Numerous growth factors, cytokines, and chemokines are secreted by human CD34(+) cells, myeloblasts, erythroblasts, and megakaryoblasts and regulate normal hematopoiesis in an autocrine/paracrine manner.

The aim of this study was to explore further the hypothesis that early stages of normal human hematopoiesis might be coregulated by autocrine/paracrine regulatory loops and by cross-talk among early hematopoietic cells. Highly purified normal human CD34(+) cells and ex vivo expanded early colony-forming unit-granulocyte-macrophage (CFU-GM)-derived, burst forming unit-erythroid (BFU-E)-derived, and CFU-megakaryocyte (CFU-Meg)-derived cells were phenotyped for messenger RNA expression and protein secretion of various growth factors, cytokines, and chemokines to determine the biological significance of this secretion. Transcripts were found for numerous growth factors (kit ligand [KL], FLT3 ligand, fibroblast growth factor-2 [FGF-2], vascular endothelial growth factor [VEGF], hepatocyte growth factor [HGF], insulinlike growth factor-1 [IGF-1], and thrombopoietin [TPO]); cytokines (tumor necrosis factor-alpha, Fas ligand, interferon alpha, interleukin 1 [IL-1], and IL-16); and chemokines (macrophage inflammatory protein-1alpha [MIP-1alpha], MIP-1beta, regulated upon activation, normal T cell expressed and secreted [RANTES], monocyte chemotactic protein-3 [MCP-3], MCP-4, IL-8, interferon-inducible protein-10, macrophage-derived chemokine [MDC], and platelet factor-4 [PF-4]) to be expressed by CD34(+) cells. More importantly, the regulatory proteins VEGF, HGF, FGF-2, KL, FLT3 ligand, TPO, IL-16, IGF-1, transforming growth factor-beta1 (TGF-beta1), TGF-beta2, RANTES, MIP-1alpha, MIP-1beta, IL-8, and PF-4 were identified in media conditioned by these cells. Moreover, media conditioned by CD34(+) cells were found to inhibit apoptosis and slightly stimulate the proliferation of other freshly isolated CD34(+) cells; chemo-attract CFU-GM- and CFU-Meg-derived cells as well as other CD34(+) cells; and, finally, stimulate the proliferation of human endothelial cells. It was also demonstrated that these various hematopoietic growth factors, cytokines, and chemokines are expressed and secreted by CFU-GM-, CFU-Meg-, and BFU-E-derived cells. It is concluded that normal human CD34(+) cells and hematopoietic precursors secrete numerous regulatory molecules that form the basis of intercellular cross-talk networks and regulate in an autocrine and/or a paracrine manner the various stages of normal human hematopoiesis.

Hepatocyte growth factor is secreted by osteoblasts and cooperatively permits the survival of haematopoietic progenitors.

Human osteoblasts (HOBs) support the growth of human haematopoietic progenitor cells, and support the survival and limited expansion of long-term culture-initiating cells. Using human CD34+ cells and the murine myelomonocytic cell line NFS-60 as targets, we previously found that one component of HOB-derived haematopoietic activity is cell-associated granulocyte colony-stimulating factor (G-CSF). However, antibody failed to neutralize all the activity, suggesting that more than one factor supports haematopoietic cells. In the present investigations, we asked whether the HOB-derived, non-G-CSF secreted activity was as a result of other known growth factors. We found that, among the cytokines expressed by HOBs, only hepatocyte growth factor (HGF) and G-CSF stimulated NFS-60 cell proliferation. HOB cells and osteosarcoma cells secreted biologically active HGF, although the levels varied considerably. Moreover, addition of neutralizing HGF antibody to CD34+ cell/HOB co-cultures resulted in a significant reduction ( approximately 50%) in the ability of the HOBs to support haematopoietic progenitor cells. These results suggest that a major component of osteoblast-derived haematopoietic activity is HGF. Secretion of HGF, in concert with cell-associated cytokines such as G-CSF, may account for the stem cell-stimulating activities of osteogenic cells and, thereby, the unique stem cell-supporting role of the osteoblasts within the bone marrow microenvironment.

Novel human malignancy-associated gene (MAG) expressed in various tumors and in some tumor preexisting conditions.

We have identified a novel human malignancy-associated gene (MAG) expressed in various malignant tumors including glioblastomas and hepatocellular carcinomas (HCCs) and in tumor preexisting conditions such as hepatitis C virus- and hepatitis B virus-induced liver cirrhosis. The expression of MAG was characterized using reverse transcription-PCR (RT-PCR), rapid amplification of cDNA ends PCR, RNA dot blotting, RNase protection assay, and Northern blot analysis. Rapid amplification of cDNA ends PCR yielded a 536-bp MAG fragment in HCC, macroregenerative liver nodules with dysplasia, and liver cirrhosis but not in normal liver or placenta. By RT-PCR, MAG expression was not found in 12 different normal tissues but found in 46 of 51 (90%) premalignant and malignant tissues of various sites. Embryonic liver and brain were positive for MAG expression together with tumors from the same organs, but the corresponding normal adult tissues were negative. By RNase protection assay, MAG mRNA was expressed in the HepG2 liver tumor cell line and in an ovarian carcinoma but not in normal liver. The estimated transcript size from Northern blot analysis was 8.8 kb. This novel gene may play a role in the progression of premalignant conditions and in the development of HCC and other cancers.

Evidence for complex nuclear inheritance in a pedigree with nonsyndromic deafness due to a homoplasmic mitochondrial mutation.

The relationship between mitochondrial genotype and clinical phenotype is complicated in most instances by the heteroplasmic nature of pathogenic mitochondrial mutations. We have previously shown that maternally inherited hearing loss in a large Arab-Israeli kindred is due to the homoplasmic A1555G mutation in the mitochondrial 12S ribosomal RNA gene [Prezant et al., 1993: Nat Genet 4:289-294]. Family members with this mutation have phenotypes ranging from profound hearing loss to completely normal hearing, and we have shown that there is genetic and biochemical evidence for nuclear gene involvement in this family [Bu et al., 1993: Genet Epidemiol 9:27-44; Guan et al., 1996: Hum Mol Genet 5:963-971]. To identify such a nuclear locus, two candidate genes were excluded through linkage analysis and sequencing, and a genome-wide linkage search in family members who all have the identical homoplasmic mitochondrial mutation, but differ in their hearing status, was performed. In two stages a total of 560 polymorphic genetic markers was genotyped, and the data were analyzed under model-dependent and model-free assumptions. No chromosomal region was identified as a major contributor to the phenotypic expression of the mitochondrial mutation. Thus, in this simplified paradigm of a homoplasmic mitochondrial mutation in a single kindred who all live in the similar environment of a small village, the penetrance of the mitochondrial mutation appears to depend on the interaction of multiple nuclear genes.

Temporal bone analysis of patients with presbycusis reveals high frequency of mitochondrial mutations.

Presbycusis is a histologically and genetically heterogenous group of disorders, which lead to progressive, primarily sensorineural hearing loss with aging. Acquired mitochondrial DNA defects have been proposed as important determinants of aging, particularly in neuro-muscular tissues. The spiral ganglion and membranous labyrinth from archival temporal bones of 5 patients with presbycusis were examined for mutations within the mitochondrially-encoded cytochrome oxidase II gene. When compared to controls, results indicate that mitochondrial mutations in the peripheral auditory system occur commonly with age-related hearing loss, that there is great individual variability in both quantity and location of mutation accumulation, and that at least a proportion of presbycusis patients have a highly significant load of mutations in auditory tissue. This work supports the hypothesis that acquired mitochondrial mutations are a determinant of hearing loss in a subgroup of presbycusis patients.

Characterization of cDNAs encoding the murine A+U-rich RNA-binding protein AUF1.

A+U-rich elements (ARE) serve to control the degradation of some proto-oncogene and lymphokine mRNAs. The protein, AUF1, which consists of two polypeptides of 37 and 40 kDa (p37 and p40, respectively) when purified from cytosol, has been implicated in ARE-directed mRNA turnover due to its binding to ARE. Molecular cloning of a cDNA (p37AUF1) corresponding to human p37 predicted a polypeptide containing two non-identical RNA recognition motifs (RRM) and a C-terminal Gln-rich domain [Zhang et al. Mol. Cell. Biol. 13 (1993) 7652-7665]. Two cDNAs, designated muAUF1-3 and muAUF1-7, were isolated from a murine fetal cDNA library, using as a probe, a fragment of the p37AUF1 cDNA encoding RRM1 and approximately half of RRM2. The muAUF1-3 open reading frame (ORF) was very homologous to human p37AUF1 with the greatest homology between the corresponding RRMs and the C-terminal Gln-rich motif. Clone muAUF1-7 was highly homologous to muAUF1-3, but was truncated within the region encoding the RNP-1 box in RRM2. Clone muAUF1-3 encoded 19 amino acids in RRM1 not encoded by either muAUF1-7 or human p37AUF1. Such alterations in sequence could modify the RNA-binding properties of these proteins and have concomitant effects on ARE-directed posttranscriptional processes.

Purification, characterization, and cDNA cloning of an AU-rich element RNA-binding protein, AUF1.

The degradation of some proto-oncogene and lymphokine mRNAs is controlled in part by an AU-rich element (ARE) in the 3' untranslated region. It was shown previously (G. Brewer, Mol. Cell. Biol. 11:2460-2466, 1991) that two polypeptides (37 and 40 kDa) copurified with fractions of a 130,000 x g postribosomal supernatant (S130) from K562 cells that selectively accelerated degradation of c-myc mRNA in a cell-free decay system. These polypeptides bound specifically to the c-myc and granulocyte-macrophage colony-stimulating factor 3' UTRs, suggesting they are in part responsible for selective mRNA degradation. In the present work, we have purified the RNA-binding component of this mRNA degradation activity, which we refer to as AUF1. Using antisera specific for these polypeptides, we demonstrate that the 37- and 40-kDa polypeptides are immunologically cross-reactive and that both polypeptides are phosphorylated and can be found in a complex(s) with other polypeptides. Immunologically related polypeptides are found in both the nucleus and the cytoplasm. The antibodies were also used to clone a cDNA for the 37-kDa polypeptide. This cDNA contains an open reading frame predicted to produce a protein with several features, including two RNA recognition motifs and domains that potentially mediate protein-protein interactions. These results provide further support for a role of this protein in mediating ARE-directed mRNA degradation.

The erythropoietin receptor gene: cloning and identification of multiple transcripts in an erythroid cell line OCIM1.

The hormone erythropoietin is necessary for the maintenance, generation, and maturation of erythroid cells. This kidney-derived glycoprotein binds to cell-surface receptors, mediating both the proliferation and differentiation of erythroid cells. We have characterized 16 human erythropoietin receptor cDNA clones from two cDNA libraries prepared from OCIM1 poly(A)-containing mRNA. Of these 16 isolates, many showed evidence of incorrect RNA splicing events; others contained small deletions yielding multiple transcripts that may play a role in their erythroleukemic state. The human erythropoietin receptor, as deduced from the sequence of these clones, encodes a 508-amino-acid molecule, including a 24-residue signal peptide, a 226-amino-acid external domain, a transmembrane spanning region of 22 amino acids, and a cytoplasmic domain of 236 amino acids. There is an 80% overall homology between the human and murine erythropoietin receptor sequences.

Sequence specificity of the P6 pairing for splicing of the group I td intron of phage T4.

Seventeen non-directed td- (thymidylate synthase-deficient) splicing-defective mutations isolated in phage T4 were localized within the catalytic core of the ribozyme. All of the mutations occur in conserved structural elements that form part of the td intron core secondary structure. Remarkably, seven of the seventeen independently isolated mutations clustered in the dinucleotide 5' element (P6[5']) of the putative two-base-pair P6 stem. An analysis of this region was undertaken by site-directed mutagenesis of the plasmid-borne td gene, leading to the following findings: First, the short P6 pairing in the td secondary structure model was verified with appropriate P6[5'] and P6[3'] compensatory mutations. Second, all P6[5'] and P6[3'] mutants are defective in the first step of splicing, guanosine-dependent 5' splice site cleavage, whereas their activity at the 3' splice site is variable. Third, residual in vitro splicing activity of the mutants altered on only one side of the P6 pairing is correlated with the ability to form an alternative two-base-pair P6 stem. Fourth, the degree to which the compensatory mutants have their splicing activity restored is highly condition-dependent. Restoration of phenotype of the compensatory P6[5']:[3'] constructs is weak under stringent in vitro conditions as well as in vivo. This sequence specificity is consistent with phylogenetic conservation of the P6 pairing elements in group I introns, and suggests either structural constraints on the P6 stem or a dual function of one or both pairing elements.

Genes within genes: independent expression of phage T4 intron open reading frames and the genes in which they reside.

The td, nrdB, and sunY introns of bacteriophage T4 each contain a long open reading frame (ORF). These ORFs are preceded by functional T4 late promoters and, in the case of the nrdB intron ORF, a functional middle promoter. Expression of phage-encoded intron ORF-lacZ fusions indicates that these T4 genes are highly regulated. The lack of translation of these ORFs from early pre-mRNAs can be accounted for by the presence of secondary structures that are absent from the late RNAs. Because translation of the intron ORFs could disrupt core structural elements required for pre-mRNA splicing, such regulation may be necessary to allow expression of the genes in which they reside.

Two domains for splicing in the intron of the phage T4 thymidylate synthase (td) gene established by nondirected mutagenesis.

Of 97 nondirected T4 thymidylate synthase-defective (td) mutations, 27 were mapped to the intron of the split td gene. Clustering of these intron mutations defined two domains that are functional in splicing, each within approximately 220 residues of the respective splice sites. Two selected mutations, tdN57 and tdN47, fell within phylogenetically conserved pairings, with tdN57 disrupting the exon I-internal guide pairing (P1) in the 5' domain and tdN47 destabilizing the P9 helix in the 3' domain. A splicing assay with synthetic oligonucleotides complementary to RNA junction sequences revealed processing defects for T4tdN57 and T4tdN47, both of which are impaired in cleavage at the 5' and 3' splice sites. Thus prokaryotic genetics facilitates association of specific residue changes with their consequences to splicing.

Processing of phage T4 td-encoded RNA is analogous to the eukaryotic group I splicing pathway.

Several features of the split td gene of phage T4 suggest an RNA processing mechanism analogous to that of the self-splicing rRNA of Tetrahymena and other group I eukaryotic introns. Previous work has revealed conserved sequence elements and the ability of td-encoded RNA to self-splice in vitro. We show here that a noncoded guanosine residue is covalently joined to the 5' end of the intron during processing. Further, we demonstrate the existence of linear and circular intron forms in RNA extracted from T4-infected cells and from uninfected Escherichia coli expressing the cloned td gene. Sequence analysis of the intron cyclization junction indicates that the noncoded guanosine and one additional nucleotide are lost from the 5' end of the intron upon cyclization. This analysis places a uridine residue upstream of the cyclization site, in analogy to three other group I cyclization junctions. These striking similarities to the splicing intermediates of eukaryotic group I introns point not only to an analogous processing pathway and conserved features of cyclization site recognition but also to a common ancestry between this prokaryotic intervening sequence and the group I eukaryotic introns.

RNA splicing and in vivo expression of the intron-containing td gene of bacteriophage T4.

The splice junction sequence of td mRNA from T4-infected cells has been determined (5'....GGU-CUA....3') and shown to be identical to that of the RNA ligation product encoded by the cloned gene [Belfort et al. Cell 41 (1985) 375-382]. The RNA processing functions, T4 RNA ligase, T4 polynucleotide kinase, and the host prr gene product appear not to be essential for exon ligation; neither are the host endoribonucleases RNase III, RNase P and RNase E required for intron excision. While these results are consistent with the autocatalytic splicing mechanism demonstrated in vitro [Chu et al. J. Biol. Chem. 260 (1985) 10680-10688], they leave unanswered the question of which protein(s), if any, might stimulate the in vivo reaction. Analysis of the products of the cloned td gene has led to identification of two td-encoded polypeptides, namely a polypeptide corresponding to the exon-I-coding sequence (NH2-TS), and the catalytically active thymidylate synthase (TS). Kinetic and nucleotide sequence data provide evidence that NH2-TS is the product of the primary transcript and that TS is encoded by spliced mRNA. These results suggest that splicing may provide a switch controlling the relative expression of NH2-TS and TS, two proteins with markedly different temporal appearances despite their identical transcriptional and translational start sites.

Processing of the intron-containing thymidylate synthase (td) gene of phage T4 is at the RNA level.

The interrupted T4 phage td gene, which encodes thymidylate synthase, is the first known example of an intron-containing prokaryotic structural gene. Analysis of td-encoded transcripts provides evidence in favor of maturation at the RNA level. Northern blotting with T4 RNA and with region-specific probes revealed three classes of RNA: diffuse premessage (ca. 2.5 kb), a low-abundance mature mRNA (ca. 1.3 kb), and an abundant free intron RNA (ca. 1.0 kb). The existence of covalently joined mature mRNA was suggested by hybridization and S1 protection experiments and was confirmed by primer extension analysis of the splice junction. In analogy to expression of interrupted eukaryotic genes, these results are consistent with an RNA processing model that would account for the direct gene transcript serving as precursor for both free intron RNA and a spliced mRNA that is colinear with the thymidylate synthase product.