Novel peptides from the RAS-p21 and p53 proteins for the treatment of cancer

Summary

We have employed a novel computer-based molecular modeling method to design peptides from the ras-p21 and p53 proteins that block proliferation of cancer cells. The rationale of our approach is to identify peptide domains from each protein that alter conformation in response to oncogenic amino acid substitutions in their polypeptide chain. We accomplish this by first generating and comparing low energy average structures for oncogenic and wild-type proteins using conformational energy calculations. Peptides are then synthesized corresponding to these domains. These domains are then linked to a trans-membrane-penetrating sequence (called penetratin) and tested against cancer and untransformed cell lines. Remarkably, we have found that two ras-p21 peptides, 35–47 and 96–110, called PNC-7 and PNC-2, respectively, can induce phenotypic reversion of ras-transformed TUC-3 pancreatic cancer cells and ras-transformed HT1080 human fibrosarcoma cells to their untransformed phenotypes. Moreover, both peptides were found to be cytotoxic to ras-transformed human MIA-PaCa-2 pancreatic carcinoma cells and human U-251 astrocytoma cells. Importantly, these peptides have no effect on the growth of their normal cellular counterparts. We have also synthesized peptides from the p53 protein corresponding to its hdm-2-binding domain sequences (residues 12–26), also linked to the penetratin sequence. Surprisingly, we have found that these peptides induce 100 percent tumor cell necrosis, not apoptosis, in 13 different human cancer cell lines but have no effect on normal pancreatic acinar cells, breast epithelial cells, and human stem cells. Moreover, these peptides are cytotoxic to TUC-3 pancreatic tumor cells in nude mice plus eradicate these tumor cells when administered at sites near these tumors. These novel peptides appear to hold much promise as new, non-toxic anti-cancer agents.

I. Introduction

Over the past several years, our group has modeled three-dimensional structures of proteins that are involved in regulation of the cell cycle. Two proteins that have received widespread attention are ras-p21 and p53. We summarize our experience designing peptides from these proteins that appear highly effective in the selective blocking of the proliferation of cancer cells.

The ras gene encodes the ras-p21 protein; the first oncogene shown to cause human cancer (Barbacid, 1987). Remarkably, only single base changes in this gene at just one of several positions results in this protein becoming transforming. These single base changes result in the encoding of ras-p21 proteins that contain single amino acid substitutions at critical positions in the polypeptide chain (Stacey and Kung, 1984; Barbacid, 1987). One of the most important of these substitutions occurs when Val-12 replaces the wild-type Gly-12. Importantly, proof of principle was demonstrated when both wild-type and Val 12-substituted ras-p21 were bacterially over-expressed and micro-injected into NIH 3T3 cells. Val 12-p21, but not its wild-type counterpart protein induced cell transformation, further suggesting mutated ras-p21 protein as the actual transforming agent (Stacey and Kung, 1984).

To further study ras-p21, we utilize a Xenopus laevis oocyte model. This oocyte expresses insulin receptors; importantly, insulin induced oocyte maturation requires activation of endogenous normal cellular ras-p21 (Deshpande et al, 1987). In this system, oncogenic, but not wild-type, ras-p21 induces maturation (completion of meiosis) of Xenopus laevis stage VI, metaphase-arrested oocytes in the second meiotic division (Birchmeier et al, 1985). Consequently, these oocytes present a unique surrogate to study mechanism as well as discriminate between agents that affect oncogenic vs. wild-type ras-p21 (Chung et al, 1992).

There are now two identifiable positions in the polypeptide chain of ras-p21, Gly 12 and Gln 61, both when substituted result in generating potent transforming proteins (Barbacid, 1987). Approximately one of every three human cancers has been found to contain a ras gene that encodes for one or another ras-p21 protein that has substitutions at either one of these two critical positions (Bos, 1989; Pincus and Brandt-Rauf, 2006). Of particular interest in this regard, over 90 percent of pancreatic cancers contain the k-ras-p21 protein containing Val in place of Gly 12 (Almoguerra et al, 1988); similarly over 50 percent of colon cancers contain this substituted k-ras p21 protein (Forester et al, 1987). Clearly, it would be of enormous value to design agents that could potentially block the actions of these oncogenic forms of ras-p21 while allowing its wild-type counterpart protein to function normally. As we will discuss in this communication, based on our molecular modeling studies of ras-p21, we have been able to design a set of peptides that achieve this objective, i.e., block oncogenic ras-induced cell proliferation but do not affect normal cellular growth (Pincus, 2004).

In contrast, p53 is an anti-oncogene protein, i.e., it blocks mitogenic signals in the nucleus. This protein contains three major domains: a transactivating domain involving residues 1–92, the DNA-binding region incorporating residues 93–312, and a tetramerization domain utilized in forming p53 tetramers that are the most active in blocking unrestricted cell proliferation (Laptenko and Prives, 2006). As with ras-p21, amino acid substitutions at critical positions in the polypeptide chain of p53 result in an “oncogenic” protein. However, unlike ras-p21, these substitutions inactivate p53 so that it can no longer counter uninterrupted cell proliferation signals. When p53 becomes activated in cancer cells, it activates apoptosis or programmed cell death by inducing expression of pro-apoptotic proteins and/or by inducing their activation (Laptenko and Prives, 2006). These proteins include annexin V, Bax, waf^p21 and caspases. The end result is programmed cancer cell death. It is likewise desirable, therefore, to devise ways of activating wild-type p53 protein in cancer cells to induce them to undergo apoptosis. This, of course, is dependent on the presence of wild-type p53 in cancer cells. However, in a wide variety of cancer cells, the p53 gene has either been homozygously deleted or mutated so that no wild-type p53 exists within these cancer cells. As will be discussed in this paper, we have also been able to design peptides derived from p53 that induce cancer cell death irrespective of the status of p53 in cancer cells.

II. Anti-oncogenic ras peptides from RAS-p21 protein

A. Mechanism of ras-p21 mitogenic signaling

In humans, the ras-p21 protein contains 189 amino acid residues and is known to be a G-protein, i.e., it becomes activated when it releases GDP and binds GTP in a nucleotide exchange process. Much is understood about the sequence of events that are then initiated by ras-p21 in cells. First, as summarized in Figure 1, ras-p21 protein must be bound to the inner cell membrane via its Cys 186 residue which becomes linked as a thioether to a farnesyl moiety promoted by the enzyme, farnesyl transferase (Almoguerra et al, 1988). To induce mitosis, therefore, it must activate other proteins in a chain of events that ends in the nucleus, through a mitogenic signal transduction pathway (Bos, 1989).

Figure 1.

Summary of known actions of activated ras-p21 in cells, beginning (top, left) when a growth factor binds to its cell receptor. This results in activation of intracytoplasmic tyrosine kinase (TK) activity on the receptor that binds to the adapter grb-2 protein that concurrently binds to and activates the guanine nucleotide exchange factor (GNEF), SOS. This, in turn, induces ras-p21 to exchange GDP for GTP, resulting in its activation. ras-p21 is bound to the inner cell membrane by a covalent farnesyl moiety attachment in thioether linkage to Cys 186. In its GTP-bound form ras-p21 binds to a number of other target proteins. One of these is GAP (GTPase activating protein) that induces GTPase activity in ras-p21 resulting in hydrolysis of GTP to GDP cycling ras-p21 into the inactive state. Both GAP and SOS may also be involved as a target in ras signaling, hence the question mark below the GAP box. Activated ras-p21 also binds to and activates raf (top middle of the figure), a 74 kDa Ser/Thr kinase protein that, in turn, activates the kinase cascade in which it activates MEK that finally activates ERK (MAP-2K in the figure) that is involved in cytoskeletal rearrangements in the cytosol and which shuttles into the nucleus where it activates the nuclear transcription factor, fos, that forms a heterodimeric complex with the nuclear transcription factor, jun. The latter protein is activated by another kinase, jun-N-terminal kinase (JNK), that occurs on a separate pathway called the stress-activated protein (SAP) pathway. As explained in the text, oncogenic ras-p21 directly activates JNK/jun that circumvents the normal, regulated wild-type ras-p21-activated pathways (right side of the figure). ras-p21 also interacts directly with phosphoinositol-3-hydroxy kinase (PI3K, left side of the figure) and induces activation of phospholipase C (PLC); both of these proteins cause increases in the second messenger molecule, inositol triphosphate (IP₃) and diacylglycerol (DAG); the former induces calcium mobilization while the latter induces activation of protein kinase C (PKC) that is especially critical to the oncogenic ras-p21 pathway. In the nucleus, the fos-jun complex, also called AP1, induces transcription of many pro-mitogenic proteins including cyclins and possible the nuclear skeletal proteins called nuclear matrix proteins (NMP’s); other nuclear proteins, like myc, also transcriptionally active, are also often activated in this process. Anti-oncogene proteins, such as p53, also become activated. This protein blocks transcription of pro-mitotic proteins and induces apoptosis in transformed cells.

Once bound to the inner cell membrane, ras-p21 can become activated via GDP/GTP exchange. This process is induced by the binding of a growth factor, such as epidermal growth factor (EGF), insulin, etc, to its respective transmembrane growth factor receptor, resulting usually in receptor dimerization. This results in activation of an intracytoplasmic tyrosine kinase that is part of the growth factor receptor. This receptor then binds to an adapter protein called grb-2 that concurrently binds to a critical protein, called SOS, a guanine nucleotide exchange promoter protein. Activated SOS binds directly to ras-p21, promoting the exchange of GDP for GTP and activation of ras-p21.

These events are modulated by the protein, GTPase activating protein or GAP. This protein binds to ras-p21 such that it enhances endogenous ras-p21 GTPase activity, resulting in hydrolysis of GTP to GDP (Figure 1). Thus SOS and GAP are thought to control the intracellular levels of activated ras-p21 at any time (Barbacid, 1987). It is also important to note, though, that, in addition to their regulatory roles, both GAP and SOS proteins are very likely targets of ras-p21 in mitogenic signaling (Nimnual and Yatsula, 1988; Yang and Widmann, 2001).

Once activated, ras-p21 activates a number of critical proteins, perhaps the most important of which is raf. This protein becomes activated when it binds to ras-p21 in the cell membrane and, in turn, activates a set of phosphorylation cascade reactions. In this cascade, activated raf-p74 directly activates mitogen (extracellular) kinase or MEK that, in turn, activates mitogen-activated protein kinase (MAPK) or extracellular mitogen response kinase (ERK). This critical protein promotes cytoskeletal rearrangements for mitosis and shuttles between the cytosol and nucleus. In the nucleus, it activates the all-important protein, fos, which forms a hetero-dimeric complex with another critical nuclear protein, jun, to form the AP1 complex that strongly promotes transcription of mitosis-promoting proteins (Barbacid, 1987).

Activated ras-p21 also induces activation of a number of other critical intracellular proteins, in particular, phospholipase C-γ (Smith et al, 1990) and phosphoinositol-3-hydroxy kinase (PI3K), the latter of which binds directly to ras-p21 (Chung et al, 1992) both activating a number of vital second messenger molecules such as diacylglycerol that activates protein kinase C (PKC), critically important for oncogenic ras-p21 action, and inositol triphosphate (IP₃) that has multiple activation targets including mobilization of calcium ions for various mitotic processes such as spindle formation.

B. Three-dimensional structures of ras-p21

In the x-ray crystallographic structure of wild-type ras-p21 bound to GTP analogues (Krengel et al, 1990), the guanine ring has multiple contacts with ras-p21 residues in the 119–125 region. The backbone NH of Gly 12, appears to interact with the bridging oxygen of the γ-phosphate residue and is thought to aid in the departure of the terminal phosphate leaving group (Krengel et al, 1990). Ser 17 and Ala 59 both interact with a magnesium ion that interacts with the terminal phosphate residue of GTP (Krengel et al, 1990). In some substituted ras-p21 proteins this alignment is disrupted, reducing the GTPase activity of ras-p21, suggesting a mechanism for prolonged ras-p21 activation (Krengel et al, 1990). Many of these interactions are not present in the x-ray structure for ras-p21 bound to GDP.

Another important consequence of binding GTP in place of GDP is the motion of an important effector domain, called the switch 1 domain, consisting of an exposed β-sheet involving residues 32–47 that makes contacts with all of the known targets of ras-p21, including raf, GAP, SOS and PI3K. Surprisingly, there are no consensus sequences in these four proteins for binding to this ras-p21 domain, and, in the x-ray structures for ras-p21 bound to these proteins, there are substantially different interactions between the switch 1 domain and the binding determinants for each complex (Nassar et al, 1995; Scheffzek et al, 1997; Margarit et al, 2003).

A second important ras-p21 binding domain, called the switch 2 domain, involves residues 55–71, and makes multiple direct contacts with GAP and SOS proteins (Moodie et al, 1993; Nassar et al, 1995; Scheffzek, K., Ahmadian et al, 1997; Margarit et al, 2003). This domain contains the critical Gln 61 residue, discussed above. A third domain, involving residues 102 and 103, of an exposed loop that also makes important contacts with SOS and, as discussed below, with JNK and jun proteins (Scheffzek et al, 1997; Margarit et al, 2003).

III. Design of anti-oncogenic RAS peptides

A. Oncogenic amino acid substitutions in ras-p21 induce changes in structure of effector domains

An important point of departure in designing anti-oncogenic ras peptides is the finding that substitution of any naturally occurring non-cyclic L-amino acid for Gly 12 or almost any amino acid residue for Gln 61 results in an oncogenic ras-p21 protein. This suggests that these amino acid substitutions cause changes in the three-dimensional structure of ras-p21 that result in permanent activation.

An alternate explanation for how amino acid substitutions in ras-p21 cause the protein to become oncogenic relates to rates of hydrolysis of GTP that activates ras-p21 when bound to it. Substitutions of amino acids for Gly 12 and for Gln 61 cause large rate reductions in GAP-induced hydrolysis of GTP (Krengel et al, 1990). This can result in prolonged activation of ras-p21. In the structure of wild-type ras-p21 bound to GTP analogues, the backbone NH of Gly 12 and the carboxamido NH of the Gln 61 side chain appears to interact with the β-γ bridging oxygen of the γ-phosphate moiety of GTP and is thought to aid in the departure of the terminal phosphate leaving group (Krengel et al, 1990). In complexes of substituted ras-p21 proteins and GTP analogues, this alignment is disrupted, suggesting a mechanism for prolonged ras-p21 activation (Krengel et al, 1990; Scheffzek et al, 1997).

However, this observation cannot explain the effects of a number of substitutions on ras-p21 activity. For example, substitution of Pro for Gly 12 is the only non-oncogenic substitution that occurs at this position (Barbacid, 1987). Interestingly, Pro 12 has no NH backbone atom that can contribute to GTP hydrolysis. On the other hand, substitution of Gly for Gln 61 is oncogenic but the rate of GTP hydrolysis of this substituted protein is not decreased (Krengel et al, 1990). Other mutated ras-p21 proteins, such as D38E (a Glu-for-Asp substitution), binds strongly to GTP and to GAP, does not undergo hydrolysis, and does not transform cells (Krengel et al, 1990).

Perhaps most convincingly, a triply-substituted ras-p21 protein with Val for Gly 10, Arg for Gly 12, and Thr for Ala 59 has been found to cause cell transformation, but this protein does not bind either GDP or GTP (Clanton et al, 1987). On the other hand, a similar protein with Gly 10, Arg 12, Val 15 and Thr 59 neither transforms cells nor binds nucleotide (Clanton et al, 1987). Since these proteins do not interact with nucleotides, differences in the rates of hydrolysis of GTP cannot explain the differences in the transforming activities of these two proteins. Thus it is plausible that the different activities of these two proteins are caused by structural differences between them.

In the original studies that demonstrated that activated ras-p21 binds directly to raf, it was found that wild-type ras-p21 bound to GDP did not bind to raf. In contrast, Val 12-ras-p21 bound to GDP was found to bind to a significant extent to this critical protein (Moodie et al, 1993). This finding further suggests that the G12V substitution in ras-p21 induces a significant change in the structure of ras-p21 enabling it to interact with intracellular targets even without binding to GTP.

Since it appeared that oncogenic amino acid substitutions in ras-p21 induce critical changes in this protein, we set out to identify the domains of ras-p21 that undergo these conformational changes and then synthesize peptides corresponding to these domains to determine if they are capable of blocking the oncogenic forms of ras-ras-p21.

B. Modeling methods

All of our computational methods are based on the assumption that the energy-minimized x-ray crystal structure of a protein is the lowest energy (global) minimum conformation for the given amino acid sequence (Scheraga et al, 2004). As discussed in the preceding section, the x-ray crystal structures for ras-p21 bound to GDP and GTP and with amino acid substitutions have been determined. The x-ray structure is subjected to energy minimization to remove any bad contacts between atoms and to optimize favorable contacts between atoms such as hydrogen bonding.

The energy-minimized x-ray structure occurs in a potential energy well in which other low energy conformations of the protein exist which have the same basic chain fold but which differ in conformation from the energy-minimized x-ray structure in local domains (Pincus, 2004). Our methods sample these conformations and then employ them to compute the average structures, which should be the observed structure in solution, for oncogenic and wild-type ras-p21 complexed with specific target proteins. These structures are then superimposed to determine which domains differ in conformation between the two complexes (Pincus, 2004).

We employ two sampling methods to generate the low energy structures that occcur around the energy-minimized x-ray crystal structure: molecular dynamics (McCammon al, 1988; Pincus, 2004) and the electrostatically-Driven Monte Carlo method (EDMC) (Piela et al, 1987; Ripol al, 1988). The molecular dynamics method samples these alternate conformations by integrating, with respect to time, Newton’s equations of motion. The dynamics trajectory for the system is followed at 300°K for 2 nsec. On the resulting trajectory, the low energy structures around the starting structure are computed (Pincus, 2004). Use of this method for low energy structure generation is described more completely (McCammon al, 1988; Pincus, 2004). For each protein structure or complex, we found that the trajectories that are computed over this time interval are such that the total energy converges to a low, constant value. The structures whose energies have converged are then used to compute the average structure.

The EDMC method samples conformations of the protein in the global minimum potential energy well by determining the electric field of the local energy minimum (Pielaet al, 1987; Ripol et al, 1988). The dipole moments of the backbone CO-NH groups are then examined to determine which dipole or group of dipoles is (are) the least optimally oriented with the field. The dihedral angles of these peptide groups are then changed to orient these dipoles to become aligned with the field, thereby perturbing the structure, and the energy of the resulting structure is minimized again (Pielaet al, 1987; Ripol et al, 1988). This procedure is repeated iteratively until the energy of the protein is lowered no further. This method is referred as the self-consistent electric field (SCEF) (Pielaet al, 1987; Ripol et al, 1988) method and has been extended so that at the end of a set of self-consistent calculations, the structure is randomly perturbed using the Monte Carlo method (Ripol et al, 1988), and the process described above is repeated. In this manner, sets of low energy structures are generated for the given starting structure. The average structure of each protein is then computed as the Boltzmann average of all of the low energy structures computed on the energy minimization “trajectory”(Liwo et al, 1994).

Once the low energy structures have been sampled, we compute the average structures and then superimpose them to detect domains that differ in conformation.

C. Results of molecular modeling studies with oncogenic and wild-type ras-p21

We performed these calculations on wild-type ras-p21 bound to GDP and then on ras-p21 bound to GTP (Monaco 1995a,b). We similarly performed these computations on the two triply-substituted, non-nucleotide-binding ras-p21 proteins and on several GTP-binding oncogenic (e.g., Val 12- and Leu 61-substituted) ras-p21 proteins (Liwoet al, 1994). Remarkably, irrespective of the oncogenic amino acid substitution or the site where it occurred, the same domains of ras-p21 were found to undergo the same changes in structure (Pincus, 2004). These results are summarized in Figure 2 (Pincus, 2004) where we show the superposition of the computed average structures of three activated forms of ras-p21 (blue, G12V-p21; green, Q61L-p21; yellow, wild-type ras-p21 bound to non-hydrolysable GTP) superimposed on the GDP-bound wild-type ras-p21 (purple). As can be seen in this Figure, the overall fold of all four proteins is the same; but the three activated ras-p21 proteins undergo local structural changes in specific domains, viz: 10–16, 35–47, 55–71, 81–93, 96–110 and 115–126. As shown for four of these domains in Figure 2, the average structures in oncogenic and activated wild-type proteins cluster together away from the structure of the inactive form of the wild-type protein. This may best be seen in the view presented in Figure 2 for the 35–47 domain that has been implicated in the binding of ras-p21 to at least four different target proteins.

Figure 2.

Upper. Superposition of the computed average structures for inactive ras-p21 bound to GDP (purple structure in each panel); for Val 12-ras-p21 (right panel, blue); wild-type ras-p21 bound to GTP (middle panel, yellow); and Leu 61-ras-p21 (left panel, green). Lower. Conformations for four ras-p21 domains, residues 10–16, 32–47, 96–110 and 115–126, from the superpositions in Figure 2, Upper, that undergo significant changes in structure in oncogenic and activated ras-p21. The color scheme is the same as for Figure 2, Upper.

One intriguing result of these calculations is the structure of the 10–16 loop domain. In the wild-type GDP-bound form of ras-p21, this loop contains a reverse turn at Ala 11-Gly 12 while in Val 12-p21, the turn shifts to Val 12-Gly 13 (Pincus et al, 1983, 1985). In prior studies, we showed that this results from the ability of Gly 12 to adopt a turn structure that is unavailable energetically to any noncyclic L-amino acid (Pincus et al, 1983). All substitutions of arbitrary L-amino acids for Gly 12 result in this shift in the reverse turn (Pincus et al, 1985). This seemingly small shift in the local structure results in a number of ineffective contacts that are made with the switch 2 domain, that lies across the phosphate binding cleft, that cause it to move to another position, causing the 32–47 effector domain also to move in position (Chen et al, 1989). Exactly the same shift is induced by substitutions at Gln 61. As shown in Figure 2 for the Leu 61 oncogenic amino acid substitution, the 10–16 loop adopts the turn structure at Gly 12-Gly 13 rather than at Ala 11-Gly 12 (green structure) resulting from close interactions between these two domains of ras-p21 (Pincus, 2004).

D. Testing of ras-p21 domain peptides in the xenopus oocyte system

We have synthesized peptides corresponding to each of these domains in ras-p21 and have tested them in the Xenopus laevis oocyte system. In these experiments, we co-inject these peptides together with Val 12-p21 and inject them into oocytes that are then subsequently incubated with insulin. We find that three of the ras-p21 peptides, 35–47 (PNC-7), 96–110 (PNC-2) and 115–126 (PNC-1) all completely block Val 12-p21-induced oocyte maturation but have only partial inhibitory effects on insulin (Pincus, 2004).

An example of this pattern is summarized in Figure 3, that shows that PNC-2 completely blocks Val 12-p21-induced oocyte maturation but has no effect on insulin-induced maturation. We obtained identical results with PNC-1. In contrast, an unrelated, negative control peptide from cytochrome P450, called X-13, had no effect on maturation induced by either agent (Pincus, 2004).

Figure 3.

Effects of PNC-2 on oncogenic ras-p21- and insulin-induced oocyte maturation. The inset explains the individual time curves.

PNC-7 produced partial inhibition of insulin-induced oocyte maturation that reached a plateau at 70 percent inhibition (Chung et al, 1992). This percentage inhibition was achieved independently of the insulin concentration and the extent of maturation, suggesting that insulin-activated wild-type ras-p21 can utilize maturation (meiosis)-inducing pathways that are not blocked by this peptide (Chung et al, 1992).

E. Oncogenic ras-p21 and activated wild-type ras-p21 induce overlapping but distinct signal transduction pathways

Since these three peptides selectively block oncogenic ras-p21 but do not affect insulin-activated wild-type ras-p21, we hypothesized that these two proteins may utilize signal transduction pathways that diverge. In work designed to discover critical intracellular proteins with which oncogenic ras-p21 may interact, we found that oncogenic ras-p21, unlike its wild-type counterpart protein, interacts strongly both with JNK and its target, jun and induces phosphorylation of JNK at least five-fold greater than does its wild-type counterpart protein (Adler et al, 1995, 1996). Since we have been able to prepare cloned, purified ras-p21, JNK and jun proteins, we have established an in vitro binding assay system in which bead-bound JNK is incubated with Val 12-p21 alone and in the presence of each of the synthetic ras-p21 peptides (Adler et al, 1995, 1996).

1. Interaction with JNK is critical for oncogenic, but not wild-type, ras-p21 signal transduction

As shown in Figure 4, bead-bound JNK binds to Val 12-p21 (Adler et al, 1995, 1996). This binding is strongly reduced by the presence of free JNK added to Val 12-p21 and by addition of JNK substrate, jun or its JNK binding 5–89 domain. Of all of the peptides tested in this assay, only two, PNC-2 (ras-p21 96–110 peptide) and PNC-1 (ras-p21 115–126 peptide) reduce this binding to the control levels (Figure 4). These results suggest that both of these peptides block Val 12-p21-JNK interactions, otherwise resulting in blocking a critical step in Val 12-p21 signal transduction. In support of this conclusion, we have found that the dose-response curve for inhibition by PNC-2 of Val 12-p21-induced oocyte maturation superimposes on that for inhibition of PNC-2 of the binding of Val 12-p21 to JNK (Pincus et al, 2000). Similar results pertain to the inhibition of the binding of Val 12-p21 to jun beads by PNC-2 although not by PNC-1 (Adler et al, 1995, 1996).

Figure 4.

Effects of ras-p21 effector peptides, identified from conformational energy calculations, on the binding of Val 12-ras-p21 protein to GST bead-bound p-GEX-JNK. The upper part of the figure explains the experimental protocol. On the left upper scheme, Val 12-p21 is incubated with JNK beads, non specifically-bound protein is washed from the beads, and the beads are then subjected to SDS PAGE and blotted with the monoclonal antibody to ras-p21, Y13–259. If ras-p21 binds specifically to JNK, the blot for ras-p21 should be positive. On the right side of the upper figure, the same scheme is shown except now the JNK beads are now incubated with ras-p21 plus another agent. If the agent competes with ras-p21 for binding to the JNK beads, then the blots with Y13–259 should show absent or diminished amounts of ras-p21. The bottom section of the figure summarizes the results for controls and ras-p21 peptides. When Val 12-ras-p21 is incubated with the JNK beads, a large band is present on the western blot (lane 1), indicating that ras-p21 binds to JNK. This binding is markedly diminished when Val 12-ras-p21 is present with the regulatory domain of jun (residues 5–89) (lane 2) and full-length jun (lane 3). The JNK fusion pGEX-2T protein does not inhibit this binding (lane 4), while JNK itself does (lane 5). Lanes 6–11 show the results from incubation of JNK beads with different ras-p21 peptides (sequence numbers are given in the figure); only ras-p21 115–126 (PNC-1), 96–110 (PNC-2), and 35–47 (PNC-7) peptides are seen to diminish the ras-p21 band to control levels indicating that these peptides interfere with the binding of ras-p21 to JNK.

In addition, we have found that the protein GST-pi strongly and specifically binds to the JNK-jun complex and blocks JNK-induced jun phosphorylation (Adler et al, 1999). Injection of this selective inhibitor with Val 12-p21 into oocytes blocks maturation while injection of this inhibitory protein into oocytes subsequently incubated with insulin has no effect on insulin-induced maturation (Amar et al, 1997; Villafania et al, 1999).

We have further determined the level of JNK phosphorylation in oocytes induced to mature with Val 12-p21 and with insulin by blotting whole cell lysates both for total JNK and for phosphorylated JNK (Ranginwale et al, 2001). We found that the level of total JNK was the same in both cases but that the level of phosphorylated JNK was markedly elevated in the Val 12-p21-matured oocytes whose level increases with increasing levels of maturation. In contrast, it was much lower in the insulin-matured oocytes that plateaus at an early stage and does not correlate with extent of maturation. These and other results suggest that Val 12-p21 utilizes a JNK-jun-dependent signal transduction pathway that is not utilized by the activated wild-type protein.

2. Site of action of PNC-7: oncogenic and wild-type ras-p21 interact with raf differently

We have likewise investigated the site of action of the 35–47 inhibitory peptide, PNC-7. This peptide does not interfere in the interaction of Val 12-p21 with JNK (Adler et al, 1995, 1996). Since it corresponds to the effector domain implicated in binding to the raf p74 protein, as discussed in Sections 1 and 2 above, we have assayed its effects on raf. We have found that c-raf and an oncogenic form of raf that lacks the amino terminal regulatory domain that contains residues 55–131, its ras-binding domain (RBD), called raf-BXB, both induce oocyte maturation (Chie et al, 2000, 2002). We have found that PNC-7 strongly inhibits c-raf induction of oocyte maturation but has no effect on raf-BXB (Chie et al, 2002). These results imply that PNC-7 blocks raf by interacting with its amino terminal regulatory domain. Surprisingly, we find that dominant negative raf blocks both Val 12-p21- and insulin-induced oocyte maturation (Chie et al, 2000). These findings suggest that PNC-7 must block a specific binding mode of oncogenic ras-p21 while allowing wild-type ras-p21 to interact with it (Pincus, 2004).

Overall, our findings suggest that oncogenic and wild-type ras-p21 utilize differing pathways allowing us to inhibit the oncogenic pathway selectively. This suggests that these peptides may block cancer but not normal cell growth.

F. PNC-2 and PNC-7 blocks cancer cell growth

We have tested our two oncogenic ras-selective peptides on ras-transformed tumor cell lines. In addition, we have developed a normal rat pancreatic acinar cell line (called BMRPA1) and a Val 12-ras-p21-induced counterpart pancreatic cancer cell line, called TUC-3, produced by stable transfection of the K-ras oncogene (encoding Val-for-Gly 12-p21) into the BMRPA1 cell line (Bao et al, 1994; Kanovsky et al, 2001).

We introduced these peptides into cells employing two methods: (1) Each peptide was synthesized attached on its carboxyl terminal end to a highly positively charged penetratin or leader sequence from the Drosophila antennapedia protein that enables peptides and proteins to cross cell membranes (Kanovsky et al, 2001, 2003). PNC-2 and 7 attached to the leader sequence are termed PNC-2 (7)-leader, respectively. (2) We also transfected lac-inducible plasmids that encode each of the peptide sequences into TUC-3 cells (Kanovsky et al, 2003). For controls, we employed the X-13 peptide attached on its carboxyl terminal end to the leader sequence, called PNC-29, and synthesized a plasmid that encoded the X-13 sequence.

The results are shown in Figure 5 for PNC-2-leader (Kanovsky et al, 2003; Pincus, 2004). Panel C shows untreated, untransformed BMRPA1 cells while Panel D shows that treatment of these cells with the ras-p21 96–110 peptide has no effect on cell viability or growth. In contrast, Panel A shows untreated TUC-3 pancreatic cancer cells that following treatment for two weeks with ras-p21-Leader peptide, at doses as low as 1 μg/ml, undergo complete reversion to the untransformed phenotype as shown in Panel B (Kanovsky et al, 2003; Pincus, 2004).

Figure 5.

PNC-2 and PNC-7 induce reversion of ras-transformed TUC-3 cells. Panel A shows TUC-3 forming colonies of non-contacted-inhibited cells, in contrast to their untransformed parent BMRPA1 cells shown in Panel C. These cells form contact-inhibited monolayers and show distinct cell boundaries in contrast to the transformed TUC-3 cells in Panel A. As shown in Panel B, after two weeks of treatment (in this figure, with PNC-2), the TUC-3 cells in Panel A are seen to revert to a morphologically untransformed phenotype where the cells now form contact-inhibited monolayers and exhibit distinct cell boundaries. As can be seen in Panel D, two weeks of treatment of untransformed BMRPA1 cells with PNC-2 has no effect on the viability of these cells. Identical results were obtained using PNC-7.

These reverted cells form contact-inhibited monolayers and do not grow when explanted into nude mice for 56 days (Kanovsky et al, 2003). In striking contrast, PNC-29 negative control peptide-treated TUC-3 cells grow rapidly when explanted into nude mice and metastasize after 3 weeks. Identical results were obtained with PNC-7-leader except that full phenotypic reversion required minimal concentrations of 125 ug/ml (Kanovsky et al, 2003).

Moreover, full phenotypic reversion was obtained when TUC-3 cells were transfected with plasmids encoding the ras-p21 35–47 and 96–110 sequences and selected on G418 media. In contrast, TUC-3 cells transfected with the Lac-inducible plasmid encoding the X13 sequence remained transformed (Kanovsky et al, 2003).

Our findings that the ras-p21 35–47-Leader and 96–110-Leader peptides, but not the X-13-Leader control, induce TUC-3 cell phenotypic reversion and that none of these three peptides have any effect on the growth of BMRPA1 cells suggests that these two ras-p21 peptides are specific for transformed cells. Furthermore, since TUC-3 cells underwent identical phenotypic reversion when transfected with the lac-inducible plasmids encoding either ras-p21 sequence but not with the X13 sequence, we conclude that the ras peptide itself induces phenotypic reversion independently of the presence of the leader (penetratin) sequence (Pincus, 2004; Kanovsky et al, 2003).

A. PNC-2 and PNC-7 block human cancer cell growth

Recently, we have tested both PNC-2-leader and PNC-7-leader on four different human cancer cell lines, three of them ras-transformed, i.e., HT1080 fibrosarcoma, SW620 colon cancer and MIA-PaCa-2 pancreatic cancer cell lines and one non ras-transformed U-251 astrocytoma cell line. The latter expresses high levels of JNK, raf, MEK and MAPK (Adler et al, 2005). In work to be published, we find that both peptides induce dramatic blockade of cancer cell growth in the first two cell lines, ultimately inducing phenotypic reversion as in TUC-3 cells. Most interestingly, incubation of these two peptides with MIA-PaCa-2 cells results in 100 percent cell death. Identical results hold for U-251 cells. Control peptides have no effect on cancer cell growth in any of these cell lines.

These results suggest that both peptides, whose mechanism of action against oncogenic ras-p21 is known, may be useful agents in treating human cancers. Their anti-cancer activity in a non-ras-transformed cell line awaits further investigation.

IV. Anti-cancer peptides from p53

A. Peptides from molecular modeling of p53

As noted in the introduction section, single amino acid substitutions at critical positions in p53 cause its inactivation leading to reduced or absent anti-proliferation signaling by this protein. The x-ray crystal structure of the DNA-binding domain (residues 93–312) of p53 has been determined (Che, et al, 1994). Using the same methodology described in the previous section on ras-p21, we have computed the changes in conformation of p53 induced by these amino acid substitutions. These included H179L, R249W and I255F (Chen et al, 1999). These substitutions are induced by the chemical carcinogen, vinyl chloride that is an important causative factor for angiosarcoma (Chen et al, 1999).

The most striking result of these calculations was the finding that each of these substituted, oncogenic proteins all showed the same structural changes in two domains: residues 94–110 and 204–217, both of which are spatially distant from each of the substituted residues (Chen et al, 1999). Our computed results for the 204–217 segment implied that this region is induced to undergo major changes in structure by oncogenic substitutions in p53. This corresponds exactly to the experimental results on p53 with an anti-p53 monoclonal antibody, PAb240, that recognizes this determinant only in the oncogenic p53 mutant forms.

Our computed results further suggest that the 94–110 domain, on the amino terminus of the DNA-binding domain, is flexible. As wild-type p53, it is exposed to the aqueous solvent while in each of the oncogenic p53 substituted forms, it moves to an unexposed position within the interior of the protein (Chen et al, 1999). Indeed, the anti-p53 monoclonal antibody, PAb1620, recognizes wild-type p53 only and does not recognize any of the substituted forms. These findings corroborate our computational results.

1. The 97–155 domain of p53 is critical in regulation of phosphorylation and activation of this protein

In an independent study, we found that a number of kinases, such as JNK, protein kinase A (PKA), and the βsubunit of casein kinase II (CKIIβ), all phosphorylate p53 in the 97–155 amino terminal domain (Adleret al, 1997). Phosphorylations induced by CKIIβ appeared to activate wild-type p53 (Adleret al, 1997). On the other hand, in mutant p53, such as R249W-p53, the level of phosphorylation in this domain decreases and is much more sensitive to temperature, decreasing as temperature decreases (Adleret al, 1997).

We synthesized three peptides from the 97–155 domain, 97–117, 115–135 and 133–153 and assayed them for their abilities to affect p53 phosphorylation. Only the 97–117 p53 peptide, called P7, was able to block kinase-induced p53 phosphorylation by JNK and PKA and to enhance CKIIβ-induced phosphorylation (Adleret al, 1997). We have attached a leader (penetratin) sequence to P7, as described above for the ras peptides, and introduced this fused peptide into fibroblasts and found a large increase in p53 phosphorylation intracellularly (Adleret al, 1997). Because of its enhancement of activating phosphorylations on p53, we are currently testing this peptide in cancer cell lines. Preliminary results suggest that this peptide causes cancer cell death. These anti-cancer effects appear to be dependent on the presence of wild-type p53 in cancer cells. As we now describe, we have developed a set of p53 peptides whose anti-cancer effects are not dependent on the presence of wild-type p53.

B. Amino terminal domain peptides from p53 are highly effective and selective anti-cancer agents

As noted in the introduction section above, p53 contains an amino terminal domain consisting of residues 1–92 in the human form for which no x-ray structure is available. Embedded within this domain is a proline-rich sub-domain that we have found, using conformational analysis, is quite flexible. Part of this subdomain, residues 12–26, has been found to be involved in the binding of p53 to an important target molecule, called human (or mouse) double minute-binding protein, or HDM-2 or MDM-2, respectively (Laptenko and Prives, 2006). HDM-2 promotes mitosis and can itself become oncogenic. When it binds to p53, it targets this protein for ubiquitination and subsequent proteolysis in the proteosome of cells (Laptenko and Prives, 2006).

We reasoned that, if we synthesized the p53 12–26 peptide and introduced this into cancer cells containing native p53, the peptide might block the HDM-2-p53 interaction prolonging the half-life of the wild-type protein. This would enable it to block proliferation and to induce apoptosis. As with the ras peptides, to introduce the p53 12–26 sequence into cancer cells, we attached a penetratin sequence to it. The question arose as to whether to place the penetratin or leader sequence on the amino or carboxyl terminus of the peptide.

To resolve this question, we utilized the x-ray structure for the p53 12–27 peptide bound to the p53-binding region of MDM-2 (Kussie and Gorina, 1996). In this structure, the p53 peptide adopts largely an α-helix in the complex. Since it is known from conformational analysis that positive charges on the carboxyl terminal ends of α-helices stabilize these structures and since penetratin sequences are all highly positively charged, we placed the penetratin sequence on the carboxyl terminus of this peptide (Kanovskyet al, 2001).

p53’s HDM-2-binding domain can be subdivided into a conserved domain (residues 12–18) and the HDM-2-contact domain (residues 17–26) (Do et al, 2003 ). We have therefore further synthesized these latter two subdomains attached to our penetratin sequence. The actual sequences are summarized below:

p53 12–26-leader:	PPLSQETFSDLWKLL-leader, called PNC-27
p53 12–20-leader:	PPLSQETF-leader, called PNC-21
p53 17–26-leader:	ETFSDLWKLL-leader, called PNC-28.

Leader or penetratin sequence: KKWKMRRNQFWVKVQRG, called leader.

1. Tests on the BMRPA1-TUC3 cell system

We first tested PNC-28 on TUC-3 cells and on their normal BMRPA1 counterpart cell line. After 3 days of treatment, all of the 1×10⁶ TUC-3 cells were completely killed (Kanovsky et al, 2001). However, PNC-28 treatment of contact-inhibited monolayers of BMRPA1 cells and of these cells in growth phase had no effect on either cell viability or of their ability to grow into contact-inhibited monolayers (Kanovsky et al, 2001). To our surprise, TUC-3 cells treated with this peptide did not express elevated levels of any of the markers for apoptosis including Bax and waf^p21 proteins, nor did they show other indications of apoptosis, such as “laddering” of DNA (Kanovsky et al, 2001).

2. PNC-28 kills many different human cancer cell lines

We proceeded to test PNC-28 on a number of human cancer cells lines, i.e., HeLa (cervical cancer), E49 (angiosarcoma) and SW1417 (colon cancer) cell lines and found that within 48 hours of treatment, PNC-28 killed all of the cancer cells (Kanovsky et al, 2001). Surprisingly, PNC-28 was effective in destroying all of the SW1417 cells, despite the fact that this is a cell line known to be p53 homozygously deleted (Kanovsky et al, 2001). Combining this result with the absence of any markers for p53-induced apoptosis in PNC-28-treated TUC-3 cells suggested that this peptide was inducing cancer cell cytotoxicity by a p53-independent mechanism.

3. PNC-28 does not affect normal stem cell growth and differentiation

One of the drawbacks to chemotherapeutic agents is their suppression of bone marrow that results in anemia and leukopenia. The latter condition often results in immunosuppression, leading to serious life-threatening infections in patients treated with these agents. We tested whether PNC-28 affected the ability of human stem cells, from the cord bloods of five human donors, to differentiate, in the presence of hematopoietic growth factors. These included colony stimulating factor (CSF) and GM-CSF (granulocyte-monocyte-stimulating factor), IL6 for lymphocytes, and erythropoietin for erythrocytes. We found that neither PNC-28 nor the negative control PNC-29 peptide had any effect on the ability of the stem cells to differentiate into and form colonies of these cell lines (Kanovsky et al, 2001). Thus it seems clear that PNC-28 would likely not suppress bone marrow functioning in humans.

4. PNC 27, PNC-28 and 21 are cytotoxic to many other cancer cell lines

Table 1 summarizes the human cancer cell lines against which we have tested our p53-derived peptides. As can be seen from this table, these peptides are lethal to every one of the cell lines listed. These include cell lines, such as SW1417 cells, that have the p53 gene homozygously deleted. It can also be seen from this table that PNC-27 and 28 both induce complete cytotoxicity to three different breast cancer cell lines in less than one hour (Do et al, 2003) when present in the incubation medium at 30 uM!

Table 1.

Times for induction of total cytotoxicity by PNC27/28 in some cancer cell lines

Cell Line	Cell Type	Time for Total Cytotoxicity (1×106 Cells)
TUC-3 (100 ug/ml)PNC-28	Pancreatic Acinar Carcinoma	72 hr.
HeLa (100 ug/ml) PNC-28	Cervical Squamous Carcinoma	72 hr.
E-49 (100 ug/ml) PNC-28	Angiosarcoma	72 hr.
SW1417 (100 ug/ml) PNC-28*	Colon Cancer	48 hr.
MDA-MB-468 (30uM) PNC-27	Breast Cancer	30 min.
MDA-MB-157 (30uM) PNC-27	Breast Cancer	30 min.
MDA-MB-453 (30uM) PNC-27*	Breast Cancer	30 min.
HT29 (30uM) PNC-27*	Colon Cancer	1 hr.
SAOS2 (30uM) PNC-27*	Osteogenic Sarcoma	1 hr.
MIA-PaCa-2 (150 ug/ml) PNC-28	Pancreatic Carcinoma	72 hr.

^*Cell line is homozygously p53-deleted

5. PNC-27, 28 and 21 are lethal to breast cancer cell lines but not to an untransformed breast epithelial cell line (Do et al, 2003)

Figure 6 summarizes the effects of our p53-derived peptides on the three breast cancer cell lines and on an untransformed human breast epithelial cell line (Do et al, 2003). PNC-27 induced total cell cytotoxicity in MDA-MB-468 and MDA-MB-157 cell lines in 30 minutes and induced extensive cell death in the MCF-7 line, which it completely killed in 2 hours. PNC-28 appears to be almost as effective as PNC-27 although it required longer times to induce total cell death. PNC-21 was less effective than the other two p53-leader peptides, although it eventually induced extensive cytotoxicity. In contrast, a “scrambled” PNC-27 sequence in which the sequence of the p53 portion was scrambled while the sequence for the leader peptide portion was maintained had no effect on the growth of any of these cell lines, suggesting that the cytotoxic effect of the three peptides was specific.

Figure 6.

Effects of incubating three different human breast cancer cell lines, MDA-MB-468, MCF-7 and MDA-MB-157, with the three p53 peptides attached on their carboxyl terminal ends with a penetratin sequence, i.e., PNC-21, 28 and 27 (conditions 5–7, respectively). The effects of controls are also shown in conditions 1–4. Also shown in this figure is the effect of the controls and the three peptides on the untransformed MCF-10-2A breast epithelial cell line.

As also demonstrated in Figure 6, other additional controls indicate that the anti-cancer effect of the three peptides is specific; the penetratin peptide by itself, the “naked” p53 17–26, and 12–26 peptides without the leader sequence (that do not traverse the cell membrane), all do not induce cytotoxicity in these human breast cancer cell lines. On the other hand, PNC-27, 28 and 21 do not affect the viability of the untransformed MCF-10-2A cell line (Do et al, 2003), again confirming that these peptides do not affect normal or untransformed cells.

6. PNC-27 and 28 induce necrosis and not apoptosis of cancer cells (Do et al, 2003)

We further investigated how these peptides induce cell death in these cell lines. A major target of the initial cascades in apoptosis is the expression of intracellular proteases or caspases. As can be seen in Figure 7, entries 4 and 5, PNC-27 did not induce expression of caspase activity when tested against the MDA-MB-468 breast cancer cell line above background caspase in control-treated cells (entries 1–3)(Do et al, 2003). When treated with low dose paclitaxel, that is known to induce apoptosis of cancer cells, however, these treated cells do express high levels of caspase activity (entry 6). The cells do not express elevated caspase activity levels when treated with high dose paclitaxel (entry 7) that is known to cause tumor cell necrosis, not apoptosis.

Figure 7.

Caspase activity as a marker for apoptosis in the MDA-MB-468 breast cancer cell line when treated with PNC-27. As shown in the inset, treatment of the cells with PNC-27 for 5 or 10 minutes (conditions 4 and 5) did not result in enhanced caspase activity when compared with the activity of controls (conditions 1–3). (PNC-27 at 30 uM induces total cell death within 1 hour in this cell line.) In contrast, low dose paclitaxel, which is known to induce apoptosis, does induce high caspase activity when incubated with these cells for 10 min (condition 6). On the other hand, high dose paclitaxel, that does not induce apoptosis but necrosis, does not induce elevated caspase activity in these cells.

In contrast to these results, as shown in Figure 8, PNC-27, but not controls, induces release of high levels of lactate dehydrogenase (LDH), a cytosolic enzyme, indicative of tumor cell necrosis in the three human breast cancer cell lines (Do et al, 2003). Significantly, treatment of MCF-10-2A untransformed cells with PNC-27 (far right in Figure 8) results in no increase of LDH above background compatible with its not affecting normal cell viability. Thus the three p53 peptides induce necrosis and not apoptosis of cancer cells by a mechanism that apparently does not involve p53!

Figure 8.

LDH release as a marker for cell necrosis when the three breast cancer cell lines and one untransformed cell line shown in Figure 6 are treated with PNC-27 or with controls. PNC-27 is seen to induce high levels of LDH release in the three cancer cell lines (graph sets 1–3) but not in the untransformed MCF-10-2A cell line (graph set 4). The controls as listed in the inset to this figure are seen not to induce enhanced release of LDH, suggesting that the effect of PNC-27 is specific.

7. PNC-27 and 28 induce lysis of the cancer cell membrane

This finding prompted us to investigate the events that transpire when the peptide enters cancer cells and normal cells. For this purpose, we prepared a fluorescent-labeled PNC-27 peptide that we incubated with MDA-MB-468 breast cancer cells and untransformed MCF-10-2A breast epithelial cells for 15 minutes that were analyzed by confocal microscopy (Do et al, 2003). We found that the fluorescence concentrated in the cell membrane, the nuclear membrane, and to a lesser extent, the nucleoli. Much less fluorescence appeared in the untransformed cell line (Do et al, 2003).

We then performed an identical experiment with PNC-27 and evaluated the cells using scanning electron microscopy (Do et al, 2003). Here we found that the peptide induced pore formation in both the cell and nuclear membranes but had no effect on the membranes of the MCF-10-2A cell line. In fact, much less peptide entered the untransformed cell line as was subsequently confirmed by Western blotting of the cells that were lysed after treatment with peptide (Do et al, 2003). Evidently, PNC-27, 28 and 21 induce membrane damage to cancer cells selectively.

8. The structure of PNC-27 provides insight into its mechanism of action

We have been able to determine the structure of PNC-27 using two-dimensional NMR both in aqueous and membrane-like solvents. The structure in both solvents is two α-helices, at right angles to one another; one in the p53 segment and one in the leader sequence domain separated by a loop at the junction of the p53 and leader peptide sequences (Rosal et al, 2004).

The structure is highly amphipathic: the hydrophobic amino acid residues occur on one face of the structure while the hydrophilic residues occur on the opposite face. These hydrophilic residues are themselves further split into a negatively charged domain in the p53 segment and a positively charged domain in the leader peptide portion of the molecule.

Thus the effect of placing the leader peptide on the carboxyl terminal end of the p53 sequence was to induce formation of a double α-helix that is highly amphipathic. This is a well-known motif for membrane active peptides such as melittin that is a major component of bee venom that intercalates in the cell membrane of red blood cells and induces hemolysis (Pincus, 2001) and magainin, the antibacterial peptide that induces lysis of bacterial cell membranes but has no effect on the red cell membrane (Dathe and Wieprecht, 1999).

PNC-27 therefore appears to be a membrane-active peptide that selectively induces lysis of cancer cell membranes. The elements of the cancer cell membrane with which this peptide interacts are unknown. We are currently actively investigating this question.

9. PNC-28 is effective in treating cancers in nude mice (Michl et al, 2006)

Since PNC-27, 28 and 21 are effective in selectively inducing cytotoxicity in cancer cells, we elected to study their effects in vivo. For this purpose, we utilized nude mice which do not possess cell-mediated or humoral immunity facilitating xenotransplantion of these tumors. We elected to transplant TUC-3 cells into these mice since this cell line has the highest metastatic potential and forms aggressively invasive tumors.

We treated the tumors in mice by implanting pumps that deliver a constant amount of peptide (PNC-28 or control X13-leader or PNC-29 peptide) over a two week period. We simultaneously implanted tumors and pumps in the peritoneal cavities of nude mice. Over a two week period, we observed no tumor growth for the PNC-28-treated mice but rapid tumor growth in the control peptide-treated group. After this period, we performed a peritoneal lavage of mice treated with PNC-28 and PNC-29 control and analyzed them using light microscopy. We found that in PNC-29-treated mice, colonies of spindle-shaped malignant cells were present. While in PNC-28-treated mice, no malignant cells, but only reactive mesothelial cells, were present (Michl et al, 2006). Thus PNC-28 eradicates pancreatic cancer cells in this animal model.

Surprisingly, we obtained identical results when we implanted the pumps outside of the peritoneal cavity in the flank region. In this case, the pumps delivering PNC-28 were far-removed from the site of tumor implantation (Michl et al, 2006). We then repeated these experiments except that we implanted the tumors and pumps on opposite sides of the nude mice, i.e., tumors in the left shoulder region, pumps in the right flank region. Figure 9 shows that PNC-28 effectively blocks tumor cell growth under these stringent conditions both during administration and two weeks post-administration. As can be seen in Figure 9, during the two week period of PNC-28 administration, no tumor growth occurred in the PNC-28-treated mice while rapid growth occurred in the control peptide-treated mice. After PNC-28 administration was discontinued, there was some small growth, but the size reached a plateau and did not increase (Michl et al, 2006). In contrast, tumors treated with control PNC-29 peptide grew rapidly and metastasized. Importantly, the mice treated with PNC-28 thrived and exhibited no toxic side effects. Their weights and nutritional-intake remained at the same level as untreated mice (Michl et al, 2006). This finding corroborates with our in vitro findings that PNC-28 (and 27 and 21) have no effects on the viabilities of untransformed cells.

Figure 9.

Effects of PNC-28 on tumors implanted into nude mice. A minimum of 1×10⁶ TUC-3 pancreatic cancer cells was xenotransplanted into the right hind leg region and minipumps delivering a total of 2 mg of PNC-28 over a 14-day period were implanted in the left front leg region. Tumor size was observed over the treatment period and then another 20 days after the pumps ceased delivering PNC-28 peptide. As shown in the inset, the open squares show tumor size over time when the negative control peptide, PNC-29, the X13 cytochrome P450 sequence linked on its carboxyl terminus to a penetratin sequence, is administered; the filled circles show the effects of PNC-28 on tumor size. The “p” values shown, over each open square, show statistical significance between tumor size obtained with PNC-28 and with PNC-29, as computed with a 3-way repeated measures ANCOVA test (54). All p values<0.05 show statistically significant differences.

Overall, these results are highly encouraging and suggest that PNC-28, and 27, may constitute an effective therapeutic anti-cancer drugs. We are currently expanding our studies to include testing against a variety of other cancers and in syngeneic animal models. Similar results with PNC-28 are now being demonstrated testing this potent novel peptide against a lethal human pancreatic cancer cell line (Bowne et al, 2007).

V. Conclusions

Molecular modeling of protein structure has resulted in the rational design of peptides from proteins that control cell proliferation that appear to block the growth of cancer cells selectively. Peptides from the ras-p21 protein block uncontrolled cell growth induced by oncogenic ras-p21 but leave the normal functioning of the wild-type protein intact. Further, they block tumor cell growth but have no effect on the growth of normal or untransformed cells. Our p53-derived peptides induce tumor cell necrosis of many different types of human tumors and block the growth of a highly metastatic pancreatic tumor in nude mice. These peptides damage the cancer cell membrane but do not affect the membranes of untransformed cells. The translational implication of these results is that both sets of peptides may be effective in the treatment of a variety of human cancers. The ras-p21 peptides have further been exceptionally important in defining differences between the mitogenic signaling pathways induced by oncogenic and wild-type ras-p21 proteins. Provided that these differences exist across different human tumor cell lines, it is possible to block components of the oncogenic pathway that are not important for normal cell signaling, allowing for the design of more discriminating anti-cancer agents.

Abbreviations

CKIIβ

βsubunit of casein kinase II

CSF

colony stimulating factor

DAG

diacylglycerol

EDMC

electrostatically-Driven Monte Carlo method

EGF

epidermal growth factor

GNEF

guanine nucleotide exchange factor

IP₃

inositol triphosphate

LDH

lactate dehydrogenase

MAPK

mitogen-activated protein kinase

NMP’s

nuclear matrix proteins

PI3K

phosphoinositol-3-hydroxy kinase

PKA

protein kinase A

PLC

phospholipase C

RBD

ras-binding domain

SCEF

self-consistent electric field

TK

tyrosine kinase