Use of human chondrocyte cell cultures to identify and characterize reactive antibodies in rheumatoid arthritis sera.

OBJECTIVE: To study the reactivity of rheumatoid arthritis (RA) sera with human chondrocyte populations isolated from normal cartilage and expanded in vitro. METHODS: Human articular chondrocytes were cultured as adherent (non-differentiated) cells on plastic dishes or in suspension (differentiated) on dishes previously coated with a thin layer of 1% agarose. Sera from 28 RA patients and 5 paired synovial fluids were tested on lysates from chondrocytes and fibroblasts as control by immunoblot. Antigen expression on the cell membrane was evaluated by flow cytometry in a few sera. RESULTS: In 9/28 RA sera IgG antibodies specific for chondrocyte antigens (97 kDa, 74 kDa, 67 kDa, 60 kDa, 54 kDa, 48 kDa and 37 kDa) were detected. Twelve sera reacted with proteins expressed both on chondrocytes and fibroblasts and 7 with fibroblasts only; two sera had no reactivity. When lysates from adherent or suspension chondrocytes were compared, RA sera reacted with higher intensity and detected more antigens on chondrocytes cultured in suspension. Flow cytometry assay demonstrated that RA sera are able to recognize antigens expressed on the cell membrane of the human chondrocytes. CONCLUSION: Our data indicate that: a) 32% of the RA sera contain antibodies reactive with antigens expressed exclusively by chondrocytes, but this value rises to 75% if antigens expressed both by chondrocytes and fibroblasts are considered; b) the reactivity of fully differentiated chondrocytes in suspension culture is higher than the reactivity of chondrocytes cultured in monolayer; and c) some of the chondrocyte-specific antigens identified are associated with the chondrocyte membrane. Thus, in vitro cultured chondrocytes may be used to study both the specificity and the biological activity of autoantibodies in RA.

Dense core secretory vesicles revealed as a dynamic Ca(2+) store in neuroendocrine cells with a vesicle-associated membrane protein aequorin chimaera.

The role of dense core secretory vesicles in the control of cytosolic-free Ca(2+) concentrations ([Ca(2+)](c)) in neuronal and neuroendocrine cells is enigmatic. By constructing a vesicle-associated membrane protein 2-synaptobrevin.aequorin chimera, we show that in clonal pancreatic islet beta-cells: (a) increases in [Ca(2+)](c) cause a prompt increase in intravesicular-free Ca(2+) concentration ([Ca(2+)]SV), which is mediated by a P-type Ca(2+)-ATPase distinct from the sarco(endo) plasmic reticulum Ca(2+)-ATPase, but which may be related to the PMR1/ATP2C1 family of Ca(2+) pumps; (b) steady state Ca(2+) concentrations are 3-5-fold lower in secretory vesicles than in the endoplasmic reticulum (ER) or Golgi apparatus, suggesting the existence of tightly bound and more rapidly exchanging pools of Ca(2+); (c) inositol (1,4,5) trisphosphate has no impact on [Ca(2+)](SV) in intact or permeabilized cells; and (d) ryanodine receptor (RyR) activation with caffeine or 4-chloro-3-ethylphenol in intact cells, or cyclic ADPribose in permeabilized cells, causes a dramatic fall in [Ca(2+)](SV). Thus, secretory vesicles represent a dynamic Ca(2+) store in neuroendocrine cells, whose characteristics are in part distinct from the ER/Golgi apparatus. The presence of RyRs on secretory vesicles suggests that local Ca(2+)-induced Ca(2+) release from vesicles docked at the plasma membrane could participate in triggering exocytosis.

The Eps15 C. elegans homologue EHS-1 is implicated in synaptic vesicle recycling.

Eps15 represents the prototype of a family of evolutionarily conserved proteins that are characterized by the presence of the EH domain, a protein-protein interaction module, and that are involved in many aspects of intracellular vesicular sorting. Although biochemical and functional studies have implicated Eps15 in endocytosis, its function in the endocytic machinery remains unclear. Here we show that the Caenorhabditis elegans gene, zk1248.3 (ehs-1), is the orthologue of Eps15 in nematodes, and that its product, EHS-1, localizes to synaptic-rich regions. ehs-1-impaired worms showed temperature-dependent depletion of synaptic vesicles and uncoordinated movement. These phenotypes could be correlated with a presynaptic defect in neurotransmission. Impairment of EHS-1 function in dyn-1(ky51) worms, which express a mutant form of dynamin and display a temperature-sensitive locomotion defect, resulted in a worsening of the dyn-1 phenotype and uncoordination at the permissive temperature. Thus, ehs-1 and dyn-1 interact genetically. Moreover, mammalian Eps15 and dynamin protein were shown to interact in vivo. Taken together, our results indicate that EHS-1 acts in synaptic vesicle recycling and that its function might be linked to that of dynamin.

Signaling through CD38 induces NK cell activation.

Human CD38 is a signal transduction molecule, and, concurrently, an ectoenzyme catalyzing the synthesis and degradation of cyclic ADP-ribose (cADPR), a potent Ca2+ mobilizer. One facet of CD38 that has not yet been addressed is its role in NK cells. To this end, the events triggered by CD38 ligation with agonistic mAb were analyzed on freshly purified human NK cells. Ligation was followed by (i) a significant rise in the intracellular level of Ca2+, (ii) increased expression of HLA class II and CD25, and (iii) tyrosine phosphorylation of discrete cytoplasmic substrates. The phosphorylation cascade involved CD3-zeta and FcepsilonRIgamma chains, zeta-associated protein (ZAP)-70 and the proto-oncogene product c-Cbl. NK effector functions were then analyzed: CD38 signaling was able (iv) to induce release of IFN-gamma and, more prominently, of granulocyte macrophage colony stimulating factor, as assessed by measuring both mRNA and protein products; and, lastly, (v) to induce cytolytic effector functions on target cells after IL-2 activation, as shown both by cytotoxicity assays and ultrastructural changes. The tyrosine-phosphorylated substrates and all the effects mediated by CD38 were similar to those observed following triggering via CD16 (FcgammaRIIIA); moreover, Ca2+ mobilization via CD38 no longer operated in NK-derived cell lines lacking CD16. These results suggest that the activation signals transduced by CD38 in NK cells elicit relevant cellular events. The effects are similar to those elicited via CD16 and possibly rely on common signaling pathways.

Cytokine gene expression and T-cell proliferative responses in lymph node mononuclear cells from children with early stage human immunodeficiency virus infection.

BACKGROUND AND OBJECTIVES: The immunologic events taking place in secondary lymphoid tissue from children with early stage human immunodeficiency virus (HIV) infection are poorly understood. The aim of this study was to investigate cytokine gene expression and proliferative responses in lymph node (LN) biopsies from five children with early stage HIV infection, in the context of LN morphology and viral load. DESIGN AND METHODS: The design of the study was approved by the local Ethical Committee. Cytokine gene expression was studied in LN biopsies and in paired peripheral blood (PB) samples from HIV-infected children by reverse transcriptase-polymerase chain reaction. T-cell proliferation was assessed by 3H-thymidine incorporation. Viral burden in germinal centers was assessed by video densitometric analysis following immunohistochemical staining for HIV p24. RESULTS: Interleukin (IL)-2, IL-4 and IL-5 mRNA were not detected in any LN or PB sample from HIV-infected children. Interferon (IFN)-gamma mRNA was found only in CD8+ cells. IL-12 p35, IL-10, transforming growth factor-(TGF)-beta1, regulated on activation normal T-cell expressed and secreted (RANTES), macrophage inflammatory protein (MIP)-1alpha, MIP-1beta and IL-16 transcripts were detected in all samples. Proliferation of LN and PB mononuclear cells to polyclonal mitogens and soluble (recall and HIV-related) antigens was impaired as compared with the responses in a group of age-matched healthy controls. INTERPRETATION AND CONCLUSIONS: Changes in cytokine gene expression and T-cell proliferative responses are already detectable in lymph nodes from HIV-infected children at an early stage of disease.

Numb is an endocytic protein.

Numb is a protein that in Drosophila determines cell fate as a result of its asymmetric partitioning at mitosis. The function of Numb has been linked to its ability to bind and to biologically antagonize Notch, a membrane receptor that also specifies cell fate. The biochemical mechanisms underlying the action of Numb, however, are still largely unknown. The wide pattern of expression of Numb suggests a general function in cellular homeostasis that could be additional to, or part of, its action in fate determination. Such a function could be endocytosis, as suggested by the interaction of Numb with Eps15, a component of the endocytic machinery. Here, we demonstrate that Numb is an endocytic protein. We found that Numb localizes to endocytic organelles and is cotrafficked with internalizing receptors. Moreover, it associates with the appendage domain of alpha adaptin, a subunit of AP2, a major component of clathrin-coated pits. Finally, fragments of Numb act as dominant negatives on both constitutive and ligand-regulated receptor-mediated internalization, suggesting a general role for Numb in the endocytic process.

Human immunodeficiency virus transactivator protein (Tat) stimulates chemotaxis, calcium mobilization, and activation of human polymorphonuclear leukocytes: implications for Tat-mediated pathogenesis.

The extracellular activities of the human immunodeficiency virus (HIV) transactivator protein (Tat) include induction of angiogenesis and stimulation of monocyte migration. Here it is shown that polymorphonuclear leukocytes (PMNL), mostly neutrophils, rapidly invade in response to Tat in vivo and initiate the formation of new vessels. In vitro, Tat was chemotactic for PMNL and induced calcium (Ca(2+)) mobilization. Tat proteins with inactivating substitutions in the arginine-glycine-aspartic acid or basic domain were still active in inducing PMNL migration, whereas Tat peptides mapped the migration and Ca(2+) mobilization activity to a cysteine-rich core domain, previously described as a Tat "chemokine-like" region (peptide CysL(24-51)). Tat and the CysL(24-51) peptide also induced PMNL superoxide production and the release of the angiogenic factors interleukin-8 and vascular endothelial growth factor from PMNL. CysL(24-51) did not induce endothelial cell migration but was angiogenic in vivo. These data indicate that the Tat activity on PMNL is mediated by its chemokine-like region and that PMNL recruitment by Tat is linked to angiogenesis.

Oa1 knock-out: new insights on the pathogenesis of ocular albinism type 1.

Ocular albinism type I (OA1) is an X-linked disorder characterized by severe reduction of visual acuity, strabismus, photophobia and nystagmus. Ophthalmologic examination reveals hypopigmentation of the retina, foveal hypoplasia and iris translucency. Microscopic examination of both retinal pigment epithelium (RPE) and skin melanocytes shows the presence of large pigment granules called giant melanosomes or macromelanosomes. In this study, we have generated and characterized Oa1-deficient mice by gene targeting (KO). The KO males are viable, fertile and phenotypically indistinguishable from the wild-type littermates. Ophthalmologic examination shows hypopigmentation of the ocular fundus in mutant animals compared with wild-type. Analysis of the retinofugal pathway reveals a reduction in the size of the uncrossed pathway, demonstrating a misrouting of the optic fibres at the chiasm, as observed in OA1 patients. Microscopic examination of the RPE shows the presence of giant melanosomes comparable with those described in OA1 patients. Ultrastructural analysis of the RPE cells, suggests that the giant melanosomes may form by abnormal growth of single melanosomes, rather than the fusion of several, shedding light on the pathogenesis of ocular albinism.

Tyrosine phosphorylation of Eps15 is required for ligand-regulated, but not constitutive, endocytosis.

Membrane receptors are internalized either constitutively or upon ligand engagement. Whereas there is evidence for differential regulation of the two processes, little is known about the molecular machinery involved. Previous studies have shown that an unidentified kinase substrate is required for endocytosis of the epidermal growth factor receptor (EGFR), the prototypical ligand-inducible receptor, but not of the transferrin receptor (TfR), the prototypical constitutively internalized receptor. Eps15, an endocytic protein that is tyrosine phosphorylated by EGFR, is a candidate for such a function. Here, we show that tyrosine phosphorylation of Eps15 is necessary for internalization of the EGFR, but not of the TfR. We mapped Tyr 850 as the major in vivo tyrosine phosphorylation site of Eps15. A phosphorylation-negative mutant of Eps15 acted as a dominant negative on the internalization of the EGFR, but not of the TfR. A phosphopeptide, corresponding to the phosphorylated sequence of Eps15, inhibited EGFR endocytosis, suggesting that phosphotyrosine in Eps15 serves as a docking site for a phosphotyrosine binding protein. Thus, tyrosine phosphorylation of Eps15 represents the first molecular determinant, other than those contained in the receptors themselves, which is involved in the differential regulation of constitutive vs. regulated endocytosis.

Molecules that inhibit T-cell functions: cytochemical localization and shuttling.

Adaptive immune responses to antigens are mediated by specific receptors expressed on B cells (BCR's) and T cells (TCR's). Effector cells and memory cells are produced following a proliferative wave that accounts for clonal expansion. If not down-regulated, clonal expansion might lead to uncontrolled lymphoproliferation that would be harmful for the organism. Several mechanisms that account for the down-sizing of activated lymphocyte clones are briefly reviewed here. We next consider in detail one such mechanism that deals with the functional characterization and the immunocytochemical localization of two T-cell inhibitory molecules, namely the Cytotoxic T Lymphocyte Antigen-4 (CTLA-4) and the HP-F1 antigen, both present in all T lymphocytes. CTLA-4 and HP-F1 inhibit CD4+ T-helper cell proliferation and the lytic ability of CD8+ T-cytotoxic cells in non-specific and in antigen-specific cytolytic assays. Interestingly, a clonal distribution exists as for the ability of CTLA-4 and HP-F1 to inhibit T-cell functions. In resting and activated T cells, both molecules are largely confined in the endosomal compartment, as shown by immunofluorescence analyses. However, upon interaction of T cells with Antigen-Presenting Cells (APC's) or with target cells that must be killed, CTLA-4 molecules are transported to the plasma membrane, at the site of cell-to-cell contact where, following interaction with ligands, they trigger inhibitory signals.

Inhibition of angiogenesis and vascular tumor growth by interferon-producing cells: A gene therapy approach.

We developed an in vivo gene therapy approach to characterize and optimize the anti-angiogenic activity of class I interferons (IFNs), using packaging cell lines producing an amphotropic LXSN-based retrovirus expressing either IFN-alpha1 (alpha1Am12), IFN-beta (betaAm12) murine cDNAs, or the vector alone (neoAm12). Pretreatment of endothelial-like Eahy926 cells in vitro with conditioned media (CM) from alpha1Am12 or betaAm12 cells for 48 hours significantly inhibited their migration and invasion as compared to neoAm12-CM-treated cells. betaAm12-CM also inhibited the formation of capillary-like structures on Matrigel by EAhy926 cells. In vivo, inclusion of the betaAm12 cells strongly inhibited, and alpha1Am12 partially inhibited, the angiogenic response in the Matrigel sponge model in both immune-competent and athymic nude mice. Electron microscopy showed a reduction of host cell infiltration in alpha1Am12- and betaAm12-containing sponges and reduction of invading tubular clefts of host cells as compared to controls. Finally, inoculation of either alpha1Am12 or betaAm12 cells (10%) along with a highly angiogenic Kaposi's sarcoma cell line (90%) resulted in a powerful reduction of tumor growth in nude mice in vivo, as did infection with the interferon-alpha-producing retroviruses. These data suggest that a gene therapy approach using class I interferons can effectively inhibit tumor angiogenesis and growth of vascular tumors.

Anatomy of the immune system: facts and problems.

In the introductory section of this report, the anatomy of the immune system, from organs and tissues to molecules, will be reviewed briefly. Cell proliferation and differentiation in the central lymphoid organs (thymus and bone marrow) yield a repertoire of T- and B-cell clones that seed into peripheral lymphoid organs (spleen, lymph nodes and Mucosa-Associated Lymphoid Tissue, MALT), where humoral and cell-mediated antigen-specific immune responses occur. The stringent process of clonal selection in the central lymphoid organs implies deletion of inappropriate cells via apoptosis. In the peripheral lymphoid organs, the potential of unlimited activation and expansion of lymphocytes in response to antigens is primarily regulated by apoptosis and anergy. These events, on the one hand, are relevant to prevent autoimmunity and lymphoproliferative disorders; on the other hand, clonal deletion and anergy provide a detrimental escape to immune recognition of malignant cells. Two major inhibitory mechanisms of the immune response have emerged recently. One is linked to the existence of bona fide suppressor cells and cytokines; the other relies on the existence of inhibitory molecules expressed by T, B and NK cells, as well as by other leukocytes. In the studies herein reported, emphasis will be given to surface membrane molecules that down-regulate T-cell-mediated immune responses. These molecules control interactions between T cells and antigen presenting cells (APC's) or target (virus-infected or mutated) cells that have to be killed. Two sets of molecules exist that either upregulate (coactivation molecules) or down-regulate (inhibitory molecules) T-cell mediated responses. The latter aspect of the immune regulation, i.e. molecules that limit the expansion of T-cell clones following specific recognition of antigens will be considered in depth. Two inhibitory molecules, CD152 (CTLA-4) and CD85/LIR-1/ILT2 are expressed in all T cells, being largely confined within intracellular compartments of these lymphocytes when they are in a resting state, but ready to be shuttled to and from the plasma membrane when cells are activated following encounter with antigen. Membrane expression of the two inhibitory molecules is transient and is regulated by an internalization process directed to endosomal compartments and to receptor degradation and/or recycling. CTLA-4 and CD85/LIR-1/ILT2 play a pivotal role in T-cell homeostasis that follows any cell-mediated immune response; their localization and functional role will be thoroughly analyzed. In the last part of this study a major question will be faced, i.e. is the containment of the possibly unlimited expansion of the immune system due to a blockade of the cell cycle? Or, else, could be apoptosis the sole mechanism responsible? Experimental data in support of the latter contention will be provided.

Ultrastructural and functional studies of the interaction between IL-12 and IL-2 for the generation of lymphokine-activated killer cells.

IL-12 promotes generation of LAK activity in short-term-cultured NK cells, but information on the structure and function of IL-12-induced LAK cells is not yet available. The latter issues have been here investigated with emphasis on interactions between IL-12 and IL-2. Peripheral blood mononuclear cells (MNC) exposed to IL-12 for 5-7 days displayed a decrease in the amount and density of the matrix of large granular lymphocyte (LGL)-associated granules. In cells cultured with IL-12 and IL-2 for 5-7 days, empty vacuoles were predominant and the electron-dense matrix was scanty. In MNC incubated with IL-2 for 5-7 days, most granules were loaded with electron-dense matrix. IL-12 and IL-2 displayed an additive effect on LAK cell cytotoxicity until approximately 48 h in culture which was followed by a sharp decline. Immunocytochemical and biochemical studies demonstrated that MNC cultured for 5-7 days with IL-12 and IL-2 displayed downregulated perforin expression and upregulated granzyme B expression. Fas ligand expression was virtually undetectable in MNC cultured for 5-7 days with or without cytokines. It appears that perforin downregulation plays a major role in the reduced cytotoxicity of MNC cultured with IL-12 and IL-2 for 5-7 days.

Ocular albinism: evidence for a defect in an intracellular signal transduction system.

G protein-coupled receptors (GPCRs) participate in the most common signal transduction system at the plasma membrane. The wide distribution of heterotrimeric G proteins in the internal membranes suggests that a similar signalling mechanism might also be used at intracellular locations. We provide here structural evidence that the protein product of the ocular albinism type 1 gene (OA1), a pigment cell-specific integral membrane glycoprotein, represents a novel member of the GPCR superfamily and demonstrate that it binds heterotrimeric G proteins. Moreover, we show that OA1 is not found at the plasma membrane, being instead targeted to specialized intracellular organelles, the melanosomes. Our data suggest that OA1 represents the first example of an exclusively intracellular GPCR and support the hypothesis that GPCR-mediated signal transduction systems also operate at the internal membranes in mammalian cells.

Expression of the extracellular fatty acid binding protein (Ex-FABP) during muscle fiber formation in vivo and in vitro.

We report that Ex-FABP, an extracellular protein belonging to the lipocalin family and involved in the extracellular transport of long-chain fatty acids, is expressed in the forming myotubes both in vivo and in vitro. The presence of the protein and of the mRNA was observed in newly formed myotubes at early stages of chick embryo development by immunohistochemistry and by in situ hybridization. At later stages of development myofibers still expressed both the mRNA and the protein. Ex-FABP expression was observed also in the developing myocardium and the muscular layer of large blood vessels. In agreement with these findings, an initial expression of the mRNA and protein secretion by cultured chicken myoblasts were observed only after the onset of myoblast fusion. Double-immunofluorescence staining of these cultured cells revealed that multinucleate myotubes were stained by antibodies directed against both the Ex-FABP and the sarcomeric myosin, whereas immature myotubes and single myoblasts were not. When added to cultured myoblasts, antibodies against the Ex-FABP induced a strong enhancement of the production of the same protein. In all experiments some cell sufferance and a transient impairment of myotube formation were also observed. The finding that the continuous removal of the Ex-FABP from the culture medium of myoblasts, due to the formation of immune complexes, resulted in an overproduction of the protein suggests a feedback (autocrine) control during myotube differentiation and maturation. We propose that the requirement for increased transport and metabolism of free fatty acid released from the membrane phospholipids and storage lipids, mediated by Ex-FABP, may be essential during differentiation of multinucleated myotubes or that an increased local demand of fatty acids and metabolites may act as a local hormone in tissues differentiating and undergoing morphogenesis.

Monocyte-derived dendritic cells and monocytes migrate to HIV-Tat RGD and basic peptides.

OBJECTIVE AND DESIGN: Extracellular Tat released from HIV-1-infected cells is a mitogenic and motogenic factor for endothelial and Kaposi's sarcoma (KS)-derived cells and is angiogenic in vivo. Here we show for the first time that Tat induces migration of human dendritic cells in a concentration-dependent manner and that the Arg-Gly-Asp (RGD) and basic Tat peptides contribute to dendritic and monocyte cell migration. In vivo, Tat stimulates invasion of macrophages into a matrigel sponge. METHODS: Monocyte and dendritic cell chemotaxis was assessed using the Boyden chamber assay. RESULTS: Tat induced migration of monocyte-derived dendritic cells at the same levels as the N-formyl-Met-Leu-Phe peptide, and of monocytes at levels comparable to RANTES. Peptide mapping of the chemotactic activity of Tat showed that the RGD domain, which has been shown to support integrin-mediated cell migration, and the basic domain which binds and activates the tyrosine kinase receptor KDR on endothelial cells, both had activity. Antibody-blocking experiments indicate that responses to the RGD domain was inhibited by beta1 and alpha vbeta3 integrin blocking antibodies. Combination of the Tat RGD and basic peptides did not show additive effects; however, Tat co-operated with macrophage-chemotactic protein or RANTES in inducing monocyte migration. CONCLUSIONS: Our results show that Tat can act as a chemoattractant for dendritic cells, and that both the RGD and basic domains are involved in this response. These same domains attract monocytes. The alpha vbeta3 and beta1 integrins are equally involved in Tat-induced monocyte migration, while the alpha vbeta3 integrin largely mediates the dendritic cell response to Tat.

Regulated expression of fibronectin, laminin and related integrin receptors during the early chondrocyte differentiation.

We have investigated the expression and localization of fibronectin, laminin, and their receptors, and we used an in vitro chick chondrocyte differentiation model to define a time hierarchy for their appearance in early chondro-genesis and to determine their role in the cell condensation process. By serum fibronectin depletion/reconstitution, or GRGDSP peptide competition experiments, we show that fibronectin contributes to the initial cell-cell interactions that occur during condensation. In later stages, a down-regulation of both fibronectin and of its alpha5beta1 integrin receptor occur, as demonstrated by mRNA and protein kinetics. Immunolocalisation studies suggest that the reduction of fibronectin in discrete areas is involved in local activation of the cell differentiation program. Furthermore, we show that laminin is expressed during the in vitro cell condensation process in areas that are negative for fibronectin staining. The types of laminin as well as the timing of expression have been determined by northern blot and RT-PCR analyses. The highest levels of expression are coincident with maximal cell aggregation. The alpha3beta1 laminin receptor, highly expressed in dedifferentiated cells, follows later on the ligand trend. During in vitro chondrogenesis, a down-regulation in the B isoform, and an up-regulation of the A isoform, of the alpha subunit of the alpha6beta1 laminin receptor occurs. Immunolocalisation studies suggest that laminin is involved in the definition of differentiating areas as opposed to non differentiating areas of the condensed region, i.e. the periphery, which eventually gives rise to the perichondrium.

HIV is trapped and masked in the cytoplasm of lymph node follicular dendritic cells.

To gain further insight into the pathogenesis of human immunodeficiency virus (HIV) infection, lymph nodes from seven asymptomatic HIV+ subjects were analyzed during the latent phase of disease. Both ultrastructural and immunohistochemical analyses revealed that, in all of the cases, plasma cells producing IgM/gamma were present in germinal centers. Secreted immunoglobulins formed extracellular deposits mimicking the follicular dendritic cell network. Immunoglobulin produced by germinal center plasma cells are specific for HIV because they bind the HIV env protein gp 120. Plasma cells producing antibodies with the same specificity were also abundant in the extrafollicular regions of lymph nodes. During the latent phase of infection, the virus largely accumulates within the germinal centers. Therefore, extracellular immunoglobulin may form immune complexes, as shown by the presence of HIV-specific antibodies, HIV particles, and complement components C3c, C3d, and C1q in the interdendritic spaces. When the ultrastructural localization of HIV in germinal centers was analyzed, abundant virus particles were found in the interdendritic spaces. In addition to this extracellular localization of HIV, receptor-mediated endocytosis of viral particles by follicular dendritic cells was observed. Complete HIV particles were found within the endosomal compartment of the follicular dendritic cells and, as complete viral particles, free in the cytoplasm, indicating that the virus may escape from the endocytic compartment. As the virus is abundant in the cytoplasm, this event leads to formation of a hidden reservoir within follicular dendritic cells. In this location, HIV escapes recognition by cytotoxic T lymphocytes. In contrast, virus budding indicating a productive infection of follicular dendritic cells that would render them susceptible to T-cell-mediated lysis has been seldom observed.