Abstract
Besides a structural role in tissue architecture, fibroblasts have been shown to regulate the proliferation and differentiation of other neighboring specialized cell types, but differently according to the anatomic site and pathologic status of their tissue of origin. In this study we report a novel regulatory function of human spleen-derived fibroblasts in the development of NK cells from adult resting blood progenitors. When CD34+ cells were cocultured with spleen-derived fibroblasts in monolayers, nonadherent CD56+CD3− NK cells were predominantly produced after 2–3 wk of culture in the absence of exogenous cytokines. Most NK cells expressed class I-recognizing CD94 and NK p46, p44, and p30 receptors as well as perforin and granzyme lytic granules. Moreover, these cells demonstrated spontaneous killing activity. Cell surface immunophenotyping of spleen-derived fibroblasts revealed a low and consistent expression of IL-15, Flt3 ligand, and c-kit ligand. Additionally, low picogram amounts of the three cytokines were produced extracellularly. Neutralizing Abs to IL-15, but not the other two ligands, blocked NK cell development. Additionally, suppressing direct contacts of CD34+ progenitors and fibroblasts by microporous membrane abrogated NK cell production. We conclude that stromal fibroblasts within the human spleen are involved via constitutive cell surface expression of bioactive IL-15 in the development of functional activated NK cells under physiologic conditions.
Human spleen, the body’s largest lymphoid organ, is made up of two components: 1) the white pulp consisting of T cell-rich periarteriolar lymphoid sheaths and B cell-rich follicles, and 2) the red pulp composed of endothelially lined sinusoids, and a voluminous extravascular compartment supported by a meshwork of densely packed reticular fibers, fibroblasts, macrophages, and cytotoxic T and NK cells, in which countless blood cells are free floating, and numerous platelets are stored (1). The capacity of the spleen for plasma cell formation upon splenic B cell activation and the easy access of the various cellular components to circulating blood facilitate the two major functions of the spleen: immune responses to blood-borne Ags and elimination of unwanted materials and cells from the circulation.
The spleen is not a site of hemopoiesis, even though under pathologic situations, notably certain myeloproliferative disorders, production of hemopoietic cells can be demonstrated, probably due to the proliferation of displaced bone marrow precursors stored in the spleen (2). Recently, we have shown that stromal fibroblasts established in culture from the spleen of patients with such an abnormal myeloid activity, directly contributed to myeloid proliferation and differentiation of patients’ blood hemopoietic progenitors (3). This raised the possibility that regulatory effects of splenic fibroblasts on blood progenitors might similarly occur under physiologic conditions. We thus investigated the influence of fibroblasts isolated from human adult spleen on the proliferation and differentiation of resting blood CD34+ progenitors. Given the broad repertoire of adhesion molecules, surface ligands, and cytokines/chemokines that human fibroblasts have been shown to express and produce (4, 5, 6, 7), we also analyzed the phenotypic characteristics of spleen-derived fibroblasts with respect to their effects on hemopoietic cells.
Our study provides the first evidence that stromal fibroblasts from human adult spleen can regulate NK cell development, and that constitutive expression of IL-15 on their cell surface plays a functional role in the process of NK cell genesis.
Materials and Methods
Isolation of fibroblasts from human spleen tissue
Normal adult spleen samples (n = 4) were obtained from both male and female donors, aged 49–70 years, from whom consent for multiple organ donation had been granted (Etablissement Français des Greffes, Kremlin-Bicêtre, France). Fibroblasts were isolated by gentle grinding of splenic tissue in a sterile glass homogenizer. Tissue homogenates were then centrifuged at 1000 rpm for 5 min. Cellular aggregates present in the pellet were washed three times with PBS and then transferred to culture flasks in a small volume of RPMI 1640 medium (Life Technologies, Cergy-Pontoise, France) supplemented with 15% FCS (Life Technologies) to adhere. The medium was changed every 3–4 days until a substantial outgrowth of cells was observed, at which point cells were trypsinized (Life Technologies) and subsequently subcultured at a 1:2 split ratio for up to three passages before analysis. The fibroblastic nature of the cultures was assessed by their homogeneous staining for anti-smooth muscle-actin with a typical stress fiber distribution characteristic of myofibroblasts (8). All cells also showed strong staining for other mesenchymal Ags, including vimentin, fibronectin, and collagen type I and III. Homogeneous cell surface expression of Ag ASO2, specific for fibroblasts (9), was also detected by flow cytometry, whereas Ag CD31, characteristic of endothelial cells, was missing.
Isolation of human CD34+ cells from adult resting peripheral blood
Mononuclear cells obtained from the peripheral blood of healthy donors (n = 9) were prepared by centrifugation over a Ficoll solution (specific gravity, 1.077; Pharmacia Biotech, Orsay, France). After overnight plastic adherence in culture medium, the nonadherent mononuclear CD34+ cell fraction was isolated using a supermagnetic microbead selection kit and varioMACS columns according to the manufacturer’s instructions (Miltenyi Biotec, Paris, France). Purity was >90%.
Culture of CD34+ cells with human spleen-derived fibroblasts
Fibroblasts were grown to confluence in 60-mm petri dishes (Falcon; BD Biosciences, Le Pont de Claix, France) or six-well and 0.4-μm pore size Transwell plates (PolyLabo, Strasbourg, France) and incubated for 3 wk with CD34+ cells (3:10 ratio with fibroblasts) in standard culture medium supplemented with 10% FCS; no exogenous cytokines were added. Half-medium changes were performed once a week. In some experiments cultures were also treated with various combinations of neutralizing Abs as follows. Anti-stem cell factor (anti-SCF3
Cell immunophenotyping
Cells were stained with various combinations of mAbs as listed in Table I⇓ according to conventional direct and indirect techniques (6). In some experiments cell membranes were permeabilized by treatment with ORTHOPermeafix (Ortho Diagnostic Systems, Roissy, France) before Ab incubation for monitoring total (surface and cytoplasmic) Ag expression. Unspecific staining was determined using identical nonreactive isotype mAbs. Cells were analyzed on a FACScan (BD Biosciences, Franklin Lakes, NJ) using CellQuest (BD Biosciences) and WinMDI software programs (Scripps Research Institute, La Jolla, CA).
Abs used in this study
Cytokine production
Multiple aliquots of cell-free supernatants were collected from fibroblasts cultured alone or with hemopoietic CD34+
Confocal microscopic analysis
For double staining, cells were first permeabilized with ORTHOPermeafix for 45 min at room temperature. Cells were then stained with anti-perforin mAb δG9 (BD PharMingen, Le Pont de Claix, France), followed by incubation with Alexa Fluor 594-conjugated goat anti-mouse Ab (Molecular Probes, Montluçon, France). Cells were then labeled with biotinylated anti-granzyme B mAb CLB-GB11 (Tebu, Le Perray en Yvelines, France), followed by streptavidin-Alexa Fluor 488 conjugate (Molecular Probes). Cytospin-stained cells were analyzed by laser scanning confocal microscopy on a Leica TCS NT/SP interactive laser cytometer equipped with confocal optics (Leica Microsystems, Wetzlar, Germany).
Cytotoxicity assay
The cytolytic activity of total nonadherent cultured cells was determined in a standard 4-h 51Cr release assay against the NK-sensitive K562 cell line, a gift from Dr. S. Chouaib (IGR, Villejuif, France). In some experiments cells were additionally incubated with 1000 U recombinant human IL-2 (Hoechst-Marion-Roussel, Romainville, France) for 48 h and then tested for cytotoxic activity.
Results
Characteristics of CD34+ progenitors isolated from resting blood
The purified CD34+ cell populations isolated from adult nonmobilized peripheral blood cells showed no evidence of myeloid or NK and B lymphoid differentiated cells, with generally <1% of the population expressing CD14, CD15, CD56, and CD19 (data not shown). In contrast, cells expressing lymphoid CD2 (19.0 ± 9.6%; range, 3–28%) and megakaryocyte CD41 (46.4 ± 26.9%; range, 22–89%) lineage cell Ags were detected, indicating that CD34+ progenitors in resting blood were most likely committed cells (data not shown). Further phenotype characterization also indicated that significant (44–76%) proportions of CD34+ cells expressed a set of surface adhesion molecules necessary for blood cell interactions, including ICAM-1, VLA-4, and LFA-1. Cell surface expression of c-kit (23%) and Flt3 (12%) was detected in a fraction of cells (data not shown).
Differentiation of resting blood CD34+ progenitors cultured with human spleen-derived fibroblasts
When purified CD34+ cells were cultured with splenic fibroblasts for 3 wk, a decline in nonadherent cell number was generally observed by wk 1, followed by a low cell expansion (Table II⇓). Phenotype characterization of nonadherent surviving cells revealed a progressive lost of cells positive for CD34, CD41, and CD2. In contrast, increasing percentages of CD56+ and, to a much lesser extent, CD15+ cells were detected. No cells positive for CD3, CD19, and CD14 could be identified, indicating the lack of T or B cell and monocyte development (data not shown). However, a fraction (∼40%) of the nonadherent cell population remained consistently negative for each of the Ags tested, indicating that other lineages might also be present.
Phenotype of nonadherent cells generated from peripheral resting blood CD34+ progenitors when cultured with human spleen-derived fibroblastsa
Phenotype and lytic activity of CD56+ cells produced with splenic fibroblasts
The production of CD56+ cells that was reproducibly obtained in a series of experiments using fibroblasts and blood CD34+ cells from different donors generally reached maximal values of 33.5 ± 6.4% by wk 3 (Fig. 1⇓a). Further analysis demonstrated that these CD56+ cells displayed a typical NK phenotype that closely resembled cells produced in stroma-free medium supplemented with IL-15, Flt3L, and c-kit ligand (c-kitL) (14, 15, 16). Thus, dual flow cytometry showed that all CD56+ cells lacked CD3, and only a small fraction of them expressed CD2 (<5%) or CD16 (∼10%; Fig. 1⇓b). Furthermore, a high (>80%) percentage of CD56+ cells coexpressed the C-type lectin receptor CD94 and the NK receptors p46 (>70%), p44 (>60%), or p30 (∼50%). The surface expression levels of these NK Ags were consistently high (mean fluorescent intensity between logs 2 and 3) in a majority of cells (Fig. 1⇓c). Other specific killer cell IgG-like receptors, including p58-1, p58-2, p70, and leukocyte Ig-like receptor-1, were lacking or expressed in only a small (<5%) percentage of cells (data not shown). Further confocal microcopic examination of NK cells demonstrated in a high proportion of nonadherent cells (Fig. 2⇓) an intracytoplasmic colocalization of perforin and granzyme B, which are thought to be lytic mediators for lymphocyte cytotoxicity (17). Expression of Fas ligand, another mediator responsible for the lytic NK cell function (18), was not detected by flow cytometry (data not shown). When total nonadherent cells were directly tested for their ability to lyse the classical class I-negative target K562, a substantial (>35%) level of lytic activity was detected at an E:T cell ratio of 5:1 (Fig. 1⇓d). Further stimulation by IL-2 did not increase cell cytotoxicity.
Characterization of NK cells generated with human spleen-derived fibroblasts. a, Kinetics of NK cell development monitored by the percentage of nonadherent CD56+ cells harvested at the indicated times. The results represent the mean ± SEM of five separate experiments. b, Percentage of CD56+ cells double-stained with the indicated labeled mAbs and analyzed by flow cytometry. c, CD56 and NKR expression levels. d, Percentage of cell lysis with unstimulated (▪) and IL-2-stimulated (□) nonadherent cells at a 5:1 E:T cell ratio. For b–d, results are representative of at least three separate experiments
Confocal microscopic analysis of intracytoplasmic localization of perforin and granzyme B in NK cells generated from blood CD34+ cells cocultured with splenic fibroblasts. a, Background staining. b, The yellow staining indicates intracellular colocalization of perforin (red) and granzyme B (green).
Expression of c-kitL, Flt-3L, and IL-15 in human spleen-derived fibroblasts
Low amounts of the two ligands c-kitL/SCF and Flt3L, related to NK progenitor amplification and differentiation (14, 15), were detected in culture supernatants of spleen-derived fibroblasts as early as day 4, with optimal values of 93.9 ± 24.3 and 19.0 ± 5.2 pg/ml, respectively, by wk 1 (Fig. 3⇓). In contrast, IL-15, the most potent cytokine for NK cell differentiation (14), was barely detectable, below or at the limit of ELISA detection and accuracy (3.9 pg/ml) during the first 4 days of culture and then reached an optimal value of 17.7 ± 0.4 pg/ml by wk 1. Other cytokines, including IL-2, TNF-α, and IFN-γ, consistently remained undetectable (data not shown). Except for SCF, no significant changes in secreted cytokines were observed under coculture conditions.
Cytokine production in human spleen-derived fibroblasts. Supernatants from splenic fibroblasts cultured alone (□) or with CD34+ cells (▪) were collected at the indicated times and assayed for cytokines. The limit (dotted line) of cytokine detection and accuracy by ELISA is shown. Results are the mean ± SEM of four separate experiments.
Low and consistent surface expression of c-kitL and Flt3L was monitored on splenic fibroblasts by flow cytometry (Fig. 4⇓a). Unexpectedly, significant surface IL-15 expression was also observed. Similarly to c-kitL and Flt3L, high levels of IL-15 expression were detected after permeabilization of the cell membrane, indicating that splenic fibroblasts constitutively expressed IL-15 both on the cell surface and intracellularly. Splenic fibroblasts express IL-15R α-chain. To determine whether the expression of IL-15 on the cell surface was membrane-bound or secreted bound IL-15, cells were treated with neutralizing anti-IL-15 from the start of culture to day 3, when IL-15 was barely detectable in culture supernatants. No surface IL-15 expression was revealed in treated cells (data not shown), indicating that the IL-15 expressed on the surface of fibroblasts was secreted, rather than unsecreted, bound IL-15. After fixation, IL-15/IL-15Rα interaction appeared stable because treatment with anti-IL-15Rα mAb did not block bound IL-15 (Fig. 4⇓b).
Expression of c-kitL, Flt3L, and IL-15 in human spleen-derived fibroblasts. a, Nonpermeabilized (non perm.) and permeabilized (perm.) cells were stained with the indicated labeled mAbs (thick histograms) or with isotype-matched control Abs (thin histograms) after 4 days of culture. b, Cells were incubated with the anti-IL-15Rα mAb M162 (2 μg/ml) or with an isotype-matched control Ab (IgG1) for 24 h, washed, then stained with anti-IL-15 mAb (thick histograms) or control Ab (thin histograms).
Role of fibroblast-derived IL-15 in the production of NK cells
To evaluate the importance of the three cytokines, IL-15, Flt3L, and SCF, in the NK differentiation process stimulated by spleen-derived fibroblasts, neutralizing mAbs directed at each cytokine were added to the cultures for 3 wk, and cells with the NK phenotype were monitored on the basis of CD56 expression. Under conditions using a combination of the three neutralizing mAbs, no CD56+ cells could be detected (Fig. 5⇓a). However, when used individually, neutralizing anti-IL-15 mAb was found to block CD56+ cell development, whereas incubation with anti-SCF or anti-Flt3L mAbs did not.
Implication of IL-15 in NK cell production with splenic fibroblasts. a, Percentage of CD56+ cells cultured with neutralizing Abs to IL-15, Flt3L, and SCF for 3 wk. b, The number of nonadherent viable cells and percentage of CD56+ cells recovered from Transwells after 3 wk. Results are representative of two separate experiments.
To determine whether the IL-15 exposed on the surface of splenic fibroblasts and that produced extracellularly were both implicated in NK cell development, cocultures of fibroblasts were performed in Transwell plates without cell contacts with CD34+ cells. Similar numbers of nonadherent viable cells were monitored after 3 wk in cultures with and without cell-to-cell contacts (Fig. 5⇑b). Despite the lack of CD34+ cells (data not shown), no CD56+ cells were recovered from the inserts, whereas a fraction (17%) of highly expressing CD56+ cells was detected under cell-to-cell contact standard culture conditions. Moreover, when exogenous rIL-15 was added to CD34+ cells at 50 ng/ml, a concentration shown to elicit NK cell differentiation in stroma-free cultures (14), CD56+ cells were produced at levels comparable to those of cell-to-cell contact standard culture conditions (data not shown). All these results indicated that contact with fibroblasts and bound IL-15 interaction were necessary and sufficient for CD34+ progenitors to differentiate along the NK cell lineage.
Discussion
Fibroblast is a generic term for a family of stromal cells whose exclusive function has long been thought to produce substantial quantities of extracellular matrix to support other more specialized cell types. However, experimental evidence has accumulateddemonstrating that fibroblasts do not act as mere supportive cells but are also able to drive differentiation of other neighboring cell types under both physiologic and pathologic conditions (19, 20). Especially in the human hemopoietic system (3, 21) it has been shown that homogeneous stromal fibroblast populations established in culture from bone marrow were able to directly contribute to the formation of blood cells from all lineages except T/NK lymphocytes, with a majority of granulomonocytes. The results presented in this work clearly show that fibroblasts isolated from human spleen also regulate blood cell differentiation. However, unlike bone marrow-derived fibroblasts, those isolated in culture from the spleen predominantly regulated lymphoid NK cell differentiation from resting blood CD34+ progenitors, further emphasizing the specialized functions of stromal fibroblasts in different tissues. Because fibroblasts displayed a homogeneous myofibroblast phenotype, and no hemopoietic cells appeared in control cultures maintained for extended periods without subcultures, it is most likely that the NK cells produced with spleen-derived fibroblasts did not arise from splenic cellular components. Furthermore, contaminating NK cells were generally not detected in the initial CD34+ cell populations cocultured with fibroblasts.
Most fibroblast-induced NK cells showed CD56 and NK receptor expression closely related to circulating blood-derived NK subsets or cultured cells produced in stroma-free medium supplemented with cytokines (14, 15, 16). Thus, they highly expressed the CD56 NK lineage Ag, the C-type lectin CD94 receptor specific for MHC class I recognition, and other newly identified NK activator receptors, p46, p44, and p30, specific for non-MHC ligands (22). Additionally, expression of perforin and granzyme lytic granules was observed intracellularly. Thus, these NK cells displayed lytic potentialities, as suggested by the expression of killer cell-activating receptors and cytoplasmic lytic organelles. Indeed, these CD56+ cells were functionally activated cells, as evidenced by their ability to kill the classical class I-negative target K562 without additional cytokine activation.
In contrast to other human and murine stroma cultures composed of heterogeneous stromal cell populations and using multiple exogenous cytokines (23, 24), spleen-derived fibroblasts were capable, by themselves, of stimulating the development of activated NK cells from blood CD34+ progenitors, indicating that they provided the appropriate microenvironment for NK cell genesis. In this context, IL-15 is now recognized as one of the major factors triggering NK cell differentiation, whereas c-kitL and Flt3L can promote IL-15-mediated expansion and differentiation of NK progenitor cells in stroma-free cultures (25) In human spleen-derived fibroblasts, we detected the three cytokines both exposed on the cell surface and produced extracellularly.
The constitutive expression of IL-15 on the surface of human spleen-derived fibroblasts was quite striking because stromal fibroblasts isolated from other tissue sources did not show such expression unless stimulated by inflammatory cytokines or derived from pathologic lesions (3, 26). In contrast, blood-derived cells, including monocytes, monoblastoid cells, and leukemic progenitors, have been reported to constitutively express membrane-associated IL-15 without any IL-15 secretion (27, 28, 29). This membrane form has been proposed to be the bioactive form of IL-15 (27) and the result of IL-15 mobilization from intracellular stores to the membrane (30). Unlike resting and stimulated cells, secreted rather than unsecreted bound IL-15 was detected on the surface of splenic fibroblasts. Even though expressed at low levels, the membrane-bound IL-15 appeared to play a predominant role over the surface expression of the other two ligands and extracellular cytokine production in the process of NK cell development reported here. First, NK cell production was blocked only when blood CD34+ progenitors and splenic fibroblasts were cocultured in the presence of neutralizing IL-15, but not c-kitL and Flt3L, mAbs. Second, culture conditions that left the cytokine microenvironment intact but prevented direct contacts between the two cell populations totally abrogated NK cell production. In stroma-free cultures it has been shown that nanomolar concentrations of exogenous IL-15 were necessary to activate NK cell proliferation and differentiation, whereas picomolar amounts were effective at maintaining NK cell survival. Indeed, we found that 50 ng/ml exogenous rIL-15 induced NK cell differentiation of resting blood CD34+ cells as efficiently as surface IL-15 expressed on splenic fibroblasts. Moreover, elevated concentrations of c-kitL in synergy with Flt3L have been found to promote NK progenitor expansion. Therefore, the picogram amounts of all three cytokines that were produced extracellularly by human spleen-derived fibroblasts were most likely efficient at ensuring cell survival, but not proliferation, and might explain why the yields of NK cells resulting from splenic fibroblasts were reduced compared with those from stroma-free cultures (16). It should be emphasized that IL-2, the cytokine initially used for NK cell development (23), and the two classical cytokines, TNF-α and IFN-γ, known to elicit lymphocyte activation, consistently remained undetectable in culture supernatants under both standard and coculture conditions.
In conclusion, our data provide the first evidence that human spleen-derived fibroblasts directly contribute to NK cell development via exposure of bioactive cell surface IL-15, strengthening the importance of stromal fibroblasts as sentinel cells in the immune system (31). It is reasonable to speculate that IL-15 exposed on the surface of fibroblasts serves as a physiologic regulator in maintaining a small number of functionally activated NK cells within the spleen.
Acknowledgments
We are indebted to Drs. A. Moretta and L. Moretta (University of Genova, Genova, Italy) for providing us the various anti-NK mAbs, Dr. M. A. van Dijk (University Medical Center, Utrecht, The Netherlands) for his gift of anti-IL-15Rα mAb M162, and the staff of Centre Transfusion Sanguine, Hôpital Paul Brousse (Villejuif, France), and of Coordination Hospitalière des Activités de Prélèvements, Hôpital Kremlin-Bicêtre, for supplying us with normal human blood and spleen tissue samples. We also thank Corine Pottin-Clémenceau for her skillful technical assistance and Julien Giron-Michel for his guidance in the confocal analysis.
Footnotes
-
↵1 This work was supported by grants from the Association de Recherche contre le Cancer (no. 5233) and the Association Nouvelle Recherche Biomédicale (Villejuif, France). D.B. is the recipient of a fellowship from the Association de Recherche contre le Cancer.
-
↵2 Address correspondence and reprint requests to Dr. Danièle Brouty-Boyé, Institut National de la Santé et de la Recherche Médical, Unité 268, Hôpital Paul Brousse, 14 Avenue Paul Vaillant Couturier, 94807 Villejuif Cedex, France. E-mail address: dbroutyb{at}infobiogen.fr
-
↵3 Abbreviations used in this paper: SCF, stem cell factor; c-kitL, c-kit ligand; Flt3L, Flt3 ligand.
- Received September 6, 2001.
- Accepted February 26, 2002.
- Copyright © 2002 by The American Association of Immunologists