Characterization of the deoxyribonuclease and ADP-ribosyltransferase activities of CRM45, a truncated homologue of diphtheria toxin.

CRM45 is a mutant form of diphtheria toxin (DTx) that lacks a 17-kDa carboxyl-terminal segment of the receptor-binding B subunit (DTB). The missing segment is a discrete structural domain of DTB that normally rests against the NAD binding pocket of the enzymically-active A subunit (DTA). Proteolytic cleavage and disulfide bridge reduction in the DTA-DTB linker region of DTx are required for optimal ADP-ribosylation of elongation factor 2 (EF-2). Here, we show that cleaved and uncleaved preparations of X-ray crystal grade CRM45 both exhibit an ADP-ribosyltransferase activity similar to that of cleaved DTx. Crystal-grade preparations of CRM45 also display a potent deoxyribonuclease activity. However, as observed with DTx, cleavage and reduction of CRM45 are not required for expression of this nuclease activity. After SDS-PAGE in a gel that contains DNA embedded in the matrix, renaturable Ca++/Mg(++)-dependent nuclease-active bands co-migrate with intact CRM45 (45 kDa) as well as with the DTA subunit (24 kDa) of CRM45. Because the 45-kDa nuclease-active band is unique to the CRM45 form of DTx, it offers direct proof that this activity is intrinsic to the DTA domain of DTx and its homologues.

Structural changes of tumor necrosis factor alpha associated with membrane insertion and channel formation.

Low pH enhances tumor necrosis factor alpha (TNF)-induced cytolysis of cancer cells and TNF-membrane interactions that include binding, insertion, and ion-channel formation. We have also found that TNF increases Na+ influx in cells. Here, we examined the structural features of the TNF-membrane interaction pathway that lead to channel formation. Fluorometric studies link TNF's acid-enhanced membrane interactions to rapid but reversible acquisition of hydrophobic surface properties. Intramembranous photolabeling shows that (i) protonation of TNF promotes membrane insertion, (ii) the physical state of the target bilayer affects the kinetics and efficiency of TNF insertion, and (iii) binding and insertion of TNF are two distinct events. Acidification relaxes the trimeric structure of soluble TNF so that the cryptic carboxyl termini, centrally located at the base of the trimer cone, become susceptible to carboxypeptidase Y. After membrane insertion, TNF exhibits a trimeric configuration in which the carboxyl termini are no longer exposed; however, the proximal salt-bridged Lys-11 residues as well as regional surface amino acids (Glu-23, Arg-32, and Arg-44) are notably more accessible to proteases. The sequenced cleavage products bear the membrane-restricted photoreactive probe, proof that surface-cleaved TNF has an intramembranous disposition. In summary, the trimer's structural plasticity is a major determinant of its channel-forming ability. Channel formation occurs when cracked or partially splayed trimers bind and penetrate the bilayer. Reannealing leads to a slightly relaxed trimeric structure. The directionality of bilayer penetration conforms with x-ray data showing that receptor binding to the monomer interfaces of TNF poises the tip of the trimeric cone directly above the target cell membrane.

Conformation and ion channel activity of lymphotoxin at neutral and low pH.

Lymphotoxin (LT or TNF-beta), a T cell-derived lymphokine with partial homology to TNF-alpha, was found to bind to dimyristoylphosphatidylcholine vesicles in a pH-dependent manner: binding increased with decreasing pH. Binding was not limited to surface association with phospholipid head groups because studies with a photoreactive membrane-restricted probe revealed protein penetration of the hydrocarbon core of the bilayer. Intramembranous photolabeling of the trimeric form of LT demonstrated maintenance of quaternary structure upon bilayer insertion. The efficiency of insertion was greatly enhanced with gel-phase bilayers compared with fluid phase bilayers even though the binding efficiency was much lower. Hence, binding and insertion represent two distinct physical processes. Intrinsic fluorescence and dye binding assays showed that the acid-facilitated membrane interactions stemmed from acid-induced changes in protein conformation. The acquisition of hydrophobic characteristics through these conformational changes supplies a physical explanation for LT conversion from a water-soluble form to a membrane-embedded structure. Moreover, the use of vesicle-embedded LT to prepare planar bilayers vs the addition of soluble LT subsequent to bilayer formation demonstrated that LT exhibits channel activity and that low pH-induced membrane insertion precedes and is distinct from expression of voltage-dependent ion gating. LT's ability to associate intimately with lipid vesicles and form ion channels mirrors the behavior of TNF-alpha. Thus, although LT and TNF-alpha are secreted by different cell types, the conservation of membrane binding, insertion, and channel-forming activities suggests a functional role in response induction.

Localization of diphtheria toxin nuclease activity to fragment A.

We describe a series of experiments that aimed to establish whether nuclease activity is actually associated with diphtheria toxin (DTx) and its A subunit (DTA), as we originally reported (M. P. Chang, R. L. Baldwin, C. Bruce, and B. J. Wisnieski, Science 246:1165-1168, 1989). Here we show that (i) trypsinization of DTx does indeed produce nucleolytically active DTA, (ii) reduction of electroeluted, unreduced, cleaved DTx (58 kDa) yields nuclease-active DTA (24 kDa), and (iii) fractionation of DTx and DTA by anion-exchange chromatography leads to coelution of nuclease activity with both forms of the toxin, even though each form elutes at a distinct salt concentration. In addition, we show that Escherichia coli-derived DTA also expresses nuclease activity. These studies confirm our initial assertion that the nuclease activity observed in DTx preparations is intrinsic to the DTA portion of DTx.

Formation of ion-permeable channels by tumor necrosis factor-alpha.

Tumor necrosis factor-alpha (TNF, cachectin), a protein secreted by activated macrophages, participates in inflammatory responses and in infectious and neoplastic disease states. The mechanisms by which TNF exerts cytotoxic, hormonal, and other specific effects are obscure. Structural studies of the TNF trimer have revealed a central pore-like region. Although several amino acid side chains appear to preclude an open channel, the ability of TNF to insert into lipid vesicles raised the possibility that opening might occur in a bilayer milieu. Acidification of TNF promoted conformational changes concordant with increased surface hydrophobicity and membrane insertion. Furthermore, TNF formed pH-dependent, voltage-dependent, ion-permeable channels in planar lipid bilayer membranes and increased the sodium permeability of human U937 histiocytic lymphoma cells. Thus, some of the physiological effects of TNF may be elicited through its intrinsic ion channel-forming activity.

Comparison of the intoxication pathways of tumor necrosis factor and diphtheria toxin.

The mechanism by which tumor necrosis factor alpha (TNF) initiates tumor cell destruction is unknown. Having established that a brief drop in extracellular pH enhances the killing activity of TNF, our next objective was to explore whether TNF-induced cell death is dependent on endosomal acidification. Diphtheria toxin (DTx), a well-characterized acid-dependent cytotoxin, served as an indicator of the effectiveness of each treatment condition. Studies with lysosomotropic agents demonstrated that the cytotoxic pathway of TNF can operate independently of low pH exposure in contrast to the lethal pathway of DTx. When NH4Cl-treated cells were exposed to TNF at low pH, the level of killing increased two- to threefold over that attained with cells maintained at neutral pH (either with or without NH4Cl). Furthermore, inhibition of metabolic processes by sodium azide in combination with 2-deoxyglucose severely reduced DTx killing but stimulated TNF killing. Despite these differences, TNF and DTx provoked extensive internucleosomal DNA cleavage in prelytic target cells. Inhibitor of nuclear poly(ADP-ribose) transferase also evoked similar levels of cellular resistance to both cytotoxins. Models for DTx- and TNF-induced cytolysis are discussed in view of these new discoveries.

Diphtheria toxin and its ADP-ribosyltransferase-defective homologue CRM197 possess deoxyribonuclease activity.

The cytotoxic mechanism of diphtheria toxin (DTx) is associated with its ability to inhibit protein synthesis by ADP-ribosylation of elongation factor 2. Although DTx intoxication leads to internucleosomal DNA cleavage and cell lysis, these events do not occur when protein synthesis is inhibited by alternative treatments (e.g., cycloheximide). Here we show that endonucleolytic degradation of DNA is an intrinsic activity of DTx and also of the crossreactive mutant protein CRM197. Assays using DNA-impregnated gels as well as linear and supercoiled DNA in solution revealed not only that CRM197 has nuclease activity but also that its specific activity is actually significantly greater than that of the wild-type molecule. Since CRM197 contains a single amino acid substitution that renders it incapable of ADP-ribosylation, we propose that the active sites for ADP-ribosyltransferase and nuclease activities are distinct.

Characterization of the deoxyribonuclease activity of diphtheria toxin.

Having discovered that the A domain of diphtheria toxin exhibits intrinsic nuclease activity (Chang, M. P., Baldwin, R. L., Bruce, B., and Wisnieski, B. J. (1989) Science 246, 1165-1168), we proceeded to examine the requirements for optimal enzymic expression. In vitro assays with linear double-stranded DNA demonstrated that optimal activity occurs at pH 7.5 and 37 degrees C. A characterization of the stringent cation-dependence of the reaction revealed increasing activity with increasing Mn2+ up to 30 mM. In contrast, activity levels with Ca2+ or Zn2+ alone peaked at 100 microM and with Mg2+ alone at 1 mM. The Zn2(+)- and Mg2(+)-stimulated activities appear to be dependent on trace amounts of Ca2+. Indeed, inclusion of 2 mM Ca2+ plus 3 mM Mg2+ in the reaction buffer promoted a high level of DNA cleavage even though very little cleavage was seen with either cation alone at 2-3 mM. Addition of 20-200 mM NaCl or KCl caused progressive inhibition. Detection of diphtheria toxin nuclease activity under physiologically relevant conditions suggests that it may be operative in vivo and supports our contention that diphtheria toxin-induced cytolysis is not a simple consequence of protein synthesis inhibition, but rather the final step in a cytolytic pathway linked to chromosomal integrity.

Second cytotoxic pathway of diphtheria toxin suggested by nuclease activity.

Diphtheria toxin (DTx) provokes extensive internucleosomal degradation of DNA before cell lysis. The possibility that DNA cleavage stems from direct chromosomal attack by intracellular toxin molecules was tested by in vitro assays for a DTx-associated nuclease activity. DTx incubated with DNA in solution or in a DNA-gel assay showed Ca2+- and Mg2+-stimulated nuclease activity. This activity proved susceptible to inhibition by specific antitoxin and migrated with fragment A of the toxin. Assays in which supercoiled double-stranded DNA was used revealed rapid endonucleolytic attack. Discovery of a DTx-associated nuclease activity lends support to the model that DTx-induced cell lysis is not a simple consequence of protein synthesis inhibition.

Internucleosomal DNA cleavage precedes diphtheria toxin-induced cytolysis. Evidence that cell lysis is not a simple consequence of translation inhibition.

Diphtheria toxin (DTx) is an extremely potent inhibitor of protein synthesis. Cell death has been generally accepted as a straightforward effect of translation inhibition. Using human U937 cells, we found that DTx intoxication leads to cytolysis; indeed, release of 51Cr- and 75Se-labeled proteins could be detected within 7 h. However, little or no cell lysis was observed over a 20-50-h period when human U937 cells were exposed to cycloheximide, amino acid-deficient medium, or metabolic poisons even though protein synthesis was rapidly inhibited to levels observed with DTx. Likewise, investigations with human K562 cells revealed full resistance to the cytolytic action of DTx over a 50-h period despite a severe reduction in translation activity. These observations establish that inhibition of protein synthesis per se is not sufficient to provoke cell lysis. A characterization of DTx-induced cytolysis revealed a long lag period (6-7 h) which could be shortened considerably by a short exposure to low pH. NH4Cl and metabolic poisons blocked the cytolytic action of DTx, indicating that endocytic uptake of toxin is required for lytic activity. Surprisingly, DTx also induced extensive internucleosomal degradation of cellular DNA, a characteristic feature of apoptosis or programmed cell death. DNA-fragmentation preceded cell lysis and did not occur in DTx-treated K562 cells or in U937 cells that were treated with the other protein synthesis inhibitors. From these observations, we conclude that DTx-mediated cytolysis is not a simple consequence of translation inhibition and that internucleosomal DNA fragmentation is a newly identified and relatively early step in the cytolytic pathway of DTx.

The acid-triggered entry pathway of Pseudomonas exotoxin A.

In this study we examined the pH requirements and reversibility of early events in the Pseudomonas toxin entry pathway, namely, membrane binding, insertion, and translocation. At pH 7.4, toxin binding to vesicles and insertion into the bilayer are very inefficient. Decreasing the pH greatly increases the efficiencies of these processes. Acid-treated toxin exhibits pH 7.4 binding and insertion levels. This indicates that hydrophobic regions that become exposed upon toxin acidficiation become buried again when the pH is raised to 7.4. In contrast, the change in toxin conformation that occurs upon membrane binding is irreversible. Returning samples to pH 7.4, incubation with excess toxin, or dilution with buffer up to 1000-fold leads to very little loss of bound toxin. Bound toxin exhibits an extremely high susceptibility to trypsin compared to free toxin (at both pH 4 and pH 7.4). At pH 4, membrane-associated toxin slowly proceeds to a trypsin-protected state; neutralization halts this process. At low pH, toxin was found to bind and insert into DMPC vesicles very efficiently at temperatures both above and below 23 degrees C, the lipid melting point. With fluid targets, the proportion of bound toxin that was photolabeled from within the bilayer peaked rapidly and then decreased with time. With frozen targets, the efficiency of photolabeling peaked but then remained fairly constant.(ABSTRACT TRUNCATED AT 250 WORDS)

Capacity of tumor necrosis factor to bind and penetrate membranes is pH-dependent.

Studies with human U937 cells as targets established that a 15-min exposure to rTNF at pH 5.3 caused a significant increase in TNF-mediated cytolysis when compared to cells exposed to TNF at pH 7.4. A detailed examination of TNF-membrane interactions revealed that although TNF bound avidly to model membrane targets, no damage was generated under any condition tested. Binding of TNF, monitored with 125I-labeled as well as unlabeled protein, was enhanced at low pH. In the pH range tested (i.e., 4 to 8), target membrane permeability actually decreased in the presence of TNF. This membrane stabilization may be a consequence of TNF insertion into the target bilayer, a process we detected through use of an intramembranous photolabeling assay; interestingly, the efficiency of TNF insertion into membranes increased dramatically with decreasing pH. We conclude that native TNF does not cause pore formation directly and that its ability to induce cell lysis, as monitored by 51Cr release, is a consequence of some as yet obscure signaling event or intracellular activity. Parallel studies were carried out with diphtheria toxin, a protein with a more thoroughly characterized pH-dependent intoxification pathway. This toxin displayed acid-enhanced activities with both biologic and artificial targets.

An endosomal model for acid triggering of diphtheria toxin translocation.

An endosomal model system was developed for studying the effects of pH on vesicle-entrapped diphtheria toxin. The "endosomes" were prepared from dioleoylphosphatidylcholine (1 mg), diphtheria toxin (0.25 mg), and lysozyme (2.25 mg) in water at pH 8.4. The method used for preparing large unilamellar vesicles was adapted from the procedure of Shew and Deamer (Shew, R. L., and Deamer, D. W. (1985) Biochim. Biophys. Acta 816, 1-8). Efficiencies of trapping (typically 45-75%) and separation from untrapped proteins (typically 95-100%) were assessed by fluorescamine assays conducted before and after column chromatography and in the presence and absence of Tergitol Nonidet P-40. Intramembranous photolabeling revealed that diphtheria toxin inserts into the vesicle bilayer when the pH is dropped to 4; surface labeling revealed that the same treatment leads to exposure of diphtheria toxin at the trans surface of the vesicles. Release of toxin to the solution was not detected under the experimental conditions employed (i.e. with nicked or unnicked toxin, +/- exogenous trypsin, pH 4 or 8.4). Preliminary results indicate that this model system will be a valuable tool for elucidating the pathway by which the ADP ribosyltransferase domain of diphtheria toxin gains access to the cytoplasmic compartment of cells after endosomal uptake.

Effect of low pH on the conformation of Pseudomonas exotoxin A.

Previously we examined factors involved in the entry mechanism of Pseudomonas exotoxin A (PTx) at the level of lipid-protein interactions (Farahbakhsh, Z. T., Baldwin, R. L., and Wisnieski, B. J. (1986) J. Biol. Chem. 261, 11404-11408). Exposure to a low pH environment appears to be an obligatory trigger of the entry pathway. In this report we describe the effect of pH upon the conformation of PTx. We have found that the intrinsic fluorescence of PTx is strongly dependent on pH, decreasing between pH 7.4 and 4.0 with a red shift in the emission lambda max. The changes are reversible and associated with the acquisition of a binding site for the fluorescent dye 1-anilino-8-naphthalenesulfonic acid (ANS). The fluorescence intensity of ANS in the presence of PTx increases with decreasing pH and is accompanied by a blue shift in emission spectra, indicative of exposure of hydrophobic surfaces. These changes are also reversible. Both the intrinsic fluorescence and ANS binding profiles show a dramatic dependence on pH, with the transitions centered on pH 5.0 and 4.5, respectively. Circular dichroism studies reveal a 9% decrease in alpha-helicity between pH 7.7 and 4. The susceptibility of toxin to trypsin cleavage is also a function of pH, increasing with decreasing pH. The pH 7.4 cleavage profile is regained when the acid-treated samples are brought back to pH 7.4. The conformational changes observed in these pH shift experiments are likely to be physiologically significant because the conditions closely resemble those that the toxin would encounter if entry into the cytoplasm of a cell involves escape from an endosomal compartment.

Pseudomonas exotoxin A. Membrane binding, insertion, and traversal.

Using vesicle targets composed of phosphatidylcholine and cholesterol (1:1 molar ratio), we found that Pseudomonas aeruginosa exotoxin A (PTx) binding and insertion are not only dependent on pH (Zalman, L.S., and Wisnieski, B.J. (1985) Infect. Immun. 50, 630-635) but also on ionic strength, reaching a maximum in pH 4 buffer that contains 150-200 mM NaCl. Insertion was monitored by photolabeling with an intramembranous probe. Higher levels of binding and insertion were attained with vesicles that contained 2.5 mol% dicetylphosphate than with neutral vesicles. Positively charged vesicles (2.7 mol% stearylamine) were the least effective targets. At pH 7.4, all binding levels were depressed. While PTx binding increased with increasing temperature, the relative proportion of the vesicle-associated toxin that was photolabeled decreased. The most likely explanation for the decrease is that the bilayer translocation rates increased with increasing temperature, and hence fewer PTx molecules were accessible at the time of photolabeling. At 37 degrees C, binding and insertion both plateaued within 10 min of lowering the pH to 4. After 10 min, the amount of bound toxin decreased slightly with time but there was a dramatic decrease in photolabeling, indicating that inserted PTx had begun to cross the bilayer. This was verified by the finding that when PTx was incubated with vesicles that contained trypsin, cleavage occurred only in those samples in which the pH was shifted down to pH 4. Entry is triggered by an acid-induced conformational change that promotes productive binding and insertion. After insertion, the kinetics of membrane traversal appear to be regulated by the physical properties of the bilayer.

Characterization of the insertion of Pseudomonas exotoxin A into membranes.

Pseudomonas aeruginosa exotoxin A (PTx) is an extremely potent inhibitor of protein synthesis, similar to diphtheria toxin in its mode of action. It is synthesized in precursor form and secreted as an Mr 66,583 protein lacking a 25-amino acid leader sequence. While the primary sequence and the nature of the enzyme activity that leads to inactivation of elongation factor 2 are known, the mechanism of PTx internalization remains obscure. To elucidate the entry pathway, we examined PTx-membrane interactions using vesicle targets of defined lipid composition. Insertion was monitored with an intramembranous photoreactive probe; pore formation was determined from liposomal swelling rates. Our results show that the efficiency of PTx binding to vesicles increases dramatically with decreasing pH. In general, the insertion efficiency correlated with the binding efficiency. At pH 4, we noted a slight decrease in binding below the melting point (23 degrees C) of the target vesicles. Not only was PTx able to insert into frozen bilayers, but the efficiency of penetration at 0 degrees C was actually somewhat higher than expected based on binding efficiency. Liposome swelling assays analyzed by the Renkin equations indicated that PTx-liposomes made at pH 4 were permeable to solutes up to 2.8 nm in diameter. Pores of a similar size were found when the liposomes were made at pH 7, but the efficiency of pore formation at this pH was very low. Chymotrypsin fragmentation profiles of PTx depended on incubation conditions, e.g., pH, presence of NAD, reducing agents, and membranes. Liposomes containing PTx cleaved at pH 4 displayed up to 40-fold more pore activity than liposomes containing uncleaved PTx or PTx cleaved at pH 7. Pore activity at pH 7 was negligible. Addition of reducing agents caused a 50 to 60% increase in pore activity. Cleaved toxin was active in target membrane insertion even at 0 degrees C, and all of the major fragments were photolabeled.

Mechanism of insertion of diphtheria toxin: peptide entry and pore size determinations.

Diphtheria toxin ( DTx ) is an extremely potent inhibitor of protein synthesis. It is secreted as a linear polypeptide, which is cleaved to produce disulfide-linked A and B fragments. Fragment A, the inhibitor of protein synthesis, requires fragment B, the recognition subunit, for entry into intact cells. Fragment B has been proposed to form a transmembrane channel through which A gains access to the cytosol. If it were demonstrated that the B subunit had an exclusive association with membrane lipid acyl chains, this might indicate that A is secluded in a proteinaceous B channel. However, our results from intramembranous photolabeling studies show that both subunits of DTx enter the hydrocarbon domain of the bilayer. Toxin cleavage is not required for penetration. Decreasing pH leads to increased binding and hence indirectly to increased penetration. Parallel permeability studies indicate that cleaved DTx does indeed form pores (24 A in diameter) and they are larger than those previously reported (5 A) with native toxin. The data suggest that these are dimeric structures. Cleaved DTx is much more effective than intact DTx at pore formation. Thus, we conclude that, while pore formation is a feature of toxin-membrane interaction, the pore structure does not protect A from contact with lipid side chains and may in fact consist of both the A and B domains in a dimeric configuration, (AB)2.

Characterization of a glycoprotein fusogen isolated from Sendai virus.

After isolation from Sendai virus, the glycoproteins HN and F retained their ability to induce hemagglutination and both heterologous and homologous cell-cell fusion. Both methods for demonstrating cell fusion indicated that the isolated HN and F glycoproteins compared favorably with whole Sendai virus as a fusogen. Conditions affecting the degree of fusion were examined and optimized. Whole virus and isolated glycoprotein preparations were characterized by electron microscopy and by SDS-polyacrylamide gel electrophoresis. Lipid analysis of the glycoprotein preparations by thin layer chromatography and gas chromatography/mass spectrometry indicated that they were partially lipid-depleted during the isolation protocol and the ratio of cholesterol to phospholipid was higher than in the whole virus. A complete fatty acid analysis was performed on lipid extracts from whole virus and from glycoprotein preparations. Detergent was removed from the glycoproteins by dialysis and by incubation with Amberlite XAD-2 resin. The detergent content of the glycoprotein preparations was monitored by gas chromatography and with [3H]Triton X-100. Both methods showed that virtually all (greater than or equal to 99.8%) of the originally added detergent was removed. Electron microscopy of the negatively-stained HN and F preparations showed primarily spherical particles 120 +/- 20 A in diameter (range 80-250 A). Since no organization reminiscent of envelopes could be demonstrated, we conclude that the fusogenic activity of Sendai virus resides in the glycoproteins per se rather than in bilayer integrated lipid-protein complexes.