The Chimeric Antigen Receptor (CAR), first devised by Gross et al in 1989 (1), has proven effective against cancer and holds promise for other therapeutic applications. Landmark work done by Carl June M.D. et al. led to a recent clinical trial in which CD-19 specific-CAR transfected autologous T cells were reinfused into a patient with chemorefractory Chronic Lymphoid Leukemia (CLL). This patient, along with another in the same trial, achieved a complete response within 30 days and remained in remission 10 months after treatment. A specific immune system response was detected and memory effector cells were generated. The only high level toxic events were Tumor Lysis Syndrome and Hypogammaglobulinemia the later expected as the CAR was specific for CD19 which is expressed on all early B cells (2). The historical significance of Dr. June’s work notwithstanding, a limitation to this approach exists stemming from the fact that the peptide containing the epitope targeted by the CAR’s variable region must exist on the surface of the tumor cells. With the exception of hematopoietic cells, as exemplified in Dr. June’s work, the peptide would preferentially be expressed only on tumor cells so off-tissue toxicity could be avoided. Unfortunately, this has already occurred during another CAR trial in which the specificity was set for erbB-2. ErbB-2 is expressed on breast and colon cancer cells, but it is also found on healthy endothelial cells; one patient died just days after treatment and on-target/off-tissue events were postulated as the root cause of death (3).
I am proposing an approach that differs from the conventional in three ways: (i) directly targeting of tumor specific or associated antigens within the malignant cells with (ii) an antibody based biological agent (iii) containing a constitutively active apoptosis effector that is inhibited or non-functional until the antibody binds its agonist. I propose investigating three possible configurations to achieve this end. The first is a configuration similar to that of a standard CAR where the transmembrane domain is replaced with an engineered protein. The second would be a simpler configuration in which a scFv is embedded into the apoptosis effector using an intein. The third would use a split effector in an approach called Sequence Enabled Reassembly of Proteins (SEER). This approach appears, at least at first, more complex, though it may provide broader scope for targeting as zinc fingers can be used to bind directly to DNA rather than a gene product which actually may present a limitation analogous to that of the standard CAR as a result of protein folding. Further, this would allow the targeting of common oncogenic mutations that do not translate into a gene product such as those in gene promoter regions.
The Chimeric Antigen Receptor
The CAR is made up of four components; the transport (leader), the antigen binding (scFv), the transmembrane link, and the signaling. Given its residence in the membrane of cytotoxic T cells, the function and method of the scFv, the signaling regions, and even the transport, are readily apparent, however the transmembrane domain is rarely mentioned and the method of signal initiation even less so.
In general there are three steps in the engineering of the gene which encodes the CAR. The first is to identify a molecule found on the surface of the cells that are to be targeted by the transfected cytotoxic T cells. The second is to find a binding molecule with the appropriate specificity such as a single chain Fv (scFv) extracted from an existing mAb or the binding region of a co-stimulatory receptor (1). For example, Herceptin is a humanized mouse mAb used in the treatment of breast and other cancers. It has binding specificity for erbB2 which is highly expressed on the surface of malignant cells though unfortunately on certain endothelial cells as well (4). Another example is the CD8 receptor which binds with high affinity to the glycoprotein GP120 present on the envelope of the HIV (5).
In most cases, a mAb is the source and segments of the genes that encode the variable regions of the heavy and light chains are linked together from the 3′ end of the light chain to the 5′ end of the heavy using a standard flexible peptide. This produces a gene which encodes a single protein designated scFv or Single Chain Fragment Variable. The scFv is linked to the signaling proteins via a hydrophobic alpha helix usually derived from CD28 (6).
The last step is to choose the signal generation proteins that will form the cytosolic end of the chimeric receptor. This step is far simpler than the first two given Janeway’s three signals and the small number of signaling receptors involved in each. However, in earlier generations of CARs, the full complement required to satisfy Janeway’s signals were not present. The first relied solely on the CD3-zeta chain of the T cell receptor complex and were also MHC presentation limited (1). Second generation CARs, for example, used the zeta (ζ) chain of the T Cell Receptor Complex joined to the scFv via the transmembrane region of a co-stimulatory molecule such as CD 8 or CD28. Third generation CARs incorporated the intracellular signaling domains from 4-1 bb or OX40 thus providing the survival co-stimulation signals resulting in enhancing survival and proliferation of the transfected T cells (7).
Once the chimeric gene is formulated and cloned, the in vitro process of stimulation, transfection, and replication is undertaken. After extraction of the autologous T cells, expansion is induced in vitro using magnetic beads coated with anti-CD28 and anti-CD3 monoclonal antibodies. The chimeric gene is then delivered to the cells via a lentiviral vector and upon positive selection and after quality tests; they are cryogenically frozen until the time of administration (2).
The iCCAR: A Derivative Approach
The main requirement for the development of a clinically successful CAR is the availability of a signature molecule on the surface of the target cells. To circumvent this inherent limitation of the CAR design, I am exploring the possibility of creating a biological agent that would amount to an intracellular chimeric antigen receptor. This new class of CAR, referred to hereafter as an intracellular Caspase-Modulating Chimeric Antigen Receptor (iCCAR), would be directly transported across the membrane to the cytoplasm or specific subcellular compartment via a Guanidinium Rich Transporter (GRT). The iCCAR is a chimeric protein comprised of a scFv, engineered with specificity for a tumor specific epitope, embedded within an intein that is integrated into an apoptosis-inducing protein. The basic idea is to deliver into somatic cells a biological agent capable of detecting signature sequences in mutant proteins and, upon detection, induce apoptosis.
The Antigen-Binding Domain
In the iCCAR, the antigen binding domain is a scFv derived from a mAb engineered with specificity for a signature epitope within the sequence of mutant protein expressed in malignant cells of a given phenotype. In most cases cells from malignant tissue from an afflicted individual will be sequenced to identify the epitope, thus requiring on-demand engineering of the applicable chimera.
However, with regard to oncogenes, the research suggests it isn’t the function of the protein product that is necessarily altered in an oncogenic mutation, but rather a change in transcription regulation, a substantial decrease in the ubiquitination-proteasome turnover rate, or its active state changing from induced to constitutive that is the actual driver of oncogenesis (8) (9). From the latter two, it seems plausible that the mutations should involve only a few sequence substitutions given that the protein itself is left functionally intact. Whether these “hotspots” arise from Single Nucleotide Polymorphisms or consistently aligned translocations, it appears that there may be epitopes that are common across cancer types. This may leave only one major question; whether or not a sufficient level of specificity is achievable to ensure off-target toxicity is not a probable outcome. The fact that it has been reported that amino acids with charged side chains can influence specificity by 1,000 fold may hold the answer, but at the very least may help focus the search (10).
Further, although malignancies may involve a large number of mutated genes, it is becoming clear that there are a relative few that are critical for survival and of those some have an active role while others passive. That is, certain mutated genes encode products that drive oncogenesis and thus are required for survival and growth while others that are not. This phenomenon, referred to as “Oncogene Addiction”, reveals that though there may be many mutant genes in a malignant cell’s genome, the targeting of the product of just one can lead to its death. There have been many cases in humans and animals where this has been successfully demonstrated (8).
Given this, it may be possible for single mAb to be engineered with high affinity for an epitope contained in an oncogene product whose sequence is conserved. The protein kinases commonly investigated in targeted cancer therapies, come to mind though they may not be suitable candidates because neither the tyrosine, threonine, nor serine involved has charged side chains; however there appears to be other candidates.
Actually, there may be many others with c-Myc looking like a good place to start. It is considered to be a primary proto oncogene with one study implicating it in approximately 70,000 cancer deaths per year in the U.S. (This figure may actually turn out to be a very low estimate given that the data was from 1995 and the number of data sets available for analysis has increased exponentially since then). (11)
In Myc mutations, translocations are common and occur to the immunoglobulin genes on chromosomes 2, 14, and 22 resulting in lymphoid malignancies. Somatic hypermutation occurring after the translocation commonly cause SNPs at Thr58, Ser 62, and, significant here, Pro57. These specific mutations have been shown to result in Burkitt’s lymphoma. Elevated expression of the c-Myc oncogene has also been reported in Lung Carcinoma and in one-third of breast and colon carcinomas. (9)
If a mAb can be engineered to target an epitope from just this one conserved sequence, widespread applicability may be the result. Works cited here as well as other not mentioned support this notion with one notable example demonstrating that an engineered mAb with specificity for a Nucleophosmin mutation was effective in the diagnosis of Acute Myeloid Leukemia (12).
The Effector – Caspase
Caspase-8 was originally chosen for the effector region because of its ability to directly activate Caspase-3 thus simplifying, it was thought, the regulatory matrix that might inhibit or otherwise negatively affect the signal downstream. Also, it is well studied and starting higher in the pathway seemed preferable as it was thought necessary to produce sufficient signal via the resultant cascade to induce apoptosis. However, after examining the signal path it became apparent that more factors are in play.
Caspase-8 is the initiator in the extrinsic apoptotic pathway that responds to external signals with a pathway that begins with the ligation and subsequent clustering of the Fas (CD95) receptors. This in turn leads to a conformational change in the receptor’s intracellular domain which exposes its Death Domain (DD). This exposure, in turn, allows the recruitment of FADD (Fas Associated via Death Domain) which is an adapter molecule. FADD’s binding with Fas produces a conformational change in FADD which exposes its Death Effector Domain (DED). This domain recruits initiator Caspases, Caspase-8 in this case, to the growing molecular complex known as the Death Inducing Signaling Domain (DISC). Increased concentrations of Caspase-8 leads to its dimerization and thus activation (dimerization also leads to the cleavage of the enzyme’s intersubunit linker though this seems only to stabilize the homodimer and does not appear to be required for activation) (13) (14).
Normally, it is only after this progression of clustering, activation, and dimerization that active Caspase-8 is released from the DISC. Once activated and released, Caspase-8 proceeds to act on its primary substrate; Caspase-3. Caspase-8 activates Caspase-3 by cleaving the Intersubunit Linker (IL) resulting in the release of two active site loops thus forming the substrate binding pocket (15). The substrates of active Caspase-3 are immediately destructive and include CAD (Caspase Activated DNAse) which is normally found in a heterodimer with its inhibitor, ICAD. After its inhibitor is cleaved, CAD then degrades chromosomal DNA (13). The substrate binding domain of CAD was briefly considered as a candidate to replace Caspase-8, however, it was conjectured that the bi-products of all Caspase-3 substrate activity may be required to encourage non-inflammatory consumption by phagocytes though no data was found to support this.
Another complication that exists in the use of Caspase-8 as the effector for a targeted cancer treatment is that the Caspase-8 mediated pathway is only active in certain cell types referred to as Type I as opposed to Type II cells in which only the intrinsic, or mitochondrially-mediated pathway has been observed (15). I haven’t been able to locate a breakdown, but it appears that Type I cells are those that, like lymphocytes, significantly change phenotype and rapidly expand and contract in population based on environmental signals.
Caspase-3, on the other hand, though downstream of the cascade, is the intersection of both Caspase-modulated apoptotic pathways making it the next logical choice for an Apoptotic effector. I didn’t consider it originally for the reason mentioned earlier, however, there is a mutant form of Caspase-3 (D3A, V266E) that is not only constitutively active but also has been shown to be immune to inhibition by XIAP (X chromosome-linked Inhibitor of Apoptosis Protein). Further, it has been shown that V266E can actually induce apoptosis in low concentrations in mammalian cells. (16)
Directed Evolution and Rational Protein Design
The interaction of forces at the atomic and quantum level that are the fabric in which proteins fold are not well understood. It is for this reason that, just as we confirm the very existence of proteins, an indirect path must be followed to engineer them.
Marc Ostermeier wrote in 2009 given that allosteric effects depend on the “protein’s sequence, structure, and energetics” one might reasonably conclude that computational modeling would be required to engineer the same. However, it has been rational design strategies that have proven most often successful. This is due, he continues, to the “modularity of protein function”. (17).
While it has been widely demonstrated that proteins are modular with defined functional domains and fusions of those domains from multiple sources yield novel chimeric proteins of predicable function, stabilizing those chimeras as well as goal-specific alterations of a given domain requires directed mutation. The predictive capacity that precludes the ability to perform directed mutation, however, requires that very understanding of protein folding we have not yet achieved (18).
However, fifteen years ago Frances Arnold described a process known as Directed Evolution by which the necessary alterations to amino acid sequences could be arrived upon without the need to completely understand protein folding. The process begins with a protein that is found in nature, the parent, and a desired function. Examples of this function include an energetic stabilization of an inserted domain into the parent or an alteration of the domain’s specificity (18).
The process by which the desired function is obtained is a mimic of natural evolution and begins with the selection of the natural protein that is to be altered. However, instead of the appropriate change being dictated by an environmental challenge a desired change is defined by the researcher. The next step in the process is to introduce one or two single nucleotide mutations in the sequence to create a library of variants. These mutations may not be purely random as they would be in nature, but rather rationally focused using structural and mechanistic information. This is highly desirable given that the majority of random mutations lead to a lowering of fitness or a complete loss of function. Past this, sheer volume is an issue as even a protein of just 100 amino acids has approximately 10130 possible sequences; or more atoms than there are in the universe (18).
Once the library of variants is established, screening or selection takes place to identify sequences of higher fitness via a range of assays. The process is then restarted using the selected proteins and continues iteratively until the desired change achieved. Interestingly, on average, it only takes five to ten “generations” to reach, no time compared to the natural process (18).
Chimeric protein engineering using a combination of Rational Protein Design and Directed Evolution has been demonstrated successful in many experiments with several notable examples including Roger Tsien’s work in florescent proteins for which he won the Nobel Prize. This combination will serve as the core strategy for the development of the proposed iCCAR (18).
An intein, or Internal Protein Sequence, is a domain that is capable of self-excision as well as catalyzing the ligation of the flanking sequences known as exteins. Found in organisms in all three domains and in viral proteins, they were first discovered 20 years ago in studies of Vacuolar Membrane proton-ATPase (VMA1) in Saccharomyces Cerevisiae (19).
From sequence analysis VMA1 was predicted to produce a protein of 118 kDa, however when estimations of the actual weight were done from SDS-PAGE gels it was only 67 kDa. Analysis of the gene sequence revealed that the n- and c- terminal regions were conserved among vacuolar membrane ATPases in other organisms, however the central region of 454 amino acids was unique. This region was found to be similar to an endonuclease and is present in the mRNA, though not in the protein after translation (19).
It was determined that the region was excised from the host protein without dependence on any external catalyst and before that event, the host protein was non-functional. After excision and ligation of the exteins, the host protein regained its normal form and function and was termed mature; the excised protein was found to be stable. This process, termed Protein Splicing has since been exploited ever since and has been particularly useful in reducing costs and improving quality in protein purification by alleviating the need for the use of site-specific endoproteases to remove affinity tags. Applications in molecular biology have also been demonstrated including the controlled expression of toxic proteins and the addition of non-canonical amino acids to existing sequences. (19)
It is now known that inteins are of three types; inteins, mini-inteins, and split inteins, with the first often containing an endonuclease absent in the mini examples. Even more interesting, split inteins can also occur In Trans. In this case the intein exists split into an n- and c- terminal region which are synthesized as part of two separate proteins. Post translational association between the two regions results in the assembly of the intein followed by its excision leaving the intein and a mature protein (19).
Antibody-Intein Fused Caspase-3 Mutant
Figure 1 – iCCAR configured with Intein-scFv switch
Controlling Proteins via Intein Function
Multiple methods have been developed to utilize intein function to manipulate protein function but work done by Wood and Skretas which established a general procedure commonly known as a Small-Molecule Controlled Inteins, is of specific relevance here (20).
Experiments conducted between 1999 and 2004 formed the foundation for their work by providing a suitable engineered intein which could be fused into their selected host protein (20) and establishing the fact that the only apparent restriction in the selection of a host protein is that it contains a Cysteine, Serine, or Threonine as the first amino acid of the c-terminal extein (19). The mini intein they chose was derived from the intein Mtu RecA (Mycobacterium Tuberculosis RecA) which originally contained an endonuclease. The endonuclease was deleted and a single amino acid switched to produce a mutant that displayed high splicing efficiency in a variety of proteins (20).
As the small molecule receptor, they chose the binding domain from the Human Thyroid Hormone Receptor β (TR) which increased in stability and underwent a conformational change when it bound its ligand. Among the factors contributing to their decision to use TR include the observation that mutations in the endonuclease domain of the WT intein effected splicing efficiency thus implying cross-talk between the regions. Also, it was shown that, in place of the endonuclease, the intein would accommodate entire folded protein domains. (20)
They found that when the binding domain was inserted the splicing capability of the intein was destroyed; however, it was restored when it ligand bound. Using the T4 thymidylate synthase (TS) gene as a host and through a temperature calibrated genetic selection system in a thymineless growth media, they were able to confirm a TS+ phenotype in TS knockout cells which were transformed with the intein over a notable temperature range when thyronine (T3) or its synthetic analog where present. (20)
Within the same work, the group created alternative chimeric assemblies to evaluate additional criteria and met continued success. For example, a tripartite fusion combining a maltose-binding protein, the ΔI-SM, and a DNA binding domain was successfully utilized in a SDS-PAGE assay to visualize the splicing activity. Their results suggest that a general approach to small-molecule controlled inteins is feasible though success using a scFv rather than TR is not implied. (20)
Inteins and Antibodies as an iCCAR
The idea of fusing a scFv-loaded intein into the apoptotic effector may provide the ideal configuration for the iCCAR. The V266E, D3A Caspase-3 mutant was found to be a suitable host protein for such a construct thus providing the rational for proposing it here as the ideal first model to explore (19). The proposed intein insertion is at the start of the dimeric interface at E266. The next amino acid, which would be the first in the c-terminal extein, is Serine thus satisfying the criteria for an efficient splicing host.
The end goal is the insertion of a scFv into an engineered intein that will then be fused to the Caspase-3 mutant thus rendering it non-functional. Upon encountering its tumor-specific ligand the resultant binding will be a conformational change that induces the intein self-excision and extein ligation. At this point the reassembled, constitutively active Caspase-3 mutants within the malignant cell will regain their ability to act on their substrates thus inducing apoptosis. The combination of directed evolution and rational protein design will provide not only a stabilization method but also a contingency in the event that the intein does not perform as desired.
When I first started researching for the iCCAR I became aware of a number of options available for transmembrane transport. Within reasonable range, the size of the conjugate did not appear a limiting factor and there are options which favor delivery into malignant cells (21). Toxicity was not reported as resulting from any of the transporters I reviewed, so the only criteria I had left to satisfy was the subcellular localization of the cargo.
Upon learning about the successful testing of a broad spectrum anti-viral therapeutic localized to the cytoplasm I focused on the transporter used in their chimeric protein. The approach, dubbed DRACO or Double-stranded RNA Activated Caspase Oligomerizer, is essentially a sensing region tied to a Caspase effector region and therefore similar to the iCCAR at least conceptually. The delivery mechanism for this highly effective and potentially revolutionary therapeutic that yielded the most favorable results was a polyarginine, or Guanidinium-Rich Transporter (22).
Guanidinium Rich Transporters
Paul Wender and his colleagues in the Departments of Chemistry and Chemical and Systems Biology at Stanford University made the realization that the uptake capability of the Tat nonamer was a result of its guanidinium ions. This, in turn, led them to develop and test a range of effective transporters all containing guanidinium thus termed Guanidinium- Rich Transporters. Turning from a peptide only paradigm, they developed a range of transporters including oligocarbamates, dendrimers, and peptoids that are not only effective, but are resistant to proteolysis (23).
An interesting revelation in their 2008 review presenting their work is not only the range of variability in viable transporter structure and composition, but also in the mechanism of cellular uptake involved. That is, not just the observed transport of the Tat 9-mer as opposed to that of octaarginine, but the same transport with cargo of varying size and conformation was shown to affect the mechanism of uptake and even whether or not it occurs at all. They list the example of an octaarginine attached to fluorescin which achieves rapid uptake, yet the same transport with a polycarboxylate will not be transported (23). What seems most likely is that there is actually more than one transport mechanism operating simultaneously on the same transporter/cargo combination.
There have been multiple efforts put forth to elucidate the internalization mechanism of the GRTs (23). At first glance the inconsistency and disparity in the results seem discouraging, however, when it is considered with the conjecture that multiple membrane transport system may be simultaneously engaged, even with a single type and cargo, they become rather informative.
As the molecule approaches the cell membrane the positively charged guanidinium head groups are drawn to negatively charged motifs in polar membrane constituents such as phospholipids and heparan sulfate proteoglycans. The rigid planer array of hydrogen bond donors in the head groups allow for the formation of bidentate (two different sites) hydrogen bonds. This aligns with experimental data showing that as the number of arginines in the transport increases, so does the rate of uptake. Increases to a point, of course, since adding too many would trap the transporter at the membrane; 15 were found to be optimal (23).
Given this and the electrostatic forces acting on the membrane constituents, the need for proper spacing of the headgroups and flexibility in the scaffold become apparent. Electrostatic repulsion along with non-uniform composition will ensure that the anionic molecules are not only not adjacent, but stochastic in their distribution (23). And herein lies the most interesting feature of this class of transporter; controlled changing the substrate and structure of the GRT lead to the ability to select the membrane transport mechanism and even the subcellular destination. The mechanisms observed to contribute include clathrin and caveolae mediated, macropinocytosis, receptor mediated, and the non-endocytotic process of direct uptake (23). Control over subcellular localization would provide not only an increase in the probability of the iCCAR encountering its target, but may also serve to decrease the probability of toxicity and premature protein metabolism.
The Dendrimeric GRT will be bound to the scFv by a (thiol-based) disulfide bond. This choice will ensure that the transporter releases from the iCCAR upon entry into the cell while maintaining the integrity of the bond while it is in ECF. The mechanism making this possible is the reversible yet stable covalent link in a disulfide bond created as a result of the oxidation of two sulfhydryl groups contained in cysteines. Disulfide bonds formed by oxidation are stable in the oxidative environment of the ECF, however the reverse is true in the reductive environment found in many subcellular areas including the cytoplasm (24).
The oxidative versus reductive environments found in different compartments within the cell are primarily the result of the action of Glutathione and, more specifically, the enzymatically maintained ratio between Glutathione (GSH) and Glutathione Disulfide (GSSG), the tripeptide in its reductive and oxidative states respectively. Glutathione and thioredoxin reductase have been shown to be responsible for the redox state found in both the cytosol, Endoplasmic Reticulum, and in some cases; endosomes. In the ER and the endosomes, Protein Disulfide Isomerase (PDI) is believed responsible for maintaining the GSH, GSSG ratio necessary for the oxidative environment, though its presence in endosomes may be cell-type dependent (24).
Dendrimeric Backbone, Destination Varied by Protein Target
Two factors led to the selection of the Dendrimeric GRT as the first to be evaluated. First, given the combined effects of spacing, guanidinium head group quantity, and cargo parameters, the Dendrimeric scaffold seemed to offer the most flexibility (25) (23). Second, it has been shown that by simply varying the length of the alkyl spacer, changes in the destination of the transport could be achieved thus offering the possibility of only a single variable to vary to achieve the targeting of the desired subcellular compartment (25).
In one study, two transport variants, delivered the same cargo, one to the nucleus and the other to the cytosol. The GRT were non-peptidic, Newkome-type dendrimers varied only by the length of the alkyl spacer and strong subcellular localization was observer in both fixed and live fibroblasts and human microvascular endothelial cells (25).
While all proto-oncogenes I examined for targetable hotspots thus far are functionally localized to the nucleus or nucleolus, the probability that targetable mutant proteins localize elsewhere is high. This makes the dendrimeric structure optimal given that it minimizes the change necessary to accommodate a change in the desired localization of the iCCAR.
CAR Analogous Design
Figure 2 – iCCAR configured with specificity for the Nucleophosmin mutant
The chimeric protein will consist of four regions; the transporter (Dendrimeric GRT), the scFv, the Autoproteolytic Linker (ApL), and the effector (V266E, D3A mutant Caspase-3). Two events will be required for the successful operation of the iCCAR in the intracellular environment. First, the transporter must be discarded upon entry into the cell to free the scFv to bind its agonist. The second event is a conformational change in the ApL resulting in the cleavage of the DICA-Caspase-3 disulfide bond, occurring when this binding takes place, resulting in the release of the Caspase-3 mutant in its uninhibited, active state.
The makeup of this alternative design can be generalized to an Intracellular Chimeric Antigen-Specific Signaling Protein (ICASSP). For example, the scFv can be replaced by another “detection” domain such Protein Kinase R (PKR1-181) used in DRACO (22), the Caspase-3 mutant replaced with a reprogrammed endonuclease. I am currently looking at how granzyme-A might be used in place of the Caspase-3 as an alternative configuration.
The granzymes are a homologous family of serine esterases that have been shown to activate at least three distinct apoptotic pathways (26). The possibility of working from its active site, or even that of Caspase 3 present alternatives that may provide a simpler molecular mechanism, a smaller polypeptide, and depending on test observations, an effector region with a lower probability of off-target toxicity.
Autoproteolytic Linker and DICA
It has been shown that active WT Caspase-3 is inhibited when DICA (2-(2, 4-Dichlorophenoxy)-N-(2-mercapto-ethyl)-acetamide) is bound to C264 of its small subunit (16). If the V266E, D3A Caspase-3 mutant is similarly affected it may be possible to utilize DICA as part of a switch/release mechanism to control the mutant. The first two mechanisms that came under consideration depend on whether the allosteric inhibitor DICA is bound in process to the Caspase or as part of the ApL itself. In either case, the scFv agonist binding will induce a conformational change in the ApL that will act on or through DICA releasing a constitutively active Caspase-3 into either the cytosol or the nucleus.
Caspase-3 has as its direct substrates the molecules that carry out apoptosis. Inhibitor of Caspase Activated DNAse (iCAD) is cleaved from CAD freeing it to carry out its function which is the degradation of chromosomal DNA within the nuclei causing chromatin condensation. Another target of the active Caspase-3 will be Gelsolin which normally serves as a nucleus for actin polymerization. Once cleaved the Gelsolin fragments will, in turn, cleave the cytoskeleton and intracellular transports thus structurally destroying the cell. (13)
It should be noted here that building the Linker around the Caspase-8 active site was considered. This moiety would remain concealed until epitope ligation in a manner similar to Death Effector Domains. At this point the iCCAR will interact with the Caspase-3 constitutively present in the cytosol of all somatic cells, activating it to induce apoptosis. The reason was abandoned is that the WT Caspase-3 is subject to the inhibitory signals of the IAP (Inhibitors of Apoptotic Proteases) family of BIR domain containing proteins. The V266E mutant has been shown to be immune to its signal and therefore may be uninhabitable (16).
Another possibility is the engineering of a V266E mutation inducing domain (MID) into the linker that will bind to the C264 on the small subunit and cause the same conformational change as the SNPs in the D3A, V266E mutation. This option, however, may be significantly more complex requiring the induction and control of more conformational states.
Split Caspase-3 Mutant and Zinc Finger Proteins
Another approach that may yield more targeting options and greater specificity would be to utilize a split mutant Caspase-3 with Zinc Finger Protein (ZFP) mediated, sequence specific reassembly. In this approach, unique marker DNA sequences would be directly targeted to achieve malignant cell specific effector function rather than relying on a gene product.
While this approach will not be fully discussed here, pioneering work done by Wataru Nomura and Carlos Barbas at the Scripps Research institute has already demonstrated the feasibility and confirmed the specificity of a ZFP based reassembly model. Attempting to achieve site specific DNA methylation using a split methyltransferase and ZFP targeting, their goal was to achieve directed site specificity where previous attempts that relied upon the targeting potential of the intact methyltransferase. They not only achieved site specific methylation, they did so, for the first time in a living cell. (27)
Given that ZFPs can be engineered to bind to virtually any DNA sequence (27), the potential for targeting unique, yet conserved mutations in a given cancer genome is exciting. I have already begun exploring this option, but the investigation will not be complete in time to be included here.
Given the successes achieved over the preceding decades in recombination, sequence reassembly, protein splicing, and other methods of chimeric protein generation along with the fitness gains resulting from Directed Evolution, the time seems right to undertaking the work proposed here. In fact, given the apparent modularity of protein domains and the ability to alter function and substrate specificity, the time seems right for yet another revolution that may see proteins of our own design challenging nanotechnology for future dominance in the world of therapeutic molecular biology. The paradigm appears to be shifting from searching for those things that will positively affect a biological system to changing that system to achieve a positive effect.
Everything that has been presented here has been demonstrated as possible, albeit in different forms and to different ends. Even within the CAR Analogous Design option, with the exception of the ApL, all molecular components proposed for inclusion in that iCCAR arrangement have been well described, tested, and even proven in approved drugs. For example, guanidinium rich transporters have been extensively studied over the past decade, found their way into at least two clinical trials, and have been demonstrated effected as recently as 2011 in MIT’s landmark work on a Broad Spectrum Antiviral Treatment (23) (22). Further, disulfide bonds are used to understand protein folding (28) and the wealth of lab and clinical data on mAb is well known. Even the ApL, while I have just begun to formulate the specifics, it appears finding a natural sourced domain capable of producing the he necessary conformational change is likely.
Two key areas as of yet unexplored are the efficacy of the iCCAR with regard to tissue penetration and the development of an extensive set of protocols to demonstrate that it can be delivered and exist within healthy cells and tissues without the potential for off-target toxicity.
1. Specific activation and targeting of cytotoxic lymphocytes through chimeric single chains consisting of antibody-binding domains and the y or C subunits of the immunoglobulin and T-cell receptors. ZELIG ESHHAR, TOVA WAKS, GIDEON GROSS, AND DANIEL G. SCHINDLER. 1993, Proc. Natl. Acad. Sci. USA, pp. Vol. 90, pp. 720-724.
2. Chimeric Antigen Receptor–Modified T Cells in Chronic Lymphoid Leukemia. June, Carl, et al., et al. 8, s.l. : Massachusetts Medical Society, August 2011, The New England Journal of Medicine, Vol. 365, pp. 725-33.
3. Case Report of a Serious Adverse Event Following the Administration of T Cells Transduced With a Chimeric Antigen Receptor Recognizing ErbB2. Morgan, Richard A, et al., et al. 4, 2010, The American Society of Gene & Cell Therapy, Vol. 18, pp. 843–851.
4. Cytostatic Drugs, Neuregulin Activation of ErbB Receptors, and Angiogenesis. Hedhli, Nadia and Russell, Kelly Strong. 2010, Current Hypertension Reports, pp. 411-417.
5. Long-term in vivo survival of receptor-modified syngeneic T cells in patients with human immunodeficiency virus infection. Walker, Robert E., et al., et al. 2, 2000, Blood, Vol. 96, pp. 467-474.
6. A Herceptin-Based Chimeric Antigen Receptor with Modified Signaling Domains Leads to Enhanced Survival of Transduced T Lymphocytes and Antitumor Activity. Zhao, Yangbing, et al., et al. 183, 2009, Journal of Immunology, Vol. 2009, pp. 5563-5574.
7. Safer CARS. Heslop, Helen E. 4, s.l. : Nature, 2010, Molecular Therapy, Vol. 18, pp. 661–662.
8. Oncogene addiction: setting the stage for molecularly targeted cancer therapy. Sharma, S. V. and Settleman, J. 24, Genes & development, Vol. 21, pp. 3214-31.
9. c-Myc hot spot mutations in lymphomas result in inefficient ubiquitination and decreased proteasome-mediated turnover. Bahram, Faud, et al., et al. 6, s.l. : The American Society of Hematology, March 15, 2000, Blood, Vol. 95, pp. 2104-2110.
10. Sequence and structure signatures of cancer mutation hotspots in protein kinases. Dixit, A., et al., et al. 10, 2009, PloS one, Vol. 4, p. e7485.
11. c-Myc Target Genes Involved in Cell Growth, Apoptosis, and Metabolism. Dang, Chi V. 1, s.l. : Molecular and Cellular Biology, 1999, Molecular and Cellular Biology, Vol. 19, pp. 1-11.
12. A monoclonal antibody against mutated nucleophosmin 1 for the molecular diagnosis of acute myeloid leukemias. Gruszka, A. M., et al., et al. 12, 2010, Vol. 116, pp. 2096-102.
13. Apoptosis: A Review of Programmed Cell Death. Elmore, Susan. 495, s.l. : SAGE, 2007, Toxicologic Pathology, Vol. 35.
14. Studies of the molecular mechanism of caspase-8 activation by solution NMR. Keller, N, Grutter, M G and Zerbe, O. 17, s.l. : Macmillan Publishers Limited, 2010, Cell Death and Differentiation, Vol. 2010, pp. 710-718.
15. XIAP discriminates between type I and type II FAS-induced apoptosis. Jost, Philipp J, et al., et al. 20, s.l. : NPG, August 2009, Nature, Vol. 460, pp. 1035-1040.
16. A constitutively active and uninhibitable caspase-3 zymogen efficiently induces apoptosis. Walters, Jad, et al., et al. 424, s.l. : Biochemical Journal, 2009, pp. 335-345.
17. Designing switchable enzymes. Ostermeier, M. [ed.] 442-448. 4, s.l. : Elsevier, 2009, Current opinion in structural biology, Vol. 19.
18. Exploring protein fitness landscapes by directed evolution. Romero, P. A. and Arnold, F. H. 12, s.l. : Nature Publishing Group, 2009, Nature Reviews Molecular Cell Biology, Vol. 10, pp. 866-876.
19. Inteins, valuable genetic elements in molecular biology and biotechnology. Elleuche, S. and Poggeler, S. 2, s.l. : Springer, 2010, Applied microbiology and biotechnology, Vol. 87, pp. 479-489.
20. Regulation of protein activity with small-molecule-controlled inteins. Skretas, G. and Wood, D. W. 2, s.l. : Wiley Online Library, 2005, Protein Science, Vol. 14, pp. 523-532.
21. Folate-targeted therapeutic and imaging agents for cancer. Low, P. S. and Kularatne, S. A. 3, s.l. : Current opinion in chemical biology, 2009, Vol. 13, pp. 256-62.
22. Broad-Spectrum Antiviral Therapeutics. TH, Rider, et al., et al. 2011, PLoS ONE.
23. The design of guanidinium-rich transporters and their internalization mechanisms. Wender, P. A., et al., et al. 4-5, 2008, Advanced drug delivery reviews, Vol. 60, pp. 452-72.
24. Drug delivery strategy utilizing conjugation via reversible disulfide linkages: role and site of cellular reducing activities. Saito, Go, Swanson, Joel A. and Lee, Kyung-Dall. 2, 2003, Advanced Drug Delivery Reviews, Vol. 55, pp. 199-215.
25. Dendritic Molecular Transporters Provide Control of Delivery to Intracellular Compartments. Huang, Kui, et al., et al. 2, 2007, American Chemical Society, Vol. 18, pp. 403-409.
26. Death by a thousand cuts: granzyme pathways of programmed cell death. Chowdhury, D. and Lieberman, J. 2008, Annual review of immunology, Vol. 26, pp. 389-420.
27. In vivo site-specific DNA methylation with a designed sequence-enabled DNA methylase. Nomura, W. and Barbas III, C. F. 28, s.l. : ACS Publications, 2007, Journal of the American Chemical Society, Vol. 129, pp. 8676-8677.
28. Disulfide Bonds and Protein Folding. Wedemeyer, William J., et al., et al. 39, s.l. : American Chemical Society, 2000, Biochemistry, pp. 4207-4216.
29. Can Antitumor Immunity Help to Explain “Oncogene Addiction”? Restifo, Nicholas P. November, s.l. : Cell Press, November 18, 2010, Cancer Cell, Vol. 18, pp. 403-405.
30. Pluripotency Redux — Advances in Stem-Cell Research. Gearhart, John, Pashos, Evanthia E and Prasad, Megana K. 15, s.l. : Massachusetts Medical Society, October 2007, The New England Journal of Medicine, Vol. 357, pp. 1469-1472.
31. Engineering allosteric protein switches by domain insertion. Ostermeier, M. 8, 2005, Protein engineering, design & selection : PEDS, Vol. 18, pp. 359-64.
32. Structural features of the reactions between antibodies and protein antigens. Braden, B. C. and Poljak, R. J. 1, s.l. : FASEB, 1995, The FASEB journal, Vol. 9, pp. 1-9.