C&EN: COVER STORY - BREACHING BARRIERS

CELIA M. HENRY, C&EN WASHINGTON

"Life doesn't exist without barriers," says Paul Wender, a chemistry professor at Stanford University. "Therapy requires that we be able to breach those barriers. To breach those barriers, we have to understand something about their composition and dynamics."

Most small-molecule drugs reach their targets because they are able to passively diffuse through the cell membrane. But reliance on passive diffusion limits the universe of drugs to those that are soluble in both the polar extracellular environment and the nonpolar cell membrane. Researchers in academia and start-up companies are working on ways to deliver all types of drugs--small molecules, peptides, proteins, and nucleic acids--without relying on passive diffusion.

A promising method is known as protein transduction. Despite the name, protein transduction can be used to deliver practically every type of molecule.

Several naturally occurring proteins have been found to enter cells easily, including the TAT protein from HIV, the antennapedia protein from Drosophila, and the VP22 protein from the herpes simplex virus. Specific short sequences within the larger molecule account for the transduction abilities of these proteins. These peptides can be used to deliver a variety of molecules by covalently linking to them.

The mechanism of protein transduction is still unclear. "We know what it isn't," says Peter Fischer, head of drug discovery at Cyclacel, a company located in Dundee, Scotland. Protein transduction "is not receptor mediated. There are no particular cell-surface or membrane receptors for this. It's some kind of physical interaction of the amphiphilic structure of the peptide with the cell membrane that leads to this permeabilization. This is not fully understood."

One sign that the process is not receptor mediated is that, in the case of the TAT peptide, the d isomers of the amino acids can be used and the sequence can be scrambled without preventing transduction. "If it were a ligand for a receptor, you could not do those things," says Steven F. Dowdy, a Howard Hughes Medical Institute researcher at the University of California, San Diego, medical school. "It's an ionic interaction with the cell membrane, likely with proteoglycans, which have these sugar groups that are modified with sulfates and sialic acids." Dowdy reported that a fusion of the TAT protein transduction domain with the protein b-galactosidase could be delivered to all tissue types in a mouse [Science, 285, 1569 (1999)].

Dowdy's work indicates that protein transduction can be used to target all cell types and that the process works regardless of the size of the cargo. Since cargo size doesn't matter, "you can deliver compounds that are twice as large as what's traditionally used in big pharma," he says. Dowdy believes that the ability to deliver larger molecules could lead to drugs with catalytic activity rather than simply inhibitors.

In addition, protein transduction appears able to deliver cargo regardless of its composition. Everything "from small molecules to oligonucleotides to peptides to proteins to iron beads to liposomes" can be delivered using protein transduction, Dowdy says. To commercialize the technology, he helped found the company Ansata Therapeutics in La Jolla, Calif. He serves on the company's scientific advisory board but is not involved in day-to-day operations.

Originally, protein transduction appeared not to go through an endocytotic pathway. In some experiments, the cargo appears to be spread throughout the cells. Other times, it appears to be trapped in vesicles.

Now, Dowdy believes that protein transduction probably occurs through a "specialized type of endocytosis that favors leakiness out of the vesicles into the cytoplasm," he says. Knowing the mechanism makes devising ways to improve it more likely. "Once you know how it works," he notes, "you can always make it better or you can remove things from your list of potential therapeutics because they won't work by that particular approach."

There are also conflicting results about the energy requirements of the transduction process, which are investigated by running experiments at two different temperatures. "The problem is when you block ATP, you see a reduction in transduction but not elimination," says Paul D. Robbins, a professor of molecular genetics and biochemistry at the University of Pittsburgh. "That means that maybe there are two mechanisms."

Wender believes that there's no doubt that protein transduction domains, which he prefers to call molecular transporters, work through a variety of mechanisms. However, he is reluctant to advocate a particular mechanism.


BIG LOAD Two live cells (nuclei stained blue) have been transduced with a protein transduction domain protein fusion (green). The cells have also been stained with a lysosomal marker (red). The majority of the protein enters the cell by endocytosis but shows little tendency to associate with lysosomes. Understanding the cell biology behind protein transduction will be key to improving intracellular drug delivery using this technology. ANSATA THERAPEUTICS

NEVERTHELESS, he believes that multiple mechanisms are at work. "The biological barriers that some of these agents penetrate are very different in composition," he says. "It's just philosophically impossible to invoke the same mechanism because you don't have the same constituents involved in the pathway."

For example, peptide sequences have been shown to get across the cell plasma membrane and the stratum corneum, the outer layer of the skin [Nat. Med., 6, 1253 (2000)]. "The plasma membrane of the cell is very different than the stratum corneum of the skin. They're both biological barriers, but the composition of those barriers is hugely different, the dynamics of those barriers are hugely different. To think of them having the same mechanism would be inappropriate," Wender says.

The naturally occurring protein transduction domains share certain features: They are cationic and tend to be rich in the amino acid arginine. Arginine is special, Wender says, because of its guanidinium head group. Arginine can form two hydrogen bonds, whereas lysine, which is also cationic, can form only one hydrogen bond.

Working in collaboration with Jonathan B. Rothbard (formerly at Stanford and now the chief scientific officer at CellGate), Wender has found that the ideal number of arginines--what he calls the "sweet spot"--ranges from a minimum of six to a maximum of 15. "We usually go to the short side so we can save on cost of goods," he says. Wender serves on the board of CellGate, located in Sunnyvale, Calif., and is also its chief scientific adviser.

Although Wender doesn't claim to know the mechanism, he has a "working hypothesis" of why that particular range of arginine residues works. He describes the molecular transporter as a three-dimensional structure with positive charges on its surface that are associating with negatively charged counterions in the extracellular region. As the complex approaches the cell surface, it exchanges those counterions for negatively charged entities on the cell surface. As the complex is pulled into the lipid bilayer, its surroundings become increasingly nonpolar, so it holds together more strongly.

"If you have too few arginines to start off with, you're not going to have good cell surface association," Wender says. "The association between the positively charged guanidinium group and the negatively charged cell surface can be thought of as a 'handshake.' If you have only one positive charge, you only have the capability of shaking hands with one partner on the cell surface. That's going to be a weak association." Each additional arginine strengthens the interaction. At seven arginines, the effect becomes similar to Velcro. "It sticks longer, so it's more likely to be taken up into the membrane and then on the other side be released," Wender says. "If you make that Velcro strip too long, it will not come off the inner leaf."

HOWEVER, other research has shown that artificial protein transduction domains can be made with a variety of cationic amino acids. In fact, even nonnatural amino acids such as ornithine can be used to make protein transduction domains. Robbins has found that a broad range of cationic peptides can work as transduction domains, including polyarginine, polylysine, polyornithine, and even polyhistidine. "We can get histidine to work as a pH-sensitive transduction domain that's almost as efficient as TAT," he says.

Ansata scientists are synthesizing a variety of new transduction domains. "We've been making composites of natural and nonnatural amino acid mixtures that work at the same efficiency, or in some cases slightly better, than the polyarginine type of transporters," says Brian A. Pollok, senior vice president of research.

"I think there have been some exaggerated claims about the efficiency of individual protein transduction domain sequences. We don't really see the tremendous differences between TAT or antennapedia--sort of the first-generation type--and the second-generation type of the more synthetic transduction sequences," Pollok says.

Vladimir P. Torchilin, a professor of pharmaceutical sciences at Northeastern University, Boston, is trying to find out just how large a cargo the TAT protein transduction domain can deliver. His group has shown that 200-nm-diameter liposomes can be delivered even in the presence of metabolic inhibitors [Proc. Natl. Acad. Sci., USA, 98, 8786 (2001)].

"The general feeling is that you cannot go indefinitely up with the size of the particle to be delivered," Torchilin says. "It looks like the delivery is somehow associated with a membrane reorganization process. Certainly, if too much membrane area is involved in this reorganization process, you might lose cell integrity or the whole mechanics might not work."

However, Torchilin doesn't believe that knowing the upper limit is important from a practical standpoint. "If you can effectively deliver 200-nm particles, you can get into the cell almost everything you need," he says. "It's interesting if you can do big particles, but I don't believe it's important from a practical point of view. It's more important to understand the mechanism--what are the limitations of that mechanism--from a basic point of view."

UNDERSTANDING the mechanism might appear to be strictly an academic debate, but Robbins believes that it's necessary to advance the technology. "I think to really develop the technology as a viable clinical approach, you have to know how it's getting in, how it's getting out, and how you may be able to improve upon that to increase the activity of the therapeutic cargo that you're delivering," Robbins says. "If you're delivering a large protein, you don't want it to be trapped in the ER [endoplasmic reticulum] when it's supposed to work in the nucleus."

Cyclacel is developing the antennapedia protein, also known as Penetratin, as a transduction method. According to Fischer, the company became interested in Penetratin as a way to introduce proteins into cells for studies unraveling the cell biology of proteins associated with the cell cycle. The company licensed the technology from its original inventors in France.

Cyclacel has been able to shorten the peptide sequence that is used for protein transduction from the original 16 residues to just the seven C-terminal residues. "We think that the 7-mer is the shortest sequence that can adopt some sort of stable amphipathic helix," Fischer says. "If you make it any shorter, you've lost all the internalization capabilities."

In collaboration with CellGate, Wender has shown that not only can the amino acid sequences be changed on molecular transporters, they don't even need a normal protein backbone.

At first, the researchers made relatively modest changes--moving the side chain to a nitrogen atom rather than a carbon atom [Proc. Natl. Acad. Sci. USA, 97, 13003 (2000)]. In that case, they were able to maintain the same spacing between the side chains as would normally occur in a peptide.

Then they got "more adventurous," Wender says. "We said, let's change the backbone radically. Let's put in a carbamate backbone where we will now have very different properties because we're no longer dealing with peptidic linkages. And let's change the spacing." With the carbamate backbone, there is an extra atom between the side chains. The oligocarbamate transporters were used to transport the model drug biotin through mouse skin [J. Am. Chem. Soc., 124, 13382 (2002)].

Wender and colleagues then investigated the role of backbone spacing in cellular uptake. They made molecular transporters with 10 amino acids, including seven arginines. They placed an amino caproic acid spacer between each residue [J. Med. Chem., 45, 3612 (2002)]. "We found that the uptake into Jurkat cells [derived from human T-cell leukemia] was enhanced by increasing the spacing between arginine groups," Wender says.

Wender wants to determine how large a payload can be delivered using transporters. He needs to use a cargo that does not change shape as a function of time, because even though proteins can be quite large, they could also be unfolded to go through the membrane, complicating the interpretation of his experiments.

Wender compares the cargo to elephants and balls of yarn. The cell membrane can be thought of as a picket fence. "It's a big ball of yarn that can't get through the picket fence, but you find the end and you pull it through like a string. On the other side, you reassemble it into a ball of yarn. It looks like you got the proverbial elephant through the picket fence," he says.

To eliminate such possibilities and get to the bottom of such issues, Wender and Stanford colleague Hongjie Dai are conjugating the transporters to carbon nanotubes. "We have evidence now that we're able to get these in," Wender says. "Now we're getting to cargos that are getting close to the dimensions of cells."

A major challenge for protein transduction is targeting it to specific cells. "The good thing is they seem to go everywhere. The bad thing is they go everywhere," Dowdy says. "You're stuck with the traditional pharmacology problem. You still have to select for tissue-specific targets or disease-specific targets."

Torchilin is also starting to combine targeting moieties with liposomes and protein transduction. "You can combine on a single drug carrier both TAT and a targeting moiety like an antibody in such a fashion that TAT is masked by an antibody so it can only go to the target," he says. "Using certain physiological features of the target, you can detach the antibody from the carrier and expose TAT. I'm just starting to find out how practical it is to do it."

Robbins has found that some protein transduction domains are tissue specific. One such peptide is able to deliver proteins into synovial tissue lining joints [Mol. Ther., 8, 295 (2003)]. Robbins says that it is unclear whether these peptides, which are uncharged, work by the same mechanism as protein transduction domains such as TAT.

The French company Synt:em, located in Nîmes, is focusing on the modifications of antimicrobial peptides to design vectors called Pep:trans that can deliver agents into cells or across the blood-brain barrier. Drugs are chemically linked to a peptide vector that is nine or 10 amino acids long. The new chemical entity could be active as is, or the drug might need to be cleaved from the peptide vector once it reaches its destination.

"Nobody knows how to predict whether a compound reengineered with such a vector will be active or to what extent its pharmacological behavior will be modified," says Michel Kaczorek, president and chief executive officer of Synt:em. Such information can only be gleaned empirically.

Some Pep:trans vectors are able to transport a very few percent of the drug directly into the brain. "I don't believe that it will be possible to design the sort of vectors that will transport 10 or 20% of the drug directly into the brain," Kaczorek says. "It's probably enough [to have a local concentration around the receptor in the brain] to have activity."

ONE OF SYNT:EM'S candidates, a morphine-related compound conjugated to a peptide vector, is currently in Phase I clinical trials. When the compound is conjugated to the Pep:trans vector, animal models have shown that not only does the candidate drug work much more rapidly and efficiently at a lower dose than morphine, it no longer exhibits morphine's side effects. The results strongly suggest that the drug gets into the central nervous system rapidly enough and in sufficient quantity without causing these side effects, Kaczorek says.

The Swedish company CePeP, based on the research of Ülo Langel, a professor in the department of neurochemistry at Stockholm University, uses an algorithm to design what it calls cell-penetrating peptides. Kjell Stenberg, CePeP's chief scientific officer, who has 25 years' experience in big pharma, has been surprised at just how nontoxic protein transduction is.

"If you bring things through cells and the transduction system is a bit promiscuous in what it can bring, you should see some toxicity," he comments. "In studies I've seen so far, with the compounds and constructs we have made, if there is toxicity, it's so transient that it's difficult to detect. I can't promise that every compound every time in every system will be nontoxic. I'm just saying that our experience is it's surprisingly nontoxic."

CePeP researchers are using cell-penetrating peptides (CPPs) in two ways. First, they are using them to deliver cargo. By coupling a therapeutic agent to a CPP, "we will be able to reduce general toxicity, increase specificity, and make a better therapy," Stenberg says.

In the second approach, CePeP is making peptides that will enter cells, mimic the inside of a cellular receptor, and turn on a cell-response system. For example, CePeP is working on peptides that mimic the GLP1 peptide, which activates a G-protein-coupled receptor involved in insulin release. Such a peptide could be a therapy for type 2 diabetes.

British company Phogen, a "virtual" company that is a partnership between the Marie Curie Cancer Care charity (which funds the Marie Curie Research Institute) and Xenova Research Limited, is focusing on methods using the herpes protein VP22. This protein has the distinction of being the only compound known to promote both import into cells and the spreading of proteins between cells.

In the process that Phogen calls transport, the VP22 gene is introduced into a cell using any transfection method. (Both viral and nonviral methods work.) The protein is then expressed in the cell. VP22 spreads from the original cell, or "producer" cell, to the surrounding cells.

"Protein is actually exiting from the producer cell and being imported into the surrounding cells," says Elizabeth A. Rollinson, vice president for business development at Xenova and commercial director at Phogen. "You will find the gene only in one cell, but you'll find the protein in lots of surrounding cells." The method works with VP22 alone or with a fusion of VP22 and another protein.

"The application is to link the VP22 to your therapeutic molecule, whether it's a vaccine antigen or a protein such as p53," Rollinson says. "In effect, you're enhancing the efficiency of gene delivery, but you're not actually getting more genes into more cells. You're enhancing efficiency by getting more biologically active protein into more cells" [Mol. Ther., 7, 659 (2003)].

In another application, Phogen condenses the C-terminal half of VP22 around nucleic acid oligomers to form structures called vectosomes. Again, the protein can be just VP22 or a fusion protein. The oligonucleotide can also be selected to have therapeutic activity.

When vectosomes enter the cell, they stay in the cytoplasm until they are activated using either light or some chemical means. For example, Phogen formed vectosomes by fusing VP22 with a peptide that promotes programmed cell death. The peptide was released by light activation, causing cell death and tumor regression in vivo [Mol. Ther., 7, 262 (2003)].


	LIGHT ACTIVATED The herpes virus protein VP22 and fluorescently labeled oligonucleotides assemble into particles called vectosomes. Cells were incubated with the vectosomes overnight and illuminated to activate the particles and release the protein. These images show the cells before (left) and after (right) illumination. PHOGEN IMAGES

VECTOSOMES can be used to deliver proteins and a variety of nucleic acids, including ribozymes, interfering RNA, and plasmid DNA. They can even be used to deliver small molecules, but Phogen is not focusing on that application currently.

"We did a little bit of work with [small molecules], but in our interaction with big pharma, they basically said, 'Why would we want to put a big gloopy protein onto our beautiful small molecule?' " Rollinson says. "I feel sure that there might be some chemical entities that this may be a good solution for. Our focus has been more on the macromolecular side because that seems to be where the key interest is."

Because the vectosomes must be activated to release their cargo, they can be administered systemically but will act only locally using photodynamic therapy. "We have been looking at incorporating different cleavage sites into the protein to use endogenous proteases to trigger release," Rollinson says.

Most protein transduction work is still in the preclinical phase. However, two companies have candidates in clinical trials. One of Synt:em's compounds is in Phase I clinical trials. CellGate has a product formulated for topical application in Phase II trials for psoriasis.

In many ways, protein transduction can still be considered to be "over the horizon," Dowdy says. "Are we too soon to take commercial advantage of this? Is it just an academic issue at this point? Or has it been developed to the point where it is commercializable?" He points to CellGate's technology as the best example so far. "If it works for one compound, it will work for two, and if it works for two, it will work for 10."

Protein transduction technology is still in its early days, but it's looking very promising, Dowdy emphasizes. "It has great potential, but it's potential with a couple of question marks," he says.