Concise Review International Journal of Cell Cloning 9:95-108 (1991)

The Leukemia Inhibitory Factor (LIF) Donald Metcalf The Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia

Key Words. Myeloid leukemia Platelets * Megakaryocytes

*

Differentiation commitment

*

Stem cells

*

Abstract. Leukemia inhibitory factor (LLF) is a glycoprotein able to enforce differentiation and/or suppress clonogenic self-renewal in a number of myeloid leukemic cell lines. When acting on normal embryonic stem cells, it has the opposite action of preventing differentiation commitment. LIF is not a proliferative factor when acting alone on normal hemopoietic cells, but can potentiate the action of interleukin 3 on blast cell and megakaryocyte precursors. When injected in vivo, LIF stimulates rises in megakaryocyte numbers and platelet levels. LIF also exhibits striking functional effects on a wide range of other cells including hepatic parenchymal cells, neurones, adipocytes, osteoblasts and gonadal cells. The polyfunctionality of LIF suggests strongly that it is normally intended to be produced locally and act as a local regulator. Despite its wide range of actions, LIF remains a promising candidate for clinical use in thrombocytopenia and myeloid leukemia.

Historical Background The leukemia inhibitory factor (LIF)has a curious history of multiple rediscovery by workers in unrelated fields which is useful to review to introduce the multiple names for LIF in the literature (Table I), and some of the complex biological actions of this regulator. Work on LIF originated in 1969 with the description by Zchikawa that cells of the murine myeloid leukemia M1 line could be induced to differentiate to granulocytes and macrophages by the addition of various types of conditioned medium [l, 21. This work coincided with the introduction of another murine myelomonocyticleukemia, the WEHI3B [3] which, when finally adapted to growth as a cloned cell line, could also be induced to exhibit differentiation in colonies growing in semisolid medium [4]. Discrepant reports appeared in subsequent years regarding the nature of the biological molecule able to induce differentiation in murine myeloid leukemic cells. Work with WEHI3B colonies led to the purification of granulocytecolony~~

~~

~~~

Correspondence: Donald Metcalf, The Walter and Eliza Hall Institute of Medical Research, P.O. Royal Melbourne Hospital, 3050 Victoria, Australia. Received December 4, 1990; accepted for publication December 4, 1990. 0737-1454/91/$2.00/0 oAlphah4ed Press

96

Metcalf

Table I. Alternative names for LIF Name

Acronym

Reference

Differentiation-inducing factor Differentiation-stimulating factor Differentation-inhibitory factor Differentiation-retarding factor Human interleukin for DA celIs Hepatocyte-stimulating factor I11 Cholinergic neuronal differentiation factor Melanoma-derived lipoprotein lipase inhibitor

D-Factor D-Factor DIA DRF HILDA HSF III CNDF MLPLI

Tornida et al. [8] Lave et al. [lo] Smith et al. [17] Xbopman and Cotton [16] Moreau et al. [IS] Baumann and Wong [25] Yamamori et al. [28] Mori et al. [26]

stimulating factor (G-CSF) as the active molecule [5]. Parallel studies using the M1 leukemia led Such and his colleagues to the purification of a molecule termed MGI-2 [6] which, on sequencing, proved to be interleukin (IL) 6 [7]. Meantime Tomidu et ul. [8], also working with M1 leukemic cells, purified from L-cellconditioned medium a differentiation-inducing factor (DIF or D-factor) with characteristics clearly different from those of either G-CSF or IL-6. In an attempt to clarify this problem, we fractionatedKrebs 11ascites tumorconditioned medium-the source often used in M1 studies-and from this material purified a molecule, leukemia inhibitory factor (LIF) [9], which seemed likely to be similar to D-factor. Subsequent cloning of the D-factor has confirmed its identity to LIF [lo]. The basis for the three groups arriving at three different active molecules has since been explained. WEHI-3B cells do not respond to LIF [11] nor does the M1 subline used in the Weizmann Institute [El while the Sapporo/Melbourne subline of M1 cells responds to both L F and IL-6, but preferentially to the former [ll]. Using nucleotide probes based on amino acid sequence data from purified murine LIF, we isolated a cDNA for LIF from a T lymphocyte library [l3] and subsequently isolated murine and human genomic LIF clones [14, 151. The biochemical properties of LIF had intriguing similaritiesto the partially established properties of a factor termed “differentiationinhibitory activity” (DIA or DRF). This factor was under characterization as being able to prevent spontaneous differentiation commitment in cultured murine embryonic stem cells [16, 17. Studies using LIF confirmed that embryonic stem cells exhibit LIF receptors and can be held indefinitely in an undifferentiated state by LIF [18]. When DIA was purified and sequenced, the amino acid sequence observed was that of LIF [17]. In a separate series of investigations, a molecule of human origin was characterized as being able to stimulate the proliferation of a subline of m u r k myeloid DA-1 cells. This factor, termed “human interleukin for DA cells” (HILDA), when purified and sequenced also proved to be LIF [19]. It had been observed that DF (LIF) was able to release calcium from rodent calvarial cultures [20] and

Leukemia Inhibitory Factor

97

work using recombinant LIF confirmed this action [21], documented the presence of receptors for LIF on osteoblasts, but not osteoclasts [22], and showed that LIF was a proliferative stimulus for osteoblasts [21] with other functional effects on bone metabolism [22]. Studies on signalling by autonomic neurones had revealed the existence of a factor able to switch signalling from adrenergic to cholinergic mode (cholinergic neuronal differentiation factor, CNDF). When purified and subjected to amino acid sequencing, this molecule was also LIF [23]. Concurrently, in studies on the production by hepatocytes of acute phase proteins, two distinct hepatocytestimulating factors (HSF) were documented. Analysis showed the monocytederived factor to be IL-6 [24], while HSF-111 was found to be LIF [25]. Finally, Mon' et af. [26] purified a polypeptide produced by human melanoma cells able to induce severe cachexia when grafted to nude mice. This polypeptide was characterized as an inhibitor of lipoprotein lipase-mediated uptake of lipide by adipocytes. Sequencing of the purified polypeptide showed this also to be LIF. In view of this diverse set of biological functions, there is obviously some difficulty in assigning to LIF an appropriate, all-embracing, name or acronym. For the present, we continue to use the name under which the cDNA and genomic clones were originally isolated-leukemia inhibitory factor (LIF).

Biochemistry and Molecular Biology of LIF and Its Receptor LIF is a highly glycosylated single chain polypeptide with a predicted M, for both the murine and human LIF polypeptide of 20,000. Glycosylation is variable according to tissue source but most often results in a glycosylated molecule of apparent M, 4562,000 [27]. There are six cysteine residues in murine, human and ovine LIF and seven in rat LIF. Disulfide bridging of these cysteines is necessary for biological activity. The number of potential N-linked glycosylation sites ranges in different species from seven to eight. The carbohydrate moiety of the molecule does not take part in receptor binding, and is not necessary for the biological action of the molecule either in vitro or in vivo. LIF is encoded by a single gene located in the mouse on chromosome 11 [28] and in man on chromosome22ql2 [29]. In both species, the gene comprises four exons although portions of only two encode RNA for the mature protein. LIF shows no sequence homology with any known growth factor or other polypeptide. The uniqueness of LIF is confirmed by the presence of specific membrane receptors for LIF of high binding affinity [30,31]. Binding of LIF to these receptors is not competed for by any known hemopoietic growth factor [30, 311. Within hemopoietic populations, LIF receptors are exhibited by cells of the monocyte-macrophageseries, receptor numbers increasing with maturation [31], although osteoclasts do not exhibit receptors [22]. Megakaryocytes also exhibit receptors for LIF and again receptor numbers increase with maturation (Metcalf D,Nicolu NA, Hilton DJ, unpublished data). No detectible receptors for LIF are

Metcalf

98

present on erythroid, granulocytic, eosinophilic or mast cells. The majority of lymphocytes, although a small subset of lymphocyte-likecells in the marrow, spleen and thymus, exhibits LIF receptors [31]. This selective pattern of receptor distribution within hemopoietic populations contrasts with a broad distribution of receptors on non-hemopoieticcells as listed in Table II[27]. LIF receptors are displayed on osteoblasts,certain neurones, liver parenchymal cells, placental syncytialtrophoblast cells, embryonic stem cells, adipocytes, adrenal cortical cells, Leydig cells and certain endothelial cells. No information is available on the size of the receptor or its molecular structure.

Sources of LIF LIF is not detectable by bioassay in normal murine or human serum nor is LIF mRNA detectable in tissues by northern analysis. Amplification using the polymerase chain reaction has given variable results, but has detected low levels of LIF mRNA in some tissues. An exception is the consistent production of LIF by the metrial gland developing in the uterine wall at the site of blastocyst implantation [32]. LIF, like other hemopoietic regulators, appears to be a potential product of many cell types if they are stimulated by suitable inductive signals. Cell types known to be able to produce LIF include fibroblasts [8], T lymphocytes[El],monocytes and macrophages [33], stromal cells [34], osteoblasts [22], astrocytes [35] and embryonic blastocyst cells [36]. LIF is also produced by a variety of neoplastic cells including Krebs ascites tumor cells, 5637 bladder carcinoma cells and a human melanoma cell line [22]. Knowledge of signals inducing LIF transcription and/or production is limited. For various cell types, these include endotoxin, IL-1, T lymphocyte mitogens, phorbol esters, IL-la! and 6 , transforming growth factor ("GF) p, tumor necrosis factor (TNF) a! and retinoic acid.

Organ Distribution of Injected LIF After intravenous(i.v.) injection in the mouse, non-glycosylated recombinant LIF has a second phase half-life of 8-10 minutes, but following intraperitoneal injection, a more sustained elevation of serum levels of LIF for 2-6 hours can be achieved [37]. After i.v. injection of radiolabeled LIF, the highest counts/mg tissue are observed initially in the liver and the kidney. In the liver, the localization is over parenchymal cells while in the kidney the glomeruli (presumably the glomerular capillary loops) are selectively labeled. The pancreas and salivary gland exhibit a slightly delayed accumulation of label. Low and relatively uniform levels of labeling are observed in other organs. On autoradiography, labeling is also observable of osteoblasts, megakaryocytes,the mesenchymal tissue in intestinal villi and cells

Leukemia Inhibitory Factor

99

Table II. Normal cells exhibiting receptors for LlF

Hemopoietic Monocytes Macrophages Megakaryocytes Lymphocytes (small subset) Non-Hemopoietic Hepatic parenchymal cells Sensory neurones (some) Autonomic neurones Endothelial cells Osteoblasts Fibroblasts Adipocytes Adrenal cortex (inner layer) Leydig cells Embryonic stem cells Syncytial trophoblasts

of the inner region of the adrenal cortex (Hilton D, MetcalfD, Nicola NA, WaringR unpublished data). Intravenous injection into pregnant mice leads to intensive labeling of maternal syncytial trophoblast cells but not their fetal equivalent and no labeling is observed of fetal tissues, indicating that LIF is unable to cross the placental barrier. LIF also appears unable to cross from the blood to the cerebrospinal fluid.

Actions on Hemopoietic Cells LIF has no direct survival-promoting or proliferative effects on normal adult or fetal murine [ll] or adult human hemopoietic cells. However, it is a proliferative stimulus for murine DA-1 cells 1191 and potentiates the stimulation by multipotential (mu1ti)-CSF of the prolifration of murine megakaryocyteprecursors (Metcalf D, unpublished data), human blast colony-forming cells [38] and two myc-transformed erythroid cell lines of murine fetal origin (CoryS,Maekawa Z Metcalf D, unpublished data). LIF (HILDA) was reported to be a functional activator of eosinophils [39], but eosinophils lack receptors for LIF and we have not observed eosinophil activation with purified recombinant LIE The most striking biological actions of LIF on hemopoietic cells result from the ability of LIF, in concentrations above 0.2 ng/ml, to induce differentiation in murine M1 leukemic cells. Culture of M1 cells with LIF leads to increased lysozyme synthesis with maturation to cells with a macrophage morphology that express Fc receptors and phagocytic activity [8] and an acquired dependency on macrophage (M)-CSF for survival [11,40]. In parallel with these actions, LIF sup-

Metcalf

100

presses the self-generation of clonogenic M1 cells and, as a consequence, reduces the size and number of colonies formed by MI cells. The overall result of these actions is an inhibition of the proliferation of M1 cells and hence the use of the term “leukemia inhibitory factor” as an appropriate description for the then known properties of the molecule [ll, 401. The actions of LIF on MI cells are receptor-mediated and appear to be direct because LIF can induce differentiation in and/or suppress single M1 cells and exhibit density-independent actions on M1 cells in culture, Evidence of LIF action is already present after 24 hours of culture when many developing clones have a dispersed morphology and contain large, opaque, cells of irregular shape and increased cell-cell adherence typical of developing macrophage precursors [ll]. Although M1 cells constitutively produce low levels of IL-6 and production of IL-6 is increased by LIF stimulation, it appears unlikely that the action of LIF is IL-6-mediated, since antibodies to IL-6 do not block LIF action on M1 cells [40]. Furthermore, with the MI subline used in this laboratory, although LIF and IL-6 induce a comparable series of changes in M1 cells, 25-fold higher concentrations of IL-6 are required to produce equivalent responses. Neither G-CSF nor M-CSF has significant differentiation-inducingactions on M1 cells, but combination of either CSF with LIF leads to enhanced differentiation as, less surprisingly, does the combination of IL-6 with LIF [40]. When using a combination of M-CSF and LIF, clear increases are observed in the number and size of differentiating colonies. This may be based on the commitment of M1 cells to an immature, M-CSF-dependent, macrophage phenotype. There are other murine myeloid or macrophage leukemic cell lines that exhibit no obvious response to LIF action, e.g., WEHI3B cells; although in some cases, low levels of binding of LIF are detectable. The differentiation-inducing actions of LIF are not restricted to cells of the myelomonocyticlineage. With the erythroleukemic cell lines derived from rnyctransformed fetal liver cells, LIF induces increased hemoglobinizationof the cells when grown in the presence of dimethylsulfoxide (DMSO). The situation with human leukemias is somewhat similar to murine leukemias. On autoradiography, receptors have so far only been consistently found on leukemic cells differentiating into the monocytic lineage, e.g., in chronic myelogenous leukemia (CML), chronic myelomonocytic leukemia (CMML) and in M4 or M5 leukemias [27], although some blast cells in these leukemias also exhibit receptors. No overt action of purified LIF has been noted in agar cultures of a limited number of primary acute myelogenousleukemias (AML), but a suppressive action of crude preparations of recombinant LIF (COS cell-conditioned medium) has been described for CML colony formation [41]. This needs reinvestigation using purified LIE LIF exhibits some suppressive action on colony formation by HL60 or U937 cells with a reduction in clonogenic cell self-renewal, but with little morphological evidence of differentiation [42, 431. Combination of LIF with G-CSF,

Leukemia Inhibitory Factor

101

GM-CSF or IL-6 leads to enhanced suppression of these leukemic cell lines. In these studies, there was no consistent hierarchy of suppressiveregulators, the most active regulator depending on the cells under test. LIF has also been reported to increase the doubling time of clonogenic cells in human AML cell lines with evidence suggesting a prolongation of the nonS-phase of the cell cycle [MI. LIF differs from the CSFs and IL-6 in having no proliferative action on any human leukemias so far examined and remains, therefore, a candidate agent for clinical use in the biological suppression of at least some forms of myeloid leukemia.

Actions of LIF on Stem Cells Normal embryonic stem (ES) cells are initially totipotential and, upon reinjection into blastocysts, can generate cells of all tissues including gonadal stem cells. ES cells are thus ideal for genetic manipulation and the ultimate generation of transgenic mice with defined gene changes. However, ES cells exhibit a predisposition in culture to differentiation commitment with loss of totipotentiality. LIF, in concentrations similar to those used to induce differentiation in MI leukemic cells, has no influence on the proliferation of ES cells, but prevents differentiation commitment with maintenance of totipotentiality [18]. This action of LIF appears to be unique, and LIF is now in widespread use for this purpose. The production ofLIF, both by blastocyst cells themselves [36] and by the metrial gland at the site of blastocyst implantation [32], indicates that LIF probably plays an important role in early embryonic development. LIF has also been reported to increase the efficacy of retroviral infection of hemopoietic stem cells [45]. We have, to date, observed no proliferative effects of LIF on purified murine hemopoietic stem cells, and it is unclear what the basis might be for this reported action of LIE

Changes in Mice with Excess LIF Levels The astonishing range of pathological changes developing in mice with excess LIF levels provided early evidence that LIF was a highly unusual molecule with significant actions on many cell types. These studies were performed in DBA/2 mice and were based on previous observationsthat mice injected i,v. with cells of the non-leukemic, but immortalized and CSF-dependent, line FDC-P1 exhibit engraftment of the cells in the bone marrow, spleen and some lymphoid tissues where the cells reach a plateau of 510%of the population [46]. Retroviral insertion of the LIF cDNA with a strong MPSV LTR promoter into FDC-P1 cells and subsequent selection of cells producing high levels of LIF provided a population of cells that could serve as a resident factory producing

Metcalf

102

high levels of LIF in recipients engrafted with such cells. Such mice developed LIF levels of approximately lo00 Unitdm1 and died prematurely after 6-8 weeks with a curious state of apparent cachexia combined with hypermobility [47]. The apparent cachexia was based on complete loss of subcutaneous and abdominal fat and was not accompanied by weight loss of organs, such as liver and kidney, although the thymus was uniformly atrophic. The mice had a neutrophil leukocytosis, an elevated erythrocyte sedimentation rate and defective clot retraction. They exhibited striking new bone formation in the medullary cavity of all bones associated with the presence of massive numbers of osteoblasts. The reduced ability of the marrow to support hemopoietic populations was associated with spleen enlargement and marked extramedullary hemopoiesis in the spleen and to a lesser degree the liver [48].Serum calcium levels were elevated and the mice exhibited calcificationin skeletal and heart muscle and less often in the liver. The liver sometimes exhibited necrotic areas with or without cirrhosis. The pancreas was friable and reduced in volume with necrosis of many pancreatic acini but apparently intact islets and no evidence of diabetes. There was also loss of the inner zone of the adrenal cortex. In females, the ovaries exhibited apparently normal follicular maturation but a severe reduction and often complete absence of corpora lutea. In males, the testes showed a complete absence of spermatogenesis.

Changes in Mice Injected with LIF In mice injected three times daily with 2 pg LIF for 14 days, doses that proved to be in the 10%lethal range, marked body weight loss occurred in the first five days, but was not progressive thereafter. This was based on complete loss of subcutaneous and abdominal fat, and was not accompanied by organ weight changes either in the liver or kidney. Paralleling the body weight loss was an excitable and hypermotile state with ruffling of hair at the nape of the neck but without fighting [37]. These mice did not develop a neutrophil leukocytosis and showed no significant new bone formation, tissue calcification, pancreatic or gonadal changes. They did develop elevated erythrocyte sedimentation rates within six hours of first injection, an elevated serum calcium level and lowered serum albumin levels, an increase in megakaryocyte progenitors and subsequently in megakaryocytes, particularly in the spleen. Beginning on day 5, a rise occurred in blood platelets reaching levels twice those in control mice, with elevated levels being sustained for the period of LIF injections. In mice injected with lower doses of LIF, no body weight loss or behavioral changes were observed, but LIF again elicited rises in megakaryocyteand platelet levels. LIF had no platelet-releasing action and the earliest elevation of platelet levels was observed after injections of LIF for 5-7 days.

Leukemia Inhibitory Factor

103

Actions of LIF on Other Tissues Possible mechanisms for many of the changes observed in mice with excess LIF levels have now been established. The demonstration by Abe et a/. [20] that LIF can release calcium from bone tissue has been confirmed and extended using recombinant LIE LIF can have no direct action on osteoclasts because osteoclasts lack LIF receptors. Therefore, the calcium release, presumably due to osteoclast action, must be a consequence of perturbationof the known close functional linkage between osteoblastsand osteoclasts. Receptors for LIF are demonstrable on osteoblasts and LIF stimulates the proliferation of osteoblasts [21] and their production of alkaline phosphatase, and plasminogen activator inhibitor by actions that appear to be direct [22]. Adipocytes express receptors for LIF and the demonstration by Mori er al. [27] that LIF inhibits lipoprotein lipase-mediated lipid transport to such cells seems likely to be the basis for the remarkable loss of fat deposits following injection of high doses of LIF or the development of excess LIF levels. Recognition that LIF is the hepatocyte-stimulating factor 111, able to release acute phase proteins from liver parenchymal cells [26], provides a likely basis for the observed early rise in erythrocyte sedimentation rates and fall in serum albumin and may be the basis, in mice with excess LIF levels, for the observed necrotic and fibrotic changes in some livers. The hyperexcitable state observed with high LIF levels may be due to the ability of LIF to switch autonomic nerve signalling from adrenergic to cholinergic mode [24], although the elevated serum calcium levels might also play some role in this state. LIF also influences the production of other neuronal signalling peptides such as the vasoactive intestinal polypeptide [49]. This may be the basis for the curious development in C3H/HeJ mice injected with high doses of LIF, of a distended stomach with pylorospasm and a spastic small bowel. The explanation of the observed changes in the pancreas, adrenals and gonads is currently unknown. Receptors for LIF are present on the relevant adrenal cortical cells and Leydig cells, so direct actions could occur. It remains possible that some of the changes are mediated by the pituitary/hypothalamicaxis although no overt pathological changes were observed in the pituitary of mice with excess LIF levels.

Basic Biological Problems Posed by LIF LIF was purified on the basis of its ability to induce differentiation commitment in myeloid leukemic cells, yet it has exactly the opposite effect of preventing differentiationcommitment in normal embryonic stem cells. Further evidence of differentiation-committingactions of LIF are its action on erythroleukemiccells. How can these opposing actions be rationalized? One possible explanation is that a relatively simple genomic mechanism operates to control the choice be-

Metcalf

104

tween self-renewal or differentiation commitment, made when a cell is about to enter the S-phase of the cell cycle. LIF-initiated signalling may lead to the activation, or induce the transcription, of a nuclear transcription factor able to influence this switch gene with the actual outcome being dictated by other genetic programs active in the cell. This would assign to LIF a role as an activating regulator without any particular ability to dictate the outcome of a differentiation commitment decision. While differentiationcommitment in multipotentialhemopoietic stem cells exhibits a stochastic pattern [50],there is evidence that extrinsic regulators can significantlybias the outcome of this differentiationcommitment [51]. With this principle established, there is no particular conceptual problem in envisaging LIF as being able to have a similar action. The only unusual feature of the action of LIF is its capacity to elicit opposite effects according to the cell type responding. As is well documented for the CSFs, hemopoietic regulators are polyfunctional and can influence functions as diverse as cell division, differentiationcommitment, maturationinduction and functionalactivation [51]. LIF exhibits a similar poIyfunctionality (Table 111)being a proliferative stimulus for at least some hemopoietic cell lines and osteoblasts, and an accessory proliferative signal for other hemopoietic cells such as blast cells and megakaryocyteprecursors. In contrast, the actions of LIF on liver parenchymal cells or autonomic nerves are best characterized as influencing the functional activity of already differentiated cells. The polyfunctionality of LIF would now be considered unremarkable were its actions restricted to one cellular population, e.g., hemopoietic cells. What makes LIF such an enigmatic molecule are its actions on so broad a range of tissue cells. There is no obvious situation in which functional changes in this diverse array of tissues need to be influenced simultaneously and it is, therefore, puzzling why the body should elect to use a single regulator to achieve these changes. The pleiotropic actions of IL-6 [52] raise a similar general problem. One possible solution to the problem posed by LIF and IL-6 is to postulate that such molecules are intrinsically useful signalling molecules because of some, as yet unrecognized,physicochemicalproperty. They can be used as perfectly satisfactory regulators provided that the production and action of the molecules are strictly local. Thus astrocyte production of LIF may be a quite efficient method for controlling neuronal function provided that the LIF so produced does not enter the circulation and elicit other unwanted responses. Similarly, LIF production locally in the uterine wall at the site of blastocyst implantation would make sense if its intended function was to promote implantation and early embryo development. Even a response with systemic consequences, such as the production by the liver of acute-phase proteins, could also be fitted in to this pattern if the induced production of LIF was essentially localized to the liver. The action of LIF in stimulating megakaryocyte and platelet formation is a response able to be elicited by circulating LIF and raises the possibility of LIF being able to act as a conventional humoral regulator. However, if LIF production is normally strictly localized, local production of LIF by stromal cells in the

Leukemia Inhibitory Factor

105

Table 111. Multiple actions of LIF

Differentiation Commitment Induction in myeloid leukemic cells - M1. - HL60 - u937 Induction in myc-initiated erythroleukemic cells Suppression in normal embryonic stem cells Proliferative Actions DA-1 hemopoietic cells Osteoblasts Potentiation of IL-3 stimulation of - blast cell precursors - megakaryocyte precursors - erythroleukemic cells Functional Actions Calcium release from bones Stimulation of new bone formation Production of acute phase protein by hepatocytes Switching of signalling by autonomic neurones Depletion of body fat Induction of gonadal dysgenesis Elevation of platelet levels

marrow or spleen could represent a workable method for controlling (at least in part) platelet formation, without LIF needing to enter the circulation and producing unwanted changes in other tissues. The general alternative to these more conventional explanations is to postulate that, with agents like LIF or IL-6, we are confronted by regulators that require a major reassessment of our ideas regarding the functions and interactions of various organs. Classical hormones, such as insulin or growth factor, cross organ systems in their actions. Possibly, LIF and IL-6 are representatives of yet another class of inter-organ regulators-the logic for which is not yet evident because our concepts of cell biology are too tissue oriented.

Clinical Implications of LIF Despite its lack of direct proliferative effects when acting alone, LIF has significant potentiating effects on the proliferation of hemopoietic blast cells and megakaryocyte progenitors, and is able to elevate platelet levels in vivo at doses having no obvious toxic effects. LIF is therefore a candidate regulator for clinical use, particularly in elevating platelet levels and possibly for accelerating early hemopoietic regeneration. Because of its lack of proliferative effects on myeloid leukemic cells, LIF remains a candidate for use in the biological suppression of at least some forms

Metcalf

106

of myeloid leukemia, probably in combinationwith other agents such as the colonystimulating factors or IL-6. The wide variety of disease states developing in the presence of chronic excess levels of LIF raises the question of whether LIF over-production, either systemic or local, might sometimes be a factor in the development of these diseases. Included in this group are: cirrhosis and hepatic necrosis, pancreatitis, pylorospasm, osteosclerosis, gonadal dysgenesis and sterility. A survey of LIF levels in these disease states is clearly warranted, examining both circulating LIF levels and levels of LIF transcription in tissue biopsies. The curious ability of LIF to induce loss of body fat raises the possible use of LIF to reverse atheromatous plaques, while the influence of LIF on neuronal signalling may be of potential interest in a variety of neurological diseases.

References 1 Ichikawa Y. Differentiation of a cell line of myeloid leukemia. J Cell Physiol 1969; 741223-234. 2 Ichikawa Y. Further studies on the differentiation of a cell line of myeloid leukemia. J Cell Physiol 1970;76:175-184. 3 Metcalf D, Moore MAS, Warner NL. Colony formation in vitro by myelomonocytic leukemic cells. JNCI 1969;43:983-1001. 4 Metcalf D. Clonal analysis of the action of GM-CSF on the proliferation and differentiation of myelomonocytic leukemic cells. Int J Cancer 1979;24:616-623. 5 Nicola NA, Metcalf D, Matsumoto M, Johnson GR. Purification of a factor inducing differentiation in murine myelomonocyticleukemic cells: identification as granulocyte colony-stimulating factor (G-CSF). J Biol Chem 1983;258:9017-9023. 6 Lipton JH,Sachs L. Characterizationof macrophage and granulocyte inducing proteins for normal and leukemic myeloid cells produced by the Krebs ascites tumor. Biochem Biophys Acta 1981;673:552-569. 7 Shabo Y, Lotem J, Rubinstein M, et al. The myeloid blood cell differentiation-inducing protein MGI-2A is interleukin-6. Blood 1988;72:2070-2073. 8 Tomida M, Yamamoto-YamaguchiY and Hozumi M. Purification of a factor inducing differentiation of mouse myeloid leukemic M1 cells from conditioned medium of mouse fibroblast L929 cells. J Biol Chem 1984;259:10978-10982. 9 Hilton DJ, Nicola NA and Metcalf D. Purification of a murine leukemia inhibitory factor from Krebs ascites cells. Anal Biochem 1988;173:359-367. 10 Lowe DG, Nunes W, Bombara M, et al. Genomic cloning and heterologous expression of human differentiation-stimulatingfactor. DNA 1989;8:351-359. 11 Metcalf D, Hilton DJ and Nicola NA. Clonal analysis of the action of the murine inhibitory factor on leukemic and n o d murine hemopoietic cells. Leukemia 1988; 2:216-221. 12 Lotem J, Shabo Y and Sachs L. Clonal variation in susceptibility to differentiation by different protein inducers in the myeloid leukemia cell line, M1. Leukemia 1989; 3:804-807. 13 Gearing DP, Gough NM, King JA, et al. Molecular cloning and expression of cDNA encoding a murine myeloid leukemia inhibitory factor (LIF). EMBO J 1987;6: 3995-4002.

Leukemia Inhibitory Factor

107

14 Gough NM, Gearing DP, King JA, et al. Molecular cloning and expression of the human homologue of the murine gene encoding myeloid leukemia inhibitory factor. Proc Natl Acad Sci USA 1988;85:2623-2627. 15 Stahl J, Gearing DP, Willson TA, Brown MA, King JA and Gough NM. Structural organization of the genes for murine and human leukemia inhibitory factor (LIF): evolutionary conservation of coding and non-coding regions. J Biol Chem 1990;265: 8833-8841. 16 Koopman P, Cotton RGH. A factor produced by feeder cells which inhibits embryonal carcinoma cell differentiation.Characterizationand partial purification. Exp Cell Res 1984;l54:233-242. 17 Smith AG, Heath JK, Donaldson DD, et al. Inhibition of pluripotential embryonic stem cell differentiation by purified polypeptides. Nature 1988;336:688-690. 18 Williams IU, Hilton DJ, Pease S, et al. Myeloid leukemia inhibitory factor (LIF) maintains the developmentalpotential of embryonic stem cells. Nature l988;336:684-687. 19 Moreau J-F, Donaldson DD, Bennett F, Witek-Gianotti JA, Clark SC and Wong GG. Leukemia inhibitory factor is identical to the myeloid growth factor human interleukin for DA cells. Nature 1988;336:690-692. 20 Abe E, Tanaka H, Ishimi Y, et al. Differentiation-inducingfactor purified from conditioned medium of mitogen-treated spleen cell cultures stimulates bone resorption. Proc Natl Acad Sci USA 1986;83:5958-5962. 21 Reid IR, Lowe C, Cornish J, et al. Leukemia inhibitory factor: a novel bone-active cytokine. Endocrinology 1990;126:1416-1420. 22 Allan EH, Hilton DJ, Bmwn MA, et al. Osteoblasts display receptors for and responses to leukemia inhibitory factor. J Cell Physiol 1990;145:110-119. 23 Yamamori T, Fukada K, Aebersold R, Korsching S, Fann M-J, Patterson PH. The cholinergic neuronal differentiation factor from heart cells is identical to leukemia inhibitory factor. Science 1989;246:1412-1416. 24 Gauldie J, Richards C, Harnish D, Lansdorp P, Baumann H. Interferon P2IB-cell stimulating factor type 2 shares identity with monocyte-derived hepatocyte-stimulating factor and regulates the major acute phase protein response in liver cells. Proc Natl Acad Sci USA 1987;84:7251-7255. 25 Baumann H, Wong GG. Hepatocyte-stimulatingfactor III shares structural and functional identity with leukemia-inhibitory factor. J Immunol 1989;143:1163-1167. 26 Mori M, Yamaguchi K, Abe K. Purificationof a lipoproteinlipase-inhibitingprotein produced by a mefanomacell line associated with cancer cachexia. Biochem Biophys Res Commun 1989;160:1085-1092. 27 Metcalf D, Gough NM, Stahl J, Hilton DJ, Nicola NA. Leukemia inhibitory factor. In: Agganval BB, Gutterman JU, eds. Human Cytokines: Handbook for Basic and Clinical Researchers. Cambridge, MA: Blackwell Scientific, 1991 (in press). 28 Kola I, Davey A, Gough NM. Localization of the murine leukaemia inhibitory factor gene near the centromere on chromosome 11. Growth Factors 1990;2:235-240. 29 Sutherland GR, Baker E, Hyland VJ, Callen DF, Stahl J, Gough NM. The gene for human leukemia inhibitory factor (LJF) maps to 22ql2. Leukemia 1989;3:9-13. 30 Hilton DJ, Nicola NA, Metcalf D. Specific binding of murine leukemia inhibitory factor to normal and leukemic monocyticcells. Proc Natl Acad Sci USA 1988;85:5971-5975. 31 Hilton DJ, Nicola NA, Metcalf D. Distribution and comparison of receptors for leukemia inhibitory factor on murine hemopoietic and hepatic cells. J Cell Physiol 1991 (in press). 32 Cmy BA, Guilbert LT, Brown MA, et al. Characterizationof @kine production by the metrial gland and granulated metrial gland cells. J Reproduc Immunoll990 (in press). 33 Anegon I, Moreau JF, Godard A, et al. Production of human interleukin for DA cells (HILDA) by activated monocytes. Cell Immunol 1990;130:50-65.

Metcalf

108

34 Wetzler M, Talpaz M, Lowe DG, Baiocchi G, GuttermanJU, Kurzrock R. D-factor, a hematopoietic and embryonal cell differentiationregulatory factor: Constitutiveexpression by human bone marrow stromal cells and modulation by IL-1,TNFa and TGFp Blood 1990 (in press). 35 Wesselingh SL, Gough NM, Finlay-Jones JJ, McDonald PJ. Detection of cytokine mRNA in astrocyte cultures using the polymerase chain reaction. Lymphokine Research 1990;9:177-185. 36 Murray R, Lee F, Chiu C-P. The genes for leukemia inhibitory factor and interleukind are expressed in mouse blastocysts prior to the onset of hemopoiesis. Mol Cell Biol 1990;10:4953-4956. 37 Metcalf D, Nicola NA, Gearing DP. Effects of injected leukemia inhibitory factor (LIF) on hemopoietic and other tissues in mice. Blood 1990;76:50-56. 38 Leary AG, Wong GG, Clark SC, Smith AG, Ogawa M. Leukemia inhibitory factor differentiation-inhibitoryactivitylhuman interleukin for DA cells augments proliferation of human hematopoietic stern cells. Blood 1990;75:1960-1964. 39 Godard A, Gascan H, Naulet J, et al. Biochemical characterization and purification of HILDA, a human lymphokine active on eosinophils and bone marrow cells. Blood 1988;71:1618-1623. 40 Metcalf D. Actions and interactionsof G-CSF, LIF and IL-6 on normal and leukemic murine cells. Leukemia l989;3:349-355. 41 Verfaille CM, McGlave BP. Leukemia inhibitory factor (LIF)/humaninterleukin for DA cells (HILDA) induces growth and proliferative hematopoieticprogenitors from normal human bone marrow but suppresses growth of progenitors from CML bone marrow. Exp Hematol 1990;18:646. 42 Maekawa T, Metcalf D. Clonal suppression of HL60 and U937 cells by recombinant leukemia inhibitory factor in combinationwith GM-CSF or G-CSF. Leukemia 1989; 3 :270-276. 43 Maekawa T, Metcalf D, Gearing DP. Enhanced suppression of human myeloid leukemic cell lines by combinations of IL-6, LIF, GM-CSF and G-CSF. Int J Cancer 1990;45:353-358. 44 Wang C, Lishner M, Minden MD, McCulloch EA. The effects of leukemia inhibitory factor (LIF) on the blast stem cells of acute myeloblastic leukemia. Leukemia 1990;4:548-552. 45 Fletcher FA, Williams DF, Miliszewski C, Anderson D, Rives A, Belmont JW. Murine leukemia inhibitory factor enhances retroviral-vectorinfection efficiency of hematopoietic progenitors. Blood 1990;76:1098-1103. 46 Diihrsen U, Metcalf D. Effects of irradiation of recipient mice on the behavior and leukemogenic potential of factor-dependenthematopoieticcell lines. Blood 1990;75: 190-197. 47 Metcalf D, Gearing DP. A fatal syndrome in mice engrafted with cells producing high levels of leukemia inhibitory factor. Proc Natl Acad Sci USA 1989;86:5948-5952. 48 Metcalf D, Gearing DP. A myelosclerotic syndrome in mice engrafted with cells producing high levels of leukemia inhibitory factor (LIF). Leukemia l990;3:847-852. 49 Nawa H, Yamamori T, Le T, Patterson PH. The generation of neuronal diversity: Analogies and homologies with hematopoiesis. Cold Spring Harbor Symposia on Quantitative Biology 1990 (in press). 50 Suda J, Suda T, Ogawa M. Analysis of differentiation of mouse hemopoietic stem cells in culture by sequential replating of paired progenitors. Blood l984;64:393-399. 51 Metcalf D. The molecular control of cell division, differentiation commitment and maturation in haemopoietic cells. Nature l989;339:27-30. 52 Kishimoto T. The biology of interleukind. Blood l989;74:1-10.