Isolation, molecular cloning, and partial characterization of a novel carboxypeptidase B from human plasma

A novel plasminogen-binding protein has been isolated from human plasma utilizing plasminogen-Sepharose affinity chromatography. This protein copurified with alpha 2 antiplasmin when the plasminogen affinity column was eluted with high concentrations of epsilon-aminocaproic acid (greater than 20 mM). Analysis by sodium dodecyl sulfate suggests this protein has an apparent Mr of 60,000. The amino-terminal amino acid sequence showed no similarity to other protein sequences. Based on the amino-terminal amino acid sequence, oligonucleotide probes were designed for polymerase chain reaction primers, and an approximately 1,800 base pair cDNA was isolated that encodes this Mr 60,000 protein. The deduced amino acid sequence reveals a primary translation product of 423 amino acids that is very similar to carboxypeptidase A and B and consists of a 22-amino acid signal peptide, a 92-amino acid activation peptide, and a 309-amino acid catalytic domain. This protein shows 44 and 40% similarity to rat procarboxypeptidase B and human mast cell procarboxypeptidase A, respectively. The residues critical for catalysis and zinc and substrate binding of carboxypeptidase A and B are conserved in the Mr 60,000 plasminogen-binding protein. The presence of aspartic acid at position 257 of the catalytic domain suggests that this protein is a basic carboxypeptidase. When activated by trypsin, it hydrolyzes carboxypeptidase B substrates, hippuryl-Arg and hippuryl-Lys, but not carboxypeptidase A substrates, and it is inhibited by the specific carboxypeptidase B inhibitor (DL-5-guanidinoethyl)mercaptosuccinic acid. We propose that the Mr 60,000 plasminogen-binding protein isolated here is a novel human plasma carboxypeptidase B and that it be designated pCPB.     

Affinity purification of ubiquitin-protein ligase on immobilized protein substrates. Evidence for the existence of separate NH2-terminal binding sites on a single enzyme

Previous studies have indicated the existence of separate binding sites of ubiquitin-protein ligase, E3, specific for basic (Type I) or bulky hydrophobic (Type II) NH2-terminal amino acid residues of proteins. Another class (Type III) of protein substrates appeared to interact with E3 at regions other than the NH2 terminus (Reiss, Y., Kaim, D., and Hershko, A. (1988) J. Biol. Chem. 263, 2693-2698). In the present study we have used affinity chromatography on immobilized protein substrates to examine the question of whether the different binding sites belong to one E3 enzyme, or to different E3 species. Another objective was to develop a procedure for the extensive purification of E3. When a crude extract of reticulocytes is applied to Type I or Type II protein substrates linked to Sepharose, E3 becomes strongly bound to the affinity columns and is not eluted with salt at high concentration. However, the enzyme can be specifically eluted by a dipeptide that has an NH2-terminal residue similar to that of matrix-bound protein substrate. A 350-fold purification is obtained in this single step. Preparations of E3 purified on either Type I or Type II protein substrate affinity columns act on both types of protein substrates, indicating that the separate binding sites for basic and hydrophobic NH2-terminal residues belong to one enzyme. Another species of E3 that acts strongly on some Type III protein substrates does not bind to Type I or Type II protein substrate affinity columns.     

Crystal structure of the his-tagged saccharopine reductase from Saccharomyces cerevisiae at 1.7-Å resolution

The three-dimensional structure of the saccharopine reductase enzyme from the budding yeast Saccharomyces cerevisiae was determined to 1.7-A resolution in the apo form by using molecular replacement. The enzyme monomer consists of three domains: domain I is a variant of the Rossmann fold, domain II folds into a alpha/beta structure containing a mixed seven-stranded beta-sheet as the central core, and domain III has an all-helical fold. Comparative fold alignment with the enzyme from Magnaporthe grisea suggests that domain I binds to NADPH, and domain II binds to saccharopine and is involved in dimer formation. Domain III is involved in closing the active site of the enzyme once substrates are bound. Structural comparison of the saccharopine reductase enzymes from S. cerevisiae and M. grisea indicates that domain II has the highest number of conserved residues, suggesting that it plays an important role in substrate binding and in spatially orienting domains I and III.     

The crystal structure of the yeast Hsp40 Ydj1 complexed with its peptide substrate

The mechanisms by which Hsp40 functions as a molecular chaperone to recognize and bind non-native polypeptides is not understood. We have identified a peptide substrate for Ydj1, a member of the type I Hsp40 from yeast. The structure of the Ydj1 peptide binding fragment and its peptide substrate complex was determined to 2.7 A resolution. The complex structure reveals that Ydj1 peptide binding fragment forms an L-shaped molecule constituted by three domains. The domain I exhibits a similar protein folds as domain III while the domain II contains two Zinc finger motifs. The peptide substrate binds Ydj1 by forming an extra beta strand with domain I of Ydj1. The Leucine residue in the middle of the peptide substrate GWLYEIS inserts its side chain into a hydrophobic pocket formed on the molecular surface of Ydj1 domain I. The Zinc finger motifs located in the Ydj1 domain II are not in the vicinity of peptide substrate binding site.     

Integron integrases possess a unique additional domain necessary for activity.

Integrons are genetic elements capable of integrating genes by a site-specific recombination system catalyzed by an integrase. Integron integrases are members of the tyrosine recombinase family and possess the four invariant residues (RHRY) and conserved motifs (boxes I and II and patches I, II, and III). An alignment of integron integrases compared to other tyrosine recombinases shows an additional group of residues around the patch III motif. We have analyzed the DNA binding and recombination properties of class I integron integrase (IntI1) variants carrying mutations at residues that are well conserved among all tyrosine recombinases and at some residues from the additional motif that are conserved among the integron integrases. The well-conserved residues studied were H277 from the conserved tetrad RHRY (about 90% conserved), E121 found in the patch I motif (about 80% conserved in prokaryotic recombinases), K171 from the patch II motif (near 100% conserved), W229 and F233 from the patch III motif, and G302 of box II (about 80% conserved in prokaryotic recombinases). Additional IntI1 mutated residues were K219 and a deletion of the sequence ALER215. We observed that E121, K171, and G302 play a role in the recombination activity but can be mutated without disturbing binding to DNA. W229, F233, and the conserved histidine (H277) may be implicated in protein folding or DNA binding. Some of the extra residues of IntI1 seem to play a role in DNA binding (K219) while others are implicated in the recombination activity (ALER215 deletion).     

Crystal structure of transhydrogenase domain III at 1.2 A resolution

The nicotinamide nucleotide transhydrogenases (TH) of mitochondria and bacteria are membrane-intercalated proton pumps that transduce substrate binding energy and protonmotive force via protein conformational changes. In mitochondria, TH utilizes protonmotive force to promote direct hydride ion transfer from NADH to NADP, which are bound at the distinct extramembranous domains I and III, respectively. Domain II is the membrane-intercalated domain and contains the enzyme's proton channel. This paper describes the crystal structure of the NADP(H) binding domain III of bovine TH at 1.2 A resolution. The structure reveals that NADP is bound in a manner inverted from that previously observed for nucleotide binding folds. The non-classical binding mode exposes the NADP(H) nicotinamide ring for direct contact with NAD(H) in domain I, in accord with biochemical data. The surface of domain III surrounding the exposed nicotinamide is comprised of conserved residues presumed to form the interface with domain I during hydride ion transfer. Further, an adjacent region contains a number of acidic residues, forming a surface with negative electrostatic potential which may interact with extramembranous loops of domain II. Together, the distinctive surface features allow mechanistic considerations regarding the NADP(H)-promoted conformation changes that are involved in the interactions of domain III with domains I and II for hydride ion transfer and proton translocation.     

Crystal structure of avian carboxypeptidase D domain II: a prototype for the regulatory metallocarboxypeptidase subfamily.

The crystal structure of domain II of duck carboxypeptidase D, a prohormone/propeptide processing enzyme integrated in a three repeat tandem in the natural system, has been solved, constituting a prototype for members of the regulatory metallocarboxypeptidase subfamily. It displays a 300 residue N-terminal alpha/beta-hydrolase subdomain with overall topological similarity to and general coincidence of the key catalytic residues with the archetypal pancreatic carboxypeptidase A. However, numerous significant insertions/deletions in segments forming the funnel-like access to the active site explain differences in specificity towards larger protein substrates or inhibitors. This alpha/beta-hydrolase subdomain is followed by a C-terminal 80 residue beta-sandwich subdomain, unique for these regulatory metalloenzymes and topologically related to transthyretin and sugar-binding proteins. The structure described here establishes the fundamentals for a better understanding of the mechanism ruling events such as prohormone processing and will enable modelling of regulatory carboxypeptidases as well as a more rational design of inhibitors of carboxypeptidase D.     

The major piscine liver alcohol dehydrogenase has class-mixed properties in relation to mammalian alcohol dehydrogenases of classes I and III.

The major alcohol dehydrogenase of cod liver has been purified, enzymatically characterized, and structurally analyzed in order to establish original functions and relationships among the deviating classes of the enzyme in mammalian tissues. Interestingly, the cod enzyme exhibits mixed properties--many positional identities with a class III protein, but functionally a class I enzyme--blurring the distinction among the classes of alcohol dehydrogenase. The two domain interfaces, affected by movements upon coenzyme binding, both exhibit substitutions in a manner thus far unique to the cod enzyme. In contrast, coenzyme-binding residues are highly conserved. At the active site, inner and outer parts of the substrate pocket show different extents of amino acid replacement. In total, no less than 7-10 residues of 11 in the substrate binding pocket differ from those of all the mammalian classes, explaining the substrate specificities. However, the inner part of the substrate pocket is very similar to that of the class I enzymes, which is compatible with the observed characteristics of the cod enzyme: ethanol is an excellent substrate (Km = 1.2 mM) and 4-methylpyrazole is a strong inhibitor (Ki = 0.1 microM). These values are about as low as those typical for the ethanol-active class I mammalian enzyme and do not at all resemble those for class III, for which ethanol is hardly a substrate and pyrazole is hardly an inhibitor. Further out in the substrate pocket, several residues differ from the mammalian classes, affecting large substrates.(ABSTRACT TRUNCATED AT 250 WORDS)     

Crystal structure of an inactive duck delta II crystallin mutant with bound argininosuccinate

Delta-crystallin, the major soluble protein component of avian and reptilian eye lenses, is highly homologous to the urea cycle enzyme, argininosuccinate lyase (ASL). In duck lenses, there are two highly homologous delta crystallins, delta I and delta II, that are 94% identical in amino acid sequence. While delta II crystallin has been shown to exhibit ASL activity in vitro, delta I is enzymatically inactive. The X-ray structure of a His to Asn mutant of duck delta II crystallin (H162N) with bound argininosuccinate has been determined to 2.3 A resolution using the molecular replacement technique. The overall fold of the protein is similar to other members of the superfamily to which this protein belongs, with the active site located in a cleft formed by three different monomers in the tetramer. The active site of the H162N mutant structure reveals that the side chain of Glu 296 has a different orientation relative to the homologous residue in the H91N mutant structure [Abu-Abed et al. (1997) Biochemistry 36, 14012-14022]. This shift results in the loss of the hydrogen bond between His 162 and Glu 296 seen in the H91N and turkey delta I crystallin structures; this H-bond is believed to be crucial for the catalytic mechanism of ASL/delta II crystallin. Argininosuccinate was found to be bound to residues in each of the three monomers that form the active site. The fumarate moiety is oriented toward active site residues His 162 and Glu 296 and other residues that are part of two of the three highly conserved regions of amino acid sequence in the superfamily, while the arginine moiety of the substrate is oriented toward residues which belong to either domain 1 or domain 2. The analysis of the structure reveals that significant conformational changes occur on substrate binding. The comparison of this structure with the inactive turkey delta I crystallin reveals that the conformation of domain 1 is crucial for substrate affinity and that the delta I protein is almost certainly inactive because it can no longer bind the substrate.     

Refined crystal structure of the potato inhibitor complex of carboxypeptidase A at 2.5 Å resolution

The exopeptidase carboxypeptidase A forms a tight complex with a 39 residue inhibitor protein from potatoes. We have determined the crystal structure of this complex, and refined the atomic model to a crystallographic R-factor of 0.196 at 2.5 Å resolution. The structure of the inhibitor protein is organized around a core of disulfide bridges. No α-helices or β-sheets are present in this protein, although there is one turn of 3₁₀ helix. The four carboxy-terminal residues of the inhibitor protein bind in the active site groove of carboxypeptidase A, defining binding subsites S′₁, S₁, S₂ and S₃ on the enzyme. The carboxy-terminal glycine of the inhibitor is cleaved from the remainder of the inhibitor in the complex, and remains trapped in the back of the active site pocket. Interactions between the inhibitor and residues Tyr248 and Arg71 of carboxypeptidase A resemble possible features of binding stages for substrates both prior and subsequent to peptide bond hydrolysis. Not all of these interactions would be available to different types of ester substrates, however, which may be in part responsible for the observed kinetic differences in hydrolysis between peptides and various classes of esters. With the exception of residues involved in the binding of the inhibitor protein (such as Tyr248), the structure of carboxypeptidase A as determined in the inhibitor complex is quite similar to the structure of the unliganded enzyme (Lipscomb et al., 1968), which was solved from an unrelated crystal form.     

X-ray crystal structure of the hypothetical phosphotyrosine phosphatase MDP-1 of the haloacid dehalogenase superfamily

The haloacid dehalogenase (HAD) superfamily is comprised of structurally homologous enzymes that share several conserved sequence motifs (loops I-IV) in their active site. The majority of HAD members are phosphohydrolases and may be divided into three subclasses depending on domain organization. In classes I and II, a mobile "cap" domain reorients upon substrate binding, closing the active site to bulk solvent. Members of the third class lack this additional domain. Herein, we report the 1.9 A X-ray crystal structures of a member of the third subclass, magnesium-dependent phosphatase-1 (MDP-1) both in its unliganded form and with the product analogue, tungstate, bound to the active site. The secondary structure of MDP-1 is similar to that of the "core" domain of other type I and type II HAD members with the addition of a small, 28-amino acid insert that does not close down to exclude bulk solvent in the presence of ligand. In addition, the monomeric oligomeric state of MDP-1 does not allow the participation of a second subunit in the formation and solvent protection of the active site. The binding sites for the phosphate portion of the substrate and Mg(II) cofactor are also similar to those of other HAD members, with all previously observed contacts conserved. Unlike other subclass III HAD members, MDP-1 appears to be equally able to dephosphorylate phosphotyrosine and closed-ring phosphosugars. Modeling of possible substrates in the active site of MDP-1 reveals very few potential interactions with the substrate leaving group. The mapping of conserved residues in sequences of MDP-1 from different eukaryotic organisms reveals that they colocalize to a large region on the surface of the protein outside the active site. This observation combined with the modeling studies suggests that the target of MDP-1 is most likely a phosphotyrosine in an unknown protein rather than a small sugar-based substrate.     

Structure and Mechanism of Metallocarboxypeptidases

Metallocarboxpeptidases cleave C-terminal residues from peptide substrates and participate in a wide range of physiological processes, but they also contribute to human pathology. On the basis of structural information, we can distinguish between two groups of such metallopeptidases: cowrins and funnelins. Cowrins comprise protozoan, prokaryotic, and mammalian enzymes related to both neurolysin and angiotensin-converting enzyme and their catalytic domains contain 500–700 residues. They are ellipsoidal and traversed horizontally by a long, deep, narrow active-site cleft, in which the C-terminal residues are cut from oligopeptides and unstructured protein tails. The consensus cowrin structure contains a common core of 17 helices and a three-stranded β-sheet, which participates in substrate binding. This protease family is characterized by a set of spatially conserved amino acids involved in catalysis, HEXXH+EXXS/G+H+Y/R+Y. Funnelins comprise structural relatives of the archetypal bovine carboxypeptidase A1 and feature mammalian, insect and bacterial proteins with strict carboxypeptidase activity. Their ～ 300-residue catalytic domains evince a consensus central eight-stranded β-sheet flanked on either side by a total of eight helices. They also contain a characteristic set of conserved residues, HXXE+R+NR+H+Y+E, and their active-site clefts are rather shallow and lie at the bottom of a funnel-like cavity. Therefore, these enzymes act on a large variety of well-folded proteins. In both cowrins and funnelins, substrate hydrolysis follows a common general base/acid mechanism. A metal-bound solvent molecule ultimately performs the attack on the scissile peptide bond with the assistance of a strictly conserved glutamate residue.     

Structure and mechanism of metallocarboxypeptidases

Metallocarboxpeptidases cleave C-terminal residues from peptide substrates and participate in a wide range of physiological processes, but they also contribute to human pathology. On the basis of structural information, we can distinguish between two groups of such metallopeptidases: cowrins and funnelins. Cowrins comprise protozoan, prokaryotic, and mammalian enzymes related to both neurolysin and angiotensin-converting enzyme and their catalytic domains contain 500-700 residues. They are ellipsoidal and traversed horizontally by a long, deep, narrow active-site cleft, in which the C-terminal residues are cut from oligopeptides and unstructured protein tails. The consensus cowrin structure contains a common core of 17 helices and a three-stranded beta-sheet, which participates in substrate binding. This protease family is characterized by a set of spatially conserved amino acids involved in catalysis, HEXXH+EXXS/G+H+Y/R+Y. Funnelins comprise structural relatives of the archetypal bovine carboxypeptidase A1 and feature mammalian, insect and bacterial proteins with strict carboxypeptidase activity. Their approximately 300-residue catalytic domains evince a consensus central eight-stranded beta-sheet flanked on either side by a total of eight helices. They also contain a characteristic set of conserved residues, HXXE+R+NR+H+Y+E, and their active-site clefts are rather shallow and lie at the bottom of a funnel-like cavity. Therefore, these enzymes act on a large variety of well-folded proteins. In both cowrins and funnelins, substrate hydrolysis follows a common general base/acid mechanism. A metal-bound solvent molecule ultimately performs the attack on the scissile peptide bond with the assistance of a strictly conserved glutamate residue.     

Molecular characterization of the principal substrate binding site of the ubiquitous folding catalyst protein disulfide isomerase

Disulfide bond formation in the endoplasmic reticulum of eukaryotes is catalyzed by the ubiquitously expressed enzyme protein disulfide isomerase (PDI). The effectiveness of PDI as a catalyst of native disulfide bond formation in folding polypeptides depends on the ability to catalyze disulfide-dithiol exchange, to bind non-native proteins, and to trigger conformational changes in the bound substrate, allowing access to buried cysteine residues. It is known that the b' domain of PDI provides the principal peptide binding site of PDI and that this domain is critical for catalysis of isomerization but not oxidation reactions in protein substrates. Here we use homology modeling to define more precisely the boundaries of the b' domain and show the existence of an intradomain linker between the b' and a' domains. We have expressed the recombinant b' domain thus defined; the stability and conformational properties of the recombinant product confirm the validity of the domain boundaries. We have modeled the tertiary structure of the b' domain and identified the primary substrate binding site within it. Mutations within this site, expressed both in the isolated domain and in full-length PDI, greatly reduce the binding affinity for small peptide substrates, with the greatest effect being I272W, a mutation that appears to have no structural effect.     

Hsc70 contacts helix III of the J domain from polyomavirus T antigens: addressing a dilemma in the chaperone hypothesis of how they release E2F from pRb

Hsc70's expected binding site on helix II of the J domain of T antigens appears to be blocked in its structure bound to tumor suppressor pRb. We used NMR to map where mammalian Hsc70 binds the J domain of murine polyomavirus T antigens (PyJ). The ATPase domain of Hsc70 unexpectedly has its biggest effects on the NMR peak positions of the C-terminal end of helix III of PyJ. The Hsc70 ATPase domain protects the C-terminal end of helix III of PyJ from an uncharged paramagnetic probe of chelated Gd(III), clearly suggesting the interface. Effects on the conserved HPD loop and helix II of PyJ are smaller. The NMR results are supported by a novel assay of Hsc70's ATP hydrolysis showing that mutations of surface residues in PyJ helix III impair PyJ-dependent stimulation of Hsc70 activity. Evolutionary trace analysis of J domains suggests that helix III usually may join helix II in contributing specificities for cognate hsp70s. Our novel evidence implicating helix III differs from evidence that Escherichia coli DnaK primarily affects helix II and the HPD loop of DnaJ. We find the pRb-binding fragment of E2F1 to be intrinsically unfolded and a good substrate for Hsc70 in vitro. This suggests that E2F1 could be a substrate for Hsc70 recruited by T antigen to an Rb family member. Importantly, our results strengthen the chaperone hypothesis for E2F release from an Rb family member by Hsc70 recruited by large T antigen. That is, it now appears that Hsc70 can freely access helix III and the HPD motif of large T antigen bound to an Rb family member.     

Rational design and improvement of the dimerization‐disrupting peptide selectivity between ROCK‐I and ROCK‐II kinase isoforms in cerebrovascular diseases

Human rho‐associated coiled‐coil forming kinases (ROCKs) ROCK‐I and ROCK‐II have been documented as attractive therapeutic targets for cerebrovascular diseases. Although ROCK‐I and ROCK‐II share a high degree of structural conservation and are both present in classic rho/ROCK signaling pathway, their downstream substrates and pathological functions may be quite different. Selective targeting of the two kinase isoforms with traditional small‐molecule inhibitors is a great challenge due to their surprisingly high homology in kinase domain (~90%) and the full identity in kinase active site (100%). Here, instead of developing small‐molecule drugs to selectively target the adenosine triphosphate (ATP) site of two isoforms, we attempt to design peptide agents to selectively disrupt the homo‐dimerization event of ROCK kinases through their dimerization domains which have a relatively low conservation (~60%). Three helical peptides H1, H2, and H3 are split from the kinase dimerization domain, from which the isolated H2 peptide is found to have the best capability to rebind at the dimerization interface. A simulated annealing (SA) iteration method is used to improve the H2 peptide selectivity between ROCK‐I and ROCK‐II. The method accepts moderate degradation in peptide affinity in order to maximize the affinity difference between peptide binding to the two isoforms. Consequently, hundreds of parallel SA runs yielded six promising peptide candidates with ROCK‐I over ROCK‐II (I over II [IoII]) calculated selectivity and four promising peptide candidates with ROCK‐II over ROCK‐I (II over I [IIoI]) calculated selectivity. Subsequent anisotropy assays confirm that the selectivity values range between 13.2‐fold and 83.9‐fold for IoII peptides, and between 5.8‐fold and 21.2‐fold for IIoI peptides, which are considerably increased relative to wild‐type H2 peptide (2.6‐fold for IoII and 2.0‐fold for IIoI). The molecular origin of the designed peptide selectivity is also analyzed at structural level; it is revealed that the peptide residues can be classified into conserved, non‐conserved, and others, in which the non‐conserved residues play a crucial role in defining peptide selectivity, while conserved residues confer stability to kinase‐peptide binding.     

Structural basis of the lisinopril-binding specificity in N- and C-domains of human somatic ACE

Angiotensin I-converting enzyme (ACE) is a dipeptidyl carboxypeptidase which converts angiotensin I into the vasopressor peptide angiotensin II and also inactivates the hypotensive peptide bradykinin, playing an important role in blood pressure regulation. The present work describes the molecular modeling of the N-terminal human somatic ACE in complex with the inhibitor lisinopril, identifying the residues involved in the inhibitor-binding pocket. The obtained results identify differences in the lisinopril lysine moiety-binding residues for N- and C-terminals of sACE domains and an important carboxy-terminal proline hydrophobic accommodations mediated by the aromatic ring of Tyr532 and Tyr1128 residues, respectively. The present model will be useful for the development of a new inhibitor family based on the natural BPP peptides and derivatives, or even to improve the binding capacities and the domain specificity of the already known inhibitors.     

Functions of the stem region of the Semliki Forest virus fusion protein during virus fusion and assembly

Membrane fusion of the alphaviruses is mediated by the E1 protein, a class II virus membrane fusion protein. During fusion, E1 dissociates from its heterodimer interaction with the E2 protein and forms a target membrane-inserted E1 homotrimer. The structure of the homotrimer is that of a trimeric hairpin in which E1 domain III and the stem region fold back toward the target membrane-inserted fusion peptide loop. The E1 stem region has a strictly conserved length and several highly conserved residues, suggesting the possibility of specific stem interactions along the trimer core and an important role in driving membrane fusion. Mutagenesis studies of the alphavirus Semliki Forest virus (SFV) here demonstrated that there was a strong requirement for the E1 stem in virus assembly and budding, probably reflecting its importance in lateral interactions of the envelope proteins. Surprisingly, however, neither the conserved length nor any specific residues of the stem were required for membrane fusion. Although the highest fusion activity was observed with wild-type E1, efficient fusion was mediated by stem mutants containing a variety of substitutions or deletions. A minimal stem length was required but could be conferred by a series of alanine residues. The lack of a specific stem sequence requirement during SFV fusion suggests that the interaction of domain III with the trimer core can provide sufficient driving force to mediate membrane merger.