Cell-cell signaling in co-cultures of macrophages and fibroblasts

Abstract

The foreign body response (FBR) comprises a general, ubiquitous host tissue-based reaction to implanted materials. In vitro cell-based models are frequently employed to study FBR mechanisms involving cell signaling responses to materials. However, these models often study only one cell type, identify only limited signals, and cannot accurately represent the complexity of in vivo inflammatory signaling. To address this issue, a cell co-culture system involving two primary effector cells of the FBR, macrophages and fibroblasts, was employed. Cell-cell signaling systems were monitored between these cell types, including long-term 1) culture of one cell type in conditioned media from the other cell type, 2) non-contacting cell co-cultures (paracrine signaling), and 3) contact co-cultures (juxtacrine signaling) of primary- and secondary-derived cells. Cell culture media and cell images were collected on Days 1, 2, 3, 7, 14, and 21 and changes in soluble protein secretion, cellular behavior, and morphology were assessed. Primary- and secondary- derived cells responded uniquely during each signaling scenario and to one another. In general higher in vitro fidelity to FBR-like responses was found in primary cell co-cultures compared to their mono-cultures and all secondary cell cultures.

1. INTRODUCTION

Implantation of any foreign material into living tissue evokes a host inflammatory response generally described as the foreign body response (FBR). The FBR -- a cascade of cell-based soluble signaling events -- reacts to and modulates the interface between an implanted device and the host tissue. This host response is associated with numerous complications including fibrosis, bone resorption, implant degradation, increased infection rates, delayed healing, pain, and general device failure [1–3].

In vitro models are commonly employed to study various mechanistic aspects of the FBR. In contrast to complex in vivo models, cell culture in various forms provides a simplified, more cost-effective and focused analysis, utilizing fewer animals and enabling higher throughput of biomaterials assays. However, as a reductionist approach, these cell culture systems are perhaps overly simplified and inaccurate, examining only limited numbers and/or types of cells and their responses, and missing much of the essential positive and negative feedback signals from key cells necessary to faithfully duplicate in vivo aspects of the FBR. Consequently, in vitro models may be largely incapable of predicting and correlating in vivo phenotypes. For example, a significant inconsistency between in vitro and in vivo models is the observed lack of inflammatory cell (e.g., macrophage) activation in vitro [4–6]. This has prompted the standard practice of adding lipopolysaccharide (LPS) [7] or other exogenous activating agents such as IFNγ [8] or phorbol esters [9] to replicate cell activation seen in vivo [10] around implants. It has been suggested that the mere presence of a material in vitro in cell culture may be insufficient to mount a meaningful cell inflammatory response without the use of LPS [5]. However, even LPS stimulation in vitro produces an attenuated response compared to in vivo LPS activation. Reichert and co-workers showed that in vitro production of the inflammatory cytokine, IL-6, by secondary macrophages after LPS stimulation appeared to be non-responsive compared to untreated groups [6]. Their subsequent study showed that in vivo stimulation with LPS produced a drastic increase in IL-6 production compared to the non-stimulated group, supporting the contention that in vitro and in vivo responses were inconsistent [11]. Similarly, Roumestan et. al. showed that after LPS stimulation, in vivo production of inflammatory cytokine, TNFα, was nearly double that secreted by in vitro primary monocytes, underscoring the inconsistency between in vivo and in vitro models [12]. Understanding the differences between in vivo and current in vitro models, and providing some basis for improvements serve as motivation for this study.

Further compounding translation between in vitro and in vivo models are variations seen in vitro between primary- and secondary- derived cells used almost interchangeably in literature reports for decades. Primary cells are derived from living tissue possess limited passaging capacity (i.e., Hayflick limit [13]), and are therefore best used almost immediately and for limited proliferative cycles. Secondary cells (cell lines) on the other hand are immortalized (often through genetic transformation), can be passaged and stored theoretically indefinitely, are frequently contaminated with other more aggressive but irrelevant cell types such as HeLa and K562 [14], and often lack much phenotypic semblance to their primary derived counterparts. For example, murine RAW 264.7 macrophages are immortalized, transformed monocyte/macrophage-like cells frequently utilized in inflammatory assays [15–17]. However, these secondary-derived immortalized cells are poorly validated and display cellular responses distinct from primary cells, such as types and levels of external receptor upregulation, excessive internalization rates, rapid proliferation with lack of contact inhibition, and distinct cytokine production dynamics [4].

Macrophages and fibroblasts are primary FBR effector cells acting in concert in local implant-associated inflammation, cell recruitment, implant degradation, fibrosis, and chronic unresolved healing [1, 18]. Proinflammatory cytokines secreted by both macrophages and fibroblasts are immediately upregulated post-injury and remain upregulated in the presence of a foreign material [5, 11]. These soluble signals are recognized by the same cell in autocrine, and neighboring cells in paracrine, responses [19]. Macrophages and fibroblasts in the FBR communicate via soluble autocrine and paracrine signals as well as juxtacrine signals associated with direct cell-cell contacts. Hence, both chemical and physical cues exchanged between macrophages and fibroblasts at the implant site modulate cell migration, proliferation, protein synthesis, and enzymatic function associated with the FBR [1, 5, 18, 20]. Fibroblasts, acting on cues from both recruited and tissue-resident macrophages, are thought to synthesize extracellular matrix, specifically collagens and keratins, resulting ultimately in implant fibrosis [1, 18, 20–22]. However, the precise mutual influences of these two cell types in the FBR are currently unknown.

This study analyzes and compares extended-duration in vitro co-cultures of primary and secondary macrophages and fibroblasts to elucidate and distinguish signaling patterns between these cell types and determine relevant feedback systems that may enable more representative in vitro cellular responses to in vivo models, or provide evidence for confounding in vitro cell behavior.

2. METHODS AND MATERIALS

2.1. In vitro secondary cell culture

Transformed murine secondary monocyte/macrophage-like cell line RAW 264.7 and fibroblast-like cell line NIH 3T3 were purchased from the American Type Culture Collection (TIB-71 for RAW and CRL-1658 for 3T3 ATCC, Manassas, USA). Frozen stocks were suspended in 20 ml of pre-warmed secondary cell control media (Dulbecco’s modified eagle medium (DMEM) with 10% fetal bovine serum (FBS), 1% antibiotic/antimycotic, and 1% 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), all sourced from Invitrogen, Carlsbad, USA) and cultured in T75 flasks (BD Falcon, San Jose, USA) at 37°C with 5% supplemental CO₂ until 80% confluency before passaging. RAW cells were passaged by scraping with a rubber policeman and used below passage 10. 3T3 fibroblasts were passaged in control media using bovine trypsin (Invitrogen) and used below passage 20. For all studies, cells were seeded into 12-well tissue culture-treated polystyrene plates (BD Falcon, San Jose, USA) and cultured for a period of 21 days in a serum-containing media as specified at 37°C with 5% supplemental CO₂. Mono-cultures in control media, mono-cultures treated with pre-conditioned media (see below), mono-cultures stimulated with LPS (see below), non-contact co-culture (paracrine), and contact-culture (juxtacrine) conditions were designed for two-dimensional (2-D) adherent cell cultures on rigid materials, see Figure 1A. Transwell^® Permeable Supports (3-μm porous plasma-treated polycarbonate inserts) (Corning, Corning, USA) were utilized for paracrine non-contact cell co-cultures. Fibroblasts (1.2×10⁵ per well) were added to mono-culture wells, juxtacrine co-cultures, and in the Transwell^® inserts of paracrine co-cultures. Macrophage-like cells (3.1×10⁵) were added to mono-culture wells, juxtacrine culture wells, and paracrine co-culture wells beneath the Transwell^® inserts containing fibroblasts. Cell seeding concentrations were selected in order for wells to reach confluency by Day 1 of the experiment.

Figure 1.

A) Depiction of the four cell signaling feedback culture scenarios utilized in this in vitro study. B) Description of the culture process implemented to investigate each signaling condition.

Mono-cultured cells (negative controls) and paracrine and juxtacrine co-cultures were grown in control media (Figure 1B). Conditioned FBS-supplemented media (20 ml) was harvested from RAW cell cultures 3 days past confluency and from 3T3 cell cultures 7 days past confluency in T75 flasks. These times were the longest periods the cells could grow before requiring a media change and were thus selected to maximize cell-produced cytokine concentrations. These time points also served to normalize cytokine production as RAW cells are more highly metabolically active than 3T3 cells, as seen by their use of media (i.e. turning the phenol red indicator in the media from red to yellow more rapidly than 3T3s). Each conditioned medium was sterile filtered and 10% v/v was added to the control media for subsequent cell treatment. For positive cell activation controls, 1μg/ml LPS (endotoxin, List Biological Laboratories Inc., Campbell, USA) was added to control media. Media (2ml) in each well was changed daily, and analyzed on Days 1, 3, 7, 13 and/or 14, and 19 and/or 21, and stored at −70°C for subsequent analysis. Media was analyzed on Days 13 and 19 due to fibroblast delamination from surfaces observed beyond those times.

2.2. Murine primary macrophage sourcing

Specific-pathogen-free female C57BL/6 mice, 6 to 8 weeks old, were purchased from Jackson Laboratory (Bar Harbor, USA). Animals were kept in the University of Utah ILAAC-approved animal facility and given water, mouse chow, bedding, and modes of enrichment ad libitum throughout this study.

2.3. Primary cell cultures

Bone marrow cells (BMCs) were collected from femurs and tibias of male C57BL/6 mice and differentiated into bone marrow macrophages (BMMΦs) using a previously described method [23, 24]. On Day 7, cells were removed from plastic culture surfaces by incubation in Ca⁺²/Mg⁺²-free PBS and scraping with a rubber policeman. Cells were spun at 500 rcf for 5 minutes to form a pellet and then resuspended in BMMΦ control media (DMEM with 10% heat inactivated FBS, 1% antibiotic/antimycotic, 1% MEM non-essential amino acids, 1% HEPES, and 1% sodium pyruvate, Invitrogen, Carlsbad, USA). Transwell^® inserts were utilized for non-contacting paracrine co-culture. BMMΦ cells were plated into mono-culture control wells (1.5×10⁵ per well), conditioned media-treated mono-culture wells, wells beneath Transwell^® inserts containing seeded fibroblasts (paracrine co-culture), and into contact culture wells (juxtacrine co-culture). NIH 3T3 cells (1.5×10⁵ per well) were plated into Transwell^® inserts, and in juxtacrine co-cultures. BMMΦs treated with conditioned media were grown in control BMMΦ media supplemented with 10% 3T3-conditioned media. Positive control media for BMMΦs was produced by adding 1 μg/ml LPS to BMMΦ control media (see Figure 1 for pictorial summary representation of culture media and conditions).

2.4. Cytokine secretion assays

BD^™ Cytometric Bead Array (CBA) assay was used to determine RANTES, TNF, MCP-1, MIP1-β, MIP1-α, IL-6, IL-2, IL-4, IL-5, IL-9, IL-10, IL-12P70, IL-13, and IFN-γ cytokine expression profiles over time. Media was collected from cells on Days 1, 2, 3, 7, 13 and/or 14, and 19 and/or 21 to be analyzed. The CBA has been shown to detect comparable levels of cytokines to an Enzyme Linked Immunosorbant Assay (ELISA) [25], and was implemented for this study due to its low technical error (attained by averaging the relative fluorescence of at least 300 beads per analyte) and ability to multiplex multiple cytokines simultaneously.

2.5. Flow cytometry analysis

Data acquisition for the CBA assay was performed using an upgraded 5-color FACScan Analyzer (BD Biosciences, Mountain View, USA), employing a benchtop analyzer with two lasers for fluorochrome excitation. The primary laser is a 15 mW argon (488 nm) laser and the secondary laser is a 25 mW red diode (637 nm) laser. The instrument uses seven detectors, two for light scattering (forward and 90°) and five for fluorescence. CellQuest 3.3 software (BD Biosciences, San Jose, USA) and Rainbow software 1.1 (Cytek, Fremond, USA), was used for data collection. WinMDI 2.9 (J. Trotter, The Scripps Research Institute, La Jolla, CA), Weasel (Walter & Eliza Hall Institute, Melbourne, Australia) and Summit software (Dako North America, Inc., Carpenteria, USA) were used for data analysis.

2.6. Cell quantification

Adherent cell counts in culture were taken from 40× objective phase contrast microscope images. For secondary cells, at least 3 frames per replicate, per condition, and per day were counted and the mean of the replicates was used for analysis. For primary cells, 10 frames per replicate, per condition, and per day were counted and the replicates averaged for analysis.

2.7. Imaging

Cells were imaged on Day 1, 3, 7, 13 and/or 14, and 19 and/or 21 using a Nikon Eclipse TE2000-U microscope (Nikon Inc., Melville, USA) equipped with fluorescent optics, CCD camera, and Metamorph (Molecular Devices, Sunnyvale, USA) and Q Capture Pro software (QImaging, Surrey, Canada). Each image presented was selected as a representative image of at least 3 independent replicates.

2.8. Cell Staining

A hematoxylin and eosin stain (Fisher Scientific, Kalamazoo, USA) was employed according to manufacturer’s instructions to stain for nuclei and cytoplasm respectively. For fluorescence staining, cells were fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, USA) for 10 minutes at room temperature and stained with rhodamine-phalloidin and counterstained with 4,6-diamidino-2-phenylindole (DAPI) (Molecular Probes, Eugene, OR) according to standard protocols. All wells were preserved with Fluoromount-G (Southern Biotech, Birmingham, USA). For live green fluorescent staining, fibroblasts were stained with Vybrant CFDA SE Cell Tracer Kit (Invitrogen) according to manufacturer’s instructions prior to seeding with macrophages.

2.9. Multinucleated cell characterization

Primary and secondary macrophages were stained using a tartrate-resistant acid phosphatase (TRAP) assay (Sigma-Aldrich) according to manufacturer’s instructions after being cultured on tissue culture polystyrene (TCPS) for 21 days in each culture condition.

2.10. Endotoxin (LPS) analysis

Levels of LPS contamination in cell culture materials, reagents, and laminar flow hood were tested using Limulus Amebocyte Lysate (LAL) Assay (Lonza, Basel, Switzerland). All solid samples were soaked for three days in pyrogen-free water, and the water was used in the LPS colorimetric assay.

2.11. Statistics

Endogenous levels of cytokines in conditioned media were subtracted from cell-produced cytokine profiles. All experimental results are presented as the mean ± SEM. Technical replicates using the CBA were taken as the geometric mean fluorescence of at least 300 beads. Replicates for secondary macrophages and fibroblasts are represented as 3 separate wells carried in parallel through the entire experimental protocol; this experiment was repeated on 3 independent occasions and revealed similar results (data not shown). Replicates for primary macrophages were harvested from 4 mice. A Single-Factor ANOVA was utilized to determine significance between groups of samples for the conditioned media and LPS test groups. A Two-Factor ANOVA without replication was utilized to determine significant differences between the co-cultures (paracrine and juxtacrine) and either the mono-cultured macrophages or fibroblasts, or sum of both mono-cultured macrophages and fibroblasts. A post-hoc student’s t-test was used to determine statistically significant differences between samples (p<0.05). Samples treated with conditioned media or LPS were compared against control mono-cultured wells, while paracrine and juxacrine co-cultures were compared against the sum total of both control mono-cultured macrophages and fibroblasts at each discrete time point. This accommodates the problem of cytokines produced from both macrophages and fibroblasts in the co-cultures that should be considered. Particular comparisons to be tested were selected in advance and were reported individually rather than as a group and therefore were not appropriate for correction by a multiple comparison procedure [26]. In graphical representations, values below the assay detection limit were set to 0. The detectable limit was determined using the assay signal value of the 0 cytokine standard plus 1.5 times the standard deviation for that cytokine assay result.

3. RESULTS

3.1. Cytokine production

From an initial screen of cytokines, IL-2, IL-3, IL-4, IL-5, IL-9, IL-10, IL-12P70, IL-13, IFN-γ, and GM-CSF were found to be below 40 pg/ml (with the exception of GM-CSF, which increased above this value only in the presence of LPS) and were not included in this study (data not shown). IL-6, TNF, MCP-1, RANTES, MIP-1α, and MIP-1β were detected above this value and were considered the most relevant to the foreign body response and were thus included in this study.

3.1.a. Conditioned Media Treated Sample

Secondary and primary macrophages primarily decreased their production of inflammatory cytokines in the presence of fibroblast-conditioned media, compared to control macrophages, with the exception that they both increased their production of MCP-1 within the first 3 days (Figures 2 and 3). In contrast, primary macrophages increased their production of MIP-1β, while secondary macrophages did not. Interestingly, fibroblasts increased their production of all inflammatory cytokines tested in the presence of macrophage-conditioned media compared to control fibroblasts, even after subtracting the background values for cytokines in the macrophage-conditioned media (Figure 2).

Figure 2.

Cytokine production profiles for secondary RAW macrophages and fibroblasts. Profiles were normalized to the sum total of all cells in the well including macrophages and fibroblasts during co-cultures. Cytokine production represents the average from 3 separate wells; each experiment was repeated four times with similar results and trends (data not shown). IL-6, TNF, MCP-1, RANTES, MIP-1β, and MIP-1α were detected using a BD^™ Cytometric Bead Array. Significant difference from each internal control condition (p<0.05) is indicated by an *. Controls are represented by solid bars, combined RAW & 3T3 control represents the mathematical sum of mono-cultured 3T3 and RAW control cells for comparison with co-cultures. Values below the assay detection limit were set to zero. All cells grown in wells were cultured on tissue culture polystyrene. Fibroblasts delaminated on Day 19. Background endogenous levels of cytokines in conditioned media were subtracted from each data set.

Figure 3.

Cytokine production profiles for cultures containing primary macrophages and secondary fibroblasts. Profiles were normalized to the sum total of all cells in the well including macrophages and fibroblasts during co-cultures. Cytokine production represents the average from 4 mice. IL-6, TNF, MCP-1, RANTES, MIP-1β, and MIP-1α were detected using a BD Cytometric Bead Array. Significant difference from each internal control (represented by solid bars) (p <0.05) is indicated by an *. Mono-cultured BMMΦ macrophages were used as the control for BMMΦs in conditioned media. The mathematical sum of mono-cultured BMMΦ and 3T3s were used as the control for paracrine and juxtacrine co-culture (3T3 mono-cultured cell control data were taken from Figure 2). All cells grown in wells were cultured on tissue culture polystyrene. Values below the assay detection limit were set to zero (see methods and materials). Fibroblast delamination during juxtacrine co-culture occurred on Day 13. Background endogenous levels of cytokines in conditioned media were subtracted from the above data.

3.1.b. Co-cultures

Macrophage responses (both primary and secondary) during paracrine and juxtacrine co-culture were more similar to each other than to their cultures in conditioned media (Figures 2 and 3). In contrast to cultures with conditioned media, secondary macrophages during co-culture with fibroblasts dramatically increased their production of IL-6, MCP-1, and RANTES compared to controls (i.e., monocultures). More similar to samples treated with fibroblast-conditioned media, primary macrophages during co-culture increased only their production of MCP-1 significantly compared to controls.

3.1.c. LPS stimulation

LPS stimulation produced the greatest amount inflammatory cytokines in both primary and secondary macrophages compared to all other culture conditions. Primary macrophages produced higher maximum concentrations of inflammatory cytokines than secondary cells upon LPS stimulation. Additionally, primary macrophages decreased inflammatory cytokine production with repeated LPS treatments, while secondary macrophages primarily increased cytokines in the presence of repeated LPS stimulation (Figure 4). Interestingly, 3T3 fibroblasts also appeared highly responsive to LPS addition, significantly increasing nearly every cytokine tested. A qualitative summary of these relative changes compared to controls is shown in Table 1.

Figure 4.

Cytokine production normalized to cell number comparing mono-cultures both with and without LPS stimulation. Mono-cultured 3T3 fibroblasts, RAW macrophages, and BMMΦ macrophages were used as the control for their respective cell type repeatedly stimulated with LPS daily. Significant difference from each internal control (represented by solid bars) (p <0.05) is indicated by an *. All cells grown in wells were cultured on tissue culture polystyrene. Values below the assay limit of detection for each cytokine were set to zero.

Table 1.

A) Summary of relative changes in inflammatory cytokine production for secondary macrophages and fibroblasts, and B) primary macrophages in the presence of conditioned media, during co-culture, and after repeated stimulation with LPS. Arrows represent relative trends over 21 days in culture, and multiple arrows indicate changes in trends over the time course of the experiment. Background endogenous levels of cytokines in conditioned media were subtracted from the data.

3.2. Adherent cell morphology

Secondary macrophages during paracrine and juxtacrine co-culture and those stimulated with conditioned media all retained similar morphologies to mono-cultured control cells, displaying a more rounded phenotype and growing near, or on top of, neighboring cells at both early (Supplementary Figure 1) and late time points (Figure 5) (also previously observed [4, 27, 28]). Primary macrophages in conditioned media and in paracrine and juxtacrine co-culture all exhibited similar morphologies at early (Supplementary Figure 1) and late culture time points (Figure 5), exhibiting more cell attachments than secondary macrophages and remaining relatively contact-inhibited. However, control BMMΦs (mono-cultured in the absence of fibroblast stimulation), by contrast, lost their contact inhibition over time, forming large dense clusters of cells in multilayers. Over time, both primary and secondary macrophages developed a few scattered larger single-nucleated cells, with RAW macrophages occasionally producing multinucleate giant-like cells (Figure 5). These multinucleated cells were far larger than co-existing mononuclear cells, with centrally clustered nuclei, and consistently stained negative for TRAP (Supplementary Figure 2). The frequency of multinucleate cells in RAW cultures remained similar for every treatment condition with the exception of LPS-stimulated cells that resulted in far more fused cells. In the presence of LPS stimulation, both primary and secondary macrophages developed highly activated morphologies compared to controls, displaying a high frequency of larger cell bodies with cytoplasmic vesicles (e.g., foamy appearance) and occasionally multiple nuclei. In all primary macrophage cultures, LPS-stimulation was the only condition that produced multinucleate cells (Figure 5).

Figure 5.

Fluorescence and phase contrast microscope images of A) secondary RAW macrophages, and B) primary BMMΦ macrophages, in various 3T3 fibroblast signaling conditions after 21 days of culture. For fluorescence staining, phalloidin (red) was used to stain cytoskeleton and DAPI (blue) was used to stain cell nuclei.

Fibroblasts maintained similar adherent morphologies in every condition tested. During secondary cell co-culture experiments, fibroblasts cultured alone, in RAW-conditioned media, and LPS media all began to delaminate from the culture surface on Day 19 (Figure 2 and 4), a phenomenon also observed in previous extended fibroblast cultures [27, 29]. During primary juxtacrine co-culture experiments, fibroblasts delaminated on Day 13 (Figure 3). Due to this delamination, few fibroblasts remained, but began to repopulate past those time points.

3.3. Cell proliferation kinetics

Secondary RAW cells in 3T3-conditioned media and in paracrine and juxtacrine co-cultures possessed similar proliferation profiles as secondary RAW cells. Primary macrophages appeared to increase in cell number slightly in the presence of 3T3-conditioned media and during co-culture. Secondary and primary macrophages displayed distinct proliferation kinetics in culture (Figure 6). Secondary macrophages proliferated rapidly within the first week, peaking at Day 3, and then decreased until Day 21, while primary macrophages proliferated at much slower rates during the first week, peaking at Day 7, and then decreased very slowly during the remaining time periods. Primary and secondary cells showed no apparent correlations between cell proliferation and cytokine production (Figure 2, 3, 4, and 6). Juxtacrine co-culture cell density dropped on Day 13 due to fibroblast delamination. Macrophage cultures below the fibroblast Transwell inserts in paracrine co-culture wells often became contaminated by rapidly growing fibroblasts around Day 13, resulting in observed dramatic cell density increases to Day 21 (see Figure 6).

Figure 6.

A) Proliferation profiles for secondary macrophage and fibroblast mono-cultures. Profiles for 3T3s stimulated with LPS and treated with conditioned media were comparable to mono-cultured 3T3s, and RAWs in conditioned media were comparable to mono-cultured RAWs (not shown). B) Proliferation profiles for primary BMMΦs in various culture conditions. Juxtacrine co-culture cell density dropped on Day 13 due to fibroblast delamination. Macrophage cultures below the fibroblast Transwell^® insert in the paracrine co-culture wells became contaminated by rapidly growing fibroblasts around Day 13, resulting in dramatic cell density increases until Day 21. Cells in all conditions reached confluence by Day 1. Error bars not seen are smaller than symbols.

3.4. LPS assay

An LAL assay, employed to detect endotoxin contamination, showed that all samples and materials had LPS levels below the assay detection limit (0.06 EU/ml) with the exception of the LPS-treated cultures (data not shown). This validates that differences observed in cell culture were not due to unwanted adventitious LPS contamination.

4. DISCUSSION

Cell co-culture systems incorporating macrophages and fibroblasts, two primary effector cell types in the FBR, might better reflect actual FBR inflammatory profiles than their respective mono-cultures. Therefore, primary and secondary cells were both co-cultured in several different communication conditions in this assessment. Macrophages were treated with fibroblast-conditioned media and vice versa in order to determine if cytokine presence alone, without direct feedback from the other cell type in real-time, produced distinctly different responses from direct contact co-culture or mono-cultures. Transwell^® inserts were used to physically separate the two cell types in co-culture, allowing cytokine transport between cell types to mimic paracrine signaling. Both cell types were co-cultured together in physical contact to mimic juxtacrine signaling. As a positive control of cell activation, primary- and secondary-derived macrophages and fibroblasts were stimulated with LPS, shown to activate macrophages [30, 31]. Each cell mono-culture served as a negative control.

Cytokines TNF, IL-6, MIP-1α and MIP-1β, and chemokines MCP-1 and RANTES are some of the most potent inflammatory signals responsible for orchestrating the cellular responses to foreign material, frequently up-regulated in the presence of a foreign material in vivo [5, 11, 32, 33]. They were monitored as a quantitative metric for cell behavioral changes arising from culture conditions and cell origin. Cell proliferation was assessed to normalize cytokine production to cell number to account for cell density changes in each condition. Cell morphology was used as a qualitative metric of phenotype. Murine cells were chosen due to the cost-effectiveness, abundant reagent base, and relevant comparisons to the many implant reaction studies performed in mice.

4.1. Cytokine Response

4.1.a. LPS-stimulated cell cultures

LPS dramatically affected 3T3 fibroblasts, statistically increasing production of all cytokines with the exception of MIP-1α. Fibroblasts are not considered primary contributors of inflammatory cytokines upon endotoxin exposure. However, others have also reported increased IL-6 and other inflammatory cytokines by fibroblasts in response to LPS [34–36]. Strong LPS response by all cell types shows that both primary and secondary macrophages and fibroblasts are not maximally activated simply by in vitro culture on a synthetic surface, another contrast to cell activation by polystyrene and nearly every other synthetic material in vivo [37, 38]. This discrepancy is most pronounced with IL-6, shown in this and other studies to be secreted at low in vitro and very high in vivo concentrations even in the absence of LPS stimulation [6, 11]. Lack of cell activation against biomaterials in vitro has been a common challenge to such biomaterials assays, requiring use of exogenous stimulants such as LPS and phorbol esters to mimic in vivo activation (vide supra). This finding suggests that even in co-culture, cells may not be capable of biomaterials-based activation levels shown in vivo.

Published in vivo studies show a general decrease in inflammatory cytokine production from macrophages after repeated LPS dosing, a phenomenon known as LPS or endotoxin tolerance [39, 40]. This effect is considered a natural cellular response to prevent uncontrolled inflammation [41–43], and is readily apparent in cultured primary-derived macrophages in this and other studies [44]. However, the opposite effect was observed in secondary-derived RAW macrophages, where this and other studies show increasing inflammatory cytokine production after repeated LPS stimulation [45]. Though RAW 264.7 macrophages have exhibited endotoxin tolerance [43, 46], it has only been over the span of a few hours and with one repeated dose, compared to the continual LPS stimulation seen in this study over the course of 21 days. This general increasing inflammatory cytokine production from secondary macrophages over a three-week period may result from their transformed oncogenic phenotype, impeding the normal down-regulation of inflammatory cytokines observed in vivo and in primary cultures.

Additionally, primary macrophages exposed to LPS in this study produced comparable relative increases of inflammatory cytokines as in vivo serum levels produced in response to LPS stimulation [11, 47] (compared to non-stimulated controls). In contrast, secondary macrophages stimulated with LPS exhibited less dramatic increases in cytokine production, most pronounced with IL-6 and MIP-1α (Figure 4). These data suggest that endotoxin sensitivity in primary cells may better reflect in vivo LPS response that in than secondary cells.

4.1.b. Conditioned media-treated cell cultures

Primary macrophage monocultures showed more dramatic decreases in IL-6, TNF, and RANTES in the presence of 3T3-conditioned media compared to controls than their paracrine and juxtacrine co-cultures (Table 1). Secondary macrophages in fibroblast-conditioned media displayed less drastic changes in cytokine production compared to primary macrophages, more similar to control cells. As RAW cells are oncogenically transformed, passaged many times, displaying a more monocytic phenotype [4, 14, 15], they may exhibit less sensitivity to cellular cues in fibroblast-conditioned media than primary cells. This decreased primary and secondary cytokine production falls below that in their co-cultures (Table 1). Distinct differences in primary and secondary cell response to conditioned media versus co-culture could result from lack of dynamic cell feedback and reaction: co-culture allows for real-time cell processing and feedback between macrophages and fibroblasts, prompting unique signaling profiles.

Another contrast is that 3T3 fibroblasts increased production of pro-inflammatory cytokines in RAW-conditioned media, while RAW and BMMΦ cells generally decreased their pro-inflammatory cytokine production in the presence of 3T3-conditioned media. Cytokines present in fibroblast-conditioned media may inhibit macrophage production of cytokines tested while macrophage-conditioned media may induce cytokine production in fibroblasts. This complex balance of cell-cell signaling determines the host’s response to an implanted biomaterial, and importantly, that factor is missing in frequently used mono-cultures of isolated cell types.

4.1.c. Contact and non-contact co-cultures

Co-cultured primary and secondary macrophages both generally decrease production of MIP-1β, MIP-1α, and TNF compared to mono-cultured controls. Previous studies also document reductions of these cytokines in the presence of fibroblast-released cytokines [48, 49]. Primary macrophages in co-culture also generally decreased RANTES and IL-6 production compared to mono-cultured controls, related possibly to mutual cytokine down-regulation (Table 1). RANTES and IL-6 down-regulation were not observed in secondary cell cultures: by contrast, IL-6 and RANTES were highly up-regulated in both paracrine and juxtacrine co-cultures compared to controls. This in vitro discrepancy between primary and secondary macrophages cytokine expression may again be due to intrinsic differences in cell sensitivity and responses to cytokine signaling.

Notably, both primary and secondary paracrine and juxtacrine macrophage-fibroblast co-cultures displayed significant increases in MCP-1 at nearly all time points (Figures 2 and 3). MCP-1 could participate in a positive feedback loop between cultured macrophages and fibroblasts (seen in a schematic of known cytokine communication between macrophages and fibroblasts, Figure 7), resulting in its increased production in the presence of the other cell type (Table 1). Importantly, this dramatic increase in MCP-1 was not seen in primary or secondary macrophage monocultures treated with fibroblast-conditioned media a – system that inherently lacks cell-cell reciprocal feedback signals found in vivo. This MCP-1 increase is consistent with a positive feedback mechanism for MCP-1 previously proposed to perpetuate the FBR [50]. Increasing levels of TNF are shown to increase MCP-1 production [51, 52] and act as a mitogen for fibroblasts [53] (Figure 7) which may also then contribute to high MCP-1 levels seen during co-culture (Table 1). Increased MCP-1 production during 3T3 co-culture with both primary and secondary macrophages relative to mono-cultured macrophages could reflect an important fibroblast transition to a fibrotic phenotype, as fibrotic fibroblasts are known to increase TNF-induced MCP-1 [54], proliferation [55], and protein secretion [56]. Though co-cultures do not replicate all comprehensive aspects of in vivo reactions, they are capable of cell signaling patterns more representative of the in vivo environment than their mono-cultures with or without conditioned media.

Figure 7.

Overview of the select cytokine signaling pathways known between fibroblasts and macrophages [12, 33, 50-56, 64–73].

Interestingly, all cytokines significantly increased in co-cultures (compared cytokines produced from the arithmetic sum total from both mono-cultured macrophages and fibroblasts) were only those appearing to be produced in greater amounts by fibroblasts over macrophages during mono-culture. Mono-cultured fibroblasts exhibited greater IL-6, MCP-1, and RANTES cytokines than mono-cultured secondary macrophages, and produced greater MCP-1 cytokine amounts than primary cells. Relatively strong fibroblast cytokine response compared to macrophages may initiate positive feedback during co-culture. As cytokines are pleiotropic [57], their dynamic production kinetics and fluctuations are expected, prompting dynamic production and fluctuations of other cytokines.

4.2. Adherent cell morphology

At early time points, adherent secondary RAW 264.7 macrophages behaved very differently from primary-derived BMMΦs, showing no contact inhibition and readily growing in multi-layers. BMMΦs maintained larger cell bodies, more cell attachments, frequent lamellipodia, and elongated morphologies, while RAWs had smaller cell bodies and maintained a more-rounded phenotype compared to primary cells during mono-culture, a characteristic of their less differentiated and more monocytic phenotype (images of all cells on Day 1 are found in Supplementary Figure 1) [4]. However, over time both cell populations transitioned to larger cell bodies, showing differentiation over time (Figure 5). RAW control cells infrequently developed into FBGCs with large cell bodies and centrally clustered nuclei, while primary control macrophages did not. These cells consistently stained negative for TRAP (Supplementary Figure 2). This supports a FBGC-like phenotype [58, 59] over an alternative osteoclastic phenotype also known to be derived from RAW 264.7 cells [60]. Osteoclasts tend to have multiple nuclei lining the cell periphery as opposed to clustered centrally (arrows, Supplementary Figure 2), have larger cell bodies than FBGCs, and stain positive for TRAP [61].

Both in the presence of conditioned media and co-culture, RAW cells maintained a similar phenotype to control cells, unlike BMMΦs in these conditions that remained contact-inhibited in the presence of fibroblast signaling. Both primary and secondary control macrophage cultures, absent of fibroblast signaling, lost their contact inhibition, began to grow much larger cell bodies, and formed dense multilayered clusters (notably not multinucleated as for LPS-stimulated cells). Perhaps without appropriate signals, these cells may not be able to retain their original macrophage phenotype in culture.

LPS activates all macrophage cultures, influencing morphology of both primary and secondary macrophages more than any other condition, leading to the production of intracellular vesicles, enlarged cell bodies, and occasionally multiple nuclei, seen previously [30]. Kyriakides et. al identified a loss of peripheral actin rings and extended lamellipodia in cells destined to become FBGCs [62]. This is witnessed in LPS-stimulated BMMΦs (see Figure 5 -- the only treatment of BMMΦs leading to FBG-like cells). Neither primary nor secondary macrophages appeared activated by culture surfaces compared to LPS-stimulated cells, consisting with cytokine data. This absence of cell activation during in vitro co-culture of macrophages and fibroblasts may be due to the absence of acute wound-derived inflammatory cues always produced in vivo during device implantation, or insufficient cell types, numbers and signals necessary to accurately recapitulate host inflammatory reactions or produce a clear foreign body response.

During contact co-culture, one cell type grew more aggressively than the other, eventually creating a layer over the other cell type. RAW cells over-layered 3T3 fibroblasts, while 3T3 fibroblasts over-layered BMMΦs, both within the first 3 days. However, even with this covering layer, the cells underneath did not die, but also did not appear to proliferate. Supplementary Figure 3 shows viable fluorescent 3T3s beneath unlabeled RAWs, still maintaining a normal morphology. But these cells did not reach confluence like mono-cultured fibroblasts. Similarly, in the primary contact co-culture on Day 13, the 3T3s delaminated, revealing a layer of viable BMMΦs underneath, which appeared to have retained their original plating density and phenotype (Supplementary Figure 4). Cell culture conditions that enabled fibroblasts to grow to confluence all resulted in eventual longer-term cell-substrate delamination seen previously [29, 63].

4.3. Cell culture proliferation kinetics

Secondary RAW macrophages proliferated rapidly in the first week of culture, and then decreased cell numbers at later time points (seen previously [28]). Primary macrophages grew at far slower rates, maintaining a relatively constant cell density. These primary macrophages did not reach confluency until Day 7, after which their density decreased very slowly. By Day 3, secondary cells were at their highest proliferative rate (Figure 6), however, they exhibited low cytokine secretion profiles during that time. Proliferation of primary cells peaked at Day 7, but their highest cytokine secretion levels were during the first 3 days. These results suggest that cell inflammatory signaling activity is independent of proliferation. Also supporting this idea, LPS-stimulated secondary and primary cells secreted the largest amounts of inflammatory cytokines, but while undergoing the least proliferation.

RAW cell treatment with fibroblast-conditioned media or direct co-culture did not appear to affect their proliferation. These cultures maintained comparable proliferation rates to control RAW cells (as RAWs proliferated at much higher rates than 3T3 fibroblasts and therefore dominated co-culture kinetics). However, primary BMMΦs in conditioned media and their paracrine co-culture with fibroblasts show slightly increased proliferation rates compared to control BMMΦs. This increased cell number compared to mono-cultured BMMΦ controls is due to cell size in each culture: control BMMΦs without fibroblast stimulation grew very large cell bodies, while those with fibroblast signaling maintained much smaller cell bodies (Figure 5), altering the cell densities. LPS-stimulated primary- and secondary- macrophage cultures displayed an overall decreased cell number compared to controls. This reduced cell density may be due to this same cell size transition to larger cells, toxic effects of LPS, or a phenotypic shift to a less proliferative cell during LPS stimulation. Cell proliferation kinetics in primary juxtacrine co-culture were comparable to mono-cultured fibroblasts, because secondary fibroblasts proliferate faster than primary macrophages. With both secondary and primary macrophage juxtacrine co-cultures, growth kinetics are determined by this more highly proliferative cell type.

5. CONCLUSIONS

Several differences in cell signaling, adherent cell morphology, and proliferation between primary- and secondary-derived macrophages are shown in cultures, with greater apparent fidelity in primary over secondary cells to in vivo responses. Primary and secondary macrophage cell sources also exhibit unique responses in culture that provide different cell-cell feedback mechanisms, with reciprocal feedback (i.e., in paracrine and juxtacrine co-cultures) eliciting more representative characteristics found in vivo than monocultured cells in conditioned media. In general, co-culture feedback signaling in primary macrophage co-cultures with fibroblasts improves on in vitro models currently employing secondary cell mono-cultures. Despite co-culture feedback, without cell-derived wounding cues or implant-associated healing cascades in cell culture, FBR in vitro assays may exclude other acute cellular reactions associated with implant placement in vivo. Additionally, 2-D monolayer cell culture on rigid substrates and absence of other inflammatory cells such as T-cells may also limit in vitro approaches to duplicate the cell signaling observed in vivo. In vitro cell assays may only be adequate to represent specific cell types, functions, and simple biomaterials-associated signaling, but may be insufficient to approach or accurately represent more dynamic cell-interactive acute inflammatory or resulting FBR mechanisms found in vivo. The many in vitro assays (especially those using secondary-derived cell cultures and/or mono-cultured cells) used for FBR models, cell toxicity screening, drug and materials testing, and basic cell signaling research should all be validated by comparisons to primary cell cultures, cell co-cultures, and in vivo contexts in order to assert fidelity and relevance to the specific in vivo phenomena they claim to represent.

Supplementary Material

01. Supplementary Figure 1.

Day 1 morphologies for primary and secondary macrophages and fibroblasts in various adherent culture conditions (40× objective).

02. Supplementary Figure 2.

Cells during co-culture for extended time stained negative for TRAP. A) Positive control: optical microscopy image of an osteoclast-like multinucleated cell differentiated with RANKL stains positive for TRAP and peripherally distributed nuclei indicated by arrows (courtesy Y. Wang, University of Utah). B) Optical micropscopy image of a negative TRAP stain for BMMΦ cells co-cultured with 3T3 fibroblasts for 21 days. C) Optical microscopy image of a negative TRAP stain for RAW 264.7 cells after 21 days of co-culture with 3T3 fibroblasts.

03. Supplementary Figure 3.

3T3 fibroblasts remain present, but non-proliferative beneath a confluent layer of RAW cells. A) phase contrast image showing a confluent layer of RAW cells co-cultured over 3T3 fibroblasts. B) the same image under fluorescence microscopy with fluorescently-labeled fibroblasts displayed beneath the macrophages, remaining at low density even after several days of culture.

04. Supplementary Figure 4.

BMMΦ cells shown underneath delaminating fibroblast cell sheet after 13 days of co-culture. Red stains both cell types for actin.