Recent Developments in Drug Delivery to the Nervous System

Dusica Maysinger, Radoslav Savic, Joseph Tarn, Christine Alien, and Adi Eisenberg

McGill University, Montreal, Quebec, Canada

NEUROACTIVE AGENTS AND THEIR DELIVERY TO THE CNS AND PNS

In the past decade, the contribution of the material sciences to drug delivery to the brain was realized mainly through the use of nonbiodegradable cylinders for intracerebral implantation of genetically engineered cells, or through the use of polymeric matrices that contained drugs. More recently, further progress has been made in the arena of pro-drugs or conjugates that can exploit existing transport systems. An understanding of the basic mechanisms of the blood-brain barrier (BBB) transport biology provides a broad platform for current and future nervous system drug targeting strategies. In general, current approaches are either invasive (e.g., neurosurgical), pharmacological (e.g., by applying lipid carriers, liposomes, or different kinds of nanoparticles), or physiological (e.g., by taking advantage of normal endogenous pathways of carrier-mediated transport or receptor-mediated transport). Lipid-soluble molecules that have molecular mass under 500 daltons access the brain via lipid-mediated transport, but hydrophilic molecules such as peptides are mainly transported via receptor-mediated endocytosis. The main concepts of and underlying strategies for the administration of clinically relevant growth factors to the PNS, (1), and to overcome the BBB in the CNS, are summarized in several reviews (2-6).

A. Problems with Hydrophilic Agents

Although the surface area of the BBB in the human brain is large [approximately 20 m2 (7)], small hydrophilic molecules cannot access the brain in pharmacologically adequate amounts when administered systemically or orally. This applies also to small peptidomimetic agents such as nerve growth factor (NGF)-mimetics (8) or neurotensine mimetics (9); hence effective delivery of these agents will require a drug delivery and targeting vehicle, or they should be conjugated to a BBB-targeting system. Development of novel drug delivery strategies requires adequate biological models to test their suitability. In vitro models include (i) primary cultures, (ii) immortalized neuronal, glial, and cerebromicrovascular endothelial cultures, (iii) hippocampal immortalized neuronal cultures (10,11), (iv) human cerebromicrovascular endothelial cell lines as a model of the BBB (12), and (v) more complex cocultures of neuronal and glial cells (13). In addition to these in vitro models, a number of in vivo model systems have been employed for testing neuroactive agents and their delivery systems. Rodent models, although indispensable and most commonly used to investigate neurological diseases, have limitations: (i) In general, they show some, but rarely all, of the pathological features of human neurological diseases; (ii) the time course of the progression of the disease is limited due to the difference in life span between two

species; and (iii) tests for verbal communication skills cannot be applied. A number of neurological disorders are associated with either a lack of neuroactive peptides (e.g., growth factors, neurotrophins) or malfunctioning of their receptors (defective binding between receptor and ligand, impaired internalization and transport of the receptor-li-gand complex, or impaired signaling pathways downstream from the receptor site) (14-20). For example, abnormal growth factor levels in the CNS and/or PNS have been associated with Alzheimer's disease, Parkinson's disease, Huntington's disease, and diabetic neuropathy (21-24). Results from preclinical studies employing both in vitro and in vivo models discussed above suggest that individual growth factors (as representatives of hydrophilic molecules) can indeed correct, prevent, or delay some of the pathological features characteristic of diabetic neuropathy, Alzheimer's, Parkinson's, and Huntington's diseases. However, due to the complexities involved in these pathologies, a simple replacement therapy employing drug delivery systems containing individual hydrophilic neurothera-peutics will most likely be used in conjunction with gene therapy and/or stem-cell therapies.

B. Problems with Lipophilic Agents

In general, lipophilic agents have little difficulty in penetrating cell membranes, including those of the BBB. The more lipophilic a drug is, the more readily it will cross the BBB and reach the brain. Thus the main problem with these agents lies not in their permeability but rather with aspects of (i) specificity and selectivity of action in the brain, (ii) neurotoxicity, and (iii) poor solubility and unfavorable pharmacokinetic properties. Some of these problems can be solved, at least partially, by incorporating the drug into a carrier polymer so that the release is slower and the toxicity is reduced. An attempt to increase the specificity and selectivity of neuroactive lipophilic drugs has been made by conjugation of the drug either with a specific ligand or with an antibody toward a protein specifically expressed at the cell surface (3). More recently, a class of lipophilic compounds, neurosteroids, i.e., steroids known to be particularly effective in the nervous system, were found to influence the brain's functions significantly, memory in particular. These agents do not have a problem in crossing the BBB or in specifically binding to their receptors. Studies by Toran-Allerand and colleagues showed that estrogen receptors are localized in central cholinergic neurons, and that signaling pathways activated by growth factors can be also activated by estrogens (25,26). Neurosteroids have been tested in several models (27), and numerous studies are currently underway to provide a proof of concept for neurosteroids as potential therapeutics in neurodegenerative diseases (27).
II. NONVIRAL DELIVERY SYSTEMS TO DELIVER AGENTS TO THE NERVOUS SYSTEM

Although the expression of specific proteins by transfec-tion with viral vectors has been a commonly used technique, this method of drug delivery has certain disadvantages (28-30). A number of nonviral approaches to drug delivery to the nervous system have been developed, including (i) intraventricular infusion of neuroactive agents, (ii) injection or implantation of polymeric systems, (iii) implantation of genetically engineered cells or stem cells, and (iv) use of liposomes. These approaches are summarized in the following sections.

A. Intraventricular Infusion of Neuroactive Agents

Poorly soluble agents and unstable peptides are often administered into the lateral ventricles either as single injections or via permanently installed cannulae (31,32). The advantages of these approaches are that the dosage and rate of drug administration can be controlled, and the results resemble a slow intravenous infusion if the drug is readily distributed into the peripheral bloodstream. However, in-tracerebroventricular (ICV) injection of drug results in distribution to the ependymal surface of only the ipsilateral brain because of the unidirectional flow of cerebrospinal fluid within the brain. The major disadvantage of ICV drug administration is its invasiveness and the possibility of infection at the site of penetration of the BBB.

B. Injectable and Implantable Polymeric Systems as Drug Carriers

1. Drug-Polymer Conjugates

Synthetic polymer materials have been used as drug carriers in several modalities (Fig. 1). Injectable drug-polymer conjugates are produced by covalent binding of water-soluble polymers to a drug. The nature of the covalent bond between the drug and the polymer should be such that the bond is strong enough to be stable in the bloodstream but easily cleaved once the conjugate has reached the target site. This is often difficult to achieve. Moreover, only a relatively small number of biologically active molecules can be attached to the polymer molecule, thus requiring relatively large amounts of drug-polymer conjugate to be injected at the site of action. Approaches overcoming some of these problems are discussed in the following sections.

2. Implants

Simple replacement therapy with polymeric implants of nerve growth factor have been implemented in animal

Figure 42.1 Some common approaches to administer neuroactive drugs. 1. Drug covalently bound to the polymer. 2. Micro-spheres (made of biodegradable polymers) containing neuroactive agents can be injected either systemically, into the lateral ventricle, or into the selected brain structure. 3. Microsponges can be impregnated with neuroactive agent and administered locally. 4. Osmotic pumps allow for steady release of neuroactive agent for a prolonged time period (1-2 weeks). 5. Injections of neuroactive agents directly into the lateral ventricle or parenchyma.

models of central cholinergic deficiencies (33,34) and of peripheral nerve impairment in diabetes (24), in both humans and several animal species (35,36). Recently nerve growth factor (NGF) was delivered locally by implantation of a small polymer pellet providing slow release at a controlled distance from the target site (37). The implants placed 1-2 mm away from the target cholinergic site were effective, whereas the same implants placed 3 mm away from the target site had no detectable effect. These findings strengthen the notion that NGF delivery within a spatially restricted area should be considered a desirable feature if the drug is to be effective. Due to the larger size of the target areas in the human brain than in rodent animal models, the concept of pharmacotechtonics has been tested. This strategy involves the creation of an array of local drug-releasing loci to create large but spatially restricted and anatomically defined fields of biological activity. Drug distribution can be more controlled, and moreover this approach lends itself to comparison with mathematical models (38). The geometry and sites of implantations can be determined by noninvasive diagnostic procedures, such as MRI, prior to the surgical procedure. Local delivery, in conjunction with pharmacokinetic modeling (39) and ste-reotaxic atlases linked to MRI scanners, will eventually allow for customized drug therapy for individual patients. Replacement of other factors such as ciliary neurotrophic factor (CNTF), lymphocyte inhibitory factor (LIF), and brain derived neurotrophic factor (BDNF) has also been

1085

achieved in different ways. However, all of these simple replacement approaches have three major limitations: (i) site-specific delivery, (ii) the amount of drug that can be administered by single administration, and (iii) susceptibility of full-length peptides to enzymatic cleavage due to the presence of various peptidases in the tissue. To solve some of the stability problems, drugs can be incorporated into biodegradable polymers, and an overview of these polymers is given in Chapter 5.

3. Osmotic Pumps

Osmotic pumps are also often used in experimental animals. For instance, implantable pumps have been used in primates to deliver dopamine or dopamine agonists (40,41). The pump reservoir is typically installed subcuta-neously, and a catheter links a cannula with the pump. There are different sizes of pumps, suitable for small rodents or larger animals (commercially available "Alzet" minipumps); the pumps are refillable, and newer models allow for the adjustment of the delivery rates. The major limitation of pumps is the possibility of a local immune reaction at the site of delivery. In addition, due to the limited diffusion of most peptidergic neurotrophic agents, the majority of the agent is degraded before reaching the intended site of action.

4. Micro- and Nanoparticles

These delivery systems were reviewed previously (4,42-48). Controlled release polymer systems not only improve drug safety and efficacy but may also lead to new therapeutics. Some of the frequently used polymers are poly(sebacic acid-co-1,3-bis(p-carboxyphenoxy)propane), poly(b-hydroxybutyrate-hydroxyvalerate), poly(lactide-co-glycolide), poly(methyl methacrylate-acrylic acid), poly(acrylamide-co-acrylic acid), and poly(fumaric-co-se-bacic anhydride). Numerous micro- and nanoparticles (some examples of which are shown in Fig. 2) have been designed and tested in vitro and in vivo to demonstrate superior effectiveness with concomitant reduction in neu-rotoxicity. Conventional oral or transdermal delivery is inadequate for the delivery of macromolecules such as proteins. Due to the short half-life of macromolecules such as growth factors, micro- and nanocontainers made of different polymers have been investigated as a means of their controlled and prolonged release. Johnson et al. (49) developed biodegradable microspheres composed of lactic co-glycolic acid polymer in which lyophilized macromolecules (human growth hormone) were complexed with zinc to solve the problem of moisture-induced protein aggregation. The system was tested in vitro and in a primate in vivo model. A release of the protein for one month was demonstrated, suggesting the possibility that such a system

Figure 42.2 Some examples of injectable nanoparticles as carriers for neuroactive agents.

may be considered for chronic clinical use. Numerous other nano- and microparticulate biodegradable and bio-compatible delivery systems have been developed in the last several years. For instance, rhodium (II) citrate, a recent member of promising antitumor agents, was com-plexed and encapsulated into poly (D,L-lactic-co-glycolic) acid (PLGA) and poly(anhydride) microspheres (50). Complexation in this case significantly increased the encapsulation efficiency and duration of release in both polymer systems (50). However, problems that need to be dealt with include the limited supply of the neuroactive agent, the invasive aspects of micro- and nanoparticle administration, and release kinetics that were not amenable to regulation by physiological changes at the site or elsewhere. Two interesting and novel approaches have been recently considered and tested: controlled release microchips and neu-rospheres. Briefly, in contrast to previous methods of controlling drug release from polymeric devices such as pulsatile stimuli by an electric or magnetic field, exposure to ultrasound, light, enzymes, changes in pH or in temperature, new biotechnological approaches have led to the development of a solid-state silicon microchip that can provide controlled release of a single or multiple agents on demand (51). Although it is too early to evaluate its usefulness for the delivery of neuroactive substances, it certainly seems promising. Neurospheres of multipotent and restricted precursors may provide solutions for a longer lasting and more physiological supply of biologically active compounds, either singly or in combination (52-54).

5. Liposomes

Cationic liposomes may have a significant potential for clinical applications in gene therapy for the disordered central nervous system (CNS) (55). Recently it has been reported that intracerebroventricular or intrathecal injection of cationic liposome-DNA complexes can produce significant levels of expression of biologically and therapeuti-cally relevant genes within the CNS such as nerve growth factor (NGF), granulocyte colony-stimulating factor (G-CSF), and choline acetyltransferase (ChAT) (56). Technical aspects to achieve maximal gene transfer into brain cells using a plasmid DNA-cationic liposome complex have been discussed by Imaoka et al. (57). These authors have administered plasmid DNA-cationic liposome (lipo-polyamine of dioctadecylamidoglycyl spermine) complex to 3-6 months old male rats using an osmotic pump. They report an increase of approximately up to two orders of magnitude in transfection efficiency compared to one obtained by a single injection. The authors propose that the continuous injection approach may be safe and effective in increasing the transfection efficiency. Another group led by Yokota (58) examined the effects of a calcium-dependent cysteine protease (calpain) inhibitor entrapped in liposomes in delaying neuronal death in gerbil hippocampal CA1 neurons following a transient forebrain ischemia. Selective neuronal damage induced by forebrain ischemia in the CA1 region of the hippocampus, and calpain-induced proteolysis of neuronal cytoskeleton, were prevented by administration of the inhibitor in a dose-dependent manner (58). Evaluation of transfection efficacy of a plasmid vector complexed with three different cationic liposomes into two experimental rodent and human malignant glioma cell lines and the mouse 3T3 fibroblast were studied by Bell et al. (59). The transfection efficacy and cytotoxicity of the liposomes were reported to vary quantitatively and qualitatively between cell lines. These authors suggest that their results support a potential application of cationic liposomes in both experimental and human malignant glioma gene transduction. Further studies on liposomal transfection of normal and neoplastic cells derived from the CNS will likely be very useful in helping to ascertain the particular merits of liposome-mediated gene transfection (59). Although the emphasis has been on utilizing liposomes in gene delivery to the CNS, this by no means limits their use to gene transfection (60-63).

C. Therapeutic Approaches Employing Cells

1. Genetically Engineered Cells

In order to provide longer term neurotrophin delivery without the need to refill the containers or reduce the frequency of reimplantation of delivery devices, several groups (5,64-66) have developed implantable polymeric devices containing genetically engineered cells that can produce, for example, a missing trophic factor (Fig. 3). This strategy has been tested in animal models, including primates (67). Either primary cultures or genetically engineered cells producing a missing factor can serve as "long term effective mini-factories," and various cell types used for these pur-

Figure 42.3 Genetically engineered cells and stem cells. Different cell lines, primary cell cultures, and genetically engineered cells producing a neuroactive agent can be directly injected into the brain as a cell suspension, or prior to administration cells can be microencapsulated in biocompatible polymers. Neural stem cells with the capacity to renew and produce the major cell types of the brain can be used for cell replacement therapy in neurological disorders. (See Section C.3, Stem Cells.)

poses have been reported and reviewed, including pheo-chromocytoma cells (PC 12) (68), fibroblasts, and NIH 3T3 cells genetically altered to produce growth factors (66,69,70). Although fully mature primary cultures or genetically engineered proliferating cells of nonneuronal origin can replace missing peptides, they are either (i) deliberately physically separated from the environment at the implantation site to prevent tumor formation (e.g., by encapsulation or by placement of cells within a retractable implantation device) or (ii) in contact with the immediate microenvironment, their phenotype not allowing them to integrate and make functional connections (e.g., PC 12 cells, fibroblasts).

3. Stem Cells

Replacement strategies using stem cells have recently become an attractive way to overcome the problems of cell integration and of acquisition of normal brain functions (71). Adult CNS stem cells can replace neurons and glia in the adult brain and spinal cord (72) and can also give rise to other cell types such as skin melanocytes and a range of mesenchymal cells in the head and neck (73). Stem cells may integrate appropriately into both the developing and the degenerating central nervous system and may be uniquely responsive to some types of neurodegenera-tive conditions (74). Neural-derived stem cells are self-renewing under the influence of mitotic agents such as fibroblast growth factor (75), epidermal growth factor (76,77), BDNF (78), and other factors (71,79-82). These cells can differentiate into either neuronal or glial cells and therefore can be used to replace neurons that are damaged or destroyed in defined neuronal structures, such as dopa-minergic nigral neurons in Parkinson's disease, or hippo-