PNAs breach the blood-brain barrier?

Peptide nucleic acids (PNAs) are the thoroughbreds of the nucleic acid world. Combining the structural motifs of proteins and DNA, their chimeric structure confers not only resistance to nucleases and proteases, but also increased binding stability¹. Although these properties give them an edge over other types of oligonucleotides as antisense agents, intracellular delivery of PNAs to their ultimate targets remains an elusive goal. In particular, cells within the central nervous system remain inaccessible because of the impermeability of the blood-brain barrier (BBB). Now, in a recent issue of Proc. Natl. Acad. Sci. USA, Beth Tyler et al.² report the surprising finding that unmodified PNAs can indeed cross the BBB after intraperitoneal injection. Their work is the first demonstration that naked PNAs can traverse the BBB in amounts sufficient to mediate appreciable behavioral effects, thus raising the prospect of systemic targeting of other brain proteins.

Antisense and antigene therapies could be valuable approaches for treating cancers, neurodegenerative diseases, and ischemic disorders of the brain. But the difficulty of delivering these agents through the BBB—a tightly sealed layer of cerebral endothelial cells that form continuous tight junctions—has so far stymied these efforts.

The BBB excludes most solutes from the brain on the basis of size, charge, and lipid solubility, such that molecules that are >500 Da, polar, or hydrophilic cannot enter cells or penetrate the tight junctions. Recent studies have shown that the BBB is much more dynamic than assumed in the past, and some passage of solutes can occur by transcytosis, carrier-mediated transport, or simple diffusion of hydrophobic substances.

As the BBB is impermeable to most naked antisense oligonucleotides and gene therapy vectors, approaches have typically relied on injecting agents either directly into the brain parenchyma, or intrathecally through the dura mater (the outer covering of the brain) into spaces where cerebrospinal fluid (CSF) can circulate. However a major limitation of intraparenchymal injection is that it leads to extremely limited distribution in the brain tissue (e.g., one study³ showed gene delivery only 5 to 25 cell diameters from the injection site). For this reason, blood and CSF are now considered the most promising routes for global delivery to the brain.

Although injection of viral vectors into the CSF ventricle has met with some success⁴, the cerebral capillary endothelium represents a more appealing target because of the much larger depth and area of accessible tissue. For example, an adult human brain contains about 100 cm² of capillary endothelium per gram, with over 12 m² of microvasculature⁵. In fact, the total end-to-end brain capillary length is about 650 km, roughly the distance from Boston, MA, to Washington, DC.

The large surface area accessible through the vasculature has thus prompted researchers to try to find ways of circumventing the BBB. In one approach, hyperosmotic mannitol is used to transiently permeabilize the membrane, allowing administration of chemotherapy or viral gene vectors through the human or rat carotid artery⁶. In another approach, Boado et al.⁷ linked PNAs to antitransferrin antibodies, which were endocytosed after binding to transferrin receptors enriched on the endothelial surface. These investigators demonstrated a 28-fold increase in the uptake of antibody-conjugated PNAs compared with naked PNAs—an efficiency comparable to that of morphine. In our own laboratory we have attempted to conjugate receptor-binding factors to recombinant adenoassociated virus vectors to enhance transvascular delivery; however, the results have thus far not shown increased uptake.

In the PNAS study², Tyler et al. chose an unlikely approach, assessing the efficacy of antisense PNAs injected into the peritoneum of rats (Fig. 1). By targeting PNA against the mRNA for rat neurotensin receptor, they were able to use behavioral tests such as neurotensin-mediated analgesia to monitor the ability of the PNA to cross the BBB and reduce expression of the receptors. To corroborate the results, they carried out radioligand binding studies, which indicated that brain homogenates collected from PNA-injected rats were 35% to 40% less effective in binding neurotensin. Quantitative reverse transcriptase polymerase chain reaction (RT-PCR) showed no decrease in neurotensin receptor mRNA levels when antisense PNAs were injected intraperitoneally, suggesting that the downregulation of receptors was mediated by a blockade in protein synthesis. Interestingly, a disparity was observed between in vivo and in vitro actions of the antisense PNA. When total RNA isolated from control rats was treated with the antisense PNA, RT-PCR detected a 65% decrease in the RT product, demonstrating that the PNA was capable of binding and impairing transcription in vitro.

Figure 1: PNAs can be administered intracranially or systemically to a rat.

In the latter case, they must traverse brain capillaries in order to reach neurons. Tight junctions (shown in black) help to prevent free passage of molecules out of the capillary. Modified PNAs can be targeted to receptors on the luminal side, or naked PNAs might traverse the endothelial cells by other, unidentified means.

The only direct evidence the authors provided for BBB passage was a gel-mobility shift assay to detect PNAs in the brain tissue. An alternative approach would have been to inject an isotope-tagged PNA into the brain, although, as the authors point out, a bulky or polar tag such as ⁹⁹Tc or ¹²⁵I might itself inhibit BBB passage.

In further experiments described in their paper, Tyler et al. tested the effectiveness of "antigene" therapy by injecting sense PNAs directly into rat brains. In contrast to the PNA antisense experiment, this treatment resulted in a 50% decrease in neurotensin receptor mRNA, suggesting that the sense PNAs blocked transcription in vivo. (The authors, however, did not report intraperitoneal injection of the sense PNAs, so the results of the sense and antisense PNA experiments cannot be directly compared.)

An important consideration for the therapeutic applications of PNAs is the duration of the antisense effect. The behavioral changes reported by Tyler et al. were short-lived (i.e., 48 hours), as were the antigene effects of directly injected PNA to the brain, which showed only a transient dip in transcription. Transient changes in gene expression may be useful in animal "knock-down" models, but clearly PNA gene therapy is still a long way from the clinic.

The results of Tyler et al. indicate just how far PNA delivery approaches have come. While early studies suggested that PNAs are unable to efficiently permeate cell membranes and penetrate into the cytoplasm and nucleus, subsequent work has indicated that naked PNAs can be taken up by cells in vitro⁸ and that uptake can be significantly enhanced by incorporating membrane-penetrating signal sequences⁹, or by conjugating antireceptor antibodies that facilitate endocytosis. More recent studies have shown that specially modified PNAs can cross the BBB in significant amounts¹⁰ and that PNAs injected intrathecally can mediate antisense effects in vivo¹¹. The data of Tyler et al. now suggest that naked antisense PNAs can reach the brain, even when injected at a location remote from the target site.

In a way, it seems fitting that these PNA hybrids of peptides and nucleic acids are performing so promisingly. We can't help wondering whether Pauling, when he first perceived the alpha helical structure of proteins, paving the way to the discovery of DNA's helical structure, could have imagined the conjoining of these two molecules for the purpose of curing human ailments.