Transglutaminase bonds in neurofibrillary tangles and paired helical filament tau early in Alzheimer's disease.

Transglutaminase-catalyzed epsilon(gamma-glutamyl)lysine cross-links exist in Alzheimer's disease (AD) paired helical filament (PHF) tau protein but not normal soluble tau. To test the hypothesis that these cross-links could play a role in the formation of neurofibrillary tangles (NFT), we used single- and double-label immunofluorescence confocal microscopy and immunoaffinity purification and immunoblotting to examine epsilon(gamma-glutamyl)lysine cross-links in AD and control brains. The number of neurons that are immunoreactive with an antibody directed at the epsilon-(gamma-glutamyl)lysine bond was significantly higher in AD cortex compared with age-matched controls and schizophrenics. PHF tau-directed antibodies AT8, MC-1 and PHF-1 co-localized with epsilon(gamma-glutamyl)lysine immunolabeling in AD NFT. Immunoaffinity purification and immunoblotting experiments demonstrated that PHF tau contains epsilon(gamma-glutamyl)lysine bonds in parietal and frontal cortex in AD. In control cases with NFT present in the entorhinal cortex and hippocampus, indicative of Braak and Braak stage II, epsilon(gamma-glutamyl)lysine bonds were present in PHF tau in parietal and frontal cortex, despite the lack of microscopically detectable NFT or senile plaques in these cortical regions. The presence of PHF tau with epsilon(gamma-glutamyl)lysine bonds in brain regions devoid of NFT in stage II (but regions, which would be expected to contain NFT in stage III) suggests that these bonds occur early in the formation of NFT.     

Increased epsilon(gamma-glutamyl)lysine crosslinking associated with increased protein synthesis in the inner layers of healing rat skin wounds.

Three days after biopsy wounds were made in the dorsal skin of rats the animals were killed and explants of wounded and unwounded skin were incubated for 7 h with either [3H]glutamine or [3H]lysine. Both incubated and fresh control explants were then dissected into three layers which were homogenized, extracted, digested and then assayed for epsilon (gamma-glutamyl)lysine. The concentration of epsilon(gamma-glutamyl)lysine was greater in all three wounded layers than in the corresponding unwounded layers. The concentration in the wounded middle (dermal) layer and in the unwounded middle layer of younger rats was greater than in the unwounded outer (keratinized) layer, which has previously been shown to contain epsilon(gamma-glutamyl)lysine crosslinks. The incorporation of label from both [3H]glutamine and [3H]lysine into buffer-insoluble protein of the middle and inner (muscle) layers was much greater in the wounded explants than in the unwounded. Except for [3H]lysine in the inner layer there was also an increase in the fraction of incorporated label which was converted to epsilon(gamma-glutamyl)lysine. These results show that increased protein biosynthesis during repair in the wounded explants is associated with increased formation of epsilon(gamma-glutamyl)lysine. In addition, they indicate that the crosslink is involved in some process in the middle and inner layers which is distinct from its known function in keratinization of the epidermis.     

Determination of epsilon (gamma-glutamyl)lysine crosslink in proteins using phenylisothiocyanate derivatization and high-pressure liquid chromatographic separation

A sensitive, reproducible high-performance liquid chromatographic method is reported for the determination of the epsilon (gamma-glutamyl)lysine isopeptide bond, which is usually formed by protein-crosslinking transglutaminases between polypeptide chains. The procedure is based on the separation and quantitation of epsilon (gamma-glutamyl)lysine isodipeptide following exhaustive proteolytic digestion of the crosslinked peptide. It involves preliminary separation steps on a cation exchanger resin and a silica HPLC column, precolumn derivatization with phenylisothiocyanate, and reversed-phase high-pressure liquid chromatographic separation on a C18 column. The derivatized isodipeptide gave a linear concentration-response relationship, with a detection limit of 10 pmol/mg of protein. The combination of the preliminary separation steps and the sensitive detection system permits the determination of the epsilon (gamma-glutamyl)lysine crosslink in complex biological systems including total tissue homogenates.     

N epsilon(gamma-glutamyl)lysine crosslinks in the blood clot of the horseshoe crab, Limulus polyphemus.

Clots were allowed to form in samples of whole blood taken from the American horseshoe crab, Limulus polyphemus, in the absence and presence of dansylcadaverine (16), and were analyzed for their contents of N epsilon(gamma-glutamyl)lysine and gamma-glutamyl-dansylcadaverine. Clots obtained without dansylcadaverine yielded significant amounts of N epsilon(gamma-glutamyl)lysine product. Clots formed in the presence of dansylcadaverine yielded only gamma-glutamyl-dansylcadaverine. Formation of these products reflects on the activity of transglutaminase released from the blood cells during coagulation.     

Binding of complement subcomponent C1q to Streptococcus pyogenes: evidence for interactions with the M5 and FcRA76 proteins

Binding of C1q, the first component of the complement system, to some human pathogens has been earlier reported. In the present study, direct binding of C1q to group A streptococci (GAS) of various serotypes as well as some other Gram-positive and Gram-negative species was demonstrated. The interaction between C1q and GAS was investigated more in detail. In hot neutral extracts of a number of GAS strains two components of 64 and 52 kDa, respectively, bound C1q; alkaline and SDS extracts yielded the 52 kDa component as the main C1q-binding substance. Trypsin treatment of the SDS extracts of two GAS strains suggested the C1q-binding component(s) to be of protein nature. C1q-binding material purified from the SDS extract of an avirulent strain, type T27, was separated in 12% SDS-PAGE and probed in Western blot with human C1q and fibrinogen, conjugated to horse radish peroxidase (HRP) as well as rabbit IgG antibodies complexed to HRP (PAP system). The 52 kDa component was non-reactive with fibrinogen or rabbit IgG. However, C1q-binding components purified from the alkaline extracts of two M-positive strains revealed strong binding of either fibrinogen (type M5) or both fibrinogen and rabbit IgG (type M76); the molecular mass of these components, 55 kDa and 43–40 kDa, respectively, was in agreement with the reported molecular mass of the M5 and FcRA76 proteins. Our findings suggest that C1q may interact with GAS through certain M-family proteins as well as by a so far unidentified surface factor of protein nature occurring in most GAS strains. The involvement of M-family proteins, regarded as virulence factors of these organisms, may suggest the interaction of GAS with C1q as biologically important.     

Platelet activation and microfilament bundling

Human platelets were obtained in the fully resting state by treating discoid populations with 1.5 mM tetracaine and in the activated state by treatment with 2 microM A-23187. After gel filtration or washing, respectively, platelet suspensions were lysed with 1% Triton X-100 at pH 6.8. The precipitates from resting platelets viewed by negative staining appeared predominantly granular with a few very short microfilaments. They contained polypeptides of 250, 100, 45, 38, 36.5, and 35 Kdaltons, and three small polypeptides including one with the mobility of profilin on SDS gels. Precipitates from activated platelets lacked this low molecular weight band and contained a major band at 200 Kdaltons with the mobility of myosin; these precipitates had significant K+, Ca++ ATPase activity absent from the precipitate of resting platelets. As seen in negative staining, precipitates from activated platelets contained microfilaments arranged as nets or bundles. The granular resting precipitates were transformed in vitro into microfilament bundles by washing the precipitates in buffer at higher pH (7.6) in the presence of 5 X 10(-5) M calcium chloride.     

A direct continuous spectrophotometric assay for transglutaminase activity

Herein we report the development of a direct and continuous spectrophotometric method for determining transglutaminase (TGase) activity by using N,N-dimethyl-1,4-phenylenediamine (DMPDA) as a gamma-glutamyl acceptor substrate and carbobenzyloxy-l-glutamylglycine (Z-Gln-Gly) as a typical peptide gamma-glutamyl donor substrate. The transamidation activity of TGase can thus be followed by monitoring the increase of absorbance of the resulting anilide product at 278 nm. The extinction coefficient of the authentic, independently synthesized anilide was determined to be epsilon = 8940 +/- 55 M(-1) cm(-1). Using this assay, we determined the apparent K(M) of DMPDA to be 0.25 mM, which compares favorably to the apparent K(M) values determined for other acceptor substrates under conditions where Z-Gln-Gly is also used as the donor substrate, such as N-acetyl-l-lysine methyl ester (9.6 mM) and methylamine (13.1 mM). Finally, the sensitivity of this assay technique was established through the measurement of irreversible inhibition constants for iodoacetamide, determined to be K(I) = 75 +/- 11 nM and k(inact) = (120 +/- 1) x 10(5) M(-1) min(-1).     

Lysine binding to activated human platelets and its similarity to fibrinogen binding

Platelet surface glycoproteins IIb-IIIa are considered to function as the binding site for fibrinogen. Fibrinogen binding is essential for platelet aggregation and several amines have been shown to inhibit this binding. The present study compares the binding properties of ¹²⁵I-fibrinogen and [³H]lysine with platelets activated by the Ca²⁺ ionophore A23187. Many lines of similarities in the binding properties are apparent; however, several differences were also found. The similarities are listed below and the differences are pointed out in parentheses. (a) Marked enhancement by platelet activation; (b) deficiency of binding by thrombasthenic platelets lacking the glycoproteins IIb-IIIa; (c) saturability (fibrinogen binding approaches saturation at more than 12 μM, within 10 min; lysine binding at more than 100 mM within 1 min); (d) Ca²⁺-dependence (at 1 mM Ca²⁺ lysine binding is minute and fibrinogen binding is half-saturated); (e) reversibility; the binding achieved within 10 min is exchangeable; dissociation depends upon time and external ligand concentration; (f) inhibition by the oligoamines His-Lys and Lys₄; (g) inhibition by serum from a thrombasthenic patient who developed anti-glycoproteins IIb-IIIa antibodies; (h) specificity; alanine neither binds to activated platelets nor inhibits fibrinogen binding; it thus appears that the lysine which associates with activated platelets is mostly bound onto the surface of the cells rather than being incorporated; Moreover, the major site of lysine binding seems to be the complexed glycoproteins IIb-IIIa.     

Molecular polymorphism and mechanisms of activation and deactivation of the hydrolytic function of the coupling factor of oxidative phosphorylation.

The 13S coupling factor of oxidative phosphorylation from Alcaligenes faecalis has a latent adenosine triphosphatase (ATPase) function that can be activated by heating at 55 degrees C for 10 min at pH 8.5 in 50% glycerol. The specific activity increases from 0.1 to 20--30 mumol min-1 mg-1. Adenosine 5'-triphosphate (ATP) is not required for stabilization at 55 degreesC when glycerol is present. Activation involves displacement of the endogenous ATPase inhibitor subunit (epsilon subunit), and readdition of this subunit results in deactivation. In the deactivation process the ATPase inhibitor subunit can be replaced by other cationic proteins such as protamine, histones, or poly(lysine). Mg2+ and H+ also are effective deactivators. The fact that every positively charged substance tested deactivated the enzyme suggests that the inhibitor subunit is complexed with the enzyme at a site containing a surplus of negative charges. The activated enzyme is not labile, but it is salt labile, having a half-life of 2-3 min in 0.1 M KI at either 25 or 0 degrees C. The activated ATPase is also inhibited by aurovertin, 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole (NBD), and by the cross-linking agent dimethyl suberimidate. Evidence for polymorphism comes from finding that the properties of the unactivated enzyme (intrinsic ATPase) are different in many ways from the properties of activated ATPase. With respect to the coupling factor's ability to hydrolyze ATP, the data in this study suggest that there are at least four distinct functional allomorphs of this enzyme: (1) the latent enzyme, which has no kinetically measurable ATPase activity, (2) intrinsic ATPase, which is catalyzed by a small percentage of the molecular population that has been activated by some natural mechanism, (3) activated ATPase, which has properties different from those of intrinsic ATPase, and (4) aged activated ATPase, in which some of the properties (Km for substrate, sensitivity to deactivation by Mg2+ and H+) spontaneously change within 30 min.     

Lysine binding to activated human platelets and its similarity to fibrinogen binding

Platelet surface glycoproteins IIb-IIIa are considered to function as the binding site for fibrinogen. Fibrinogen binding is essential for platelet aggregation and several amines have been shown to inhibit this binding. The present study compares the binding properties of 125I-fibrinogen and [3H]lysine with platelets activated by the Ca2+ ionophore A23187. Many lines of similarities in the binding properties are apparent; however, several differences were also found. The similarities are listed below and the differences are pointed out in parentheses. Marked enhancement by platelet activation; deficiency of binding by thrombasthenic platelets lacking the glycoproteins IIb-IIIa; saturability (fibrinogen binding approaches saturation at more than 12 microM, within 10 min; lysine binding at more than 100 mM within 1 min); Ca2+-dependence (at 1 mM Ca2+ lysine binding is minute and fibrinogen binding is half-saturated); reversibility; the binding achieved within 10 min is exchangeable; dissociation depends upon time and external ligand concentration; inhibition by the oligoamines His-Lys and Lys4; inhibition by serum from a thrombasthenic patient who developed anti-glycoproteins IIb-IIIa antibodies; specificity; alanine neither binds to activated platelets nor inhibits fibrinogen binding; it thus appears that the lysine which associates with activated platelets is mostly bound onto the surface of the cells rather than being incorporated. Moreover, the major site of lysine binding seems to be the complexed glycoproteins IIb-IIIa.     

Stabilization of the nuclear matrix by disulfide bridges: identification of matrix polypeptides that form disulfides.

The molecular structure of the nuclear matrix is still poorly understood. We have tried to assess which proteins are important structural elements by examining the process of stabilization of the nuclear matrix by sodium tetrathionate. Sodium tetrathionate stabilizes the nuclear matrix by oxidizing sulfhydryl groups to disulfides. We show that tetrathionate-stabilized matrices are disassembled in buffers containing SDS, indicating that the stabilized nuclear matrix is not a continuous network of cross-linked proteins. Using monobromobimane, a thiol-specific fluorescent reagent, we show that many protein thiols in the stabilized matrix are oxidized. By chromatography on activated thiol-Sepharose we estimated that about 50% of the matrix proteins had oxidized sulfhydryl groups. The protein composition of the material bound to activated thiol-Sepharose was similar to that of the not-bound material. A few proteins are highly enriched in the fraction that was bound to the column. This indicates that many matrix protein species are partially oxidized and that some proteins are completely oxidized. The oxidized protein thiols are found in relatively large complexes as determined by SDS gel-electrophoresis under nonreducing conditions. These results are interpreted in terms of protein-protein interactions in the matrix. The possible role of thiols and disulfides in the in vivo organization of the nucleus is discussed.     

Acceptor-donor relationships in the transglutaminase-mediated cross-linking of lens beta-crystallin subunits

Following the isolation of the N epsilon-(gamma-glutamyl)lysine-containing polymers from human cataracts, our efforts were directed to induce such cross-links experimentally in rabbit lens, and evidence was obtained for the selective reactivities of certain beta-crystallin subunits in this transglutaminase-catalyzed event. In the present work, we examined the enzymatic cross-linking of purified crystallins individually (alpha, beta H, beta L, and gamma) and in combinations, with particular emphasis on forming the approximately 55K dimer. This species was the primary product in the cross-linking of beta H-crystallins; beta L also reacted with transglutaminase. Neither alpha- nor gamma-crystallins formed appreciable amounts of cross-linked structures with transglutaminase. Dansylcadaverine, known to compete against the reactive lysines of proteins in forming N epsilon-(gamma-glutamyl)lysine cross-bridges, was shown to inhibit the generation of dimeric and higher ordered oligomers from beta H and beta L. The fluorescent amine specifically labeled only two subunits in beta H (approximately 29-30K and approximately 26K) and one in beta L (approximately 26K), identifying these substrates as possessing transglutaminase-reactive endo-gamma-glutaminyl residues. An antiserum to bovine beta Bp recognized the approximately 23K subunit of rabbit beta-crystallins and also the approximately 55K dimer, suggesting that the approximately 23K protein participates as a lysine donor in generating the cross-linked dimer with transglutaminase. Inasmuch as the same antiserum reacts with an approximately 50K material reported to appear in increasing amounts with age in human lens, the results lend added support to the physiological significance of transglutaminase in the aging of lens.     

Human factor VIII: purification from commercial factor VIII concentrate, characterization, identification and radiolabeling

Human factor VIII was purified from commercial factor VIII concentrate with a 12% yield. The specific coagulant activity of purified factor VIII was 8,000 units/mg. In the presence of SDS the purified factor VIII consisted of a variety of polypeptides on polyacrylamide gels, ranging between Mr 80,000 and Mr 208,000. In the absence of SDS the purified factor VIII showed an apparent molecular weight of 270,000 upon Sephadex G200 gel-filtration. The purified factor VIII could be activated by thrombin, which resulted in the disappearance of Mr 108,000-208,000 polypeptides in favor of an Mr 92,000 polypeptide. Treatment with factor Xa also activated factor VIII, whereas treatment with activated protein C resulted in the inactivation of coagulant activity. Coagulant-active 125I-factor VIII was prepared using a lactoperoxidase radioiodination procedure. This 125I-factor had the same characteristics as unlabeled factor VIII. All polypeptides could be precipitated with monoclonal antibodies directed against factor VIII. With 125I-factor VIII a pIapp of 5.7 was found in the presence of urea.     

Studies on factor VIII-related protein. I. Ultrastructural and electrophoretic heterogeneity of human factor VIII-related protein.

Human factor VIII-related protein precipitates with specific heterologous anti-bodies directed against purified factor VIII and supports ristocetin-induced aggregation of washed platelets. We purified human factor VIII from cryoprecipitate by subsequent gel filtration on crosslinked large-pore agarose. Factor VIII-related protein appeared as a large aggregate following electrophoresis on 3% polyacrylamide gels in the presence of sodium dodecyl sulfate (SDS). The same material was separated into multiple bands (molecular weight in excess of several millions) following electrophoresis on SDS-1% agarose gels. After complete disulfide reduction of factor VIII-related protein and electrophoresis on SDS-5% polyacrylamide gels a single subunit chain (Mr approximately equal to 200 000) was revealed. Analysis of this protein, in its non-reduced state, by negative contrast electron microscopy showed filaments of markedly variable size. The calculated molecular weight of such filaments ranged from about 0.6.10(6) to 20.10(6). We conclude that size heterogeneity is an essential feature of human factor VIII-related protein.     

Thrombocytin, a serine protease from Bothrops atrox venom. 2. Interaction with platelets and plasma-clotting factors.

Thrombocytin, a serine protease from Bothrops atrox venom, caused platelet aggregation and release of platelet constituents at a concentration of 10(-7) M and clot retraction at a concentration of 2 x 10(-9) M. Thrombocytin was slightly more active when tested on platelets in plasma than on washed platelets suspended in Tyrode--albumin solution. Thrombin was 5 times more active than thrombocytin when tested on platelets in plasma and 50 times more active when tested on washed platelets. The patterns or release induced by thrombocytin and thrombin were similar. Prostaglandin E1 (10(-5) M) produced complete inhibition of platelet release induced by thrombocytin and thrombin. Indomethacin (10(-4) M) was without any effect. Antithrombin III, in the presence of heparin, inhibited the action of thrombocytin on platelets and on a synthetic peptide substrate (Tos-Gly-Pro-Arg-pNA.HCl). formation of an antithrombin III--thrombocytin complex was demonstrated on NaDodSO4--polyacrylamide gel electrophoresis. Hirudin and alpha 1-antitrypsin did not inactivate thrombocytin. Thrombocytin had a low fibrinogen-clotting activity (less than 0.06% that of thrombin). Thrombocytin also caused progressive degradation of the alpha chain of human fibrinogen, and it cleaved prothrombin, releasing products similar to intermediate 1 and fragment 1 produced by thrombin. Thrombocytin activated factor XIII by limited proteolysis and increased the procoagulant activity of factor VIII in a manner analogous to that of thrombin.     

Affinity purification of a rat-brain junctional protein, connexin 43.

Immunocytochemical investigations have previously shown that antibodies specific for mammal connexins labeled in situ rat and mouse brain gap junctions. However brain gap-junction proteins have neither been identified with certainty, nor purified. By immunoblotting, anti-peptide antibodies directed against rat heart connexin 43 (CX43) detect a major protein of 41 kDa in rat brain homogenates. The specificity of these antibodies made it possible to establish an affinity-chromatography purification procedure of the 41-kDa protein. Purified antibodies specific for the sequence SAEQNRMGQ (residues 314-322) of rat heart CX43 were covalently bound to a protein-A-Sepharose-CL-4B matrix. Rat brain homogenates were recycled through the immunomatrix and the material specifically bound to the matrix was then competitively eluted with the peptide SAEQNRMGQY. Analysis by SDS/PAGE of eluates demonstrated that they contain a 41-kDa protein associated with low amounts of high-molecular-mass proteins. By immunoblotting, these proteins were shown to be specifically recognized by antibodies directed against residues 5-17, 55-56, and 314-322 of rat heart CX43. The NH2-terminal partial sequence for the 41-kDa protein was determined by microsequencing and shown to be similar to alpha 1 connexins. This is the first successful purification of a junctional protein from brain tissue and provides direct evidence that the 41-kDa protein is a CX43 gene product.     

Modulation of extracellular-matrix synthesized by cultured stromal cells from normal human breast tissue by epidermal growth factor

A routine, reproducible procedure was developed for the preparation and characterization of stromal cells from normal human breast tissue obtained by reduction mammaplasty. Isolates (n = 15) all exhibited enhanced rates of proliferation, even in the presence of 20% fetal calf serum, when exposed to epidermal growth factor or transforming growth factor a (both 10(-8) M). Cellular responsiveness to these growth factors was consistent with expression of specific surface receptors for epidermal growth factor (approximately 10(4)/cell). In cultures, stromal cells elaborated an extensive, cross-linked, insoluble extracellular matrix which remained firmly associated with the plastic surface of tissue culture ware upon lysis of cells. The insoluble matrix material was analyzed using enzymatic digestion procedures following incorporation of radiolabelled precursors into macromolecular material prior to lysis and preparation. The relative proportion of glycoconjugate (glycopeptides and proteoglycans) and collagenous material present in matrix material was approximately 45% and approximately 55%, respectively, and this was modulated by inclusion of epidermal growth factor into culture medium to approximately 60% and approximately 40%, respectively. Under similar culture conditions stromal cells synthesized twice as much hyaluronate as was produced by control cultures. By use of specific antibody preparations we identified at least four species of glycopeptide present in stromal matrices (namely, fibronectin, laminin, tenascin, and thrombospondin) as well as three types of collagen (types I, III, and IV). The rapid and reproducible procedure for the preparation of radiolabelled insoluble matrix material from normal human breast tissue allows for the study of cellular interaction involving extracellular matrix turnover and degradation.     

Mechanism of insulin incorporation into alpha 2-macroglobulin: implications for the study of peptide and growth factor binding.

In recent years, many studies have suggested a direct role for alpha 2-macroglobulin (alpha 2M), a plasma proteinase inhibitor, in growth factor regulation. When coincubated in the presence of either trypsin, pancreatic elastase, human neutrophil elastase, or plasmin, 125I-insulin rapidly formed a complex with alpha 2M which was greater than 80% covalent. The covalent binding was stable to reduction but abolished by competition with beta-aminopropionitrile. Neither native alpha 2M nor alpha 2M pretreated with proteinase or methylamine incorporated 125I-insulin. Experiments utilizing alpha 2M cross-linked with cis-dichlorodiammineplatinum(II) indicated that 125I-insulin must be present during alpha 2M conformational change to covalently bind. A maximum stoichiometry of 4 mol of insulin bound per mole of alpha 2M and the short half-life of the alpha 2M intermediate capable of covalent incorporation were consistent with thiol ester involvement. Protein sequence analysis of unlabeled insulin-alpha 2M complexes, together with results of beta-aminopropionitrile competition, confirmed that insulin incorporation occurs via the same gamma-glutamyl amide linkage responsible for covalent proteinase and methylamine binding to alpha 2M. Although intact insulin apparently incorporated through its sole lysine residue on the B chain, we found that isolated A chain also bound covalently to alpha 2M. Phenyl isothiocyanate derivatization of the N-terminus had no effect on A-chain binding, supporting the possibility of heretofore unreported gamma-glutamyl ester linkages to alpha 2M.     

The carboxyl terminus of the epsilon subunit of the chloroplast ATP synthase is exposed during illumination

The epsilon subunit of the chloroplast ATP synthase is an inhibitor of activity of the enzyme. Recombinant forms of the epsilon subunit from spinach chloroplasts lacking the last 10, 32, or 45 amino acids were immobilized onto activated Sepharose. A polyclonal antiserum raised against the epsilon subunit was passed over these immobilized protein columns, and the purified antibodies which were not bound recognized the portions of the epsilon subunit missing from the recombinant form present on the column. The full polyclonal antiserum can strip the epsilon subunit from the ATP synthase in illuminated thylakoid membranes [Richter, M. L., and McCarty, R. E. (1987) J. Biol. Chem. 262, 15037-15040]. Exposure of illuminated thylakoid membranes to antibodies recognizing the last 32 amino acids of the epsilon subunit collapses the proton gradient and hinders ATP synthesis with similar efficiency as the full polyclonal preparation. These results indicate that antibodies against the last 32 amino acids of the epsilon subunit are capable of stripping the subunit from the ATP synthase in illuminated membranes. Neither of these effects was seen when the membranes were exposed to the antibodies in the dark. This is direct evidence that the chloroplast ATP synthase undergoes a conformational shift during its activation by the electrochemical proton gradient which specifically alters the conformation of the carboxyl-terminal domain of the epsilon subunit from protected to solvent-exposed. The relation between this shift and activation of the enzyme by the electrochemical proton gradient is discussed.