Heterogeneous expression of apolipoprotein-E by human macrophages

Abstract

Apolipoprotein-E (apoE) is expressed at high levels by macrophages. In addition to its role in lipid transport, macrophage-derived apoE plays an important role in immunoregulation. Previous studies have identified macrophage subpopulations that differ substantially in their ability to synthesize specific cytokines and enzymes, however, potential heterogeneous macrophage apoE expression has not been studied. Here we examined apoE expression in human THP-1 macrophages and monocyte-derived macrophages (MDM). Using immunocytochemistry and flow cytometry methods we reveal a striking heterogeneity in macrophage apoE expression in both cell types. In phorbol-ester-differentiated THP-1 macrophages, 5% of the cells over-expressed apoE at levels more than 50-fold higher than the rest of the population. ApoE over-expressing THP-1 macrophages contained condensed/fragmented nuclei and increased levels of activated caspase-3 indicating induction of apoptosis. In MDM, 3–5% of the cells also highly over-expressed apoE, up to 50-fold higher than the rest of the population; however, this was not associated with obvious nuclear alterations. The apoE over-expressing MDM were larger, more granular, and more autofluorescent than the majority of cells and they contained numerous vesicle-like structures that appeared to be coated by apoE. Flow cytometry experiments indicated that the apoE over-expressing subpopulation of MDM were positive for CD14, CD11b/Mac-1 and CD68. These observations suggest that specific macrophage subpopulations may be important for apoE-mediated immunoregulation and clearly indicate that subpopulation heterogeneity should be taken into account when investigating macrophage apoE expression.

Introduction

Apolipoprotein-E (apoE) is a ∼34 000 MW glycoprotein that plays an important role in lipoprotein metabolism, cellular lipid transport and immunoregulation.¹ ApoE is expressed at high levels in monocyte-derived macrophages (MDM), Kupfer cells and microglia.¹ Earlier studies established that apoE is a major secreted product of macrophages, accounting for 10–25% of total secreted protein.^2,3 An important immunoregulatory function of the macrophage is proposed to be via the apoE-mediated suppression of T-cell activation and proliferation.^4–7 Many studies have confirmed that apoE suppresses mitogen-stimulated lymphocyte proliferation (see ⁸ and references cited therein) and studies in apoE gene knockout (apoE^−/−) mice have extended these observations by providing evidence that apoE plays a functionally important role in modulating the immune response in vivo. Such studies showed that apoE^−/− mice displayed: higher antigen-specific IgM after immunization with tetanus toxoid; increased lymphocyte proliferation and lesion development in experimental autoimmune encephalomyelitis; and impaired delayed type hypersensitivity responses.^8–10 In addition, the interferon-γ-induced expression of major histocompatibility complex (MHC) class II molecules I-A^b and the costimulatory surface proteins CD40 and CD80 was significantly increased on macrophages derived from apoE^−/− mice compared to wild type control animals.⁸ These latter studies indicate that macrophage-derived apoE not only suppresses immune activation via the inhibition of lymphocyte activation and proliferation but also through autocrine inhibition of MHC and costimulatory protein expression on antigen-presenting cells.

An important focus for macrophage apoE research has been its potential role in the prevention of atherosclerosis. In this context, apoE promotes cholesterol efflux from macrophage foam cells present in atherosclerotic lesions and likely suppresses chronic T-cell mediated inflammation.¹¹ While the immunoregulatory mechanisms underlying the antiatherosclerotic functions of macrophage apoE remain to be defined, bone marrow transplant studies in atherosclerosis-prone apoE^−/− mice^12,13 indicate that macrophage-specific apoE production inhibits atherosclerosis.^14,15 The factors that regulate macrophage apoE expression and immunoregulatory function thus remain the subject of continuing in vivo and in vitro studies by many investigators.

Monocyte to macrophage differentiation,¹⁶ cholesterol loading,² specific glycosphingolipids and fatty acids,^17,18 apolipoprotein-AI,¹⁹ and several growth factors and cytokines regulate apoE expression via mechanisms that range from transcriptional through to post-translational levels.^20–25 It is becoming increasingly clear that the transcriptional regulation of specific inducible macrophage genes is subject to a remarkable degree of heterogeneity.^26–29 For example, using cDNA microarrays and the RAW264 macrophage cell line, it was shown that lipopolysaccharide-inducible genes are not transcribed to the same extent in all cells.²⁶ Earlier work showed that the secretion of lysozyme and transforming growth factor-β (TGF-β) by human macrophages and of interleukin-1 (IL-1) by monocytes is also predominantly caused by subpopulations of cells.^28,29 The possibility that macrophage apoE expression is heterogeneous has not been addressed. In the present work we examined the expression of apoE in human THP-1 cells and MDM and reveal a striking heterogeneity in both the amount and subcellular distribution patterns of apoE in these cell types.

Materials and methods

Cell culture

Human THP-1 monocytes were cultured in RPMI-1640 medium containing 10% (v/v) fetal calf serum (FCS) supplemented with 2 mm l-glutamine and 100 U/ml penicillin and 100 mg/ml streptomycin at 37° in 5% CO₂. Differentiation to a macrophage phenotype was induced by culturing 2 × 10⁶ cells on sterile glass cover slips placed in 22-mm diameter dishes in the presence of 50 ng/ml phorbol 12-myristate 13-acetate (PMA) for 3 days. Human monocytes were isolated from donor buffy coats using Ficoll-Paque³⁰ as described previously.³¹ Monocytes were allowed to differentiate into macrophages by culture for up to 2 weeks in RPMI-1640 medium containing 10% (v/v) human serum supplemented with l-glutamine, penicillin and streptomycin as above. Cell viability was assessed by trypan blue exclusion. Where indicated, cells were rinsed in phosphate-buffered saline (PBS) and incubated for up to 16 hr in the presence of 0·1 µm staurosporine (Sigma, St Louis, MO) in serum-free RPMI-1640 medium. Staurosporine was stored as a 0·2 mm stock in ethanol and diluted in medium immediately prior to addition to cells.

Immunocytochemistry

Macrophages were fixed in 4% paraformaldehyde in PBS for 20 min at 4°, rinsed in PBS, and permeabilized in 0·1% saponin and 5% fetal bovine serum in PBS (0·1% SB) for another 20 min at 22°. The cells were then treated with 1 : 500 diluted murine anti-human apoE monoclonal antibody (Biogenesis, Poole, UK) in a humidifier at 4° overnight. After rinsing in 0·1% SB, anti-mouse immunoglobulin G (IgG) fluorescein isothiocyanate (FITC) conjugate (1 : 100; Calbiochem, Kilsyth, Australia) was added for 1 hr at 22°. The samples were rinsed again in 0·1% SB, PBS and distilled H₂O and mounted in Vectashield containing 4,6-diamidino-2-phenylindole (DAPI). Specimens were studied with a Microphot-SA fluorescence microscope (Nikon, Tokyo, Japan) using a blue excitation filter (450–490 nm) and a 520 nm barrier filter and photographed using a digital video camera (Hamamatsu, Tokyo, Japan). Cells incubated either in the absence of anti-human apoE monoclonal antibodies or in the presence of non-immune murine IgG did not result in significant staining. THP-1 macrophages were also studied by dual labelling as described above with the addition of 1 : 100 diluted rabbit anti-human active caspase-3 polyclonal antibody (Pharmingen, San Diego, CA). Highly cross-adsorbed anti-rabbit IgG AlexaFluor 633 (1 : 100, Molecular Probes, Eugene, OR) was used for detection of the anti active caspase-3. In these experiments a Leica TCS FP laser scanning confocal microscope was used in combination with an argon laser (488 nm blue light and 633 nm red light) and a 20× objective.

Human MDM were also examined under an Olympus LSM-GB 200 laser scanning confocal microscope using an argon laser (488 nm blue light and 543 nm green light) and a 60× oil objective. Cells were immunostained as described above using anti-apoE monoclonal antibody. Cells were then exposed to propidium iodide (5 µg/ml) for 5 min during the final wash in PBS and mounted in Vectashield without DAPI. The specificity of the immunostaining method for apoE was previously demonstrated.³²

Macrophage apoE immunostaining fluorescence intensity was semiquantitatively determined by analysing digital monochrome images that were originally collected using Adobe Photoshop 5 software and quantified using the public domain National Institutes of Health Image program (http://rsb.info.nih.gov/nih-image/). Individual cells were analysed and the amount of fluorescence per cell was calculated as the product of average pixel value by cell area as described previously.³³

Caspase-3 activation

Caspase-3-like activity was assessed in aliquots of cell lysates (containing approximately 12 µg protein) after addition of 20 mm Ac-DEVD-7-amino-4-methylcoumarin and measurement of the liberated fluorophore 7-amino-4-methylcoumarin (AMC) at Ex 380; Em 435 nm according to the manufacturer's instructions (Pharmingen). Caspase-3 activity has been previously reported in PMA-treated THP-1 cells and human MDM in the absence of generalized apoptosis.^34,35

Western blotting

Macrophage apoE was detected by Western blot as previously described.²² Cell protein was measured using the bicinchoninic acid method and equal amounts of protein were mixed with Laemmli sample buffer and run on 12% polyacrylamide–sodium dodecyl sulphate (SDS) gels using a Mini-Protean II system (Bio-Rad, Hercules, CA) and subsequently transferred to nitrocellulose membranes (Bio-Rad). Loading equivalence and transfer efficiency were monitored by Ponceau S staining. The membranes were then incubated for 16 hr at 4° with a rabbit anti-human apoE polyclonal antibody (Cat. No. A0077, DAKO, Glostrup, Denmark) diluted 1/1000, followed by a 1-hr incubation with horseradish peroxidase-conjugated goat anti-rabbit IgG antibody (Cat. No. P0448, DAKO). Blots were developed by using enhanced chemiluminescence (Amersham Pharmacia, Castle Hill, Australia), and the membranes were exposed to X-ray film (Fuji), developed, scanned and signal intensity quantified using NIH Image 1.62 software.²²

Flow-cytometric analysis

Human MDM were cultured for 2 weeks in 75 cm² flasks, rinsed several times with PBS (60 ml in total) and incubated in serum-free RPMI-1640 for 4 hr to minimize contamination of the MDM with human serum-derived apoE. The cells were then prefixed with 4% paraformaldehyde in PBS and permeabilized using 0·5% (w/v) saponin, 0·1% BSA in PBS.³⁶ The MDMs were then gently scraped from the flasks and transferred to plastic tubes for immunolabelling using anti-human apoE monoclonal antibody 1/500 dilution (7·2 µg/ml) and FITC-conjugated F(ab′)₂ goat anti-mouse IgG (F(ab′)₂-specific, 10 µg/ml) 4° in the dark as described previously.³⁶ MDM were similarly analysed for CD68 expression using anti-human CD68 monoclonal antibody 1/100 dilution (10 µg/ml, DAKO) with the same FITC-conjugated anti-mouse IgG. Expression of CD14 and CD11b/Mac-1 was analysed using FITC- and phycoerythrin (PE)-conjugated monoclonal antibodies, respectively (Pharmingen). MDM were also incubated with the appropriate IgG isotype non-immune control FITC- and PE-conjugated antibodies (Pharmingen, San Diego, CA and Serotech, Raleigh, NC). After immunolabelling, MDM were washed twice with PBS, fixed with 1% paraformaldehyde in PBS, and analysed by using a FACScan flow cytometer (Becton Dickinson, San Jose, CA). MDM prepared using this method were also examined by Normarsky interference microscopy and found to be structurally intact and free of cellular aggregates. Attempts to remove MDM from flasks without fixation (e.g. by scraping into ice-cold PBS) led to alterations in cell morphology and a significant loss of cell viability as assessed by uptake of trypan blue stain. Macrophages are well known to exhibit autofluorescence, which can be quenched by crystal violet in order to reduce background signal in immunostaining/flow cytometry analysis.^37,38 We did not quench macrophage autofluorescence in the present work as it can also be used to discriminate between human MDM subpopulations.³⁹ In the present work, we analysed unlabelled MDM for autofluorescence, then adjusted the flow cytometer voltage to remove the autofluorescence signal while permitting optimal detection of fluorophore signal.

Results

Analysis of THP-1 macrophage apoE expression by immunocytochemistry

The expression of apoE in phorbol-ester treated THP-1 cells is well known.^40–42 We wanted to know if heterogeneity existed between individual cells with respect to PMA-induced apoE expression. Using immunocytochemical microscopy, we found that essentially all THP-1 cells were positive for apoE (Fig. 1a). Most cells stained with a perinuclear, presumably Golgi, pattern as would be predicted for a secretory protein. Even among cells with a very similar morphology, there was some degree of heterogeneity in apoE expression (Fig. 1a). We also detected a very dramatic over-expression of apoE in a minor proportion of cells. Semiquantitative immunofluorescence analysis indicated that these cells expressed up to 50-fold higher levels of apoE than the average level. The THP-1 cells over-expressing apoE displayed either one or more of several specific morphological traits. First, all cells that displayed micronuclei (small chromatin-containing structures within a nuclear envelope that are separated from the principal nucleus) over-expressed apoE (Fig. 1a, arrowed). Cells that contained condensed/fragmented nuclei, that are classic markers of apoptosis,⁴³ also greatly over-expressed apoE in all cases (Fig. 1c, d). Obvious nuclear alterations and apoE over-expression was detected in approximately 5–10% of PMA-treated THP-1 cells. Multinucleated giant cells and binucleated cells were also present and these cells also over-expressed apoE (Fig. 1b, e). Note that the multinucleated cell shown in Fig. 1(b) also contains several micronuclei.

Figure 1.

Analysis of apoE expression in THP-1 macrophages. THP-1 cells were treated with PMA 50 ng/ml for 72 hr then assessed for apoE expression by immunocytochemical staining and nuclear morphology by DAPI staining. (a–f) ApoE expression. The nuclear morphology for the series is shown directly below each panel. (a) Typical apoE expression in most cells is compared to high expression in the arrowed cell which contains distinct micronuclei in addition to the major nucleus. (b) ApoE expression is high in multinucleated giant cells. The arrowed cell contains six intact nuclei in addition to several clustered micronuclei. (c) In the arrowed cell, apoE is highly over-expressed and the nucleus shows a typical apoptotic morphology as it is condensed and fragmented. (d) Higher magnification of apoE over-expressing cell with condensed/fragmented nucleus. The apoE appears to encapsulate the fragmented nuclei. (e) Binuclear cell over-expressing apoE. (f) Staurosporine treated THP-1 macrophages displaying cytoplasmic shrinkage and relocalization of intracellular apoE to the plasma membrane. Note that in (f), the nuclear morphology panel is overlayed with a phase contrast image in order to illustrate plasma membrane alterations. Bar = 25 µm: (a, b, c and f) same magnification; (d and e) higher magnification all croped and resized.

Previous studies indicate that PMA can induce apoptosis in THP-1 cells. For example, it was shown that 5% of PMA-treated THP-1 cells were TdT-mediated biotin–dUTP nick-end labelling-positive⁴⁴ and that 7% of PMA-treated THP-1 cells contained sub-G1 phase DNA after propidium iodide staining and flow cytometric analysis.⁴⁵ The association of high levels of apoE in macrophages with fragmented nuclei may be related to the induction of apoptosis in these cells. To gain further evidence that apoE over-expression is associated with apoptosis in PMA-treated THP-1 cells, we conducted dual labelling experiments for potential colocalization with active caspase-3, an important protease in the execution phase of apoptosis. Confocal microscopy images indicated that cells that over-expressed apoE were also positive for active caspase-3 (Fig. 2).

Figure 2.

ApoE over-expression is colocalized with active caspase-3 in THP-1 macrophages. THP-1 cells were treated with PMA 50 ng/ml for 72 hr then assessed for apoE and active caspase-3 expression by immunocytochemical staining and laser scanning confocal microscopy. (a) ApoE expression detected with an FITC-conjugated secondary antibody. (b) Active caspase-3 expression detected with an Alexa Fluor 633-conjugated secondary antibody. In the arrowed cell, both apoE and active caspase-3 are over-expressed. Bar = 50 µm

Staurosporine (a PKC inhibitor), induces apoptosis in many cell types, including murine RAW264 macrophages,^46,47 but not human MDM.⁴⁸ We used staurosporine to further investigate the potential relationship between macrophage (apoptotic) morphology and apoE expression. Staurosporine induced a dramatic alteration in THP-1 morphology including cytoplasmic shrinkage. This was accompanied by a generalized redistribution of the perinuclear apoE pools to the plasma membrane (Fig. 1f). Interestingly, while caspase-3 activity was present in PMA treated cells and increased moderately with staurosporine treatment (as compared to human foreskin fibroblasts for example, Fig. 3a), the formation of fragmented nuclei was not as evident (Fig. 1f) as might be predicted based on caspase-3 activation in other cell types.^49,50 Staurosporine treatment also moderately increased total apoE levels in THP-1 cells as assessed by Western blotting (Fig. 3b).

Figure 3.

ApoE expression and caspase-3 activation in staurosporine-treated THP-1 cells. (a) THP-1 macrophages and MDM were treated with 0·1 µm staurosporine for the times indicated and caspase-3 activation was assessed by hydrolysis of the caspase-3 substrate Ac-DEVD-AMC to yield the fluorophore AMC in cell lysates (containing 12 µg protein per sample). Both THP-1 macrophages (○) and MDM (•) contained caspase-3 activity and this was either moderately increased (THP-1) or unchanged (MDM) with staurosporine treatment. In contrast, fibroblasts (□) contained very low caspase-3 activity initially but dramatically increased their activity after staurosporine treatment. The data for human foreskin fibroblasts are based on our previously published data generated using an identical experimental protocol and are included here for the purposes of comparison only.³² Data are means ± SE, n = 3. (b) THP-1 macrophages were treated with 0·1 µm staurosporine for 8 h and cell lysates (34 µg protein per sample) were analysed for apoE by Western blot. Recombinant human apoE3 was included as a positive control and the positions of molecular weight markers are indicated. Analysis of integrated optical densities of the apoE bands shown indicated that staurosporine (Stsp) treatment increased apoE levels by 64%.

These data show for the first time that THP-1 macrophages exhibit a remarkable degree of heterogeneity in their expression of apoE. Furthermore, this heterogeneity is clearly associated with nuclear morphological alterations. Such nuclear alterations are to be expected in cancer cell lines and are associated with the genetic instability of tumour cells in general.⁵¹ We therefore also assessed the expression of apoE in human MDM which would not be predicted to exhibit such nuclear alterations or genetic instability.

Analysis of human MDM apoE expression by immunocytochemistry

Immunocytochemical analysis of human MDM revealed that essentially all cells expressed apoE. As was observed in THP-1 macrophages, apoE immunostaining was generally in a perinuclear, presumably Golgi, compartment (Fig. 4). A partial localization of apoE at the plasma membrane was detectable in most cells (Fig. 4f), and confocal microscopy studies revealed a predominant plasma membrane apoE pool was relatively rare (see below). Approximately 5% of MDM (see also quantitative flow cytometry data below) dramatically over-expressed apoE (Fig. 4). In contrast to THP-1 cells, the MDM that over-expressed apoE did not display fragmented nuclei. Nuclear morphology was essentially normal with the exception that the apoE over-expressing MDM were often binuclear and/or exhibited a large and well defined nucleolus (Fig. 4g). The most conspicuous defining morphological feature of MDM that over-expressed apoE was their larger than average size. This size difference was clear in apoE over-expressing MDM after either 1 or 2 weeks of culture in vitro (Fig. 4). ApoE appeared in the over-expressing MDM as densely packed ring-like structures that were presumably associated with cytoplasmic vesicles (Fig. 4i). Semi-quantitative immunofluorescence analysis indicated that these larger MDM contained up to 50 times more apoE than the average of the remaining cells. Treatment of MDM with staurosporine for up to 16 h did not induce a typical apoptotic response; however, apoE subcellular distribution was altered with a loss of perinuclear staining and the formation of numerous punctate structures in the cytosol (Fig. 4l). Caspase-3 activity was present in MDM but was not increased with staurosporine treatment (Fig. 3a).

Figure 4.

Analysis of apoE expression in human MDM. MDM were cultured for either 7 or 14 days and assessed for apoE expression by immunocytochemical staining and nuclear morphology by DAPI staining. (a, d, g and j) Nuclear morphology. The corresponding apoE expression for each panel is shown directly to the right (b, e, h and k) and the arrowed cells are also shown at higher magnification (c, f, i and l). (a, b, c) Typical 7-day-old MDM are compared to an apoE over-expressing (arrowed) MDM. (d, e, f) Fourteen-day-old MDM. Clear Golgi and plasma membrane staining are demonstrated in the arrowed cell. (g, h, i) An apoE over-expressing 14-day-old MDM is compared with typical neighbouring cells. At higher magnification and using shorter exposure time, apoE is detected as numerous ring-like structures, which are presumably coating vesicles. (j, k, l) Treatment of 14-day-old MDM with 0·1 µm staurosporine for 16 hr resulted in disruption of the typical staining pattern (Golgi and plasma membrane) and the formation of numerous large apoE-containing intracellular deposits (l). Bar = 25 µm. Note: (c, f, i and l) at higher magnification.

Previous biochemical studies have characterized a cell surface apoE pool that accounts for approximately 8% of MDM apoE.⁵² In order to more accurately examine potential heterogeneity with respect to the proportional distribution of MDM apoE at the cell surface, we utilized laser scanning confocal microscopy. In the vast majority of cells, the distribution of MDM apoE was predominantly intracellular (data not shown), confirming previous data.⁵² However, in a small proportion of MDM that over-express apoE, the cell surface pool accounted for almost all of the cellular apoE (Fig. 5). We estimate that MDM with this predominant cell surface apoE localization account for <0·5% of the entire population.

Figure 5.

Demonstration of predominant MDM plasma membrane apoE immunostaining. MDM were grown under standard culture conditions for 2 weeks, immunostained using an anti-apoE monoclonal antibody and examined under an Olympus LSM-GB 200 laser scanning confocal microscope. Cells were treated with propidium iodide after immunostaining and mounted in Vectashield medium. Optical serial sections of 2 µm are shown from the basal to apical cell surface (a–f).

Analysis of human MDM apoE expression by flow cytometry

In order to obtain additional quantitative data related to the heterogeneous expression of apoE in MDM, we used an immunostaining/flow cytometry approach. Analysis of 2-week-old MDM by forward and right-angle light scatter revealed a minor subpopulation (fraction R2) that was distinct from the major (R1) population (Fig. 6a). In agreement with previous data^37,39 autofluorescence was clearly detected at different levels throughout the MDM populations (Fig. 6b). The larger and more granular R2 population displayed increased levels of autofluorescense as compared to the majority of the MDM. When apoE immunostaining was measured, three obvious levels of staining intensity (M1, M2 and M3) were identified (Fig. 6c). The majority of cells were moderately positive (71·5%, M1) for apoE and a second population of cells contained increased apoE content (25·3%, M2). The M1 and M2 populations clearly overlapped and together represented 96·8% of the MDM present. A third, minor population of MDM (3·2%, M3) displayed approximately 50–100-fold higher levels of apoE immunostaining as compared to the M1/M2 populations (Fig. 6c). Incubation of MDM with an isotype control non-immune IgG did not result in significant staining (Fig. 6d), thereby ruling out potential non-specific effects caused by IgG binding to MDM cell surface receptors. When apoE immunostaining was assessed in MDM gated for R2, we found that 94% of the M3 MDM was present in the R2 subpopulation. Similarly, forward and right-angle light scatter analysis of the M3 population revealed an obvious clustering of these cells in the larger, more granular R2 subpopulation (Fig. 6e), while the M2 population were clustered in the R1 fraction (Fig. 6f). The increased apoE expression in the R2 subpopulation was not caused by increased non-specific IgG binding (Fig. 6g). To further characterize the R2 subpopulation, we analysed expression of CD11b/Mac-1 and CD14. Essentially all cells in the R2 subpopulation were positive for both CD antigens (Fig. 6h, i, respectively). The R2 subpopulation were also positive for CD68 (data not shown).

Figure 6.

Flow cytometric analysis of apoE in 14-day-old MDM. MDM were cultured for 14 days and analysed for size, granularity and for expression of apoE, CD14 and CD11b/Mac-1 by immunocytochemical staining and flow cytometric analysis. (a) Forward and side scatter analysis revealed a subpopulation of large granular MDM ‘R2’. (b) Analysis of MDM autofluorescence. (c) Three obvious levels of apoE immunostaining were present with a minor fraction ‘M3’ highly over-expressing apoE. (d) IgG control for apoE immunostaining did not reveal significant staining. (e) The M3 MDM population over-expressing apoE were clustered within the larger granular R2 MDM subpopulation. (f) The M2 MDM coincided with the major R1 population. (g, h and i) The R2 subpopulation were positive for apoE, CD11b and CD14, respectively. Fluorescence resulting from the relevant IgG isotype control conditions are also shown. Except for cell counts, data are log scale.

The flow cytometric analysis of MDM confirm our immunocytochemical microscopy data and show for the first time that there is marked heterogeneity in the expression of apoE by human macrophages.

Discussion

The aim of the present work was to investigate the potential heterogeneous expression of apoE in human macrophages. In addition to a moderate degree of heterogeneity (several-fold differences) in the bulk of the macrophage populations (e.g. as shown in M1 and M2 MDM populations, Fig. 6c), our data clearly show that both MDM and THP-1 macrophages contain subpopulations of cells that highly over-express apoE. The identification of this heterogeneity is important for several reasons. Previous studies that have assessed the regulation of both constitutive and inducible macrophage apoE expression in vitro have analysed the entire macrophage population without considering specific subpopulations that may contribute more apoE. Using our flow cytometry data, we calculate that the apoE over-expressing MDM population, that accounts for approximately 3% of the total population, may contribute as much apoE to the total pool than the remaining 97% (where we estimate that on average, the over-expressing cells generate 50-fold more apoE than the remainder). CD68 expression confirmed the macrophage phenotype of the R2 subpopulation as this is a well known macrophage marker. Previous studies in vitro and in vivo have shown that the expression of CD11b and CD14 is not universal for all MDM and that subpopulations of cells can be identified that express these markers to varying degrees.^53,54 Our data indicated that MDM in the R2 subpopulation were positive for CD11b/Mac-1 and CD14 (Fig. 6h, i). Because CD14 is not expressed on monocyte-derived dendritic cells it is unlikely that R2 is a dendritic cell subpopulation.⁵⁵

In studies that are analogous to our finding that macrophage apoE expression is heterogeneous, the secretion of lysozyme and TGF-β by human macrophages and of IL-1 by monocytes is also predominantly due to subpopulations of cells.²⁸ Interestingly, only 1·8% of macrophages released TGF-β (assessed by a reverse haemolytic plaque assay), and stimulation by PMA increased the proportion of TGF-β-secreting cells to 3·1% while increasing the amount of TGF-β released by these cells (assessed by plaque size) more than 10-fold.²⁸ In these same studies, the secretion of lysozyme by macrophages was also heterogeneous with larger cells generating the greatest amount of enzyme.²⁸ It is possible that the larger, more granular cells that we have identified as over-expressing apoE may also represent a similar population of MDM with hypersecretory capacity. Indeed, the high expression of apoE appeared to be associated with numerous vesicle-like structures that may be secretory vesicles.

The factors that lead to the heterogeneous expression of apoE and other inducible proteins in human MDM are incompletely understood. Earlier work suggested that specific subpopulations of monocytes may give rise to MDM that are functionally heterogeneous via a deterministic differentiation pathway.²⁸ Consistent with this idea was the observation that only a small fraction (4·2%) of freshly isolated human monocytes had the capacity to release IL-1; this implies that heterogeneous subpopulations of monocytes may give rise to functionally distinct macrophage subpopulations.²⁸ As was observed for macrophages, stimulation of monocytes with PMA significantly increased the amount of IL-1 released from this monocyte subpopulation while only marginally increasing the size of the IL-1 secreting population to 5–6%.²⁸ More recently, studies of lipopolysaccharide-inducible gene expression in RAW264 macrophages demonstrate that heterogeneous macrophage phenotypes arise as a result of the transcriptional probability associated with inducible genes.²⁶ It is possible that apoE expression is also controlled in a similar fashion either directly or possibly via the transcription of other genes that could influence macrophage apoE levels. It will be important to study the heterogeneous apoE expression in order to better understand how apoE levels are regulated by numerous cytokines, growth factors and pharmacological agents.^19,22,24

The THP-1 cell line is widely used for in vitro studies of human macrophage biology.⁵⁶ Stimulation of THP-1 cells with PMA and other structurally related phorbol esters promotes differentiation to a macrophage phenotype and concomitantly up-regulates apoE expression.⁴⁰ Our studies revealed a marked heterogeneity in apoE expression in THP-1 macrophages with a notable over-expression in approximately 5% of cells. The underlying mechanisms resulting in apoE over-expression in THP-1 macrophages have not been addressed here; however, from a morphological point of view, the factors appear to be different to those regulating MDM apoE over-expression. In THP-1 macrophages we found that apoE was over-expressed in cells that contained condensed or fragmented nuclei. These morphological changes are hallmarks of apoptosis and are know to be dependent on activation of caspases, caspase-3 in particular.^49,50 Previous studies indicated that approximately 50% of PMA treated THP-1 macrophages express caspase-3 activity yet only approximately 5% of these cells become non-viable, through apoptotic cell death.^44,45 Our data also indicate that 5% of THP-1 macrophages exhibit apoptotic nuclear morphology and, for the first time, reveal that this is associated with increased apoE expression.

Treatment of many cell types with staurosporine induces apoptosis via a caspase-3-dependent pathway.⁵⁷ Interestingly, staurosporine treatment moderately increased THP-1 macrophage caspase-3 activity and this was associated with increased apoE expression. Previous work has shown that macrophages are resistant to certain pro-apoptotic stimuli. For example, in THP-1 macrophages, PMA treatment increased caspase-3 and Fas-ligand expression while increasing cellular resistance to the pro-apoptotic agent, NaF.³⁴ The resistance of macrophages to apoptosis was correlated with class-A scavenger receptor type I expression³⁴ and may also be partially caused by the constitutive expression of the so called inhibitor of apoptosis proteins and the anti-apoptotic caspase-2 short isoform that are expressed in human macrophages.^58,59 Whether or not apoE plays a role in macrophage apoptosis remains to be determined. ApoE may facilitate lipid membrane redistribution in apoptotic macrophages and could potentially facilitate the clearance of apoptotic debris via receptor mediated mechanisms, for example through its well characterized interaction with the low-density lipoprotein receptor-related protein^60,61 or dampen local inflammation through the immunoregulatory pathways described above (see Introduction). Interestingly, the induction of apoptosis in human fibroblasts is also correlated with increased apoE expression.³² Similarly, the insect structural homologue of apoE, apolipophorin III, is up-regulated during apoptosis where it is thought to facilitate lipid membrane remodelling during eclosion of the tobacco hawkmoth.⁶²

In human MDM, caspase-3 activity was present but these cells showed no morphological signs of apoptosis and neither caspase-3 or apoptotic morphological alterations were induced by staurosporine. As indicated above, MDM over-expressing apoE did not show obvious changes in nuclear morphology, although the nucleus was often larger and contained an enlarged nucleolus. These nuclear features have been observed in senescent cells and it is possible that the apoE over-expressing MDM represent an older subpopulation. While the expression of apoE in senescent macrophages has not been studied, there is evidence that apoE gene expression is increased in the brain of ageing mice.⁶³ Furthermore, we have shown that apoE levels are up-regulated in human fibroblasts with a senescent phenotype, as assessed by accelerated accumulation of the age pigment lipofuscin.³² Elucidation of the potential roles of apoE in apoptosis and senescence and of apoE over-expressing macrophage subpopulations in immunoregulation will clearly require further work.

In conclusion, these studies reveal a marked heterogeneity in human macrophage apoE expression, with minor subpopulations (accounting for approximately 5% of the total population) expressing up to 50-fold more apoE than the majority of the cells. These findings are important for the interpretation of in vitro studies where specific modulators of macrophage apoE expression may have global or selective effects and indicate the potential for heterogeneous macrophage apoE expression and subpopulation-specific immunoregulation to occur in vivo.