Marrow stromal stem cells

Marrow stromal tissue is a network of cells and extracellular matrix which physically supports the haemopoietic cells and influence their differentiation (Dexter, 1982). Macrophages and endothelial cells are often included in the stromal cell population (Allen & Dexter, 1982, 1984). However, in the present chapter, the term stromal will be restricted to cells of the stromal ‘fibroblastic’ system of marrow. By definition these include osteoblasts and preosteoblasts found near to bone surfaces, fibroblasts and reticular cells, terms commonly used for the soft connective tissue cells of marrow and blood vessel walls, and marrow adipocytes (Owen, 1985). Current evidence suggests that haemopoietic, stromal and endothelial cellular systems are histogenetically distinct in the post-natal animal under normal conditions (Le Douarin, 1979; Wilson, 1983; Simmons et al. 1987).

As a working hypothesis for differentiation in the marrow stromal system we have proposed a scheme analogous to that in the haemopoietic system, where stromal stem cells give rise to committed progenitors for different cell lines (Fig. 1) (Owen, 1978, 1985). The main evidence for stem cells comes from an in vivo assay of tissue formed following transplantation of stromal cells under the renal capsule or in diffusion chambers. The formation of fibroblastic colonies from a suspension of marrow cells plated in vitro, each colony being derived from a single fibroblastic colony forming cell (FCFC) or colony forming unit fibroblastic (CFU-F), was demonstrated more than a decade ago (Friedenstein et al. 1970; Friedenstein, 1976) but clonal in vitro assays, which have been so successful for investigating progenitor cells in the haemopoietic system, have been little explored for the stromal system. The number and hierarchy of the stromal fibroblastic lines in marrow are not yet completely identified. The proposed lines have been designated fibroblastic, reticular, adipocytic and osteogenic. Lines with these characteristics can be established in vitro from the soft connective tissues of marrow and bone surfaces. Osteogenic, fibroblastic and adipocytic lines have been derived both from new born mouse calvarial and marrow tissues (Kodama et al. 1982; Lanotteeř al. 1982; Sudoet al. 1983; Ziporiet al. 1985; Benayahu et al. 1987). In this paper, the evidence supporting the stromal system hypothesis will be reviewed.

The demonstration that suspensions of single cells derived from marrow form a fibrous osteogenic tissue when incubated within diffusion chambers implanted in vivo was an important advance. This showed that the differentiating capacity of marrow stromal cells was not dependent on the structural relationship of the cells in situ and that precursor cells determined in an osteogenic direction were present in marrow stroma (Friedenstein, 1973). Cells of the host do not penetrate the chambers and vascularization of the tissue does not occur. The resorbing cells of bone are derived from the haemopoietic system (Marks, 1983; Chambers, 1985) and there is no evidence of bone resorption within the chambers (Ashton et al. 1980). Consequently the diffusion chamber is suitable for studying differentiation of the stromal cells per se.

The stages of tissue formation when a suspension of about 10⁷ single marrow cells (rat or rabbit) is incubated within diffusion chambers implanted intraperitoneally is shown diagrammatically in Fig. 2 (Ashton et al. 1980; Bab et al. 1986; Mardon et al. 1987). Within a few days stromal fibroblasts have attached to the millipore filter and fibrous tissue grows from this surface. Early signs of osteogenesis, from about 8 days onwards, are localized areas of fibroblastic cells which assume a more polygonal shape and express intense staining in the cytoplasmic membranes for alkaline phosphatase activity (Fig. 3). By about 3 weeks both bone and cartilage have developed in association with these areas and the tissue is morphologically identical to its skeletal counterparts according to both light and electron microscopic criteria (Ashton et al. 1980; Bab et al. 1984).

The fibrous tissue formed initially, stains strongly for type III collagen, laminin and fibronectin, weakly for type I and is negative for type II collagen. However, as osteogenesis develops within the fibrous anlage, collagen types I and II are found exclusively in bone and cartilage respectively. Fibrous tissue which does not become osteogenic continues to stain for type III; in fact loss of expression of collagen type III is considered a good marker for the appearance of the osteoblastic phenotype. Staining for laminin and fibronectin also decreases to negligible levels as osteogenic tissue develops. The sequential expression of these extracellular matrix components is similar to that seen in osteogenesis in the embryo and it was concluded that connective tissue generation in diffusion chambers from precursor cells present in adult marrow resembles a bone developmental process (Mardon et al. 1987).

Using histomorphometric techniques analysis of cell kinetics in chambers inoculated with 10⁷ marrow cells showed that there was an increase in the total stromal cell population by more than six orders of magnitude between 3 and 20 days (Bab et al. 1986). The number of haemopoietic cells decreased to O•O5 % of the initial inoculum during the same period. At 3 days an average of only 15 stromal fibroblastic cells was identified within the chamber and it was concluded that the mixture of connective tissues formed is generated by a small number of cells with high capacity for proliferation and differentiation, i.e. cells with the characteristics of stem cells.

When an intact piece of marrow is transplanted ectopically, for example either subcutaneously or under the renal capsule, stromal cells survive, proliferate and differentiate into bone and marrow stroma and provide the appropriate microenvironment for haemopoiesis by invading host haemopoietic stem cells, resulting in formation of a bone and marrow organ (Tavassoli & Crosby, 1968; Friedenstein, 1976). By the use of different antigenic and chromosomal markers in donor and host it was shown that the marrow stromal cells of the newly formed bone and marrow organ are derived from the donor and the haemopoietic tissue from the host (Friedenstein et al. 1978). Freshly isolated suspensions of marrow cells or fibroblastic cells cultured from marrow, grafted under the kidney capsule within porous sponges, also form a bone and marrow organ in heterotopic transplants (Friedenstein et al. 1982). With the morphological methods used it was not possible to be certain of the exact sequence of events in regeneration of bone and marrow tissue. A review of the relevant literature suggested that initially the stromal elements differentiate into osteoid and bone trabeculae and later reconstruct the connective tissue lining cells of the microvasculature (Owen, 1980).

The clonal origin of the fibroblastic colonies formed when suspensions of dispersed marrow cells are cultured in vitro (Friedenstein et al. 1970; Castro-Malaspina et al. 1980) has been confirmed using thymidine labelling, time-lapse photography and chromosome markers (Friedenstein, 1976; Friedenstein et al. 1987; Latsinik et al. 1987). A definitive answer to the question of whether stem cells are present in the stroma of bone and marrow must come from an analysis based on the self-replicating ability and multipotentiality of individual FCFC or CFU-F. This has been investigated by assay of single colonies in open transplant and diffusion chambers in vivo (Friedenstein, 1980; Friedenstein et al. 1987). By transplanting individual clones under the renal capsule it is possible to examine whether a single cell gives rise to the variety of cell lines of bone and marrow stroma necessary to support haemopoiesis.

Friedenstein and his colleagues have studied the potential for differentiation in open transplant of single fibroblastic clones grown in vitro from mouse marrow cells (Friedenstein, 1980). Clones were grown for 16 to 30 days on thin collagen gels. Individual clones were excised, placed between small pieces of millipore filter and grafted under the renal capsule of syngeneic recipients. The tissue formed was examined histologically. About 15 % of the fibroblastic clones transplanted produced a bone and marrow organ, containing osteogenic tissue, typical adipose cells, and a well-formed microvasculature together with associated host-derived haemopoiesis. About 15 % of the clones yielded only bone tissue. The remaining clones formed a soft connective tissue or did not form any tissue (Friedenstein, 1980).

It is proposed that the clonogenic precursors which yielded a bone marrow organ, i.e. gave rise to several different stromal lines, are putative stromal stem cells. Clones producing either bone or soft connective tissue only arise from precursor cells with limited potential. A proportion of clones has a low proliferative activity and does not survive. Although these results provide strong circumstantial evidence for the existence of a stromal cell system and stromal stem cells, more definitive proof of the clonality of colonies capable of forming the mixture of bone and marrow stromal lines is necessary and requires the use of uniquely marked stem cells in the donor material.

Using cultures of rabbit marrow cells a high ability of CFU-F to self-renew was demonstrated (Friedenstein et al. 1987). Fibroblasts were grown to confluence in culture from a number of CFU-F (FCFC) and after about eighteen passages in vitro these cells were assayed in diffusion chambers. The inoculum of harvested fibroblasts able to form fibrous-osteogenic tissue in the chamber was defined as one osteogenic unit. The number of osteogenic units which could be harvested from culture after passaging ranged up to more than a thousand times the number of initiating CFU-F, thus demonstrating that precursors of the fibrous-osteogenic tissues are highly proliferative and that they retain their capacity for differentiation after extensive culture and passaging in vitro. The ability of fibroblasts grown from a single colony in vitro, harvested after two or three passages, to form osteogenic tissue in diffusion chambers was demonstrated and evidence for a common precursor for bone and cartilage was also obtained in these experiments by assay of clonal progeny in conditions where either bone only or both bone and cartilage are formed.

In vitro colony assays have been used to study both multipotent and lineage-committed progenitor cells in the haemopoietic system (Metcalf, 1984). Recent studies to investigate differentiation of CFU-F at the colony level have been initiated with a view to developing in vitro clonal methods for studying lineage in the marrow stromal system (Owen et al. 1987). When a suspension of rabbit marrow cells was cultured in vitro with foetal calf serum the fibroblastic colonies formed varied widely in size and level of expression of alkaline phosphatase activity (Fig. 4). The enzyme was used as an osteogenic marker, since its appearance is an early indicator of developing osteogenic tissue when rabbit marrow cells are cultured in diffusion chambers in vivo (Bab et al. 1986). Alkaline phosphatase activity appeared to originate in the centre of the colony where the cells first reach confluence and where cell growth is arrested. Labelling with tritiated thymidine showed growing cells located mainly at the edge of the colony (Owen & Friedenstein, 1988). The level of enzyme expression varied: some colonies were entirely negative, whereas in others a large proportion of the cells was positive for the enzyme (Fig. 4).

The heterogeneity of the colonies formed is consistent with their derivation from a population of stem and progenitor cells at different stages of a tissue developmental system, (Fig. 1). It was assumed that colonies expressing a high level of alkaline phosphatase activity were initiated by CFU-F approaching committed progenitor status for the osteogenic line. Only 2•5 % of colonies fell into this category and although not proven it seemed likely that the majority of colonies in the cultures are probably from relatively early precursors with more than one lineage potential. In these cultures hydrocortisone (HC), which is known to increase alkaline phosphatase activity in a number of osteogenic systems (Rodan & Rodan, 1984), stimulated the number of colonies formed in all size ranges and the enzyme in colonies at all levels of expression (Fig. 5A). Addition of epidermal growth factor (EGF) increased average colony size and reduced expression of the enzyme in colonies at all levels to negligible amounts (Fig. 5B). These results indicate that both HC and EGF act on a spectrum of early precursors in the stromal system.

As is well known the survival of cell populations in culture is highly dependent on culture conditions. In the current literature culture conditions for marrow stromal cells fall mainly into two categories. (1) As in the experiment just described, marrow cells cultured in medium with foetal calf serum form fibroblastic colonies from precursors designated CFU-F. Under these conditions adipogenesis is rarely seen and the cells are commonly referred to as marrow fibroblasts (Castro-Malaspina et al. 1980). (2) Using the conditions which establish the microenvironment necessary for support of haemopoietic stem cells in long term marrow cultures (horse serum with HC), reticular-type cells with a high capacity for adipogenesis are predominant in the cell layer and few fibroblastic cells are present (Dexter et al. 1977; Singer et al. 1985). With a modification of these conditions, the growth of colonies with a reticular-fibroblastoid morphology, many with lipid inclusions, from precursors designated CFU-RF has been reported (Lim et al. 1986). In the latter study a comparison of macromolecules synthesized by progeny from CFU-F and CFU-RF was made using immunofluorescent antibody techniques. Both cell populations were positive for collagen types I, HI and V; however, cells from CFU-RF were, in addition, positive for collagen type IV and laminin. The cells of both populations were negative for markers for endothelial and myeloid cells. It is not known how CFU-F are related to CFU-RF and the possibility cannot be ruled out that they are derived from common precursors and that the different patterns of macromolecules synthesized are a result of culture conditions.

An important question is, how are cells in culture related to their in vivo counterparts (Fig. 6)? Unfortunately there is no standard terminology for the different cell types. The best characterized cells of the marrow stromal system are osteogenic cells. Both in vivo and in vitro they are identified by high levels of alkaline phosphatase activity, synthesis of type I collagen (Sandberg & Vuorio, 1987), and of the bone-specific protein, osteocalcin (Bronckers et al. 1987), and the presence of receptors for parathyroid hormone (Rodan et al. 1988). By comparison there has been little characterization of other stromal cell types. Collagen types I and III are found in situ throughout the reticular-fibroblastic network and connective tissues of blood vessel walls (Bentley et al. 1981, 1984). Type IV collagen and laminin are mainly associated with basement membrane and localized in the walls of small blood vessels. It might be speculated that stromal cells which synthesize collagen type IV and laminin in vitro may be derived from perivascular cells.

There is little understanding of the relationship of perivascular cells (pericytes, smooth muscle cells and fibroblasts of vessel walls) to other cells of the stromal system (Fig. 6). Cells which express muscle actin have been identified in long term marrow cultures (Charbord et al. 1985) and a cloned cell line derived from foetal rat calvaria was reported to have myogenic, osteogenic and adipocytic potential in culture (Grigoriadis et al. 1986). Pericytes are thought to be precursors of smooth muscle cells (Ham, 1974a). Whether a myogenic cell line is a component of the marrow stromal system however, must await the outcome of future experiments. Stromal stem cells are not identifiable morphologically and their location in bone marrow is not known. In young adult rabbits the CFU-F are more concentrated in the cell layers near bone surfaces than throughout the marrow cavity (Ashton et al. 1984). On the basis of morphological evidence, it has been suggested in the past that pluripotential mesenchymal cells are present in soft connective tissues throughout the body often in perivascular locations (Ham, 19746).

The best available assay for marrow stromal stem cells is tissue analysis following transplantation in an open system under the renal capsule where all cell lines of a bone and marrow organ may be formed. The most direct evidence for stromal stem cells has been obtained from assay of single colonies using this system. Studies with diffusion chambers and preliminary investigations in vitro also support the stromal system hypothesis. However, for more rigorous proof of the stem cell hypothesis, marked stem cells and an ability to follow the marker in the stem cell progeny is necessary (Abramson et al. 1977). Gene transfer techniques which have recently been used for haemopoietic stem cells could be applied in the present case (Lemischka et al. 1986; Price, 1987).

Few experiments have been made to investigate the possibility of developing clonal ‘in vitro assays for the stromal system but preliminary studies indicate their feasibility (Owen et al. 1987). In liquid cultures with foetal calf serum a proportion of colonies formed appears to be derived from early precursors and investigations in vitro of events occurring prior to production of committed progenitor cells should be possible. The existence of irreversibly committed progenitor cells for the different cell lines of the marrow stromal system is still hypothetical. Nor is there any solid information on the hierarchy of the different lines in this system, e.g. whether committed progenitors exist for each cell line or are common to two or more lines is not known. The availability of more markers for the different cell lines of the stromal system and of culture methods which allow their differentiation in vitro are needed before the stimulatory factors necessary for the development of specific phenotypes can be fully identified. In this respect the use of soft-gel media (Ernst & Froesch, 1987; Guenther et al. 1988) and the bone nodule method (Bellows et al. 1986, 1987) are giving promising results.

It is possible that the stromal fibroblastic system of bone and marrow is part of a wider stromal cell system of the body, each organ containing pluripotent and restricted stem cells which generate the cell lines of the particular organ concerned (Fig. 7). In discussing this an attempt will be made to clarify some of the current nomenclature. In one known situation, stromal cells can be induced to differentiate in a direction different from the tissue in which they are located. This is induction of osteogenesis in many tissues outside the skeleton, by factors derived from bone matrix and transitional epithelium (Friedenstein, 1973; Reddi, 1981; Urist et al.1983). The cells involved were named inducible osteogenic precursor cells (IOPC); on the other hand the cells in marrow which form bone spontaneously in ectopic transplants and in diffusion chambers in vivo were called determined osteogenic precursor cells (DOPC) (Friedenstein, 1973) and more recently osteogenic stem cells (Friedenstein et al. 1987). In our present state of knowledge it is not possible to distinguish between DOPC, osteogenic stem cells and the marrow stromal stem cells of the present chapter. All are components of a population which contain specific stem cells and early progenitors of bone and marrow stromal lines whereas IOPC are likely to be components of a pluripotent stromal stem cell compartment (Fig. 7). There is little information on the stromal cell system in organs other than bone and marrow except that in haemopoietic organs stromal cells are specific for their organ of origin in terms of the type of haemopoiesis which they support (Friedenstein et al. 1974; Trentin, 1976).

Finally, little is known about the relative numbers of marrow stromal stem cells with age or in different diseases and their role in normal physiology. Marrow CFU-F are practically non-cycling in vivo (Keilis-Borok et al. 1971) and the importance of stromal stem cells therefore may lie mainly in the role they play in connective tissue regeneration, e.g. in re-establishment of the haemopoietic microenvironment and in bone repair, and possibly in certain connective tissue diseases.