Localization of endotoxin in the rat intestinal epithelium.

High levels of circulating lipopolysaccharide (LPS) cause intestinal inflammation and increased permeability to bacteria and toxins. To better understand the effects of LPS on the gut, confocal microscopy and immunofluorescence staining were used to examine the distribution of LPS in the rat intestine after intravenous or enteral administration. LPS was localized in macrophages in the lamina propria from 1 h to >28 days after intravenous injection. LPS was also detected in the epithelial cells from 8 h to 7 days after injection. In contrast, LPS administered enterally was found in the gut lumen in close proximity to the mucosa but was not detected in enterocytes at any time. The concentration of LPS in enterocytes near the villus tip provides a mechanism for the clearance of endotoxin, by the turnover and shedding of LPS-containing enterocytes into the gut lumen, that has not been previously described.

Analysis of the F-actin binding fragments of vinculin using stopped-flow and dynamic light-scattering measurements.

Using amino acids 884-1066 and 884-1012 expressed from chicken vinculin as fusion proteins with schistosomal glutathione S-transferase, we determined the binding kinetics of the protein fragments with F-actin. We established by the stopped-flow method a two-step binding process: an initial rapid reaction followed by a slower process. The latter is attributed to F-actin cross-linking and/or bundling, which was previously detected by viscometry and electron microscopy [Johnson, R. P. & Craig, S. W. (1995) Nature 373, 261-264]. This is also supported by dynamic light-scattering measurements, indicating dramatic changes in the internal actin filament dynamics, i.e. in bending undulations due to thermal noise. The similar size of the binding reaction for both fusion proteins with F-actin indicates that the F-actin binding site(s) on vinculin are located between residues 884-1012. No binding of pure glutathione S-transferase or its fusion protein with vinculin peptide 1012-1066 with F-actin was detected by either method.

Inhibition of cellular proliferation by the Wilms tumor suppressor WT1 requires association with the inducible chaperone Hsp70.

The Wilms tumor suppressor WT1 encodes a zinc finger transcription factor that is expressed in glomerular podocytes during a narrow window in kidney development. By immunoprecipitation and protein microsequencing analysis, we have identified a major cellular protein associated with endogenous WT1 to be the inducible chaperone Hsp70. WT1 and Hsp70 are physically associated in embryonic rat kidney cells, in primary Wilms tumor specimens and in cultured cells with inducible expression of WT1. Colocalization of WT1 and Hsp70 is evident within podocytes of the developing kidney, and Hsp70 is recruited to the characteristic subnuclear clusters that contain WT1. The amino-terminal transactivation domain of WT1 is required for binding to Hsp70, and expression of that domain itself is sufficient to induce expression of Hsp70 through the heat shock element (HSE). Substitution of a heterologous Hsp70-binding domain derived from human DNAJ is sufficient to restore the functional properties of a WT1 protein with an amino-terminal deletion, an effect that is abrogated by a point mutation in DNAJ that reduces binding to Hsp70. These observations indicate that Hsp70 is an important cofactor for the function of WT1, and suggest a potential role for this chaperone during kidney differentiation.

Differences in F9 and 5.51 cell elasticity determined by cell poking and atomic force microscopy.

We studied the elasticity of both a wild type (F9) mouse embryonic carcinoma and a vinculin-deficient (5.51) cell line, which was produced by chemical mutagenesis. Using cell poking, we measured the effects of loss of vinculin on the elastic properties of these cells. F9 cells were about 20% more resistant to indentation by the cell poker (a glass stylus) than were 5.51 cells. Using the atomic force microscope to map the elasticity of wild type and vinculin-deficient cells by 128 X 128 force scans, we observed a correlation of elasticity with cell poking elastometric measurements. These findings, as well as previous atomic force, rheologic, and magnetometric measurements [Goldmann and Ezzell, Exp. Cell Res. 226 (1996) 234-237; Ezzell et al., Exp. Cell Res. 231 (1997) 14-26], indicate that vinculin is an integral part of the cytoskeletal network.

Differences in elasticity of vinculin-deficient F9 cells measured by magnetometry and atomic force microscopy.

We have investigated a mouse F9 embryonic carcinoma cell line, in which both vinculin genes were inactivated by homologous recombination, that exhibits defective adhesion and spreading [Coll et al. (1995) Proc. Natl. Acad. Sci. USA 92, 9161-9165]. Using a magnetometer and RGD-coated magnetic microbeads, we measured the local effect of loss and replacement of vinculin on mechanical force transfer across integrins. Vinculin-deficient F9Vin(-/-) cells showed a 21% difference in relative stiffness compared to wild-type cells. This was restored to near wild-type levels after transfection and constitutive expression of increasing amounts of vinculin into F9Vin(-/-) cells. In contrast, the transfection of vinculin constructs deficient in amino acids 1-288 (containing the talin- and alpha-actinin-binding site) or substituting tyrosine for phenylalanine (phosphorylation site, amino acid 822) in F9Vin(-/-) cells resulted in partial restoration of stiffness. Using atomic force microscopy to map the relative elasticity of entire F9 cells by 128 x 128 (n = 16,384) force scans, we observed a correlation with magnetometer measurements. These findings suggest that vinculin may promote cell adhesions and spreading by stabilizing focal adhesions and transferring mechanical stresses that drive cytoskeletal remodeling, thereby affecting the elastic properties of the cell.

Relationship of tissue and cellular interleukin-1 and lipopolysaccharide after endotoxemia and bacteremia.

Distributions of immunoreactive interleukin-1 (IL-1) and lipopolysaccharide (LPS) were studied in the tissues of rats after intravenous injection of purified LPS or live Escherichia coli bacteria. IL-1 staining in the spleen peaked at 4-8 h, colocalized with LPS in marginal zone macrophages, and was undetectable 24 h after injection, whereas LPS staining peaked at 24 h and was detectable for 4 weeks. The tissue IL-1 response was similar for LPS and live bacteria. Thus, tissue IL-1 is down-regulated within hours despite maintenance of LPS in the same cells for weeks. Macrophages in liver and lung had only slight IL-1 staining despite intense staining for LPS. Tissue IL-1 production appears to be differentially regulated after gram-negative bacteremia; LPS cleared by liver and lung macrophages elicit minimal IL-1, whereas there is high local IL-1 production in the marginal zone of the spleen that may increase immune responses to bacterial wall antigens.

Viscoelasticity of actin-gelsolin networks in the presence of filamin.

Cross-linking of actin filaments by filamin by means of frequency-dependent rheology yields an increase in the filament's elasticity and stiffness. Higher cross-linker (filamin) ratios are required for mean actin-filament lengths of 5-6 microm than for random-length distribution of actin filaments. The loss modulus (i.e. the viscous portion) in the region of the internal-chain dynamics [G"(omega) approximately omega(alpha)] is influenced by the cross-linking of filaments, and with an increasing molar ratio of filamin/actin a reduction of alpha is observed. Rheological measurements reveal that actin networks are already formed at the polymerizing stage at a molar ratio of filamin/actin of less than 1:100, and electron micrographs show phase separation of actin/filament networks of low density and of actin/filament bundles.

Digital image processing for microscopy.

The author introduces some of the basic concepts in the theory and mechanics of digital signal processing techniques as applied to microscopy. With the right software and a sound approach, a great deal of information can be retrieved from noisy or degraded images. The author also discusses manipulation of images in order to produce a clearer images for the observer.

Acidosis induces hyperpermeability in Caco-2BBe cultured intestinal epithelial monolayers.

We previously demonstrated that ileal mucosal acidosis in pigs reversibly increases intestinal permeability to hydrophilic macromolecules, even in the absence of tissue hypoxia [A.L. Salzman, H. Wang, P.S. Wollert, T.J. Vandermeer, C.C. Compton, A.G. Denenberg, and M.P. Fink. Am. J. Physiol. 266 (Gastrointest, Liver Physiol, 29): G633-G646, 1994]. In an effort to further explore the mechanism(s) underlying this phenomenon, we examined the effect of acidic pH on the permeability characteristics of cultured Caco-2BBe (human intestinal epithelial) cells grown as monolayers on permeable supports. Permeability was determined by measuring the mucosal-to-basolateral flux of fluorescein disulfonic acid (FS; mol wt 478 Dal, fluorescein isothiocyanate-labeled dextran (FD4; average mol wt 4 kDa), or [3H]mannitol. Incubation of monolayers under hypercarbic conditions or with acidific bicarbonate-free medium significantly increased permeability to FS, FD4, and mannitol in a manner dependent on both time and pH. Incubation in medium at pH 5.43 for 24 h increased the release of lactate dehydrogenase and decreased the intensity of staining with calcein-acetoxymethyl ester, findings that are indicative of plasma membrane injury; nevertheless, the percentage of nonviable cells did not increase. Ultrastructural analyses revealed evidence of increased paracellular trafficking of horseradish peroxidase after incubation of monolayers under acidic conditions. Fluorescence confocal microscopy and temperature studies demonstrated that incubation at pH 5.43 induced an increase in both the intracellular uptake of FD4 and the activation energy for FS permeation across Caco-2BBe monolayers, respectively, suggesting increased transcellular permeation. Exposure to acidic conditions also decreased cellular levels of ATP. We conclude that acidosis increases both paracellular and transcellular permeability to hydrophilic macromolecules and leads to depletion of ATP.

Vinculin promotes cell spreading by mechanically coupling integrins to the cytoskeleton.

Mouse F9 embryonic carcinoma 5.51 cells that lack the cytoskeletal protein vinculin spread poorly on extracellular matrix compared with wild-type F9 cells or two vinculin-transfected clones (5.51Vin3 and Vin4; Samuels et al., 1993, J. Cell Biol. 121, 909-921). In the present study, we used this model system to determine how the presence of vinculin promotes cytoskeletal alterations and associated changes in cell shape. Microscopic analysis of cell spreading at early times, revealed that 5.51 cells retained the ability to form filopodia; however, they could not form lamellipodia, assemble stress fibers, or efficiently spread over the culture substrate. Detergent (Triton X-100) studies revealed that these major differences in cell morphology and cytoskeletal organization did not result from differences in levels of total polymerized or cross-linked actin. Biochemical studies showed that 5.51 cells, in addition to lacking vinculin, exhibited slightly reduced levels of alpha-actinin and paxillin in their detergent-insoluble cytoskeleton. The absence of vinculin correlated with a decrease in the mechanical stiffness of the integrin-cytoskeleton linkage, as measured using cell magnetometry. Furthermore, when vinculin was replaced by transfection in 5.51Vin3 and 5.51Vin4 cells, the levels of cytoskeletal-associated alpha-actinin and paxillin, the efficiency of transmembrane mechanical coupling, and the formation of actin stress fibers were all restored to near wild-type levels. These findings suggest that vinculin may promote cell spreading by stabilizing focal adhesions and transferring mechanical stresses that drive cytoskeletal remodeling, rather than by altering the total level of actin polymerization or cross-linking.

Intestinal epithelial restitution. Involvement of specific laminin isoforms and integrin laminin receptors in wound closure of a transformed model epithelium.

Disruptions in the mucosal lining of the gastrointestinal tract reseal by epithelial cell migration, a process termed restitution. We examined the involvement of laminin isoforms and their integrin receptors in restitution using the intestinal epithelial cell line T84. T84 cells express primarily laminins 5, 6, and 7 as indicated by immunostaining using laminin subunit-specific monoclonal antibodies (MAbs). A MAb (BM2) specific for the laminin alpha 3 subunit, a component of laminins 5, 6, and 7, completely inhibited the closure of mechanical wounds in T84 monolayers. Confocal microscopy using MAbs BM2 (laminin alpha 3 subunit) and 6F12 (laminin beta 3 subunit) revealed that laminin-5 is deposited in a basal matrix that extends into the wound. The MAbs 4E10 (laminin beta 1 subunit) and C4 (laminin beta 2 subunit) stained the lateral membranes between T84 cells. This staining was enhanced in cells adjoining wounds. Because T84 cells stained faintly with MAbs 4C7 (laminin alpha 1 subunit) and with MAbs 4F11 and 1B4 (laminin alpha 2 subunit), we suggest that expression of laminins 6 and 7 is enhanced in response to wounding. The alpha 3 beta 1 integrin and the alpha 6-containing integrins function in wound closure because MAbs specific for the beta 1 integrin subunit (MAb13), the alpha 3 subunit (IVA5), and the alpha 6 subunit (2B7) potently inhibited T84 migration into wounds. Immunofluorescence using UMA9, a beta 4-integrin-specific MAb, revealed that alpha 6 beta 4 integrin exists in a Triton-X-100-insoluble structure at the basal surface and that the staining of this structure is enhanced in cells adjoining wounds. In addition, a Triton-X-100-soluble pool of alpha 6 beta 4, as well as alpha 3 beta 1 and presumably alpha 6 beta 1, was found along lateral surfaces of T84 cells. On flattened cells adjoining wounds, staining for these integrins was distributed diffusely, suggesting a redistribution that accompanies cell migration. Taken together, these data suggest that wound-induced epithelial cell migration is a finely tuned process that is dependent upon the regulated function and localization of specific laminins and their integrin receptors.

Müllerian inhibiting substance inhibits branching morphogenesis and induces apoptosis in fetal rat lung.

Müllerian inhibiting substance (MIS) is a glycoprotein hormone required for normal male reproductive tract development; it is presumed to signal through a heteromeric complex of type I and type II receptors. MIS exposure produces a paracrine-mediated regression of the embryonic Müllerian duct with histological changes consistent with apoptosis. MIS has also been shown to inhibit fetal lung development in vitro and in vivo, although the mechanism of this inhibition is unknown. The primordial lung and gonad are anatomically proximate on embryonic day 13.5, raising the possibility of a paracrine-mediated influence of MIS in male embryos on lung as well as MIS effecting dissolution of the Müllerian duct. We hypothesized that a negative regulatory event(s) might occur in the lung, as occurs in the duct, at the onset of MIS protein expression; thus, apoptosis and branching morphogenesis were studied in explanted fetal rat lungs incubated with proteolytically activated MIS. MIS exposure resulted in reduced total lung bud number as well as lung perimeter length. Explanted lungs exposed to MIS also exhibited numerous apoptotic bodies. To assess whether this MIS-induced phenomenon in lung might be mediated by the MIS type II receptor (MIS RII), reverse transcriptase-PCR performed on multiple fetal rat lung RNA samples using oligonucleotide primers designed from the 3'-untranslated region of rat MIS RII complementary DNA showed a product of the expected size that when sequenced was nearly identical to rat MIS RII. Northern blot analysis using polyadenylated fetal rat lung RNA and a 3'-MIS RII probe revealed a 2-kilobase transcript that was also seen in testicular messenger RNA. These studies show that the putative ligand binding receptor for MIS is expressed in embryonic lung, where MIS negatively modulates branching and activates apoptosis. We speculate that the mechanism of MIS-induced inhibition of lung development in the male fetus begins with MIS binding to the MIS RII, followed by a signaling cascade resulting in delayed airway branching temporally associated with enhanced apoptosis.

Examining F-actin interaction with intact talin and talin head and tail fragment using static and dynamic light scattering.

We examined the binding kinetics of intact talin and talin head and tail fragment with F-actin at pH 7.0 and at low ionic strength. We observed by a transient kinetic method a fast followed by a slower binding process for intact talin and talin tail fragment with filamentous actin. The latter can be attributed to F-actin cross-linking and/or bundling, which was observed in cosedimentation assays as well as by low shear viscometry and electron microscopy [Zhang, J., Robson, R. M., Schmidt, J. M. & Stromer, M. H. (1996) Biochem. Biophys. Res. Commun. 218, 530-537]. This finding is supported by dynamic light scattering measurements, indicating changes in internal actin filament dynamics due to cross-linking/bundling events with intact talin and talin tail fragment. No binding of the talin head fragment with F-actin was detected by either method.

Viscoelasticity in wild-type and vinculin-deficient (5.51) mouse F9 embryonic carcinoma cells examined by atomic force microscopy and rheology.

We have been studying mouse F9 embryonic carcinoma cells which contain no detectable vinculin protein (5.51 cells), and compared them with F9 wild-type cells. Employing atomic force microscopy, we probed the elastic properties of individual F9 wild-type and 5.51 cells by measuring the dynamic response of controlled loads of the cantilever tip. An elastic modulus (Young) of approximately 3.8 and approximately 2.5 kPa was calculated for wild-type and 5.51 cells, respectively. Using disc rheometry, we detected a marked change in shear of a 1000g pellet of approximately 55 x 10(6) cells between wild-type and 5.51 mutants. These differences are attributed to the loss of vinculin and altered cytoskeletal organization in these cells.

Human fetal enterocytes in vitro: modulation of the phenotype by extracellular matrix.

The differentiation of small intestinal epithelial cells may require stimulation by microenvironmental factors in vivo. In this study, the effects of mesenchymal and luminal elements in nonmalignant epithelia] cells isolated from the human fetus were studied in vitro. Enterocytes from the human fetus were cultured and microenvironmental factors were added in stages, each stage more closely approximating the microenvironment in vivo. Four stages were examined: epithelial cells derived on plastic from intestinal culture and grown as a cell clone, the same cells grown on connective tissue support, primary epithelial explants grown on fibroblasts with a laminin base, and primary epithelial explants grown on fibroblasts and laminin with n-butyrate added to the incubation medium. The epithelial cell clone dedifferentiated when grown on plastic; however, the cells expressed cytokeratins and villin as evidence of their epithelial cell origin. Human connective tissue matrix from Engelbreth-Holm-Swarm sarcoma cells (Matrigel) modulated their phenotype: alkaline phosphatase activity increased, microvilli developed on their apical surface, and the profile of insulin-like growth factor binding proteins resembled that secreted by differentiated enterocytes. Epithelial cells taken directly from the human fetus as primary cultures and grown as explants on fibroblasts and laminin expressed greater specific enzyme activities in brush border membrane fractions than the cell clone. These activities were enhanced by the luminal molecule sodium butyrate. Thus the sequential addition of connective tissue and luminal molecules to nonmalignant epithelia] cells in vitro induces a spectrum of changes in the epithelial cell phenotype toward full differentiation.

Hyperpermeability and ATP depletion induced by chronic hypoxia or glycolytic inhibition in Caco-2BBe monolayers.

Previous studies have documented that the barrier function of epithelial or endothelial monolayers is deranged when cellular ATP levels are rapidly decreased to very low levels by inhibitors of mitochondrial and glycolytic function. We hypothesized that lesser degrees of ATP depletion also might affect epithelial permeability, particularly if the perturbation were sustained for a prolonged interval. Using Caco-2BBe cells grown on permeable supports mounted in bicameral chambers, we assessed permeability by measuring the apical-to-basolateral clearance (flux divided by apical compartment concentration) of fluorescein disulfonic acid. ATP was depleted by incubating cells in glucose-free (Glu-) medium containing 10 mM 2-deoxyglucose (2-DOG) for 12, 24, or 48 h or under an anoxic atmosphere for 24, 48, or 72 h. Although both models of energy depletion were characterized by significant derangements in barrier function, metabolic inhibition with 2-DOG/ Glu- resulted in greater increases in permeability and more profound decrements in cellular ATP content. Morphological studies using electron and confocal fluorescence microscopy showed structural changes in individual cells and derangements in the normal distribution of perijunctional actin after monolayers were incubated with 2-DOG/Glu- but not after incubation under an anoxic atmosphere. Addition of 10 mM lactic acid (final pH 6.7) provided significant protection against both hyperpermeability and ATP depletion induced by 2-DOG/Glu-. We conclude that moderate degrees of ATP depletion are sufficient to increase the permeability of Caco-2BBe monolayers and that lactic acidosis helps to preserve ATP content, barrier function, and morphological integrity in hypoxic intestinal epithelial cells.

Mouse preimplantation blastocysts adhere to cells expressing the transmembrane form of heparin-binding EGF-like growth factor.

Previous studies have shown that heparin-binding epidermal growth factor (EGF)-like growth factor (HB-EGF) mRNA is synthesized in the mouse uterine luminal epithelium temporally, just prior to implantation, and spatially, only at the site of blastocyst apposition (Das, S. K., Wang, X. N., Paria, B. C., Damm, D., Abraham, J. A., Klagsbrun, M., Andrews, G. K. and Dey, S. K. (1994) Development 120, 1071-1083). HB-EGF is synthesized as a transmembrane protein (HB-EGF TM) that can be processed to release the soluble growth factor. An antibody that cross-reacts only with the transmembrane form detected HB-EGF TM in uterine luminal epithelium in a spatial manner similar to that of HB-EGF mRNA. HB-EGF TM is a juxtacrine growth factor that mediates cell-cell contact. To ascertain if HB-EGF TM could be an adhesion factor for blastocysts, a mouse cell line synthesizing human HB-EGF TM was co-cultured with mouse blastocysts. Cells synthesizing HB-EGF TM adhered to day-4 mouse blastocysts more extensively than parental cells or cells synthesizing a constitutively secreted form of HB-EGF. Adhesion of cells synthesizing HB-EGF TM to blastocysts was inhibited by excess recombinant HB-EGF but less so by TGF-alpha. Adhesion was also inhibited by the synthetic peptide P21 corresponding to the HB-EGF heparin binding domain, and by incubating the blastocysts with heparinase. In addition, adhesion to delayed implanting dormant blastocysts, which lack EGF receptor (EGFR), was diminished relative to normal blastocysts. These results suggested that adhesion between blastocysts and cells synthesizing HB-EGF TM was mediated via interaction with both blastocyst EGFR and heparan sulfate proteoglycan (HSPG). It was concluded that HB-EGF TM, which is synthesized exclusively in the luminal epithelium at the site of blastocyst apposition, and which is a juxtacrine adhesion factor for blastocysts, could be one of the mediators of blastocyst adhesion to the uterus in the process of implantation.

Culture matrix configuration and composition in the maintenance of hepatocyte polarity and function.

Several extracellular matrix (ECM) configurations involving type I collagen and Matrigel were examined for their ability to support differentiated function and polarity of cultured adult rat hepatocytes. Collagen sandwich- and Matrigel-based cultures yielded superior and comparable albumin secretion for at least 2 weeks. In collagen sandwich, hepatocytes were polygonal, and formed multicellular arrays. Collagen sandwich was also found to promote in vivo-like polarization of F-actin, cell adhesion molecules (E-cadherin), and lateral (Na+, K(+)-ATPase, glucose transporter) and apical (dipeptidyl peptidase, aminopeptidase) membrane polarity markers, but not the expression of the gap junction protein connexin 32 and the epidermal growth factor (EGF) receptor. In contrast, hepatocytes cultured in or on Matrigel were more rounded and formed aggregates. Matrigel-based cultures also elicited detectable levels of connexin and EGF receptor and an altered distribution of F-actin, E-cadherin, and apical and lateral membrane proteins. Composite sandwich configurations containing collagen I and Matrigel restored markers lacking in the collagen sandwich, and showed a variable morphology and membrane polarity. Hepatocyte polarity could thus be manipulated by the overall ECM composition. Furthermore, in composite sandwich cultures, these manipulations can be effected largely independent of changes in hepatocyte morphology and albumin secretion.

F-actin, a model polymer for semiflexible chains in dilute, semidilute, and liquid crystalline solutions.

Single actin filaments were analyzed in solutions ranging from dilute (0.2 microgram/ml), where filaments interact only with solvent, to concentrations (4.0 mg/ml) at which F-actin forms a nematic phase. A persistence length of approximately 1.8 microns and an average length of approximately 22 microns (Kaufmann et al., 1992) identify actin as a model for studying the dynamics of semiflexible polymers. In dilute solutions the filaments exhibit thermal bending undulations in addition to diffusive motion. At higher semidilute concentrations (1.4 mg/ml) three-dimensional reconstructions of confocal images of fluorescently labeled filaments in a matrix of unlabeled F-actin reveal steric interactions between filaments, which account for the viscoelastic behavior of these solutions. The restricted undulations of these labeled chains reveal the virtual tube formed around a filament by the surrounding actin. The average tube diameter scales with monomer concentration c as varies; is directly proportional to c-(0.5 +/- 0.15). The diffusion of filaments in semidilute solutions (c = (0.1-2.0) mg/ml) is dominated by diffusion along the filament contour (reptation), and constraint release by remodeling of the surrounding filaments is rare. The self-diffusion coefficient D parallel along the tube decreases linearly with the chain length for semidilute solutions. For concentrations > 2.5 mg/ml a transition occurs from an isotropic entangled phase to a coexistence between isotropic and nematic domains. Analysis of the molecular motions of filaments suggests that the filaments in the aligned domains are in thermal equilibrium and that the diffusion coefficient parallel to the director D parallel is nearly independent of filament length. We also report the novel direct observation of u-shaped defects, called hairpins, in the nematic domains.