Parathyroid hormone (PTH) facilitates macrophage efferocytosis in bone marrow via pro-resolving mediators Resolvin D1 and Resolvin D2

Abstract

Bone marrow macrophages stimulate skeletal wound repair and osteoblastic bone formation by as yet poorly defined mechanisms. Specialized pro-resolving mediators of inflammation drive macrophage efferocytosis (phagocytosis of apoptotic cells) and resolution but little is known regarding this process in the bone marrow. In the present report, metabololipidomic profiling via liquid chromatographic mass spectrometry revealed higher levels of specialized pro-resolving mediators (SPM) in the bone marrow relative to the spleen. The endocrine and bone anabolic agent parathyroid hormone (PTH) increased SPM levels including resolvins (Rv) in bone marrow. Human and murine primary macrophages efferocytosed apoptotic osteoblasts in vitro, and resolvin (Rv) D1 and RvD2 (10 pM-10 nM) enhanced this process. These findings support a unique profile of specialized lipid mediators in bone marrow that contributes to a feedback system for resolution of inflammation and maintenance of skeletal homeostasis.

Introduction

A primary function of bone is to house marrow, the major hematopoietic organ in mammals. Bone as an organ is composed of various cell types including those traditionally considered ‘bone’ cells (osteoblasts, osteoclasts, osteocytes), hematopoietic cells (hematopoietic stem, progenitor cells, megakaryocytes), immune cells (lymphocytes, macrophages, dendritic cells), and stromal cells (mesenchymal stem cells). Osteoblasts are ‘repair’ cells responsible for development, secretion and extracellular matrix mineralization. Mesenchymal stem cells are recruited to bone surfaces where they differentiate into osteoblasts. Once committed to their lineage, osteoblasts are destined to embed into the mineralized matrix as osteocytes, become flattened lining cells, or apoptose. Studies have evaluated the trajectory toward these fates, but none have detailed the aftermath of apoptotic osteoblasts. Inefficient apoptotic cell clearance leads to a local inflammatory environment (1). Hence, there are likely mechanisms for the efficient removal of apoptotic osteoblasts and a likely facilitator is the resident macrophage. The term ‘osteomac’, has been attributed to macrophages credited with stimulating bone formation through mechanisms that link to their phagocytic capacity (2). Macrophages facilitate apoptotic cell clearance in a process termed efferocytosis, which converts the dead and dying cell toxic environment to an anti-inflammatory environment.

With the identification of resolution phase mediators, recent efforts have focused on the resolution phase, an active process orchestrated by special mediators that direct cellular and biochemical pathways to facilitate return to homeostais (3). Such specialized pro-resolving mediators of inflammation include lipoxins (LX), resolvins (Rv), protectins (PD) and maresins (MaR)(4). Among other actions, these mediators are potent stimulators of macrophage efferocytosis but little is known about their role in bone remodeling. The purpose of this investigation was to determine the lipid mediator profiles of two lymphoid organs: bone marrow and spleen, assess their modulation by parathyroid hormone (PTH), and determine their ability to facilitate macrophage efferocytosis of apoptotic osteoblasts. The implications identify a new feedback process in bone which has potential impact in normal bone remodeling as well as wound healing in bone.

Materials and Methods

In vivo models and lipidomics

Male C57B6 mice (4–5wks; Jackson Labs, Bar Harbor, ME) were injected once with rhPTH 1–34 (50 μg/kg) (Bachem, Torrance, CA) or vehicle (saline) 2h prior to sacrifice. Spleens and long bones were frozen in liquid nitrogen then placed in ice cold methanol containing dueterated internal standards (d₈-5S-hydroxyeicosatetraenoic acid (HETE), d₄-leukotriene (LT) B₄, d₄-prostaglandin (PG)E₂ and d₅-lipoxin (LX) A₄; 500pg each) and homogenized using a PTFE dounce (Kimble Chase). Proteins were precipitated (4°C), solid-phase extracted using Biotage RapidTrace_®⁺(5), and analyzed using liquid chromatography-ultraviolet-tandem mass spectrometry, QTrap 5500 (ABSciex, Framingham, MA) equipped with an Agilent HP1100 binary pump (Santa Clara, CA). An Agilent Eclipse Plus C18 column (100mm × 4.6 mm × 1.8 μm) maintained at 50°C was used with a gradient of methanol/water/acetic acid of 55:45:0.01 (v/v/v) to 100:0:0.01 at 0.4 ml/min flow rate. Multiple reaction monitoring (MRM) with signature ion fragments for each molecule was used with six diagnostic ions employed for identification, and quantification achieved using calibration curves (5). Principal component analysis (PCA) was performed using SIMCA 13.0.3 software (Umetrics, Umea, Sweden) following mean centering and unit variance scaling of lipid mediators (LM).

Human mononuclear cell isolation, culture, and efferocytosis assays

Peripheral human blood mononuclear cells (PBMCs) were isolated from human venous blood by density gradient centrifugation (6). Blood was obtained from healthy human volunteers under a Partners Human Research Committee approved protocol. Cells were cultured in RPMI with 10% FBS, Pen/Strep and GM-CSF (R&D Systems) (10ng/ml) 7d prior to efferocytosis assays. Macrophages differentiated from PBMCs were plated overnight (96 well plates; 50,000/well) then incubated with 1 pM-100 nM of RvD1 or RvD2 15mins prior to the addition of apoptotic human osteoblasts (5:1 osteoblasts/macrophages). Human fetal osteoblasts (hFOB, ATCC, Manassas, VA) cultured in DMEM F12 media (10% FBS 1% P/S, glutamine) (Invitrogen, Life Technologies, Grand Island, NY), were labeled with carboxyfluorescein diacetate, succinimidyl ester (CFSE, 3 μM) (Invitrogen) 15mins prior to induction of apoptosis via ultraviolet radiation. Cells were exposed to UV light for 30mins then returned to 37°C for 2h, harvested and enumerated via trypan blue dye exclusion. Greater than 90% of cells were trypan blue positive reflecting cell death. The hFOB cells were resuspended in PBS and incubated with macrophage cultures 4h, followed by incubation with trypan blue to quench extracellular fluorescence and exclude cell associated non-phagocytosed cells (i.e. not internalized), prior to rinsing and scanning at 487–517 nM on a microplate reader (SpectraMax M3, Molecular Devices, Sunnyvale, CA). Results are expressed as the fold increase in CFSE above control, as a reflection of cell engulfment.

Murine macrophage efferocytosis

Marrow was flushed from 4wk C57B6 mice, expanded with 30ng/ml M-CSF (eBioscience) in α-MEM media(10%FBS P/S, glutamine) 5–7d to enrich for macrophages, plated at 10,000/well in 8-well chambers, attached overnight, stained 15min with 2μM Green BODIPY in PBS, then returned to complete medium and incubated 1hr to finalize dye crosslinking. MC3T3-E1 cells (previously stained as above with Red CMPTX, Invitrogen) were induced to undergo apoptosis for 48h with 50mM etoposide (Sigma, St. Louis, MO) and 40,000/well were co-incubated with macrophages for 3h. Cells were washed, fixed with ice cold methanol, washed, then overlayed with Vectashield (Vector Laboratories, Burlingame, CA) to visualize DAPI nuclei. Z-stack images were performed on a Leica Inverted SP5X Confocal Microscope System using Leica Application Suite Software (Leica, Wetzlar, Germany) and compiled to 3D movie format using Imaris (Bitplane Scientific Software, South Windsor, CT). Videography was performed using the DeltaVision RT Live Cell Imaging System with SoftWoRx 3.5.1 Software (Applied Precision, Issaquah, WA).

Results and Discussion

Lipid mediator profiles from spleen and bone were obtained and compared using LC-MS-MS based lipid mediator metabololipidomics. Specific mediators from the arachidonic acid (AA), docosahexaenoic acid (DHA), and eicosapentaenoic acid (EPA) bioactive metabolomes were identified including resolvin (Rv) D1 and RvD2 from DHA, RvE1 and RvE3 from EPA and LXA₄ and LXB₄ from the AA bioactive metabolomes (Fig 1). These mediators were identified with matching retention times, MS-MS fragmentation patterns as well as at least 6 characteristic and diagnostic ions as illustrated for RvD1 and RvD2 (Supplemental Fig 1). Quantification of individual lipid mediators demonstrated elevated AA, EPA and DHA-derived pro-resolving mediator levels in bone marrow from unchallenged mice compared to spleens. RvD2, RvE1and LXB₄ among others were significantly elevated in bone marrow. Principle component analysis demonstrated that lipid mediator profiles obtained with mouse bone marrows clustered separately from those obtained with mouse spleens (Fig. 1A) e.g. with Protectin D1 with spleen and RvD2 with bone marrows (Fig 1B). These results expand recent identifications of lipid mediator biosynthesis in murine spleen (7) and bone marrow (8) with the identification of novel SPMs that with the present results include RvD2.

Figure 1. Pro-resolving lipid mediator levels in hematopoietic tissues.

Lipid mediator levels were assessed following solid phase extraction by LC-MS-MS based lipid mediator metabololipidomoics. (A) 3D score plot of bone marrow (green) and spleen (purple) endogenous lipid mediators, (B) 3D loading plot displaying the mediators that correlate with bone marrow (green) and spleen (blue), n=5/gp, (C, D) Mice were administered vehicle (CT) or PTH (50 mg/kg) and 2h later bone marrow lipid mediator levels were assessed, (C) 3D score plot of mouse bone marrow lipid mediator profiles for vehicle (purple) and PTH (green) treated mice. Ellipses in A, C denote 95% confidence region. (D) 3D loading plot displaying the mediators that correlate with vehicle (purple) and PTH (green) treated mice, n=8–9/gp.

The spleen and bone marrow serve as stem and leukocyte cell reservoirs with an important difference that the marrow is enclosed by a skeletal boundary and hence hematopoietic and leukocytic cells are juxtaposed with osteoblasts, osteoclasts, and a mineralized extracellular matrix. That prostaglandins were increased in bone marrow versus spleen was not surprising as years of work have shown the diverse actions of prostaglandins in bone (9). Unique to this study, were findings that members of the DHA bioactive metabolome, i.e. the resolvins were higher in marrow. This suggests bone is poised to facilitate resolution of inflammation. Indeed, injurious and infectious insults to the marrow are more deleterious than to the spleen and hence protective mechanisms for resolving inflammation in a timely manner would be essential for survival.

In the present study, PTH administration led to a rapid (2h) and selective increase in pro-resolving lipid mediator levels in marrow including RvD1 and RvD2, giving characteristic LM profiles as demonstrated by PCA analysis (Fig 1C, D, 2 and Supplemental Fig 1), but no significant difference in the spleen (data not shown). This regulation included D-series and E-series resolvins. The D-series resolvins facilitate macrophage efferocytosis of apoptotic neutrophils and reduce the resolution interval post inflammatory challenge (5). Of note, statistically significant differences were not observed for LM levels 5 days after PTH administration (data not shown).

Figure 2. Upregulation of bone marrow pro-resolving lipid mediators by PTH.

Mice were administered vehicle (CT) or PTH (50 mg/kg) and 2h later femurs and tibias were collected and lipid mediator levels assessed following solid phase extraction by LC-MS-MS based lipid mediator metabololipidomoics. Bioactive lipid mediator families significantly increased by PTH are shown as mean ± SEM. n= 8–9 mice per group, Results are expressed as mean ± SEM, n=4–5/gp. *P<0.05, **P<0.01 vs CT mice

One of the prime actions of SPMs is to facilitate macrophage efferocytosis (1,3). Efferocytosis studies most commonly involve macrophage clearance of apoptotic neutrophils and results in macrophage production of anti-inflammatory factors such as TGFβ (10). If dying cells are not cleared, their intracellular contents are expelled creating an unfavorable environment (11). In the bone marrow, the fate of osteoblasts that undergo apoptosis is unclear. Experiments performed where labeled osteoblasts were incubated with macrophages identified that macrophages rapidly efferocytose apoptotic osteoblasts (Fig 3, Videos 1–5).

Figure 3. Murine macrophage efferocytosis.

A) Four microscopic still images from video with macrophages (Green BIODIPY label) phagocytosing apoptotic osteoblasts (white arrow, Cell Tracker orange label) taken over 24hrs, B) Images from co-incubation of apoptotic osteoblasts (Red CMPTX label with blue DAPI nuclear stain) and bone marrow macrophages (Green BIODIPY label) at various states of macrophage ingestion of apoptotic osteoblasts from left (early) to right (late). Supplemental materials include video for A (Video 1) and 3D movie format for B (Videos 2–5).

Studies to determine whether increased bone marrow SPMs facilitate macrophage efferocytosis revealed RvD1 and RvD2 increased uptake of apoptotic osteoblasts (Fig 4). Of note, RvD1 (100 pM-10 nM) and RvD2 (100 pM-100 nM) increased efferocytosis of osteoblasts (30%–60%) to a similar extent as efferocytosis of apoptotic PMNs (~30%)(12). These experiments demonstrate the ability of macrophages at physiologically relevant concentrations based on our LC-MS-MS quantification to engulf apoptotic osteoblasts, a process facilitated by resolvins produced in the local environment. Other studies have identified that RvD1 reduced macrophage derived TNFα promoting the resolution of inflammation and that the process of efferocytosis involves subsets of macrophages reprogrammed from CD11b(high) to CD11b(low) (13,14).

Figure 4. Resolvins increase macrophage efferocytosis.

Macrophage enriched PBMCs were incubated with concentrations of RvD1 (open circles) or RvD2 (closed squares) 15mins prior to the addition of CFSE-labeled apoptotic hFOB osteoblasts (4h), then rinsed and scanned at 487–517 nM. Results are expressed as the CFSE levels above control, as a reflection of cell engulfment, for RvD1 or RvD2. Experiments were performed in quadruplicate for each donor and dose and data shown as mean ± SEM from 6–7 donors performed at different times. *RvD1 or **RvD2 <0.05 vs. 1 pM dose.

The process of bone remodeling has been likened to inflammation where the inflammatory stage of ‘injury’ correlates with pressure and microfracture. The ‘reaction’ stage correlates with osteoclastic bone resorption, and the ‘repair’ stage with osteoblastic bone formation (15). PTH is a potent inducer of bone remodeling and well known to increase bone formation through mechanisms that are still not entirely clear. PTH drives robust increases in bone accrual during wound healing (16,17) and hence our interest was in determining whether PTH was associated with lipid mediators responsible for the resolution of inflammation. PTH stimulates IL-6 and CCL2 production from osteoblasts which recruits and activates myeloid cells (18,19). During wound healing macrophages phagocytose debris and apoptotic cells. PTH has robust anabolic actions during traumatic injuries associated with increased apoptosis such as bone marrow ablation (20). Hence, PTH could act in the marrow to recruit macrophages and promote the release of specialized pro-resolving mediators that facilitate efferocytosis. Such a proresolving circuit would be enhanced when PTH is administered during osseous wound healing where PTH then expedites skeletal homeostasis in favor of anabolism. In summary, the bone marrow has a unique profile of specialized lipid mediators and PTH increases select SPMs. Bone marrow macrophages efferocytose apoptotic osteoblasts which is facilitated by RvD1 and RvD2. Taken together with reports that macrophages produce TGFβ during efferocytosis, and that TGFβ recruits mesenchymal stem cells to renew osteoblasts (21), these results suggest a new feedback system for the resolution of inflammation and restoration of skeletal homeostasis.