The bioactive molecule sphingosine-1-phosphate (S1P) binds with high affinity to five recognized receptors (S1P1–5) to affect various tissues, including cellular responses of cardiac fibroblasts (CFbs) and myocytes. CFbs are essential components of myocardium, and detailed study of their cell signaling and physiology is required for a number of emerging disciplines. Meaningful studies on CFbs, however, necessitate methods for selective, reproducible cell isolations. Macrophages reside within normal cardiac tissues and often are isolated with CFbs. A protocol was therefore developed that significantly reduces macrophage levels and utilizes more CFb-specific markers (discoidin domain receptor-2) instead of, or in addition to, more commonly used cytoskeletal markers. Our results demonstrate that primary isolated, purified CFbs express predominantly S1P1–3; however, the relative levels of these receptor subtypes are modulated with time and by culture conditions. In coculture experiments, macrophages altered CFb S1P receptor levels relative to controls. Further investigations using known macrophage-secreted factors showed that S1P and H2O2 had minimal effects on CFb S1P1–3 expression, whereas transforming growth factor-β1, TNF-α, and PDGF-BB significantly altered all S1P receptor subtypes. Lowering FBS concentrations from 10% to 0.1% increased S1P2, whereas supplementation with either PDGF-BB or Rho-associated protein kinase inhibitor Y-27632 significantly elevated S1P3 levels. S1P2 and S1P3 receptor levels are known to regulate cell migration. Using cells isolated from either normal or S1P3-null mice, we demonstrate that S1P3 is important and necessary for CFb migration. These results highlight the importance of demonstrating CFb culture purity in functional studies of S1P and also identify conditions that modulate S1P receptor expression in CFbs.
sphingosine-1-phosphate (S1P) is a biologically active, cell membrane-associated sphingolipid that is secreted by various cells upon activation. S1P binds with high affinity to five distinct G protein-coupled receptors, also referred to as endothelial differentiation gene (EDG) receptors. S1P receptors are ubiquitously expressed on mammalian cells and can affect such cellular responses as proliferation, differentiation, and migration (reviewed in Refs. 1, 12, 13, 39, 40, and 45). To characterize S1P regulation, it is therefore necessary to have pure populations of the target cells.
Consistently obtaining highly purified cell cultures, which do not include any significant contaminating cell types, remains a considerable challenge when working with primary isolates from organs and tissues. However, this initial step is a requirement for many current molecular and cellular physiology studies in the cardiovascular system. Examples include 1) receptor characterization/signaling studies, where S1P receptors may be common among various cell types; 2) cytokine and growth factor studies, where contaminating cell types might secrete stimulatory or inhibitory factors, thus affecting the targeted cell population; or 3) electrophysiological studies in which cell coupling through gap junctions could complicate the interpretation of underlying ion channel-based mechanisms.
Mammalian cardiac tissue consists of a number of different cell types. Although myocardium is mainly composed of myocytes (∼50%), in terms of cell mass, nonmyocytes comprise the majority (65–70%) of cells within the ventricle. These nonmyocytes include endothelial cells, pericytes, smooth muscle cells (SMCs), and cardiac fibroblasts (CFbs) (36). In the mammalian myocardium, nonmyocytes also provide essential functions. For example, CFbs, which constitute the significant majority nonmyocyte cell type, secrete and maintain the interstitial collagen (collagen types I and III) (10). The collagen matrix helps to anchor and align myocytes. CFbs also secrete a variety of cytokines and growth factors that can have autocrine and paracrine effects. In both ventricles, CFbs are found interspersed between myofibers and are attached to the ECM, to myocytes, and to other CFbs (15, 36). In addition, cells specialized for immunological roles [e.g., macrophages (MΦ) and dendritic cells] are present in the myocardium, as well as within the circulatory system (e.g., monocytes, neutrophils, and lymphocytes).
The literature is limited regarding the presence of MΦ in cardiac cultures. This is surprising, since MΦ are normal resident cells within the tissues and organs of the cardiovascular system. MΦ can also differentiate from circulating peripheral blood monocytes (23). In the heart, MΦ are found within the pericardial sac and are interspersed within the myocardium where they are closely associated with myocytes and CFbs (4, 36). On this basis, it is expected that MΦ should be regularly obtained during isolation procedures for CFbs.
MΦ are known to secrete paracrine factors that are capable of rapidly changing target cell behavior and phenotype. These include interleukins, cytokines, and growth factors (3, 20, 31, 32, 42, 43). For example, MΦ secrete TNF-α, which can induce apoptosis in a variety of cells (14), transforming growth factor-β1 (TGF-β1), which can stimulate fibroblast differentiation to a myofibroblast phenotype (27), and PDGF, which can stimulate fibroblast proliferation (38). MΦ may also generate reactive oxygen species (e.g., H2O2) (57). MΦ are also a source of S1P (33).
An essential goal of our study was to improve CFb culture purity since MΦ can affect CFb phenotype. Methods for reducing or removing MΦ from cardiac cultures, and thereby enriching the population of CFbs, have been developed using more specific markers [e.g., the recently described CFb-specific marker, discoidin domain receptor-2 (DDR-2)] (15, 35) than those commonly used for the identification of CFbs (e.g., the cytoskeletal proteins vimentin and smooth muscle α-actin).
The results presented here demonstrate characterization of CFb phenotype via expression levels of S1P receptors and how CFb phenotype may be affected by MΦ. This has relevance not only for better understanding of paracrine signaling in native tissue, where CFbs and MΦ are in close approximation, but also in conditions of wound healing (e.g., granulation tissue or infarcted myocardium), in which MΦ play a prominent role. Purified CFbs were evaluated both separately and, after coculture with MΦ, to investigate the potential for coisolated MΦ to affect the differentiation and phenotype of CFbs. We have also studied various culture conditions and the addition of selected MΦ-associated factors. This was done using purified CFbs to determine better what factors contribute to altered S1P receptor expression in CFbs. We investigated the contributions of S1P2 and S1P3 receptor levels for migration of CFbs using culture conditions to modulate S1P receptor expression levels and through the use of CFbs isolated from S1P3-null mice.
Hearts from heparinized adult C57Bl/6 or S1P3-null mice (22) were isolated under isoflurane anesthesia in accordance with protocols approved by the Institutional Animal Care and Use Committee. These methods adhered to Guidelines for the Care and Use of Laboratory Animals (1996). Isolated hearts were placed in Ca2+-free Tyrode buffer containing (in mM) 130 NaCl, 5.4 KCl, 0.3 Na2HPO4, 1 MgCl2, 10 HEPES, and 5.5 glucose (pH 7.2–7.4). After cannulation of the aorta with a blunted 21-gauge needle, the heart was retrograde perfused with Tyrode buffer to clear the coronary arteries. Each heart was then connected to a Langendorff apparatus and perfused at 3–5 ml/min for 5 min with oxygenated Tyrode buffer (37°C). After an additional 10–15 min of enzymatic digestion in Tyrode buffer containing collagenase type 2 (1 mg/ml, Worthington Biochemical, Lakewood, NJ) and 40 μM CaCl2, the ventricles were removed, gently triturated, and filtered through 100-μm nylon mesh into modified Krebs-Henseleit buffer containing (in mM) 100 K+-glutamate, 10 K+-aspartate, 25 KCl, 10 KH2PO4, 2 MgSO4, 0.5 EGTA, and 5 HEPES and 1% BSA, 20 mM glucose, 20 mM taurine, and 5 mM creatine (pH 7.2–7.4). Myocyte populations were removed by mild centrifugation (1 min, 50 g, repeated twice). The CFb-containing supernatant was pelleted (5 min, 750 g), resuspended, and pelleted again through a density gradient (4% BSA in Tyrode buffer) to remove debris. The CFb-containing pellet was resuspended in 37°C culture medium [Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS, l-glutamine, sodium pyruvate, and antibiotic-antimycotic], plated onto tissue culture plastic, and allowed to incubate for 1–2 h in a humidified 5% CO2-95% atmospheric air incubator at 37°C. After this attachment period, nonadherent cells were removed by aspiration. Remaining cells were gently washed twice with Ca2+- and Mg2+-free PBS, and the cells were incubated with fresh culture medium.
Peritoneal MΦ were isolated from anesthetized adult mice by intraperitoneal lavage with MΦ buffer (Ca2+- and Mg2+-free PBS containing 2 mM EDTA and 0.5% BSA) and gentle abdominal massage for 1 min.
The CFb-containing pellet was resuspended in ice-cold MΦ buffer (5 ml) and incubated on tissue culture plastic for 30 min at 4–8°C. The nonadherent cells (containing CFbs) were collected and layered onto the 4% BSA density gradient and then processed as described in Cell isolation. The adherent MΦ were washed and retained for further analysis of defined cell type from the same heart.
An alternative procedure for removal of MΦ was accomplished via cell sorting using magnetic particle-conjugated antibody technology (Miltenyi Biotec, Auburn, CA). Cells were pelleted and resuspended in cold MΦ buffer and reacted with anti-CD11b antibodies as per manufacturer's instructions. CD11b (integrin αM or complement C3b receptor) reacts with myeloid-derived cells, including monocytes and MΦ. Both the positive (MΦ-containing) and negative (CFb-containing) cell fractions were collected for analysis.
Selected cell populations were seeded onto glass chamber slides and fixed in phosphate-buffered paraformaldehyde (PFA) (81 mM Na2HPO4, 19 mM NaH2PO4, and 4% PFA) for 10–20 min at room temperature and then stored at 4°C in 1% PFA. Cells were permeabilized in 0.3% Triton X-100 detergent and blocked in BSA and nonimmune sera. Primary or directly conjugated antibodies were as follows: DDR-2 (Santa Cruz Biotechnology, Santa Cruz, CA), F4/80-FITC (Caltag, Carlsbad, CA), F4/80-PE (eBioscience, San Diego, CA), vimentin (Sigma, St. Louis, MO), and smooth muscle α-actin (Sigma). Cells were washed with Ca2+- and Mg2+-free PBS containing 0.05% Tween-20 and 1% BSA (Rockland, Gilbertsville, PA). Visualization was achieved using fluorochrome-conjugated secondary antibodies as follows: rabbit anti-goat IgG FITC (Sigma) or sheep anti-mouse IgG FITC (Sigma). Isotype IgG control samples were incubated at comparable immunoglobulin concentrations. Finally, nuclei were visualized by counterstaining with 4′,6′-diamidino-2-phenylindole dihydrochloride hydrate (DAPI; Sigma). To control for nonspecific antibody binding by MΦ fragment, crystallizable (Fc) receptors, MΦ were preincubated with an anti-FcγIII/II receptor (CD16/CD32; BD Biosciences, San Jose, CA)-blocking antibody before the addition of primary antibody.
Digital images were captured on an ORCA-285 charge-coupled device camera (Hamamatsu Photonic Systems, Bridgewater, NJ) using ImagePro Plus 5.0 (MediaCybernetics, Silver Spring, MD), an Olympus IX-81 inverted epifluorescent microscope, and appropriate filter sets. Identical exposure times were used for isotype controls and samples, and all images were subjected to equivalent postimage processing, if any, to control for brightness and contrast.
Cells in suspension were fixed and processed as discussed in Immunostaining or using IntraCyte FACS kit reagents (Orion Biosolutions, Carlsbad, CA). Stained cells and their isotype controls were resuspended in FACScan buffer (BD Biosciences) and analyzed using a BD FACScan flow cytometer. Cell populations of interest were gated to count a minimum of 10,000 events based on forward and side scatter of control samples. Histogram data sets were analyzed by CellQuest software (BD Biosciences) by selective gating compared to isotype controls.
Electrophoresis and immunoblotting.
Whole cell lysates from NIH3T3 fibroblasts and human umbilical vein endothelial cells were collected for SDS-PAGE and immunoblotting as previously described (56). In brief, confluent cultures (10-cm dishes) were washed in ice-cold PBS, lysed in 750 μl ice-cold lysis buffer, and clarified by centrifugation. Equivalent protein amounts (determined by Bradford assay) were separated on 12% SDS gels and transferred to nitrocellulose. Immunoblotting was performed with anti-EDG-1 antibodies (Santa Cruz Biotechnology) after blocking with 5% BSA, and the protein bands were visualized by secondary-conjugated horseradish peroxidase antibodies and enhanced chemiluminescence (Amersham, Piscataway, NJ).
Cells were lysed, total RNA was isolated by silica-gel columns using RNeasy kits (Qiagen, Valencia, CA) and eluted in water, and cDNA was reverse transcribed from mRNA by oligo-dT priming and Omniscript reverse transcriptase (Qiagen). Expression levels of gene targets were characterized by SYBR green (2) quantitative real-time PCR (qPCR) of cDNA using specific primer sets (see Table 1) for GAPDH, DDR-2, CD64, CD68, and MΦ scavenger receptor-1 (MSR-1). Expression levels of S1P receptors were performed using Applied Biosystems (Foster City, CA) TaqMan primer/probe sets as follows: S1P1 (Mm00514644_m1), S1P2 (Mm01177794_m1), S1P3 (Mm00515669_m1), and GAPDH (Mm99999915_g1). Primer sets for genotyping S1P3-null mice were as previously reported (22).
Amplicon lengths for each primer set were confirmed by gel electrophoretic separation (2% agarose) with ethidium bromide visualization, and all SYBR green experimental samples were subjected to melt curve analyses following amplification. All experimental data were reported relative to GAPDH levels using the 2−ΔΔCt calculation methods (where ΔΔCt is delta delta method threshold cycle), which was considered appropriate based on primer set efficiency determinations (29).
S1P expression experiments.
Adult mouse CFbs were purified on the basis of CD11b sorting the day following isolation and plated at subconfluent (∼30–50%) densities to multiwell plates in 10% FBS DMEM culture medium. Because the intent of these investigations was to evaluate changes in acutely isolated CFbs, cells were used in experiments without subsequent passaging. Cultures were ∼80–100% confluent when studied.
In separate wells, cardiac MΦ (either CD11b purified or as mixed CFb cultures) were seeded onto microporous cell culture inserts (BD Falcon, Bedford, MA; Corning Costar, Acton, MA; or Nalge Nunc, Rochester, NY) and allowed to attach. After this attachment period, the inserts were transferred to wells containing CFbs, and fresh culture medium was supplied. The cells were cocultured for an additional 3 days (5% CO2-95% atmospheric air, 37°C), at which time they were washed in PBS and collected for PCR.
Alternatively, CD11b-purified CFbs were seeded to multiwell plates, allowed to adhere, then dosed with various substances for 48–72 h (5% CO2-95% atmospheric air, 37°C). These included TGF-β1 (Sigma), TNF-α (Sigma), S1P (Avanti Polar Lipids, Alabaster, AL), H2O2 (Rite Aid, Harrisburg, PA), PDGF-BB (Sigma), IGF-1 (Sigma), the Rho-associated protein kinase inhibitor Y-27632 (Sigma), or a Rac1 inhibitor (Calbiochem, La Jolla, CA).
Selected cell populations were seeded at relatively high density on 8-μm pore size-modified Boyden migration chambers (BD Falcon or Corning Costar) and allowed to attach. The cells were then washed with PBS and serum starved by overnight culture in 0.1% FBS DMEM culture medium. The medium was replaced with either 0.1% FBS DMEM control medium or treatment medium in the lower chamber (to establish a concentration gradient), and the cells were cultured for an additional 24 h. These preparations were then fixed and stained for MΦ (anti-F4/80) and counterstained with DAPI, and the nonmigrating cells on the upper membrane surface were removed by physical abrasion using a cotton swab. The surface of the lower membrane of the migration chamber was then analyzed using epifluorescent microscopy as described previously. In addition, captured images of cell nuclei were counted using ImagePro Plus software.
Results from replicate experiments were averaged and are presented as means (SD). Results were analyzed for ANOVA (with Bonferonni t-test post hoc testing) or Student's t-test. Significant differences were accepted for results with P < 0.05.
Expression of S1P receptor subtypes in freshly isolated CFbs.
Primary isolated CFbs from adult mice were found to express mRNA predominantly for the S1P1, S1P2, and S1P3 receptor subtypes (Fig. 1A). Expression levels of S1P4 and S1P5 were ∼50-fold and 150-fold lower, respectively, than either S1P1–3 (data not shown). Accordingly, S1P4 and S1P5 mRNA levels were omitted from analysis in the studies reported here. The relative S1P receptor expression levels for 1-day-old CFbs, when normalized to S1P1, showed that S1P3 was most abundant and that S1P2 was expressed the least (Fig. 1B). With extended time in culture (4 days), however, the levels of S1P2 and S1P3 (relative to S1P1) mRNA changed. S1P2 increased, whereas S1P3 was noted to decrease slightly with time.
Modulation of S1P receptor expression in CFbs by known MΦ-secreted factors.
MΦ can secrete various paracrine factors, including interleukins, cytokines, and growth factors, and often reside in close approximation with fibroblasts. To investigate whether such MΦ-secreted factors could cause alterations of S1P receptor mRNA expression in CFbs, the effects of a number of purified factors were studied one at a time. We first evaluated the effects of TGF-β1. Purified CFbs were seeded on multiwell dishes and incubated in standard culture medium (DMEM with 10% FBS) or in culture medium with TGF-β1 (5 ng/ml) for 48–72 h. TGF-β1 reduced the relative mRNA expression levels (vs. normalized control values) of all three S1P receptor subtypes (Fig. 2A). However, when TNF-α, another MΦ-secreted factor, was added at either 0.1 or 1 ng/ml (Fig. 2A), all three S1P receptor subtype expression levels increased relative to normalized control levels. Furthermore, there was a reproducible and significant rank ordering of the changes. After TNF-α treatment, S1P3 message showed significant increases (P < 0.05) compared with S1P1 and S1P2 levels, which exhibited slight increases. Finally, TGF-β1 and TNF-α were coadministered to CFbs. The presence of TGF-β1 abolished the observed upregulation by TNF-α. Under these conditions, S1P receptor levels approximated those of TGF-β1 alone (Fig. 2A).
Exposure to physiologically relevant concentrations of the bioactive molecule S1P (0.1 μM) failed to alter S1P receptor mRNA expression levels in CFbs (Fig. 2B). Increasing the S1P dosing concentration 10-fold resulted in only a slight increase in S1P1 expression. Similarly, exposure of CFbs to H2O2 (0.1–1 μM) failed to alter S1P receptor expression levels compared with control cells. In contrast, PDGF-BB (50 ng/ml) resulted in a significant increase in S1P1 (P < 0.05) and a moderate increase in S1P3 (Fig. 2C). As in Fig. 2A, all factors were supplemented in 10% FBS-containing medium. Our S1P results agree with published reports (30), which describe the absence of an effect of S1P on S1P1 expression in mouse embryonic fibroblasts; however, whereas mouse embryonic fibroblasts did not respond with an increase in S1P1 upon exposure to PDGF, our results using adult mouse cardiac fibroblasts demonstrated a significant increase in S1P1.
These CFb studies necessitated the use of qPCR to assess changes in S1P receptor isotype expression levels due to limited numbers of available cells and the lack of high-quality, commercially available antibodies. To address whether changes in S1D receptor message expression actually reflect changes in protein expression, NIH3T3 fibroblasts were exposed to 10% or 1% FBS for 2 days and analyzed by qPCR and Western blot analysis for expression of S1P1. As shown in Fig. 2D, 1% FBS induced an increase in both message and protein levels of S1P1 in NIH3T3 cells.
Modulation of S1P receptor expression in CFbs by small MΦ coculture.
In some experimental settings, having a percentage of monocytes and MΦ within a culture of CFbs may not have any significant bearing on the interpretation of experimental findings. However, this depends heavily on the objective of the planned studies. A representative study is our interest in physiological responses of CFbs to S1P. S1P receptors are ubiquitously expressed in many cell types. They include differing cellular responses depending on the specific receptor subtype(s) expressed. We therefore investigated whether the presence of MΦ in CFb cultures can significantly affect the measurement of the expression of S1P receptor subtypes on CFbs.
We examined the effect of coculturing MΦ with CFbs on S1P receptor expression (Fig. 3). Unpurified (CFbs/MΦ) or purified (CD11b+ MΦ) cultures of mouse cardiac MΦ were seeded into separate microporous inserts, allowed to attach, then placed into wells, which had been preplated with purified mouse CFbs and cocultured for 3 days. When CFbs were exposed to mixed CFb and MΦ populations, S1P3 mRNA levels decreased and S1P1 mRNA levels increased. These changes were shown to be significantly different from each other (P < 0.05). However, when MΦ were first purified (by CD11b sorting) and then cocultured with CFbs, S1P2 expression decreased. In contrast, S1P1 and S1P3 exhibited slight increases in expression levels.
MΦ have the capacity to alter S1P receptor mRNA expression levels in CFbs by coculture experiments (see Fig. 3). Furthermore, several factors that are produced by MΦ (see Fig. 2) were also shown to affect CFb S1P receptor expression when investigated using purified cell products. Since MΦ are a frequent contaminating cell type in CFb isolations, our results emphasize the necessity of having pure CFb populations. Accordingly, the development of purification procedures following initial CFb isolation was undertaken. These methods are presented in the following section.
Detection and quantification of MΦ following routine CFb isolation procedures.
In previous studies using collagenase-isolated CFbs from the ventricles of adult mice, heterogeneous cell phenotypes were observed during phase-contrast microscopic examination. Specifically, in addition to the adherent cells having characteristic fibroblast morphology, there were also rounded adherent cells and phase-contrast dense cells of varied shapes and sizes (Fig. 4A). Initial screening of these nonfibroblast cell types by PCR and further immunological characterization using the highly specific mouse MΦ surface receptor F4/80 demonstrated the presence of MΦ in cultures of mouse CFbs (Fig. 4B). In our study, flow cytometry analysis was employed to quantify the numbers of monocytes and MΦ present in freshly digested myocardial tissues. Adult mouse ventricles were digested as described, stained with antibodies against F4/80, and counted. The percentage of F4/80-positive cells that produced signals which exceeded isotype controls was 7.04% (SD 2.61). Thus MΦ are present in mouse CFb isolations, and any meaningful study of CFbs requires awareness of this potential problem.
Removal of MΦ by scavenger receptor adherence and antibody-mediated separation.
Since MΦ are obtained during “CFb isolations” from mouse myocardial tissue (5), we developed methods that would augment their removal. Trypsinization has been employed in other cell isolation methods as a means to help separate MΦ from fibroblasts (7). Consistent with these reports, we found that when CFb cultures were trypsinized within a few days after their initial isolation, MΦ numbers were reduced; that is, many trypsin-resistant MΦ remain adhered to the tissue culture plastic (Fig. 4, C and D). However, this procedure removes only a small subset of these contaminating cells (i.e., those adhered to the tissue culture plastic surface). Notably, it does not remove MΦ that have adhered to CFbs (see Fig. 4B). For this reason, MΦ could still be detected in unpurified CFb cultures after the first passage (Fig. 4, E and F).
Fibroblasts attach to ECM molecules via cell-surface integrin receptors, and this process requires divalent cations. In addition to this integrin-mediated attachment, MΦ can attach to culture surfaces in integrin-independent ways that depend on scavenger receptors (17). We have taken advantage of this to develop an additional purification step during the CFb isolation procedure. The MΦ buffer used during our preplating step contains no serum (and hence no serum-derived attachment factors such as fibronectin or vitronectin) but does include BSA to block nonspecific protein binding. Addition of EDTA (2 mM) chelates divalent cations. An incubation period of 30 min at 4–8°C was shown to be optimal since shorter time periods removed is fewer MΦ. Incubations for up to 2 h in this solution failed to result in attachment of NIH3T3 mouse embryonic fibroblasts (used as control cells). NIH3T3 cells remained in the nonadherent fraction (i.e., suspended in the solution, data not shown).
When freshly digested mouse myocardial isolates were subjected to a MΦ preplating step, the percentage of F4/80-positive MΦ was significantly reduced from 7.04% (SD 2.61) to 1.50% (SD 0.017) (P < 0.02) of total nonmyocytes, as determined by flow cytometry (Fig. 5A). Analysis by qPCR of the adherent and nonadherent fractions following the MΦ preplating step confirmed that MΦ were significantly enriched in the adherent fraction (i.e., they were able to attach via their scavenger receptors) and that CFbs were enriched in the nonadherent fraction (i.e., they were unable to attach and remained suspended in the solution). In mouse myocardial digests, there was a 7-fold increase in MΦ-specific MSR-1 expression levels in the adherent fraction and a 20-fold increase in CFb-specific DDR-2 expression levels in the nonadherent fraction (Fig. 5B).
This MΦ preplating step has proven to be straightforward and cost effective, and it is easily incorporated into the overall isolation procedure. However, it relies on cells that express active scavenger receptors and may therefore be less effective at eliminating monocytes or undifferentiated MΦ. As a result, some nonadherent monocytes and MΦ may be carried through with the CFb-containing fraction. They become activated, and eventually adhere to the tissue culture plastic along with the CFbs. To address this concern, we used magnetic-activated cell sorting methods to remove monocytes and MΦ from cells in suspension.
Magnetic particle-conjugated anti-CD11b antibodies were reacted with CFb cultures immediately following the MΦ preplating step. After centrifugation to remove excess free antibody, the labeled cells were loaded onto a column containing packed metallic beads under an induced magnetic field (MiniMACS Separator, Miltenyi Biotec). The CD11b+ cells were retained within the column, whereas the CD11b− cells flowed through and were collected from the reservoir tip. After several washes, the column was removed from the magnet and the bound cells (CD11b+) were eluted. Typical results of CD11b-sorting experiments are presented in Fig. 6.
We now routinely perform CD11b sorting 1–2 days postisolation to minimize the potential effect(s) of MΦ coculture on the CFbs (see Figs. 2 and 3). Performing the sorting procedure immediately following the initial enzymatic digestion process was shown to be less effective. Dead myocytes and debris interfere with the elution of CFbs from the column (data not shown). Furthermore, allowing the CFbs to adhere initially onto tissue culture plastic surfaces takes advantage of differential plating times, thus for minimizing endothelial cell contamination.
Adult mouse peritoneal MΦ were utilized to validate the CD11b-sorting method initially. As shown in Fig. 6A, strong mRNA expression of MΦ markers CD64, MSR-1, and CD68 was found in the CD11b+ fraction. In contrast, very distinct expression of the CFb marker DDR-2 was found in the CD11b− fraction. When adult mouse CFb cultures were sorted, a similar expression profile was obtained (Fig. 6B). These results demonstrate successful separation of monocytes/MΦ and CFbs in isolations from mouse hearts. To determine whether MΦ can remain in unpurified CFb cultures for extended periods, cultures at various times after the initial isolation were analyzed by CD11b sorting. As depicted in Fig. 6C, MSR-1 continued to be strongly expressed in CD11b+ fractions of adult mouse CFb cultures for up to 14 days (the longest time evaluated). The detection of MΦ expression patterns was assessed by qPCR in these studies, since many of the MΦ have been already removed by the preplating step, and flow cytometry was not considered sensitive enough for the analysis of this sorting procedure.
Immunological assessment of fibroblast markers.
Results of immunological assessment of CFbs are presented in Fig. 7. CFbs are often identified by positive immunoreactivity for vimentin and smooth muscle α-actin as markers of undifferentiated and differentiated (i.e., myofibroblasts) phenotypes, respectively (11, 54). Vimentin is an intermediate filament that is abundant in migrating cells. In the next set of experiments, CFbs were isolated from adult mice and stained with anti-vimentin antibodies. As expected, strong reactivity was observed in all CFbs examined as early as 2 days after isolation. Similarly, we also observed strong immunoreactivity against smooth muscle α-actin in CFbs isolated from adult mice.
However, both vimentin and smooth muscle α-actin have also been reported to be expressed in monocytes and MΦ (18, 34). When using mouse peritoneal MΦ as a positive control for the MΦ cell type, we consistently observed positive staining for both vimentin and smooth muscle α-actin. To further explore this finding, we examined vimentin and smooth muscle α-actin expression in CD11b+ MΦ isolated from mouse hearts. Dual labeling with antibodies against F4/80 and the cytoskeletal proteins vimentin or smooth muscle α-actin demonstrated colocalization in mouse MΦ that had been isolated from CFb preparations. Although the staining in the MΦ was less intense and more peripheral than that observed with CFbs, MΦ were, nonetheless, immunoreactive against these proteins that are routinely used as selective fibroblast markers.
Although CD11b is expressed primarily on monocytes and MΦ, it can also be found on other myeloid-derived cells such as granulocytes (i.e., neutrophils, basophils, eosinophils, and dendritic cells) (49). Thus the CD11b sorting method employed may be capable of removing these cell types as well. However, our immunostaining results using F4/80 suggest that the CD11b+ cells were almost exclusively monocytes and MΦ.
Recently, efforts to identify a more specific marker of CFbs have identified the fibrillar collagen receptor DDR-2 (15). DDR-2 has been reported to be expressed on CFbs but not on endothelial cells, SMCs, or myocytes within normal cardiac tissue. Based on our observations that MΦ can coisolate with CFbs and that the (15) study on DDR-2 did not specifically evaluate MΦ, we characterized both CFbs and MΦ for immunoreactivity to DDR2. We observed positive labeling of adult mice CFbs with anti-DDR2 antibodies. Importantly, no immunoreactivity of DDR-2 toward CD11b+ adult mouse peritoneal MΦ or CD11b+ MΦ from cardiac isolations was detected. Thus DDR-2 does appear to be a cardiac-specific immunomarker for the fibroblast population.
Modulation of S1P receptor expression in CFbs cultured in low-serum conditions.
Many experimental procedures include a serum withdrawal and deprivation step before initiating treatment. For example, this is routinely performed before the addition of a chemotactic agent in migration-based assays or before the addition of a mitogenic factor in proliferation-based assays. We therefore wanted to determine whether S1P receptor mRNA expression levels would be affected during such pretreatment steps. We have noted that when CFbs were exposed to culture medium containing only 0.1% FBS, S1P2 expression was increased over 2.5-fold relative to control cells cultured in 10% FBS-containing medium (P < 0.05 vs. S1P1 or S1P3) (Fig. 8A). In addition under these conditions, S1P3 levels were reduced to less than one-fourth of that of control cells, and S1P1 was diminished by ∼50% compared with control levels.
The relative levels of S1P2 and S1P3 have been associated with negative and positive influences, respectively, on cell migration (47). We investigated agents that could either reduce or minimize the increases observed in S1P2 levels or, alternatively, increase levels of S1P3 when exposed to reduced serum concentrations. S1P2 utilizes a signaling pathway through G12/13 that activates Rho. Therefore, we investigated whether this pathway could be interrupted using the Rho-associated protein kinase inhibitor Y-27632. This inhibitor was selected since it had been shown previously to inhibit S1P2-mediated Rho-dependent stress fiber formation in transfected Chinese hamster ovary cells (46). There was a slight increase in S1P2 mRNA in CFbs treated with the Rho kinase inhibitor (relative to control cells), a significant increase in S1P3 was noted in the treated cells (P < 0.05 vs. S1P1 or S1P2; Fig. 8B).
PDGF-BB was selected as a candidate for supplementation. It has been shown to cross-activate with signaling pathways (6). When serum-deprived (i.e., cultured in 0.1% FBS) CFbs were compared with serum-deprived CFbs supplemented with PDGF-BB (50 ng/ml), the mRNA expression level of S1P3 was over 12-fold higher (P < 0.05 vs. S1P1 or S1P2; Fig. 8C). This response was the most dramatic of any of the evaluated factors. To determine whether this increase was specific to PDGF-BB, we evaluated the response of CFbs to IGF-1, another known tyrosine kinase receptor agonist. IGF-1 did not upregulate S1P3 expression. In fact, this caused a slight reduction in expression levels (Fig. 8C).
Since interfering with the S1P2 signaling pathway using a Rho-associated protein kinase inhibitor resulted in a decrease in S1P2 mRNA expression, we investigated whether interfering with the S1P3 signaling pathway might induce similar downregulation. PDGF-BB was used as a positive modulator of S1P3 expression. When CFbs were coincubated with PDGF-BB and a Rac1 inhibitor, S1P3 mRNA expression levels were significantly reduced to those of controls (P < 0.05; Fig. 8D). Thus it appears that downstream interference of S1P receptor pathways can alter expression of the very same isotypes in CFbs.
Finally, we evaluated exposure of CFbs to S1P under conditions of serum deprivation. As before, S1P at low concentrations (0.1 μM) failed to influence S1P receptor subtype mRNA expression in CFbs. S1P at higher concentrations (1 μM) had a similar effect (i.e., slight increases) on S1P1 and S1P3 (Fig. 8E).
S1P3 has been shown to augment cell migration via Gi signaling (46, 46). Because we planned to conduct comparative migration studies using CFbs from normal and S1P3-null mice, we characterized S1P3-null CFbs to determine how they would respond to several of the S1P receptor modulations that had been demonstrated in normal CFbs. First, we confirmed the genotype of S1P3-null mice (Fig. 9A). We next exposed CFbs isolated from these mice to PDGF-BB (50 ng/ml) or to 0.1% FBS for 2–3 days. Similar to normal CFbs (see Figs. 2C and 8A), S1P3-null CFbs increased expression of S1P1 mRNA in response to PDGF-BB and increased S1P2 mRNA in response to reduced serum (Fig. 9B). When compared with control S1P3-null cells cultured in 0.1% FBS-containing medium, S1P3-null cells exhibited attenuated S1P2 mRNA expression levels after being cultured in 0.1% FBS supplemented with the Rho-associated protein kinase inhibitor Y-27632, PDGF-BB, or IGF-1 (Fig. 9C). These results agree with the responses obtained in normal CFbs under similar conditions (see Fig. 8, B and C).
MΦ can bias results in CFb migration assays.
Functional migration assays revealed another example of how MΦ contamination of CFb cultures can complicate data interpretation. Motile cells are often evaluated in terms of their ability to exhibit chemotaxis-directed migration toward a concentration gradient of a chemical stimulus. In our experiments, CFbs that had not been purified by CD11b sorting to remove monocytes and MΦ were seeded onto microporous inserts, serum starved overnight, and then subjected to 10% FBS culture medium for an additional 24 h. The inserts were counterstained with DAPI to quantify total number of migrating cells and immunostained for MΦ using anti-F4/80 antibodies (Fig. 10A). This experimental test revealed that many of the cells that migrated in the unpurified CFb cultures were, in fact, MΦ and not CFbs.
S1P chemotaxis in adult CFbs.
S1P can have stimulatory or inhibitory effects on cell migration depending on S1P concentration, cell type, and relative levels of S1P receptor subtypes. Since little is known with regard to the effects of S1P on CFb migration, we performed experiments on the migration response of CFbs to S1P. Purified adult mouse CFbs (4 days old) were seeded to migration chambers, serum deprived for 24 h, and exposed to S1P (0.05–1 μM). When compared with cells receiving only 0.1% FBS, S1P failed to elicit any fold change above these controls (Fig. 10B). However, additional chambers containing PDGF-BB (50 ng/ml) demonstrated a positive migratory response, thus ensuring that the evaluated CFb populations were not defective in this regard (Fig. 10B).
PDGF-BB chemotaxis and S1P3 receptor involvement in CFbs.
To ascertain the involvement of S1P3 in migration of CFbs, we isolated CFbs from S1P3-null mice and performed migration assays using PDGF-BB as the stimulus. In contrast with normal CFbs, which were capable of migrating toward PDGF-BB, migration assays using S1P3-null CFbs failed to show a similar response (Fig. 10C). To determine whether this was a generalized response in cells lacking S1P3, we isolated peritoneal MΦ from S1P3-null mice and performed migration assays. As shown in Fig. 10D, S1P3-null peritoneal MΦ migrated toward PDGF-BB. Thus the lack of responsiveness toward PDGF-BB observed using CFbs appears to be cell-type specific. In follow-up studies, peritoneal MΦ were shown to express almost exclusively message levels for S1P1 (data not shown). Furthermore, peritoneal MΦ do not increase S1P3 expression upon exposure to PDGF-BB (data not shown). Thus knockout of the S1P3 gene in peritoneal MΦ had little impact on their migration activity. This finding further illustrates S1P receptor isotype differences and their functional relationship to migration in differing cell types.
Summary of main findings.
The main objective of this study was to identify the S1P receptor subtypes (isoforms) expressed in distinct types of cells that comprise the working myocardium. Our initial emphasis was on studies of fibroblasts and myofibroblasts isolated from adult mouse ventricles. However, it became clear that MΦ were present and represented a significant source of contamination, both with regard to their ability to directly (i.e., through expression of similar S1P receptors) and indirectly (i.e., through secretion of modulating growth factors) interfere with the experimental analysis.
Our results show that the presence of MΦ in CFb cultures can affect both CFb phenotype and the interpretation of CFb functional responses. We demonstrate that the exposure of primary isolated CFbs to MΦ can alter S1P receptor message levels in CFbs. Furthermore, we demonstrate that exposure to purified factors, which are known to be released by MΦ, also affect S1P receptor subtype expression. For example, TNF-α can increase all S1P receptor subtypes, whereas TGF-β1 can decrease all S1P receptor subtypes evaluated. Furthermore, TGF-β1 can abolish the effects of TNF-α when coadministered. In contrast, other factors secreted by MΦ, such as S1P itself and H2O2, had no effect on CFb S1P receptor expression. In other experiments, we show that withdrawal of serum can increase S1P2 levels but that S1P2 levels are not increased when cells are treated with the Rho-associated protein kinase inhibitor Y-27632. Y-27632, instead, altered the relative expression levels in favor of S1P3, similar to increases after exposure to PDGF-BB. This was shown to be specific for PDGF-BB and not a general activation of tyrosine kinase receptors since IGF-1 was unable to mimic the increases in S1P3. Interfering with S1P3 signaling pathway via a Rac1 inhibitor reduced S1P3 receptor expression, suggesting that S1P receptor expression levels are regulated by downstream events. Finally, we demonstrate that CFbs do not migrate toward S1P and that S1P3 is necessary for migration of CFbs in response to PDGF-BB. MΦ have naturally low expression levels of S1P3, and migration in this cell type is not affected by the loss of this receptor isotype.
For these reasons, our success with these experiments depended on first developing new methods significantly to reduce MΦ levels. MΦ are a significant contaminating cell type in CFb cultures of the mouse heart. Effective removal of MΦ from CFb enzymatic digests can be accomplished by preplating in cold cation-chelated buffer followed by magnetic-activated cell sorting using anti-CD11b antibodies to more effectively deplete MΦ and enrich CFbs. Because MΦ express commonly used cytoskeletal markers (e.g., vimentin and smooth muscle α-actin), a more CFb-specific marker, DDR-2 (15), was used to characterize the effectiveness of our chosen isolation and culture procedures.
MΦ presence in CFb cultures and the significance for cell physiology and cell signaling in CFbs.
Isolation of CFbs from adult myocardium is commonly performed by retrograde perfusion through the aorta and enzymatic digestion of the ECM. Following physical separation of myocytes using centrifugation, the CFbs are allowed to attach to tissue culture plastic for 1–2 h (to minimize endothelial attachment). It is worth noting that CFb isolation protocols that this 1- to 2-h-preplating step in serum-containing culture medium would be expected to include monocyte and MΦ attachment. This is because monocytes have been shown to adhere within 1 h in PBS alone and MΦ readily plate out in 10% serum-containing medium within 1 h (24).
We have demonstrated that monocytes and MΦ are routinely identified in CFb isolations. The related literature, however, is extremely limited. Most published procedures for CFb isolation and culture are based on what is termed “a high degree of purity,” and one report that specifically mentions MΦ being present in enzymatic digestions fails to provide any further quantitative details (5). Identification of CFbs is largely based on phase-contrast examination and positive immunolabeling to selected cytoskeletal proteins. Our results (Fig. 7) confirmed previous reports (18, 34) that MΦ express vimentin and smooth muscle α-actin. For these reasons, we evaluated alternative markers for CFb cell-type specificity.
The plasmalemmal receptor DDR-2, a nonintegrin tyrosine kinase receptor that binds to fibrillar collagen (primarily types I and III), has been proposed as a specific marker for the identification of CFbs. It is present in embryonic adult hearts but is absent in SMCs, endothelial cells, and myocytes (15, 35). Furthermore, MΦ are reported to express DDR-1 only (19). We have found that message levels for DDR-2 was enriched in the CD11b− fractions and that CFbs demonstrated very selective immunoreactivity to anti-DDR-2 antibodies; that is, DDR2 immunoreactivity of MΦ (i.e., CD11b+ peritoneal MΦ and CD11b+ cardiac MΦ) was below our detection level. We therefore concur that DDR-2 be considered instead of, or in conjunction with, the traditional cytoskeletal proteins (i.e., vimentin) used to identify CFbs.
CFbs are an essential cell type within myocardium. They are an increasing focus of studies on normal and pathological physiology signaling and remodeling events. Why should investigators be concerned with MΦ contamination of CFb cultures? First, it has been shown that MΦ can survive for several passages when cocultured with fibroblasts. This is likely due to reciprocal feedback of various cytokines and growth factors since MΦ became growth arrested in the absence of fibroblasts or fibroblast-conditioned medium (26, 48). Therefore, once a CFb culture becomes contaminated, it is likely that the MΦ will persist for extended periods unless proactive steps are undertaken to remove them.
Second, both CFbs and MΦ secrete a number of factors that influence wound healing and remodeling. Activated MΦ can secrete TGF-β1 (as can activated peripheral blood monocytes), TNF-α, PDGF, granulocyte MΦ colony-stimulating factor, adrenomedullin, angiontensin II, and S1P (3, 33, 37, 50, 51). As previously stated, TGF-β1 is an important inducer of myofibroblast differentiation (27) and can regulate collagen synthesis, increase stress fiber formation, increase atrial natriuretic peptide levels, and induce connective tissue growth factor (8, 25, 28). S1P, which can be secreted by MΦ, has been shown to induce myofibroblast differentiation in lung fibroblasts (52). Fibroblasts cocultured with monocytes or MΦ have been shown to increase matrix metalloprotease activity and increase collagen gel contraction (9, 58). In summary, since MΦ can modulate myofibroblast and fibroblast phenotyope and/or differentiation, their presence in CFb cultures may complicate the interpretation of studies of CFb function.
As has been shown with vimentin and smooth muscle α-actin cytoskeletal proteins, MΦ may express proteins in common with those expressed in CFbs, an example of which are the S1P receptors. S1P receptors are found ubiquitously on many cell types, and their downstream responses can affect proliferation, motility, intracellular Ca2+ regulation, cell survival, and ion channel activation. The responses depend on the cell type, the complement of S1P receptor expression, and the associated G proteins (reviewed in Refs. 12, 13, 39, 40, and 45). Thus an improved understanding of S1P-mediated responses in CFbs requires highly purified cultures, almost completely free of contaminating cell types that may coexpress similar receptor subtypes.
Involvement of S1P3 in migration of CFbs.
S1P has been shown to be involved both directly and indirectly in the migration of cells. Endothelial cells migrate in response to S1P, whereas SMCs are either not induced or are inhibited to migrate in the presence of S1P (41, 55). It is known that endothelial cells have a relative absence of S1P2 and an abundance of S1P1 and S1P3. In contrast, cultured SMCs have an abundance of S1P2.
S1P2 and S1P3 are involved in Rac- and Rho-mediated migration. Although S1P2 and S1P3 can both activate Gi and G12/13 signaling pathways, S1P2 largely acts through G12/13-Rho. In contrast, S1P3 largely acts through Gi-Rac. Excess levels of S1P2 may therefore inhibit migration not only through positive regulation of Rho but also through negative control of Rac-associated cofactors (41, 46, 47).
S1P levels are increased in cells treated with PDGF or serum (38), and S1P has been shown to be involved in cross-communication of PDGF receptors. Cross-communication of PDGF receptors with S1P can have subsequent effects on migration (6, 44). Although there are conflicting studies regarding the necessity of S1P1 on PDGF receptor activation and PDGF-induced migration (21, 53, 55), S1P3 has been shown to be involved (6).
We did not observe any positive chemotactic response of CFbs to S1P when using migration assays. This may have been due to the combined effects of increased S1P2 (from the serum-deprivation pretreatment step) and the fact that S1P itself had no effect on altering S1P receptor expression levels. CFbs did respond to PDGF-BB as the chemotactic agent, as expected. These results may be explained by a reciprocal change in S1P2 and S1P3 expression levels upon exposure to PDGF-BB (i.e., S1P2 decreases and S1P3 increases after PGDF-BB treatment).
Our results using S1P3-null CFbs confirmed that S1P2 is also upregulated in the absence of serum and that these cells respond similarly to treatments that attenuate an increase in S1P2 in normal CFbs (e.g., the Rho-associated protein kinase inhibitor Y-27632 and PDGF-BB). Furthermore, S1P3-null CFbs did not migrate toward PDGF-BB, demonstrating that S1P3 is involved in PDGF-directed chemotaxis in adult mouse CFbs. However, other studies using S1P3-null mouse embryonic fibroblasts (16) have shown positive migration in response to PDGF, which was further enhanced upon deletion of S1P2. The discrepancy between these sets of results may be related to age differences (i.e., embryonic vs. adult) or that fibroblasts isolated from different anatomical regions do not behave similarly. Interestingly, our results using S1P3-null MΦ showed that this cell type migrated in response to PDGF-BB, similar to S1P3-null mouse embryonic fibroblasts. Since normal MΦ express negligible levels of S1P3 or S1P2, their migration in response to PDGF-BB is not dependent on S1P3. It is therefore possible that mouse embryonic fibroblasts are similarly regulated. Alternatively, it is interesting to speculate whether preparations of mouse embryonic fibroblasts yield highly purified cell populations.
In summary, the complex interplay of S1P receptor subtypes with regard to migration, and the ability of multiple factors to alter expression levels and ratios can complicate the interpretation of functional studies using CFbs.
Limitations of our work.
These studies were conducted using cells isolated from adult mouse ventricles. Although we have unpublished data to suggest that MΦ are also present in cultures of rat CFbs, this may not be applicable to other species. It should also be noted that the conclusions of our studies were largely based on expression levels of mRNA. We performed comparative analysis of mRNA expression levels with protein levels by Western blot analysis to demonstrate that S1P1 is similarly modulated in NIH3T3 fibroblast. However, it cannot be assumed that protein levels for the specific receptor subtypes would be similarly expressed either in relative quantity or with time in CFbs. Nonetheless, the limited sample size (i.e., total cell numbers available from each isolation) and the desire to study S1P receptor expression in early nonpassaged cells, made qPCR an appropriate, but also a necessary, methodology. Lastly, signaling pathways of S1P are increasingly being identified. These are complex and can vary between systems (e.g., cell type, overexpression methods, and concentrations of S1P used).
This work was supported in part by a grant from the American Heart Association Western States Affiliate Program. L. Landeen was supported by a graduate research fellowship from the National Science Foundation.
We sincerely thank Dr. Richard L. Proia of the National Institute of Diabetes and Digestive and Kidney Diseases for the use of his S1P transgenic mice. We also thank Drs. Wilbur Lew and Masahiko Hoshijima of the University of California, San Diego (UCSD) for allowing us the use of essential pieces of equipment and Kim Weldy (UCSD) for maintenance of the S1P3-null breeding colonies.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society