The invention relates generally to the field of modulating the activity of the apoptotic protease-activating factor 1 (Apaf-1), which is an essential component of the apoptotic mechanism in mammalian cells.
Programmed cell death, or apoptosis, is essential to the development and homeostasis of metazoans (Danial et al., 2004, Cell 116:205-219; Horvitz, 2003, Chembiochem. 4:697-711). Abnormal inhibition of apoptosis is a hallmark of cancer and autoimmune diseases, and excessive activation of cell death is implicated in neuro-degenerative and other disorders (Hanahan et al., 2000, Cell 100:57-70; Yuan et al., 2000, Nature 407:802-809; Green et al., 2002, Cancer Cell 1: 19-30). Caspases, named after cysteine proteases that cleave after an aspartate residue in their substrates, are central components of the apoptotic response. The conserved mechanism of apoptosis across species involves a cascade of sequential activation of initiator and effector caspases (Riedl et al., 2004, Nature Rev. Mol. Cell. Biol. 5:897-907).
The caspase activation cascade downstream of mitochondria is controlled by Apaf-1, which is responsible for the activation of the initiator caspase-9 and subsequent activation of effector caspases-3 and -7 (Zou et al., 1997, Cell 90:405-413; Li et al., 1997, Cell 91:479-489). Apaf-1 has an essential role in the regulation of programmed cell death in mammalian development and in oncogene- and p53-dependent apoptosis (Cecconi et al., 1998, Cell 94:727-737; Yoshida et al., 1998, Cell 94:739-750; Soengas et al., 1999, Science 284:156-159; Fearnhead et al., 1998, Proc. Natl. Acad. Sci. 95:13664-1366). The importance of Apaf-1-mediated apoptosis is manifested by the observation that Apaf-1 is frequently inactivated in cancers such as malignant melanoma (Soengas et al., 2001, Nature 409:207-211).
In response to a wide range of intrinsic cell death stimuli, Apaf-1 interacts with cytosolic cytochrome c that is released from mitochondria and, in the presence of dATP or ATP, forms an oligomeric complex dubbed the apoptosome (Li et al., 1997, Cell 91:479-489; Zou et al., 1999, J. Biol. Chem. 274:11549-11556; Saleh et al., 1999, J. Biol. Chem. 274:17941-17945; Hu et al., 1999, EMBO J. 18:3586-3595). The mechanistic role of ATP/dATP-binding to Apaf-1 is unknown, although it is essential to the formation of the apoptosome (Li et al., 1997, Cell 91:479-489; Zou et al., 1999, J. Biol. Chem. 274:11549-11556; Saleh et al., 1999, J. Biol. Chem. 274:17941-17945; Hu et al., 1999, EMBO J. 18:3586-3595; Jiang et al., 2000, J. Biol. Chem. 275:31199-31203). The apoptosome, in turn, recruits and activates procaspase-9 and forms a holoenzyme with the processed caspase-9 (Rodriguez et al., 1999, Genes Dev. 13:3179-3184). In Drosophila, the Apaf-1 orthologue Dark (also known as Dapaf-1 and Hac-1; Rodriguez et al., 1999, Nat. Cell Biol. 1:272-279; Kanuka et al., 1999, Mol. Cell 4:757-769; Zhou et al., 1999, Mol. Cell 4:745-755) is critically important for activation of the initiator caspase Dronc (a caspase-9 orthologue). In C. elegans, CED-4 exhibits significant sequence homology to Apaf-1 and is indispensable for the activation of CED-3 (Zou et al., 1997, Cell 90:405-413; Yuan et al., 1992, Development 116:309-320), the only apoptotic caspase in worms.
Apaf-1 is a 140-kilodalton, multi-domain protein, consisting of an N-terminal caspase recruitment domain (CARD), a central nucleotide-binding domain, and 12-13 repeats of the WD40 domain at the C-terminal half. The WD40 repeats are thought to be responsible for binding to cytochrome c and are believed to have a regulatory role in Apaf-1 function, because the removal of the WD40 repeats resulted in a constitutively active Apaf-1 protein that activated caspase-9 in a cytochrome c-independent manner (Hu et al., 1998, J. Biol. Chem. 273:33489-33494; Srinivasula et al., 1998, Mol. Cell 1:949-957). However, the underlying molecular mechanisms of how Apaf-1 interacts with ATP/dATP during formation of the apoptosome and activation of caspase-9 were not previously understood.
Apaf-1 is a representative member of the nucleotide-binding oligomerization (NOD) family of proteins that, in addition to Dark and CED-4, also include Ipaf, Nod1, Nod2, and a large family of disease-resistant proteins in plants (Inohara et al., 2001, Oncogene 20:6473-6481; Poyet et al., 2001, J. Biol. Chem. 276:28309-28313; Inohara et al., 1999, J. Biol. Chem. 274:14560-14567; Ogura et al., 2001, J. Biol. Chem. 276:4812-4818; Dangl et al., 2001, Nature 411:826-833). The hallmark of these proteins is the central NOD domain flanked by an N-terminal homotypic interaction motif and a C-terminal ligand-sensing domain. The shared domain structure suggests conserved mechanisms of action. However, the lack of structural information on any member of the NOD family proteins severely restricts our understanding on the mechanisms of the NOD family of proteins.
Efforts to study Apaf-1 protein have been hampered by inability of others to generate significant quantities of the protein in a form sufficiently stable, soluble, and pure to allow such study. For instance, there is no published protocol that allows bacterial expression and purification of a soluble recombinant Apaf-1 fragment longer than 200 amino acids. In addition, there is no published protocol that allows the preparation of a soluble, stable, recombinant Apaf-1 fragment longer than 200 amino acids, except for the full-length Apaf-1 protein in baculovirus-infected insect cells.
As a consequence of the lack of availability of reasonable quantities of Apaf-1 protein for research studies, little work has been done to identify compounds which can modulate the activity of Apaf-1. Furthermore, there has been an absence of three-dimensional structure information for any fragment of Apaf-1 other than the soluble N-terminal CARD domain. Knowledge of the physical structure of Apaf-1 protein would significantly aid design and screening of compounds that can modulate the activity of Apaf-1.
The present invention overcomes prior limitations by providing a method of producing stable, soluble, pure, and active Apaf-1 protein. The invention includes a description of the three-dimensional structure of Apaf-1 and methods of screening compounds to assess their ability to modulate Apaf-1 activity.
The invention relates to soluble, activatable fragments of metazoan apoptotic protease-activating factor 1 (Apaf-1) proteins. Although human Apaf-1 is exemplified herein, the invention relates to analogous fragments of any metazoan Apaf-1, such as that of any chordate or of any mammal. The fragments comprise substantially the entire alpha/beta fold domain and at least most of the helical domain I of the factor and lack at least a portion of the WD40 repeat domain of the factor. Unlike the full-length Apaf-1 protein, the fragments described herein are soluble. Functionality is also preserved by including in the fragment at least the adenine nucleotide-binding region of Apaf-1. Caspase-activating activity of Apaf-1 is preserved if the CARD domain is also inclued. For instance, a soluble fragment of human Apaf-1 can be made that includes at least residues 94-349 of the human Apaf-1 sequence disclosed herein (i.e., residues 94-349 of SEQ ID NO: 1). Other examples of soluble Apaf-1 fragments include those which include at least residues 1-349, at least residues 94-589, or at least residues 1-591 of that sequence. Most or all of the WD40 repeat domain should be excluded from the fragments so that solubility of the fragment is not degraded.
The invention includes Apaf-1 fragments having conservative amino acid substitutions relative to the Apaf-1 sequences disclosed herein. The substitutions are those which do not significantly reduce the activity or solubility of the fragment, relative to the native sequence. Several suitable substitutions are disclosed herein, and others can be readily made by a skilled artisan.
The invention also relates to methods of making soluble recombinant activatable fragments of metazoan Apaf-1 proteins. In these methods, a host organism is transformed with an expression vector that is operable in the host organism. The vector includes a coding segment that is operably linked with the promoter/regulatory sequences of the vector. The coding segment encodes a protein fragment as described herein. In one embodiment, the coding segment also encodes a metal binding sequence linked to the fragment. Enhanced expression of the fragment can be obtained by culturing the host organism at a temperature lower than 37 degrees Celsisus, such as a temperature in the range from 12-32 degrees Celsius. When the fragment is recombinantly expressed in the host organism, it can be isolated therefrom. Such isolation preferably is performed in the presence of a metal chelating agent.
The invention includes a variety of methods for assessing the ability of a compound to modulate activity of a metazoan Apaf-1. These methods can be performed by assessing the ability of the compound to affect an Apaf-1 protein fragment or an activity exhibited by such a fragment. For example, the ability of the compound to modulate the conformation of the fragment, hydrolysis of ATP catalyzed by the fragment, or activation of an apoptotic caspase (e.g., caspase-9) catalyzed by the fragment. In each instance, observing the phenomenon in the presence and absence of the phenomenon indicates the effect of the compound on the fragment and, by extension, on native Apaf-1.
The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.
Abnormal inhibition of apoptosis is a hallmark of cancer and autoimmune diseases, whereas excessive activation of cell death is implicated in neuro-degenerative disorders such as Alzheimer's Disease. Apoptosis involves the enzymatic activity of a family of specialized proteases known as “caspases.” Caspases are synthesized as inactive zymogens and must be proteolytically processed to become fully active.
Human apoptotic protease activating factor 1 (Apaf-1) is a known activator of caspase proteins and an important component of the apoptotic response in human cells. Analogs of human Apaf-1 appear to exist in most, if not all, metazoans. Apaf-1 is responsible for activation of the initiator caspase-9 and subsequent activation of effector caspases-3 and -7. Apaf-1 plays an essential role in the regulation of programmed cell death in mammalian development and in oncogene- and p53-dependent apoptosis. The critical importance of Apaf-1-mediated apoptosis is manifested by the observation that Apaf-1 is frequently inactivated in cancers such as malignant melanoma.
For these reasons, there has been intense interest in studying Apaf-1, its activity, and its role in human disease. However, because the protein could not previously be isolated from human cells in a practical way and because no practical method existed previously for bacterial production of soluble recombinant forms of the protein, knowledge of Apaf-1 and its significance has been severely limited.
The invention relates to discovery of a soluble recombinant form of Apaf-1 protein and its analogs that retains its nucleotide-binding and caspase-activating activities. The invention also relates to methods of producing these recombinant proteins and methods of using them, for example to identify compounds capable of modulating their activity.
The invention also relates to the crystal structure of Apaf-1 bound to ADP, which provides the first glimpse of this important protein at atomic resolution (2.2 Angstroms). The structure reveals, unexpectedly, that the nucleotide-binding pocket of Apaf-1 is much larger than the bound ADP and that the pocket is lined by amino acid residues that are positioned to make specific interactions. These structural observations indicate that particular nucleotide analogs can be synthesized to occupy the nucleotide-binding pocket of Apaf-1. Biochemical evidence presented herein shows that binding to this pocket directly impacts the ability of Apaf-1 to activate caspase-9. Hence these novel structural features can be used to design nucleotide analogs that either enhance or inhibit activation of caspase-9 by Apaf-1. This information has profound implications for the treatment of cancer and other apoptosis-related diseases.
Definitions
As used herein, each of the following terms has the meaning associated with it in this section.
An Apaf-1 fragment is “soluble” if at least at least 10% (by weight) of the fragment remains in the supernatant following lysis of cells used to produce the fragment and centrifugation of the lysate at 20,000×g for 40 minutes and if the fragment in the supernatant does not aggregate, as assessed by gel filtration chromatograpy, for example. Preferably, solubility of the fragment is assessed in a lysate of a bacterial culture used to produce the fragment at a concentration of at least 1 milligram per liter of bacterial culture.
An Apaf-1 fragment is “funtional” if the fragment exhibits both i) the ability to bind ADP and ATP and ii) the ability to activate caspase-9 following binding of ATP to the fragment.
Soluble Recombinant Protein Fragments
Previously, others have been unable to isolate recombinant Apaf-1 protein in a soluble, functional form from bacteria. It has been discovered that a recombinant fraction of Apaf-1 protein can be overexpressed in bacteria and isolated in a soluble, functional form. This soluble fragment retains the nucleotide-binding and caspase-activating activities of native Apaf-1 protein. Furthermore, given the amino acid sequence homology and similarity among metazoan Apaf-1 homologs (e.g., the similarity of human, fish, and fly Apaf-1 and worm CED-4 proteins is shown in
Human Apaf-1 consists of several distinct protein domains. Beginning at the amino-terminus, these domains are a CARD domain, an alpha/beta fold domain, an alpha-helical domain designated helical domain I, a winged helix domain, a second alpha-helical domain designated helical domain II, and a carboxyl-terminal WD40 repeat domain. A bridging helix designated alpha8 is interposed between the CARD and alpha/beta fold domains. It has been determined that the adenine nucleotide-binding region of human Apaf-1 involves the alpha/beta fold domain and at least most of helical domain I (possibly excluding alpha helix 19, as shown in
In one embodiment, the invention includes an isolated, soluble, activatable fragment of human Apaf-1 protein that includes at least residues 144-333 of SEQ ID NO: 1 (i.e., a portion of the alpha/beta fold domain that does not include alpha helix 9 and a portion of helical domain I that does not include alpha helix 19). Preferred fragments include fragments having the amino acid sequence of SEQ ID NO: 1, beginning with an amino terminus at any of residues 1, 94, and 108 and having a carboxyl terminus at any of residues 333, 349, 469, 554, 567, 570, 589, 591, 611, and 650.
Human Apaf-1 exhibits a splice variant that affects the sequence of residues 99-143 of SEQ ID NOs: 1, 3, and 5 and residues 99-132 of SEQ ID NOs: 2 and 4 (compare residues 99-143 of the sequence identified as “SF” with residues 99-132 of the sequence identified as “S” in
Residue 115 (i.e., a cysteine residue) in SEQ ID NO: 1 can be substituted with a serine residue. Residue 568 in SEQ ID NO: 1 (i.e., a cysteine residue corresponding to residue 557 in SEQ ID NOs: 2 and 4) can also be substituted with a serine residue. Substitution of residue 160 in SEQ ID NO: 1 (i.e., a lysine residue corresponding to residue 149 in SEQ ID NOs: 2 and 4) with an arginine residue lowers expression and solubility of the resulting fragment, but nonetheless results in a soluble fragment.
The CARD domain of Apaf-1 is necessary in order for a recombinant fragment of Apaf-1 to exhibit its caspase-activating activity. In order to obtain a soluble recombinant protein fragment, it is necessary that most or all of the carboxyl-terminal WD40 repeat domain be absent from the protein. Suitable recombinant protein fragments preferably include substantially the entire helical domain I, and can include the winged helix and helical domain II domains of the factor.
The Apaf-1 fragments of the invention lack most (preferably all) of the WD40 repeat domain of the corresponding native protein. For example, the WD40 repeat domain begins in the neighborhood of residues 600-620 of SEQ ID NOs: 1 and 3, and extends substantially through to the carboxyl-terminus of the protein, including 12-13 copies of WD40 domains (depending on the splice variant). The data presented herein demonstrate that soluble, functional fragments can be generated which completely lack any portion of the WD40 repeat domain (e.g., fragment 1-591 in Table 2). Fragments containing a portion of the WD40 repeat domain extending through at least residue 611 of SEQ ID NO: 3 have been generated and exhibit suitable solubility and function, and it is believed that inclusion of a portion of the WD40 repeat domain extending through at least residue 650 of SEQ ID NO: 3 will exhibit suitable solubility and function for the purposes described herein. A skilled artisan will understand that the maximum amount of the WD40 repeat domain that can be included in the protein fragments described herein is substantially limited by the desolubilizing effect of including increasingly large portions of this domain. Nonetheless, the skilled artisan will recognize that residue 650 of SEQ ID NO: 3 is not a definite boundary for the portion of the WD40 repeat domain that can be included, and that determination of such a boundary is a matter of routine experimentation, in view of the solubility required for any particular application.
Although only several Apaf-1 homologs from metazoans are explicitly exemplified in this application, it is apparent that soluble active protein fragments can be made from the same regions of substantially any metazoan Apaf-1 homolog that shares an analogous domain structure. By way of example, the sequences of regions of analogs of human Apaf-1 are shown in
The invention includes a soluble, activatable fragment of a metazoan Apaf-1, the fragment comprising substantially the entire alpha/beta fold domain and at least most of the helical domain I of the factor and lacking at least a portion of the WD40 repeat domain of the factor. The fragment preferably includes the bridging helix, and can include one or more of the CARD domain, the winged-helix domain, and the helical domain II of the factor. The fragment must lack at least most of, and preferably at least substantially all of, the WD40 repeat domain.
It is evident that the identity of many amino acid residues is not critical to the solubility, stability, or activity of human Apaf-1 protein. A skilled artisan is able, in view of the structural information presented herein, to select residues of human Apaf-1 or another metazoan Apaf-1 that can be substituted with a different amino acid residue. Similarly, in view of the information presented herein (e.g., in the figures and in Example 2), a skilled artisan is able to identify amino acid residues having relatively high significance for the structure, stability, or activity of Apaf-1 protein and avoid making amino acid substitutions at those positions. By way of example, substitution of a serine residue in place of one of the cysteine residues at positions 115 or 568 of SEQ ID NO: 1 does not significantly adversely affect the solubility or activity of the corresponding recombinant Apaf-1 protein fragment. Substitution of an arginine residue in place of the lysine residue at position 160 of SEQ ID NO: 1, by contrast, has a relatively greater detrimental effect on the solubility and expression of the corresponding fragment.
The soluble fragments of metazoan Apaf-1 described herein do not occur in nature. They can be produced by way of substantially any recombinant protein production method known in the art. That is, the fragments can be made by including a nucleic acid encoding the fragment in a suitable expression vector and transforming a host organism with the vector. The host organism and vector are selected to be compatible, such that the fragment is expressed from the vector in the host organism. A wide variety of host organisms and corresponding expression vectors are known in the art, and substantially any compatible host organism—expression vector pair can be used to generate the protein fragments described herein. A suitable expression vector encoding a fragment described herein will include a coding segment (encoding the fragment) operably linked with promoter/regulatory sequences sufficient to achieve expression of the fragment in the host organism. The expression vector can optionally include other transcription regulatory sequences (e.g., regions known to enhance or inhibit transcription in the host organism), indicator regions (e.g., an antibiotic-resistance gene to identify host organisms transformed with the vector), or other expression vectors known in the art. The transcript of the vector encoding the fragment can optionally include a region encoding a polypeptide linked to the fragment, such as a fusion protein, a polypeptide that directs the translated peptide to a particular cellular compartment or to the exterior of the cell (e.g., a signal peptide), or a polypeptide for facilitating recovery of the expressed fragment (e.g., a poly-histidine polypeptide suitable for binding a metal affinity chromatography medium). A wide variety of such polypeptides are known in the art, and substantially any of them may be linked with the Apaf-1 fragment described herein. Optionally, the peptide can be linked with the fragment by a specifically cleavable polypeptide region to facilitate removal of the peptide after expression of the peptide-linked fragment.
A recombinantly-expressed fragment can be recovered using substantially any protein purification method known in the art. Preferably, the fragment is linked with a ligand suitable for binding with an affinity chromatography medium, such as a hexahistidine sequence which is capable of complexation with a bound metal ion (e.g., see Sharma et al, 1991, Biotechnol. Appl. Biochem. 14:69-81). Use of an affinity tag of this sort simplifies recovery of the recombinant fragment and can improve the purity of the recovered fragment, relative to other protein purification methods. By way of example, a hexa-histidine tag can be added an end of the recombinant sequence to facilitate metal affinity chromotographic purification of the recombinant protein, and the tag can thereafter be cleaved using thrombin.
It has been discovered that reduction (or substantial elimination) of free metal ions in the media used for purification of the recombinant fragment can improve recovery of the functional fragment in a soluble form. Such reduction can be achieved by exclusion of metal ions from the media or, preferably, by including one or more chelating agents in the media. Many suitable chelating agents are known in the art, of which ethylenediamine tetraacetic acid (EDTA) is a suitable example. The media preferably include a chelating agent in significant excess (e.g., two-fold or more) of the known or anticipated concentration of metal ions in the media.
Another way to improve the yield of soluble, functional fragment obtained from a recombinant organism is to maintain a low incubation temperature. It has been discovered that improved recovery of functional recombinant fragment can be realized as the temperature at which the host organism is incubated during production of the recombinant fragment is reduced below 37 degrees Celsius, and preferably below 32 degrees Celsius. Greater recoveries were achieved at incubation temperatures of about 20 degrees Celsius and at 12 degrees Celsius than at 37 degrees Celsius. A skilled artisan is able to select a suitable incubation temperature based on the identity and temperature tolerance of the host organism selected for production of the fragment. In general, however, lower temperatures tend to increase recovery of the recombinant fragment in a soluble, functional form.
The invention includes a nucleic acid that encodes a soluble fragment of a metazoan Apaf-1 of the type described herein. The nucleic acid can have the nucleotide sequence of a naturally-occurring Apaf-1, separated from at least most of the WD40 domains-encoding portion of the naturally-occurring sequence. In view of the redundancy of the genetic code, a skilled artisan recognizes that the sequence of a nucleic acid encoding a particular fragment can vary from the naturally-occurring nucleotide sequence without altering the amino acid sequence of the encoded fragment. By way of example, a synonymous codon (i.e., one which encodes the same amino acid residue) can be used in place of a naturally-occurring codon when the nucleic acid is to be expressed in a host organism preferentially translates the synonymous codon relative to the naturally-occurring codon. Similarly, a skilled artisan will recognize that conservative amino acid residue substitutions can be encoded, corresponding to portions of the fragment for which the amino acid sequence is not identified herein as being important to functionality of the fragment. Thus, the invention include nucleic acids which encode a soluble, activatable fragment of a metazoan Apaf-1 encoded by a nucleic acid, wherein the nucleic acid is capable of hybridizing to a polynucleotide having the sequence of one of SEQ ID NOs: 1-8 under stringent hybridization conditions (e.g., during washing with a solution consisting of 15 millimolar sodium chloride, 1.5 millimolar sodium citrate, and 0.1% (w/v) sodium dodecylsulfate at 50 degrees Celsius).
Significant insolubility is a characteristic shared by many NOD proteins (Apaf-1 is a NOD protein). Many NOD proteins are believed to have a repeating carboxy-terminal portion that, in many instances, is not believed to significantly affect the activity of other protein domains nearer the amino terminus. By analogy to the methods described herein for making soluble, functional the Apaf-1 fragments, one can prepare recombinant fragments of other NOD proteins by removing most or all of the carboxy-terminal repeat domains from the NOD protein and recombinantly expressing the remainder of the protein in a bacterium. In this way, soluble, functional fragments of other NOD proteins can be prepared which are suitable for crystallization or other purposes described herein.
Screening Methods
The data reported herein indicate that activation of human Apaf-1 protein involves a conformation change that occurs upon binding and/or hydrolysis of an adenine nucleotide with Apaf-1. The binding site of the adenine nucleotide is identified in the data reported herein as being located in a portion of Apaf-1 that includes substantially the entire alpha/beta fold domain and at least most of the helical domain I of Apaf-1. Because this region of Apaf-1 appears to mediate most, if not all of the interaction of Apaf-1 with adenine nucleotides, observation of the effect of a nucleotide analog (or another compound) on this region is sufficient to predict the effect of the analog (or other compound) on activation of Apaf-1, and thereby the effect of the analog on caspase activation. Screening of compounds against a soluble fragment of Apaf-1 that includes at least this region is therefore a suitable method for assessing the effect of the compound on activation of Apaf-1 and its associated caspases. Such screening can be used to identify compound capable of therapeutic use in apoptosis-related disorders.
Because prior preparations of recombinant Apaf-1 from bacteria yielded little or no soluble, functional protein, they could not be practically used in such screening methods. The soluble, functional protein fragments described herein therefore represent an important advance in methods of identifying compounds suitable for treatment of disease conditions in which Apaf-1 has a role. Examples of such disorders include cancers such as malignant melanoma and neuro-degenerative disorders such as Alzheimer's disease.
The invention includes a method of assessing the ability of a compound to modulate activation of a metazoan Apaf-1 homolog. In this method, the compound is contacted with a soluble, activatable fragment of a metazoan Apaf-1 described herein and the conformation of the fragment is observed. Attainment of an open conformation by the fragment (i.e., a conformation wherein the fragment can form the apoptosome) is an indication that the compound activates the factor and attainment of a closed conformation by the fragment (analogous to the conformation of the ADP-bound Apaf-1 fragment described herein) is an indication that the compound inactivates the factor. The conformation of the fragment can be assessed in the presence of the compound alone. Alternatively, the conformation of the fragment can be assessed in the presence of an adenine nucleotide (e.g., ADP or ATP) having a known effect on the conformation, and the influence of the presence and absence of the compound on that known effect can be assessed.
Substantially any known method can be used to assess the conformation of the fragments described herein in the presence or absence of a compound. Examples of suitable techniques include fluorescence spectroscopy, light scattering analysis, size exclusion chromatography, and circular dichroism spectroscopy. X-ray crystallography and nuclear magnetic resonance spectroscopy can also be used to assess conformation. The ability of a compound to affect interaction of a human Apaf-1 fragment of the type described herein (or a soluble fragment of another metazoan Apaf-1) with an adenine nucleotide can be assessed directly (i.e., by assessing binding of a fluorescently- or radio-labeled nucleotide with the fragment, for example) or indirectly, by assessing Apaf-1-mediated hydrolysis of a hydrolyzable nucleotide in the presence and absence of the compound. Furthermore, the binding affinity of Apaf-1 toward an adenine nucleotide (e.g., ADP or ATP) can be assessed in the presence and absence of the compound. Lowered binding affinity for the adenine nucleotide is an indication that the compound binds with Apaf-1 in a way that inhibits adenine nucleotide binding. The kinetics of binding inhibition can be assessed in routine ways to yield further information about the mechanism (e.g., competetive or non-competetive) by which the compound inhibits adenine nucleotide binding, which can provide further information about the portion of the fragment with which the compound interacts.
If the fragment used in the screening assay comprises the CARD domain of the corresponding Apaf-1, then a caspase activation assay (of which many are known in the art) can be used to assess caspase-activating activity of the fragment in the presence and absence of the compound.
Therapeutic Compositions
The soluble, functional Apaf-1 fragments described herein can be administered to a cell or tissue of a patient in need of caspase activation. Such fragments can be suspended in a pharmaceutically acceptable carrier and administered to a body location at which Apaf-1 activity is needed. For tissues to which administration of a protein fragment described herein is not expected to be practical (e.g., where the fragment must be transported across the cytoplasmic membrane of a cell, a nucleic acid encoding such a fragment can be administered instead. A variety of suitable vectors (e.g., virus vectors and other vectors suitable for delivering a gene to the interior of a mammalian cell are known in the art, and substantially any of these vectors can be used to deliver an expression vector encoding a protein fragment described herein to a cell.
The three-dimensional Apaf-1 fragment structure disclosed herein can be used in computerized rational drug design (i.e., molecular modeling and molecule-molecule interaction modeling) methods to identify candidate compounds that bind with an active portion of Apaf-1. A variety of computerized drug design programs capable of modeling the interaction of a candidate compound with, for example, atomic coordinates described herein, are known in the art, and the operation of such programs is within the ken of the ordinary artisan in the field of rational drug design. Such methods can be used to design compounds that interact with the adenine nucleotide-binding portion of Apaf-1, the caspase-interacting (CARD) domain of Apaf-1, or portions of Apaf-1 which interact with one another (i.e., to disrupt Apaf-1 structure).
Candidate compounds can be designed based on the structure of a known ligand of Apaf-1. By way of example, adenine nucleotide analogs can be designed based on the structure of ADP or ATP for the purpose of designing an analog that will interact with the adenine nucleotide-binding portion of Apaf-1. Alternatively, the interaction with Apaf-1 of compounds in a library of known or modeled compounds can be assessed using the same rational drug design software. Iterative design methodologies can be employed, whereby a later generation of candidate compounds can be designed based on modeling data obtained with an earlier generation of candidate compounds (e.g., by modifying the structure of an earlier-generation compound for which the modeling software indicates strong Apaf-1-binding affinity).
A compound identified by rational drug design methods as likely exhibiting desirable interaction with Apaf-1 can thereafter be synthesized or purchased, and the compound's ability to interact with Apaf-1 in a desirable manner can be assessed experimentally. For example, compounds which are intended to modulate the activity of Apaf-1 can be assessed using any of the methods described herein.
Using the methods described herein, the crystal structure of a soluble Apaf-1 fragment having a candidate compound bound therewith can be assessed. This crystal structure information can be used to confirm that the compound binds with Apaf-1 in the manner suggested by the molecular modeling software used to design the compound. The crystal structure of a compound-bound soluble Apaf-1 fragment can also be used to design modifications of the compound structure that bind more or less strongly with Apaf-1, as described herein.
The invention is now described with reference to the following Examples. These Examples are provided for the purpose of illustration only, and the invention is not limited to these Examples, but rather encompasses all variations which are evident as a result of the teaching provided herein.
Preparation of Soluble Recombinant Apaf-1 Protein
Prior to the experiments described herein, there was no published protocol that allows bacterial expression and purification of a soluble recombinant Apaf-1 fragment longer than 200 amino acids. There was also no published protocol that allows the preparation of any soluble, stable, recombinant Apaf-1 fragment longer than 200 amino acids except that for the full-length Apaf-1 protein produced in baculovirus-infected insect cells. The full-length protein produced in baculovirus exhibited low yield and significant conformational heterogeneity, and its usefulness for many purposes is limited.
In the experiments described herein, several factors were discovered that enhance the expression of soluble Apaf-1 protein fragments in bacteria.
Length of Apaf-1 Protein Fragment
It was discovered that an Apaf-1 fragment having its carboxyl-terminal boundary between amino acids 550 and 650 of the native protein exhibits better solution behavior than those with carboxyl-terminal boundaries prior to amino acids 550. The structure of Apaf-1 was used to determine the length of appropriate Apaf-1 fragments. For example, Apaf-1 (residues 1-589) exhibits greater solubility than Apaf-1 (1-349) or Apaf-1 (1-469) because Apaf-1 (1-349) and Apaf-1 (1-469) failed to maintain the structural integrity of Apaf-1. For example, the following recombinant Apaf-1 proteins, residues 1-611, 1-591, 1-589, 1-570, 1-567, and 1-554, all exhibit good solubility and solution behavior because they allow the fragment 1-591 to form an intact structural entity, as seen in the crystal structure. A summary of Apaf-1 fragments and their relative degrees of expression and solubility is included herein as Table 2.
The structure of Apaf-1, described herein for the first time, reveals the domain organization of Apaf-1. This information provides a useful guideline for the design of soluble recombinant constructs for Apaf-1. For example, the primary nucleotide-binding site is located between the alpha/beta fold (residues 108-284) and helical domain I (residues 285-365), indicating that any functional recombinant Apaf-1 fragment must contain, minimally, the entire alpha/beta fold and the bulk of the helical domain I in order to be soluble and functional. Indeed, non-functional recombinant Apaf-1 (residues 1-286) protein is largely insoluble while Apaf-1 (residues 1-349) is soluble to significant extent.
Skilled artisans will appreciate that conservative amino acid substitutions that do not significantly disrupt the domain structure of Apaf-1 or alter residues shown herein to be significant for inter-domain interactions or binding of adenine nucleotides can be made without significantly altering the functionality of the protein. By way of example, point mutations K160R, C115S, and C568S were made in SEQ ID NO: 1 without significant loss of activity or solubility.
Growth Conditions
Recovery of soluble, active recombinant Apaf-1 protein fragments can be increased by decreasing the temperature at which the recombinant organism is cultured below 32 degrees Celsius. Lowering the temperature progressively facilitates the production of soluble recombinant Apaf-1 protein. For example, at 27 degrees Celsius, some recombinant Apaf-1 protein can be expressed in soluble form, whereas at ambient temperatures (i.e., between 20 and 24 degrees Celsius), a greater portion of recombinant Apaf-1 protein is soluble. A growth temperature below 20 degrees Celsius (e.g., 4, 10, or 15 degrees Celsius) further improves expression and production of Apaf-1 protein.
Purification of Recombinant Apaf-1 in the Presence of EDTA
It was discovered that incubation with excess amounts (here, 2-5 millimolar) of EDTA allows recombinant Apaf-1 protein to be recovered as a more homogeneous population of molecules than if EDTA is not used. Without being bound by any particular theory of operation, the beneficial effect of EDTA is believed to be attributable to the metal-chelating ability of EDTA. Metal chelation can reduce the ability of recombinant Apaf-1 to change between various conformations having differing chromatographic and other behaviors. It is believed that substantially any chelator of metal ions will have a similar effect. Although the exact amount of chelator used is not critical, the amount should be sufficient to chelate substantially all metals present in the medium from which Apaf-1 is to be purified.
Three-Dimensional Structure and Mechanism of Apaf-1
In this example, the 2.2-Angstrom three-dimensional structure of a functional fragment of Apaf-1 (residues 1-591) bound to adenosine diphosphate (ADP) is reported. This structure reveals, surprisingly, a closed conformation of Apaf-1. This structure provides a framework for understanding Apaf-1 function, apoptosome assembly, and caspase-9 activation. The information can also be used to decipher the general mechanisms of the NOD family of proteins.
The materials and methods used in this example are now described.
Protein Preparation
All constructs for protein expression were generated using a standard PCR-based cloning strategy, and the identity of individual clones was verified by double-stranded plasmid sequencing. A soluble Apaf-1 fragment (residues 1-591, sequence (SEQ ID NO: 1) shown in
Recombinant full-length caspase-9 was also over-expressed in E. coli strain BL21(DE3) using pET29 having the sequence described in Srinivasula et al., (1998, Cell 1:949-957) operatively ligated therein. Recombinantly-expressed caspase-9 was purified in a manner analogous to that used for the soluble Apaf-1 fragment, with addition of a gel-filtration chromatography step (using SUPERDEX™-200 chromatography medium, Pharmacia).
Crystallization
Purified soluble Apaf-1 fragment obtained from anion-exchange chromatography was used directly for crystallization. The fragment was initially present at a concentration of about 3 milligrams per milliliter in 20 millimolar HEPES buffer, pH 7.8, containing 2 millimolar DTT. Crystals were produced using the sitting-drop vapor-diffusion method. Protein samples having a volume of 2 microliters were mixed with an equal volume of a reservoir solution containing 100 millimolar HEPES (pH 7.1), 250 millimolar ammonium acetate, and 19% (w/v) polyethylene glycol (PEG-3350), and with 1 microliter of 100 millimolar DTT. Crystals appeared after one- to three-weeks of storage at 4 degrees Celsius, with a typical crystal size of 0.1 millimeter×0.1 millimeter×0.1 millimeter.
Derivative crystals were obtained by soaking crystals overnight in mother liquor containing 1 millimolar mercury thimerosal followed by back-soaking for 5 minutes. Native and derivative crystals were then equilibrated in a cryo-protectant buffer containing well buffer plus 5% glycerol and were flash-frozen in liquid nitrogen. Derivative crystals belong to the space group P21, with unit cell dimensions of a=47.83 Angstrom, b=76.01 Angstrom, c=81.15 Angstrom, beta=91.33 degrees, and contained one protein molecule per asymmetric unit. Native crystals belong to the space group P1, with unit cell dimensions of a=75.96 Angstrom, b=92.88 Angstrom, c=94.99 Angstrom, alpha=62.96 degrees, beta=89.99 degrees, and gamma=90.05 degrees, and contained four molecules of Apaf-1 in one asymmetric unit (the unit cell).
Data Collection and Structure Determination
Anomalous diffraction data were collected at the National Synchrotron Light Source (NSLS) beamline X25 using three wavelengths corresponding to inflection point, high energy remote, and peak of a Hg-MAD experiment. In addition, a 2.2-Angstrom native data set was collected at the Cornell High Energy Synchrotron Source (CHESS) beamline A1. The data sets were collected at 100 K using Quantum 210 CCD detectors. Data were integrated, reduced, and scaled using HKL2000 (Otwinowski et al., 1997, Methods Enzymol. 276:307-326). The initial structure was determined using 3.1-Angstrom Hg-MAD data using SOLVE/RESOLVE (Terwilliger et al., 1999, Acta Crystallogr. D55:849-861). Model building and TLS-refinement were performed using the 2.2-Angstrom native dataset using 0 (Jones et al., 1991, Acta Crystallogr. A47:110-119) and REFMAC5 (Collaborative Computational Project, 1994, The CCP4 suite: programs for protein crystallography, Acta Crystallogr. D50:760-763). Tight and medium NCS restraints were imposed for main chain and side chain atoms, respectively. Data collection and refinement statistics are summarized in Table 1. The final atomic model includes four monomers (chains A, B, C, and D; corresponding to Apaf-1 residues 1-586), four ADP molecules, and 782 water molecules in the asymmetric unit. No electron density was observed for residues 587-591, or for residues 95-103 in chains A and B.
ATPase Assay
ATPase activity was determined using a thin layer chromatography (TLC) assay designed to quantitatively measure the conversion of (alpha-32P)ATP to ADP. Reactions were initiated by mixing the soluble Apaf-1 fragment at a final concentration of 2 micromolar with a solution containing 20 millimolar HEPES, pH 7.5, 10 millimolar DTT, 10 micromolar magnesium chloride, 0.1 micromolar (alpha-32P)ATP, and 8 micromolar ATP. The reaction was incubated at room temperature for the indicated duration and then quenched by addition of an equal volume of developing solvent (1 molar formic acid, 0.5 molar LiCl). Samples were evaluated based on the differential mobility of ATP versus ADP on TLC using polyethyleneimine-cellulose F plates.
Caspase-9 Activation Assay
A fluorogenic assay was developed to monitor caspase-9 activation by the soluble Apaf-1 fragment. ATP or another nucleotide or nucleotide analog was mixed with buffer (20 millimolar HEPES, pH 7.5, 100 millimolar KCl, 5 millimolar DTT) to a final concentration of 1 millimolar in the presence or absence of 1 millimolar EDTA. Then, caspase-9 and Apaf-1 were added to a final volume of 200 microliter and final concentrations of 0.2 micromolar and 2 micromolar, respectively. After incubation for 5 minutes at room temperature, the caspase-9 substrate LEHD-AMC (a tetrapeptide coupled to 7-amino 4-methoxycoumarin) was added to the reaction at a final concentration of 200 micromolar. The conversion of the fluorogenic substrate was then measured using a Hitachi F2500 fluorescence spectrophotometer with an excitation wavelength of 380 nanometers and a fluorescence detection wavelength of 440 nanometers.
The results obtained from the experiments in this example are now described.
Crystallization and Structure Determination
Full-length (ca. 140-kilodalton) Apaf-1 protein exhibited considerable conformational flexibility that impeded crystallization efforts. Because the removal of carboxyl-terminal WD40 repeats does not affect the ability of Apaf-1 to bind ATP/dATP, to form the apoptosome, or to activate caspase-9 (Hu et al., 1999, EMBO J. 18:3586-3595; Hu et al., 1998, J. Biol. Chem. 273:33489-33494; Srinivasula et al., 1998, Mol. Cell 1:949-957), functional fragments of Apaf-1 devoid of the WD40 repeats were selected for study.
We assessed the micro-conformational heterogeneity of the recombinant protein. We found that, when expressed in bacteria or in insect cells, Apaf-1 exists in more than one conformation, as judged by its solution behavior on ion-exchange chromatography media and in caspase-9 activation assays. It was determined that the conformational heterogeneity is largely contributed by the status of nucleotide binding and the intrinsic ATPase activity of Apaf-1. It was discovered that incubation of Apaf-1 with EDTA during recombinantly-produced protein purification yielded recombinant Apaf-1 protein in a single conformation. The homogeneous Apaf-1 protein (residues 1-591) was crystallized in the space group P1, with four molecules in each asymmetric unit.
The structure was determined by multi-wavelength anomalous dispersion (MAD) using anomalous signals from mercury (properties listed Table 1). The final atomic model of Apaf-1, which includes 586 amino acid residues, was refined at 2.2-Angstrom resolution. The atomic coordinates determined from the modeling are deposited with the Protein Data Bank and assigned the accession number ______.
The four molecules of Apaf-1 in each asymmetric unit exhibit identical structural features important for this discussion. For simplicity, the remaining discussion describes just one of the four molecules.
Overall Structure of Apaf-1
A ribbon plot of the three-dimensional structure of the soluble Apaf-1 fragment (i.e., residues 1-591 of full length Apaf-1) is shown in
The N-terminal CARD domain (residues 1-107; colored green in
The alpha/beta fold domain (residues 108-284; colored blue in
A short alpha-helical domain (helical domain I, residues 285-365; colored cyan in
The C-terminal extended domain (helical domain II, residues 451-591; colored red in
The centrally located winged-helix domain is capped by the alpha/beta fold domain and helical domain I on the top, helical domain II at the bottom, and CARD on the side. Together, these five domains give rise to a relatively compact structure, having a length of about 80 Angstroms, a width of about 55 Angstroms, and a thickness of about 65 Angstroms. ADP is bound at the interface between three domains: the alpha/beta fold, helical domain I, and the winged-helix domain. Strikingly, the bound ADP molecule is deeply buried and inaccessible to even small molecules unless the conformation is changed. The inaccessible nature of the nucleotide-binding site indicates that the structure of Apaf-1 bound to ADP is in a closed conformation, unable to activate caspase-9 until opened.
To facilitate structural analysis, an internet-based search for structural homologues of Apaf-1 was performed using the program DALI (Holm et al., 1993, J. Mol. Biol. 233:123-138). Two of the most homologous structures were found to be the hexamerization D2 domain of n-ethylmaleimide-sensitive fusion protein (NSF), which is an essential ATPase required for intracellular vesicle fusion (Lenzen et al., 1998, Cell 94:525-536), and p97 (Zhang et al., 2000, Mol. Cell 6:1473-1484).
Both NSF and p97 belong to the AAA (ATPases associated with various cellular activities) family of ATPases (Lupas et al., 2002, Curr. Opin. Struct. Biol. 12:746-753). The conserved regions encompass the entire alpha/beta fold and helical domain I, with a root-mean-square deviation (RMSD) of approximately 4.7 Angstroms over 201 aligned C-alpha atoms between Apaf-1 and NSF or 4.4 Angstroms over 198 C-alpha atoms between Apaf-1 and p97. The presence of a short helical domain following the alpha/beta fold is a hallmark of the AAA family of ATPases, in which the helical domain energetically contributes to nucleotide binding (Lupas et al., 2002, Curr. Opin. Struct. Biol. 12:746-753). This analysis, in conjunction with sequence features identifiable in
CARD in a Closed Conformation
The CARD domain of Apaf-1 interacts with the prodomain of caspase-9 and this interaction is essential for the recruitment and subsequent activation of caspase-9 (Li et al., 1997, Cell 91:479-489; Qin et al., 1999, Nature 399:547-555). In the Apaf-1 structure, the N-terminal CARD domain stacks closely against the alpha/beta fold and the winged-helix domain through a large interface involving helices alpha2, alpha4, and alpha5 of the CARD domain, as shown in
Helix alpha2 of the Apaf-1 CARD domain is required for binding to the prodomain of caspase-9 (Qin et al., 1999, Nature 399:547-555; Zhou et al., 1999, Proc. Natl. Acad. Sci. USA 96:11265-11270). Simulated docking of caspase-9 prodomain to the CARD domain of Apaf-1 indicated significant steric clash between caspase-9 prodomain and the alpha/beta fold of Apaf-1.
Thus, in order for Apaf-1 to interact with caspase-9, the relative orientation of CARD with respect to the rest of Apaf-1 structure must be changed. In addition, the packing of CARD with other domains of Apaf-1 further restricts access to the bound ADP molecule (e.g., see
The interactions between CARD and the alpha/beta fold and the winged-helix domain primarily involve a network of 13 inter-domain hydrogen bonds, with limited van der Waals contacts (as shown in
The extensive inter-domain packing results in the burial of an approximately 2550 square Angstrom solvent-exposed surface area between CARD and the alpha/beta fold and the winged-helix domain. These CARD-mediated interactions are dominated by hydrogen bonds, which are conducive to regulation due to their fast kinetic rates. In addition, these interactions appear to significantly stabilize the limited contact between the alpha/beta fold and the more C-terminal winged-helix domain, allowing His438 of the winged-helix domain to make a direct hydrogen bond to stabilize the bound ADP molecule (see
Effect of Caspase Binding to CARD
To examine whether CARD in the closed conformation of Apaf-1 is capable of binding to caspase-9, we incubated the soluble Apaf-1 fragment (i.e., residues 1-591 of full length Apaf-1) with the full length, processed caspase-9 (residues 1-416) and compared the behavior of the complex with either component on gel filtration chromatography. The elution volume for Apaf-1 (residues 1-591) corresponds to a molecular mass of a monomer (i.e., 65 kilodaltons). The elution volume for caspase-9, which contains an extended linker segment between its prodomain and the caspase unit, corresponds to a molecular mass of about 90 kilodaltons, almost twice of that expected for a monomer (47 kilodaltons). This discrepancy is likely caused by the enlarged radius of hydration of caspase-9 due to the extended linker segment, because caspase-9 exists exclusively as a monomer in solution. When equi-molar amounts of Apaf-1 and caspase-9 were used, the center of the protein peak was shifted to an earlier fraction, which corresponds to a molecular mass of approximately 130 kilodaltons, consistent with a 1:1 complex between Apaf-1 and caspase-9. The formation of the Apaf-1/caspase-9 complex appeared to be inefficient, as judged by trailing fractions that likely contained non-interacting Apaf-1 or caspase-9.
Next, we investigated whether formation of a 1:1 complex between Apaf-1 and caspase-9 had any effect on the catalytic activity of caspase-9. Surprisingly, the activity of caspase-9 was significantly and reproducibly elevated in the presence of Apaf-1 despite the absence of ATP and the absence of the apoptosome. This observation suggests that, even in a 1:1 complex, Apaf-1 is able to allosterically enhance the catalytic activity of caspase-9. The formation of the apoptosome can greatly accelerate this process.
Nucleotide Binding
A surprising revelation of the Apaf-1 structure is that the bound nucleotide is ADP rather than ATP. The binding pocket for ADP is formed at the junction of four domains, CARD, alpha/beta fold, helical domain I, and winged-helix domain. ADP binding appears to help bring together these adjoining four domains. Consequently, the bound ADP molecule is buried and the only narrow channel from ADP to solvent is blocked by the packing of the CARD domain. This structural organization strongly indicates that unpacking of CARD, which can be achieved through interaction with the prodomain of caspase-9, leads to a more accessible nucleotide binding pocket.
Consistent with this analysis, dATP/ATP binding to Apaf-1 was shown by Jiang et al. (2000, J. Biol. Chem. 275:31199-31203) to be facilitated by the presence of caspase-9. Our structural analysis also suggests that even a slight perturbation to the nucleotide-binding pocket, such as exchange or hydrolysis of nucleotide, may result in the disruption of the inter-domain packing among the four adjoining subunits. This suggests that nucleotide analogs can significantly modulate the activity of Apaf-1, either enhancing or inhibiting the activity, depending on the effect of the particular analog on interactions between the subunits. The structure of Apaf-1 described herein indicates the types and structures of ADP/dADP/ATP/dATP analogs that can be expected to modulate Apaf-1 activity.
The adenine base of the bound ADP exists in the anti conformation with respect to the ribose (see
A few well-ordered water molecules appear to play an important role in binding to ADP. The side chain of Arg129 and the carbonyl group of Gly159 make a water-mediated hydrogen bond to the N7 atom of the adenine base; whereas Val125 and Ser422 make water-mediated hydrogen bonds to the adenine base and the ribose, respectively (see
Compared to other members of the AAA+ATPase family, the unique feature of Apaf-1 is the involvement of the winged-helix domain in the coordination of ADP, with His438 and Ser422 contributing two hydrogen bonds. Consequently, the ADP molecule is deeply buried. The significant yet weak interactions between ADP and the winged-helix domain suggests that this domain may be prone to conformational shifts.
ATP Hydrolysis
Both structural comparison with other AAA ATPases such as p97 (Zhang et al., 2000, Mol. Cell 6:1473-1484) and sequence features indicate that Apaf-1 belong to the AAA+ family of ATPase (Neuwald et al., 1999, Genome Res. 9:27-43), consistent with an earlier suggestion (Jaroszewski et al., 2000, Proteins 39:197-203). The sequences of Apaf-1 contain all the key elements of an ATPase that are required for activity, including the P-loop and the Walker B motif. However, whether Apaf-1 is a bona fide ATPase remained unresolved, largely due to the technical difficulty in eliminating trace amounts of contaminating ATPases (Zou et al., 1999, J. Biol. Chem. 274:11549-11556; Saleh et al., 1999, J. Biol. Chem. 274:17941-17945; Jiang et al., 2000, J. Biol. Chem. 275:31199-31203). It was also unknown whether ATP hydrolysis is essential to the function of Apaf-1 and, if so, how ATPase activity is related to the function of Apaf-1.
To definitively address these issues, we over-expressed the soluble Apaf-1 fragment (residues 1-591) described in this example in E. coli and purified this recombinant protein to homogeneity. Unlike full-length Apaf-1 that was expressed and purified from baculovirus-infected insect cells, the bacteria-derived Apaf-1 is free of any detectable contaminating proteins as judged by SDS-PAGE, amino acid analysis, HPLC, and mass spectroscopy. This Apaf-1 protein fragment is folded well, as judged by gel filtration and circular dichroism analyses.
Using thin layer chromatography (TLC), we reconstituted an in vitro ATPase assay to detect the hydrolysis of (alpha-32P)-labeled ATP, and raw results obtained using this assay method are shown in
Next, we examined whether ATPase activity is essential to the function of Apaf-1 using an in vitro reconstituted caspase-9 activation assay. These results are shown in
dATP has been reported to be a more potent activator of the apoptosome-mediated caspase-9 activation than ATP (Li et al., 1997, Cell 91:479-489). At a concentration of 2 micromolar, dATP promoted greater activation of caspase-9 than did ATP. We also discovered that, in contrast to ATP, dATP exhibits an optimal range of concentration at 1-5 micromolar. At higher concentrations of dATP, Apaf-1-mediated activation of caspase-9 is reduced slightly. Similar to ATP, dATP-mediated activation is dependent on its hydrolysis, because blocking the ATPase activity of Apaf-1 in the presence of 1 millimolar EDTA and 2 micromolar dATP also completely abolished caspase-9 activation (e.g., compare
Effect of Nucleotide Analogs
The dATP analog 2-chloro-deoxyadenosine (2CdA) is an effective agent in the treatment of several forms of cancer such as chronic lymphocytic leukemia (CLL) and hairy cell leukemia (Juliusson et al., 1996, J. Clin. Oncol. 14:2160-2166). The cytotoxicity of 2CdA largely depends on the accumulation of its 5′-triphosphate metabolite 2CdATP (Juliusson et al., 1996, J. Clin. Oncol. 14:2160-2166; Kawasaki et al., 1993, Blood 81:597-601). The chemotherapeutic effect of 2CdATP cannot be completely attributed to its deleterious incorporation into newly synthesized DNA, because 2CdATP also induces apoptosis in non-dividing lymphocytes (Juliusson et al., 1996, J. Clin. Oncol. 14:2160-2166). Interestingly, 2CdATP was shown to cooperate with cytochrome c and Apaf-1 to activate caspase-3 in a cell-free system (Leoni et al., 1998, Proc. Natl. Acad. Sci. USA 95:9567-9571; Genini et al., 2000, J. Biol. Chem. 275:29-34).
To further investigate the molecular mechanism by which 2CdATP regulates the function of Apaf-1, we examined the effect of various concentrations of 2CdATP on caspase-9 activation. As anticipated, 2CdATP is a potent activator for caspase-9 activation, with an optimal working concentration of approximately 100 micromolar. Similar to ATP and dATP, the hydrolysis of 2CdATP by Apaf-1 is essential to caspase-9 activation, as EDTA incubation led to loss of this response (compare
In the ADP-binding site of Apaf-1, the adenine- and ribose-binding pocket contains five well-ordered water molecules, which mediate hydrogen bonds between ADP and amino acid residues from Apaf-1 (see
Implications on Apoptosome Assembly
The role of ATP hydrolysis is manifest from the structural analysis presented in this example: hydrolysis of ATP at the junction of four subunits results in the reorganization of these domains and subsequent formation of the apoptosome. Indeed, the AAA+ family of proteins, such as the large T antigen of the DNA tumor virus SV40, is known to couple ATP hydrolysis with conformational changes that result in important functional consequences (Gai et al., 2004, Cell 119:47-60; Gai et al., 2004, J. Biol. Chem. 279:38952-38959; Wang, 2004, J. Struct. Biol. 148:259-267).
In order to identify conformational changes to Apaf-1 that ATP hydrolysis brings about, we studied known AAA+ ATPases and observed that their oligomerization (frequently hexamerization) is mediated by a conserved mode of domain organization. For example, for both NSF and p97, the far end of the alpha/beta fold of one protomer stacks against the wedge between the alpha/beta fold and its carboxyl-terminal helical domain of the adjacent protomer. This interaction is repeated six times, resulting in the hexamerization of NSF. The same general interface topology was also observed in a number of other AAA+ ATPases, including the bacterial protein HslU and the viral protein SV40 large T antigen. Thus the overall packing arrangement at the oligomerization interface of AAA+ ATPases is conserved.
The ADP-bound Apaf-1 exists in a closed conformation. In agreement with the biochemical and structural analysis presented herein, superposition of the alpha/beta fold of Apaf-1 into that of one NSF protomer in the hexameric structure results in extensive overlap of structures among neighboring Apaf-1 molecules. To remove these severe steric clash, the winged-helix domain and helical domain II must be dislocated from their current positions and rotated by approximately 45-50 degrees. This re-organization allows the formation of a symmetric oligomer for Apaf-1 (
The proposed conformational change in Apaf-1 is supported by all available structural, biochemical, and sequence information. First, the structural organization of the four subunits at the ADP-binding site strongly suggests an ATP hydrolysis-induced conformational change. His438 from the winged-helix domain directly coordinates the beta-phosphate of ADP. Second, biochemical characterization indicates that ATP hydrolysis is coupled to apoptosome formation and subsequent caspase-9 activation. Third, Apaf-1 shares significant structural homology with the hexamerization domain of NSF and p97, suggesting a similar mode of oligomerization. More importantly, the interface organization of the AAA+ family of proteins appears to be conserved, allowing us to predict conformational changes in Apaf-1 based on known structures.
In this example, the crystal structure of Apaf-1 bound to ADP at 2.2-Angstrom resolution has been described. In the structure, ADP is located at the junction of and coordinated by four domains. Binding to ADP has an essential role in organizing the structure of Apaf-1 into a closed form, in which the CARD domain packs against three other subunits and the ATPase domain exhibits a conformation that is incompatible with formation of the apoptosome.
The full activation of caspase-9 appears to comprise two stages. In the first stage, Apaf-1-mediated binding of caspase-9 in the absence of ATP/dATP leads to significant elevation of the caspase-9 activity. This observation indicates that monomeric Apaf-1 can allosterically improve caspase-9 activity, either through direct modification of the caspase-9 active site or by relieving the negative impact of caspase-9 prodomain on its own catalytic activity (Shi, 2004, Cell in press).
In the second stage, binding and hydrolysis of ATP/dATP induces a conformational change in Apaf-1 that propels its oligomerization and formation of the apoptosome holoenzyme. This reorganization depends on the energy released by the hydrolysis of bound nucleotide. Importantly, nucleotide analogs such as 2CdATP can also substitute ATP/dATP to activate Apaf-1. The binding of such nucleotide analogs is safeguarded by the enlarged ADP-binding pocket, which further suggests synthetic approach to improve the binding affinity and hence potency of the nucleotide analogs. Such analogs have important uses the therapeutic treatment of cancer, for instance.
aRmerge = Σ |I − <I>|/ΣI, where I is the observed intensity and <I> is the average intensity from multiple observations of symmetry-related reflections, the value in parentheses correspond to the highest resolution shell.
bRcryst = Σ |(Fobs) − (Fcalc)|/Σ(Fobs),
cRfree = same as Rcryst but comprises a test set (5.1% total Reflections), which was not used in model refinement.
The disclosure of every patent, patent application, and publication cited herein is hereby incorporated herein by reference in its entirety.
While this invention has been disclosed with reference to specific embodiments, it is apparent that other embodiments and variations of this invention can be devised by others skilled in the art without departing from the true spirit and scope of the invention. The appended claims include all such embodiments and equivalent variations.
This application is entitled to priority pursuant to 35 U.S.C. §119(e) to U.S. provisional patent application 60/628,000, which was filed on 15 Nov. 2004, with attorney docket no. 7616/99/127.
This work was supported in part by grants from the National Institutes of Health (NIH Grants Nos. RO1 CA090269 and RO1 CA095218) and the U.S. Government may therefore have certain rights in this invention.
Number | Date | Country | |
---|---|---|---|
60628000 | Nov 2004 | US |