The Selective Estrogen Receptor Modulators, Tamoxifen and Raloxifene, Impair Dendritic Cell Differentiation and Activation

DC culture reagents

DC were cultured in either hormone-deficient (DEF) or normal (NORM) RPMI 1640 medium. DEF medium contained 10% charcoal dextran-stripped FBS (Omega Scientific) lacking phenol red. Charcoal dextran stripping of FBS extracts steroid hormones and reduces their levels below the detection limits of a standard RIA (33). Phenol red was omitted because it can have weak estrogenic activity at the concentration that is present in RPMI 1640 medium (34). NORM medium contained 10% regular FBS (Omega Scientific) with phenol red. Aside from these two differences, both medium formulations were the same and consisted of 2 mM glutamine, 10 mM HEPES buffer, 1 mM sodium pyruvate, 50 μM 2-ME, 100 U/ml penicillin, and 100 μg/ml streptomycin. Medium conditioned by the GM-CSF-producing cell line, J558L (35), was added to DC culture medium at 5% to promote DC development. J558L cells were kindly provided by Dr. R. Steinman (Rockefeller University, New York, NY). Estrogen, in the form of E2 (Sigma-Aldrich), was solubilized in 95% ethanol and was diluted into DEF medium cultures at the concentrations indicated from a 1 μM stock solution. Stock solutions of tamoxifen, 4-hydroxytamoxifen (4OH), and raloxifene or comparable amounts of vehicle alone were prepared and added to cultures at the indicated concentrations. Tamoxifen (Sigma-Aldrich) stock solutions were prepared in DMSO or 95% ethanol at 10 mM; 4OH (Sigma-Aldrich) stock was prepared at a concentration of 50 mM in 95% ethanol, and raloxifene (Sigma-Aldrich) stock was prepared at a concentration of 10 mM in DMSO.

DC generation from murine BM

DC were generated from murine BM precursors as previously described (36). BM was isolated from 6- to 12-wk-old male and female 129S6 mice (Taconic), which were housed at the City of Hope Animal Resource Center (Duarte, CA) in compliance with federal and institutional guidelines. BM was flushed from femurs and tibiae with PBS, resuspended in medium containing GM-CSF at 2 × 10⁵ cells/ml, and plated at a cell density of roughly 3.3–3.5 × 10⁴ cells/cm². Cells were cultured in standard 100-mm nontissue culture-treated petri dishes, 100-mm Teflon dishes (Savillex), or 6-well plate Teflon inserts (Savillex) to reduce cell adherence (for 6-well inserts, BM cells were resuspended at 1.65 × 10⁵ cells/ml). Cultures were grown for a total of 7–8 days to allow for sufficient DC differentiation from BM precursors. Various concentrations of E2, tamoxifen, and raloxifene were added alone or in combination from the first day of culture (day = 0) and cells were fed every 3 days with medium containing GM-CSF. On day 3, cells were fed by adding a volume of fresh medium equivalent to that of the culture. For feeding on day 6, one-half of the culture supernatant was first removed and centrifuged to recover any cells before disposal. The cell pellet was then resuspended in an equal volume of fresh medium and added back to the original culture. At each feeding, E2, tamoxifen, and raloxifene were replenished at the indicated concentrations. For DC maturation, cultures were incubated for 12–14 h over a time period spanning days 6–7 or days 7–8 with 0.2–2.0 μg/ml LPS from Escherichia coli serotype O55:B5 (Sigma-Aldrich).

Flow cytometry

For flow cytometric analyses, murine DC cultures were analyzed on day 7 or 8 by three- or four-color surface staining. FACS staining buffer (PBS, 5% newborn calf serum, 0.1% sodium azide) was used for all washes and incubations. FACS staining was performed by standard procedures and is later described. Following harvest, 2 × 10⁵ cells were incubated for at least 5 min with an unlabeled anti-CD16/32 mAb (2.4G2) to block FcRs. Then surface staining was performed by incubating samples for 15 min with combinations of optimally titered, fluorochrome, or biotin-conjugated mAbs specific for CD11c, CD11b, CD86, CD80, MHC class II, CD40, and TLR4. After one wash with 2 ml of buffer, allophycocyanin- or PerCP-conjugated streptavidin was added for another 15 min incubation to detect biotin conjugates. This step was followed by one wash with 2 ml of buffer. In most instances, cells were fixed with 1% paraformaldehyde before analysis with the exception being when anti-CD40 was included in staining combinations. Isotype control Abs tested by this staining procedure did not bind nonspecifically to cells from our cultures.

Unlabeled anti-CD16/32, anti-MHC class II (Y3P) biotin, and anti-CD11c (N418) Alexa 488 were produced by the corresponding hybridomas (American Type Culture Collection), and were purified and labeled in our laboratory. Streptavidin-allophycocyanin, streptavidin-PerCP as well as the following mAbs: anti-CD80 biotin (16-10A1), anti-CD86 biotin and PE (GL1), anti-CD40 FITC (3/23), and anti-CD11b PerCP-Cy5.5 (M1/70) were from BD Pharmingen. Anti-CD11b PE (M1/70) and anti-TLR4 biotin (MTS510) were from eBioscience. Anti-TLR4 PE (MTS510) was from BioLegend. Samples were run on a BD Biosciences FACSCalibur instrument capable of four-color detection and analyzed with CellQuest or FlowJo software.

OVA and dextran uptake

Chicken OVA labeled with Alexa 488 (OVA₄₈₈) and dextran-labeled (10,000 m.w.) fluorescein (DexFITC) were obtained from Molecular Probes. The conditions for Ag uptake were chosen based upon methods used in similar studies (37, 38). At day 7, BMDC were harvested from DEF medium cultures and washed with PBS. Then 2 × 10⁵ cells were resuspended in fresh DEF medium at a concentration of 1.5 × 10⁶ cells/ml and were placed back into culture with either 50 μg/ml OVA₄₈₈ or 100 μg/ml DexFITC for 30 min at 37°C. Ag uptake controls were conducted at 0°C with all other conditions being equal. The samples were then washed two times in FACS staining buffer and surface stained for CD11c, CD11b, and MHC class II as described and analyzed by FACS analysis. No E2, tamoxifen, raloxifene, ethanol, or DMSO was included in the medium during the 30 min Ag uptake.

Raloxifene and tamoxifen inhibit DC differentiation in the presence of E2

Normal levels of E2 in mouse serum vary based upon sex, the stage of the menstrual cycle, and pregnancy. Male mice possess lower levels of E2 (8–15 pg/ml) than females in diestrus (25–35 pg/ml) and in estrus (100–200 pg/ml) (39, 40). We performed competition experiments between E2 and SERM to measure effects on DC development. Based upon the results from our previous study of the E2 dependency of BMDC, we used E2 at either 0.1 or 1.0 nM (27 and 272 pg/ml), which are at its K_D values for ER. These physiological amounts of E2 were efficient at supporting DC differentiation. We chose to use raloxifene from 1 to 200 nM and tamoxifen from 50 to 500 nM in our BM cultures because these concentrations are similar to the levels that have been detected in the plasma of humans taking SERM orally, 1.36 ng/ml (∼2.5 nM) for raloxifene and 120 ng/ml (∼200 nM) for tamoxifen (according to Physicians Desk Reference). The relative ER binding affinities, expressed as the percentage of E2 affinity for ER, have been estimated to be 34% for raloxifene (10) and 3% for tamoxifen (34) from competitive binding assays with ER⁺ MCF-7 cells.

Results from a characteristic 8-day culture experiment are shown in Fig. 2⇓. BM cultures were treated with vehicle only (a combination of ethanol, the vehicle for E2 and DMSO, the vehicle for SERM), E2 plus DMSO, or E2 plus variable doses of SERM. As before, it was found that in control cultures containing deficient medium not supplemented with E2 (Fig. 2⇓A), few DC developed (9% CD11b^high and 2% CD11b^int). However, addition of E2 alone (Fig. 2⇓B) promoted the differentiation of both CD11b^high (32%) and CD11b^int (30%) DC. Interestingly, tamoxifen and raloxifene inhibited the E2-mediated development of both of these DC populations, suggesting that these drugs competed with E2 for ER binding, and that excess ER-SERM complexes did not act similarly to ER-E2 complexes to promote DC differentiation. Although significant inhibition of CD11b^high DC was observed, the most remarkable decreases were seen in the CD11b^int population. 100 nM tamoxifen (Fig. 2⇓C) did not have an effect in competition with 1 nM E2 in this example, however, 500 nM tamoxifen (Fig. 2⇓D), 50 nM raloxifene (Fig. 2⇓E), and 100 nM raloxifene (Fig. 2⇓F) all induced a dramatic reduction in DC. For example, CD11b^high DC represented 32% of cultures given E2 alone; this was reduced to 24% by 500 nM tamoxifen (a 25% reduction) and 9–10% by 50–100 nM raloxifene (∼70% reduction). CD11b^int DC in the same cultures were reduced from 30% with E2 alone, down to 13% for 500 nM tamoxifen (∼56% reduction) and around 4–5% for 50–100 nM raloxifene (∼85% reduction). As shown in this experiment, a third population of CD11c⁺CD11b^low DC was repeatedly identified but was relatively unaffected by the presence or absence of E2. For this reason, CD11b^low DC have been excluded from most of the subsequent analyses of DC subpopulations.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Tamoxifen and raloxifene competed with estrogen and inhibited the differentiation of DC from BM precursors. BM cultures were grown for 8 days in DEF medium supplemented with 1 nM E2 and various concentrations of SERM. At the end of the culture period, cells were harvested and stained with mAbs specific for CD11b and CD11c for FACS analyses. A, In the vehicle control (V) of ethanol and DMSO, few DC developed, especially the CD11b^int DC. B, In the presence of 1 nM E2 plus DMSO, CD11b^high and CD11b^int DC development were both promoted. C, Addition of E2 plus 100 nM tamoxifen (T100) did not significantly affect DC development in this instance. D, Addition of E2 plus 500 nM tamoxifen (T500) reduced the frequency of both DC populations. Addition of E2 plus+ 50 nM raloxifene (R50) (E) or E2 plus 100 nM raloxifene (R100) (F) impaired DC development to a similar extent. Numbers indicate the percentage of total cells within the boxed region. The greatest effects were on the CD11b^int DC population. Data are representative of 11 independent experiments.

E2 often acts as a regulator of growth and differentiation. In general ovariectomy causes immune organs to enlarge whereas treatment with E2 results in significant shrinkage presumably due to the reported negative effects of E2 on lymphoid progenitors (41). Thus it was possible that ER antagonism by SERM might have led to an increase in numbers of undifferentiated lymphoid or myeloid progenitors in BM cultures. To confirm that the apparent decrease in the percentage of DC accurately reflects a real drop in DC numbers, we back calculated the total number of DC recovered from the cultures. Results from three independent experiments under various culture conditions are shown in Fig. 3⇓. The results from Fig. 2⇑ have been redisplayed in Fig. 3⇓A, showing that SERM treatment of BM cultures in DEF medium supplemented with 1 nM E2 does indeed reduce the actual numbers of DC. CD11b^high DC tended to be less dependent on E2 than CD11b^int DC.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

The total number of DC in culture was reduced by exposure to tamoxifen or raloxifene. BM cultures were grown under several different conditions to assess SERM effects on DC differentiation. DC were differentiated in the presence of variable doses of tamoxifen (T), 4OH, raloxifene (R), or the DMSO vehicle (V). A, Differentiation for 8 days in DEF medium plus 1 nM E2 on nontissue culture-treated plastic petri dishes. Ethanol (EtOH), in which E2 was solubilized, and DMSO were added as a double vehicle control (V) to assess E2-independent DC differentiation. B, Differentiation for 7 days in DEF medium plus 0.1 nM E2 on Teflon. C, Differentiation for 7 days in NORM medium on nontissue culture-treated plastic. At the end of the culture period, cells were harvested, surface stained with mAbs specific for CD11b and CD11c, and analyzed by FACS. Total DC numbers were derived from cell counts and FACS plot data. The total number of cells recovered from one dish was very different among the three experiments depicted and is probably reflective of variations in culture conditions such as duration, type of plate used, and E2 concentration.

Only the nonadherent and loosely adherent cell fractions were analyzed in our initial studies (Fig. 3⇑A) using petri dishes to which some DC adhered strongly, and previous experiments indicated that they were representative of the entire culture. To alleviate a potential sampling error, we began to use Teflon cultureware, which has been reported to result in significantly less cell adherence. A representative 7-day BM culture in Teflon with 0.1 nM E2 added to DEF medium is shown in Fig. 3⇑B. These experiments demonstrated that tamoxifen and raloxifene had significant inhibitory effects under these conditions as well, and the increased sensitivity of the CD11b^int population to SERM was still observed.

We also investigated the effects of a well-studied metabolite of tamoxifen, 4OH in NORM medium, in which all the E2 present is derived from FBS. Upon direct comparison of the effects of 4OH to tamoxifen (Fig. 3⇑C), we found 1 nM 4OH to be nearly as effective as 500 nM tamoxifen at inhibiting CD11b^int DC development in NORM medium. The 1 and 10 nM 4OH and 50 nM tamoxifen slightly increased the number of CD11b^high DC relative to the control but this effect was diminished when cultures were treated with 500 nM tamoxifen. The opposing effects of 50 and 500 nM tamoxifen on CD11b^high DC in this system were an unexpected but reproducible finding for which we have no explanation. It is interesting to note that overall, there was less inhibition of DC differentiation by SERM in NORM medium cultures compared with DEF medium cultures that contained charcoal dextran stripped FBS, to which only E2 was added back. This implies that, in addition to E2, there are important factors influencing DC differentiation, which are removed by the charcoal dextran treatment process. Under all culture conditions, we found that the development of CD11b^int DC was the most severely inhibited by SERM.

Raloxifene and tamoxifen augment Ag uptake by DC

Ag uptake capacity and surface expression of maturation markers are two parameters by which to gauge the maturation status of DC. Immature DC are efficient at Ag uptake and this function is reduced after maturation. We observed that immature DC from tamoxifen- and raloxifene-treated cultures had lower levels of MHC class II and CD86 compared with untreated DC suggesting that they are more immature (see discussion below and Fig. 6⇓A). Therefore, we reasoned that a greater percentage of the DC that develop in the presence of SERM, relative to the control, might exhibit the ability to internalize high levels of soluble Ags. To test this, DC differentiated for 7 days in the presence of 0.1 nM E2, with or without one of the two SERM, were allowed to internalize OVA₄₈₈ or DexFITC for 30 min at 37°C. Ag incubations also were done at 4°C to allow for the distinction between signals from surface binding and internalization because uptake is arrested at low temperatures.

As expected, both raloxifene and tamoxifen augmented the uptake of OVA₄₈₈ and DexFITC by BMDC, defined in this study as all CD11c⁺ cells (Fig. 4⇓). There was a difference in Ag uptake capacity observed between the CD11b^high and CD11b^int cells; this is demonstrated in Fig. 5⇓ from an independent repeat of the experiment. Immature CD11b^high DC were found to be the most efficient at internalizing Ags in all treatments. This finding is consistent with the observation that CD11b^high cells are relatively less mature based upon their overall lower expression of MHC class II and CD86 (31). However, the effect of both SERM to increase Ag uptake was most apparent in the CD11b^int DC. The results were similar whether OVA₄₈₈ or DexFITC (data not shown) were used for uptake. These data indicate that SERM treatment of BM cultures during DC differentiation leads to DC that are functionally more immature.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Tamoxifen or raloxifene treatment increased the percentage of DC capable of Ag uptake. BM cells were cultured for 7 days in DEF medium with 0.1 nM E2 and either DMSO, 500 nM tamoxifen (Tam), or 100 nM raloxifene (Ral). Cells were harvested, washed, and incubated with 50 μg/ml Alexa 488-labeled OVA or 100 μg/ml FITC-dextran for 30 min at 37°C. Ag uptake was allowed to occur in the absence of DMSO and SERM. The cells were then washed and surface stained for CD11b and CD11c. Internalization of labeled Ags by all DC (CD11c⁺ cells) is shown. The numbers above the marker in each plot indicate the percentage of DC that internalized Ag. Tamoxifen and raloxifene increased the percentage of DC that internalized OVA (left panels) and Dex (right panels) relative to the DMSO control. The negative control for each experiment consisted of DMSO-treated DC given Ag and incubated for 30 min at 4°C. Data are representative of three independent experiments.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

SERM augmentation of OVA uptake was greatest in CD11b^int DC. DC were derived from BM cultures and were treated as described in Fig. 4⇑. Histograms show OVA uptake by either CD11b^high (top row) or CD11b^int (bottom row) DC. The percentage above the marker in each plot indicates the DC population that internalized Ag. Overall, raloxifene (Ral) and tamoxifen (Tam) increased OVA uptake relative to the control (DMSO). CD11b^high DC were the most efficient at OVA uptake and the percentage of the population that internalized Ag was elevated in Tam and Ral cultures. CD11b^int DC were reduced in number in Tam and Ral cultures but the percentage of the population that internalized OVA increased more dramatically relative to CD11b^high cells. Data are representative of three independent experiments.

Raloxifene and tamoxifen render DC hyporesponsive to LPS maturation

To determine whether the effects of tamoxifen and raloxifene on DC influence the final phase of differentiation, activation by inflammatory stimuli, we studied the effects of SERM exposure on expression of key surface markers important for DC to function as APC. As before, BM cultures were supplemented with 1 nM E2 and given vehicle or SERM. Relative to control immature DC, treatment of BM cultures with tamoxifen (500 nM) or raloxifene (100 nM) for 8 days led to reduced levels of MHC class II and CD86 on immature DC (Fig. 6⇓A). We next tested whether these two SERM might prevent DC acquisition of a mature phenotype following activation. Therefore, we induced DC maturation with the bacterial cell wall component, LPS, and examined these cells for expression of MHC class II and CD86. Nearly 90% of control DC expressed high levels of MHC class II and CD86 after LPS exposure indicating that they had acquired a mature phenotype (Fig. 6⇓B). When DC were treated with 500 nM tamoxifen, only 60% of the population expressed high levels of MHC class II and CD86 (Fig. 6⇓B). Treatment with 100 nM tamoxifen did not significantly impair the expression of MHC class II and CD86 on immature or activated DC (data not shown). In the case of 100 nM raloxifene, only 42% of DC exhibited the mature phenotype (Fig. 6⇓B). Hence, tamoxifen and raloxifene rendered DC hyporesponsive to LPS induced maturation. Effects were seen on both CD11b^high and CD11b^int DC, and as was observed during DC differentiation, phenotypic effects were often greater on the CD11b^int DC population (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Tamoxifen and raloxifene reduced the expression of MHC class II and CD86 on immature DC and decreased the percentage of mature DC that expressed elevated levels of each marker. BM cultures were grown for 8 days in DEF medium with 1 nM E2 plus either DMSO, 500 nM tamoxifen (Tam), or 100 nM raloxifene (Ral). For each treatment, parallel cultures of DC were either left unstimulated (immature) or LPS activated (mature) before harvest and FACS analyses. Histograms show MHC class II or CD86 expression on all DC (CD11c⁺ cells). A, Tam and Ral both reduced the expression of CD86 (top) and MHC class II (bottom) on immature DC. B, LPS maturation resulted in MHC class II and CD86 up-regulation on the majority of DMSO-treated DC (upper panels). The percentage at each marker represents DC that fall within the marker range. Treatment with Tam (middle panels) and Ral (lower panels) reduced the percentage of DC that up-regulated MHC class II and CD86 in response to LPS. CD11b^int DC were most dramatically affected by SERM (data not shown). These data are representative of 13 independent experiments.

We then tested titrated amounts of raloxifene, from 1 to 100 nM, against 0.1 nM E2 for two reasons: 1) to demonstrate that the observed effects were dose-dependent; and 2) to determine the inhibitory activity of doses similar to the low concentrations that have been measured in the plasma of patients undergoing raloxifene therapy (∼2.5 nM). We found that low doses of raloxifene were able to inhibit DC differentiation and maturation in response to LPS. Weak effects of raloxifene on CD40 and CD80 were observed at 10 and 100 nM although to a lesser degree than that observed for MHC class II and CD86 (Fig. 7⇓A); only the results with the 100 nM concentration are depicted. Raloxifene significantly inhibited the expression of MHC class II and CD86 by LPS-activated DC at concentrations as low as 1 nM (Fig. 7⇓B). In the control, 77% of DC expressed elevated CD86. This was decreased to 56, 31, and 35% of the population by 1, 10, and 100 nM raloxifene, respectively. Similarly, MHC class II was found to be elevated in 74% of mature DC in the control and this was reduced to 60% by 1 nM, 41% by 10 nM, and 45% by 100 nM raloxifene. In Fig. 7⇓, the total percentage of DC that developed in cultures was reduced from a maximum of 74% in the control to 70, 59, and 56% by 1, 10, and 100 nM raloxifene, respectively (data not shown). These results demonstrate that low doses of raloxifene (10 nM) maximally impaired the ability of DC to mature in response to LPS.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Low concentrations of raloxifene inhibited the LPS-induced maturation of DC. BM cells were cultured for 8 days in DEF medium supplemented with 0.1 nM E2 and either DMSO or raloxifene (1, 10, or 100 nM). Cells were matured with LPS, harvested, surface stained with combinations of mAbs against lineage and activation markers, and analyzed by FACS. Histograms show activation marker expression on total DC (CD11c⁺ cells). Raloxifene (Ral) treatment inhibited the expression of cell activation markers that are normally up-regulated after LPS exposure. A, Weak reductive effects were observed on CD40 and CD80 with 100 nM Ral. B, Ral significantly reduced the percentage of DC that expressed high levels of CD86 and MHC class II in a dose dependent manner. The percentage DC within the marked region is indicated. Maximum effects were seen at the 10 nM dose. Data are representative of two independent experiments.

TLR4 expression is not reduced by raloxifene or tamoxifen

DC express surface receptors including TLR family members that allow them to sense pathogens by recognizing molecules that contain pathogen associated molecular patterns. TLR4 is involved in the detection of bacterial LPS (42). We hypothesized that SERM-induced hyporesponsiveness to LPS is related to effects on TLR4 expression levels. Interestingly, there were differences in the level of TLR4 expression between the CD11b^high and CD11b^int populations on control DC (Fig. 8⇓A); CD11b^high DC exhibited overall higher levels of TLR4 expression relative to CD11b^int DC. There were slight differences in TLR4 expression following 100 nM raloxifene or 500 nM tamoxifen treatment. Exposure of DC to SERM did not significantly alter the levels of TLR4 on CD11b^high DC (Fig. 8⇓B). In the case of CD11b^int DC, it was sometimes observed that SERM actually increased TLR4 expression to nearly the level seen on CD11b^high DC (Fig. 8⇓B). CD11b^int DC were the most susceptible to influence by ER ligands, and if LPS hyporesponsiveness were due to reduced or absent TLR4 expression, we would expect to see lower levels on these cells. Our data indicate that raloxifene and tamoxifen do not impair the LPS induced maturation of DC through modulation of TLR4 expression levels.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

TLR4 expression was not impaired by tamoxifen or raloxifene. DC were prepared as described in Fig. 4⇑ and surface stained for CD11b, CD11c, and TLR4 for FACS analysis. Histograms show TLR4 expression on CD11b^high or CD11b^int DC. Binding of the isotype control mAb (shaded histogram) is indicated. A, The highest levels of TLR4 were seen on the CD11b^high DC subset in the DMSO-treated cultures. B, Treatment of BM cultures with 100 nM raloxifene (Ral) or 500 nM tamoxifen (Tam) did not reduce TLR4 expression on CD11b^high or CD11b^int DC. In some experiments, SERM treatment slightly increased TLR4 expression on CD11b^int DC. The data are representative of four independent experiments.