Interdependence of profilin, cation, and nucleotide binding to vertebrate non-muscle actin.

The interaction of profilin and non-muscle beta,gamma-actin prepared from bovine spleen has been investigated under physiologic ionic conditions. Profilin binding to actin decreases the affinity of actin for MgADP and MgATP by about 65- and 13-fold, respectively. Kinetic measurements indicate that profilin binding to actin weakens the affinity of actin for nucleotides primarily due to an increased nucleotide dissociation rate constant, but the nucleotide association rate constant is also increased about 2-fold. Removal of the actin-bound nucleotide and divalent cation produces the labile intermediate species in the nucleotide exchange reaction, nucleotide free actin (NF-actin), and increases the affinity of actin for profilin about 10-fold. Profilin binds NF-actin with high affinity, K(D) = 0.013 microM, and slows the observed denaturation rate of NF-actin. Addition of ATP to NF-actin weakens the affinity for profilin and addition of Mg(2+) to ATP-actin further weakens the affinity for profilin. The high-affinity Mg(2+) of actin regulates binding of both nucleotide and profilin to actin and is important for actin interdomain coupling. The data suggest that profilin binding to actin weakens nucleotide binding to actin by disrupting Mg(2+) coordination in the actin central cleft.

Cross-linked dimers with nucleating activity in actin prepared from muscle acetone powder.

A covalently linked actin dimer is identified in solutions of actin prepared from an acetone powder from skeletal muscle. This actin dimer acts as an actin nucleating factor (ANF), decreasing the half-time for spontaneous actin polymerization. ANF reacts with antibodies to both the N- and C-terminal portions of actin on Western blots and migrates during reduced polyacrylamide gel electrophoresis like actin cross-linked with N, N'-p-phenylenebismaleimide. The origin of the cross-linked dimer appears to be related to the presence of carbonyl groups in purified actin. A large number of carbonyls (approximately 0.3/actin) are introduced into actin during the prolonged treatment with acetone in the preparation of the muscle acetone powder from which actin is extracted. Actin extracted from acetone powder prepared by a single acetone wash and actin prepared from bovine spleen, which is not washed with acetone, both contain fewer carbonyl groups (approximately 0.05 carbonyl/actin). ANF forms spontaneously in solutions of polymer actin containing 0.3 carbonyl/actin. We speculate that a reaction between a carbonyl on one actin polymer subunit and a lysine on a neighboring subunit is responsible for ANF formation. The presence of cross-linked actin dimers in commonly used skeletal muscle actin preparations could certainly affect studies of actin polymerization and, particularly, studies of the nucleation reaction. The physiological relevance of ANF is not clear, but given the large cellular concentration of actin, similar reactions yielding ANF could occur in vivo when increased levels of reactive oxygen species are present.

Impact of profilin on actin-bound nucleotide exchange and actin polymerization dynamics.

We have investigated the effects of profilin on nucleotide binding to actin and on steady state actin polymerization. The rate constants for the dissociation of ATP and ADP from monomeric Mg-actin at physiological conditions are 0.003 and 0.009 s-1, respectively. Profilin increases these dissociation rate constants to 0.08 s-1 for MgATP-actin and 1.4 s-1 for MgADP-actin. Thus, profilin can increase the rate of exchange of actin-bound ADP for ATP by 140-fold. The affinity of profilin for monomeric actin is found to be similar for MgATP-actin and MgADP-actin. Continuous sonication was used to allow study of solutions having sustained high filament end concentrations. During sonication at steady state, F-actin depolymerizes toward the critical concentration of ADP-actin [Pantaloni, D., et al. (1984)J. Biol. Chem. 259, 6274-6283], our analysis indicates that under these conditions a significant number of filaments contain terminal ADP-actin subunits. Addition of profilin to this system increases the polymer concentration and increases the steady state ATPase activity during sonication. These data are explained by the fast exchange of ATP for ADP on the profilin-ADP-actin complex, resulting in rapid ATP-actin regeneration. An important function of profilin may be to provide the growing ends of filaments with ATP-actin during periods when the monomer cycling rate exceeds the intrinsic nucleotide exchange rate of monomeric actin.

Ca2+ regulation of gelsolin activity: binding and severing of F-actin.

Regulation of the F-actin severing activity of gelsolin by Ca2+ has been investigated under physiologic ionic conditions. Tryptophan fluorescence intensity measurements indicate that gelsolin contains at least two Ca2+ binding sites with affinities of 2.5 x 10(7) M-1 and 1.5 x 10(5) M-1. At F-actin and gelsolin concentrations in the range of those found intracellularly, gelsolin is able to bind F-actin with half-maximum binding at 0.14 microM free Ca2+ concentration. Steady-state measurements of gelsolin-induced actin depolymerization suggest that half-maximum depolymerization occurs at approximately 0.4 microM free Ca2+ concentration. Dynamic light scattering measurements of the translational diffusion coefficient for actin filaments and nucleated polymerization assays for number concentration of actin filaments both indicate that severing of F-actin occurs slowly at micromolar free Ca2+ concentrations. The data suggest that binding of Ca2+ to the gelsolin-F-actin complex is the rate-limiting step for F-actin severing by gelsolin; this Ca2+ binding event is a committed step that results in a Ca2+ ion bound at a high-affinity, EGTA-resistant site. The very high affinity of gelsolin for the barbed end of an actin filament drives the binding reaction equilibrium toward completion under conditions where the reaction rate is slow.

Severing of F-actin by the amino-terminal half of gelsolin suggests internal cooperativity in gelsolin.

Gelsolin is a Ca2+-regulated actin-binding protein that can sever, cap, and nucleate growth from the pointed ends of actin filaments. In this study we have measured the binding of the amino-terminal half of gelsolin, G1-3, to pyrene-labeled F-actin as a function of Ca2+ concentration. The rate of binding is shown to be dependent on micromolar concentrations of Ca2+. Independent experiments demonstrate that conformational changes in G1-3 are induced by micromolar concentrations of Ca2+. Titrations of pyrene-F-actin with G1-3 and gelsolin show that the quenching of pyrene fluorescence is identical in extent and stoichiometry for both G1-3 and gelsolin. In contrast, severing of F-actin by G1-3 is found to be much less efficient than is severing by gelsolin. In experiments in which F-actin severing is quantitatively measured, the filament number is found to be proportional to the 1.35 power of the G1-3 concentration. This deviation from linearity may be explained by cooperativity; the binding of two G1-3 molecules in close proximity may lead to cooperative severing of the polymer, thus increasing the severing efficiency. This model is supported by experiments that show that the efficiency of G1-3 severing of F-actin increases with increasing G1-3:F-actin ratios. Extrapolating from these results, we conclude that G4-6, the carboxyl-terminal half of gelsolin, has an active role in the severing of F-actin by intact gelsolin. Whereas F-actin severing by G1-3 is enhanced by cooperative binding of two separate G1-3 molecules, cooperativity is inherent to intact gelsolin because the cooperative partners are covalently linked.

Kinetics of gelsolin interaction with phalloidin-stabilized F-actin. Rate constants for binding and severing.

The kinetics of gelsolin interaction with actin filaments have been investigated using two fluorescent probes, tetramethylrhodamine isothiocyanate-labeled phalloidin bound to F-actin and N-(1-pyrenyl)iodoacetamide-labeled actin. We have also analyzed the F-actin severing by gelsolin using an assay for actin filaments which measures the polymerization rate of monomeric actin added to the gelsolin-severed filaments. Phalloidin-stabilized actin filaments were used in order to minimize the depolymerization reaction and thus simplify the kinetic analysis. Because gelsolin activity is Ca(2+)-activated, experiments were conducted in the presence of 0.5 mM CaCl2 to ensure maximal activity. We show that the interaction of gelsolin with F-actin may be separated into two distinct kinetic phases which correspond to binding and severing events. Using a two-step model of gelsolin activity, we have determined that gelsolin binds to F-actin with an association rate constant of 2 x 10(7) M-1 s-1, dissociates with a rate constant in the range 0.4-1.2 s-1, and subsequently severs phalloidin-stabilized F-actin with a first-order rate constant of 0.25 s-1. Characterization of the binding and severing reactions will facilitate further investigation of gelsolin activity and its regulation.

Actin-bound nucleotide/divalent cation interactions.

At this point, it may be worthwhile to list, in summary form, the important aspects of divalent cation and nucleotide binding to actin that have been reviewed here: 1) High affinity divalent cation binding to actin is very tight, with equilibrium dissociation constant KCa approximately 1 nM and KMg approximately 5 nM at pH 7.0. 2) The binding kinetics of Ca++ are diffusion limited. Dissociation is slow, with k-Ca approximately 0.015 sec at pH 7.0 (and low ionic strength). 3) The binding kinetics of Mg++ are limited by the characteristics of the Mg++ aquo-ion and are much slower than for Ca++; k-Mg approximately 0.0012 at pH 7.0. 4) Increase in pH or ionic strength weakens divalent cation binding at the high affinity site, primarily by increasing k-Ca and k-Mg. 5) Exchange of Mg++ for Ca++ (or vice versa) at the high affinity site is by a competitive pseudo-first order process with an apparent rate constant (kapp) intermediate between k-Ca and k-Mg and dependent upon the cation concentration ratio [Ca]/[Mg] present. 6) High affinity ATP binding is modulated by the high affinity divalent cation. The cation concentration range over which this modulation occurs is about 100-fold higher for Mg++ than for Ca++, again because of the different characteristics of the Mg++ and Ca++ aquo-ions. 7) At low divalent cation concentrations, ATP dissociation from actin is limited by dissociation of the tightly-bound divalent cation. 8) At high divalent cation concentrations, ATP dissociation probably occurs via dissociation of the divalent cation-nucleotide complex and is quite slow, with dissociation rate constant approximately 0.0005 sec-1. 9) Competitive nucleotide exchange on actin may be described by a pseudo-first order model analogous to that for divalent cation exchange. The pseudo-first order rate constants depend upon the divalent cation concentration. The overall nucleotide exchange rate constant kex depends upon these constants and the solution nucleotide concentration ratio, e.g. [ATP]/[ADP]. The following circumstances develop from the characteristics of the high affinity binding of divalent cation and nucleotide to actin: 1) The standard methods for actin preparation convert in vivo Mg-actin into Ca-actin. 2) Converting Ca-actin back to Mg-actin is not easy. A very low ratio of [Ca]/[Mg] is necessary, which usually requires the use of Ca-cheltors, and a long time (5-10 min) must be allowed for complete exchange.(ABSTRACT TRUNCATED AT 400 WORDS)

Influence of the high affinity divalent cation on actin tryptophan fluorescence.

This study demonstrates that the intrinsic tryptophan fluorescence of actin provides an effective and convenient way of measuring divalent cation exchange and polymerization of native actin. In the measurement of divalent cation exchange, this method is as sensitive as those previously used (8) and provides further evidence of the importance of the tightly bound divalent cation to the properties of actin. In monitoring polymerization, this method cannot compete with the sensitivity of the commonly used pyrene-actin fluorescence. However, some proteins (e.g., profilin) and other agents (e.g., cytochalasin D) that bind to actin are affected by the presence of fluorescent labels, making actin intrinsic fluorescence potentially useful in investigating the interaction of these agents with actin, and in validating data obtained using labeled actin. Our laboratory routinely checks the quality of our native actin preparations using this technique and, more recently, we have used actin tryptophan fluorescence to monitor the nucleating and severing effects of gelsolin on actin. The simplicity of the technique is most appealing, and we expect that a variety of innovative and routine uses will be developed.

Vascular clearance and organ uptake of G- and F-actin in the rat.

This study comparatively evaluated the kinetics of removal and organ distribution of circulating G- and F-actin. Both F- and G-actin were cleared in two phases (fast component with a t1/2 of 3-5 min and a slow component with a t1/2 of hours). There was no effect of dose on either the fast- or slow-compartment clearance kinetics at the doses tested (5-100 micrograms/100 g body wt). However, at the same challenging dose of F- and G-actin, more F-actin was removed during the rapid phase. Although the time constants (Tfast) for F- and G-actin removal from the vasculature during the initial rapid phase were the same, during the slow phase the time constants (Tslow) for removal of F-actin were less (P < 0.001) than that of G-actin. The fraction of F-actin removed during the rapid phase ranged from 33 to 63% and was significantly greater (P < 0.01) than the fraction of G-actin removed during this phase (10-33%). The liver was the main organ of localization, and autoradiographic studies of liver tissue demonstrated that G-actin monomers were removed by Kupffer cells, whereas F-actin was predominantly removed by hepatic sinusoidal endothelial cells. In vivo endotoxin activation of Kupffer cells enhanced the rate of G-actin removal and increased liver localization of G-actin but had no effect on F-actin removal. This further supports a role for Kupffer cells in the clearance of G-actin. These studies therefore demonstrate that F- and G-actin clearance mechanisms are different. G-actin removal, presumably mediated by its binding to vitamin D binding protein, is accomplished by Kupffer cells, whereas F-actin removal at the same doses is due mainly to hepatic endothelial cell uptake.

Actin filament annealing in the presence of ATP and phalloidin.

The re-formation of actin filaments after fragmentation by sonication in the presence of phalloidin and ATP has been found to follow second-order kinetics. The data are described by a model in which the rate of actin filament annealing is proportional to the square of the number concentration of actin filaments and the rate of fragmentation is proportional to the actin polymer concentration. In the presence of 100 mM KCl, 1 mM MgCl2, and equimolar phalloidin, the second-order rate constant for annealing of actin filaments is 2.2 x 10(6) M-1 s-1 and the first-order rate constant for fragmentation is 7 x 10(-7) s-1. In addition, the observed pseudo-first-order rate constant for annealing was found to increase with increasing ionic strength. Thus, annealing may play a major part in the length redistribution phase of actin polymerization and may be important for actin filament rearrangement in the cell.

Nucleotide exchange and rheometric studies with F-actin prepared from ATP- or ADP-monomeric actin.

It has recently been reported that polymer actin made from monomer containing ATP (ATP-actin) differed in EM appearance and rheological characteristics from polymer made from ADP-containing monomers (ADP-actin). Further, it was postulated that the ATP-actin polymer was more rigid due to storage of the energy released by ATP hydrolysis during polymerization (Janmey et al. 1990. Nature 347:95-99). Electron micrographs of our preparations of ADP-actin and ATP-actin polymers show no major differences in appearance of the filaments. Moreover, the dynamic viscosity parameters G' and G" measured for ATP-actin and ADP-actin polymers are very different from those reported by Janmey et al., in absolute value, in relative differences, and in frequency dependence. We suggest that the relatively small differences observed between ATP-actin and ADP-actin polymer rheological parameters could be due to small differences either in flexibility or, more probably, in filament lengths. We have measured nucleotide exchange on ATP-actin and ADP-actin polymers by incorporation of alpha-32P-ATP and found it to be very slow, in agreement with earlier literature reports, and in contradiction to the faster exchange rates reported by Janmey et al. This exchange rate is much too slow to cause "reversal" of ADP-actin polymer ATP-actin polymer as reported by Janmey et al. Thus our results do not support the notion that the energy of actin-bound ATP hydrolysis is trapped in and significantly modifies the actin polymer structure.

Nucleotide binding to actin. Cation dependence of nucleotide dissociation and exchange rates.

We have reinvestigated nucleotide binding to actin in order to resolve conflicts regarding the mechanism of nucleotide dissociation and exchange. We present evidence that supports a mechanism for nucleotide binding to actin in which the tightly bound divalent cation (Ca2+ or Mg2+) directly interacts with the bound nucleotide. The dissociation rates of ATP or ADP from actin are limited by the dissociation of the high affinity divalent cation from actin and vary inversely with free Ca2+ or free Mg2+ concentration. The divalent cation concentration range over which attenuation of the ATP dissociation takes place is about 100-fold greater for Mg2+ than that for Ca2+ due to the much slower association rate constant for Mg2+ compared with Ca2+. The relative affinity for ATP versus ADP is 200:1 for Ca-actin in 100 microM free [Ca2+], and 4:1 for Mg-actin in 100 microM free [Mg2+]. Actin without a tightly bound divalent cation has about a 3-fold greater affinity for ATP than ADP. At constant free divalent cation concentration, the rate of nucleotide exchange on actin is described by competitive binding kinetics.

Tightly-bound divalent cation of actin.

Actin is known to undergo reversible monomer-polymer transitions that coincide with various cell activities such as cell shape changes, locomotion, endocytosis and exocytosis. This dynamic state of actin filament self-assembly and disassembly is thought to be regulated by the properties of the monomeric actin molecule and in vivo by the influence of actin-associated proteins. Of major importance to the properties of the monomeric actin molecule are the presence of one tightly-bound ATP and one tightly-bound divalent cation per molecule. In vivo the divalent cation is thought to be Mg2+ (Mg-actin) but in vitro standard purification procedures result in the preparation of Ca-actin. The affinity of actin for a divalent cation at the tight binding site is in the nanomolar range, much higher than earlier thought. The binding kinetics of Mg2+ and Ca2+ at the high affinity site on actin are considered in terms of a simple competitive binding mechanism. This model adequately describes the published observations regarding divalent cation exchange on actin. The effects of the tightly-bound cation, Mg2+ or Ca2+, on nucleotide binding and exchange on actin, actin ATP hydrolysis activity and nucleation and polymerization of actin are discussed. From the characteristics that are reviewed, it is apparent that the nature of the bound divalent cation has a significant effect on the properties of actin.

Thermodynamics of actin polymerization; influence of the tightly bound divalent cation and nucleotide.

Previous work by this laboratory has shown that the tightly bound divalent cation of actin affects the enthalpy of the polymerization reaction for ATP-actin (Selden et al. (1986) J. Muscle Res. Cell Motil. 7, 215-224). In the present study, we have measured the temperature dependence of polymerization for actin containing ATP or ADP as the bound nucleotide and Mg2+ or Ca2+ (Mg-actin or Ca-actin) as the tightly bound divalent cation. In contrast to the marked effect of the tightly bound divalent cation on enthalpy and entropy changes for the polymerization of ATP-actin, ADP-actin polymerization is affected very little by the tightly bound divalent cation. The Arrhenius and van't Hoff plots for polymerization of Ca-ATP-, Mg-ADP- and Ca-ADP-actin were found to be non-linear. The free energy data for actin polymerization have been analyzed as a second order function of absolute temperature (Osborne et al. (1976) Biochemistry 15, 317-320). The values of the enthalpy change and activation enthalpy change for Ca-ATP-, Mg-ADP- and Ca-ADP-actin polymerization were found to be temperature-dependent, in contrast to those for Mg-ATP-actin, which were nearly constant over the temperature range studied. These results suggest that (1) polymerization of actin which does not contain both Mg2+ and ATP may be a multi-step reaction including a rate-limiting step and (2) Mg-ATP-actin has a unique conformation which enhances its ability to polymerize.

High affinity divalent cation exchange on actin. Association rate measurements support the simple competitive model.

Each actin molecule has one high affinity site which binds a divalent cation. It has been proposed that an isomerization of the actin molecule is involved in divalent cation exchange at this site ("isomerization model," Frieden, C. (1982) J. Biol. Chem. 257, 2882-2886); we have maintained that exchange is by a simple competitive mechanism (Estes, J. E., Selden, L. A., and Gershman, L. C. (1987) J. Biol. Chem. 262, 4952-4957). Here, using fluorescent-labeled actin, we measure the apparent rate constant for exchange (kapp) as a function of the ratio of free Ca2+ and Mg2+ concentrations, ([Ca]/[Mg]), and show that both models are consistent with the data. The major parameter controlling this relationship in the simple competitive exchange model, the ratio of the association rate constants for Ca2+ and Mg2+ to actin (kCa/kMg), is found to have a value of about 90. We have verified this parameter by direct measurements of kCa and kMg, finding that kCa = 1.9 x 10(7) M-1 s-1 and kMg = 2.3 x 10(5) M-1 s-1, consistent with the characteristics of the Ca2+ and Mg2+ aquo ions. The corresponding parameter derived from the isomerization model is not verifiable. We conclude that high affinity divalent cation exchange on actin proceeds by a simple competitive mechanism.

Growth factors derived from a human malignant glioma cell line, U-251MG.

A human malignant glioma cell line, U-251 Mg, cultured under serum free conditions, was shown to produce a growth factor for BALB/c 3T3 cells (glioma-derived growth factor-1, GDGF-1). The biological activity of GDGF-1 resided in a heat- and acid-resistant protein with a molecular weight (MW) of 25 kDa estimated by gel permeation chromatography. GDGF-1 activity was neutralized by a goat anti-human platelet derived growth factor (PDGF) antibody, indicating that the two factors were immunologically related. Furthermore, U-251 Mg cells constitutively expressed c-sis mRNA. When U-251 Mg cells were stimulated with bacterial lipopolysaccharide, 2 novel growth factors (GDGF-2 and GDGF-3) were produced in addition to the PDGF-like substance. GDGF-2 was determined to be greater than 100 kDa MW and was not neutralized by the goat anti-PDGF antiserum. The biological activity of GDGF-3 was also heat- and acid-resistant with an apparent 14 kDa MW. This factor also did not show any common antigenicity with PDGF. GDGF-2 and GDGF-3 are currently under investigation and evidence as to their natures will be published elsewhere. Our findings with this glioma cell line provide further evidence that inappropriate expression of growth factor-related genes could play important autocrine role(s) in the processes leading to malignant transformation and/or uncontrolled proliferation and may provide a paracrine stimulus for such processes as glioma neovascularization.

High affinity divalent cation binding to actin. Effect of low affinity salt binding.

Monomeric actin labeled with the fluorescent probe N-iodoacetyl-N'-(5-sulfo-1-naphthyl)ethylenediamine (1,5-I-AEDANS-actin) displays a fast fluorescence intensity increase immediately upon addition of salt and then a slow fluorescence intensity change concurrent with Ca2+/Mg2+ exchange at the high affinity divalent cation binding site on actin. The fast change appears to reflect competitive binding of K+ at low affinity (nonspecific) sites and of Mg2+ or Ca2+ at low and intermediate affinity sites. Binding of cation at the low affinity sites (but apparently not at the intermediate affinity sites) results in an increase in k-Ca and k-Mg and thus a decrease in affinity for divalent cations at the high affinity site. The effect of Mg2+ on k-Ca is twice that of K+ for equal fractional saturations of the low affinity binding, and the effect of K+ and Mg2+ together on k-Ca reflects competitive binding at the low affinity sites. Thus the affinity and kinetics of divalent cation binding at the high affinity site of actin are significantly affected by concurrent cation binding at low affinity sites.

Preparation and polymerization properties of monomeric ADP-actin.

An improved method for the preparation of Mg-ADP-actin and Ca-ADP-actin which minimizes denaturation of the protein has been developed. Using ADP-actin prepared by this method, we have measured the polymerization characteristics of Mg-ADP-actin and Ca-ADP-actin. In contrast to the significant difference in Mg-ATP-actin and Ca-ATP-actin polymerization characteristics that we reported previously (J. Muscle Res. Cell Motility 7 (1986) 215-224), we show here that values for the critical concentration, the relative rate constant of elongation (mk+) and the relative rate constant of depolymerization (mk-) for Mg-ADP-actin are similar to those for Ca-ADP-actin. The value of mk+ for Mg-ATP-actin is about 8-fold higher than that for Mg-ADP-actin and the value of mk- for Mg-ADP-actin is 3-4-fold higher than that for Mg-ATP-actin. These factors may help explain the observation that the spontaneous nucleation rates of both types of ADP-actin are low in contrast to the rapid nucleation of Mg-ATP-actin.

Human platelet-derived growth factor preparations contain a separate activity which potentiates follicle-stimulating hormone-mediated induction of luteinizing hormone receptor in cultured rat granulosa cells: evidence for transforming growth factor-beta.

The ability of platelet-derived growth factor (PDGF) preparations to potentiate FSH-mediated LH receptor induction in rat granulosa cell cultures was shown to be due to a component distinct from PDGF. Purification of heat-treated platelet lysate by carboxymethyl-Sephadex C-50 and Cibacron blue-Sepharose chromatography, followed by Bio-Gel P-60 chromatography, resulted in the separation of two activities: 1) a growth-promoting activity, P60-PDGF, defined on the basis of increased DNA synthesis in BALB/c-3T3 cells, and 2) a differentiation-promoting activity which enhanced FSH-dependent LH receptor induction in granulosa cells. On the basis of electrophoretic mobility on sodium dodecyl sulfate-polyacrylamide gels, inhibition of tritiated thymidine uptake by epithelial cells, and attenuation of LH/hCG receptor expression in the presence of antitransforming growth factor-beta (anti-TGF beta) immunoglobulin G, the differentiation-promoting component of the preparations appears to be TGF beta. The Bio-Gel fractions that contained TGF beta did not stimulate LH receptor induction of cAMP production in the absence of FSH. PDGF prepared free of TGF beta did not potentiate receptor induction. We conclude, therefore, that the differentiative effects of PDGF previously described in this system are due to TGF beta.