Stress response genes: the genes that make cancer metastasize

Cancer is characterized by dysregulated growth control, overcoming of replicative senescence, and metastasis formation. The topology of cancer spread is mediated by a set of developmentally nonessential genes which are physiologically involved in stress responses, inflammation, wound healing, and neovascularization. The function of these gene products is extensively modified posttranscriptionally. In cancer, metastasis genes are dysregulated at the levels of expression or splicing. These genes constitute a unique group of cancer-related biomolecules.

Costimulation by extracellular matrix proteins determines the response to TCR ligation

Although ligation of the T-cell antigen receptor (TCR) is central to the responsiveness and antigen specificity of T-cells, it is insufficient to elicit a response. To determine whether the need for costimulation reflects inadequate strength of signal transduction through the TCR or an absolute block of signaling in the absence of a coligand, we studied T-cell activation under serum-free conditions eliminating costimulation by various extracellular matrix proteins which otherwise have an omnipresent and frequently overlooked effect. Engagement of the TCR leads to induction of Fas, but not to measurable IL-2 secretion or apoptosis. Those activation parameters are induced by costimulation through integrin alphaVbeta3. Furthermore, T-cell survival or elimination is determined by the type of ligand binding to this coreceptor with vitronectin, fibronectin, and fibrinogen efficiently inducing apoptosis and IL-2 production while osteopontin and entactin mediate IL-2 secretion comparably without causing programmed cell death. Consistent with the cytokine properties of these ligands, differential costimulation depends on their presentation in soluble rather than immobilized form. The determination of elimination versus survival of activated T-cells by coligation of beta3-integrins may have bearing on the fundamental postthymic mechanisms that shape the T-cell repertoire.

Molecular mechanisms of tumor dissemination in primary and metastatic brain cancers

Cancer is characterized by dysregulated growth control, overcoming of replicative senescence, and metastasis formation. Tumor dissemination distinguishes malignant from benign neoplasms and is mediated by homing receptors, their ligands, and proteinases. The homing receptor CD44 is frequently expressed on primary brain tumors and brain metastases. Its engagement by osteopontin physiologically induces macrophage chemotaxis, a mechanism that may be utilized by metastatic brain tumors in the process of dissemination. In host defense, osteopontin and its receptors, CD44 and integrin alpha(V)beta(3), play key roles in mediating delayed type hypersensitivity responses by activating macrophages to induce Th1 cytokines while inhibiting Th2 cytokines. Other metastasis associated gene products similarly contribute to host defenses. Hence, cancer spread is regulated by a set of developmentally non-essential genes which physiologically mediate stress responses, inflammation, wound healing, and neovascularization. Function of the relevant gene products is extensively modified post-transcriptionally and their dysregulation in cancer occurs on the levels of expression and splicing. Consistent patterns of organ preference by malignancies of particular tissue origin suggest a necessary connection between loss of growth control and senescence genes and expression of genes mediating the dissemination of tumor cells.

Eta-1 (osteopontin): an early component of type-1 (cell-mediated) immunity

Cell-mediated (type-1) immunity is necessary for immune protection against most intracellular pathogens and, when excessive, can mediate organ-specific autoimmune destruction. Mice deficient in Eta-1 (also called osteopontin) gene expression have severely impaired type-1 immunity to viral infection [herpes simplex virus-type 1 (KOS strain)] and bacterial infection (Listeria monocytogenes) and do not develop sarcoid-type granulomas. Interleukin-12 (IL-12) and interferon-gamma production is diminished, and IL-10 production is increased. A phosphorylation-dependent interaction between the amino-terminal portion of Eta-1 and its integrin receptor stimulated IL-12 expression, whereas a phosphorylation-independent interaction with CD44 inhibited IL-10 expression. These findings identify Eta-1 as a key cytokine that sets the stage for efficient type-1 immune responses through differential regulation of macrophage IL-12 and IL-10 cytokine expression.

Signal transduction of mechanical stimuli is dependent on microfilament integrity: identification of osteopontin as a mechanically induced gene in osteoblasts

Mechanical perturbation has been shown to modulate a wide variety of changes in second message signals and patterns of gene expression in osteoblasts. Embryonic chick osteoblasts were subjected to a dynamic spatially uniform biaxial strain (1.3% applied strain) at 0.25 Hz for a single 2-h period, and osteopontin (OPN), an Arg-Gly-Asp (RGD)-containing protein, was shown to be a mechanoresponsive gene. Expression of opn mRNA reached a maximal 4-fold increase 9 h after the end of the mechanical perturbation that was not inhibited by cycloheximide, thus demonstrating that mechanoinduction of opn expression is a primary response through the activation of pre-existing transcriptional factors. The signal transduction pathways, which mediated the increased expression of opn in response to mechanical stimuli, were shown to be dependent on the activation of a tyrosine kinase(s) and protein kinase A (PKA) or a PKA-like kinase. Selective inhibition of protein kinase C (PKC) had no effect on the mechanoinduction of osteopontin even though opn has been demonstrated to be an early response gene to phorbol 12-myristate 13-acetate (PMA) stimulation. Mechanotransduction was dependent on microfilament integrity since cytochalasin-D blocked the up-regulation of the opn expression; however, microfilament disruption had no effect on the PMA induction of the gene. The microtubule component of the cytoskeleton was not related to the mechanism of signal transduction involved in controlling opn expression in response to mechanical stimulation since colchicine did not block opn expression. Mechanical stimulus was shown to activate focal adhesion kinase (FAK), which specifically became associated with the cytoskeleton after mechanical perturbation, and its association with the cytoskeleton was dependent on tyrosine kinase activity. In conclusion, these results demonstrate that the signal transduction pathway for mechanical activation of opn is uniquely dependent on the structural integrity of the microfilament component of the cytoskeleton. In contrast, the PKC pathway, which also activates this gene in osteoblasts, acts independently of the cytoskeleton in the transduction of its activity.

Identification of the phosphorylated sites of metabolically 32P-labeled osteopontin from cultured chicken osteoblasts

Osteopontin (OPN) is one of the major secretory phosphoproteins in both calcifying and non-calcifying tissues. Evidence has accumulated for the biological importance of the phosphoproteins and, in particular, the phosphate groups in bone formation, resorption, and calcification. The precise locations of the phosphate groups in the OPN molecule were determined by metabolically labeling OPN with 32P in cultured chicken osteoblasts, followed by purification to homogeneity. N-terminal sequencing showed a single sequence of WPVSKRQHAISA, consistent with that deduced from both cDNA, and previous amino acid sequencing of the protein isolated from chicken bone. Three 32P-labeled peptides were isolated by reverse-phase high performance liquid chromatography of thrombin-digested, 32P-labeled OPN. The N-terminal sequencing of each of these thrombin fragments gave single sequences as follows: WPVSKSRQHAIS, SHHTHRYHQDHVD, and ASKLRKAARKL, with approximate molecular masses of 5, 30, and 20 kDa. These data demonstrate that 32P was incorporated throughout the N- to C-terminal sequence of the protein. Thrombin specifically cleaved chicken OPN at two sites: between Arg-22 and Ser-23, which generated the 5-kDa N-terminal end fragment, and another between Lys-138 and Ala-139, which generated the 30- and 20-kDa fragments. To further define the exact locations of the phosphorylated amino acids and the surrounding amino acid sequences, OPN was digested with trypsin, which generated seven major 32P-labeled peptides whose amino acid sequences were determined. The phosphorylated peptide regions of osteopontin were identified as amino acids 8-18 (QHAIS*AS*S*EEK), 39-54 (LASQQTHYS*S*EENAD), 150-171 (LIEDDAT*AEVGDSQLAGLWLPK), 179-191 (ELAQHQSVENDSR), 194-205 (FDS*PEVGGDSK), 214-219 (ES*LASR), and 239-248 (HSIENNEVTR). The phosphorylated amino acid sites are followed by an asterisk (*). Of the seven identified phosphorylated peptide regions, three were localized on the N-terminal end of the osteopontin molecule (with five phosphorylated serines) and contained the sequence motifs that were phosphorylated by casein kinase II type(s), whereas the remaining four peptides are concentrated toward the C-terminal half of the molecule (with five phosphorylated residues) and contained recognition motifs for other kinases as well as casein kinase II.

Interaction between CD44 and osteopontin as a potential basis for metastasis formation

Malignant growth has been associated with oncogene activation, telomerase activity, and expression of CD44 splice variants on the cell surface. Though dysregulation of growth control due to expression of oncogene products is fairly well understood, the mechanism of CD44-mediated homing and colony formation in specific tissues has remained cryptic. We have identified the cytokine osteopontin as a ligand for CD44. Osteopontin binds to naturally expressed and stably transfected CD44 in a manner that is specific, dose-dependent, inhibitable by anti-CD44 antibodies, insensitive to competition by Gly-Arg-Gly-Asp-Ser, and sensitive to competition by hyaluronate. The receptor-ligand interaction mediates chemotaxis or attachment, depending on presentation of osteopontin in soluble or immobilized form. In contrast, binding of CD44 to hyaluronate mediates aggregation or attachment but not chemotaxis. We found that two events occurring in malignancy-secretion of osteopontin and expression of CD44v-are linked in such a way that they may cause migration of tumor cells to specific sites of metastasis formation.

Protein kinases of cultured osteoblasts: selectivity for the extracellular matrix proteins of bone and their catalytic competence for osteopontin

The enzyme activities of the major kinases found within the cytosolic and microsomal fractions of embryonic avian calvaria osteoblasts were assayed for their specificity for various noncollagenous extracellular matrix (ECM) proteins of bone. At least 6 proteins with M(r)'s of 66, 58, 50, 36, 30, and 22 kD out of more than 30 of the noncollagenous proteins of the bone ECM were phosphorylated by the kinase(s) found in both osteoblast cellular fractions. The purification and N-terminal sequence analysis of three of the above proteins, M(r)'s 66 and 58 kD (+50 kD), identified them as chicken bone sialoprotein (BSP) and osteopontin (OPN), respectively. Heparin, a specific inhibitor of factor-independent protein kinase (FIPK) activity, blocked the phosphorylation of all six ECM proteins by the microsomal kinase(s) but only inhibited the phosphorylation of the 66, 50, and 36 kD by the cytosolic enzyme(s). Casein kinase II (a known FIPK) showed a similar phosphorylation pattern of the same bone ECM proteins as the FIPK(s) found in osteoblast cell extracts, while purified cyclic adenosine monophosphate (cAMP)-dependent protein kinase did not phosphorylate any of the ECM proteins. Use of dephosphorylated casein showed that in comparison with casein kinase II, casein was a poor substrate for the FIPK found in the osteoblast cellular extracts. Further studies, using FIPK(s) of osteoblasts and purified chicken OPN or bacterially produced recombinant murine OPN as a substrate, showed that both species of OPN were excellent substrates for the FIPK(s) found in osteoblasts. The phosphorylation of the purified chicken and recombinant mouse OPNs were evaluated by quantitative analysis using commercially available protein kinases. cAMP-dependent kinase showed no phosphorylation of either protein, and cyclic guanodine monophosphate (cGMP)-dependent kinase and protein kinase C incorporated 1.2 and 0.5 mol phosphate/mol OPN, respectively. However, both chicken and mouse OPNs were significantly phosphorylated by casein kinase II (9.3 and 9.0 mol of phosphate/mol of OPN, respectively). These results demonstrate that the noncollagenous proteins of the bone ECM, and in particular OPN, are predominantly phosphorylated by FIPK(s), and this class of kinase is the major enzyme found within the microsomal fraction of osteoblasts.

Receptor-ligand interaction between CD44 and osteopontin (Eta-1)

The CD44 family of surface receptors regulates adhesion, movement, and activation of normal and neoplastic cells. The cytokine osteopontin (Eta-1), which regulates similar cellular functions, was found to be a protein ligand of CD44. Osteopontin induces cellular chemotaxis but not homotypic aggregation, whereas the inverse is true for the interaction between CD44 and a carbohydrate ligand, hyaluronate. The different responses of cells after CD44 ligation by either osteopontin or hyaluronate may account for the independent effects of CD44 on cell migration and growth. This mechanism may also be exploited by tumor cells to promote metastasis formation.

Protein kinases of cultured chicken osteoblasts that phosphorylate extracellular bone proteins

Cytosolic and microsomal protein kinase preparations from cultured chicken osteoblasts were found to phosphorylate up to six major proteins with Mrs 66, 58, 50, 36, 32, and 22 kDa in chicken bone extract. Use of heparin led to the conclusion that these proteins were predominantly phosphorylated by factor-independent protein kinase (FIPK) present both in microsomal and cytosolic preparations. It was confirmed that microsomal preparation contained predominantly FIPK, whereas cytosolic preparation contained additional kinases, that can phosphorylate the bone proteins. Use of purified chicken bone osteopontin (OPN) (58 kDa) and recombinant OPN led to the same conclusions. The identify of the protein kinases was clearly established by using a series of synthetic peptide substrates. Quantitative analysis utilizing pure protein kinases and purified chicken bone OPN, recombinant mouse OPN, and bovine bone OPN and BSP led to introduction of approximately 9 moles of phosphate/mole of OPN and 6.6 moles phosphate/mole bovine bone sialoprotein (BSP) by casein kinase II. cGMP-dependent protein kinase and protein kinase C both introduced 0.5-1.2 moles phosphate/mole of OPN and BSP, whereas cAMP-dependent protein kinase led to no significant phosphorylation of OPN or BSP. Consistent with the above results, sites of phosphorylation identified for OPN (metabolically labeled) and BSP (labeled by casein kinase II) revealed that predominant phosphorylated sites have recognition sequences for FIPK.

Gene expression and phosphorylation of mouse osteopontin

Osteopontin is expressed in many different cell types and has been proposed to play several functions. Distinct forms of the protein have been detected. Various tissues and cell lines from mouse, however, exhibit two classes of transcripts with different 5'-untranslated ends but with an identical coding region (exons II through VII). These transcripts do not arise from the alternative splicing of coding exons. These results suggest that posttranslational modifications of osteopontin, such as phosphorylation, are a major mechanism to generate different forms of the protein. Mouse osteopontin was expressed in E. coli and used as a model to study its phosphorylation.

Mouse osteopontin expressed in E. coli exhibits autophosphorylating activity of tyrosine residues

Osteopontin is a secreted glycosylated phosphoprotein found in various normal and transformed tissues. Mouse osteopontin expressed in bacteria has been found to autophosphorylate in vitro using ATP or GTP as phosphoryl donors. The reaction does not occur using inorganic orthophosphate as the donor. Only tyrosine residues are phosphorylated. Neither serine nor threonine residues, both of which are found phosphorylated in osteopontin extracted from bone, is autophosphorylated in vitro. The autophosphorylation of tyrosine residues by a secreted protein such as osteopontin may provide additional insight into its biological functions.

In vitro phosphorylation of mouse osteopontin expressed in E. coli

To understand the role of post-translational modifications on the structure and function of osteopontin, a secreted glycosylated phosphoprotein, we expressed mouse osteopontin in E. coli as a fusion protein with glutathione-S-transferase (GST). The purified fusion protein was cleaved by factor Xa generating GST (26 kDa) and recombinant osteopontin (60 kDa). The fusion protein was phosphorylated in vitro by cytosolic, microsomal, and casein kinase II fractions from mouse kidney homogenates. The fusion protein and recombinant osteopontin were also phosphorylated by the catalytic subunit of cAMP-dependent protein kinase. The suitability of the fusion and recombinant proteins as model substrates for the study of the function(s) and post-translational modifications of osteopontin is discussed.

Regulation of gluconeogenesis in swine kidney proximal tubule cells

Proximal tubule cells were isolated from swine kidney and cultured for periods of more than 30 days. The cells formed confluent monolayers after plating on a collagen surface and they were passaged more than 5 times on this matrix. The cells maintain several metabolic functions of proximal tubule cells, including gluconeogenesis and the ability to respond to epinephrine and parathyroid hormone. Gluconeogenesis, a principal metabolic pathway in proximal tubule cells, was examined as a function of days in culture. The isolated cells showed a nearly constant rate of gluconeogenesis from 14C-lactate, 14C-alanine and 14C-glycerol with no significant loss of activity for at least 30 days in culture. Likewise, the activities of several cytosolic and membrane associated enzymes including, alkaline phosphatase, delta-glutamyltransferase, fructose-1,6-bisphosphatase and phosphofructokinase were nearly constant over the same time period. The cells responded to treatment with epinephrine and parathyroid hormone, and the rate of gluconeogenesis from 14C-lactate doubled in the presence of these hormones. The morphological and biochemical evidence obtained in these studies show that the proximal tubule cells isolated from swine kidney provide an excellent well defined system for studying the hormonal regulation of carbohydrate metabolism in this tissue.

Interaction of the P1c1 repressor with P1 DNA: localization of repressor binding sites near the c1 gene

The c1 repressor of phage P1 was previously shown (B.R. Baumstark and J.R. Scott, 1980, J. Mol. Biol. 140, 471-480) to bind specifically to P1BamHI-9, a 1.4-kb fragment that is closely linked to the c1 structural gene and spans the ends of the P1 genetic map. The position of the repressor binding site(s) relative to the ends of the genetic map and the c1 gene was investigated by testing cloned fragments of EcoRI-7 and BamHI-9 for c1 expression and repressor binding. Although sequences in both BamHI-9 and the adjacent 2.7-kb EcoRI/BamHI fragment were found to be required for the production of the c1 protein, c1 expression could be restored to the 2.7-kb fragment by the addition of a heterologous promoter (ptac). These observations are consistent with the localization of the c1 reading frame to the 2.7-kb fragment and at least part of the c1 promoter region to BamHI-9. The c1 repressor was shown to bind in vitro to two distinct cloned fragments of BamHI-9 derived from the far right side of the P1 map, indicating the presence of at least two recognition sites in this region. DNA sequence analysis revealed that these two fragments share a 23-bp region of homology. A synthetic DNA containing an 11-bp sequence from this region acts as an effective competitor for repressor binding in vitro, suggesting that at least part of the sequence shared by the fragments is involved in repressor-DNA recognition.

Evidence for a specific phosphoryl binding site in swine kidney phosphofructokinase

Phosphofructokinase (PFK) from swine kidney was purified by a procedure which included affinity chromatography on Cibacron blue F3GA-Sepharose 4B and ATP-Sepharose 4B columns in order to examine its binding properties. The homogeneous enzyme was purified more than 3000-fold with a yield of 30% and it had a specific activity of 39.8 mumol/min/mg of protein at 25 degrees C. The molecular weight of the native enzyme was 360 000 and it contained 4 identical subunits of molecular weight 88 000. The principal catalytically reacting form of the enzyme had a S20,w of 13.7 S which corresponds to a molecular weight of 360 000 +/- 6 000. The initial velocity patterns in the forward and reverse directions suggested a sequential mechanism for the reaction. The Km values for fructose 6-phosphate, ATP, fructose, 1,6-bisP and ADP were 33 microM, 8.3 microM, 460 microM and 110 microM, respectively. The homogeneous native enzyme binds specifically to phosphoryl groups immobilized in cellulose phosphate columns. ATP and fructose 6-phosphate interacted with the enzyme and decreased its affinity for phosphoryl binding sites. Other metabolites including fructose 1,6-bisP, glucose 6-phosphate and various nucleotides, alone or in various combinations, were ineffective in promoting the dissociation of the enzyme. Allosteric effectors of the enzyme, such as citrate and AMP were also inactive. However, the cooperatively altered the concentration of ATP required to dissociate the enzyme from phosphoryl groups. The bound enzyme was enzymatically inactive. The enzyme was also inactivated when it was treated with pyridoxal 5'-phosphate and reduced with sodium borohydride and the inactive enzyme no longer bound to cellulose phosphate. These effects were not observed when treatment with pyridoxal 5'-phosphate was carried out in the presence of fructose 6-phosphate. These observations and the results of similar studies with swine kidney fructose 1,6-bisphosphatase (FBPase) show that both enzymes share the unique property of binding specifically to phosphoryl groups. FBPase interacts through its allosteric AMP binding site and PFK binds through its fructose 6-P binding site. This specific binding of both enzymes through these sites result in the inactivation of PFK and the desensitization of FBPase to allosteric inhibition by AMP. In the unbound state PFK may be active and FBPase can be inhibited by AMP. Taken collectively, these binding effects could play a role in the reciprocal regulation of these enzymes during gluconeogenesis in kidney.

Localized elaboration of IgA in normal and neoplastic murine mammary cell cultures: relationship to fibronectin matrix

IgA secretion by the lactating mammary gland culminates a complex sequence of biologic events both within the gland and at distant sites. Because of the many extraglandular influences, it is difficult to investigate IgA secretion at the tissue and cellular levels in the intact animal. In this study, with the use of immunohistofluorescence, we have observed elaboration of IgA by primary monolayer cultures of mammary cells from the glands of mid-pregnant mice and from mammary tumors. In cultures of normal cells, IgA appeared first in vesicular structures on the upper surfaces of the monolayers. Vesicular IgA was maximal at day 5 in culture, and at that time, was observed only over mounds (domelike structures) that were covered with fibrillar fibronectin (FN), and eventually developed a subpopulation of fusiform cells. It appears that the IgA observed was secreted in vitro, that normal mammary epithelial cells must form multicellular FN-bearing structures to secrete IgA in culture, and that by mid-pregnancy the murine mammary gland contains all of the lymphoid cells required for IgA secretion. In contrast, primary cultures of mammary tumors displayed minimal amounts of IgA and FN. The small amount of cell-associated IgA that was detected on tumor cultures was not localized to any multicellular structures nor was it associated with FN, but instead appeared as minute, punctate accumulations on individual cells scattered across the epithelioid areas. This finding is consistent with the loss of specialized functions and the increased autonomy typical of malignant cells. The study in cultured cells of a function as specialized as IgA secretion should permit greater understanding of both the process itself and the despecializations that accompany malignancies of secretory epithelia.

Localized elaboration of IgA in normal and neoplastic murine mammary cell cultures: relationship to fibronectin matrix

IgA secretion by the lactating mammary gland culminates a complex sequence of biologic events both within the gland and at distant sites. Because of the many extraglandular influences, it is difficult to investigate IgA secretion at the tissue and cellular levels in the intact animal. In this study, with the use of immunohistofluorescence, we have observed elaboration of IgA by primary monolayer cultures of mammary cells from the glands of mid-pregnant mice and from mammary tumors. In cultures of normal cells, IgA appeared first in vesicular structures on the upper surfaces of the monolayers. Vesicular IgA was maximal at day 5 in culture, and at that time, was observed only over mounds (domelike structures) that were covered with fibrillar fibronectin (FN), and eventually developed a subpopulation of fusiform cells. It appears that the IgA observed was secreted in vitro, that normal mammary epithelial cells must form multicellular FN-bearing structures to secrete IgA in culture, and that by mid-pregnancy the murine mammary gland contains all of the lymphoid cells required for IgA secretion. In contrast, primary cultures of mammary tumors displayed minimal amounts of IgA and FN. The small amount of cell-associated IgA that was detected on tumor cultures was not localized to any multicellular structures nor was it associated with FN, but instead appeared as minute, punctate accumulations on individual cells scattered across the epithelioid areas. This finding is consistent with the loss of specialized functions and the increased autonomy typical of malignant cells. The study in cultured cells of a function as specialized as IgA secretion should permit greater understanding of both the process itself and the despecializations that accompany malignancies of secretory epithelia.