Relationship between placenta growth factor 1 and vascularization, dehydroepiandrosterone sulfate to dehydroepiandrosterone conversion, or aromatase expression in patients with rheumatoid arthritis and patients with osteoarthritis

Authors


Abstract

Objective

Proliferating pannus is in many aspects similar to placental tissue. Both fibroblast-rich tissues have high vascularity, and tissue from patients with rheumatoid arthritis (RA) and patients with osteoarthritis (OA) demonstrates conversion of androgenic prehormones to downstream estrogens. We undertook this study to investigate similarities between proliferating pannus and placental tissue by focusing on angiogenic placenta growth factor 1 (PlGF-1) in patients with OA and patients with RA.

Methods

We used immunohistochemistry to study the presence of PlGF-1, its synovial distribution, and the PlGF-1–expressing synovial cell type. The relationship between PlGF-1 and conversion of the biologically inactive placental prehormone dehydroepiandrosterone sulfate (DHEAS) to the biologically active dehydroepiandrosterone (DHEA) was investigated in mixed synovial cells. The effects of DHEA on PlGF-1 expression were studied by intracellular fluorescence-activated cell sorting analysis.

Results

PlGF-1–positive cells were detected in the lining and sublining areas in patients with RA and patients with OA, and cellular density was similar. Double staining revealed that PlGF-1–positive cells were macrophages. In RA and OA, the density of PlGF-1–positive cells correlated positively with the density of macrophages and the density of type IV collagen–positive vessels. The supernatant concentration of 3H-DHEA after conversion from 3H-DHEAS and the density of aromatase-positive cells were positively correlated with the density of PlGF-1–positive cells only in OA. Low DHEA concentrations (≤10−9M) had stimulatory effects on PlGF-1 when compared to serum concentrations (10−8M to 10−7M) in the monocytic cell line THP-1 and in primary mixed synovial cells.

Conclusion

PlGF-1 functions similarly in inflamed synovium and in the placenta. It is related to vessel formation and, in OA patients, to androgen/estrogen conversion. Evolutionarily conserved functions of PlGF-1 for placental phenomena are obviously also present in synovial inflammation.

Proliferating synovial tissue has histologic and functional aspects of placental tissue, as demonstrated by 5 key factors. 1) Synovial fibroblasts are important elements in the inflammatory process in rheumatoid arthritis (RA) (1, 2), and decidual fibroblasts form the largest part of developing placental tissue (3). 2) In tissue from RA patients and patients with osteoarthritis (OA), androgenic prehormones are typically converted to downstream estrogens, particularly to 16α-hydroxylated estrogens such as 16α-hydroxyestrone and 16α-hydroxylated 17β-estradiol (16αOH-17β-estradiol [estriol]) (4–6), an important mechanism that is also used in the placenta since estriol is a key hormone for fetal development (7, 8). 3) Despite dense vascularization, synovial tissue of patients with RA is devoid of sympathetic innervation (9), which is also an important characteristic of the placenta (10, 11). It seems that nerve repellent factors such as semaphorins play a role in the reduction of innervation in synovial tissue and the placenta (11–14). 4) Neoangiogenesis is a mainstay of the inflammatory process in synovial tissue (15), and it is the key factor during the process of placental decidualization (16). The most important vascular growth factor in synovial and placental tissue is vascular endothelial growth factor (VEGF) (16, 17). Together with VEGF, placenta growth factor (PlGF) is also an important angiogenic factor in the placenta (16). 5) A critical element for neoangiogenesis in the uterus (e.g., during early pregnancy) and in inflamed rheumatoid synovium is hypoxia leading to secretion of hypoxia-inducible factors (HIFs), which promote the secretion of angiogenic factors such as VEGF (18, 19).

Identical pathways in the synovial compartment and in the fetoplacental unit are present for similar functions during tissue remodeling (20). In the present study, we focused on PlGF, which has never been investigated in relation to synovial vascularization, inflammation, and hormone conversion and action but was found in the synovial fluid and tissue of patients with RA (21, 22). In RA, PlGF demonstrated proinflammatory activities by inducing tumor necrosis factor (TNF) and interleukin-6 (IL-6) from synovial fluid mononuclear cells, which was dependent on calcineurin signaling (22). PlGF is a member of the VEGF family of growth factors, encoding a protein with a 50% identity to VEGF in the platelet-derived growth factor domain (23).

Due to alternative splicing, PlGF occurs in at least 3 isoforms: PlGF-1 (PlGF149), PlGF-2 (PlGF170), and PlGF-3 (PlGF221) (24, 25). PlGF-1 binds to VEGF receptor 1 (flt-1), which is expressed predominantly by vascular endothelial cells (26) but also by the synovial fluid mononuclear cells of RA patients (22). In contrast, PlGF-2 binds to neuropilin 1, which is a receptor for semaphorins expressed on nerve fibers and endothelial cells (27). Originally, PlGF was identified in the placenta, where it was proposed to control angiogenesis and trophoblast growth (26). However, PlGF is also expressed during vascular development in other tissues such as heart, lung, and skin (26). Furthermore, PlGF plays a critical role in the control of cutaneous inflammation (26). Although PlGF has been described in RA patients (21, 22), a link to typical placental PlGF functions has not been reported in arthritis.

Other investigators have demonstrated that hypoxia induces VEGF and PlGF from RA synovial cells (28). Activation of angiogenic molecules is mediated by HIF-1α (29). Hypoxia is a critical factor in the synovial tissue of RA patients, as recently investigated in vivo, and blockade of TNF leads to improved tissue oxygenation (30). Thus, expression of angiogenic factors is expected under these conditions.

In this study, we hypothesized that PlGF-1 may play similar roles in both synovial tissue and placental tissue, because it is needed during tissue remodeling. We aimed at studying the cell type expressing PlGF-1 protein in RA and OA synovial tissue. Furthermore, we investigated the relationship between expression of PlGF-1 and synovial vascularity and inflammation. Since availability of biologically active androgens and estrogens is a hallmark of placental physiology, we investigated whether expression of PlGF-1 is linked to conversion of the biologically inactive dehydroepiandrosterone sulfate (DHEAS) to the biologically active dehydroepiandrosterone (DHEA), and whether PlGF-1 expression is linked to the presence of aromatase-positive cells. In addition, we investigated whether DHEA changes the expression of PlGF-1 in the monocytic cell line THP-1 and in primary mixed synovial cells of patients with OA and patients with RA.

PATIENTS AND METHODS

Patients.

In this study, we included 30 patients with longstanding RA fulfilling the American College of Rheumatology 1987 revised criteria (31) and 50 patients with OA. These patients underwent elective knee joint replacement surgery, and they were informed about the purpose of the study and gave written consent. The study was approved by the Ethics Committee of the University of Regensburg. Basic clinical and laboratory data are shown in Table 1. Parameters such as the erythrocyte sedimentation rate, the C-reactive protein level, and rheumatoid factor were measured by standard techniques.

Table 1. Characteristics of the study patients*
 OA patients (n = 50)RA patients (n = 30)
  • *

    Except where indicated otherwise, values are the mean ± SEM. RA = rheumatoid arthritis; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate; RF = rheumatoid factor; NSAIDs = nonsteroidal antiinflammatory drugs; NA = not applicable.

  • Disease duration in patients with osteoarthritis (OA) is often underestimated.

  • P < 0.01 versus OA patients.

  • §

    P = 0.06 versus OA patients.

Age, years69.8 ± 1.165.0 ± 2.0
No. women/no. men (% women/% men)40/10 (80/20)25/5 (83/17)
Disease duration, years4.5 ± 1.013.9 ± 1.7
CRP, mg/liter5.6 ± 0.520.6 ± 5.3
ESR, mm/hour15.8 ± 1.924.9 ± 4.8§
RF positive, no. (%)1 (2.0)16 (53.3)
Synovial lining layer thickness, cells2.0 ± 0.14.4 ± 0.4
Synovial cellular density, cells/mm21,119 ± 701,749 ± 148
Synovial CD3+ T cell density, cells/mm218.2 ± 3.145.7 ± 12.3§
Synovial macrophage density, cells/mm244.5 ± 5.380.1 ± 14.0§
Synovial vascularity, vessels/mm228.2 ± 2.130.3 ± 2.5
Density of sympathetic nerve fibers, fibers/mm21.4 ± 0.20.5 ± 0.1
Therapy  
 NSAIDs, no. (%)21 (42.0)21 (70.0)
 Prednisolone, mg/dayNA5.0 ± 0.8
 Prednisolone, no. (%)NA21 (70.0)
 Methotrexate, no. (%)NA7 (23.3)
 Leflunomide, no. (%)NA8 (26.7)
 Azathioprine, no. (%)NA1 (3.3)
 Sulfasalazine, no. (%)NA3 (10.0)
 Chloroquine/hydroxychloroquine, no. (%)NA4 (13.3)

Synovial tissue preparation.

Synovial tissue samples were obtained immediately after opening the knee joint capsule, the preparation of which was described previously (9). A piece of synovial tissue up to 9 cm2 in size was dissected. A larger piece of the synovial tissue was used to isolate mixed synovial cells (see below) and to perform superfusion on a piece of synovial tissue (see below), and ∼8 pieces of the same synovial area were used for histology. Samples intended for hematoxylin and eosin (H&E) staining and immunohistochemistry were immediately placed in protective freezing medium (Tissue-Tek; Sakura Finetek) and then quick-frozen. Tissue samples for the detection of nerve fibers were fixed for 12–24 hours in phosphate buffered saline (PBS) containing 3.7% formaldehyde and then incubated in PBS with 20% sucrose for 12–24 hours. Thereafter, tissue was embedded in Tissue-Tek and quick-frozen. All tissue samples were stored at −80°C.

Histologic evaluation and determination of synovial innervation.

Histologic evaluation was carried out as described previously (9). From 5–7-μm thick sections, we determined the cell density and lining layer thickness of ∼45 sections from at least 3 different tissue samples per patient (H&E staining). The overall cell density was determined by counting all stained cell nuclei in 17 randomly selected high-power fields (hpf) (400×). The lining layer thickness was analyzed by averaging the number of cells in a lining layer cross section at 9 different locations (400×). To determine the number of T cells (CD3+; DakoCytomation), macrophages (CD163+; DakoCytomation), and vessels (type IV collagen positive; DakoCytomation), 8 cryosections were investigated using alkaline phosphatase–anti–alkaline phosphatase staining, and the number of identified structures was averaged from 17 randomly selected hpf (400×). The number of investigated hpf was derived from a pioneering histologic study by Bresnihan et al (32).

For the determination of synovial innervation, 6–8 cryosections (5–7-μm thick) were used for immunohistochemistry with a primary antibody against tyrosine hydroxylase (the key enzyme for norepinephrine production in sympathetic nerve endings; Chemicon) (9) and an Alexa Fluor 546–conjugated secondary antibody (Molecular Probes). The numbers of sympathetic nerve fibers per mm2 were determined by averaging the number of stained nerve fibers (minimum length 50 μm, determined through a micrometer eyepiece) in 17 randomly selected hpf (400×).

Immunohistochemistry of PlGF-1 and VEGF165.

Human PlGF-1 and VEGF165 were detected and quantified using immunohistochemistry. Unfixed (for PlGF-1 detection) and acetone/methanol-fixed (for VEGF165 detection) cryosections were treated briefly with 3% H2O2 in 1× PBS for 10 minutes to block endogenous peroxidase. After blocking with 10% fetal calf serum (PAN-Biotech) and with 10% goat serum (Dako) for 45 minutes, cryosections were incubated with primary antibody (2 μg/ml rabbit polyclonal anti–PlGF-1 IgG [Abcam], 5 μg/ml mouse monoclonal anti-VEGF165 IgG2b [R&D Systems]) or isotype control (for PlGF-1, 2 μg/ml rabbit IgG [Abcam]; for VEGF165, 5 μg/ml mouse IgG2b [Chemicon]) for 3 hours at room temperature. After incubation with biotinylated secondary antibody (for PlGF-1, biotinylated polyclonal goat anti-rabbit IgG [Dako]; for VEGF165, biotinylated polyclonal goat anti-mouse IgG [Dako]) for 90 minutes, the sections were treated with streptavidin–horseradish peroxidase (GE Healthcare). Visualization was performed with diaminobenzidine tetrahydrochloride hydrate (ImmPACT DAB; Vector). Quantification of the PlGF-1– and VEGF165-positive cells was done as described above.

For the localization of the PlGF-1–positive cells, the same polyclonal antibody against PlGF-1 was incubated together with monoclonal antibodies against T cells (anti-CD3), activated macrophages (anti-CD163), or fibroblasts (anti–prolyl 4-hydroxylase) (all from DakoCytomation). After 3 washes, staining of the positive cells was achieved by incubating the sections with the respective secondary Alexa Fluor 488– and Alexa Fluor 555–conjugated F(ab′)2 fragments (Molecular Probes). Nuclei were stained with Vectashield mounting medium with DAPI (Vector).

Steroid hormone conversion assay and thin-layer chromatography (TLC) analysis.

Mixed synovial cells were isolated by enzymatic digestion of fresh synovial tissue for 1–2 hours at 37°C (Dispase Grade II; Roche Biochemicals). The synovial cells were resuspended in serum-free RPMI 1640 medium (Sigma) supplemented with 1% penicillin/streptomycin (Life Technologies), 0.1% amphotericin B (Bristol-Myers Squibb), and 8 μg/ml ciprofloxacin (Bayer). A total of 35,000 cells were seeded into 48-well plates in a volume of 250 μl and cultured for 24 hours (primary early culture of mixed synovial cells without further selection) in a humidified atmosphere with 5% CO2 at a temperature of 37°C.

To measure conversion of DHEAS to DHEA, 1,2,6,7–3H-DHEAS (PerkinElmer Life Sciences) was added to each dish to achieve a final concentration of 0.2 μmoles/liter (normal serum concentration ∼6 μmoles/liter, serum concentration in RA patients 1.2 μmoles/liter). The synovial cells were cultured with 1,2,6,7–3H-DHEAS for an additional 24 hours in a humidified atmosphere with 5% CO2 at a temperature of 37°C.

Steroids in the supernatant were extracted twice with ethyl acetate at pH 1.0 as described previously (33). After centrifugation, the organic phase was removed and evaporated to dryness by a nitrogen stream. Dried extracts were dissolved in 50 μl 100% ethanol. Samples of 10 μl were then spotted on silica gel 60 F254 TLC aluminum sheets (Merck). A mixture of unlabeled carrier steroids (DHEAS, DHEA; all from Sigma) was also spotted on each sheet in order to control chromatography later by ultraviolet illumination. The silica gel aluminum sheets were then developed in 1 dimension with the solvent system chloroform:diethylether (1:1). After developing, aluminum sheets were marked with 1,2,6,7–3H-DHEAS at 1 μM, 0.1 μM, and 0.01 μM in order to produce a standard curve on every aluminum sheet as described previously (33). Establishing a standard curve on every TLC sheet allows the exact determination of molar amounts of converted 3H-DHEA. Then, the sheets were exposed to tritium storage phosphor screens (GE Healthcare) for 18–24 hours. For the measurement of the activated screens, we used a PhosphorImager SI system (GE Healthcare). Spontaneous 3H-DHEA generation from 3H-DHEAS and background radioactivity were subtracted from each radioactive spot. TLC spots were assigned to steroid hormones by comigration with authentic standards in the solvent system and visualized by heating to 130°C after dipping plates into 3% copper(II) acetate in 8% phosphoric acid.

Intracellular quantification of PlGF by fluorescence-activated cell sorting (FACS) analysis.

Primary mixed synovial cells from patients with RA and patients with OA were cultured in 24-well ultra-low attachment plates (Corning Life Sciences) for 24 hours with or without stimulation. Cells were harvested and centrifuged for 5 minutes at 300g. Cells were fixed with 100 μl 1.5% formaldehyde in PBS for 10 minutes at room temperature. Then, 1 ml of 100% ice-cold methanol was added, and cells were kept at −20°C for 20 minutes (at this point cells could be stored for up to 1 month in a freezer). Methanol makes cells permeable to enable intracellular antigen staining. To remove residual methanol, cells were washed with PBS/1% bovine serum albumin (BSA). Synoviocytes were stained for 30 minutes with 100 μl of a polyclonal rabbit anti-PlGF antibody (0.5 μg/ml in PBS/1% BSA; Abcam). Cells were washed and incubated with secondary antibody for 30 minutes (biotinylated polyclonal goat anti-rabbit IgG, 5 μg/ml in PBS/1% BSA). Thereafter, cells were washed and incubated with 3 μg/ml streptavidin–RPE (Dako) for 30 minutes in the dark. Analysis of the cells was performed using a Beckman Coulter EPICS XL-MCL flow cytometer and WinMDI (freeware).

Stimulation of THP-1 cells with DHEA and expression of PlGF-1.

In order to detect a direct effect of DHEA on the expression of PlGF-1 in macrophages, we used the monocytic cell line THP-1 stimulated with 500 units/ml interferon-γ (Roche Diagnostics). A total of 150,000 cells were seeded in 3 ml using 24-well plates for 2 days with different concentrations of DHEA (10−9, 10−8, and 10−7 moles/liter). At the end of incubation, cells were fixed with cold acetone on glass slides and subjected to immunocytochemistry.

Superfusion technique for the study of synovial tissue samples.

In order to use additional soluble readout parameters of synovial inflammation, IL-6 and IL-8 were detected in the superfusate of synovial tissue samples. The superfusion technique has been described previously in detail (34). Briefly, 6 pieces of synovial tissue samples were placed in superfusion chambers with 80-μl volume and then superfused with serum-free culture medium (RPMI 1640, 25 mM HEPES, 1% penicillin/ streptomycin, 30 μM mercaptoethanol, 0.57 mM ascorbic acid, 1.3 mM calcium; all from Sigma). Superfusion was performed for 2 hours at 37°C using a flow rate of 66 μl/minute (1 sample per chamber, 8 chambers in parallel). Superfusate was collected at 2 hours and stored at −20°C for later bulk analysis of IL-6 and IL-8 by enzyme-linked immunosorbent assay (ELISA) (antibody pairs were from BD Biosciences PharMingen).

Statistical analysis.

All data are given as the mean ± SEM. Groups were compared by the nonparametric Mann-Whitney test. Correlations were calculated by Spearman's rank correlation analysis and demonstrated as linear regression lines using SPSS software, version 17. P values less than 0.05 were considered significant.

RESULTS

Localization and cell type of PlGF-1–expressing cells.

Immunohistochemical staining demonstrated that in patients with RA and patients with OA, PlGF-1–positive and VEGF165-positive cells were found in the proliferating lining and sublining areas (Figure 1). Particularly in RA but also in OA, vessels also stained positive for PlGF-1 and VEGF165 (Figure 1).

Figure 1.

Immunohistochemical analysis of placenta growth factor 1 (PlGF-1)–positive and vascular endothelial growth factor 165 (VEGF165)–positive cells in synovial tissue from patients with rheumatoid arthritis (RA) and patients with osteoarthritis (OA). The control is also shown. Original magnification × 200.

Immunohistochemical double staining demonstrated that PlGF-1–positive cells did not costain for prolyl 4-hydroxylase–positive synovial fibroblasts and CD3+ T cells (Figures 2A and B). However, positive costaining was found in CD163+ macrophages (Figure 2C). Most of the CD163+ macrophages were positive for PlGF-1 (Figure 2C).

Figure 2.

Localization of PlGF-1–positive cells in synovial tissue from patients with OA and patients with RA. A, Double immunohistochemical staining of prolyl 4-hydroxylase (a fibroblast marker; green fluorescence) and PlGF-1 in synovial tissue from an OA patient. The result was identical for synovial tissue from an RA patient (not shown). B, Double immunohistochemical staining of cells positive for CD3 (a T cell marker; green fluorescence) and PlGF-1 in synovial tissue from an RA patient. The result was identical for synovial tissue from an OA patient (not shown). In A and B, no double staining is visible in the overlay. C, Double immunohistochemical staining of CD163+ macrophages (green fluorescence) and PlGF-1–expressing cells in synovial tissue from an OA patient and an RA patient. The overlay indicates that most macrophages are also positive for PlGF-1 (arrowheads, yellow color). Original magnification × 400. Bars = 70 μm. See Figure 1 for definitions.

In order to link PlGF-1 expression and macrophages, a correlation analysis was carried out for the density of PlGF-1–positive cells and macrophages. The density of PlGF-1–positive cells and CD163+ macrophages correlated positively in OA and RA (Figures 3A and B). This positive correlation was more pronounced in patients with RA (Figure 3B). Both findings indicate that macrophages produce and express PlGF-1 in the natural environment of intact synovial tissue. In patients with RA and patients with OA, there was no correlation between the density of PlGF-1–expressing cells and lining layer thickness, the density of CD3+ T cells, and overall synovial cellularity (data not shown).

Figure 3.

A and B, Correlation of the density of PlGF-1–positive cells with the density of CD163+ macrophages in OA patients (A) and RA patients (B). RRank = Spearman's rank correlation coefficient. C, Comparison of density of PlGF-1–expressing cells in male and female OA patients and RA patients. Data are shown as box plots. Each box represents the 25th to 75th percentiles. Lines inside the boxes represent the median. Whiskers outside the boxes represent the 10th and the 90th percentiles. Statistical significance in C was determined using the Mann-Whitney test. See Figure 1 for other definitions.

Sex difference in the density of PlGF-1–expressing cells.

Immunohistochemical staining demonstrated that male patients with OA compared to female patients with OA (Figure 3C). Similar results were found for RA patients, in whom the difference was not statistically significant (Figure 3C). Despite a marked difference in markers of inflammation between RA and OA (Table 1), there was no statistically significant difference in the density of PlGF-1–positive cells between RA patients and OA patients (mean ± SEM 2.3 ± 0.4 cells/mm2 and 3.4 ± 0.6 cells/mm2, respectively; P = 0.58).

PlGF-1 and vascularity.

In both OA patients and RA patients, the density of PlGF-1–expressing cells correlated positively with the density of type IV collagen–expressing vessels, which was more obvious in RA patients than in OA patients (Figures 4A and B). Interestingly, no such correlation existed for the density of VEGF165-expressing cells with the density of type IV collagen–expressing vessels in RA patients (Figure 4D), whereas both the density of PlGF-1–positive cells and the density of VEGF165-expressing cells correlated closely with the density of type IV collagen–expressing vessels in OA patients (Figures 4A and C).

Figure 4.

Correlation of the density of PlGF-1–expressing cells (A and B) or of VEGF165-expressing cells (C and D) with the density of type IV collagen–positive vessels in OA patients (A and C) and RA patients (B and D). RRank = Spearman's rank correlation coefficient (see Figure 1 for other definitions).

In this chronically inflamed tissue, vascularity did not differ between RA patients and OA patients (Table 1). Furthermore, in OA patients and RA patients, the density of PlGF-1–expressing cells did not correlate with the density of VEGF165-expressing cells (rs = 0.075 and rs = 0.294, respectively; P not significant [NS]). The density of VEGF165-expressing cells was significantly higher than that of PlGF-1–expressing cells, in both RA patients and OA patients (in RA patients, 20.5 ± 1.7 cells/mm2 versus 2.3 ± 0.4 cells/mm2; P < 0.001) (in OA patients, 21.7 ± 2.5 cells/mm2 versus 3.4 ± 0.6 cells/mm2; P < 0.001), but VEGF165 expression did not differ in the chronically inflamed tissue of RA patients and OA patients (20.5 ± 1.7 cells/mm2 and 21.7 ± 2.5 cells/mm2, respectively; P NS) (see also Figures 4C and D).

Another marker of inflammation was superfusate IL-6 and IL-8 from superfused synovial tissue. A Spearman analysis did not demonstrate a correlation between superfusate IL-6 and the density of PlGF-1–positive cells in OA and RA (for OA, rs = 0.105; for RA, rs = 0.079) (P NS) or between superfusate IL-8 and the density of PlGF-1–positive cells (for OA, rs = 0.063; for RA, rs = 0.213) (P NS).

PlGF-1 expression, androgen conversion, and aromatase expression.

Conversion of biologically inactive DHEAS to active DHEA is an important characteristic of the developing placenta. Similar conversion exists in synovial tissue, which can be blocked by TNF (33). Since the usual serum concentration is between 10−7M and 10−8M, TNF-induced blockade of DHEAS conversion to DHEA can lead to much lower levels of active DHEA in inflamed tissue (expected concentration ≤10−9M). We hypothesized that there is a link between expression of PlGF-1 and androgen and estrogen conversion in primary cells of the synovium.

In patients with OA, there was a strong positive correlation between the density of PlGF-1–expressing cells and DHEAS to DHEA conversion (Figure 5A). Such a correlation was not observed in RA synovial tissue (Figure 5B). In a further analysis, we observed a positive correlation between the density of aromatase-positive cells and PlGF-1–expressing cells in OA (rs = 0.332, P = 0.030) but not in RA (rs = 0.187, P = 0.40). This indicates that similar to DHEAS conversion (by DHEAS sulfatase), PlGF-1 is positively linked to an additional converting enzyme (for local estrogen generation).

Figure 5.

Interrelationship of dehydroepiandrosterone (DHEA) and PlGF-1. A and B, Correlation of the density of PlGF-1–expressing cells with DHEA (converted from DHEA sulfate [DHEAS]) in OA patients (A) and RA patients (B). Converted 3H-DHEA (*DHEA) is shown as a percent of applied 3H-DHEAS (*DHEAS) per 1 million investigated synovial cells. RRank = Spearman's rank correlation coefficient; NS = not significant. C and D, DHEA modulation of PlGF-1 protein in primary mixed synovial cells from OA patients (C) and RA patients (D). The dotted line indicates the control level without DHEA in the supernatant. Data are shown as box plots. Each box represents the 25th to 75th percentiles. Lines inside the boxes represent the median. P values at bottom of panels represent the comparison of all groups by analysis of variance on ranks. See Figure 1 for other definitions.

DHEA modulates PlGF-1 protein.

This abovementioned positive correlation between the density of PlGF-1–expressing cells and DHEAS to DHEA conversion (Figure 5A) prompted us to investigate the influence of DHEA on expression of PlGF-1 in monocyte-derived macrophages, because this steroid hormone might be involved in production of PlGF-1 (in the placenta, DHEA is an important mediator of placental function). In a culture assay with monocyte-derived macrophages (THP-1 cells), DHEA influenced expression of PlGF-1 in a concentration-dependent manner, where low concentrations of DHEA led to increased expression, while physiologic concentrations led to inhibition (further information is available at www.uni-regensburg.de/Fakultaeten/Medizin/Innere_1/aknei/Lowin_PlGF.htm).

Using a PlGF-1 ELISA, we were unable to detect the protein in the supernatant, which can be explained by typical rapid degradation, binding, or internalization. However, intracellular FACS analysis of PlGF-1 protein demonstrated a similar inverse dose-response curve when using primary mixed synovial cells from patients with OA (Figure 5C). DHEA had no effect on PlGF-1 protein, as tested in primary mixed synovial cells of patients with RA (Figure 5D). We also tested the effect of PlGF-1 on DHEAS to DHEA conversion, but PlGF-1 did not influence this particular step of steroid hormone metabolism (data not shown).

Association between medication and presence of PlGF-1–expressing cells.

With respect to all drugs shown in Table 1, no association was observed between the presence/absence of a drug therapy and the density of PlGF-1–positive cells (data not shown).

DISCUSSION

This study investigated expression of PlGF-1 in the synovial tissue of patients with RA and patients with OA. It was demonstrated that PlGF-1–expressing cells were present in the tissue samples of all RA patients and OA patients investigated. The major cell producing PlGF-1 was the macrophage. Typically, cells were located in proliferating sublining and lining areas. However, despite a clear difference in histologic inflammation markers, no significant difference was found in the density of PlGF-1–positive cells between the 2 patient groups. Since macrophages are twice as dense in RA tissue as in OA tissue, expression of PlGF-1 protein might be stronger in OA macrophages.

The presence of PlGF-1 has been described in synovial fluid and plasma samples from patients with RA and patients with other inflammatory arthropathies (21). Bottomley et al further reported that synovial fluid PlGF and VEGF levels correlated positively with the total leukocyte count (21). Another group found PlGF-1 in the synovial tissue of patients with RA (22). They also reported that synovial fibroblasts start to produce PlGF-1 when cultured over several days, and they did not find PlGF-1 in mononuclear cells of synovial fluid (22). Despite these findings, we hypothesized that macrophages are the most important source of PlGF-1. Accordingly, double immunofluorescence staining demonstrated PlGF-1 in synovial tissue macrophages but not in synovial tissue fibroblasts and T cells. In addition, density of PlGF-1–expressing cells correlated positively with the tissue density of macrophages but not with lining layer thickness, the density of T cells, overall synovial cellularity, and tissue superfusate IL-6 or IL-8.

From our data, we hypothesize that released PlGF is most likely secreted by synovial macrophages, which are capable of producing a multitude of different paracrine factors. These results seem to contradict the recent report by Yoo et al (22); however, that group studied RA synovial fibroblasts in culture only and not isolated synovial macrophages. In addition, they did not perform double immunohistochemical staining. It might well be that synovial fibroblasts start to secrete PlGF only after isolation; however, in situ, this angiogenic factor is mainly produced by CD163+ macrophages.

Similar to VEGF, PlGF was described to be an important angiogenic factor in the placenta (16). Both mediators belong to the VEGF family of growth factors. The angiogenic role of VEGF in RA has been extensively described (17, 19), but the role of the associated angiogenic molecule PlGF-1 remains unclear. In our analyses, we found a positive correlation between the density of PlGF-1–expressing cells and vascularity, particularly in RA patients, which, in the light of its placental functions, is indicative of an angiogenic activity. The positive correlation between vessel density and expression of PlGF was even stronger than the positive correlation of vascularity and VEGF expression. These data indicate that PlGF has similar roles in inflamed and placental tissue.

In addition, in RA patients and OA patients, the density of PlGF-1–expressing cells was higher in male patients than in female patients. This is of particular interest since only very few markers were found to be different in male and female patients. However, the implication of the different expression needs to be established. One might speculate that in such a situation PlGF-1–induced angiogenesis may be higher in men than in women, although vessel density was not different between men and women.

Placental function is tightly coupled to the presence of biologically active DHEA, which can be converted to downstream 16α-hydroxylated estrogens such as estriol (16α-hydroxylated 17β-estradiol). Vascular growth and hormone availability are also tightly linked in the placenta, where estriol is able to induce VEGF expression (20, 35). In the present study, we demonstrated that expression of PlGF is positively correlated with the important conversion step of biologically inactive DHEAS to biologically active DHEA, which was observed in OA patients only. Furthermore, in OA patients but not in RA patients, the density of PlGF-1–positive cells was positively correlated with the density of aromatase-expressing cells. This association with important hormone conversion steps existed only in OA patients. At this point, we have to mention that disease duration was longer in RA patients than in OA patients, which might have influenced the results in the 2 patient groups (i.e., more chronic situation in RA).

In a further study in monocyte-derived macrophages and primary mixed synovial cells from patients with OA (but not from patients with RA), low DHEA concentrations (≤10−9M) induced higher intracellular PlGF-1 expression compared to DHEA concentrations typical in serum (10−8M to 10−7M). Thus, blockade of DHEAS to DHEA conversion—the activation of biologically inactive DHEAS—might result in higher PlGF-1 expression, because low DHEA concentrations are to be expected. TNF can block the conversion from DHEAS to DHEA, as previously demonstrated (33). It might well be that low DHEA concentrations yield different intracellular followup products as compared to high concentrations. Interestingly, the observed correlation obviously does not exist in RA patients. This may depend on different hormone conversion in local synovial cells, as we recently demonstrated for DHEAS to DHEA conversion in RA patients versus OA patients (33). In that study, it was demonstrated that the downstream conversion product DHEA is significantly lower in RA patients than in OA patients, which can explain some discrepancies.

In conclusion, the positive correlation between PlGF-1 expression and vascularity indicates that PlGF-1 exerts similar effects in synovial tissue and placenta. Furthermore, PlGF-1 seems to be linked to androgen activation (generation of the biologically active DHEA) at least in OA patients (conversion is generally lower in RA patients). Both findings, angiogenesis and androgen allocation, are indicative of placental functions. In addition, the active androgen DHEA increases PlGF-1 protein, which might depend on DHEA conversion into angiogenic estrogens, which could be a therapeutic target. We have demonstrated a concept explaining why most of the changes observed particularly during the symptomatic phase of an inflammatory disease are not specific for a certain inflammatory process (20, 36). This theory, the presentation of which goes beyond the scope of this article, can explain why many similar phenomena appear in very different chronic inflammatory diseases (36). Our results support the hypothesis that some phenomena positively selected for other inflammatory processes, in this case placenta formation, are also used in synovial inflammation in patients with RA and OA.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Straub had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Straub.

Acquisition of data. Lowin, Weidler, Jenei-Lanzl, Capellino, Baerwald, Buttgereit, Straub.

Analysis and interpretation of data. Lowin, Weidler, Jenei-Lanzl, Capellino, Baerwald, Buttgereit, Straub.

Ancillary