NEDD9/Arf6-dependent endocytic trafficking of matrix metalloproteinase 14: a novel mechanism for blocking mesenchymal cell invasion and metastasis of breast cancer

NEDD9 is an established marker of invasive and metastatic cancers. NEDD9 downregulation has been shown to dramatically reduce cell invasion and metastasis in multiple tumors. The mechanisms by which NEDD9 regulates invasion are largely unknown. In the current study, we have found that NEDD9 is required for MMP14 enzymatic recovery/recycling through the late endosomes to enable disengagement of tissue inhibitor of matrix metalloproteinase 2 (TIMP2) and tumor invasion. Depletion of NEDD9 decreases targeting of the MMP14/TIMP2 complex to late endosomes and increases trafficking of MMP14 from early/sorting endosomes back to the surface in a small GTPase Arf6-dependent manner. NEDD9 directly binds to Arf6-GAP, ARAP3, and Arf6 effector GGA3 thereby facilitating the Arf6 inactivation required for MMP14/TIMP2 targeting to late endosomes. Re-expression of NEDD9 or a decrease in Arf6 activity is sufficient to restore MMP14 activity and the invasive properties of tumor cells. Importantly, NEDD9 inhibition by Vivo-Morpholinos, an antisense therapy, decreases primary tumor growth and metastasis in xenograft models of breast cancer. Collectively, our findings uncover a novel mechanism to control tumor cells dissemination through NEDD9/Arf6-dependent regulation of MMP14/TIMP2 trafficking, and validates NEDD9 as a clinically relevant therapeutic target to treat metastatic cancer.

Differential localization of myosin II isoforms in resting and activated osteoclasts

Osteoclasts resorb bone through a cyclical process of attachment to matrix, polarization, retraction, and migration. Although this process requires major alterations in the organization of actin structures, little is known about roles that myosins play in osteoclast cytoskeletal dynamics. We performed immunolocalization of myosin II using antibodies against heavy chain isoforms IIA and IIB and found that osteoclasts expressed the isoforms in distinct subcellular locations. Myosin IIA was enriched in dynamic cytoskeletal compartments, including the sealing zones of polarized and unpolarized osteoclasts. In contrast, myosin IIB was generally absent from these regions and maintained a comparatively static distribution during different phases of the osteoclast activation cycle. Inhibition of myosin II in osteoclasts by treatment with 2,3-butanedione monoxime caused detachment of unpolarized, but not polarized, cells from the bone matrix. These results suggest that myosin IIA is critical to development of an activated osteoclast phenotype.

Phosphorylation of myosin light chain kinase by p21-activated kinase PAK2

Phosphorylation of myosin II regulatory light chains (RLC) by Ca(2+)/calmodulin-dependent myosin light chain kinase (MLCK) is a critical step in the initiation of smooth muscle and non-muscle cell contraction. Post-translational modifications to MLCK down-regulate enzyme activity, suppressing RLC phosphorylation, myosin II activation, and tension development. Here we report that PAK2, a member of the Rho family of GTPase-dependent kinases, regulates isometric tension development and myosin II RLC phosphorylation in saponin permeabilized endothelial monolayers. PAK2 blunts tension development by 75% while inhibiting diphosphorylation of myosin II RLC. Cdc42-activated placenta and recombinant, constitutively active PAK2 phosphorylate MLCK in vitro with a stoichiometry of 1.71 +/- 0. 21 mol of PO(4)/mol of MLCK. This phosphorylation inhibits MLCK phosphorylation of myosin II RLC. PAK2 catalyzes MLCK phosphorylation on serine residues 439 and 991. Binding calmodulin to MLCK blocks phosphorylation of Ser-991 by PAK2. These results demonstrate that PAK2 can directly phosphorylate MLCK, inhibiting its activity and limiting the development of isometric tension.

Phosphorylation of non-muscle myosin II regulatory light chain by p21-activated kinase (gamma-PAK)

Myosin regulatory light chain (RLC) phosphorylation has been implicated in Rho-mediated stress fibre formation. The recent observation that Rho kinase phosphorylates RLC in vitro suggests that serine/threonine kinases other than those in the myosin light chain kinase (MLCK) family have the potential to activate myosin II. In this study we report that gamma-PAK, which is activated by the GTP-binding proteins Cdc42 and Rac, catalyses phosphorylation of intact non-muscle myosin II and isolated recombinant RLC. gamma-PAK phosphorylated endothelial cell myosin II to 0.85 +/- 0.02 mol PO4 per mol RLC. Phosphorylation is Ca2+/calmodulin-independent and the enzyme has a K(m) and Vmax for myosin II regulatory light chain of 12 microM and 180 nmol/min/mg respectively. No myosin II heavy chain phosphorylation was detected. Phosphopeptide maps and phosphoamino acid analysis revealed that gamma-PAK phosphorylates Ser-19 but does not phosphorylate Thr-18. A panel of recombinant RLC mutants was used to confirm that Ser-19 is the only phosphorylation site modified by gamma-PAK. On substitution of both Ser-19 and Thr-18 with Ala or Glu, no phosphorylation of other Ser/Thr residues in the RLC was detected. Similar to MLCK, Arg-16 is required for interaction of gamma-PAK with the substrate, since converting Arg-16 to Ala significantly reduced RLC phosphorylation. Endothelial cell monolayers permeabilized with saponin retract upon exposure to either Cdc42 or trypsin-activated gamma-PAK and ATP. Activation of gamma-PAK is required to initiate Ca2+/calmodulin-independent cell retraction and actin rearrangement. Taken together, these data suggest that myosin II activation by the p21-activated family of kinases may be physiologically important in regulating cytoskeletal organization.

Myosin phosphorylation by human cdc42-dependent S6/H4 kinase/gammaPAK from placenta and lymphoid cells

The p21-activated kinase (PAK) family includes protein phosphotransferases regulated by the GTPases rho, rac, and cdc42. Sequence homology, activation mechanism, and substrate specificity suggest that the well-characterized human placenta S6/H4 kinase is a member of this family. In these studies, S6/H4 kinase purified to homogeneity from human placenta was activated in vitro by cdc42-GTP, or protease incubation and MgATP-dependent autophosphorylation. The cdc42-activated enzyme demonstrated an Mr 60,000, and shares sequence homology with the gammaPAK family. Antipeptide antibodies against one of the autophosphorylation site sequences recognized a single p60 protein in the purified placenta preparation or Jurkat cell extracts. An autophosphorylated Mr 40,000 protein, previously identified as the catalytic domain of the enzyme, was also detected by the antibody after protease activation. Crude PAK60 obtained from Mono Q chromatography of Jurkat cell extracts and purified placenta enzyme catalyzed phosphorylation of histone H4 and myelin basic protein as well as a variety of synthetic peptides previously identified as S6/H4 kinase substrates. In addition, Jurkat myosin II and the regulatory myosin light chain were phosphorylated by the Jurkat and placenta gammaPAK. Synthetic peptides were used to demonstrate that the site of light chain phosphorylation occurs at the serine which results in ATPase activation. The data suggest that human gammaPAK may regulate cell motility by a GTP-dependent and calcium-independent mechanism.

Stimulated neutrophils induce myosin light chain phosphorylation and isometric tension in endothelial cells

The mechanism or mechanisms by which polymorphonuclear leukocytes (PMN) penetrate junctions between neighboring endothelial cells (EC) to traverse endothelial barriers remain unresolved. We report that chemoattractant-stimulated PMN induce a coordinate increase in both phosphorylation of serine 19 and threonine 18 of EC myosin regulatory light chains and isometric tension generation by EC monolayers. Unstimulated PMN had no effect on either parameter. These findings, coupled with our previous report (Huang et al., J. Cell Biol. 120: 1371-1380, 1993) that chemoattractant-stimulated PMN cause a rise in EC cytosolic free Ca2+, provide strong presumptive evidence that myosin light chain kinase is the EC enzyme responsible for initiating myosin light chain phosphorylation, EC contraction, and isometric tension generation in response to chemoattractant-stimulated PMN. We suggest that, by inducing phosphorylation of EC cytoskeletal proteins, chemoattractant-stimulated PMN induce EC to open their intercellular junctions, thereby facilitating transendothelial movement of these leukocytes.

Myosin light chain kinase-regulated endothelial cell contraction: the relationship between isometric tension, actin polymerization, and myosin phosphorylation

The phosphorylation of regulatory myosin light chains by the Ca2+/calmodulin-dependent enzyme myosin light chain kinase (MLCK) has been shown to be essential and sufficient for initiation of endothelial cell retraction in saponin permeabilized monolayers (Wysolmerski, R. B. and D. Lagunoff. 1990. Proc. Natl. Acad. Sci. USA. 87:16-20). We now report the effects of thrombin stimulation on human umbilical vein endothelial cell (HUVE) actin, myosin II and the functional correlate of the activated actomyosin based contractile system, isometric tension development. Using a newly designed isometric tension apparatus, we recorded quantitative changes in isometric tension from paired monolayers. Thrombin stimulation results in a rapid sustained isometric contraction that increases 2- to 2.5-fold within 5 min and remains elevated for at least 60 min. The phosphorylatable myosin light chains from HUVE were found to exist as two isoforms, differing in their molecular weights and isoelectric points. Resting isometric tension is associated with a basal phosphorylation of 0.54 mol PO4/mol myosin light chain. After thrombin treatment, phosphorylation rapidly increases to 1.61 mol PO4/mol myosin light chain within 60 s and remains elevated for the duration of the experiment. Myosin light chain phosphorylation precedes the development of isometric tension and maximal phosphorylation is maintained during the sustained phase of isometric contraction. Tryptic phosphopeptide maps from both control and thrombin-stimulated cultures resolve both monophosphorylated Ser-19 and diphosphorylated Ser-19/Thr-18 peptides indicative of MLCK activation. Changes in the polymerization of actin and association of myosin II correlate temporally with the phosphorylation of myosin II and development of isometric tension. Activation results in a 57% increase in F-actin content within 90 s and 90% of the soluble myosin II associates with the reorganizing F-actin. Furthermore, the disposition of actin and myosin II undergoes striking reorganization. F-actin initially forms a fine network of filaments that fills the cytoplasm and then reorganizes into prominent stress fibers. Myosin II rapidly forms discrete aggregates associated with the actin network and by 2.5 min assumes a distinct periodic distribution along the stress fibers.

Isometric contraction by fibroblasts and endothelial cells in tissue culture: a quantitative study

We have used an isometric force transducer to study contraction of two types of nonmuscle cells in tissue culture. This method permits the quantitative measurement of contractile force generated by cells of defined type under the influence of external agents while allowing detailed morphological observation. Chick embryo fibroblasts (CEF), which form a contractile network inside a collagen matrix, and human umbilical vein endothelial cells (HUVE), which are located in a monolayer on the surface of the collagen matrix, were studied. CEF and HUVE in 10% FCS produce a substantial tension of 4.5 +/- 0.2 x 10(4) dynes/cm2 and 6.1 x 10(4) dynes/cm2, respectively. Both cell types contract when stimulated with thrombin, generating a force per cell cross-sectional area of approximately 10(5) dynes/cm2, a value approximately an order of magnitude less than smooth muscle. The integrity of the actin cytoskeleton is essential for force generation, as disruption of actin microfilaments with cytochalasin D results in a rapid disappearance of force. Intact microtubules appear to reduce isometric force exerted by CEF, as microtubule-disrupting drugs result in increased tension. Contraction by HUVE precedes a dramatic rearrangement of actin microfilaments from a circumferential ring to stress fibers.

Regulation of permeabilized endothelial cell retraction by myosin phosphorylation

Permeabilized endothelial cell monolayers retracted on exposure to ATP and Ca2+. ADP, inosine triphosphate (ITP), GTP, adenosine 5'-(gamma-thio)triphosphate (ATP-gamma S), and 5'-adenylylimidodiphosphate failed to support retraction. However, ATP gamma S, a substrate for myosin light-chain kinase (MLCK) but not myosin adenosinetriphosphatase (ATPase), combined with ITP, a substrate for myosin ATPase but not MLCK, supported retraction. Two MLCK pseudosubstrate peptides, M5 and SM-1, inhibited endothelial cell retraction equally and more effectively than myosin kinase-inhibitory peptide with a sequence based on the phosphorylated site of myosin light chain. M5 was shown to inhibit thiophosphorylation of endothelial cell myosin light chains. Endothelial cells incubated with exogenous unregulated kinase in the presence of ethylene glycol-bis(beta-aminoethyl ether)-N,N,N',N'-tetra-acetic acid retracted on addition of ATP. This retraction was accompanied by thiophosphorylation of the 19 kDa myosin light chains in the presence of ATP gamma 35S. The N-ethylmaleimide-modified subfragment 1 of myosin heads, a specific inhibitor of actin-myosin interaction, prevented retraction. These data add support to the proposal of a central role for MLCK activation of myosin in endothelial retraction.

Involvement of myosin light-chain kinase in endothelial cell retraction

Permeabilized bovine pulmonary artery endothelial cell monolayers were used to investigate the mechanism of endothelial cell retraction. Postconfluent endothelial cells permeabilized with saponin retracted upon exposure to ATP and Ca2+. Retraction was accompanied by thiophosphorylation of 19,000-Da myosin light chains when adenosine 5'-[gamma-[35S]thio]triphosphate was included in the medium. Both retraction and thiophosphorylation of myosin light chains exhibited a graded quantitative dependence on Ca2+. When permeabilized monolayers were extracted in buffer D containing 100 mM KCl and 30 mM MgCl2 for 30 min, the cells failed to retract upon exposure to ATP and Ca2+, and no thiophosphorylation of myosin light chains occurred. The ability both to retract and to thiophosphorylate myosin light chains was restored by the addition to the permeabilized, extracted cells of myosin light-chain kinase and calmodulin together but not by either alone. These studies indicate that endothelial cell retraction, as does smooth muscle contraction, depends on myosin light-chain kinase phosphorylation of myosin light chains.

Inhibition of endothelial cell retraction by ATP depletion

To determine the relationship among endothelial cell (EC) retraction, cell adenosine triphosphate (ATP), and the status of cellular actin, ATP levels, F-actin content, and cytochemical redistribution in bovine pulmonary artery endothelial cells were assessed. EC monolayers 7 days after confluence were exposed to ethchlorvynol (ECV), histamine, or cytochalasin B (cyto B) for time intervals from 5-90 minutes. All 3 agents resulted in endothelial cell retraction without significant effect on cellular ATP content. Sixty-minute incubation of monolayers in glucose-free media containing antimycin A and 2-deoxyglucose depleted cellular ATP to less than 10% of control levels. ATP depleted monolayers failed to retract when incubated with ECV, histamine, or cyto B. ATP depletion resulted in loss of the prominent EC margins but only a rare gap between adjacent cells. When ATP levels were allowed to recover, the ability of EC monolayers to retract was restored. Actin filaments in control monolayers localized to a dense peripheral band of actin, a paranuclear complex, and bundles of microfilaments orientated parallel to the long axis of the cell. ECV induced complete loss of the dense peripheral band and other changes in the actin disposition. Monolayers exposed to histamine showed a retraction of the dense peripheral band of actin to a subcortical position. Cyto B caused loss of the dense peripheral band and the longitudinal microfilament bundles. Monolayers depleted of ATP lost their dense peripheral band and exhibited a disorganized, tangled web of microfilaments. Neither histamine nor ECV modified the actin distribution in ATP-depleted monolayers, whereas exposure to cyto B resulted in substantial change in actin with formation of a rim inside the cell membrane and considerable loss of actin filaments. ECV or histamine induced a small reduction in F-actin content while cyto B resulted in a 50% decline in 15 minutes. ATP depletion resulted in a 19% decrease in F-actin, with no further reduction on subsequent exposure to histamine or ECV. Cyto B treatment of ATP-depleted monolayers caused a drop in F-actin content equivalent to that observed in cells with normal ATP levels. These studies indicate that ATP is essential for changes in actin filament distribution and endothelial cell retraction produced by ECV, histamine or cyto B, and make it unlikely that any of these agents acts simply by depolymerization of actin filaments or modification of the dense peripheral band, although disruption of the dense peripheral band may facilitate retraction in the presence of adequate levels of cell ATP.

Ca2+-dependent and Ca2+-independent pathways for release of arachidonic acid from phosphatidylinositol in endothelial cells

The pathways for degradation of phosphatidylinositol (PI) were investigated in sonicated suspensions prepared from confluent cultures of bovine pulmonary artery endothelial cells. The time courses of formation of 3H-labeled and 14C-labeled metabolites of phosphatidyl-[3H]inositol ([3H]Ins-PI) and 1-stearoyl-2-[14C] arachidonoyl-PI were determined at 37 degrees C and pH 7.5 in the presence of 2 mM EDTA with or without a 2 mM excess of Ca2+. The rates of formation of lysophosphatidyl-[3H]inositol ([3H]Ins-lyso-PI) and 1-lyso-2-[14C] arachidonoyl-PI were similar in the presence and absence of Ca2+, and the absolute amounts of the two radiolabeled lyso-PI products formed were nearly identical. This indicated that lyso-PI was formed by phospholipase A1, and phospholipase A2 was not measurable. In the presence of EDTA, [14C]arachidonic acid release from 1-stearoyl-2-[14C]arachidonoyl-PI paralleled release of glycerophospho-[3H]inositol ([3H]GPI) from [3H]Ins-PI. Formation of [3H]GPI was inhibited by treatment with the specific sulfhydryl reagent, 2,2'-dithiodipyridine, and this was accompanied by an increase in [3H]Ins-lyso-PI. In the presence of Ca2+, [14C] arachidonic acid release from 1-stearoyl-2-[14C]arachidonoyl-PI was increased 2-fold and was associated with Ca2+-dependent phospholipase C activity. Under these conditions, [3H]inositol monophosphate production exceeded formation of [14C]arachidonic acid-labeled phospholipase C products, diacylglycerol plus monoacylglycerol, by an amount that was equal to the amount of [14C]arachidonic acid formed in excess of [3H]GPI. Low concentrations of phenylmethanesulfonyl fluoride (15-125 microM) inhibited Ca2+-dependent [14C]arachidonic acid release, and the decrease in [14C] arachidonic acid formed was matched by an equivalent increase in 14C label in diacylglycerol plus monoacyclglycerol. These data supported the existence of two pathways for arachidonic acid release from PI in endothelial cells; a phospholipase A1-lysophospholipase pathway that was Ca2+-independent and a phospholipase C-diacylglycerol lipase pathway that was Ca2+-dependent. The mean percentage of arachidonic acid released from PI via the phospholipase C-diacylglycerol lipase pathway in the presence of Ca2+ was 65 +/- 8%. The mean percentage of nonpolar phospholipase C products of PI metabolized via the diacylglycerol lipase pathway to free arachidonic acid was 28 +/- 3%.

Inhibitor of the factor VIIa-tissue factor complex is reduced in patients with disseminated intravascular coagulation but not in patients with severe hepatocellular disease

Inhibition of Factor VIIa-tissue factor activity by a plasma component(s) that requires factor Xa has been described recently. In this communication, we have developed a specific radiometric assay (which utilizes 3H-Factor IX and is sensitive to less than 1% of plasma level) for this inhibitor and have measured its activity in various disease states. Strikingly, the levels of this inhibitor were found to be normal in patients with advanced chronic hepatocellular disease but low in patients with disseminated intravascular coagulation (DIC). When endotoxin was used to induce DIC in rabbits, the levels of this inhibitor fell by 25-90%. Human umbilical vein endothelial cells (HUVE), bovine pulmonary artery endothelial cells, and a human hepatoma cell line (HepG2) all synthesized and secreted this inhibitor, whereas a promyelocytic cell line (HL-60) did not and a monocytic cell line (U937) appears to synthesize only small amounts. When ammonium sulfate-fractionated human plasma and serum-free conditioned media from both HUVE and HepG2 cells were electrophoresed on sodium dodecyl sulfate acrylamide gels, two activity peaks corresponding to Mr approximately 45,000 and Mr approximately 33,000 were eluted in each case. These observations suggest that (a) the inhibitor is consumed in DIC and that (b) endothelial cells (or other cells) synthesize sufficient amounts of this inhibitor in vivo to compensate for any decreased production by liver cells.

Phosphatidylcholine metabolism in endothelial cells: evidence for phospholipase A and a novel Ca2+-independent phospholipase C

The metabolism of phosphatidylcholine (PC) was investigated in sonicated suspensions of bovine pulmonary artery endothelial cells and in subcellular fractions using two PC substrates: 1-oleoyl-2-[3H]oleoyl-sn-glycero-3-phosphocholine and 1,2-dipalmitoyl-sn-glycero-3-phospho[14C]choline. When these substrates were incubated with the whole cell sonicate at pH 7.5, all of the metabolized 3H label was recovered in [3H]oleic acid (95%) and [3H]diacylglycerol (5%). All of the 14C label was identified in [14C]lysoPC (92%) and [14C]phosphocholine (8%). These data indicated that PC was metabolized via phospholipase(s) A and phospholipase C. Substantial diacylglycerol lipase activity was identified in the cell sonicate. Production of similar proportions of diacylglycerol and phosphocholine and the low relative activity of phospholipase C compared to phospholipase A indicated that the phospholipase C-diacylglycerol lipase pathway contributed little to fatty acid release from the sn-2 position of PC. Neither phospholipase A nor phospholipase C required Ca2+. The pH profiles and subcellular fractionation experiments indicated the presence of multiple forms of phospholipase A, but phospholipase C activity displayed a single pH optimum at 7.5 and was located exclusively in the particulate fraction. The two enzyme activities demonstrated differential sensitivities to inhibition by p-bromophenacylbromide, phenylmethanesulfonyl fluoride and quinacrine. Each of these agents inhibited phospholipase A, whereas phospholipase C was inhibited only by p-bromophenacylbromide. The unique characteristics observed for phospholipase C activity towards PC indicated the existence of a novel enzyme that may play an important role in lipid metabolism in endothelial cells.

Enhancement of human polymorphonuclear leukocyte adherence to plastic and endothelium by phorbol myristate acetate. Comparison with human C5a

The adherence of circulating neutrophils to vascular endothelium represents a necessary step in the chemotactic emigration of neutrophils to extravascular inflammatory sites. Studies of neutrophil adherence induced by phorbol myristate acetate (PMA) were undertaken to determine the ability of a nonchemotactic neutrophil stimulus to provoke increased adherence. The authors found that the adherence of human neutrophils to plastic surfaces or confluent monolayers of endothelial cells is enhanced in a concentration-dependent fashion by exposure of neutrophils to PMA. The effect of PMA concentration (0.1-5.0 ng/ml) on increased neutrophil adherence parallels that observed for superoxide anion generation and release of lysosomal enzymes from specific granules. Whereas complement C5a-treated neutrophils exhibited a fourfold to fivefold increase in adherence to endothelial cells, PMA-treated neutrophils showed a 10-fold to 20-fold increase. The ability of PMA to cause increased neutrophil adherence to endothelium appeared to be directed primarily at the neutrophil. Pretreatment of neutrophils with PMA was as effective as coincubation in causing increased adherence to plastic surfaces or confluent cultured endothelial cells, but pretreatment of endothelial cells with PMA failed to promote neutrophil adherence. Alteration of neutrophil cytoskeletal structures by cytochalasin B treatment did not prevent subsequent PMA-stimulated neutrophil adherence. These results demonstrate that increased neutrophil adherence to surfaces can be induced by a nonchemotactic stimulus and that neutrophil adherence is independent of organized microfilaments.