CRV conjugation increases the therapeutic efficacy of PSL against lung injury

Next, we evaluated the impact of CRV conjugation on PSL efficacy to attenuate lung inflammation and injury. At day 4 post LPS stimulation, ALI lungs showed injury histology characterized by patchy areas of neutrophilic infiltration, alveolar hemorrhage, interstitial thickening, and lung consolidation (Figure 4A). ALI mice were treated with PSL, Cys-PSL, and CRV-PSL at a dosage of 0.01 mg kg⁻¹ and 0.1 mg kg⁻¹ (equivalent PSL weight) every 2 days for 4 days. At the end of treatment, hematoxylin and eosin (H&E) staining was performed on the lung tissue, and the lung injury score was quantified according to the changes of alveolar septae, alveolar hemorrhage, intra-alveolar fibrin, and intra-alveolar infiltrates as described by Matute-Bello et al.³⁰ As shown in Figures 4A and 4B, CRV-PSL displayed a stronger efficacy than PSL and Cys-PSL at both dosages to attenuate lung injury induced by LPS stimulation. Other than histology evaluation, we also examined the inflammatory responses in the lung by quantifying the percentage of infiltrated inflammatory cells in the lung as evaluated by CD64, expression of myeloperoxidase, and iNOS. While ALI lungs showed increased infiltration of inflammatory cells and elevated expression of inflammatory proteins compared with healthy control in agreement with previous reports,^31,32,33 CRV-PSL treatment resulted in stronger suppression of these inflammatory markers than PSL and Cys-PSL (Figures 4D–4F and S10). Taken together, these results demonstrate that CRV conjugation increases the PSL efficacy to suppress inflammation and thus attenuates injury in ALI lungs.

Figure 4.

CRV conjugation increases the therapeutic efficacy of PSL against lung injury

(A) Diagram of treatment timeline in ALI mice. (B) Representative hematoxylin-eosin (H&E)-stained lung sections subjected to histopathological analyses. Left: lung of healthy (PBS) and ALI mouse with no treatment. Right: lung of ALI mouse treated with free PSL, Cys-PSL, and CRV-PSL at a dose equivalent of 0.01 mg kg⁻¹ (top) and 0.1 mg kg⁻¹ (bottom) of PSL. (C) Quantification of inflammatory response in the lungs following H&E staining. (D) Percentage of CD64⁺ cells. Quantification of (E) myeloperoxidase (MPO) and (F) iNOS expression in the lungs by IHC staining. The mice were given the free PSL, Cys-PSL, and CRV-PSL at a dose equivalent of 0.01 mg kg⁻¹ or 0.1 mg kg⁻¹ of PSL intravenously every 2 days for 4 days after intratracheal LPS challenge, and the tissues were collected on day 4 for histopathological analyses. All experiments were performed in 3–4 mice per group. Data are presented as mean ± SD. Statistical significance was calculated using Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; ns, not significant. A.U., arbitrary units. Scale bar, 100 μm.

CRV conjugation reduces short-term side effects of PSL in healthy organs

In this study, we mainly focused on the therapeutic benefits after a short-term treatment (less than a week). The commonly seen side effects after a short-term exposure of glucocorticoid drugs include reduced thymus weight and lymphocyte count due to increased apoptosis.^34,35,36,37 To evaluate the impact of CRV conjugation on side effects of PSL, healthy mice were intravenously injected with PSL, Cys-PSL, or CRV-PSL every 2 days for 5 days with an equivalent dose of PSL at 0.1 and 1 mg kg⁻¹, which was lower than used in previous reports.³⁸ At 4 h after the third injection on day 5, we collected various organs for subsequent analysis and used flow cytometry to quantify the total lymphocyte count and T cell subsets in the circulation. While PSL caused the biggest reduction of thymus weight, Cys conjugation rescued it to some extent and CRV conjugation was able to restore it almost back to the level of untreated animals (Figure 5A and Table S1). We then performed TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) staining to quantify the apoptosis in thymus and other organs. The result showed that PSL treatment increases the level of apoptosis mainly in the thymus, and CRV conjugation helps to reduce it back to the level of the untreated group (Figures 5B and S11A).

Figure 5.

CRV conjugation reduces the acute side effects of PSL in healthy organs

(A) Thymus weight in mice receiving 0.1 mg kg⁻¹ or 1 mg kg⁻¹ equivalent dose of PSL by free PSL and CRV-PSL treatment, respectively. (B) Quantification of immunofluorescent signals of TUNEL-positive cells showing apoptotic cells in tissues of mice receiving different treatment. (C and D) Flow-cytometry analysis of blood cell populations in mice with different treatment. The count of (C) blood lymphocytes and (D) T cells in mice receiving different treatment were normalized to the vehicle group. (E and F) Plasma levels of liver enzymes (E) alanine transaminase (ALT) and (F) aspartate transaminase (AST). The healthy mice were given free PSL or CRV-PSL conjugate every 2 days for 5 days with an equivalent dose of PSL at 0.1 mg kg⁻¹ and 1 mg kg⁻¹, respectively. The weight of thymus was measured, and the blood was collected for flow cytometry 4 h after the third dose of treatment. All experiments were performed in 3–4 mice per group, and the experiment was repeated three times. Data shown are mean ± SD. Student’s t test was performed for statistical analysis. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. There is no statistical difference between any other groups.

Similarly, we also quantified the toxicity toward lymphocytes. While PSL treatment indeed reduced the count of total lymphocytes and T subsets, CRV conjugation was able to restore it back to the level of untreated group (Figures 5C, 5D, and S11B; Table S1). Other than the above parameters, we also quantified the levels of liver enzymes aspartate transaminase (AST) and alanine transaminase (ALT) in the blood, which showed no significant difference among all groups (Figures 5E and 5F). Overall, these results supported the notion that CRV conjugation attenuates the short-term side effects of PSL after systemic administration.

Discussion

In this study, we explored the usefulness of peptide-guided delivery to improve the therapeutic efficacy and safety of a glucocorticoid drug, PSL, for treating ALI. We found that CRV peptides can selectively accumulate in the inflamed lung of ALI mice upon intravenous injection, which may be attributed to the elevated expression of RXRB on the surface of myeloid infiltrates. With such targeting ability, we investigated whether CRV can improve the biodistribution of PSL more toward the inflamed lungs of ALI mice. CRV was covalently conjugated with PSL by an ROS-responsive linker. This conjugate remained intact in the plasma until entry into the cell, where PSL was cleaved from CRV and thus exerted immunosuppressive activity in vitro at a level similar to that of native PSL. In vivo, CRV conjugation increased the accumulation of PSL in the ALI lung and thus improved its efficacy to reduce inflammation and rescue lung injury. On the other hand, CRV conjugation reduced or had no impact on PSL accumulation in healthy organs of ALI mice, and lowered the acute side effects of PSL. Overall, we provide here a peptide-guided delivery technology that may improve the efficacy and safety of glucocorticoid drugs in general and may also be applicable to other drugs for ALI treatment.

Myeloid-lineage immune cells play crucial roles in the pathogenesis of ALI.^39,40,41 Upon injury/insult, pulmonary macrophages are activated to release cytokines and chemokines, leading to the recruitment of neutrophils and alteration of alveolar membrane permeability followed by the destruction of alveolar structures and functions.¹ Agreeing with others, we found here that neutrophils are the predominant immune infiltrates in the LPS-induced ALI model, followed by macrophages, while lymphocytes account for only a very small portion.^13,18 Although neutrophils are recruited to the lung as early as 3–6 h after LPS stimulation, a robust and persistent influx of neutrophils into the air spaces peaks at day 3 with a higher amount than macrophages.¹⁷ Together with the fact that macrophages peak around 3–5 days after ALI onset,^16,17 we used day 4 in this study as the primary time point for analyzing the immune populations and CRV binding. Upon LPS stimulation, the overall expression of RXRB was elevated in the lung tissue. We also found that the cell-surface level of RXRB was elevated mainly in the myeloid infiltrates of ALI lungs compared with healthy ones, which was not seen with control organs (blood and spleen), in agreement with our previous study.⁸ The differential expression of RXRB, particularly at the cell surface, may at least partially explain the targeting specificity of CRV. The peptide binding to different immune subsets upon intravenous injection further strengthens this speculation. We further showed the CRV has a preferential binding to AM compared with IM. The development of ALI depends on the site-specific recruitment of neutrophils and monocytes from circulation, and these myeloid cells possess high plasticity of gene expression in response to the local environment.^42,43,44,45 Therefore, it is likely that RXRB expression in these cells was increased only after entering the lung tissue in response to the inflammatory status. The exact role of RXRB in the biology of these immune cells, and the pathophysiological functions of RXRB-positive cells, remain to be further investigated.

Besides murine models, we also validated this finding with patient specimens. In human, ALI/ARDS develops from various pulmonary disorders including ILD.⁴⁶ Myeloid infiltrates are present at high levels in the lung of ILD patients, which is associated with disease severity.^47,48,49 Besides confirming this result, we found that the overall level of RXRB is significantly increased in the lung of ILD patients compared with healthy individuals. This result indicates the clinical relevance of our study and may open new doors toward understanding immune infiltrate subsets and immune dysfunction during the pathogenesis of ALI.

PSL and other glucocorticoids are important steroid drugs for treating a variety of autoimmune, allergic, and severe inflammatory disorders. Despite their potent anti-inflammatory activities and broad clinical implications, glucocorticoids cause various side effects (e.g., protein catabolism, hyperglycemia, immunosuppression, infection, and osteoporosis), especially at high dosage when managing acute inflammation or over the long term when treating chronic inflammation.^34,37,50,51 One reason is their nonspecific accumulation in healthy organs upon systemic administration, likely due to the hydrophobic nature of these compounds.⁵² Delivery strategies have been explored to improve the targeting specificity. NPs such as liposomes were used to encapsulate PSL and other glucocorticoid drugs.^53,54,55 However, owing to their sizes, a significant portion of these formulations were engulfed by the reticuloendothelial system such as the liver and spleen.⁵⁶ On the other hand, glucocorticoid conjugates with peptides have been developed for treating rheumatoid arthritis,³⁶ ocular diseases,⁵⁷ and obesity.⁵⁸ Drug modification with PEG was also shown to increase the half-life and reduce nonspecific accumulation.^59,60,61 Therefore, despite CRV-pSiNPs being shown to improve drug delivery, we instead explored the approach of direct conjugation between CRV and PSL with a PEG linker. In addition, as a hydrophobic compound, PSL can diffuse freely through cellular membranes. An ROS-responsive linker was thus used to ensure that PSL can be released mainly after entry into target cells.

Agreeing with previous reports,^23,24 the conjugation with PEG and peptide increased the plasma half-life and AUC of PSL. The ROS linker remained intact in the plasma of ALI mice and was cleaved upon H₂O₂ incubation. The main goal of in vitro characterization was to verify that CRV-PSL possesses immunosuppressive activity. First, using LPS-induced inflammation in RAW 264.7 cells, we validated that CRV-PSL exhibited an immunosuppressive activity equivalent to that of free PSL, which can only occur when PSL is cleaved off from PEG. Additionally, we detected the presence of PSL metabolites at similar levels after cells were incubated with PSL or CRV-PSL, which suggests the cleavage of PSL from the conjugates. All these results support the notion that CRV-PSL conjugate is stable until they exposure to oxidative conditions inside inflammatory cells, at which time PSL can be released and thus exert its immunosuppressive function.

After in vitro characterization, we continued to investigate the effect of CRV conjugation on PSL biodistribution in vivo. Besides pharmacokinetic profiles, CRV conjugation increased the amount of the whole conjugate and free PSL in the ALI lungs but not healthy ones. For healthy organs, CRV conjugation reduced the amount of free PSL in the liver and had little effect in others. Overall, CRV conjugation was able to prolong the plasma half-life of PSL and shift the drug biodistribution more toward the target tissue (inflamed lung), both of which may result in a better therapeutic outcome with fewer side effects. Therefore, we next evaluated the therapeutic efficacy of PSL conjugates against ALI.

Judging by the intensity of lung injury and the expression of inflammatory markers, CRV-PSL showed the highest activity to reduce the lung inflammation and injury of ALI mice. Cys-PSL also showed a better therapeutic efficacy than free PSL, likely due to the prolonged half-life in the blood. CRV conjugation provided the highest amount of free PSL in the ALI lung, and thus CRV-PSL showed the highest activity in reducing lung injury and inflammation compared with Cys-PSL and PSL. These results demonstrate that targeting is indeed needed for the best efficacy, at least in this case. Besides the efficacy in reducing lung inflammation, one important parameter to evaluate the success of CRV-PSL is the side effects in healthy organs, which has been the main barrier to clinical use of glucocorticoids. Here, we mainly focused on the acute side effects such as thymus and lymphocyte toxicity, which is relevant to the clinical scenario when treating ALI/ARDS. Our studies showed that while PSL causes toxicity indicated by decreased thymus weight and lymphocyte count, CRV-PSL can reduce this damage back to untreated levels (Figure 5 and Table S1). Cys-PSL also showed reduced toxicity compared with PSL, likely due to less accumulation in healthy organs. Overall, our CRV conjugation enables a more selective drug accumulation in disease sites and drug release only after entering the target cells. Thus, CRV-PSL exhibits stronger efficacy in attenuating inflammation and lung injury and nearly eliminates the acute side effects of PSL. In the future, this strategy may be applied to other drugs for ALI/ARDS treatment and other diseases treatable by glucocorticoid drugs in order to improve drug efficacy and/or attenuate side effects over the short and long term.

Materials and methods

Chemicals and reagents

Carboxyfluorescein-conjugated peptides <FAM>-<Ahx>-CRVLRSGSC (FAM-CRV) and <FAM>-<Ahx>-GGSGGSKG (FAM-GGS), <FAM>-Cys, scrambled CRV <FAM><Ahx>-CSRSLGVRC (FAM-sCRV), and <FAM>-Cys-<Ahx>-CRVLRSGSC (FAM-Cys-CRV) were purchased from LifeTein (Somerset, NJ). Maleimide-PEG2000-hydroxyl (HO-PEG2000-MAL) was purchased from Nanosoft Polymers (Winston-Salem, NC). Prednisolone was purchased from Acros Organics (Carlsbad, CA). 2,2′-Thiodiacetic acid, N,N′-dicyclohexylcarbodiimide (DCC), 4-dimethylaminopyridine (DMAP), and N,N-dimethylformamide (DMF) were purchased from Sigma-Aldrich (St. Louis, MO). All other solvents and reagents used were of analytical quality. RPMI-1640 medium was from Sigma-Aldrich. All reagents and compounds were used without further purification or modification.

Synthesis of CRV-PSL conjugates

FAM-Cys-CRV-PSL (FAM-CRV-PSL) and FAM-Cys-PSL were designed and synthesized as shown in Figures S5A and S5B. In brief, to construct the ROS-responsive drug delivery system, equimolar of PSL and 2,2′-thiodiacetic acid were solved in DMF with magnetic stirring under 80°C followed by adding the solution of 1.5 equivalents of DMAP and 3 equivalents of DCC in DMF. After overnight reaction, the DCC-derived urea was removed by a cellulose/cotton filter. The product was purified by a preparative HPLC (mobile phase: acetonitrile [ACN] and water; gradient elution: 10% ACN → 90% ACN; column: Luna 5 μm 250 × 10 mm C18). The dried product was mixed with equimolar of HO-PEG2000-MAL and solved in DMF followed by 1.5 equivalents of DMAP and 3 equivalents of DCC in DMF with magnetic stirring at room temperature. After 2 h of reaction, the DCC-derived urea was removed by a cellulose/cotton filter. To precipitate the product, 30–40 times the volume of ethanol and diethyl ether were added and washed three times. The sedimentary product was dissolved in PBS with 10% DMF and then mixed with 3 equivalents of FAM-Cys/FAM-Cys-CRV. The carboxyfluorescein dye FAM was attached through the linker to the N terminus of CRV peptide for tracking. After 2 h of reaction, the final product (prodrug) was purified with a 2 kDa dialysis bag and dried by lyophilization. The chemical structures of the prodrugs were verified by ¹H NMR for molecular identifications.

Cell culture

RAW 264.7 cells were purchased from American Type Culture Collection (ATCC; Manassas, VA). The cells were cultured in RPMI-1640 medium containing 50 U mL⁻¹ streptomycin, 100 U mL⁻¹ penicillin, and 10% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA). THP-1 cells were purchased from ATCC (TIB-202™) and cultured in RPMI 1640 medium as above. To induce differentiation, the THP-1 cells were incubated with 100 ng mL⁻¹ phorbol-12-myristate-13-acetate (Sigma-Aldrich) for 24 h. The cells were cultured at 37°C in a humidified incubator with 5% CO₂. All cell cultures were maintained in 25-cm² or 75-cm² cell culture flasks, or 10-cm culture dishes, for use.

Animals

All animal studies were carried out in compliance with the National Institutes of Health guidelines and an approved protocol from University of Minnesota (UMN) Animal Care and Use Committee. The animals were housed in a specific pathogen-free facility with free access to food and water at the Research Animal Resources facility of UMN.

C57BL/6Ncrl mice were purchased from Charles River Laboratory (Wilmington, MA). The LPS-induced ALI model was established as previously described.¹³ In brief, C57BL/6Ncrl male mice at 8–9 weeks old were given intratracheally 25 μg of LPS (from Escherichia coli O111:B4, Sigma) in 50 μL of PBS. Sham-operated animals underwent the same procedure with intratracheal injection of PBS as controls. The mice were euthanized at 4 days after LPS challenge for tissue collection and analysis.

Human lung of interstitial lung disease

The study using lung tissue from human subjects with ILD was approved by the UMN IRB #STUDY00009145. The lung tissues of human subjects were obtained from the Clinical & Translational Science Institute of UMN. The tissues were collected from the patients with ILD by clinical diagnosis, and hematoxylin-eosin (H&E) slides associated with the case were sent to pathology in UMN for a quality control (QC) to confirm diagnosis. The controls were normal lungs obtained after a QC process by histology.

HPLC-MS/MS assay

HPLC-tandem mass spectrometry (MS/MS) analysis was performed using the chromatographic system consisting of an UltiMate 3000 RSLCnano System (Thermo Fisher Scientific) and a ZORBAX C18 column (5 μm, 150 mm × 0.5 mm; Agilent, Santa Clara, CA). The mobile phase was a mixture of H₂O/acetonitrile with a gradient elution (from 90:10 to 10:90 and return to 90:10, v/v) at a flow rate of 15 μL min⁻¹. The eluent was introduced directly into the electrospray source of a tandem quadrupole mass spectrometer (TSQ VANTAGE, Thermo Fisher Scientific) that was operated in the positive mode. The spray voltage was set at 3,000 V. The compounds were analyzed by multiple reaction monitoring (MRM) of the transitions of mass to charge ratio (m/z) (ESI+) 361.24 → 147.06 for PSL.

Intracellular drug release

Intracellular ROS-responsive drug release of CRV-PSL was investigated by HPLC-MS/MS assay. The confluent RAW cells were treated with 10 μM free PSL or CRV-PSL, with or without 10 ng mL⁻¹ LPS, for 4 h. After incubation, the cells were washed twice with cold PBS. The cells were then mixed with 500 μL of methanol/dichloromethane (DCM) (1:1) by vortex for 1 min and centrifuged at 16,000 × g for 10 min. The upper liquid layer was collected, blown dry by nitrogen, and dissolved in 20 μL of a mixture of ACN/water (90:10, v/v), and 10 μL was injected into the system for HPLC-MS/MS analysis. The PSL metabolites were analyzed by MRM of the transitions of m/z (ESI+) 363 → 345 for 20-α-OH-PSL and 20-β-OH-PSL.

In vivo pharmacokinetics assay

Healthy C57BL/6 male mice (8–9 weeks old; 25–30 g body weight) were deprived of food overnight with free access to water. The mice were randomly divided into three groups and injected via lateral tail vein with free PSL, Cys-PSL, and CRV-PSL with an equivalent dose of 1 mg kg⁻¹ PSL, respectively. Approximately 100 μL of blood was collected into tubes via venous plexus at 5, 10, 30, and 60 min and 2, 6, 12, 24, and 48 h, then centrifuged at 500 × g for 10 min to obtain plasma. Ten microliters of 0.1 M sodium hydroxide were added and vortexed for 30 s. The plasma samples were then mixed with 200 μL of methanol/DCM (1:1) by vortex for 1 min and centrifuged at 16,000 × g for 10 min. The upper liquid layer was collected, blown dry by nitrogen, and dissolved in 100 μL of ACN for HPLC-MS/MS analysis.

In vivo drug distribution

After the establishment of ALI by intratracheal injection of LPS, C57BL/6 male mice (8–9 weeks old, 25–30 g body weight) were deprived of food overnight but with free access to water. The mice were randomly divided into three groups and were given free PSL, Cys-PSL, and CRV-PSL with an equivalent dose of PSL at 1 mg kg⁻¹ via intravenous injection, respectively. After 2 h, the mice were euthanized by transcardial perfusion with PBS and tissues including the heart, lung, liver, spleen, and kidney were harvested and weighed. The tissues were then homogenized and dissolved in 200 μL of methanol/DCM (1:1). After a 1-min vortex followed by centrifugation at 16,000 × g for 10 min, the upper layer was taken and blown dry by nitrogen. The sample was then dissolved in 100 μL of ACN for HPLC-MS/MS assay.

Real-time RT-PCR

Total RNA was extracted from cells with TRI reagent (Sigma-Aldrich). First-strand cDNA was synthesized from total RNA using an iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA). Quantitative amplification by PCR was performed using PowerUp SYBR Green qPCR Master Mix (Thermo Fisher Scientific) by a StepOne Real-time PCR System (Applied Biosystems, Foster City, CA). The ΔΔCt method was used to calculate the results. The housekeeping gene mouse TATA binding protein (TBP) or human ribosomal protein L37a (RPL37A) was used as an endogenous control for quantification as appropriate. The primers were as follows. Mouse (m)TBP: 5 GAA GAA CAA TCC AGA CTA GCA GCA 3 and 5 CCT TAT AGG GAA CTT CAC ATC ACA G 3; mIL-1β: 5 CCT TCC AGG ATG AGG ACA TGA 3 and 5 TGA GTC ACA GAG GAT GGG CTC 3; miNOS: 5 GAG ACA GGG AAG TCT GAA GCA C 3 and 5 CCA GCA GTA GTT GCT CCT CTT C 3. mTGF-β1: 5 AAA CGG AAG CGC ATC GAA 3 and 5 GGG ACT GGC GAG CCT TAG TT 3; mIL-6: 5 CCA GAA ACC GCT ATG AAG TTC CT 3 and 5 CAC CAG CAT CAG TCC CAA GA 3; mMCP-1: 5 GTT GGC TCA GCC AGA TGC A 3 and 5 AGC CTA CTC ATT GGG ATC ATC TTG 3; mColI-α1: 5 CAA CCT GGA CGC CAT CAA G 3 and 5 CAG ACG GCT GAG TAG GGA ACA 3; human (h)RPL37A: 5 ATT GAA ATC AGC CAG CAC GC 3 and 5 AGG AAC CAC AGT GCC AGA TCC 3; hTNF-α: 5 CCT CTC TCT AAT CAG CCC TCT G 3 and 5 GAG GAC CTG GGA GTA GAT GAG 3; hIL-6: 5 ACT CAC CTC TTC AGA ACG AAT TG 3 and 5 CCA TCT TTG GAA GGT TCA GGT TG 3; hMCP-1: 5 GAT CTC AGT GCA GAG GCT CG 3 and 5 TTT GCT TGT CCA GGT GGT CC 3.

Flow cytometry

To identify subpopulations in myeloid cells, the single-cell suspensions were collected from the lung with a lung dissociation kit for mouse (Miltenyi Biotec, Auburn, CA) and other tissues including blood and spleen for flow-cytometry analysis. The single-cell suspension with 0.5–1 × 10⁶ cells from tissue digests were stained with fluorochrome-conjugated antibodies as indicated in the results section. The antibody panel is shown in Table S2. Flow-cytometry analysis was performed on BD Fortessa X-20 (BD Biosciences, East Rutherford, NJ). Data were analyzed with FlowJo software (Tree Star, San Carlos, CA).

Western blotting

Fifty micrograms of protein from each cell lysate was subjected to electrophoresis in SDS-PAGE on 4%–15% precast protein gels (Bio-Rad, Hercules, CA) and then transferred to polyvinylidene fluoride (0.2 μm pore size) membranes using a miniblot apparatus (Bio-Rad). The membranes were probed with primary antibodies including mouse antibodies to IL-1β and tubulin (Cell Signaling Technology, Danvers, MA); and rabbit polyclonal antibody to iNOS (Thermo Fisher Scientific) and monoclonal antibody to glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Cell Signaling) as appropriate with constant shaking overnight at 4°C. After washing, the bound antibodies were detected with the secondary antibodies IRDye 680RD donkey anti-rabbit immunoglobulin G (IgG) and IRDye 800CW donkey anti-mouse IgG (Li-COR, Lincoln, NE). The membranes were scanned via the Odyssey CLx imaging system (LI-COR), and the images with blots were quantified by ImageJ software.

H&E staining and lung injury score

The lung tissues were harvested from ALI mice after 3 days of treatments with free PSL, Cys-PSL, and CRV-PSL, respectively. In brief, the formalin-fixed tissues underwent tissue processing and then were embedded in paraffin to create a formalin-fixed, paraffin-embedded block for histology. H&E staining was performed on the paraffin-embedded sections to assess the morphologic changes in injured lung tissue. After deparaffinization and rehydration, 4-μm sections of the lung tissues were stained with H&E according to standard procedures. To evaluate the lung injury score, each slide was examined by an independent investigator in a blinded manner. A total of 300 alveoli were counted on each slide at 40× magnification as described previously.³⁰ The injury score was calculated according to the following formula: injury score = [(alveolar hemorrhage points/no. of fields) + 2 × (alveolar infiltrate points/no. of fields) + 3 × (fibrin points/no. of fields) + (alveolar septal congestion/no. of fields)]/total number of alveoli counted × 100.

Immunohistochemistry staining

IHC staining was performed on the paraffin-embedded sections. In brief, after deparaffinization and rehydration, all sections were incubated with 0.3% H₂O₂ solution for blockade of endogenous peroxidase activity. The sections were then blocked with 5% donkey serum blocking solution (with 0.1% Triton X-100), followed by incubation with primary antibodies overnight at 4°C. The primary antibodies included polyclonal rabbit anti-RXRB (GeneTex, Irvine, CA), anti-fluorescein/Oregon green (Invitrogen, Waltham, MA), anti-iNOS (PA1-036), anti-myeloperoxidase antibody (PA5-16672, Thermo Fisher Scientific), and rat anti-CD11b and CD64 (Thermo Fisher Scientific). After treatment with secondary anti-rabbit (horseradish peroxidase [HRP]) antibody for 1 h and then 3,3′-diaminobenzidine (DAB) peroxidase (HRP) substrate (Vector Labs, Burlingame, CA) for 1 min, hematoxylin counterstaining was performed followed by dehydration in ethanol and xylene, and mounting with Permount (Thermo Fisher Scientific). Images of randomly selected areas from each section were collected by the microscope. Five views per tissue were captured from each section for semi-quantification by ImageJ software according to the HRP-DAB signal.

Immunofluorescence staining

IF staining was performed on the paraffin-embedded or frozen tissue sections. In brief, the sections were first treated with PBS containing 1% BSA and 0.1% Triton X-100 (blocking buffer) at room temperature for 1 h. The sections were then washed three times with PBS followed by the incubation with primary antibodies with a 1:200 dilution in blocking buffer at 4°C overnight, followed by the appropriate secondary antibodies diluted (1:200) in blocking buffer at room temperature for 1 h. The primary antibodies are as follows: rabbit anti-fluorescein/Oregon green (Invitrogen), anti-RXRB (GeneTex), rat anti-CD11b, CD64, and F4/80 (Thermo Fisher Scientific). After washing with PBS, sections were stained with Hoechst or DAPI for nuclei, mounted in mounting medium, and covered with a coverslip. The sections were examined under the fluorescence microscope EVOS M5000 (Thermo Fisher Scientific).

TUNEL staining

TUNEL staining was performed using an In Situ Cell Death Detection Kit (Roche Diagnostics, Mannheim, Germany). The paraffin sections were stained with TUNEL reaction mixture for 1 h at 37°C followed by Hoechst staining for nucleus. The images were captured by using the EVOS M5000 fluorescence microscope, and the fluorescence intensity of TUNEL-positive cells was evaluated by ImageJ software for semi-quantification.

Biochemical toxicity analysis

Aminotransferase quantification was performed to determine toxicity effect of different formulations of PSL on the liver. Blood samples were collected by cardiac puncture after the treatment. The levels of AST and ALT in plasma were measured using assay kits from Stanbio (Boerne, TX).

Statistical analysis

A test of normal distribution of the variants was performed, showing that the variants were normally distributed before one-way analysis of variance or Student’s t test, where appropriate, to assess differences among groups. The qPCR results were expressed as means ± standard error of mean (SEM). Other results were expressed as mean ± standard deviation (SD). A p value of less than 0.05 was considered significant.

Data availability

All data generated or analyzed in this study are included in this published article and its supplemental information files.