| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Laboratory of Experimental Rheumatology and Neuroendocrino-Immunology, Division of Rheumatology, Department of Internal Medicine I, University Hospital, 93042 Regensburg, Germany
Correspondence: Address all correspondence and requests for reprints to: Rainer H. Straub, M.D., Professor of Experimental Medicine, Laboratory of Experimental Rheumatology and Neuroendocrino-Immunology, Department of Internal Medicine I, University Hospital, 93042 Regensburg, Germany. E-mail: rainer.straub{at}klinik.uni-regensburg.de
 |
Abstract |
|---|
This review delineates how the effects of estrogens are dependent on criteria such as: 1) the immune stimulus (foreign antigens or autoantigens) and subsequent antigen-specific immune responses (e.g., T cell inhibited by estrogens vs. activation of B cell); 2) the cell types involved during different phases of the disease; 3) the target organ with its specific microenvironment; 4) timing of 17ß-estradiol administration in relation to the disease course (and the reproductive status of a woman); 5) the concentration of estrogens; 6) the variability in expression of estrogen receptor 
and ß depending on the microenvironment and the cell type; and 7) intracellular metabolism of estrogens leading to important biologically active metabolites with quite different anti- and proinflammatory function. Also mentioned are systemic supersystems such as the hypothalamic-pituitary-adrenal axis, the sensory nervous system, and the sympathetic nervous system and how they are influenced by estrogens.
This review reinforces the concept that estrogens have antiinflammatory but also proinflammatory roles depending on above-mentioned criteria. It also explains that a uniform concept as to the action of estrogens cannot be found for all inflammatory diseases due to the enormous variable responses of immune and repair systems.
B 
—studies in vitro |
I. Introduction |
|---|
In a previous review, the role of the menopause on proinflammatory cytokine activity has been extensively studied (6). This review focused on the increase of proinflammatory cytokines with the menopause (the fall of estrogens and other gonadal steroids), but this review only marginally touched chronic inflammatory diseases. Another review on gonadal steroids and T and B cell immunity was presented 10 yr ago, but since then, a lot of new information has been generated (2). This is particularly true with respect to chronic diseases that formerly have not been allocated to "inflammatory diseases" such as bone resorption. This is important because incidence rates over age for osteoporosis match incidence rates over age for, e.g., rheumatoid arthritis (RA). Furthermore, the progress in animal models of chronic inflammation was immense, and a good portion of this information has been published after 1996.
There is still the unresolved paradox with respect to the immunomodulating role of estrogens. On one side, we recognize inhibition of bone resorption and suppression of inflammation in several animal models of chronic inflammatory diseases. On the other side, we realize the immune supportive role in trauma/sepsis and the proinflammatory effects in some chronic autoimmune diseases in humans (as an initiating or perpetuating factor). This review examines possible causes for this paradox and suggests a solution in the Discussion (Section XV).
Because the author has been confronted by an enormous quantity of literature (approximately 5200 references in the primary retrieval), this review only focuses on estrogens. It is evident that androgens and progesterone are important in inflammation, but presentation of these subjects was not possible due to space constraints. Throughout the text, tables, and figures, the Système International unit for the concentration is given as moles/liter (abbreviated with M).
|
II. Estrogen Receptors in Inflammation and Hypoxia |
|---|
and ERß, is of outstanding importance because a preponderance of one ER subtype over the other might change estrogen effects (7, 8). For instance, in synovial tissue of patients with RA, macrophage-like and fibroblast-like synoviocytes were positive for ER (9, 10, 11) and ERß (12). Importantly, one study demonstrated higher density of ERß+ cells than of ER+ cells (13). Others demonstrated a higher density of ERß+ cells in relation to ER+ cells in RA synovial tissue compared with controls (12). ERß preponderance was observed in all three synovial compartments investigated: in the lining cell layer, in fibroblasts, and in inflammatory cells (12). Similarly, the amount of ER was lower in T cells of patients with systemic lupus erythematosus (SLE) than controls, but the quantity of ERß was similar, which indicates a relative increase of ERß in relation to ER in SLE patients (14). In animals subjected to trauma and hemorrhage, a general inflammatory condition, ERß mRNA expression was increased, whereas ER expression was decreased (15). These studies suggest inflammation-dependent up-regulation of ERß relative to ER.
It was further demonstrated that hypoxia, which typically accompanies inflammatory conditions, reduced expression of ER (16, 17), and oxidative stress increased the expression of ERß (18). In endothelial cells (E304), oxidative stress increased ERß relative to ER. In activated macrophages, lipopolysaccharide (LPS) plus interferon (IFN)- in the presence of hypoxia increased expression of ERß but not of ER (18). These studies support a concept of up-regulation of ERß relative to ER under hypoxic conditions, which might lead to a preponderance of signaling through ERß pathways.
The preponderance of ERß relative to ER under inflammatory and hypoxic conditions might influence estrogen effects. One might hypothesize that this depends on the time point of estrogen application in relation to the state of an inflammatory disease. A situation before disease outbreak or at the beginning of a disease with slight tissue inflammation and little hypoxia might be governed by a balance of ER and ERß, whereas in the chronic phase of a disease with much inflammation and a higher degree of hypoxia, ERß is increased relative to ER. In such a situation, ERß-mediated cross-modulation of ER, e.g., demonstrated as ERß-mediated inhibition of ER-stimulated IL-1 secretion (8), might play an important role in the chronic disease course.
|
III. Estrogen Modulation of Specific Target Cells |
|---|
A. Estrogens and human peripheral blood mononuclear cells or whole cell cultures
In human peripheral blood mononuclear cells (PBMCs) in the presence of LPS (at 1 ng/ml 
EC50 of LPS), E2 inhibited TNF at concentrations of 10–10 to 10–7 M in male subjects and at 10–8 to 10–7 M in female subjects, but E2 had a stimulating effect in the absence of LPS (20). In whole human blood cultures, E2 at 10–10 to 10–8 M decreased spontaneous secretion of IL-6, TNF, IL-1 receptor antagonist (IL-1ra), IL-1ß, and the ratio of IL-1ß/IL-1ra compared with control, but E2 did not strongly change LPS-stimulated cytokine release (LPS at 500 ng/ml, which is largely higher than the EC50 of LPS) (21). In PBMCs, E2 inhibited TNF release in postmenopausal women with fractures in a dose-dependent manner between 10–12 and 10–6 M but had no consistent effect on PBMCs derived from men or premenopausal women (22).
From these data, it seems that E2 at periovulatory to pregnancy levels inhibited proinflammatory cytokines from PBMCs, which is not unchallenged (23). The contrasting results might be due to divergent effects of E2 on different subtypes of immune cells. Necessarily, it is important to examine the effects of E2 on immune cell subtypes, which is demonstrated in the subsequent sections.
B. Estrogens, the B cell, and antibodies
Mitogen stimulation of T and B lymphocytes resulted in accumulation of IgM-containing and IgM-secreting cells, which was further enhanced by E2 (late follicular to pregnancy levels). This E2 effect was mediated by inhibiting T cell-mediated suppression of B cells (24). A similar phenomenon was observed with human PBMCs stimulated with sheep red blood cells, whereby addition of 10–9 to 10–7 M E2 (maximum at 2 x 10–9 M) augments the antigen-specific immune response by inhibiting CD8+ T cell-mediated suppression of B cells (blocked by tamoxifen, no effect at 5 x 10–10 M) (25). E2 increased antibody-forming cells against sheep erythrocytes (maximum at 1.8 x 10–9 M, no effect at 3.7 x 10–11 M) (26). In PBMCs from normal donors, E2 at pregnancy levels enhanced mitogen-induced generation of antibody-producing cells (27). In human PBMCs, E2 at 10–10 to 10–8 M (IC50, 10–9 M) enhanced IgG and IgM production without altering cell viability and proliferation, and anti-IL-10 antibody partially blocked the E2 effect (28). A similarly enhanced production of anti-double-stranded DNA antibodies was observed with E2 at 10–10 to 10–8 M (IC50, 10–9 M), and E2 increased the B cell-stimulating cytokine IL-10 from monocytes (29). In PBMCs of female rhesus macaques, E2 at 3.7 x 10–9 M increased the frequency of Ig-secreting cells (30).
In women on oral contraceptives (OC), serum levels of IgA and IgG were increased (31). In cycling women, the largest quantities of Ig were detected before ovulation (31). In female rhesus macaques, the frequency of Ig-secreting cells was high in tissues collected from animals in the periovulatory period of the menstrual cycle (30).
In mice, administration of E2 (pregnancy levels) markedly augments the ability of CD5+ B cells to express their autoimmune potential (32). In male MRL lpr/lpr mice (a lupus model), administration of E2 (pregnancy levels) for 31 wk increased the frequency of IgG- and IgM-producing cells (33). In Swiss male mice, administration of E2 (pregnancy levels) for 4 wk increased serum levels of IgG1. In female C3H/N mice, the same treatment increased serum IgG2 levels (33), which was confirmed by others in nonautoimmune C57BL/6J mice with respect to anti-double-stranded DNA antibodies (34). In the collagen-induced arthritis model, E2-treated mice (pregnancy levels) demonstrated an increase in the levels of IgG1 anti-collagen type II antibodies (35).
In Balb/c mice transgenic for the H chain of anti-double-stranded DNA antibodies, E2 (pregnancy levels) rescued naive autoreactive B cells that normally are deleted and caused them to mature to a marginal zone phenotype (36). E2 (at pregnancy levels) further led to the activation of this population, causing an elevation of serum anti-double-stranded DNA antibody titers and renal disease (36). E2 enhanced transitional B cell resistance to apoptosis and expanded the population of marginal zone B cells (36). Estriol (E3) treatment at high pregnancy doses increased IgG1 antibodies in the experimental autoimmune encephalitis (EAE) model in mice most probably by an E3-induced increase of IL-10 production from T cells (37).
From above-mentioned studies, it is clear that E2 can stimulate antibody production by B cells, probably by inhibiting T cell suppression of B cells. In contrast, E2 at high concentrations leads to a suppression of B-lymphocyte lineage precursors (38). In addition, IL-7-responsive B lineage precursors were greatly expanded in genetically hypogonadal female mice. Estrogen replacement in these mice resulted in a dose-dependent reduction in B cell precursors (39). Similarly, pregnancy led to a clear decline in B cell precursors in the bone marrow, which was blocked by ICI 182,780, and an increase was observed in hypogonadal mice (40). E2 at pregnancy levels induced a down-regulation of B lymphopoietic cells in bone marrow of young ovariectomized mice, which is mediated through both ER and ERß (41).
In conclusion, E2 at periovulatory to pregnancy serum levels is able to stimulate antibody secretion under healthy conditions but also in autoimmune diseases, whereas similar serum levels of E2 lead to a suppression of bone marrow B cell lineage precursors. Today we know that autoimmune diseases are not uniform, and sometimes B cells play a major role (as substantiated by the success of anti-CD20 antibody treatment in humans, Rituximab). In chronic inflammatory disorders, where B cells play a decisive role, E2 would promote the disease when autoaggressive B cells are already present, whereas chronically elevated E2 would inhibit initiation of an autoimmune disease when no such B cells are available. This might be a good reason why particularly B cell-dependent diseases such as SLE, mixed connective tissue disease (Sharp syndrome), IgA nephropathy, dermatitis herpetiformis, gluten sensitive enteropathy, myasthenia gravis, and thyroiditis appear in women in the reproductive years, predominantly, in the third or fourth decades of life. It also explains why sometimes these particular diseases are started during or after pregnancy. These ideas are further developed at the end of this review.
C. Estrogens and the T cell
For more than 20 yr, the so-called T helper type 1 (Th1) immune response by CD4-positive T cells was thought to drive cell-mediated immunity leading to tissue damage as in chronic inflammatory autoimmune diseases (42). In contrast, T helper type 2 (Th2) CD4-positive cells drive certain antibody-mediated responses, particularly those that are involved in allergy (42). It is also known that these two pathways reciprocally inhibit each other. Cytokines such as IFN-, IL-12, and TNF were allocated to Th1 reactions, and IL-4, IL-5, and IL-10 to Th2 responses. In recent years, this viewpoint has changed due to the appearance of antiinflammatory T regulatory cells producing TGF-ß and proinflammatory T helper type 17 cells (Th17) producing IL-17 (43). Today, Th17 cells are thought to be the main responsible cells for chronic inflammatory tissue destruction in autoimmune diseases (43). Importantly, this clarifies some unresolved questions such as the ameliorating effects of IFN- in several autoimmune diseases, because IFN- inhibits Th17 cells, which is opposite for TNF that stimulates Th17 cells (Fig. 1
) (43). In addition, IL-4 from Th2-positive cells also inhibits Th17 cells, leading to amelioration of chronic inflammatory autoimmune diseases (Fig. 1) (43). The interested reader is referred to the important recent revision of this subject by L. Steinman (43). With this information in mind, effects of estrogens on T cells need to be revised (Fig. 1). Unfortunately, no direct effects of estrogens on Th17 cells or IL-17 secretion have been described until now. Thus, an influence of estrogens on Th17 cells is deduced from estrogen effects on other T cell cytokines (Fig. 1).
|
was also increased at the same dose. In contrast, the effect of E2 on antigen-stimulated secretion of TNF was biphasic, with enhancement occurring at lower doses at 3.6 x 10–9 and 1.8 x 10–8 M, and inhibition present at high concentrations of 3.6 x 10–8 to 3.6 x 10–7 M (44). The same group reported that estrone (E1) and E3 at pregnancy levels enhance secretion of antigen-induced or anti-CD3-stimulated IL-10 and IFN- from CD4+ cells of MS patients, almost identical to that of E2 (45). In addition, E1 and E3, like E2, had a biphasic effect on TNF secretion, with lower concentrations stimulatory and high doses inhibitory (45). In T cells of patients with MS, E3 at pregnancy levels stimulated secretion of IL-10 and inhibited TNF [via inhibition of nuclear factor-B (NF-B)] (46). As compared with the pretreatment situation, relapsing remitting MS patients treated with oral E3 demonstrated a marked decrease in T cell-dependent delayed type hypersensitivity responses to tetanus (47).
These findings were corroborated by studies in mice. In ovariectomized mice, E2 replacement (pregnancy levels) suppressed the T cell-mediated delayed type hypersensitivity reaction (48, 49, 50). Treatment of CD4+ T cells from nonobese diabetic mice with E2 at pregnancy levels increased secretion of IFN- (51). E2 at pregnancy levels increased -galactosylceramide-induced IFN- synthesis by invariant natural killer T cells (52). Ovariectomized C57BL/6 mice were given E2 replacement (pregnancy levels), which increased IFN- mRNA in concanavalin-A-activated enriched splenic T cells but there was no effect on IL-4 mRNA (53). In ovariectomized mice, E2 replacement (proestrus to early pregnancy levels) results in the selective development of IFN--producing cells (via ER) (54). All these experiments clearly demonstrate the positive effect of E2 at pregnancy levels on IFN- production in T cells. Because IFN- has been allocated a Th17-inhibiting role (Fig. 1
), its increase by E2 at pregnancy doses and the E2-mediated inhibition of TNF must be viewed as a favorable effect in chronic inflammation.
Other important indications for the estrogen-meditated inhibition of the proinflammatory TNF came from bone research. Ovariectomy enhanced T cell production of TNF and subsequent bone loss, which was abolished by E2 replacement therapy (proestrus or early pregnancy levels) (55). Ovariectomy-induced bone loss, which was absent in T cell-deficient nude mice, was restored by adoptive transfer of wild-type T cells but not by reconstitution with T cells from TNF-deficient mice (56).
In contrast to the inhibitory effects of E2 on TNF, E2 can increase antiinflammatory T cell responses. In C57BL/6 mice, E2 at pregnancy levels (not at 10–10 or 10–11 M) increased IL-4 secretion and GATA-3 mRNA levels from ER+ CD4+ T cells, whereas this effect was abrogated in ER-deficient T cells (57). Forkhead box P3 (FoxP3) in CD4+CD25+ cells is an important indicator of T regulatory cells (Fig. 1). In C57Bl/6 mice, E2 (pregnancy levels) is capable of augmenting FoxP3 expression in vitro and in vivo and increases CD25+ cell number, which is reduced in ER-deficient animals. Pregnancy stimulated the expression of FoxP3 protein to a similar degree (58), but progesterone might also play a supportive role.
In addition to effects of estrogens on peripheral CD4+ T cells, estrogens have a strong influence on the development and maintenance of thymic function and, thus, on generation of naive CD4+ and CD8+ T cells. ER-deficient mice have smaller thymi than their wild-type littermates; and E2-induced thymus atrophy was reduced in ER-deficient mice (59, 60). In one study, the CD4+/CD8+ phenotype was not changed in ER-deficient mice (59), but a higher frequency of CD4+CD8+ thymocytes was found in another study (60). In ER-intact animals, E2 (pregnancy levels) induced profound thymus involution in normal and adrenalectomized mice (49, 61). However, E2 did not induce lymphocytopenia in the peripheral organs but stimulated extrathymic T cell numbers in the liver and other organs (61). E2 treatment (pregnancy levels) for 4 wk decreased thymus cellularity and the percentage of CD4+ cells in the spleen, which was not observed in double ER-deficient mice (62). Whether this E2-induced thymus involution is positive or negative in chronic inflammation remains to be established because autoaggressive but also autotolerant T cells might be eliminated.
In conclusion, in humans and mice, E2 at periovulatory to pregnancy levels stimulates IL-4, IL-10, and IFN- but inhibits TNF from CD4+ T cells. In humans and mice, E3 and E2, respectively, at pregnancy levels inhibit T cell-dependent delayed type hypersensitivity. In mice, E2 at pregnancy levels increases the Th2-relevant GATA-3 and the T regulatory cell-relevant FoxP3, which might shift the immune response toward tolerance. Because we know that increased IL-4, IL-10, and IFN- in the presence of low TNF support an antiaggressive immune response (Fig. 1), collectively, these data suggest that E2 at periovulatory to pregnancy levels might be a favorable hormone leading to down-regulation of T cell-dependent immunity. The question appears as to what would happen if E2 falls to postmenopausal levels (metestrus) before initiation or during the course of an autoimmune disease. One might assume that the protecting effects of E2 are getting lost under these conditions.
D. Estrogens and the monocyte/macrophage/dendritic cell
In human promonocytic cells (U937), E2 (late pregnancy levels) inhibited LPS-stimulated IL-6 secretion (63). In human monocytic cells (THP-1), the absence of E2 led to an increase of surface CD16 (stimulatory Fc -receptor III), which after cross-linking stimulated secretion of TNF, IL-1ß, and IL-6 (64). This indicates that E2 normally suppresses these cytokines stimulated via cross-linking of CD16.
In mouse splenic macrophages, pretreatment for 16 h with E2 between 3.6 x 10–11 and 1.8 x 10–9 M decreased LPS (1000 ng/ml, EC50 of LPS in mice and rats)-induced TNF production, which was associated with a decreased NF-B-binding activity (65). In murine bone marrow-derived macrophages, preincubation with E2 at pregnancy levels (maximum at 10–8 M) inhibited LPS-induced TNF release (no effect on cells in the absence of LPS) (66). In murine peritoneal macrophages and RAW 264.7 cells, E2 at pregnancy levels inhibited LPS-stimulated nitrite production and TNF secretion (67). In the mouse macrophage cell line RAW 264.7, preincubation with E2 (proestrus levels) and subsequently activated with LPS inhibited nitric oxide (NO) and TNF release (68).
E2 treatment (pregnancy levels) decreased TNF and IL-12 production in mature mouse dendritic cells (69). This study demonstrated an interesting shift toward cytokines such as IL-4 and IL-10. Transfer of E2 (pregnancy levels)-exposed splenic dendritic cells from Lewis rats obtained on d 12 after immunization with myelin basic protein prevented the expansion of CD4+ T cells and increased proportions of regulatory T cells producing IL-10 and CD4+CD28- suppressor T cells, accompanied with increased IL-10 and IFN- (see explanation in Fig. 1), and reduced TNF production (70). This tolerating effect might be mediated by up-regulation of indoleamine 2,3-dioxygenase (70, 71). These inhibitory activities of E2 might be different at lower levels.
Indeed, in human peripheral monocytes, E2 (early follicular to periovulatory levels) resulted in maximal IL-1 stimulation. At higher concentrations of E2 (late pregnancy levels), a significant reduction in IL-1 activity was observed (72). In LPS-stimulated human monocytic cells, E2 (periovulatory levels) increased IL-1ß and IL-1 mRNA expression (73). The mouse macrophage cell line RAW 264.7 was stably transfected with the human ER and an IL-1ß promoter reporter construct. Pretreatment with E2 at 1.8 x 10–8 to 1.8 x 10–6 M (pregnancy levels, no effect at 1.8 x 10–10 M) markedly enhanced LPS-induced IL-1ß promoter-driven activity (blocked by ICI 182,780) (74). In adherent rat peritoneal macrophages, E2 at proestrus levels (little effect at 3.7 x 10–11 M) stimulated secretion of IL-1 with and without LPS, and IL-1 secretion was higher in cells from female compared with male rats (75). These studies demonstrate that lower levels of E2 particularly increase IL-1ß secretion.
In conclusion, in human and mouse/rat monocyte/macrophage-like cells, secretion of IL-1ß is increased at periovulatory/proestrus to early pregnancy levels, whereas IL-1 secretion is inhibited at high pregnancy levels. It is further obvious that proestrus to pregnancy levels of E2 decreased LPS-stimulated TNF secretion in mouse/rat cells. Thus, with respect to TNF, E2 at pregnancy levels exerts similar effects in macrophages compared with T cells, which in both cell types is most probably due to inhibition of NF-B. The dichotomous effect of E2 on IL-1ß and TNF at high and low concentrations is most probably due to inhibition of NF-B at high concentrations (see Section IV.A).
E. Estrogens and macroglial/microglial cells
In primary rat astrocytes, pretreatment with high doses of E2 at 10–6 M reduced the amyloid ß-peptide and LPS-induced activation of NF-B (76). In primary rat astrocytes, E2 at pregnancy levels inhibited the receptor for activated C kinases-1 (RACK-1, the anchoring protein involved in protein kinase C shuttling), protein kinase C activation, LPS-stimulated TNF mRNA, and inducible NO synthase (iNOS) expression (77). In primary rat cortical astrocyte cultures, E2 (proestrus to pregnancy levels) increased neuronal cell survival and both the expression and release of the neuroprotective cytokines TGF-ß1 and TGF-ß2 (blocked by ICI 182,780), which is mediated by a membrane-bound ER (78).
In primary rat microglia and a microglial cell line (N9), pretreatment with E2 at proestrus to pregnancy levels attenuated phorbol-12-myristate-13-acetate-stimulated superoxide release, LPS-stimulated phagocytic activity, and iNOS protein expression, but did not alter NF-B activation (blocked by ICI 182,780). E2 at similar concentrations induced rapid phosphorylation of the p42/p44 MAPK, and the MAPK inhibitor PD 98059 blocked the antiinflammatory effects of E2 (79). In LPS-stimulated microglial cells (N9), E2 at pregnancy levels increased IL-10 while decreasing release of TNF (similar as in T cells). In the same cell type, E2 at pregnancy levels attenuated the LPS-induced increase of membrane major histocompatibility complex class I molecules, CD40, and CD86, and E2 also decreased the percentage of nonstimulated cells positive for major histocompatibility complex class I and II, CD40, and CD152, Fas, and Fas ligand (FasL), whereas CD80 cell surface staining was increased (80).
In primary rat microglia cells, E2 at 10–10 to 10–8 M (proestrus levels, maximum at 10–9 M, minimum at 10–11 M) inhibited iNOS expression and reduced the accumulation of nitrites and nitrates upon various inflammatory stimuli. Under the same conditions, production of prostaglandin E2 and matrix metalloproteinase (MMP)-9 were also inhibited (blocked by ICI 182,780) (81). In murine microglial cells (BV-2), which naturally express only ERß, E2 at 10–13 to 10–9 M (also E3 at 10–10 to 10–8 M) decreased LPS-stimulated NO production and iNOS expression and expression of cyclooxygenase-2 (82).
In conclusion, experiments with mouse and rat macroglial and microglial cells demonstrate that E2 at proestrus to pregnancy levels exerts neuroprotective effects by increasing TGF-ß and by inhibiting iNOS and NO release, and reducing expression of proinflammatory cytokines and prostaglandin E2 production.
F. Estrogens and natural killer cells
An early study in BALB/c mice described that E2 treatment (proestrus to pregnancy levels) led to unresponsiveness of natural killer cells to poly I.C., which was not mediated by changes of IFN (83). Others demonstrated that E2 treatment (pregnancy levels) inhibits IFN--stimulated natural killer cell activity (84). This was corroborated by a study in different mouse strains, which demonstrated that E2 (pregnancy levels) reduced natural killer cell cytotoxicity. A high degree of suppression was observed in C3H/N, DBA/1, and NZB/W mice, whereas other strains displayed a low degree of suppression (C57BL/6 and MRL lpr/lpr mice) (33). In rhesus monkeys, E2 deprivation by ovariectomy (serum levels, 3.7 x 10–11 M, postmenopausal) reduced natural killer cell activity (85). These studies indicate that E2 has bimodal effects inhibiting natural killer cells at pregnancy levels but stimulating these cells at diestrus to proestrus levels.
G. Estrogens and fibroblast
In human uterine fibroblasts, E2 at 10–10, 10–8, and 10–6 M (no effect at 10–12 M) increases the concentration of basic fibroblast growth factor (86). In human cultured fibroblast-like synoviocytes from patients with RA, E2 at pregnancy levels had no effect on constitutive production of IL-6, but E2 increased IL-1ß-induced IL-6 production (87). Because E2 has some stimulatory effects on IL-1ß secretion of monocyte/macrophages at low levels (see Section III.D), the E2-stimulated IL-1ß-induced IL-6 secretion might be a proinflammatory signal.
In primary human adipose fibroblasts, E2 at pregnancy levels increased TNF receptor 1 mRNA and protein levels (no effect at 10–11 to 10–9 M), whereas E2 at the same concentration decreased TNF receptor 2 mRNA (increase at 10–11 to 10–10 M) (88). Assuming that a similar modulation of TNF receptor 1 and TNF receptor 2 also happens in other cell types, and because the TNF receptor 1 is mainly responsible for tissue-specific autoimmunity and the TNF receptor 2 is mainly responsible for proproliferative effects and tissue damage (89), E2 at pregnancy levels might initiate an autoimmune disease, whereas E2 at postmenopausal levels might support tissue damage. This needs to be investigated in B/T cells and macrophages/dendritic cells.
In human endometrial stromal cells (fibroblasts), E2 at pregnancy levels inhibited the secretion of monocyte chemoattractant protein (MCP)-1 with an IC50 of 10–8 M (slight increase at 10–12 M, no inhibition at 10–11 M and 10–10 M, maximum inhibition at 10–6 M) (90). In murine fibroblasts, E2 (maximum at 3.7 x 10–12 M and 9.9 x 10–10 M, no effect in the middle range of 3.7 x 10–11 M) inhibited serum-stimulated MCP-1 (91). In a mouse stromal cell line (ST-2), E2 (pregnancy levels) induced up-regulation of osteoprotegerin, an osteoprotective factor. The effect was dependent on ER, but not of ERß. Moreover, estrogen withdrawal after 5-d pretreatment, mimicking the event occurring in vivo at menopause, dramatically diminished the expression of osteoprotegerin (92). In rabbit uterine cervical fibroblasts, E2 at pregnancy levels (not at 10–10 M) decreased secretion of procollagenase (precursor of MMP-1) and prostromelysin (precursor of MMP-3), whereas the natural antagonists, the tissue inhibitor of MMPs (TIMP), increased in the presence of E2 (93).
In conclusion, in normal human and mouse/rat fibroblasts, E2 exerts tissue-protective effects by up-regulation of basic fibroblast growth factor, osteoprotegerin, and TIMP, and by down-regulation of the proforms of MMP-1 and MMP-3. In contrast, in synovial fibroblast of patients with RA, E2 stimulates the proinflammatory cytokine IL-6 in the presence of IL-1ß, the latter is most probably not derived from fibroblasts. This demonstrates that a change of the environmental context with an increase of IL-1ß from macrophages might render E2 a proinflammatory factor at lower levels.
H. Estrogens and vascular smooth muscle cells
In human vascular smooth muscle cells, E2 at 10–10 to 10–8 M (also E3) inhibited mRNA expression of platelet-derived growth factor-A, IL-1, and IL-6 but not of TGF-ß (94). Additionally, E1 sulfate at 10–10 to 10–8 M inhibited mRNA expression of platelet-derived growth factor-A, IL-1, and IL-6 (94). These antiinflammatory effects of E1 sulfate are to be expected in the presence of steroid sulfatase (leading to estrogen activation). In human vascular smooth muscle cells, steroid sulfatase expression levels were found to be higher in female aortas with mild atherosclerotic changes than in those with severe atherosclerotic changes and in male aortas regardless of atherosclerotic changes (95). In contrast, expression levels of estrogen sulfotransferase (leading to estrogen inactivation) were higher in female aortas with severe atherosclerotic changes and in male aortas than in female aortas with mild atherosclerotic changes (95). Furthermore, IL-1ß markedly inhibited the expression of the sulfatase but stimulated the expression of estrogen sulfotransferase. This was accompanied by IL-1ß-induced inhibition of E2 production from E1 sulfate (95). This confirms a similar effect of the proinflammatory cytokine TNF on the sulfatase in macrophages and fibroblasts (96, 97). By way of inhibiting the sulfatase and stimulating the sulfotransferase, these cytokines add to a proinflammatory milieu.
In human coronary artery smooth muscle cells (express ER and ERß), E2 at 10–9 to 10–7 M (not at 10–11 and 10–10 M) inhibited MCP-1 mRNA expression and protein production (98). In primary aortic smooth cells, E2 at 10–9 M (no other concentration tested) inhibited iNOS expression and reduced the accumulation of nitrites and nitrates consequent to various inflammatory stimuli (99).
In rat vascular smooth muscle cells, E2 did not influence stimulated IL-6 production (100). In rat vascular smooth muscle cells, E2 at pregnancy levels inhibited cell proliferation and both constitutive and IL-1ß-stimulated NF-B activation (101).
In conclusion, E2 at pregnancy levels by way of several pathways contributes to an antiinflammatory milieu in vascular smooth muscle cells. However, presence of a proinflammatory cofactor such as IL-1ß might change availability of E2 by inhibiting E2 allocation and stimulating E2 degradation.
I. Summary
Figure 2 summarizes the most important effects of high and low concentrations of estrogens on different types of cells involved in chronic inflammation. At pregnancy levels, E2 inhibits important proinflammatory pathways such as TNF, IL-1ß, IL-6, MCP-1, iNOS expression, production of MMPs, and activity of natural killer cells, whereas E2 at the same concentration stimulates antiinflammatory pathways such as IL-4, IL-10, TGFß, TIMP, and osteoprotegerin. At lower concentrations, E2 stimulates TNF, IFN-, IL-1ß, and activity of natural killer cells. This dichotomy of E2 at high vs. low concentrations is not observed for B cells, because antibody production is stimulated throughout the concentration range (Fig. 2). Consequences of these E2 effects for chronic inflammation are summarized in the discussion (Section XV).
|
|
IV. Estrogens and Inflammatory Factors |
|---|
BB is an important factor in proinflammatory signaling that can interact with ER pathways (summarized in Ref. 102). Since the overview of McKay and Cidlowski (102), a substantial amount of new data have been generated in cells relevant for inflammation.
In mouse monocytic cells (RAW 264.7), E2 (pregnancy level) blocked LPS-induced DNA binding and transcriptional activity of the p65 subunit of NF-B by preventing its nuclear translocation (103). Similarly, in primary rat astrocytes, pretreatment with high doses of E2 at 10–6 M reduced the LPS-induced activation of NF-B (76). In lymphocytic HeLa cells, E2 (pregnancy levels) almost completely inhibited phorbol-12-myristate-13-acetate-induced NF-B activation and inhibitor of NF-B (IB) protein degradation (104). In rat vascular smooth muscle cells, E2 (pregnancy levels) inhibited both constitutive and IL-1ß-stimulated NF-B activation (101). In human aorta endothelial cells, E2 at 10–10 to 10–7 M (IC50, 10–9 M) inhibited NF-B-mediated luciferase reporter activity and IL-6 secretion (105). In different cell types, it was demonstrated that ER impairs TNF-induction of IL-6 by preventing binding of c-rel and, to a lesser extent, RelA proteins to the NF-B site on the IL-6 promoter (106). This was corroborated by others who demonstrated that E2 inhibited IL-6 production by interfering with the function of NF-B at the IL-6 promoter (107). In macrophages, E2 blocked LPS-induced DNA binding and transcriptional activity of c-rel (103). All these studies clearly show reciprocal inhibition of ER and NF-B pathways in different cell types. In most occasions, periovulatory to pregnancy levels of E2 have been used. However, other important proinflammatory pathways might be linked to ERs as well.
NF-B activity is suppressed in PBMCs from pregnant women compared with age-matched nonpregnant women (108). In a burn injury model in vivo, E2 treatment resulted in a decrease in splenic NF-B activation (109). In rats receiving transient middle cerebral artery occlusion in vivo, substantial apoptosis and inflammatory responses were observed, including IB phosphorylation, NF-B activation, and iNOS overexpression (110). In this model, E2 treatment (short-term pregnancy levels) produced strong protective effects by reducing infarct volume and neuronal apoptosis and by inhibiting NF-B activation and iNOS overexpression (110). In cerebral arteries of ovariectomized rats, E2 treatment at pregnancy levels reduced cerebrovascular NF-B activity (111).
In conclusion, E2 at periovulatory to pregnancy levels inhibits NF-B activation, which must be viewed as an antiinflammatory signal. The necessary E2 levels needed for NF-B inhibition have been analyzed in human cells infected with an adenoviral NF-B luciferase reporter. It was shown that E2 concentrations equal to or above 10–10 M are necessary to inhibit NF-B activation. At 10–11 M or below (postmenopausal level), no effects were observed (112). Necessarily, only at these higher E2 concentrations is the antiinflammatory effect to be expected.
B. Adhesion molecules
Adhesion molecules are important factors during inflammation because they enable inflammatory cells to migrate to inflamed compartments. Many studies with human and mice/rat cells in vitro demonstrated that estrogens inhibited expression of adhesion molecules at pregnancy levels, which seems to be opposite at low estrogen levels.
For instance, cultured human umbilical vein endothelial cells (HUVECs) were propagated in steroid hormone-free medium and were pretreated with E2 (pregnancy levels) for 48 h before IL-1 activation. In these cells, E2 strongly inhibited IL-1-induced expression of membrane E-selectin, vascular cell adhesion molecule-1 (VCAM-1), and intercellular adhesion molecule-1 (ICAM-1) (113). In HUVECs, E2 at pregnancy levels decreased the expression of TNF-induced VCAM-1 (114). The E2 metabolite 17-epiestriol, which can be generated in vivo, is much more potent than E2 in inhibiting VCAM-1 expression in these cells (U-shaped response curve, no effect at 10–11 and 10–8 M, maximum at 3 x 10–10 M) (attenuated by ICI 182,780) (114).
In an ER overexpressing endothelial cell line (ECV-304), E2 at pregnancy levels inhibited the effects of TNF on VCAM-1 expression and U937 cell adhesion (115). In normal human female iliac artery endothelial cells, E2 at pregnancy levels decreased IL-1ß-stimulated VCAM-1 and ICAM-1 membrane expression (116). In human umbilical cord endothelial cells (ECV-304), the E-selectin promoter is down-regulated by E2 (no effect at 10–11 M, increasing effect between 10–10 and 10–5 M) (117). In cultured brain endothelial cells, E2 at suprapregnancy levels inhibited the basal and IL-1ß-stimulated expression of ICAM-1 (118).
In a functional assay with human monocytic THP-1 cells, 48-h pretreatment with E2 at suprapregnancy doses inhibited nonstimulated and cytokine-stimulated adhesion to human aortic endothelial cells (119). E2 at pregnancy levels decreased adhesion of U937 promonocytes to TNF-stimulated HUVECs (120). In an in vitro model of the vasculature (parallel plate laminar flow chamber), E2 at pregnancy levels potently inhibited monocyte adhesion, and, in parallel, E2 down-regulated Rac1 GTPase activity in monocytes (121). Transfection of monocytic cells with dominant-negative Rac1 significantly decreases adhesion to human endothelial cells (121).
In a rat myocardial ischemia reperfusion injury model, E2 treatment (proestrus levels) lowered myocardial staining of ICAM-1 (122). In a colitis model in mice, E2 treatment (pregnancy levels) decreased expression of ICAM-1 (123). In a rat model of carrageenan-induced pleurisy, E2 (pregnancy levels) inhibited expression of P-selectin and ICAM-1 (124). These studies clearly demonstrate the inhibitory effect of E2 on expression of various adhesion molecules, which appears at proestrus/ovulatory to pregnancy levels. However, the situation might be different at low concentrations of estrogens.
For instance, E2 (diestrus levels) has been shown to up-regulate L-selectin and 4-integrin dependent adhesion pathways in endothelium and lymphocytes (125). Postmenopausal women with coronary artery disease demonstrated increased serum levels of soluble E-selectin, ICAM-1, and VCAM-1, which is an indication of up-regulated membrane-bound adhesion molecules. An increase of E2 serum levels from approximately 3 x 10–11 to 1.5 x 10–10 M decreased serum levels of these soluble adhesion molecules (126). A similar effect of hormone replacement therapy was corroborated by others (127, 128, 129, 130, 131, 132). In addition, treatment of ovariectomized monkeys with conjugated equine estrogens (CEE) decreased serum levels of soluble VCAM-1 but not of soluble E-selectin (133).
In conclusion, E2 at periovulatory/proestrus to pregnancy levels decreased membrane expression of adhesion molecules and cell adhesion to endothelial cells. Some studies indicate that diestrus to ovariectomy/postmenopausal levels of E2 even increased these adhesion molecules. It is interesting that the small increase of E2 from 3 x 10–11 to 1.5 x 10–10 M under estrogen replacement therapy (ERT) in postmenopausal women already reduced soluble forms of adhesion molecules, which most probably indicates also down-regulation of membrane-bound adhesion molecules. Again, a dichotomy of estrogen effects at low and high levels can be observed.
C. IL-1, IL-2, IL-6, IL-8, IL-12, TNF, IFN-—studies in vitro
E2 effects on cytokine secretion depend on cell types, conditions in the milieu, and estrogen concentrations. Tables 1–4 delineate the effects of estrogens on proinflammatory cytokines (particularly of E2 in vitro). Aspects in humans in vivo are extensively demonstrated in Section XII.
|
|
|
|
). Generally, E2 effects are only observed at midfollicular (diestrus) to pregnancy levels. Five studies definitely mention that estrogens at low doses (postmenopausal/metestrus levels) did not lead to marked effects (74, 75, 134) or even exert stimulatory effects on IL-1 (72, 135). Particularly, the latter two studies indicate that E2 at low concentrations might stimulate IL-1ß secretion, whereas it inhibits its secretion at pregnancy levels. This indicates that the well-known dichotomy of estrogen effects is also present here. Only two of 15 studies demonstrate no effects of E2 on IL-1 secretion at all (23, 136). With respect to IL-8, E2 at periovulatory to pregnancy levels inhibited secretion of this chemotactic IL as demonstrated in five studies (Table 1), and no opposite results were reported.
The situation is relatively homogenous with respect to IL-6, where most studies indicate a suppressive effect of E2 at periovulatory to pregnancy levels but little or no effects at early follicular or postmenopausal levels (Table 2). This is substantiated by the fact that in human osteoblastic cells the IL-6 promoter is inhibited by E2 in the absence of a functional ER binding site, which is mediated via down-regulation of NF-B and C/EBP (137). Interestingly, E2 had stimulatory effects on IL-6 secretion in synoviocytes (normal or RA), which happens at pregnancy concentrations (Table 2). Only two of 20 studies demonstrated no effects of E2 on IL-6 secretion at all (23, 136).
Similar to IL-6, most studies demonstrated inhibitory effects of estrogens on TNF secretion at midfollicular to pregnancy levels (Table 3). Of the five studies that demonstrate stimulatory effects of E2 on TNF secretion, two studies delineate that stimulation appears only at lower levels of E2, whereas high pregnancy levels inhibit TNF secretion (44, 45). The latter two studies have often been cited to explain the dual role of estrogens at low vs. high concentrations. Only one of 17 studies demonstrated no effect of E2 at all (23).
E2 effects on IFN- secretion are heterogeneous. A unifying concept seems to exist because E2 stimulates IFN- from T cells but inhibits IFN- from macrophages and dendritic cells (Table 4). This separation might be important because quite different effects of estrogens are to be expected depending on the immune cell involved.
In conclusion, important proinflammatory cytokines are typically inhibited at periovulatory (proestrus) to pregnancy levels of E2, which is evident for IL-6, IL-8, and TNF. Nevertheless, low E2 concentrations were demonstrated to have no or even stimulatory effects. This renders a woman in the postmenopausal phase to a more proinflammatory situation, which might well contribute to the manifestation of chronic inflammatory diseases after the menopause.
D. IL-4, IL-10, TGF-ß—studies in vitro
The interleukins IL-4, IL-10, and TGF-ß are assumed to exert antiinflammatory effects as long as tissue-specific autoimmune diseases are considered. However, these cytokines might have a proinflammatory role in B cell-oriented autoantibody-driven autoimmunity, allergic diseases, or diseases with an overshooting fibrotic repair process (TGF-ß).
It is important to mention that some chronic inflammatory disease can demonstrate different types of immune or inflammatory reactions, whereby T cells, B cells, fibroblasts, or macrophages might play a decisive role depending on the time point of inspection or the subtype of the disease. This has been nicely demonstrated in patients with RA and MS (164, 165, 166, 167). With these prerequisites, we might better understand the role that IL-4, IL-10, and TGF-ß play in chronic inflammatory diseases. The respective immune reaction at a certain time point, at a certain location, and in a given patient defines the set of cytokines used.
With respect to IL-4, all available studies demonstrated that E2 stimulated IL-4 secretion (Table 5). Of the seven studies on IL-10, five demonstrated a stimulating effect of E2 at midfollicular to pregnancy levels. The two studies that report an inhibitory effect on IL-10 (at pregnancy levels) also delineated that E2 at periovulatory doses had no effect (Table 5). All studies devoted to TGF-ß demonstrated that E2 at pregnancy levels stimulated secretion of this cytokine (Table 5).
|
E. Estrogens and growth factors
Studies on TGF-ß mentioned in the previous section are not listed here (78, 171, 172, 174). Similar stimulatory effects of E2 at pregnancy levels were demonstrated for TGF-ß3 in osteosarcoma cells (175, 176) and TGF-ß in human kidney carcinoma cells (177). E2 at follicular to pregnancy levels (not at 10–12 M) increased the concentration of basic fibroblast growth factor (86) and keratinocyte growth factor (178). E2 treatment (diestrus to pregnancy levels) caused a dose- and time-dependent increase in keratinocyte growth factor levels in peripubertal (5-wk-old) and mature (11-wk-old) mice (179). In mouse mesenchymal stem cells (C3H10T1/2), E2 at pregnancy levels activated bone morphogenic protein-2 promoter mainly through ER and only little via ERß, which requires a classical ER element site, but not activator protein (AP)-1 or Sp-1 sites (180). In human osteoblast cell lines, E2 at pregnancy concentration increased steady-state levels of bone morphogenic protein-6 (181).
In human macrophages, E2 at follicular to pregnancy levels and membrane-impermeable BSA-conjugated E2 enhanced nerve growth factor secretion and mRNA expression (via AP-1 binding sites on the promoter) (182).
In conclusion, most of these studies indicate that E2 at periovulatory to pregnancy levels increased important growth factors such as TGF-ß, basic fibroblast growth factor, keratinocyte growth factor, and bone morphogenic proteins. This is in agreement with the important growth-supporting role of E2 during pregnancy. However, several studies also demonstrated that E2 at postmenopausal levels had no effect. In inflammatory diseases, in which the overshooting fibrotic repair response plays a decisive pathogenic role, an E2-stimulated increase of these growth factors might be harmful, whereas it can be positive for tissue repair.
F. Estrogens and nitric oxide
These aspects have been extensively covered in previous reviews from 1997 (183) and 2002 (184). In general, it has been identified that estrogens enhance NO production by the endothelial isoform of NO synthase (NOS) due to increases in expression and level of activation (184). These effects are mediated via ERs, but they are independent of ER element action (nongenomic) (184). Furthermore, it has been demonstrated that E2 increases other NOS isoforms (185, 186, 187, 188, 189). This presentation here only examines relevant aspects with respect to inflammatory diseases.
The known stimulatory effect of E2 on NO production at pregnancy levels (little at 10–10 M) was also observed in human granulocytes and monocytes via the activation of an estrogen surface receptor that is coupled to increases in intracellular calcium (190, 191). In these studies, results were independent of a proinflammatory stimulus. This might be largely different in a situation when a proinflammatory stimulus is added to the system.
For instance, in LPS-stimulated mouse macrophages (RAW 264.7), preincubation with E2 at proestrus to pregnancy levels inhibited NO and TNF release (68). Similarly, E2 at pregnancy levels inhibited IL-1ß-induced NO production in rabbit lacrimal gland acinar cells (192). In primary microglia, aortic smooth cells, and in pleural cells after induction of carrageenan pleurisy, E2 at pregnancy levels inhibited iNOS expression and reduced the accumulation of NO consequent to various inflammatory stimuli (99). E2 and E3 at pregnancy levels inhibit LPS-induced TNF and NO production in primary rat microglia and mouse N9 microglial cells via iNOS inhibition (193). Transient middle cerebral artery occlusion in rats led to a strong inflammatory response with iNOS overexpression, which was strongly inhibited by E2 injection 2 h before occlusion (pregnancy levels) (110). In a pleurisy model, the close interrelation between TNF and IL-1ß and iNOS has been demonstrated, and E2 treatment (pregnancy levels) inhibited production of all three molecules (124). This was also demonstrated in rat primary astrocytes (77).
From this point of view, E2 exerts quite different effects depending on absence or presence of additional inflammatory stimuli such LPS or IL-1ß. In the absence of inflammatory stimuli, E2 increases NO production by stimulating the expression and activity of different isoforms of NOS. However, in the presence of an inflammatory stimulus, E2 at pregnancy levels typically inhibits NO production elicited by cytokine-stimulated iNOS activity. In the latter situation, this might reflect the direct effect of E2 on NF-B activation and proinflammatory cytokine secretion.
G. Estrogens and oxygen radicals
The generation of reactive oxygen species (ROS) is an important proinflammatory factor evolutionarily conserved to overcome infections. However, the generation of ROS is also found in proinflammatory diseases. Thus, it is important to know the effects of E2 on ROS generation.
Several studies using different cell types or species have described that ovariectomy levels of E2 or ovariectomy increased ROS production, whereas treatment with E2 at pregnancy levels decreased ROS production (194, 195, 196, 197, 198, 199). E2 at pregnancy levels protected cardiac cells expressing ERß from H2O2-induced apoptosis concomitant with an increase in the activity of Akt. E2 increased the expression of glutaredoxin as well as -glutamylcysteine synthase, a rate-limiting enzyme for the synthesis of glutathione (blocked by ICI 182,780) (200). In primary rat microglia and a microglial cell line (N9), pretreatment with E2 at proestrus to pregnancy levels attenuated phorbol-12-myristate-13-acetate-stimulated superoxide release (79). A relevant factor for the generation of oxygen radicals is the NADPH oxidase. In different cell lines, E2 at pregnancy levels decreased expression of the NADPH oxidase, which significantly decreased their capacity to generate ROS (201, 202). Another important factor in generating oxygen radicals is the myeloperoxidase. E2 at pregnancy levels blocked stimulated myeloperoxidase expression (203).
In conclusion, the overwhelming majority of studies demonstrate that E2 at proestrus to pregnancy levels inhibits ROS formation. It might well be that this inhibitory activity of E2 might be reverted depending on the cell type and the microenvironmental conditions leading to catechol estrogen formation (204). In addition, ROS formation might well happen below periovulatory/proestrus to pregnancy levels, which lead to a more proinflammatory situation at low concentrations of E2 in the postmenopausal state.
|
V. Estrogens and Apoptosis |
|---|
In human peripheral T lymphocytes, E2 at periovulatory to pregnancy levels for 24 h reduced TNF-induced lactate dehydrogenase release and caspase 3/7 activities, which must be viewed as an antiapoptotic effect (205). In human promonocytic cells (U937), E2 at 10–10 to 10–7 M (not at 10–13 to 10–11 M) increased survival and prevented apoptosis induced by TNF in both undifferentiated and macrophage-like phorbol-12-myristate-13-acetate differentiated cells (206). In human monocytic cells (THP-1), E2 at pregnancy levels prevented apoptosis (207). In Balb/c mice transgenic for the H chain of anti-dsDNA antibodies, E2 (at pregnancy levels) rescues naive autoreactive B lymphocytes normally deleted (36). Support of survival of these proinflammatory cell types is most probably an unfavorable signal in autoimmune diseases.
In chronic brain inflammation, the effect of E2 on neuronal or macroglial/microglial cell apoptosis might be important. In rat primary cortical neurons, 24-h pretreatment with E2 and E3 at pregnancy levels reduced N-methyl-D-aspartate-induced toxicity and the N-methyl-D-aspartate-induced apoptotic changes by enhancing Bcl-2 levels (208). In primary cortical neurons,
Top
Abstract
I. Introduction
, monocyte/macrophage; NK cell, natural killer cell; OPG, osteoprotegerin; PDGF, platelet-derived growth factor; VSMC, vascular smooth muscle cell.