Leukemia (1998) 12, 1937–1943  1998 Stockton Press All rights reserved 0887-6924/98 $12.00 http://www.stockton-press.co.uk/leu

Phenotypic effect correlating with loss of a novel tumor suppressor gene: towards cloning by complementation WD Cook and BJ Alexander University of Melbourne Dept of Surgery, Royal Melbourne Hospital, Australia

We intend to use a gene complementation approach to clone a tumor suppressor gene on mouse chromosome 2, the loss of which contributes to myeloid leukemia. An in vitro model system has been generated using a clonal cell line, in which tumorigenic chromosomal lesions have been selected along with myeloid differentiation. Among these lesions are deletions of chromosome 2. Comparison of subclones with deleted vs intact chromosomes 2 has allowed the identification of a growth related phenotypic pattern which correlates with the deletion, viz the retention of a marker of immature cells, resistance to inhibition by lipopolysaccharide (LPS), even in the presence of markers of mature myeloid cells, such as resistance to killing by apoptosis-inducing agents. The phenotype is shared by chromosome 2-deleted cell lines derived from conventional tumors. We have begun to investigate the mechanism of the phenotype. The LPS resistance does not correlate with lack of mRNA for CD14, a known cell surface receptor for this agent, or with failure to induce TNF␣ or nitric oxide synthase in response to its binding. The system should allow cloning of the gene using complementation of this phenotype in transfected cell lines. Keywords: tumor suppressor gene; myeloid leukemia; mouse chromosome 2; gene complementation

Introduction The identification of tumor suppressor genes (TSG) is essential to progress in identifying the many paths by which normal cells become malignant. Because the existence of a TSG is heralded by its absence from certain tumors,1 locating and cloning a new one is a formidable exercise. The first stage is relatively straightforward, ie to define as small as possible a commonly deleted region (CDR) in a large number of independent tumors. Using chromosomally mapped microsatellite markers,2 a CDR of a centimorgan (cM) or less can be generated. However, this still presents a challenging second stage, generally carried out by sequence analysis of a large fraction of the CDR DNA. In order to avoid the most laborious aspects of this, we have developed a gene complementation approach. We have started with the proposition that it is possible to determine the biological effect of an unknown TSG by comparing the phenotype of tumor cells bearing the chromosomal deletions of interest with that of appropriate non-deleted control cells. This proposition depends on three preconditions: the availability of suitable control cells, the detectability of the ‘signal’, ie the biochemical effect of the gene product, and a lack of ‘background’, ie competing biochemical effects of other genes lost in the deletions. Control cells for this exercise need to be clonal populations of non-deleted cells that are otherwise phenotypically and genotypically comparable to the deleted tumor cells. An ideal

Correspondence: WD Cook, C/- Dept of Surgery, Royal Melbourne Hospital, Victoria 3050, Australia; Fax: 61 3 9347 6488 Received 19 March 1998; accepted 19 August 1998

model would be a clonal cell line in which the deletions could be selected. Our approximation of this is based on a murine hemopoietic cell line which we have previously described, RB22.2.3 Although the undifferentiated parent line is transformed by Abelson leukemia virus, the continued rapid growth of differentiated macrophage-like subclones seems to depend on the acquisition of tumorigenic karyotypic abnormalities,4 viz trisomy of chromosomes (chrs) 5 or 12,5 translocation of chr 166 or deletion of chr2.7–10 The chr2 aberrations have been known for 20 years to contribute to myelogenous leukemia.7 We have confirmed the interpretation of others that deletion of a putative TSG on this chromosome is an important tumorigenic event.4,11 We have chosen to compare our chr2-deleted and non-deleted clonally related lines to help identify and clone the TSG lost in this deletion. The second requirement is that appropriate culture conditions can be created with which to reveal the biological effect of the TSG in vitro. The TSG thus far identified show a limited and overlapping range of mechanisms of action, including inhibition of growth (P53, Rb, WT-1, p16, CDK4, BRCA1, MET, NF1), induction of apoptosis (p53, WT1), and repair of DNA damage (MLH1, MSH2, PMS1 and 2, BRCA1 and 2) (reviewed in Ref. 12). Since the chr2 deleted tumors lack the microsatellite instability indicative of loss from the last group of genes (Ref. 13 and our unpublished results), we chose to examine our cell lines for responses to growth and apoptotic signals. It is likely that the third requirement for this approach, ie that there is minimal background effect caused by loss of surrounding genes, will be realized given the expectation that both alleles of a TSG must be inactivated to induce a tumorigenic effect.1,14 This most often occurs by deletion of a large (ie cytogenetically detectable) region of one chromosome, combined with a more specific mutational inactivation of the second allele of the TSG on the homologous chromosome. Thus the deletion in any given tumor signals inactivation of single alleles of most genes in the region, but of both alleles of the TSG. It is a reasonable expectation that the growth or apoptosis-related biological effect of homozygous loss of a TSG will be greater than that of hemizygous loss of any of the surrounding genes. Thus, if a common positive effect on cell growth or survival can be found for the loss of many overlapping but non-identical deletions in independent tumors, this effect is very likely to result from the loss of the TSG. Here we describe how, by comparing chr2-deleted subclones of RB22.2 with their non-deleted counterparts, we have identified a growth-related phenotype associated with the deletion. We have previously described the generation of an independent series of conventional radiation-induced myeloid tumors, from which cloned cell lines bearing chr2 deletions have been derived.11 These have been used to confirm the phenotype ascribed to the deletion. Thus we have developed a cell biological assay which may allow the cloning of the chr2 TSG by complementation.

Effect of chromosomal deletion WD Cook and BJ Alexander

1938

Materials and methods The origin and maintenance of tumors and cell lines have been described previously.4,11 Briefly RB22.2 is a BALB/c hemopoietic tumor cell line transformed by Abelson leukemia virus.3 Its derivatives were selected as adherent macrophagelike subclones appearing either spontaneously or after various drug treatments.4 The RSS and RXS lines were derived from spleen tumors induced by radiation in SJL/J and SJL/J×CBA mice, respectively.11 Subclones were derived by limiting dilution. RSS clones 1p5.1, 4p2.1, 5p1.4, 6p2.1, 7p2.3 and 8p2.1, used throughout this study, are referred to without passage and subclone numbers. Maintenance medium was DME with 10% calf serum supplemented with iron (Life Technologies, Melbourne, Australia) and 50 ␮M 2-mercaptoethanol.

Growth inhibition assays Cells were plated at 1000 (RB22.2 derivatives) or at 10 000 (radiation-induced tumor lines) cells per well in 96-well flat bottom plates in growth medium. Groups of six replicate wells were either left untreated or exposed to varying doses of agents as described in the Figure legends. Irradiation was performed before plating, using a 60 Co source at the doses indicated. Lipopolysaccharide (LPS) was from E. coli strain F583 (Sigma, Castle Hill, Australia). Purified recombinant murine TNF-␣, IFN␥, GM-CSF and IL-6 were kindly provided by J Rasko and N Nicola (WEHI). On day 2 for the NaN3-treated cells, or day 3 for all other assays, cell concentrations were estimated using an MTS colorimetric assay (Promega, Sydney, Australia) according to the maker’s instructions, with OD readings at 496 nm using a Titertek Multiskan (Labsystems, Finland) plate reader.

primers.17 Probes for TNF␣ and GAPDH were as previously described.4,11 The MK probe was a 1 kb fragment spanning the coding region,18 and kindly supplied by Dr Hiroshi Maruta (Ludwig Institute, Melbourne, Australia).

Results

In vitro derived deletions of chromosome 2 The derivation and the phenotypic and cytogenetic characterization of RB22.2 and its macrophage-like subclones have been described previously.3,4 Briefly, the panel used here consists of three groups: Group 1: The parent clone, RB22.2 and its parental-type subclone .5bl2. These display an immature myeloid/ lymphoid phenotype and approximately diploid karyotype, including intact chrs2. Group 2: The macrophage-like RB22.2 subclones (a71, b2.1, d3.1, d3.1.3, d2.2, 13.1.1) in which chrs2 appear not to be deleted, but which display trisomy of chr5 or 12, or translocation of chr16, among other karyotypic aberrations. Group 3: The macrophage-like subclones bearing chr2 deletions: Stella (a subclone of 13.1.1) and Rd5.1 (a subclone of .5bl2). Our aim was to find a growth or apoptosis-related phenotypic criterion, or combination thereof, which consistently distinguished between groups 2 and 3, ie between the two groups of macrophage-like RB22.2 subclones, bearing intact vs hemizygously deleted chrs2, respectively.

The phenotypic pattern of resistance to both LPS and apoptosis-inducing agents distinguishes chr2-deleted RB22.2 derivatives

Northern blot analysis PolyA(+) mRNA was extracted and analyzed by Northern blot as previously described (Ref. 4, except that Genescreen filters were probed in the presence of Quikhyb (Stratagene, La Jolla, CA, USA), and washed as described by Chomczynski.15 Filters were reprobed with GAPDH to control for RNA loading. Signal intensities were measured using a phosphorimager (Fuji, Melbourne, Australia).

Quantitative Southern blot analysis Quantitative Southern blot analysis was performed as described,11 with the following modifications: DNA samples were digested with HindIII enzyme, and quadruplicate tracks were electrophoresed for each sample. Filters were probed and washed as for Northern blots described above. Normalization of signal was achieved by comparison of intensities of the 4 kb band containing the Midkine (MK) gene and the 3 kb band containing the pseudo-Midkine (−MK) gene, which is known to be encoded on chr11 in the mouse.16

Molecular probes A probe for murine CD14 was generated by RT-PCR amplification from bone marrow macrophage mRNA using published

Subclones from each group were compared by several phenotypic criteria. The myeloid cell cytokines IL-6 (10–100 U/ml), GM-CSF (60–6000 U/ml), TNF␣ (20–200 U/ml) and IFN␥ (1– 100 U/ml) failed to distinguish between the groups in growth assays described in Materials and methods (data not shown). The cell lines were not examined to determine whether this could be explained by lack of receptor expression. Three characteristics distinguished all the macrophage-like subclones (groups 2 and 3) from the immature group 1 lines. Regardless of the presence of chr2 deletions, all the macrophage-like clones grew more slowly than the parental clones (group 1) both in vitro, and in vivo in syngeneic (BALB/c) mice (data not shown). We had previously shown that the macrophage-like clones expressed increased levels of myeloidspecific markers.4 Similarly, the macrophage-like lines were several-fold more resistant to all apoptotic agents tested, ie dexamethasone, irradiation and NaN3. Again, the degree of resistance was not related to the presence or absence of chr2 deletions (Figure 1a). Bacterial LPS is known to inhibit growth of many mammalian cell types.19–23 In the myeloid lineage, it has been shown to inhibit growth and to stimulate differentiation of mature human and mouse macrophages, both normal24,25 and transformed,26–28 and by contrast to co-stimulate growth of some early monocytic progenitors.25 IFN␥ has been shown to synergize with LPS in this activity.29 Consistent with this, the non-deleted group 2 macrophage-like clones were consider-

Effect of chromosomal deletion WD Cook and BJ Alexander

ably inhibited, showing approximately 50% and 20–40% of the growth of untreated cells in response to LPS alone (data not shown) and to LPS and IFN␥, respectively, and the immature group 1 subclones were resistant to inhibition by either treatment. By contrast, the group 3 chr2-deleted clones, which showed other characteristics of mature macrophages, were resistant to, and in some cases stimulated by, LPS with or without IFN␥ (Figure 2, upper panel). The mechanism of the sensitivity to LPS was not induction of cell death. Sensitive sublines could be cloned in the presence of 30 ␮g/ml LPS. Though their cloning efficiency (15– 60%) was reduced compared to that of resistant sublines (70– 90%), the most noticeable effect was the reduction in cell number per colony in the LPS-treated plates to approximately 10% of untreated.

LPS can select for chr2 deletion We were able to use cell growth inhibition by LPS to select a third in vitro-deleted subclone from .5d2.2, one of the sensitive subclones. Compared to control untreated cultures, the cloning efficiency of .5d2.2 in LPS was 53%, but average clone size was considerably reduced. Of a total of 84 LPStreated clones examined, four stood out as growing rapidly. These were harvested, grown and re-assayed for LPS sensitivity. Only one, designated .5d2.2LR1 (LR1), proved comparably resistant to the deleted subclones (Figure 2, upper panel). Since deletion of chr2 was not detectable in LR1 by cytogenetic analysis (data not shown), it was assessed molecularly by quantitative Southern blot analysis with probing for the midkine gene (MK), which was predicted to be within the CDR on chr2. The left panel of Figure 3 confirms that MK is hemizygously deleted in the two radiation-induced tumors RXS12 and RXS10.1 which as previously described, define a deleted region of 7.6 cM.11 The right panel shows that whereas the parent line RB22.2 and the LPS-sensitive .5d2.2 each bear two copies of the MK gene, the LPS-resistant subclone LR1 is hemizygously deleted at this locus.

LPS-resistant phenotype is present in radiationinduced leukemic cell lines bearing in vivo-derived chr2 deletions

Figure 1 Apoptotic agents select for maturity, whether in RB22.2 subclones or radiation-induced lines, and irrespective of chr2 deletions. Microtiter cultures were initiated at 1000 cells (RB22.2 derivatives) or 10 000 cells (RSS lines) per well in sextuplicate and exposed to the agents shown. Growth was assessed at 2 days (NaN3), or 3 days (dex, irradiation) by an MTS assay as described in Materials and methods. Results are plotted as mean percent (±s.d.) of growth of untreated wells for each cell line. (a) As indicated in the key in the lowest panel, dashed lines represent group 1 subclones RB22.2 and .5bl2; dotted lines, group 2 macrophage-like subclones with chrs2 intact; solid lines, group 3 chr2-deleted macrophage-like subclones Stella and Rd5.1. Since subclones within the groups responded so similarly, they have not been individually identified. (b) Similar analysis to the top panel of (a), comparing radiation-induced RSS cell lines to RB22.2 and its group 3 subclone Rd5.1. Note that the only radiation-induced tumor cell line to show inhibition was RSS1.

We have described previously the induction of myeloid leukemia by radiation according to the protocols of others, and the adaptation of the tumors to cell culture.11 We examined 15 such cell lines, 14 of which were cloned, for sensitivity to LPS and IFN␥. As shown for seven of the lines in the lower panel of Figure 2, every line was resistant to growth inhibition, comparable to the deleted RB22.2 subclones, and several of the lines were even stimulated by the treatment. Similarly, we confirmed the observation of Rasko et al30 that growth of the chr2-deleted tumor cell line IGM36 is stimulated by LPS (data not shown). We assessed the responses of the radiation-induced myeloid leukemia cell lines to the apoptosis-inducing agent, dexamethasone. All but one of these lines were resistant (Figure 1b), comparable with the macrophage-like RB22.2 derivatives.

The LPS effect is not explained by lack of CD14 LPS receptors or failure to induce autocrine gene products The cellular response to LPS has been shown to be complex and remains incompletely understood.31,32 Though it is

1939

Effect of chromosomal deletion WD Cook and BJ Alexander

1940

Figure 2 LPS selects for immature and chr2-deleted macrophage-like lines. Upper panel: RB22.2 and derivatives were plated as for Figure 1 but in the presence of 100 U/ml IFN␥, treated with the indicated doses of LPS and assayed after 3 days. The subclone .5bl2 represents group 1; 13.1.1 and d2.2 are group 2; Stella, Rd5 and d2.2LR1 are group 3. Note that derivation of d2.2LR1 is described in Results. Lower panel: radiation-induced tumor cell lines RSS1,4,5,6,7,8, solid lines; RB22.2, dashed line; group 2 subclone b2.1, dotted line; and group 3 subclone Rd5.1, solid line with filled triangles. All were treated and assayed as for the upper panel.

thought that there may be several cell surface receptors for LPS, only one, CD14, has been clearly demonstrated and cloned.33,34 We tested whether the resistance of our deleted lines could be due to a lack of expression of this receptor. As shown in Figure 4, this was unlikely to be the case. Although the group 1 RB22.2 lines indeed showed minimal expression consistent with their immaturity (columns 1 and 2), among the macrophage-like RB22.2 derivatives and the radiationinduced leukemia cell lines there was no correlation between sensitivity to LPS and the levels of mRNA for CD14. A mechanism for the synergy between LPS and IFN␥ has been proposed by others.29 LPS induces the expression of cellular TNF␣, which cooperates with IFN␥ to induce response genes including inducible nitric oxide synthase (iNOS). Therefore, we tested whether the resistance to LPS of our deleted lines might be due to a failure of LPS to induce either TNF␣ or iNOS. Although this was true for the majority of the deleted lines, there was one exception, RSS5, which was resistant to growth inhibition by LPS and IFN␥ despite

abundant induction of both TNF␣ and iNOS (Figure 5 and unpublished results). This suggests that the putative TSG product does not act via these pathways, consistent with the conclusion that its loss also confers resistance to LPS alone. Discussion Despite the fact that TSG are generally highly conserved,35,36 the chr2 deletion is rare in being identified in mouse tumors.7 The pattern of synteny with human chromosomes is complex in the vicinity of the deletion,37 so that prediction of the location of the human homologue is difficult. Meanwhile, the mouse model has allowed us to take the dual approaches of fine mapping of the gene, as we have done in parallel work (Ref. 11 and unpublished data) and gene complementation. As a first step in the latter approach, the experiments described in this paper were designed to identify a biological effect of the deletion, in order to generate a rapid in vitro assay for the gene product.

Effect of chromosomal deletion WD Cook and BJ Alexander

1941

Figure 3 Chr2 is deleted in LPS-selected subclone. Quantitative Southern blots were performed as described in Materials and methods, with probing for the MK gene. Results shown are mean relative intensities (±s.d.) of MK compared with ␺MK, which is encoded on chr11. Columns represent results for DNA extracted from: 1, (SJL/J CBA) F1 mouse kidney; 2, RXS12 tumor; 3, RXS10.1 cell line; 4, BALB/c mouse kidney; 5, RB22.2; 6, .5d2.2; 7, .5d2.2LRl.

Figure 4 Growth inhibition by LPS does not correlate with production of mRNA for CD14. PolyA(+) mRNA was isolated from logarithmically growing cells. After electrophoresis and Northern gel blotting of 2 ␮g of each mRNA, the filter was sequentially probed, without stripping, for CD14 and GAPDH (a), and results expressed as relative intensities of these two signals as measured by phosphor- image analysis (b). Cell lines: Intact chrs2: lane 1, RB22.2; 2, .5bl2; 3, 13.1.1; 4,a7l; 5, .5b2.1; 6, .5d2.2; 7 .5d3.1; 8, .5d3.1.3; Deleted chr2: lane 9, stella; 10, Rd5.1; 11, RSS4; 12, RSS8; 13, RXS14.1.

Using our in vitro model, we have shown that the deletion correlates with a phenotype which is consistent with loss of a TSG, namely resistance to the growth inhibitor LPS. Given that such resistance is also a characteristic of immature myeloid cells (Ref. 25 and this paper), it has been important to establish an independent marker to distinguish our deleted cells from these. Resistance to apoptosis-inducing agents serves this purpose. Thus, whereas the immature RB22.2 lines show resistance to LPS and susceptibility to apoptosis, the mature macrophage-like lines show the reverse pattern, along with other markers of macrophage differentiation. The chr2-

Figure 5 Susceptibility to growth inhibition by LPS does not correlate with induction of mRNA for TNF␣. Logarithmically growing cultures of the cell lines indicated were treated with 30 ␮g/ml LPS for the indicated times before harvesting for isolation of polyA(+) mRNA. 2 ␮g of each mRNA was analyzed by Northern blot, with successive probing, without stripping, for TNF␣ and GAPDH (a). (b) Results are expressed as relative intensities of these two signals, as measured by phosphorimage analysis. Cell lines: a71 and d3.1 are RB22.2 group 2, and sensitive to inhibition by LPS. RSS5, RSS6 and RXS8.1 are radiation induced chr2 deleted lines, and resistant to LPS.

deleted lines conform with the macrophage phenotype, with the exception that they are resistant to LPS. Our working hypothesis is therefore that the TSG encodes a product which participates in a growth inhibitory response to LPS. Thus far we have not identified the relevant response pathway. Our observations suggest, but do not prove, that it is neither control of CD14 receptor expression nor induction of TNF␣ or iNOS. Our observations also do not exclude the possibility that the putative TSG product can be activated by the binding of other ligands, such as cytokines, including those for which our cell lines may have lost receptors. For example, Resnitzky et al38 claimed that this TSG is involved in IL6 signalling, although they examined many fewer (three) deleted cell lines, and only one, unrelated, control cell line (M1). We confirmed that the 15 radiation-induced leukemia cell lines bearing chr2 deletions share the LPS resistant phenotype. Although the hematological classification of these tumors has not been completed, their morphologies vary, and include myeloblastic, promyelocytic and monoblastic features (B Alexander, PhD thesis, University of Melbourne 1997). Given that the LPS responses of the equivalent normal cells are unknown, the full significance of the LPS resistance of the tumor cells is uncertain. Nevertheless, they are resistant, and all but one of them are also distinguishable from the RB22.2 immature cell type in their response to apoptosis-inducing agents. These observations are consistent with our hypothesis. The implications for our gene complementation experiments are clear: we choose to transfect deleted lines, of both RB22.2 class 3 and the radiation-induced tumors, which show resistance to both LPS and apoptosis. We select for transfectants using a vector marker such as G418, and we examine these for loss of resistance to LPS along with retention of resistance to apoptosis. Given the complexity of growth control mechanisms, it is a reasonable expectation that loss of the putative chr2 TSG

Effect of chromosomal deletion WD Cook and BJ Alexander

1942

would be one of several genetic events which could render sensitive cells resistant to LPS. We indeed isolated a novel deleted subclone by selection for growth in LPS, consistent with, but not proof of, the hypothesis. We are in the process of applying a second test, ie whether the LPS resistance is recessive, as predicted for gene loss. Perhaps the most remarkable aspect of this work is that an effect of the deletion could be detectable at all in the RB22.2 subclones, given that they contain and express the v-abl oncogene. This phenomenon presumably depends on the compromised growth of the macrophage derivatives of the cell line. It implies that this same model can be used to identify the genes involved in the other ‘rescuing’ genetic events we have documented in the RB22.2 macrophage subclones, particularly translocation of chr16.4 It also implies that other carcinogenic genetic events could be selected and analyzed in established tumor cell lines. Acknowledgements This work was supported by grants from the Anti-Cancer Council of Victoria, the Lyndal and Jean Skea Foundation and the Walter and Eliza Hall Institute for Medical Research, to the last of whom we are also grateful for gifts of growth factors. WDC is indebted to Louise Buckle for many very helpful discussions, and to Alan Harris (WEHI) for reading the manuscript. References 1 Levine A. The tumour suppressor genes. Ann Rev Biochem 1993; 62: 623–651. 2 Whitehead Institute for Biomedical Research/MIT Center for Genome Research web site: http:www.genome.wi.mit.edu/. 3 Cook W, Balaton A. T cell receptor and immunoglobulin genes are rearranged together in Abelson virus-transformed pre-B and pre-T cells. Mol Cell Biol 1987; 7: 266–272. 4 Alexander B, Berger R, Day L, Hogarth PM, Feneziani, A, Cook WD. Tumor-associated karyotypic lesions coselected with in vitro macrophage differentiation. Genes, Chromos Cancer 1992; 5: 286–298. 5 Voncken JW, Morris C, Pattengale P, Dennert G, Kikly C, Groffen J, Heisterkamp N. Clonal development and karyotype evolution during leukemogenesis of BCR/ABL transgenic mice. Blood 1992; 79: 1029–1036. 6 Moroy T, Fisher P, Lee G, Achacoso P, Wiener F, Alt FW. High frequency of myelomonocytic tumors in aging E L-myc transgenic mice. J Exp Med 1992; 175: 313–322. 7 Azumi JI, Sachs L. Chromosome mapping of the genes that control differentiation and malignancy in myeloid leukemic cells. Proc Natl Acad Sci USA 1977; 77: 3630–3634. 8 Hayata I, Seki M, Yoshida K, Hirashima K, Sado T, Yamagiwa J, Ishihara T. Chromosomal aberrations observed in 52 mouse myeloid leukemias. Cancer Res 1983; 43: 367–373. 9 Trakhtenbrot L, Krauthgarner R, Resnitzky P, Haran-Ghera N. Deletion of chromosome 2 is an early event in the development of radiation-induced myeloid leukemia in SJL/J mice. Leukemia 1988; 2: 545–550. 10 Resnitzky P, Esterov Z, Haran-Ghera N. High incidence of AML in SJL/J mice after X-irradiation and corticosteroids. Leukemia Res 1985; 9: 1519–1528. 11 Alexander B, Rasko J, Morahan G, Cook WD. Gene deletion explains both in vivo and in vitro generated chromosome 2 aberrations associated with murine myeloid leukemia. Leukemia 1995; 9: 2009–2015. 12 Fearon ER. Human cancer syndromes: clues to the origin and nature of cancer. Science 1997; 278: 1043–1050. 13 Thibodeau SN, Bren G, Schaid D. Microsatellite instability in cancer of the proximal colon. Science 1993; 260: 816–819.

14 Cavenee WK, Dryja T, Phillips R, Benedict W, Godbout R, Gallie B, Murphree A, Strong L, White R. Expression of recessive alleles by chromosomal mechanisms in retinoblastoma. Nature 1983; 305: 779–784. 15 Chomczynski P. One hour downward alkaline capillary transfer for blotting of DNA and RNA. Anal Biochem 1992; 201: 134–139. 16 Obama H, Matsubara S, Guenet J, Muramatsu T. The midkine (MK) family of growth/differentiation factors: structure of an MKrelated sequence in a pseudogene and evolutionary relationships among members of the MK family. J Biochem (Tokyo) 1994; 115: 516–522. 17 Lee JD, Kato K, Tobias P, Kirkland T, Ulevitch R. Transfection of CD14 into 70Z/3 cells dramatically enhances the sensitivity to complexes of LPS and LPS binding protein. J Exp Med 1992; 175: 1697–1705. 18 Maruta H, Bartlett P, Nurcombe V, Nur-E-Kamal M, Chomienne C, Muramatsu T, Muramatsu H, Fabri L, Nice E, Burgess A. Midkine (MK), a retinoic acid (RA)-inducible gene product, produced in E. coli acts on neuronal and HL60 leukemia cells. Growth Factors 1993; 8: 119–134. 19 Brunson KW, Nicolson G. LPS effects on sensitive and resistant variant Chinese hamster ovary cell lines. J Supramol Struct 1978; 9: 231–242. 20 Bartold PM, Narayanan A, Page R. Platelet-derived growth factor reduces the inhibitory effects of LPS on gingival fibroblast proliferation. J Periodontal Res 1992; 27: 499–505. 21 Wille JJ, Park J, Elgavish A. Effects of growth factors, hormones, bacterial LPS, and lipotechoic acids on the clonal growth of normal ureteral epithelial cells in serum-free culture. J Cell Physiol 1992; 150: 52–58. 22 Goureau O, Lepoivre M, Becquet F, Courtois Y. Differential regulation of inducible nitric oxide synthase by fibroblast growth factors and transforming growth factor in bovine retinal pigmented epithelial cells: inverse correlation with cellular proliferation. Proc Natl Acad Sci USA 1993; 90: 4276–4280. 23 Feder LS, Laskin DL. Regulation of hepatic endothelial cell and macrophage proliferation and nitric oxide production by GM-CSF, M-CSF and IL-1 following acute endotoxemia. J Leukoc Biol 1994; 55: 507–513. 24 Vairo G, Royston A, Hamilton J. Biochemical events accompanying macrophage activation and the inhibition of CSF1-induced macrophage proliferation by TNF, IFN and LPS. J Cell Physiol 1992; 151: 630–641. 25 Kreisberg R, Detrick M, Moore R. Opposing effects of tumor necrosis factor and leukemia inhibitory factor in LPS-stimulated myelopoiesis. Infect Immun 1993; 61: 418–422. 26 Pelner L, Fowler G, Nauts H. Effects of concurrent infections and their toxins on the course of leukemia. Acta Med Scand 1958; 338 (Suppl): 1–47. 27 Fujii Y, Yeki N, Takeichi N, Kobayashi H, Miyazaki T. Differentiation therapy of a myelomonocytic leukemia (c-WRT-7) in rats by injection of LPS and daunomycin. Cancer Res 1987; 47: 1668–1673. 28 Treon SP, Anand B, Chou I, Broitman S. Growth restraint and differentiation by LPS/TNF/IFN reorganization of the microtubule network in human leukemia cell lines. Leukemia 1992; 6 (Suppl 3): 141S–145S. 29 Ohmori Y, Hamilton T. Cooperative interaction between IFN stimulus response element and B sequence motifs controls IFN and LPS-stimulated transcription from the murine IP-10 promoter. J Biol Chem 1993; 268: 6677–6688. 30 Rasko JE, Metcalf D, Alexander B, Strasser A, Begley CG. Establishment of multipotential and antigen presenting cell lines derived from myeloid leukemias in GM-CSF transgenic mice. Leukemia 1997; 11: 732–742. 31 Vogel SN, Manthey C, Perera P, Li Z, Henricson B. Dissection of LPS-induced signaling pathways in murine macrophages using LPS analogs, LPS mimetics, and agents unrelated to LPS. Prog Clin Biol Res 1995; 392: 421–431. 32 Barber SA, Perera P, McNally R, Vogel SN. The serine/threonine phosphatase inhibitor, calyculin A, inhibits and dissociates macrophage responses to LPS. J Immunol 1995; 155: 1404–1410. 33 Ferrero E, Goyert SM. Nucleotide sequence of the gene encoding the monocyte differentiation antigen, CD14. Nucleic Acids Res 1988; 16: 4173.

Effect of chromosomal deletion WD Cook and BJ Alexander

34 Setoguchi M, Nasu N, Yoshida S, Higuchi Y, Akizuki S, Yamamoto S. Mouse and human CD14 (myeloid cell-specific leucine-rich glycoprotein) primary structure deduced from cDNA clones. Biochim Biophys Acta 1989; 1008: 213–222. 35 Dequiedt F, Kettmann R, Burny A, Willems L. Nucleotide sequence of the ovine P53 tumor suppressor cDNA and its genomic organization. DNA Seq 1995; 5: 255–259. 36 Bernards R, Schackleford G, Gerber M, Horowitz J, Friend S, Schartl M, Bogenmann E, Rapaport J, McGee T, Dryja T, Weinberg R. Structure and expression of the murine retinoblastoma gene and

characterization of its encoded protein. Proc Natl Acad Sci USA 1989; 86: 6474–6478. 37 Carver EA, Stubbs L. Zooming in on the human–mouse comparative map: genome conservation re-examined on a high resolution scale. Genome Res 1997; 7: 1123–1137. 38 Resnitzky P, Goren T, Shaft D, Trakhtenbrot L, Peled A, Resnitzky D, Zipori D, Haran-Ghera N. Absence of negative growth regulation in three new murine radiation-induced myeloid leukemia cell lines with deletions of chromosome 2. Leukemia 1992; 6: 1288–1295.

1943