Differential endocytotic characteristics of a novel human B/DC cell line HBM-Noda: effective macropinocytic and phagocytic function rather than scavenging function

Abstract

In order to characterize a novel human B cell-lineage dendritic cell line (B/DC line) as an antigen-presenting cell (APC), we compared three types of endocytosis (micropinocytosis via a clathrin-coated pit, macropinocytosis via membrane ruffling, and phagocytosis) among myeloid-related, macrophage (Mφ) cell lines and a B/DC line. In the present examination, we used a unique human dendritic cell (DC) line, HBM-Noda (Noda). Flow cytometric and immunocytochemical analyses revealed that Noda not only expresses some DC markers, but also it expresses some B-cell associated markers. Noda shows strong capacities to stimulate allogenic T cells, to produce immunoglobulin G (IgG), and to perform immunoglobulin gene rearrangment. These data strongly suggest that Noda is a B-cell lineage DC line. The endocytic differences among these cell lines were as follows. (1) The level of micropinocytosis of Noda was significantly less than that of conventional human Mφ cell lines, and the formation of a clathrin-coated pit was not observed in Noda. (2) The level of macropinocytosis of Noda was also smaller than that of conventional Mφ cells indicating that the active membrane ruffling of Noda induces rapid recycling. (3) Phagocytosis of opsonized sheep red blood cells (SRBC) was performed more efficiently in Noda than in other Mφ cell lines. Collectively, these data suggest that in human bone marrow cells, we can identify a unique DC subtype, B/DC line, which develops through a lymphoid DC-differentiation pathway, and DC in this lineage plays an important role in the host immune response because of its effective uptake of a variety of size of antigens by using the skilful membrane ruffling and surface receptors

Introduction

In the current differentiation model dendritic cells (DC) and macrophages (Mφ) are bone-marrow-derived professional antigen-processing and -presenting cells (APC) that initiate and regulate immune responses, and both are considered to be a common progenitor-derived myeloid-lineage group.^1,2 However, recent studies have demonstrated that both DC and Mφ in situ can exist in different maturational stages and developmental pathways. Particularly, dedicated studies on DC differentiation have made it possible to classify DC into a number of subtypes including interdigitating cells in lymphoid organs,³ peripheral blood DC,⁴ Langerhans' cells in the epidermis of the skin^5,6, dermal DC,⁷ and thymic DC.⁸ In a recent report, Grouard et al.⁹ has revealed that plasmacytoid T cells that are present in the T-cell zone of lymphoid tissue also belong to the DC subtype. On the other hand, Dawe and Potter¹⁰ have identified the CD5⁺ lymphoblastic lymphoma cell line, P388, coexpressing B and Mφ properties. Additionally, there are some publications describing about the developmental potency of B lineage cells to DC. By concomitant expression of the c-myc and v-raf oncogene,¹¹ or phosphorylated membrane glycoproteins such as CD44 and CD45R, major histocompatibility complex (MHC) class-II,^12,13 B lineage cells develop into DC both morphologically and functionally. Alternatively, Björck et al.¹⁴ demonstrated that, not only CD4^low precursor cells from thymus, but also CD19⁺ pro-B cells develop into DC when cultured in some cytokines. By the careful studies about the lineage-specific genes of B lymphocytes, Rolink et al.¹⁵ have demonstrated that Pax5 gene is essential for the progression of adult B lymphopoiesis beyond an early progenitor (pre-BI) cell stage. Additionally they proposed that Pax5^–/– pro-B cells, but not wild type pro-B cells, can give rise to multilineage populations such as macrophages, osteoclasts, DC, granulocytes and natural killer (NK) cells in the presence of appropriate cytokines. Galy et al.¹⁶ demonstrated the existence in normal human adult and fetal bone marrow, of a novel CD34⁺ progenitor cells which have the ability to produce B cells, T cells, NK and DC. Previously, we also have established a unique cell line, HBM-Noda (Noda),¹⁷ which was derived from bone marrow cells of a myelodysplastic syndrome patient. The most characteristic feature of Noda is that it has some B-cell specific properties (immunoglobulin G (IgG) production, immunoglobulin gene rearrangement, some B-cell surface markers) as well as DC specific properties (dendritic morphology, alloantigen stimulation, DC surface markers). Therefore, it is considered that Noda corresponds to a novel lymphoid-related, B/DC line.

In order to characterize this novel B/DC line as APC, in comparison with a human Mφ cell line (HPL-Hod-1)¹⁸ and a human histiocyte cell line (U937),¹⁹ we focused on three types of endocytosis. In these days, endocytosis is divided into two types.²⁰ One is receptor-mediated endocytosis for large molecules, and the other is fluid phase uptake of small non-particle molecules.^21,22 The former, which is so-called phagocytosis, is the efficient antigen uptake process, and it is exclusively provided for the APC (B cell, DC, Mφ). In contrast, smaller molecules can be taken up by fluid phase pinocytosis and presented by APC but with a much lower efficiency than phagocytosis. Fluid phase uptake can be performed via two distinct mechanisms. One is micropinocytosis, which is the ingest of small vesicles via a clathrin-coated pit, and the other is macropinocytosis, which is the ingestion of large vesicles formed by membrane ruffling. Usually, low-density lipoproteins (LDL)^23–31 or lucifer yellow (LY)^32,33 are employed for clathrin-dependent micropinocytosis, or for clathrin-independent macropinocytosis, respectively. LDL can be internalized into the cytoplasm via either LDL receptors expressed on all cells or scavenger receptors expressed on Mφ. Receptor and ligand complex of LDL is trapped into a coated pit and transported from the endosome to the lysosome. Receptor dissociated from ligand under the acidic condition of lysosome is recycled to the cell surface. In contrast, LY accumulates within the macrophage vacuolar system, and the uptake of LY does not require the specific receptors but is dependent on membrane ruffling, which is achieved by the polymerization of cytoskeletal proteins.^34,35

Our findings show that the level of LDL micropinocytosis of Noda is less than that of Hod-1 and enormous LDL micropinocytosis of Hod-1 is dependent on the formation of a clathrin-coated pit during the incubation with LDL. On the other hand, while the LY accumulation of Noda is also smaller than that of Hod-1 and U-937, this lower amount seems to result from its more vigorous membrane ruffling compared with that of conventional Mφ cell lines. These endocytic differences possibly relate to the actual APC functions of B/DC and conventional Mφ. Moreover, B/DC can engulf the opsonized sheep red blood cells (SRBC) mediated by their long projections and their surface IgG, indicating that their effective and skilful phagocytic manner may correspond to the antigen presenting function of B/DC. We conclude that the lymphoid-related DC exists in human bone marrow haematopoietic cells and demonstrates the elaborated macropinocytosis and phagocytosis which has not been reported with conventional Mφ.

Materials and Methods

Cell lines

HBM-Noda (Noda) was established from bone marrow-aspiration fluid of a 68-year-old myelodysplastic syndrome patient in 1990 by Morikawa et al.¹⁷ Ruffled projections and invaginated nucleus were observed in Noda by the phase contrast microscopy as well as by light microscopy. Noda had locomotive and adhesive capacities in the Petri dish. Noda showed a strong capacity to stimulate E rosette-purified allogenic T cells and the capacity was more effective than that of peripheral blood non-T cells or malignant B lymphoma cell lines.¹⁷ HPL-Hod-1 (Hod-1) was established from pleural effusion of a 57-year-old Hodgkin's disease patient in 1972 by Morikawa et al.¹⁸ U937 was established from pleural effusion of a 37-years-old diffuse malignant histiocytic lymphoma patient in 1975 by Sundström and Nillson.¹⁹ These cell lines were maintained in RPMI-1640 (Gibco, Grand Island, NY) containing 2 mm l-glutamine, 10 mm Hepes, 10 mm sodium pyruvate, 100 U/ml penicillin, 10% heat-inactivated fetal calf serum (FCS), at 37° in a 95% air−5% CO₂ incubator. Hod-1 and U937 were diluted 30-fold once a week and Noda was diluted about twofold every other day.

Media and reagents

Acetylated LDL labelled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyamime perchlorate (Dil-Ac-LDL) was purchased from Biomedical Technologies, Inc. (St. Stoughton, MA). Lucifer yellow CH (LY) was purchased from Aldrich Chem. Co. (Milwaukee, WI). Phorbol myristate acetate (PMA) and rhodamine-labelled phalloidin (RP) were purchased from Sigma Immunochemicals (St. Louis, MO). Cytochalasin B was purchased from Nakalai Tesque, Inc. (Kyoto, Japan). Anti-human IgG rabbit antibody was purchased from DAKO Japan Co., Ltd. (Kyoto, Japan). SRBC were purchased from Nihon Seibutsugaku Kenkyusho (Shizuoka, Japan).

Monoclonal antibodies

Monoclonal antibodies (mAb) of the following specificity were used in this study. CD10 (SS2/36), CD11c (KB90), CD14 (TUK4), CD19 (HD37), CD20 (L-26), CD25 (ACT-1), CD30 (Ber-H2), and CD68 (KP1) were purchased from DAKO Japan Co., Ltd. (Kyoto, Japan). CD7 (T55), CD23 (H107), and HLA-DR (Nu-l) were purchased from Nichirei (Tokyo, Japan). CD86 (IT2.2) was purchased from Pharmingen (San Diego, CA). CD11a (G43-25B), CD18 (6.7), CD44 (L178), CD54 (LB-2), CD58 (L306.4), and CD80 (L307.4) were purchased from Becton Dickinson (Mountain View, CA).

Immunocytochemistry

Cells suspended in medium (1–2 × 10⁵) were smeared on slide glasses which were dried and fixed with acetone at room temperature. The slide glasses were immersed in phosphate-buffered saline (PBS) and then in 0·5% skimmed milk for the prevention of non-specific stain. After washing with PBS, each sample was incubated with the mAb described above for 30 min at room temperature in a humidified box. After washing with PBS, they were incubated with biotin-labelled anti-mouse IgG(H + l) (Vector, Burlingame, CA) under the same condition as the first antibody. After the second wash, samples were incubated with avidin-labelled alkaline phosphatase (Alp; DAKO). Alp localization was revealed using naphtol AS-BI (6-bromo-2-hydroxy-3-naphthoic acid 2-methoxyanilide; Sigma) substrate, resulting in purple staining.

Flow cytometry

Cells (1–2 × 10⁵) suspended in Hank's balanced salt solution (HBSS) were incubated with each monoclonal antibody for 30 min on ice. When unlabelled mAb was used as the primary reagent, fluoroscein isothiocyanate (FITC)-labelled goat anti-mouse IgG F(ab′)₂ was used as a secondary reagent. Flow cytometrical analysis was performed with a FACScan flow cytometer (Beckton Dickinson) equipped with a single laser emitting at 488 nm.

Pinocytosis of Dil-Ac-LDL and LY

Cells suspended in RPMI-1640 supplemented with 10% FCS (1–2 × 10⁶) were cultured in the presence of 10 µg/ml Dil-Ac-LDL for 48 hr in a CO₂ incubator. Cells harvested at the indicated time were extensively washed with cold HBSS and analysed for FACScan (Beckton Dickinson). LY was dissolved in PBS and spun in a microfuge before use to eliminate aggregates. LY were added at a final concentration of 1 mg/ml for different times if not otherwise indicated. In the case of cell stimulation, cultured cells washed with medium were resuspended in RPMI-1640 and stimulated with PMA at a final concentration 4 ng/ml in medium for 2 or 14 hr. Afterwards, the stimulated cells were harvested and cultured with LY as described above.

Fluorescence microscopy

Cells incubated with Dil-Ac-LDL for 24 hr were fixed with 3% formaldehyde at room temperature for 15 min, and cell suspensions were dropped onto the slide glass followed by sealing with nail varnish. LY endocytic cell suspensions were prepared as described above without fixation. These samples were observed by a fluorescent microscope and photographed on Microphot-FXA (Nikon, Tokyo, Japan) using Fujichrome RD135 film (Fuji Film Co., Tokyo, Japan).

Blocking of actin polymerization by cytochalasin B

PMA stimulated or non-stimulated cells were harvested and washed twice with the medium. These cells were incubated with LY as described above in the presence or absence of 10⁻¹ mm cytochalasin B for 2 hr and analysed for FACScan as described above.

Preparation of anti-human IgG-coated SRBC

SRBC in Alsever's solution were extensively washed with saline four times at 4°. Packed SRBC (100 µl) by centrifugation at 800 g. were resuspended in 66 µg/ml CrCl₃ solution and anti-human IgG rabbit antibody, and incubated for 30 min at 37°. After then non-binding antibody was removed by the extensive wash with medium containing no serum.

Binding and phagocytosis assay for SRBC

Cells (1–2 × 10⁵) resuspended in RPMI-1640 without FCS were mixed with opsonized SRBC for 30 min at 37°. The binding of opsonized SRBC to the cultured cells was able to be observed microscopically and the number of cultured cells of rosette formation with SRBC were counted. For the identification of SRBC phagocytosis, cultured cells were prepared for the transmission electron microscopy as below.

Transmission electron microscopy (EM)

For the EM examination of SRBC phagocytosis, cells were fixed in 2·5% glutaraldehyde and 2% paraformaldehyde in 0·1 m phosphate buffer on ice for 90 min. After five washes with 8% sucrose in 0·1 m phosphate buffer, cells were post fixed in 1% osmium tetraoxide on ice for 60 min. Finally, they were embedded in 1% agar and cut into small pieces. These samples were dehydrated through a graded ethanol series and embedded in Epon. The sections were cut on a microtome (Sorvall Ultra Microtome, MT5000, RMC, Tucson, AZ) and stained with uranyl acetate and lead citrate, and visualized and photographed using an electron microscope operation at 80 kV (Topcon, Tokyo, Japan).

Results

Differences of surface phenotype between Noda and conventional Mφ cell lines (Hod-1 and U937): flow cytometric and imunocytochemical analyses

Based on our findings that Noda demonstrates DC-like morphology, but simultaneously has B-cell specific properties such as immunoglobulin-gene rearrangement and IgG production,¹⁷ it can be suggested that Noda can be classified into B/DC lineage. Therefore, to determine the precise lineage of Noda, we evaluated the lymphoid, myeloid, and monocyte marker expressions in Noda and compared them with other conventional Mφ cell lines by flow cytometer and immunocytochemistry. (Table 1.) B-cell markers, CD19, CD20, and CD25 were expressed only on Noda but not on other conventional Mφ cell lines. The lymphocyte common antigen markers CD45RA, CD45RO were also expressed on Noda, and only CD45RO was expressed on U937, but these markers were never observed on Hod-1. Among the myeloid markers, CD68 were expressed on Noda and U937 but other myeloid markers, CD14, CD11c were not expressed on Noda. MHC class-II, human leucocyte antigen (HLA)-DR was strongly expressed on Noda and U937. Moreover, it is very interesting that Noda significantly expresses costimulatory molecules such as CD80 and CD86, which play an essential role in T-cell priming by mature DC. Noda also significantly expresses adhesion molecule, such as CD11a, CD18, CD44, CD54, CD58. In contrast, Hod-1 failed to express these molecules and U937 expressed only adhesion molecules. From these phenotypic characteristics, it is suggested that newly established Noda can be classified into B-cell lineage DC lines rather than myeloid-lineage DC lines.

Table 1. Analysis of the reactivities to monoclonal antibodies of conventional Mφ and B/DC cell lines.

		Hod-1	U937	HBM-Noda
B-cell lineage
CD19	F	−	−	+
CD20	F/IC	−	−	+
CD25	F	−	−	+
Myeloid, Mφ				−
CD11c	F	N.D.	N.D.	−
CD14	F	−	−	−
CD68	IC	+−	++	+
Lymphohaematopoietic
CD10 (CALLA)	F	−	−	−
CD45 (common)	F	−	+	+
CD45 (RA)	F	−	ND	+
MHC molecule
HLA-DR	F	−	++	++
Receptors
CD7 (FcµR)	F	−	−	−
CD23(FcεR)	F	−	ND	++
CD25(IL-2R)	F	−	−	+−
Co-stimulatory/cell adherence
CD11a(LFA-1a)	F	−	++	+
CD18(LFA-1ß)	F	−	++	++
CD30	F	−	−	++
CD44	F	−	++	++
CD54 (ICAM-1)	F	−	++	++
CD58 (LFA-3)	F	++	++	++
CD80	F	−	−	++
CD86	F	−	ND	++

Each cell line was cultured without any growth factors as described in Materials and Methods, and analysed by flow cytometry (F), immunocytochemistry (IC) or both (F/IH). +, ++, positive, ± non-prominently positive, – negative.ICAM, intracellular adhesion molecule; LFA, leucocyte function-associated antigen; ND, not determined.

Comparison of micropinocytic capacity between Noda and conventional Mφ cells

Because LDL was known to be a useful probe for receptor-mediated micropinocytosis,^23–31 we used fluorescence labelled-modified LDL in our micropinocytic assay. As shown in Fig. 1, within 48 hr of incubation, Hod-1 accumulated a great deal of LDL in a linear fashion and final amounts of pinocytosed LDL of Hod-1 were significantly greater than that of Noda. In contrast, LDL accumulation in Noda was linear in the first 24 hr, but afterwards the accumulation rate was slightly decreased. The degree of LDL accumulation by U937 was as same as Noda at the final stage of incubation and it was much lower than that of Hod-1. These results suggested that the amount of endocytosed LDL within 48 hr was different among these cell lines, and Hod-1 could most vigorously and continuously internalize it. From the fluorescence microscopical observations after the LDL incubation, a number of large vesicular pinosomes distributed throughout the cytoplasm were observed in Hod-1, but a few number of small and localized pinosomes were observed in Noda. (data not shown) Therefore, it is expected that each cell line not only has a different micropinocytic capacity but also a different micropinocytic traffic.

Figure 1.

Kinetic analysis of LDL micropinocytosis among Noda and conventional Mφ lines.The result represents the mean fluorescence intensity of Hod-1, Noda and U937 incubated with Dil-AC-LDL (10 µg/ml) for various time periods. Data shown are representative of three different experiments.

Clathrin-coated pits contribute to effective LDL micropinocytosis in Hod-1 but not in Noda

As previously noted, receptor-mediated pinocytosis is induced by the clathrin-coated pit. In order to determine whether the formation of the clathrin-coated pit or clathrin coated-vesicle is related to active pinocytic activity of Hod-1 or not, we attempted EM examination of cells cocultured with LDL for 24 hr. While we have repeatedly observed the characteristic dendritic morphology of Noda,¹⁷ further morphological properties were obtained from EM examination. As shown in Fig. 2, EM examination of Noda showed that many long and ruffled dendritic projections, and their lobulated nucleus had an irregular outline with many invaginations, a dense perinuclear chromatin and often a prominent nucleus. Additionally, the cytoplasm contains a well-developed Golgi complex, many dark homogeneous membrane-bound granules similar to lysosomes, and an important number of large mitochondria. Sometimes, multivesicular bodies were observed near the nucleus (Fig. 2a,b). No Birbek or Birbeck-like granules could be found in Noda. By contrast, Hod-1 had few projections and contained poor cytoplasmic organelles. However, as shown in Fig. 2(c), a number of clathrin-coated pits and clathrin-coated vesicles, which were constitutively picked off from the cell surface, were observed in Hod-1. Clathrin-coated vesicles were prominent in Hod-1 with the highly magnified EM photographs (Fig. 2d). These results suggest that active LDL micropinocytosis, via the clathrin-coated pit, contributed to the vigorous endocytic capacity of Hod-1 but not B/DC, and this might be closely related to the scavenging capacity of conventional Mφ.

Figure 2.

EM examination of the clathrin-coated pit and clathrin-coated vesicle in Hod-1. Noda (a), which were processed for EM examination immediately after the harvest from culture dish, demonstrated many characteristic ruffled projections and an invaginated nucleus. Intracytoplasmic organelles of Noda such as Golgi-complex, lysosomes and mitochondria were more clearly observed with high magnification (b). Some clathrin-coated pits and clathrin-coated vesicles were recognized in Hod-1 (c), and in the right lower corner of (c), there were two clathrin-coated pits. These could be more easily observed in high magnification (arrow) (d) G, Golgi-complex; Ly, lysosome; M, mitochondria; ER, rough endoplasmic reticulum. Bar represents 1 µm.

Comparison of clathrin-independent macropinocytosis between Noda and conventional Mφ cell lines

In order to examine the alternative vesicular pathway, we used LY which is known to be an excellent probe for macropinocytosis.^32,33 As shown in Fig. 3(a), LY accumulation was temperature dependent in Noda and Hod-1, and the whole population of Hod-1 could accumulate a comparable amount of LY only after 60 min. The whole population of Noda shifted to the right like Hod-1, while the amount of LY accumulation was less than that of Hod-1 at least 4 hr. Although LY was accumulated with a curvilinear pattern in both Hod-1 and Noda, the total amount of LY accumulation in Hod-1 was greater than that in Noda. (Fig. 3b) These results indicate that LY retention capacity of Noda is less than that of Hod-1 because of the rapid influx and efflux of LY by active membrane ruffling of Noda.

Figure 3.

Analysis of LY accumulation of Noda and Hod-1. Noda (a, c), Hod-1 (b, d) were incubated at 37° (a, b) or 4° (c, d) for 1 hr or 4 hr in the presence or absence of LY (1 mg/ml). After washing with cold HBSS, cells were analysed by fluorescence-activated cell sorting (FACScan). Solid lines represent the profile of cells incubated in the absence of LY. Dotted lines or bold lines represent the profile of cells incubated in the presence of LY for 1 hr or 4 hr, respectively, and (e) represents the result of mean fluorescence intensity (MFI) of internalized LY by Hod-1 (dotted line) and Noda (solid line). Data shown are representative of three different experiments.

Since macropinocytosis is formed by membrane ruffling driven by actin cytoskeleton,³² we intended to investigate whether or not the magnitude of LY accumulation is affected by the stimulation of the membrane ruffling. As shown in Fig. 4, LY accumulation in Noda was enhanced about fourfold by PMA stimulation, however, it was not affected or only slightly enhanced both in Hod-1 and U937. Additionally, we could observe the appearance of a fluorescence bright population in Noda but not in Hod-1 and U937 by flow cytometry. (Fig. 5) In the pulse–chase analysis, fluorescence intensity expressed by the PMA sensitive, fluorescence-bright populations of Noda was not reduced during the 4 hr of the chase periods, suggesting that the pathway for the delay of LY excretion was induced in the PMA sensitive populations of Noda (data not shown). These results show the different behaviours for macropinocytosis among each cell line examined in the present study. The retention capacity of macropinocytosis in Noda was lower than that in other Mφ cell lines unless the PMA stimulation was used, but Noda was the most sensitive cell line to PMA stimulation among them. Similarly, the effect of PMA stimulation on macropinocytosis was not only the induction of membrane ruffling but also the alteration of the LY traffic pathway.

Figure 4.

The effect of PMA stimulation on LY macropinocytosis. After PMA (4 ng/ml) stimulation for 2 hr, LY uptake was remarkably enhanced in Noda, but was not changeable in Hod-1 and U937. The y-axis expresses the ratio of fluorescence intensity of PMA stimulated cells versus unstimulated cells in each cell line. Data shown are representative of three different experiments.

Figure 5.

PMA stimulation induces the fluorescence bright population in Noda.After 0 hr (a, b, c), 2 hr (d, e, f) or 14 hr (g, h, i) of PMA stimulation, Noda (a, d, g), Hod-1 (b, e, h), and U937 (c, f, i) were incubated with 1 mg/ml LY at 37° at various time points, and analysed by FACScan. Solid lines represent the profile of cells incubated in the absence of LY, dotted lines, dashed lines and bold lines represent the profile of cells incubated in the presence of LY for 10 min, 60 min and 120 min, respectively. Data shown are representative of three different experiments.

For further investigation as to the mechanism involved in the enhancement by PMA stimulation, we have evaluated the effect of cytochalasin B, which blocks actin polymerization.³⁶ As shown in Fig. 6, when compared with the PMA untreated case, the fluorescent bright population was increased in Noda by PMA stimulation and the effect was not completely but partially inhibited by the cytochalasin B treatment. Taken together, it is suggested that the effect of PMA on LY macropinocytosis is not only the stimulation of membrane ruffling by actin polymerization but also the alteration of LY traffic pathway which induces the enhancement of LY retention.

Figure 6.

Cytochalasin B partially blocks the enhancement of fluorescence bright population in Noda. After the incubation in the presence (a, b) or absence (c) of PMA (4 ng/ml) for 2 hr, Noda was incubated with LY (1 mg/ml) in the presence (b) or absence (a, c) of 10⁻¹ mm cytochalasin B at 37° for 2 hr. These cells were washed with cold HBSS and analysed by FACScan. Dotted lines represent the background fluorescence intensity.

Phagocytosis of opsonized SRBC by B/DC cell but not conventional Mφ cell lines

As previously noted, phagocytosis of small particulates such as carbon particles or latex beads as well as large particles, such as senescent erythrocyte, is one of the specific properties of Mφ. However, as the diameter of erythrocytes is about 10 times larger than that of carbon particles or latex beads, it is considered that well-elaborated mechanisms are needed for their complete ingestion. To explore the differences of their erythrophagocytotic manner, in the preliminary experiment, we examined the binding capacities of each cell line with anti-human IgG-coated SRBC. A typical SRBC-rosette formation was only observed in Noda but not in other conventional Mφ cells. The SRBC binding and rosette formation by Noda was completely or partially blocked by human IgG pre-treatment or cytochalasin-B pre-treatment, respectively (data not shown). These data suggested that the sequential events such as the receptor-ligand binding and the actin polymerization highly contribute to the SRBC binding and rosette formation by Noda. To further attempt to determine whether SRBC phagocytosis actually occurred or not, we carried on EM-examination for cell incubated with opsonized SRBC. As shown in Fig. 7, transmission electron microscopic sections of Noda cultured with anti-human IgG coated-SRBC revealed the various stages of SRBC phagocytosis, in which some SRBC are bound to the cell surface or partially surrounded by extended surface projections, and others are completely engulfed into Noda. In addition, we found that the requirement of SRBC deformity and the close attachments to Noda prior to complete engulfment. Although we could not find the SRBC phagocytosis by Hod-1 or U937, Hod-1 contained several phagosomes in which membranous or small vesicular structures were involved (data not shown). Cytoskeletal actin polymerization might be necessary for these elaborated membrane movements,³⁷ therefore, we examined the effect of actin polymerization on SRBC phagocytosis. As shown in Fig. 8, after the incubation with SRBC, the rhodamin stained dendritic projections were more clearly demonstrated on the surface of Noda with the time dependent manner and some of them were looked like wrapping SRBC (Fig. 8a, b). By contrast, Hod-1 have never demonstrated the extension of dendritic projections even after the incubation with SRBC (Fig. 8c). Taken together, B/DC cells were provided with efficient phagocytic machinery by which co-ordinated sequential events, such as SRBC binding and the complete ingestion of them, were carried out. Additionally, it is speculated that the binding to the surface IgG and then the actin polymerization highly contributed to the erythrophagocytosis by Noda. Instead, phagocytic activity of conventional Mφ cells is not so strong suggesting that fragmentation of them prior to ingestion is necessary.

Figure 7.

SRBC phagocytosis by Noda. After incubation with antihuman IgG-coated SRBC for 30 min at 37°, Noda (a, b) and Hod-1 (c) were processed for EM examination. Noda revealed the various stages of SRBC phagocytosis. Some SRBC were completely (small arrow) or partially (large arrow) engulfed by Noda. (a) Phagocytic arms of Noda extended to surround the SRBC and before the complete engulfment of SRBC, Noda and SRBC closely attached at several surface points. (b) The SRBC phagocytosis by Hod-1 was never observed in this experimental condition. (c) Bar represents 1 µm.

Figure 8.

Actin polymerization contributes to the SRBC phagocytosis of Noda. Noda (a, b) or Hod-1(c) incubated with anti-human IgG-coated SRBC at 37° were stained with rhodamin-labelled phalloidin. Noda extended the long projections from the cell surface within 2 min incubation (a), and the circumscribing of SRBC was observed after 5 min (b) Contrast to Noda, the cell surface of Hod-1 was smooth and the extension of projections could not be observed (magnification × 400).

Discussion

Previously, both DC and Mφ have been recognized to be myeloid-associated APC;^1,2 however, some disputes about the current differentiation models have been raised. In particular, DC have been carefully classified based on their differentiation pathway or maturational state in situ. From a differentiation point of view, Noda is considered to be a novel lymphoid-associated DC, B/DC line because of its high expression of B or lymphoid-associated markers as well as mature DC markers.¹⁷ In contrast, Hod-1 and U937 are myeloid-associated conventional Mφ lines.^18,19 In this paper we have studied the heterogeneity of three types of endocytosis namely phagocytosis, micropinocytosis and macropinocytosis observed in the B/DC line, Noda, and the conventional Mφ lines, Hod-1, and U937.

For the isolation of pure DC or Mφ, the use of some growth factors^38–40 and complicated isolation steps are required because both DC and Mφ consist of very tiny populations in situ. In such cases, the possibility of phenotypic changes by in vitro stimulation has to be excluded, moreover, cell purity should be carefully examined. Because we carried out the present study using long-term cultured human cell lines that were maintained without any growth factors which might affect DC maturation state, the results obtained in this report directly reflect the original cell properties. Additionally, our present study, with its related article,¹⁷ is the first document in which the existence of human B-cell lineage DC in bone marrow was demonstrated and the characteristic of endocytosis of B/DC was evaluated.

In comparison to receptor-mediated micropinocytosis, the magnitude of LDL internalization in Hod-1 was greater than that in Noda, and the ingested LDL was distributed throughout the cytoplasm in Hod-1. Because LDL micropinocytosis is known to be mediated by cell surface receptors,^41,42 it is important to evaluate the number of LDL receptors which are de novo synthesized or recycled to the cell surface. We have not yet measured the number of LDL receptors on each cell line used in this study; however, the result, which demonstrates the linear fashion of LDL accumulation in Hod-1, indicates that a large amount of LDL accumulation is caused by the increase of de novo synthesis and/or the rapid and continuous recycling of receptors. In contrast, the decrease of LDL accumulation in Noda from 24 to 48 hr was probably caused by the saturation of receptors. This was also confirmed by EM observation. From our present results dealing with LDL pinocytosis, it is suggested that the manner of micropinocytosis in Hod-1 is so appropriate for the tissue scavenging that it can play an essential role in the elimination of destructive tissues or apoptotic cells. In contrast to Mφ cells, B/DC lack a well-developed scavenging pathway for complete digestion of endocytosed materials into amino acids in lysosomes.

In macropinocytosis the level of LY accumulation was significantly larger in Hod-1 than in Noda, indicating that both influx and efflux of LY by membrane ruffling was quickly performed in Noda. The sensitivity of LY macropinocytosis to PMA treatment was also different between conventional Mφ cells and Noda, and some populations of Noda acquired the ability to accumulate a larger amount of LY after PMA stimulation, but this was not observed in the other conventional Mφ lines.

Although macropinosomes may interact with other organelles such as endosomes, lysosomes or MHC class II-rich vesicles, the large volume of internalized fluid must be disposed at some point.⁴³ Steinman and Swanson⁴⁴ have proposed that this could occur in the following two ways. When the macropinosomes themselves recycle to the plasma membrane, most of the internalized membrane and fluid would reflux and solute would not accumulate. Alternatively, macropinosomes would shrink inside the cell, with membrane recycling to the cell surface via smaller vesicles or shrunken multivesicular bodies, and water leaving by diffusion across membranes. In this case, internalized solutes might remain trapped inside the shrinking macropinosome for longer periods. From our present observations, it is postulated that, like the rapid recycling model by Levine and Chain,⁴⁵ Noda have the direct recycling pathway by which a large volume of internalized fluid is rapidly dispersed. In contrast, Hod-1 and U937 have the other recycling pathway. According to Steinman et al.⁴⁶, in the case of Mφ cells, macropinosomes fuse with lysosomes and most of the internalized solutes remain in the cell. Although we could not determine the precise mechanisms of LY macropinocytosis in each cell lines, it is speculated that the LY macropinocytosis of Noda is closely linked to its antigen-presenting capacity.

Swanson et al.⁴⁷ have found that PMA dramatically alters the pinocytotic activity of thioglycollate-elicited mouse peritoneal macrophage. In such a case, the volume of LY accumulation linearly increases and is four to seven times the steady-state rate of accumulation by unstimulated cells. However, an important finding is that the actual volume of LY influx is increased by only 1·6 times by PMA stimulation. This contradiction may be explained by the result of our pulse–chase experiment: that the delay of LY excretion results in the increase of net effect of LY accumulation. Additionally, the delay of LY excretion probably resulted from the alteration of LY transport from the endosomal pathway to the lysosomal one. Sullusto et al.⁴⁸ reported that peripheral blood-derived DC stimulated with granulocyte–macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4) can accumulate a great deal of LY by macropinocytosis. These results are consistent with earlier findings that growth factor enhances the macropinocytotic capacity in the cells.^49,50 The stimulation by GM-CSF and IL-4 as well as by PMA may induce the activation of DC; for example, the polymerization of cytoskeletal proteins which may induce the active membrane ruffling of DC. However, the precise mechanisms concerning the activation mechanisms are not resolved and should be investigated.

We used opsonized SRBC in order to study the phagocytosis in which large particles were ingested via surface receptors. In contrast to conventional Mφ, Noda had the prominent capacity to bind and engulf them in a well elaborated fashion. Griffin et al.³⁷ have found that erythrocytes which were functionally opsonized with only one ‘hemisphere’ that could bind to mouse peritoneal macrophage but not be ingested. From these facts, they concluded that the sequential and circumferential attachments with SRBC surface are required for the phagocytosis of opsonized SRBC. They referred to this phenomenon as a ‘zipper effect’. Although we could not duplicate a ‘zipper effect’ finding, EM sections of Noda incubated with anti-human IgG-coated SRBC revealed the partial binding or complete engulfment of SRBC. Accordingly, we postulate the factors, which are probably triggered by the SRBC binding, evoke the phagocytic arm to bend and to fuse with each other in order to engulf SRBC. Additionally, for these continuous processes to phagocytose SRBC, polymerization of actin filament is essential, and this was proved by our blocking experiments by cytochalasin B and by the staining with rhodamine-labelled phalloidin. Recently, mammalian rho proteins, which are guanosine triphosphate (GTP)-binding proteins and hydrolyse GTP, have been shown to regulate the formation of actin stress fibre in fibroblast.⁵¹ In addition, CDC42Sc, the Saccharomyces cerevisiae homologue of CDC42Hs, is involved in controlling cell polarity, which is linked to the spatial organization of polymerized actin. Ridley and Hall⁵² reported that rho proteins are essential components of signal transduction pathways in order to organize actin polymerization. Although we do not know the additional factors that are triggered by the interaction of Noda and opsonized SRBC, we think the activation of several kinase proteins or the rearrangement of cytoskeletal proteins is required for the complete ingestion of SRBC.

In summary, we have identified B-cell lineage DC in human bone marrow cells. This population has the capacity to ingest extracellular molecules by active macropinocytosis and phagocytosis via membrane ruffling and co-ordinated membrane movement. In the present study, we can demonstrate that these endocytic properties of B/DC are quite different from those of conventional Mφ. These results suggest that endocytic differences might be relevant to the functional differences of APC. The further characterization of Noda should be necessary for the classification of the DC subpopulation and the functional separation of B/DC from other APC.