Morphological, enzymic and antigenic data are presented regarding a human bone-marrow stromal cell line maintained for 10 months and subcultured weekly. The main characteristics are a fibroblastoid morphology, diffuse growth in collagen gels, no colony formation in soft gel media, contact inhibition of growth and conversion to adipocytes when treated with hydrocortisone. The cells are non-phagocytic and membrane Fc receptors (i.e. aggregated human immunoglobulin G receptors) are absent, but they show diffuse cytoplasmic non-specific esterase activity, a strong acid phosphatase reaction, and a negative immunofluorescence (direct and indirect) against factor VIII antigen. Other cell lines also have been isolated and maintained in culture and present similar characteristics. These cell lines are thought to be derived from the acid-phosphatase-positive marrow stroma directly associated with bone trabecular matrix and probably represent a component of the haemopoietic inductive microenvironment. As such, they may provide a useful tool for studies in vitro of cell interactions and regulatory processes in the control of human bone-marrow haemopoiesis.

All components of the haemopoietic system (bone marrow, spleen, thymus and lymph nodes) show a complex association of stromal and haemopoietic moieties and numerous models have been described concerning the role of such stromal elements in the regulation of haemopoietic cell proliferation and differentiation. For example, the haemopoietic inductive microenvironment (mM) concept (Trentin, 1970) proposes that stem-cell differentiation is regulated at a local level and the environmental milieu determines the cell lineages produced; e.g. the microenvironment of the marrow favours granulopoiesis while the microenvironment of the spleen favours erythropoiesis (La Pushin & Trentin, 1977).

Results obtained by the long-term liquid bone-marrow culture method (Dexter & Testa, 1976; Dexter, Allen & Lajtha, 1977) also indicate the necessity for a functional mM for CFU-S proliferation and concomitant differentiation and proliferation of the haemopoietic lineages. Indeed, the main feature of this method is the initial establishment of an adherent cellular environment containing epithelial and fibroblastic-like cells as well as endothelial, phagocytic cells and large adipocytes (Allen & Dexter, 1976a). However, although the long-term culture system maintains stem-cell proliferation and the production of differentiated progeny for many months, there is still controversy as to which stromal cells are important regulatory components. Furthermore, little is known of the nature of the stromal/haemopoietic cell interactions or the inductive mechanisms involved. It has been reported (Friedenstein et al. 1974, 1976) that cultured fibroblastic cells from rabbit or guinea-pig marrow will transfer the microenvironment of their original haemopoietic tissues when transplanted under the kidney capsule of autologous animals. However, the contribution of the donor stromal elements (versus migrating host cells) was not determined. Although the cellular nature of the mM has not so far been resolved, recently, 2 types of stromal cells have been differentiated in rodent bone marrow (Westen & Bainton, 1979): an alkaline-phosphatase-positive (Alk-Pase) reticulum cell and an acid-phosphatase-positive (Ac-Pase) cell associated with granulocytic and erythroid precursors, respectively.

However, all of these studies discussed have been performed in rodents and relatively little information is available on the features of human bone-marrow stroma, although preliminary observations in human long-term cultures show that stromal elements play an important role (Toogood et al. 1980; Hocking & Golde, 1980; Gartner & Kaplan, 1980). In order to analyse further the relative importance of the different stromal cell types and to characterize the various factors being produced, it seemed to us that the production of a cloned cell line would be of great benefit. Using such cell lines it may be possible to recreate the environment necessary for haemopoiesis and to analyse in vitro the various haemopoietic disorders based upon environmental or haemopoietic cells.

In this paper, we report morphological, enzymic and antigenic patterns of normal human bone-marrow stromal cell lines, which have now been maintained and subcultured weekly for 10 months.

Donors

Rib specimens were obtained from patients undergoing cardiac surgery. The patients had a normal haemopoietic status and had not been treated with radiation or chemotherapy.

Treatment of the specimens

The rib fragments were kept cold and dissected under sterile conditions. The bone was carefully cleaned of muscle fragments and connective tissue. The rib bone was cut around the periphery with scissors, then opened (this procedure avoids contamination of the marrow with extraneous cellular elements). Small pieces of haemopoieitic bone trabeculae were detached with forceps and teased in culture medium.

Conditions of culture

Liquid culturel

Large cell aggregates were removed from the cell suspension by sedimentation at 1 g at room temperature for 3–5 min. A total of 105 to 106 cells were cultured in plastic tissue culture flasks (Falcon) containing 10 ml Fischers’ medium (Gibco) supplemented with L-glutamine (2 mM), 20% foetal calf serum (FCS) and antibiotics, and maintained at 37 °C in an atmosphere of air+ 5% CO2. In some cases horse serum was used in place of FCS.

For subculturing, the cultures were trypsinized immediately before confluency, and split 1/4 or 1/10. Cell growth was estimated by counting on a haemocytometer. The effects of some growth promoters and inhibitors were tested: fibroblast growth factor (FGF, 10 μg), epidermal growth factor (EGF, 100 μg) were purchased from Collaborative Research Waltham, Mass., carrageenan (Marine Colloids - Seakem G. Rex 8074), cholera toxin (Sigma), insulin (Wellcome, London) (Green, 1978).

Collagen cultures

Collagen gels were established as previously described (Schor, 1980; Lanotte, Schor & Dexter, 1981). Fragments of bone trabeculae were cultured on such gels in 1 ml supplemented Fischers’ medium. When specified, single-cell suspensions of stromal cells were plated in the collagen gel matrix in order to evaluate colony formation. When necessary, cells were recovered by treatment with collagenase (0·1 mg/ml collagenase, Sigma) for 1 h at 37 °C. For morphology, the collagen gels were dried and the cells stained in situ with May-Grunwald-Giemsa.

Transmission (TEM) and scanning (SEM) electron microscopy

The methods were as described previously (Allen & Dexter, 1976a). Briefly they are as follows : for TEM the adherent layer was fixed in situ in 3% glutaraldehyde in phosphate buffer, post-fixed with osmium tetroxide in the same buffer, dehydrated and embedded in situ in Luft’s Epon. Sections were cut at right angles to the growing surface. For SEM the flasks were fixed as above, and several regions of the growing surface removed with a warmed cork borer, critical-point dried from CO2 using amyl acetate as the transitional fluid, and sputter-coated with 200 A of gold.

Test of phagocyte function

Phagocytic cells in the adherent cell populations cultured from human bone marrow were evaluated employing 3 different techniques :

  1. Opsonized yeast (2 × 107 particles/cm2) was incubated in the cultures for 2 h at 37 °C in medium supplemented with 15% non-heat-activated FCS (or supplemented with 5% guinea-pig complement, diluted 1/30, Gibco N.Y.).

  2. Sheep red blood cells (SRJ3C) fixed with glutaraldehyde.

  3. Latex particles (1–5 μm) at 2 mg/ml in the culture medium.

The layers were washed twice after incubation, then fixed and stained.

Histochemical analysis of enzymic markers

Acid phosphatase and alkaline phosphatase activity

The Ac-Pase procedure used was based upon Barka’s method (Barka & Anderson, 1962). Cell layers were fixed and incubated in a solution containing naphthol AS-BI phosphoric acid and Fast Garnet GBC (Kaplow & Burstone, 1964). After 60 min incubation at 37 °C in the dark, the slides were washed in deionized water for 3 min. For Alk-Pase, cells were stained using naphthol AS phosphate and Fast Blue BBN (Kaplow, 1955).

Naphthol ASD chloroacetate esterase

A modification of the method described by Moloney (Moloney, McPherson & Fliegelman, 1960) was used. The staining solution was made by dissolving 10 mg naphthol ASD chloroacetate substrate in 1·6 ml dimethylsulphoxide (DMSO) and adding 12 ml distilled water,-1 ml propylene glycol, 12 ml 0·1 M-barbiturate buffer (pH 7·4). Fast Garnet salt GBC was added to the solution, then it was filtered. The slides were stained for 30 min at room temperature.

Naphthol ASD acetate esterase

The staining solution contained: 10 mg naphthol ASD acetate dissolved in 1 ml acetone and 1 ml propylene glycol. The substrate solutions were mixed progressively with 0·1 M-phosphate buffer, propylene glycol (2%) (20 ml). Fast Blue salt was added and the slides stained 30 min at 37 °C.

α-Naphthyl acetate esterase

The incubation solution was made as follows: 10 mg a-naphthyl acetate was dissolved in 0·25 ml acetone and added to 20 ml of 0·15 M phosphate buffer (pH 7·4) then shaken for 1–2 min. The solution was filtered directly onto the wet slides and incubated for 15 min at room temperature. The slides were washed with distilled water.

Factor VIII antigen immunofluorescence labelling

Goat anti-rabbit antisera and rabbit anti-human factor VIII antigen antisera were obtained from Miles (Immunochemicals and Miles Lab. Ltd, UK).

Stromal cells cultured on glass coverslips were washed twice with PBS. Cells were incubated with labelled or unlabelled rabbit antifactor VIII antigen antisera after 30 min fixation with acetone (4 °C). Slides were incubated in unlabelled antisera, and then incubated (indirect method) with goat anti-rabbit fluorescent antisera for 30 min at room temperature, washed and mounted in glycerophosphate. Similarly treated human umbilical-cord endothelial cells were used as a positive control.

Incorporation of [14C] palmitic acid into cellular lipids; cellular triglyceride extraction and silica gel thin-layer chromatography

Lipids were labelled with 14C-uniformly-labelled palmitic acid (403 mCi/mmol) in hexane solution (Amersham, U.K.). Cultures were incubated in 5 ml of medium containing 0·5 μCi [14C]palmitic acid at 37 °C; 30 min later, then subsequently, at hourly intervals, 50-μl samples were taken from the incubation medium for an evaluation of isotope clearance. After the period of incorporation, the cultures were washed. The lipids were extracted as described earlier (Green & Kehinde, 1974).

Binding of protein A-(SRBC) by aggregated human IgG

The culture medium was discarded from the culture flask and the adherent cells washed twice by incubation with phosphate-buflered saline (PBS) for 15 min. One ml of aggregated human IgG (200 μg/ml or 1 mg/ml) in solution in PBS supplemented with 10% FCS (Papa-michail et al. 1979) was added to each 25 cm2 culture flask and incubated for 30—45 min at 4 °C. The incubation mixture was decanted and the layers were washed twice with PBS (each was followed by 15 min incubation at room temperature). Two ml of a protein A-coated SRBC suspension (2% cells, v/v) was added to each flask and incubated without disturbance for 45 min at room temperature. The supernatants were gently discarded and replaced by a PBS/FCS solution and the number of cells forming rosettes was determined at x 60 magnification.

Irradiation experiments

Cells in suspension in serum-free medium were chilled to 4 °C on ice and submitted to irradiation using a ls’Cs source delivering a dose of 400 rad per min. After irradiation cells were appropriately diluted in complete culture medium and immediately plated. When cells were cultured at low concentration a feeder of lethally irradiated cells (102 cells/cm2, 1500 rad) was used to promote colony growth.

Isolation of a pure stromal cell population

Primary cultures of bone-marrow cells were established using various concentrations of cells from 104 to 5 × 106 per culture (10 ml). The cells were incubated for 4 days at 37 °C and subsequently all the floating cells were removed. At the low cell concentration small clusters of cells were observed. The cultures were fed with fresh medium and incubated in the same conditions. After 2 weeks the cultures were examined with a phase-contrast microscope. Where low numbers of cells were originally plated there was little or no growth of stromal cells. At the higher concentration a confluent layer of fibroblast-like cells was formed. We selected a culture containing a single large colony of fibroblastic cells and a few attached macrophages. The culture was trypsinized and the detached cells were plated without dilution in a new culture flask. Subsequently we found that a ‘pure’ population of stromal cells could be obtained after 4 passages; the contaminating macrophages being progressively eliminated by dilution at each passage. One of the cell lines isolated has now been maintained for 10 months by serial passage and will be discussed in some detail.

General characteristics

The morphology of the cells was homogeneous. The cells were fibroblast-like adherent cells (Fig. 1 A-F), generally bipolar but sometimes tripolar, proliferating to confluence as a monolayer and becoming contact-inhibited. However, some irregularities were observed in the monolayer. Underlapping cells were seen (Fig. 1E-F) and a multi-layer pattern was even observed in some areas. When regularly subcultured, the cell line did not show any sign of ageing or degeneration. However, if cultured for several weeks after confluence is reached, the morphology changes and the cells become epithelial-like (Fig. 1c). When these cells were subsequently subcultured, they produced multinucleated cells with a very low capacity to proliferate. When subcultured before confluence, the growth was exponential. Routinely, between 5 × 102 and 103 cells per cm3 were passaged at each interval and subculture was performed when the concentration was about 104 cells per cm2. The doubling time of the population was calculated over a period of 3 months of culture (corresponding to 12 passages and an estimated number of 65 generations). The doubling time of the population varied between 22 and 42 h (Fig. 2).

Fig. 1.

Growth morphology of the human stromal cells. When grown on plastic the cells have a characteristic bipolar shape (A, SEM, X750) and the cell surface shows numerous microvilli (B, SEM, × 7500). When the cells achieve contact inhibition of growth (confluence), the morphology becomes epithelial-like (c, living cell, phasecontrast, × 350). After a few weeks of confluence the monolayer is often disorganized in some areas; over-lapping cells are observed (E, × 200). In such areas, large amounts of extracellular matrix can be observed by phase-contrast or after May-Grunwald staining (F, × 600), The cells do not form discrete colonies when cultured in collagen gel matrix, but form a typical 3-dimensional network. Most frequently the cells are bi- or tripolar but some multipolar figures are observed (D, × 200).

Fig. 1.

Growth morphology of the human stromal cells. When grown on plastic the cells have a characteristic bipolar shape (A, SEM, X750) and the cell surface shows numerous microvilli (B, SEM, × 7500). When the cells achieve contact inhibition of growth (confluence), the morphology becomes epithelial-like (c, living cell, phasecontrast, × 350). After a few weeks of confluence the monolayer is often disorganized in some areas; over-lapping cells are observed (E, × 200). In such areas, large amounts of extracellular matrix can be observed by phase-contrast or after May-Grunwald staining (F, × 600), The cells do not form discrete colonies when cultured in collagen gel matrix, but form a typical 3-dimensional network. Most frequently the cells are bi- or tripolar but some multipolar figures are observed (D, × 200).

Fig. 2.

Kinetics of growth of the human stromal cell line. Effect of hydrocortisone (10−6 M) in FCS or HS-supplemented medium. The cell line could be regularly passaged for months when cultured in medium supplemented with FCS (○ – ○)-However, in the presence of FCS (▴ – ▴) or HS (• – •) supplemented with io’ M-HC, the doubling time increased and eventually no further growth occurred.

Fig. 2.

Kinetics of growth of the human stromal cell line. Effect of hydrocortisone (10−6 M) in FCS or HS-supplemented medium. The cell line could be regularly passaged for months when cultured in medium supplemented with FCS (○ – ○)-However, in the presence of FCS (▴ – ▴) or HS (• – •) supplemented with io’ M-HC, the doubling time increased and eventually no further growth occurred.

Effects of serum and growth promoters

The cell growth was low when medium was supplemented with a concentration of 5% FCS. Between 5 and 10% FCS, the rate of growth progressively increased, the optimal concentration being between 10 and 15%. Horse serum (HS) at concentrations greater than 10% led to arrested growth after one or two subcultures. Similar effects were observed with medium containing FCS and 10−6 M-hydrocortisone (HC). Medium containing HS and HC (10−6M) was inhibitory, (Fig. 2, Table 1). Cells arrested in growth by HS or HC showed a pattern of ageing similar to that reported for long periods of confluence. Growth promoters such as epidermal growth factor (EGF; 100 ng/ml), fibroblast growth factor (FGF; 10 ng-ml) or cholera toxin (CT; 10−10 M) had no effect on cell growth. Horse serum (HS; 10–20%) with hydrocortisone (10−8 M) or CT (at high concn, 10−7 M) were inhibitory. Carrageenan, a potent inhibitor of macrophage and monocytes, had no effect (Table 1).

Table 1.

Conditions of growth and effects of growth promoters

Conditions of growth and effects of growth promoters
Conditions of growth and effects of growth promoters

Growth in soft gels and plastic, and radi0sensitivity

The cells did not grow in 0·3% agar gel, but did proliferate in collagen gel to form a dispersed cell matrix (Fig. ID). When plated on plastic, at 4·5 × 102 cells per cm2, active growth occurred but the cells migrated and rarely formed colonies. When plated at very low densities (10–50 cells per cm2) no growth occurred, although irradiated cell feeders (102 cells/cm2) could be used to promote cell proliferation. In this case, after 4–7 days of culture discrete foci were observed. Although these were difficult to count, the assay provided a means of testing the radiosensitivity of the stromal cells.

A single cell suspension obtained by trypsinization was submitted to serial doses of 137Cs irradiation at 4 °C. The cells were then plated at cell concentrations ranging from 102 to 103 cells per 10 ml flask (on irradiated feeder layers) and examined 2 weeks later for colony growth. Alternatively, cells were plated at 5 × 103 cells per culture flask (in the absence of irradiated feeders). Before confluence occurred in control cultures, the cells were trypsinized and counted.

Similar results were obtained in both assays, giving a D37 between 100 and 150 rad (Fig. 3).

Fig. 3.

Sensitivity to 137Cs irradiation. Irradiated cells were plated at concentrations ranging from 102 to 103 cells per flask, and 2–5 × 103 lethally irradiated cells as feeders after 2 weeks. In a similar experiment 5 × 103 cells were plated in a 25 cm* culture flask and cultured until confluency occur in the control bottle. Cells were trypsinized and counted.

Fig. 3.

Sensitivity to 137Cs irradiation. Irradiated cells were plated at concentrations ranging from 102 to 103 cells per flask, and 2–5 × 103 lethally irradiated cells as feeders after 2 weeks. In a similar experiment 5 × 103 cells were plated in a 25 cm* culture flask and cultured until confluency occur in the control bottle. Cells were trypsinized and counted.

Phagocytic ability

Tests for phagocytosis were carried out on the cells after the fourth passage and after 6 months of serial sub-culture. We used the 3 different techniques described in Material and methods (opsonized yeast, glutaraldehyde-fixed SRBC, and latex particles). A primary culture (not subcultured) was also tested. It was found that primary cultures show variable numbers of phagocytic cells, but the stromal cell line was negative after both 4 passages and 6 months of serial passages.

Histochemistry

The results are summarized in Table 2. The cells were negative for membrane Alk-Pase activity but the cytoplasm was stained strongly for Ac-Pase, mainly in the perinuclear region. The acid phosphatase reaction was not diffuse but restricted to clusters of heavily stained granules (Fig. 4 A, B). The pattern of staining is characteristic of a strong Ac-Pase lysosomal reaction.

Table 2.

Cytoplasmic and membrane markers: comparative studies of the stromal cell line with primary cultures

Cytoplasmic and membrane markers: comparative studies of the stromal cell line with primary cultures
Cytoplasmic and membrane markers: comparative studies of the stromal cell line with primary cultures
Fig. 4.

Histochemical analysis. Stromal cell monolayer stained for acid phosphatase (A, × 100). The activity can be detected in all cells and is concentrated in the perinuclear region. The cytoplasmic extensions of cells are weakly positive or negative (B, × 400). For comparative purposes, a population of stromal cells derived from bone endosteum was cultured for 3 weeks in collagen gel matrix. The bone trabecule was removed by dissection before staining for Ac-Pase (c). These cells show a uniform strong Ac-Pase activity. D. The diffuse non-specific esterase reaction in the cytoplasm of the human stromal cells, E. The deposit of extracellular materials on plastic by growing stromal cells (× 7500). F. The thin membranous labelling of reticulin by silver impregnation (Pearse, 1960).

Fig. 4.

Histochemical analysis. Stromal cell monolayer stained for acid phosphatase (A, × 100). The activity can be detected in all cells and is concentrated in the perinuclear region. The cytoplasmic extensions of cells are weakly positive or negative (B, × 400). For comparative purposes, a population of stromal cells derived from bone endosteum was cultured for 3 weeks in collagen gel matrix. The bone trabecule was removed by dissection before staining for Ac-Pase (c). These cells show a uniform strong Ac-Pase activity. D. The diffuse non-specific esterase reaction in the cytoplasm of the human stromal cells, E. The deposit of extracellular materials on plastic by growing stromal cells (× 7500). F. The thin membranous labelling of reticulin by silver impregnation (Pearse, 1960).

Thin sections by TEM clearly showed large numbers of both primary and secondary lysosomes within the cytoplasm (Fig. 5). The cells were negative for chloroacetate esterase, but clearly positive for naphthyl esterase (Fig. 4D). Appropriate controls were done on primary cultures.

Fig. 5.

Vertical sections through stromal cell cultured on a plastic substratum and embedded in situ. A. Cytoplasmic region showing profiles of endoplasmic reticulum and numerous primary lysosomes (l). Extracellular matrix is also apparent, × 23000. B. Cytoplasmic region of a different cell showing numerous secondary lysosomes (s) and residual bodies. A dense aggregation of sub-plasmalemmal microfilaments is also apparent (arrows), × 22000.

Fig. 5.

Vertical sections through stromal cell cultured on a plastic substratum and embedded in situ. A. Cytoplasmic region showing profiles of endoplasmic reticulum and numerous primary lysosomes (l). Extracellular matrix is also apparent, × 23000. B. Cytoplasmic region of a different cell showing numerous secondary lysosomes (s) and residual bodies. A dense aggregation of sub-plasmalemmal microfilaments is also apparent (arrows), × 22000.

Membrane receptors for binding aggregated human immunoglobulin G (IgG) were investigated using 2 concentrations of IgG (0·2 mg/ml, 1 mg/ml). The highest concentration was reported to be sensitive enough to detect low concentration or low activity receptors on non-lymphoid cell membrane (Papamichail et al. All the cells were negative. Peripheral-blood adherent leukocytes were used as a positive control. The presence of factor VIII antigen was investigated by the direct and indirect immunofluorescence techniques. The tests were negative in both cases.

Collagen synthesis was demonstrated by silver impregnation, a specific histological staining method (Pearse, 1960). Before confluence, individual cells were labelled with a thin black membraneous deposit characteristic of reticulin (Fig. 4F). In confluent cultures, deeply stained bundles of extracellular collagen fibres were apparent. A fibrous extracellular matrix can be shown on the cell layer (Fig. IF) as well as a deposit between the cell and the plastic surface (Figs. 4E, 5 A).

Induction of adipocyte differentiation by hydrocortisone

Cells were cultured in Fischers’ medium with FCS and HC (10−6 M) or HS and 10−6 M-HC. The cells were maintained at confluence and fed weekly. Thirty days after confluence occurred the first adipocytes were observed in cultures containing HC in the culture medium. Small refractile lipid vacuoles accumulating in the perinuclear region were characteristic of a preadipocyte stage. Eventually, the fat vesicles enlarged, first in the proximity of the nuclei, then progressively in peripheral extensions of the cell cytoplasm. The cell, originally very flat, increased in volume and rounded up when giant fatty vacuoles developed (Fig. 6B, C). Finally the cells in most cases retracted from the plastic.

Fig. 6.

Adipocyte conversion of stromal cells, A. Cluster of cells at various stages of adipocyte differentiation, × 750. B. Early adipocyte with small fat vesicles, × 3000. c. Fully mature adipocyte, × 10000.

Fig. 6.

Adipocyte conversion of stromal cells, A. Cluster of cells at various stages of adipocyte differentiation, × 750. B. Early adipocyte with small fat vesicles, × 3000. c. Fully mature adipocyte, × 10000.

It is of some interest that transformation to adipocytes is not synchronous but has a tendency to occur at foci (Fig. 6A). The cells do not seem to be uniform in their capacity to form large round adipocytes ; some cells remain for very long periods in culture, with the appearance of preadipocytes (small lipid vacuoles). We have always found that the induction of adipocyte conversion needs arrested growth (by confluence). It is of some interest, however, that the subculture of a confluent layer, after 2 weeks growth in the presence of HC, does not prevent the formation of fat cells.

The kinetics of [14C]palmitic acid incorporation into fat was investigated before and after adipose induction. We observed a faster uptake of the label in the fat cell culture (Fig. 7). However, after the lipids were extracted, the distribution of the label in lipid and non-lipid material in the induced and non-induced cells was found to be similar. Therefore, although both induced and non-induced cells take up the label, the rate of uptake rather than the distribution is the distinguishing feature.

Fig. 7.

Rate of [14C]palmitic acid uptake by stromal cells before (•–•) and after (▴ – ▴) adipocyte conversion.

Fig. 7.

Rate of [14C]palmitic acid uptake by stromal cells before (•–•) and after (▴ – ▴) adipocyte conversion.

Other cell lines

Apart from the cell line discussed above, we have isolated and maintained 3 other cell lines with similar if not identical characteristics to those described above.

In this study, we have isolated several human bone-marrow stromal cell lines. One of these lines has now been maintained in culture for 10 months and has a fibroblastic morphology on plastic and in collagen gels. It grows as a monolayer and achieves contact inhibition but does not clone in soft agar. It synthesizes reticulin and then large amounts of collagen after confluence is reached. The cells do not express factor VIII antigen but are positive for non-specific esterase and strongly positive for Ac-Pase. However, the cells are not phagocytic, do not express Fc receptors (i.e. aggregated human IgG receptors) and are Alk-Pase negative. The cells can be induced to differentiate to adipocytes by addition of hydrocortisone.

The characteristics listed above make it difficult to categorize the cells into any known cell lineage. Consequently, we will discuss each set of data separately.

(a) The cell line was derived from a fibroblastic colony (CFU-F) grown in primary bone-marrow stromal culture. In many of its features it may be classified as a fibroblast-like cell: in its morphology, growth pattern, active biosynthesis of collagen. However, we have observed a very high Ac-Pase activity, which seems to be localized in lysosomes. In addition, the cell line is also clearly α-naphthol esterase positive. These data may indicate that the cell line belongs to the monocyte-macrophage lineage. However, the cells are not phagocytic and do not express Fc receptor. Apart from the fact that the cells are Ac-Pase positive, our cell line has all the characteristics of the CFU-F progeny observed (Castro-Malaspina et al. 1980) in primary culture of human marrow.

It has been reported recently (Westen & Bainton, 1979) that phosphatase activity can be used to classify the murine stromal elements into 2 different types: (1) a fibroblast-type of reticulum cell, concentrated near the endosteum and characterized by having the Alk-Pase activity associated with the plasma membrane; (2) a macrophage-type of reticulum cell with lysosomal acid phosphatase activity, which is uniformly distributed throughout the marrow. Unfortunately, our cell line does not fit into this simple classification. It might be classified as a fibroblast-type reticulum cell but is clearly Ac-Pase positive. Alternatively, it might be a macrophage-like cell but is not phagocytic and does not express Fc receptor. In this respect, it is significant that when endosteum-associated stromal cells were cultured in a 3-dimensional collagen gel matrix (Fig. 4c) the stroma cells proliferated. These cells, also having a fibroblastoid morphology, were uniformly Ac-Pase positive and were not phagocytic.

It is possible that our cell fine is similar to the cell types described first by Frieden-stein (Friedenstein et al. 1974, 1976) and other authors (Blackburn & Goldman, 1980; Toogood et al. 1980; Castro-Malaspina et al. 1980; Wilson, O’Grady, McNeil & Muhum, 1974; Wilson et al. 1978), which are thought to be mesenchymal elements. These cells are supposed to be multipotent; apart from having a self-perpetuating capacity, they are able to produce bone in vivo, and form a haemopoietic inductive stroma.

The pertinent features in the cell line we obtained are: firstly the Ac-Pase activity, and secondly the striking observation of bundles of extracellular collagen. We observed that the cells contained rough ergastoplasmic reticulum and also regions rich in primary lysosomes (in some places the cytoplasm of these cells was also filled with numerous secondary lysosomes) (Fig. 5). This seems to indicate that the cells simultaneously express 2 activities, which are usually mutually exclusive : biosynthetic and degradative activity. Such situations (however uncommon) have been reported previously (Napolitano, 1964; Ten Cate, 1972; Deporter & Ten Cate, 1973). It has also been reported that chondrocytes, in the preliminary stage of calcification of their matrix, express Ac-Pase in association with membrane vésicules when they are actively involved in glycosaminoglycan synthesis (Hall, 1978). In the case of the human stromal cell line reported above, the strong Ac-Pase activity observed may be explained as a physiological response to an excessive synthesis of extracellular matrix in relation to the condition of growth in culture. It will be interesting to determine if this switch is reversible.

If the above argument is accepted, it follows that all Ac-Pase cells detected in the bone marrow must not be assigned definitely to a macrophage-type reticulum cell. A pertinent observation (Westen & Bainton, 1979) was that the Ac-Pase stromal cells are associated with extracellular reticulin; and secondly, although a rare occurrence, such cells are capable of mitosis.

(b) The human stromal cell lines do not have the characteristic morphology of endothelial cells. Furthermore, we did not find Weibel-Palade bodies and the cells were negative for factor VIII antigen. However, even if they are not derived from the endothelial components of the sinus, it still remains difficult to exclude all relationship with the pericyte-like elements of the marrow.

(c) The cells undergo a conversion to adipocytes when treated with hydrocortisone. The morphological features of these adipocytes are very similar to those observed in primary culture and in long-term bone marrow cultures (Allen & Dexter, 1976 a, b;Dexter et al. 1977; Toogood et al. 1980; Gartner & Kaplan, 1980). The occurrence of fat cells is a well-documented observation in bone marrow in vivo as well as in long-term cultures, although as yet there is no agreement about the origin of the fat cells and their role in the inductive environment. In long-term marrow cultures the adipocyte conversion seems to depend upon corticosteroid hormones, since hydrocortisone can be used successfully to supplement the adipocyte-promoting properties of many batches of serum. We have also observed that murine fibroblastic stromal-cell colonies (unpublished results), growing in collagen gel, can be induced to differentiate into fat cells after 3 weeks incubation with hydrocortisone. It is of some interest that insulin (shown to facilitate adipocyte conversion in other systems; Green & Kehinde, 1975) has no such effect upon marrow stromal cells (Greenberger, 1978; Lanotte, unpublished observations).

In this respect, it may be relevant that monocytes from human peripheral blood spontaneously differentiate into adipocytes in soft agar gels (Zucker-Franklin, Grusky & Marcus, 1978). The precursor cells were phagocytic and possessed lysosomes and Fc receptors. It has also been demonstrated by electron microscopic studies that bone marrow pericytes from sinus can differentiate into fat cells (lyama, Ohzono & Usuku, 1979). Consequently, the fact that our cell line can be induced to differentiate into adipocytes does not help to assign it a place in the mesenchyme or haemopoietic lineage. However, such a cell line may serve to facilitate the study of adipose conversion in the marrow and investigations of defects in marrow adipogenesis. It will serve also as a tool for understanding the processes of the replacement of medullary haemopoietic tissue by fat tissue in physiological as well as pathological conditions.

Our results, with primary collagen cultures and the cell lines described here, indicate that caution must be exercised in assigning a cell to a particular cell lineage based upon limited aspects of morphology or cytochemistry. In the cell lines reported, a multi-parameter analysis has demonstrated the difficulty in classifying the cells as belonging to the fibroblast, endothelial, adipocyte or monocyte/macrophage pathways. Because of the lack of similar information on other ‘fibroblast’ cell lines studied following culture of marrow cells, we cannot make comparisons with our data. However, at present we are attempting to define the growth conditions necessary for the development of other marrow stromal elements, e.g. fibroblasts and endothelial cells. In addition, we are attempting to subclone the cell type reported here in order to investigate heterogeneity in the phenotype.

Finally, the availability of such a human bone-marrow stromal cell line will be useful for studies in vitro on radiation damage to the marrow environment. Hopefully, as an element of this environment, it will be useful for the construction in vitro of a functional human haemopoietic inductive microenvironment and the studies of cellular interactions or regulators produced therein. In a following paper, the role of this human stromal cell line in granulopoiesis in vitro, and its synergestic effects on colony-stimulating factors, will be reported.

The authors wish to thank Dr Williams (Pathology Department) who helped in the histochemical work and Drs S. Kumar, A. Schor (Medical Oncology Department) and J. Garland for their assistance in the anti-factor VIII test and Agg. Hu IgG binding assay.

This work was supported by the Medical Research Council and Cancer Research Campaign. T.M.D. is a fellow of the Cancer Research Campaign. M.L. is a visiting fellow supported by the CNRS (France) and the Royal Society (U.K.).

Allen
,
T. D.
&
Dexter
,
T. D.
(
1976a
).
Cellular interrelationships during in vitro granulopoiesis
.
Differentiation
6
,
191
194
.
Allen
,
T. D.
&
Dexter
,
T. M.
(
1976b
).
Surface morphology and ultrastructure of murine granulocytes and monocytes in long term liquid cultures
.
Blood Cells
2
,
591
606
.
Barka
,
T.
&
Anderson
,
P. J.
(
1962
).
Histochemical method for acid phosphatase using hexazonium panarosanilin as coupler
.
J. Histochem. Cytochem
.
10
,
741
753
.
Blackburn
,
M. J.
&
Goldman
,
J. M.
(
1980
).
Increased survival of haemopoietic progenitor cells in vitro induced by human marrow stromal cells
.
Expl Hemat
.
8
,
suppl
.
7
,
139
.
Castro-Malaspina.
H.
,
Gay
,
R.
,
Resnick
,
G.
,
Kaspoor
,
N.
,
Meyers
,
P.
,
Mckenzie
,
S.
,
Broxmeyer
,
H.
&
Moore
,
M. A. S.
(
1980
).
Characterization of human bone marrow fibroblasts colony forming cells (CFU-F) and their progeny
.
Blood
56
,
289
301
.
Deporter
,
D. A.
&
Ten Cate
,
A. R.
(
1973
).
Fine structural localization of acid and alkaline phosphatase in collagen containing vésicules of fibroblasts
.
J. Anat
.
114
,
457
467
.
Dexter
,
T. M.
,
Allen
,
T. D.
&
Lajtha
,
L. G.
(
1977
).
Conditions controlling the proliferation of haemopoietic stem cells in vitro
.
J. Cell Physiol
.
91
,
335
344
.
Dexter
,
T. M.
&
Testa
,
N. G.
(
1976
).
Differentiation and proliferation of haemopoietic cells in culture
.
In Methods in Cell Biology
, vol.
14
(ed.
D.
Prescott
), pp.
387
405
.
New York
:
Academic Press
.
Friedenstein
,
A. J.
,
Challa-Kyan
,
R. K.
,
Latsinik
,
N. V.
,
Panasuk
,
K. A. F.
&
Keiliss-Borok
,
I. V.
(
1974
).
Stromal cells responsible for transferring the microenvironment of the haemopoietic tissue
.
Transplantation
17
,
331
340
.
Friedenstein
,
A. J.
,
Gorskaja
,
U. F.
&
Kulogina
,
N. N.
(
1976
).
Fibroblast precursors in normal and irradiated mouse hematopoietic organ
.
Expl Hematol
4
,
267
274
.
Gartner
,
S.
&
Kaplan
,
H.
(
1980
).
Long-term culture of human bone marrow cells
.
Proc, natn. Acad. Sci. U.S.A
.
77
,
4756
4759
.
Green
,
H.
(
1978
).
Cyclic AMP in relation to proliferation of epidermal cell: a new view
.
Cell
15
,
801
811
.
Green
,
H.
&
Kehinde
,
O.
(
1975
).
An established preadipose cell line and its differentiation in culture. Factor affecting the adipose conversion
.
Cell
5
,
19
27
.
Greenberger
,
J.
(
1978
).
Sensitivity of corticosteroid dependent, insulin resistant lipogenesis in marrow preadipocyte of diabetic db/db mice
.
Nature, Lond
.
275
,
752
754
.
Hall
,
B. K.
(
1978
).
Developmental and Cellular Skeletal Biology
, p.
123
.
London, New York
:
Academic Press
.
Hocking
,
W. G.
&
Golde
,
D. W.
(
1980
).
Long-term human bone marrow cultures
.
Blood
56
,
118
124
.
Iyama
,
K. L
,
Ohzono
,
K.
&
Usuku
,
G.
(
1979
).
Electron microscopical studies on the genesis of white adipocytes: Differentiation of immature percipytes into adipocytes in transplanted preadipose tissue
.
Virchows Arch. path. Anat. Physiol
.
31
,
143
155
.
Kaplow
,
L. S.
(
1955
).
A histochemical procedure for localizing and evaluating leucocyte alkaline phosphatase activity in smears of blood and marrow
.
Blood
10
,
1023
1031
.
Kaplow
,
L. S.
&
Burstone
,
M. S.
(
1964
).
Cytochemical demonstration of acid phosphatase in hematopoietic cells in health and in various hematological disorders using azo dye techniques
.
J. Histochem. Cytochem
.
12
,
805
812
.
Lanotte
,
M.
,
Schor
,
S.
&
Dexter
,
T. M.
(
1981
).
Collagen gels as a matrix for haemopoiesis
.
J. Cell Physiol
.
106
,
269
278
.
La Pushin
,
R. W.
&
Trentin
,
J. J.
(
1977
).
Identification of distinctive stromal elements in erythroid and neutrophil granuloid spleen colonies: light and electron microscope study
.
Expl Hemat
.
5
,
505
522
.
Moloney
,
W. C.
,
Mcpherson
,
K.
&
Fliegelman
,
L.
(
1960
).
Esterase activity in leucocytes demonstrated by the use of naphthol ASD chloroacetate substrate
,
J. Histochem. Cytochem
.
8
,
200
207
.
Napolitano
,
L.
(
1964
).
Cytolysosomes in metabolically active cells
,
J. Cell Physiol
.
18
,
479
481
.
Papamichail
,
M.
,
Page-Faulk
,
W.
,
Gutierrez
,
C.
,
Temple
,
A.
&
Johnson
,
P. M.
(
1979
).
Binding of native and aggregated human y-globulin by mouse lymphoid cells and fibroblasts
.
Clin. Immunol-Immunopath
.
12
,
436
442
.
Pearse
,
A. G. E.
(
1960
).
Silver impregnation for reticulin
.
In Histochemistry : theoretical and applied
, pp.
817
818
.
London
:
J. A. Churchill Ltd
.
Schor
,
S.
(
1980
).
Cell proliferation and migration in collagen substrata in vitro
,
J. Cell Sci
.
41
,
159
175
.
Ten Cate
,
A. R.
(
1972
).
Morphological studies of fibrocytes in connective tissue undergoing rapid remodelling
.
J. Anat
.
112
,
401
414
.
Toogood
,
I. R. G.
,
Dexter
,
T. M.
,
Allen
,
T. D.
,
Suda
,
T.
&
Lajtha
,
L. G.
(
1980
).
The development of a liquid culture system for the growth of human bone marrow
.
Leukaemia Res
.
4
,
449
461
.
Trentin
,
J. J.
(
1970
).
Influence of hemopoietic organ stroma (hemopoietic microenvironment) on stem cell differentiation
.
In Regulation of Hematopoiesis
, vol.
1
(ed.
A. S.
Gordon
), pp.
161
186
.
New York
:
Meredith
.
Westen
,
H.
&
Bainton
,
D. F.
(
1979
).
Association of alk-phosphatase reticulum cells in bone marrow with granulocytic precursors
.
J, exp. Med
.
150
,
919
937
.
Wilson
,
F. D.
,
Greenberg
,
B. R.
,
Konrad
,
R. N.
,
Klein
,
A. K.
&
Walling
,
P. A.
(
1978
).
Cytogenetic studies on bone marrow fibroblasts from a male/female hematopoietic chimera
.
Transplantation
25
,
87
88
.
Wilson
,
F. D.
,
O’grady
,
L.
,
Mcneill
,
C. J.
&
Muhum
,
S. L.
(
1974
).
The formation of bone marrow derived fibroblastic plaques in vitro : Preliminary results contrasting these populations to CFU-C
.
Expl Hemat
.
2
,
343
354
.
Zucker-Franklin
,
D.
,
Grusky
,
G.
&
Marcus
,
A.
(
1978
).
Transformation of monocytes into ‘fat* cells
.
Lab. Invest
.
38
,
620
628
.