Figure 1 gives an overview of the two main immunological barriers, namely BBB and BCSF and their different components. We can see how BBB acts as a neuroprotective shield by protecting the brain from most substances in the blood, supplying brain tissues with nutrients, and filtering harmful compounds from the brain back to the bloodstream 30 . BBB is constituted by the brain endothelial cells which form the anatomical substrate called cerebral microvascular endothelium. It regulates the transport of solutes and other substances including drugs in and out of the brain, leukocyte migration, and maintains the homeostasis of the brain microenvironment, which is crucial for neuronal activity and proper functioning of CNS. The cerebral microvascular endothelium, together with astrocytes, pericytes, neurons, and the extracellular matrix, constitute a "neurovascular unit" that is essential for the health and function of the CNS 31 . The transport of solutes and other substances across BBB is strictly constrained through both physical tight junctions (TJs) and adherents junctions (AJs) and metabolic barriers (enzymes, diverse transport systems) and hence excluding very small, electrically neutral and lipid soluble molecules. Thus, conventional pharmacological drugs or chemotherapeutic agents are unable to pass through the barrier.

TJs between endothelial cells of the BBB possess also an intricate complex of transmembrane proteins (junctional adhesion molecule-1 (JAM-1), occludin, and claudins) with cytoplasmic accessory proteins (zonula occludens-1 and -2 (ZO-1 and 2), cingulin, AF-6, and 7H6) and hence acts as physiological and pharmacological barrier, thereby preventing influx of molecules from the bloodstream into the brain. As shown in Figures 1 and 2, BBB is characterized by two membranes, namely luminal and abluminal, facing blood capillary and brain interstitial fluids (ISF), respectively. Another especial feature of BBB is the structural differences that exist between the endothelia of the brain capillaries and endothelia in other capillaries, such as tight junctions between adjacent endothelial cells 31 32 , a lack of fenestrations (perforations) and a lack of pinocytotic vesicles 33 34 35 36 37 38 . Furthermore, in addition to the BBB and BCSF, there exists other CNS barrier shielding the delicate brain tissue from the outer world, but which may play a role in drug transport, such as the blood tumour barrier and the blood retina barrier 39 40 , formed of pigment epithelium enclosing the retina, and thereby acting as a barrier interface between the systemic blood vessels of the neighbouring choroid and the retina. Finally targeting of tumour tissue is often constricted by the blood tumour barrier 39 .

Moreover, BCSF is the second important feature of the CNS next to the BBB, and is formed by the epithelial cells of the choroid plexus. BCSF controls the penetration of molecules within the interstitial fluid of the brain parenchyma by closely regulating the exchange of molecules between the blood and CSF. Previous reports have demonstrated the following mechanisms of transport pertaining to the choroid plexus: facilitated diffusion (efflux) and active transport into the CSF, as well as active transport (efflux) from CSF to the blood 41 42 43 .

The treatment of intractable CNS disorders such as HIV, dementia, epilepsy, CNS-based pain, meningitis and brain cancers depend mainly on the ways to achieve higher drug concentration in the targeted tissues of the brain. The ability of a substance to penetrate the BBB or be transported across BBB is mainly dependent on its physiochemical properties. The total brain exposure, and thus the pharmacological efficacy of a drug or drug candidate, depends on its drug uptake which in turn depends on a combination of factors, including the physical barrier presented by the BBB and the BCSF and the affinity of the substrate for specific transport systems located at both sides of these interfaces 26 27 . The efflux transporters present in the BBB and BCSF, limit brain penetration as well as the intra- and extracellular distribution of a variety of endogenous and exogenous compounds 28 .

The efflux transporters role, both as a homeostatic agents against endogenous substances and protective agents against the exogenous substances, have been extensively studied and three classes of transporters have been implicated in the efflux of drugs from the brain: multidrug resistance transporters, monocarboxylic acid transporters, and organic ion transporters 44 . Kabanov et al. 45 have reviewed the inhibition of efflux transporters by Pluronic^® block copolymers to enhance the penetration of drugs for CNS delivery. Drug efflux transporters not only cause elimination of the drugs from the brain, but also affects its absorption and tissue distribution 46 . Owing to the growing emphasis on identification and discovery of influx transport proteins (from blood to brain) and efflux transport proteins (from brain to blood) in last years, BBB is now considered to be a dynamic interface that controls the influx and efflux of a wide variety of substances, including endogenous nutrients and exogenous compounds to maintain a favourable environment for the CNS 47 .

Deguchi and co-workers demonstrated that the rat organic anion transporter 3 (rOat3) mediated brain-to-blood transport of uremic toxins, as well as that rat organic anion transporting polypeptide (rOatp2) is involved in efflux of 3-carboxy-4-methyl-5-propyl-2-furanpropionate 48 . Sun and co-workers investigated the transport of carbamazepine and drug interactions with cultured rat brain microvascular endothelial cells (rBMEC) as an in vitro model of the BBB 49 . They concluded that some specific ABC (ATP-binding cassette, ABC) efflux transporters may be involved in the transport of carbamazepine across the BBB.

The fact that many of the lipophilic drugs show negligible brain uptake can be attributed to the substrates of drug efflux transporters such as the organic anion transporting polypeptides and the BBB active drug efflux transporters of the ATP-binding cassette gene family, e.g. P-glycoprotein (Pgp), multidrug resistance proteins (MRPs) and breast cancer resistance protein (BCRP) 45 48 50 51 52 53 , that are overexpressed by the endothelial or epithelial cells of these barriers 52 . The combined action of these carrier systems results in rapid efflux of xenobiotics from the CNS and they also account for the cellular localization, specificity, regulation, and potential inhibition at the BBB and BCSF barriers.

Efflux transporters act as a major impediment factor to CNS access by restricting a number of solutes. The future of CNS drug delivery is highly dependent on novel strategies towards modulation of these efflux transporters by designing nanocarriers with tuned affinity for these transporters 45 52 53 . The following section brings a more detailed account of transport mechanisms.

Receptor mediated endocytosis (RME) or clathrin-dependent endocytosis is a highly specific and energy mediated transport enabling eukaryotic cells to selective uptake macromolecules as specific cargo. For the BBB receptor-specific ligands have also been shown to be very effective to transport endogenous peptides like insulin and transferrin, albumin, and opioid peptides e.g. deltorphins, [D-penicillamine 2,5] enkephalin (DPDPE) and deltorphin II 68 69 70 71 72 . That is why receptor-mediated drug delivery, is also a promising Trojan horse approach for the release of therapeutics into neuronal cells, and tissues. Nanocarriers conjugated to different types of ligands of cell surface receptors expressed on brain endothelial cells, can accumulate and eventually be internalized by cells on the vascular side of the brain through the mechanism of receptor-mediated endocytosis. By direct and indirect conjugation of endogenous and chimeric peptides to nanocarriers or receptors of BBB, significant improvement in drug delivery has been reported 66 73 74 . The desirable fate of targeted receptors after endocytosis can be seen in the in the following way. Upon binding to the receptors, the ligand conjugated nanocarrier gets collected in specialized areas of the plasma membrane known as coated pits. These clathrin coated pits invaginate to form coated vesicles, upon endosomal processing of the vesicle, clathrin and associated proteins dissociate from the vesicle membrane (early endosome), to form new coated pits at the cell surface 70 . The receptor dissociates from the ligand conjugated nanocarrier due to the acidification of the vesicle in the late endosome, and the nanocarrier complex degrades, hence releasing the drug to the cell.

In addition three different mechanisms supporting the ligand conjugated nanocarrier based transport of drugs such as neuropeptides have been proposed: (i) the adsorption of uptake promoting apolipoproteins, (ii) the modulation of tight junctions, and (iii) the inhibition P-glycoprotein, playing a key role in drug resistance 75 . Kreuter et al. suggested that the apolipoproteins B and E may be chiefly involved in the transport of NP-bound drugs into the brain. They concluded that by coating the NPs with polysorbate 80, apolipoproteins B and E get adsorbed onto the NP surface from the blood after injection and thus seem to mimic lipoprotein particles that could be taken up by the brain capillary endothelial cells via receptor-mediated endocytosis 76 77 78 .

After endocytosis, drugs may be released within the endothelium cells and undergo further transportation into the brain by diffusion or through transcytosis 27 . For instance, Liu et al. used chelator-NP system and the chelator-NP system complexed with iron to devise effective therapeutic strategy for Alzheimer's disease which is characterized by dyshomeostasis of metal ions with abnormally high levels of iron in affected areas of the brain 79 . They reported preferential adsorbtion of apolipoprotein E and apolipoprotein A-I in the in vitro studies, thereby suggesting the RME transport of chelators and chelator-metal complexes by the NPs across the BBB. Further studies are needed to investigate whether these metal chelators conjugated to NPs can play a role in solubilizing amyloid- [beta] deposits in Alzheimer disease. This can open new pathways to the treatment of neurodegerative diseases and also to study the ways of neural repair using efficiently conjugated nanocarrier system.

Kim et al. recently reported the blocking of low-density lipoprotein receptors (LDLR). Their study is based on brain endothelial cells involving cellular internalization of Poly(methoxy-polyethyleneglycol cyanoacrylate-co-hexa-decyl-cyanoacrylate) (PEG-PHDCA) NPs preincubated with apolipoprotein E. It strengthens the hypothesis of the preponderant role of the LDLR-mediated transport in the endocytosis of PEG-PHDCA NPs 80 . Using protamine-oligonucleotide NPs (proticles) coated with Apolipoprotein A-I (apoA-I), Kratzer et al. observed increased particle uptake and transcytosis in an in vitro model of the BBB 81 . These findings were further supplemented by Petri et al. who used Poly(butyl cyanoacrylate) NPs coated with poloxamer 188 (Pluronic^® F68) bounded to doxorubicin and reported enhanced anti-tumour effect of doxorubicin against an intracranial glioblastoma in rats 82 . They hypothesized that this may be facilitated by the interaction of apolipoprotein A-I, present on the surface of the NPs, with the scavenger receptor class B, type I, the prime receptor for high density lipoprotein/apoA-I that is expressed on brain capillary endothelial cells (BCEC) 81 . Further research is required to reveal the mechanisms behind the interaction between SR-B1 and apoA-1 and their possible role in enhancing the drug delivery via RME pathway. Moreover, the possibility of more than one mechanism, implicated in the interaction of nanocarrier based drug delivery systems with the brain endothelial cells, cannot be ruled out 83 .

In a novel approach, Demeule et al. reported the design of a family of Kunitz domain-derived peptides called Angiopeps as a potential brain drug delivery system. Using a in vitro model of the BBB and in situ brain perfusion, they demonstrated that these peptides, and in particular Angiopep-2, exhibited higher transcytosis capacity and parenchymal accumulation than other receptors such as transferrin, lactoferrin, and avidin. Furthermore, they suggested that the Angioprep-2 endocytosis may be mediated by the low-density lipoprotein receptor-related protein-1 (LRP1) 84 .

Adsorptive-mediated endocytosis (AME) is a transport mechanism that has gained significant importance recently, and many new drug delivery technologies focus on AME 61 85 . The underlying principle of AME based transport is the electrostatic interaction between a positively charged substance (e.g. cationized peptide such as albumin) and the negatively charged sites on the brain endothelial cell (BEC) surface (e.g. glycoprotein) 61 86 . Dos Santos et al. studied the nature and distribution of anions on the BEC surface in vitro and in situ and found that the the predominant anion detected on BEC was heparan sulphate (HS) in comparison to the anionic locations observed in endothelia from aorta and epididymal fat micro-vessels 87 .

The hypothesis that phagocytic cells of the innate immune system, mainly neutrophils and monocytes, can be exploited as transporters of drugs to the brain has been studied by Afergan et al. in vitro, in rats and rabbits by using negatively-charged nano-sized liposomes with double-radiolabeled 3H (in the membrane) and 14C-serotonin (in the core), and fluorescent markers (membrane and core) 88 . They observed a higher brain uptake of liposomal serotonin, 0.138% ± 0.034 and 0.097% ± 0.011, vs. 0.068% ± 0.02 and 0.057% ± 0.01, 4 h and 24 h after IV administration in rats, serotonin liposomes and in solution, respectively. They concluded that monocytes act as key players for the transport of serotonin liposomes.

Alkaloids like cocaine are well-known stimulants of the central nervous system, and its effect upon the BBB has been studied extensively. Alas, little exploited for drug delivery, it actually relaxes tight junctions and induces leukocyte migration. For instance, Liu et al. reported enhanced BBB permeability and pharmacological activity of the endogenous opioid receptor agonist, endomorphin (EM)-1 68 . A series of EM-1 analogs were tested, e.g. N-terminal cationization, C-terminal chloro-halogenation, and unnatural amino acid (D-Ala, Sar, and D-Pro-Gly) substitutions in position 2. They found that in comparison with EM-1, the four D-Ala-containing tetrapeptides and the chloro-halogenated D-Pro-Gly-containing pentapeptide elicited significant and prolonged central-mediated analgesia upon subcutaneous administration. This fact might be interpreted as more peptides reaching the CNS, thus bringing greater analgesic effect. They also reported that the guanidino- [D-Ala2, p-Cl-Phe4]EM-1 showed 3 times more analgesia than the parent peptide following intra cerebral-ventricular injection.

Adsorptive-mediated transport (AME) based transport has been exploited to facilitate gene delivery into brain tumours. Lu et al., for instance, has incorporated plasmid pORF-hTRAIL (pDNA) into cationic albumin-conjugated PEGylated NPs (CBSA-NP) to evaluate the efficacy of CBSA-NP-hTRAIL as a nonviral vector for gene therapy of gliomas 86 . They observed that 30 minutes after IV administration of CBSA-NP-hTRAIL to BALB/c mice bearing IC C6 gliomas. These NPs co-localized with glycoproteins in brain and tumour microvasculature. And, more importantly, cells accumulated in tumour cells. In addition, they reported apoptosis of brain tumour cells in vivo and significantly delayed tumour growth. The above results suggest adsorptive-mediated transport is a very promising route of drug and gene delivery across BBB for CNS disorders. More investigation is required to explore other anionic sites on the BEC surface that can be used to design efficient strategies for delivery using nanocarrier systems through adsorptive-mediated transport. Despite of possessing a lower affinity than RME, AME provides a higher capacity than receptor-mediated endocytosis.

As a field on its own, nano-drug delivery requires proper functionalization, profound knowledge of the range of possible routes to and from the central nervous system, as well as ways to verify whether drugs and nanocarriers reach their final destination. We proceed to review some of the most exciting trends in functionalization, delivery and imaging of nanomaterials.

One of the most important challenges in nano-based diagnostics and drug delivery is the functionalization of NPs. Firstly, we need to develop effective conjugation strategies to combine, in a highly controlled way, specific biomolecules to the surface of NPs. Figure 4 shows an example of a PEGylated, multilayer NP (polyethylene glycol, PEG, a popular choice for biocompatible nanocarriers.

Some of the most prominent candidate biomolecules are cell penetrating peptides (CPP) such as SynB vectors, penetratin and Tat that facilitate enhanced intracellular delivery 91 92 93 94 95 , fluorescent dyes (rhodamine, alexa, Cy5.5), tumoural markers for brain and gene therapeutic agents for genetic therapy such as siRNA 96 97 98 99 100 101 . Figure 5 show two kinds of mouse tumour models, namely Xenograft and genetically engineered mouse model (GEMM) 102 .

Functionalization itself requires a profound knowledge of the target organ and its transport mechanisms. The BBB has several transport molecules that can potentially increase the efficiency and kinetics of nanocarriers towards brains 103 , such as, growth factors (e.g. epidermal growth factor 58 , vascular endothelial growth factor 104 , basic fibroblast growth factor 105 , insulin-like growth factors (IGF-I and -II) 106 ), biotin-binding proteins (avidin, streptavidin, or neutravidin) 107 , insulin 59 69 , albumin 108 109 110 , leptin 111 112 , lactoferrin 103 113 , iron binding protein p97 (melanotransferrin) 114 , transferrin 68 115 and Angiopep-2 84 . Some agents play a pivotal role in enhancing the permeability of nanoprobes through BBB 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 . A list of agents/condition and their effects on BBB are summarized in Table 1.

Moreover, by altering the surface of polymeric NPs on coating them with different hydrophilic surfactants, such as polysorbate 80 (Tween^® 80) or other polysorbates with 20 polyoxyethylene units, biocompatible coatings of non-viral gene delivery systems e.g. by poly ethylene glycol (PEG) attachment for siRNA delivery show significant advantage in brain targeting 98 .

Due to its high specificity, NP provides an ideal platform for the transport of drugs across the BBB. The current status of some NP drug delivery platforms and the corresponding encapsulated drugs is summarized in Table 2. All entries refer to in vivo experiments.

NP-mediated drug transport to the brain strongly depends on the type of surfactant. In Kreuter and Schroeder's labs, 12 different surfactants were coated onto the surface of poly (butylcyanoacrylate) (PBCA) NPs were injected intravenously into mice to evaluate the influence of surfactant on the analgesic effects. The authors reported that only the NPs with polysorbate 20, 40, 60 and 80 coatings produced significant effect and the maximum effect was observed for the PBCA NPs bearing polysorbate 80 coating 133 134 135 136 . PBCA NPs coated with surfactants have been successfully used in the delivery of number of drugs across the BBB 137 138 , including the peptides (hexapeptide dalargin and the dipeptide kytorphin), anti-tumour antibiotic doxorubicin (DOX), loperamide, the NMDA receptor antagonist MRZ 2/576, and tubocurarine 137 138 139 140 141 142 .

Calvo et al. employed a novel strategy by using PEGylated polycyanoacrylate, NPs (PEG PHDCA) as vector for drug delivery in experimental model of Prion disease 143 . The work showed that the PEG PHDCA particles produced a higher uptake by the spleen and the brain which are both the target tissues of PrPres (an abnormal isoform which is characterized by the accumulation of the host-encoded Prion protein (PrP) in the brain of experimental Prion diseases mice) in comparison to the non-PEGylated NPs. Wilson et al. have used polymeric NPs for drug delivery of anti-Alzheimer's drugs such as tacrine and rivastigmine in the brain of rats 144 145 .

Toxicity of conjugated drug-nanocarriers has always been a concern. Gelperina et al. 139 studied the toxicity of DOX bound to polysorbate 80-coated PBCA NPs in healthy rats, and rats with intracranial glioblastoma. No drug-induced mortality occurred with a dose of 3 × 1.5 mg/kg of the DOX NPs formulation on days 2, 5, 8 after tumour implantation. They concluded that the toxicity of DOX bound to NPs was similar, or even lower, than that of free DOX. Other studies aimed at investigating the toxicological profile of doxorubicin bound to NPs employing different dose regimens correlates with the results of this study 146 . Based on the above findings, Pereverzeva et al. hypothesized that the lower toxicity of the nanoparticulate formulation may be due to the altered biodistribution of the drug mediated by the NPs 146 . Wang et al. applied a unique 1% polysorbate-80 coated gemcitabine PBCA NPs (GCTB-PBCA-NPs) to investigate its inhibitory effects in C6 glioma cells in vitro and in vivo with Sprague Dawley rats 147 . They observed significant increase in the survival time of the rats injected with the formulation compared with the saline control (P < 0.05).

In an interesting approach, You et al. 148 investigated feedback regulated paclitaxel delivery by using pH-Sensitive poly (N, N-dimethylaminoethyl methacrylate (DMAEMA)/2-hydroxyethyl methacrylate (HEMA)) NPs for the triggered release of paclitaxel within a tumour microenvironment. Driven by the fact that the tumours exhibit a lower extracellular pH than normal tissues, the authors found that the paclitaxel release from DMAEMA/HEMA particles can be actively triggered by small, physiological changes in pH (within 0.2-0.6 pH units). It seems to be a promising way to facilitate drug delivery by regulating the tumour microenvironment. Further studies are thus required to explore other factors within tumour microenvironment that can be exploited to enable controlled release of drugs in brain tumours.

Liposomes and related lipid structures have long been employed for drug delivery. Lipid NPs, however, are alternative carrier system to traditional colloidal carriers, such as emulsions, liposomes and polymeric particles. These novel carriers have been employed for brain tumour targeting purposes and reviewed in 76 . NPs based on solid lipids come in different types such as "solid lipid NPs" (SLN), "nanostructured lipid carriers" (NLC) and "lipid drug conjugate" (LDC) 149 150 . The breakthrough in advanced conjugation strategies have further led to the emergence of the newer forms of SLN such as polymer-lipid hybrid NPs, nanostructured lipid carriers and long-circulating SLN 151 . Because of its physiochemical characteristics, SLNs have been very successful in comparison to polymeric NPs due to the lower cytotoxicity, higher drug loading capacity, and best production scalability 152 .

Back in 2002, Olbrich et al. reported, for the first time, the use of LDC-polysorbate 80 NPs for brain delivery of diminazene to treat second stage human African trypanosomiasis (HAT) 153 . They obtained NPs with a very high drug load of 33% (w/w), despite of the highly water-soluble drug diminazenediaceturate. They concluded that by transforming water-soluble hydrophilic drugs into LDC, NPs got prolonged drug release and targeting to specific sites by intravenous injection. In an another study published shortly afterwards, Wang et al. synthesized 3',5'-dioctanoyl-5-fluoro-2'-deoxyuridine (DO-FUdR) and incorporated it into solid lipid NPs (DO-FUdR-SLN) by a thin-layer ultrasonication technique in order to deliver the drug 5-fluoro-2'-deoxyuridine (FUdR) to the brain. With the average particle size of 76 nm, drug loading of 29.02% and entrapment efficiency of 96.62%, DO-FUdR-SLN proved to be very efficient in in vivo brain targeting 154 .

More recently, Kuo et al. evaluated the permeability of anti-human immunodeficiency virus (HIV) agents, including stavudine (D4T), delavirdine (DLV), and saquinavir (SQV), across an in vitro model of BBB and incorporating them with PBCA NPs, methylmethacrylate-sulfopropylmethacrylate (MMA-SPM) NPs, and SLNs. Their experimental results revealed an enhanced BBB permeability 155 . Their work suggests that the PBCA, MMA-SPM, and SLNs seem promising for the drug delivery and clinical applications in neuro-AIDS treatment.

No review of drug delivery across BBB is complete without looking at the broad picture of administration routes. A direct drug administration to the brain region, painless and safe, will definitively improve the scenario. However, in the meantime intravenous administration is most popular choice in clinical studies. Some approaches, however, that have been gaining considerable attention, such as oral route, inhalation or intra-tracheal instillation (IT), intranasal drug delivery, convection-enhanced diffusion and intrathecal/intraventricular drug delivery systems in addition to the conventional modes like intravenous administration. Therefore, the administration route of NPs becomes an important criterion of consideration so as to overcome the physiological barriers of the brain and to achieve high drug concentrations therein 58 156 157 158 159 160 161 .

Interestingly, Semmler-Behnke et al. have recently reported the uptake of 1.4 and 18 nm gold NPs in secondary target organs like the brain following intra-tracheal or intravenous application 158 . Moreover, Wang et al. used fluorescence-labeled bovine serum albumin (FBSA) loaded in biodegradable poly(lactic acid-co-glycolic acid) (PLGA) for intraspinal administration of Glial cell line derived neurotrophic factor (GDNF) following contusive spinal cord injury (SCI) and for in vitro study 162 . PLGA-FBSA NPs were well absorbed by neurons and glia, indicating that PLGA as a considerable nanovehicle for the delivery of neuroprotective polypeptide into injured spinal cord. Also, local administration of PLGA-GDNF effectively preserved neuronal fibers and led to the hind limb locomotor recovery in rats with SCI. The research opened a new route nanocarrier administration by intraspinal administration.

Two different modes of NP administration in brain tumour mouse models are shown in Figure 5. Once administered the NPs, they reach the site of tumour, and localize it. Once they cross the BBB, the specific ligands or peptides get attached to the specific surface markers expressed on the tumours. Hence, by functionalising NPs with fluorescent dyes could naturally provide in vivo imaging of the ongoing biological events during the drug administration as well may act as potential diagnostics labels for early detection and localization of brain tumours.

Within the requirements of size and charge of effectively deliver drugs via NP carrier systems, there are other challenges that need further attention. Although, much of the work has been focused towards drug delivery with NPs, relatively few studies have focused on the interaction of NPs and their hosts in terms of biodistribution, organ accumulation, degradation and/or toxicology like possible damage of cellular structures or inflammatory foreign body effects. Nanomedicine may find itself at crossroads. It might not be wise to ignore possible adverse effects or toxicity of nanocarriers 4 8 163 164 165 .

Till recently, no pan-European initiative was addressing these concerns. Noteworthy, the European Commission has established the Registration, Evaluation, Authorisation and Restriction of Chemical substances (REACH) which provide safety regulation on substances. Further, Borm et al. have extensively reviewed the potential risks of use of NPs, in a review report commissioned under the European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC) 166 . We do expect similar commissions worldwide shortly. Nanodrug delivery is seen in its infancy, and works are mostly focusing on particular aspects rather than holistic approaches, e.g. ADME or DMPK. Well-established research protocols like absorption, distribution, metabolism and elimination (ADME), and drug metabolism and pharmacokinetics (DMPK) will surely be part of nanodrug delivery research in the near future 167 .

On the distribution side, for instance, Kreyling et al. have extensively studied translocation kinetics and particle size dependency of NPs 77 164 168 169 170 . In general, smaller NPs show superior translocation kinetics. But, because of their small size might on the other hand cause toxicological effects, see a review by Oberdörster 171 . It is all about a trade-off between drug potency and immunologic surveillance. For example, NPs of size <100 nm need to be used to circumvent macrophage clearance in the lungs 172 . Furthermore, several authors have reported that intrinsic characteristics of NPs, such as aspect ratio and surface area, can be pro-oxidant and pro-inflammatory 31 165 173 174 . Here, the ultra high surface to mass ratio together with new, and often unexpected nanosize specific, material properties related to extreme radii of curvature deserve closer attention 175 176 . Therefore, the use of biopersistent carbon-based, e.g. single or multi-wall carbon nanotubes, or metallic nanocarriers in nano medicine is debatable. These important findings need not discourage genuine efforts in nanodrug delivery, but strength the selection process of materials, shapes and surface treatments 3 4 8 164 . Biodegradable, non-toxic multi-block co-polymers like those based on poly(image-lysine), PEG copolyester and nanogels (e.g. polyethylenimine-PEG) are thus advantageous.

Depending on their functionalization, biodegradable nanocarriers can take a number of paths within tissues. What are the possible trajectories nanocarriers take inside the brain? Pharmacokinetics and excretion are key points that demand an exhaustive research. Figure 6 shows the main ways drugs and nanocarriers take within the extra cellular space of the brain. Following their release, drugs can take different mechanisms and may be transported within (and outside) the brain. One of the mechanisms is their transport by diffusion due to drug concentration gradients as shown in Figure 6 (i); or they may be transported because of the convection due to fluid pressure gradients (ii). Figure 6 (iii, a) shows drug migration into ventricular space via pial or ependymal surface. The drug molecules may also undergo circulation in the sub-arachnoid mater or ventricular spaces (iii, b). Subsequently, it is possible to diffuse back into the brain interstitium (iii, c). The drug molecules may also undergo permeation through the endothelium (iv, a); followed by the circulation in the cerebral blood vessels (iv, b); and eventually may re-enter the brain interstitium by permeation (iv, c) 177 .

The exact path, drugs and biodegradable nanocarriers take, depends on many factors and its in vivo imaging is perhaps the next milestone for nanodrug delivery, as discussed in the following section..

Over the last three decades, X-Ray CT, MRI, and PET have been commonly utilized for visualization of distribution and therapeutic effects of drugs.

X-Ray CT has emerged as a major imaging modality for imaging pharmacokinetics and treatment monitoring, mainly based on indirect tracking of morphological changes. For instance, Rabin et al. reported enhanced in vivo imaging of the vasculature, the liver and lymph nodes in mice using a polymer-coated Bi₂S₃ NP formulation as an injectable CT imaging agent 178 . Maier-Hauff et al. used CT in order to noninvasively monitor the local drug release in a rabbit radiofrequency (RF) ablation model 179 . Overall, the application of NP based imaging probes to X-ray CT imaging could have a significant impact on health care, owing to the ubiquitous nature of CT in the clinical setting as well as the increasing use and development of micro-CT and hybrid systems that combine PET and SPECT with X-ray CT. Most common CT contrast agents are based on small iodinated molecules, which are indeed effective in absorbing X-rays; but nevertheless, their non-specific distribution, rapid pharmacokinetics and low sensitivity have rather limited their targeting performance.

With the distinct advantage of functional-imaging capabilities as well as better contrast among soft tissues in comparison to the CT, MRI has emerged as a tool in oncological imaging and imaging of the diseased nervous system 180 . Yet, MRI has relatively low sensitivity to exogenous agents, therefore the choice of contrast approaches is of paramount importance in the field of in vivo brain imaging. Manganese is gaining importance as T1 contrast neural tracer for MRI. In this role, it was used to study three-dimensional (3-D) connectivity patterns in the rat somatosensory system in vivo 181 . To this end, magnetic NPs (MNPs) are of considerable interest as contrast agents for MRI and carriers for drug delivery 182 . Superparamagnetic iron oxide NPs (SPIONs), paramagnetic contrast agent (gadolinium) or perfluorocarbons have already been established as major players in tracking single or clusters of labeled cells within target tissues 183 . Multifunctional nanoplatforms, based on protein cage architectures loaded with imaging agents (fluorophore and MRI contrast agent) onto cells, have also been developed for both diagnostics and targeted treatment 184 . By including gadolinium-loaded liposomes (GDL) with adeno-associated viral vectors (AAV), real time MRI imaging and tracking of convection-enhanced delivery (CED) of viral vectors to the three different regions of non-human primate brain (corona radiata, putamen and thalamus) was achieved 185 .

Another non-invasive imaging technology, the positron emission tomography (PET), enables visualization of biodistribution of positron emitter-labelled compounds. PET has certain advantages over CT and MRI, because of its high sensitivity. For instance, Ukrami et al. 186 designed labelled lipid NPs to study in vivo distribution of liposome-encapsulated haemoglobin determined by PET. Plotkin et al. 187 employed PET for targeting the intra-tumourally injected magnetic NPs in patients with glioblastoma. Indeed, since its introduction in the late 70's, PET has become a powerful imaging modality with the ability for highly sensitive detection of molecular tracers and is currently utilized in diagnosis, therapy monitoring, and imaging gene expression using diverse reporter genes and probes. However, high costs and other complications associated with PET and SPECT equipment limit their applicability. Moreover, the images acquired by these techniques have poor spatial resolution and hence accurate identification of regions of uptake is difficult to achieve.

In summary, high costs, low sensitivity, and/or low spatial resolution associated with the existing well-accepted clinical imaging modalities promoted the search for new approaches for in vivo visualization of brain-targeting nanocarriers, such as methods based on highly sensitive and specific optical contrast.

Imaging with light has unique advantages associated with simplicity, low-cost and small size of the equipment. Visible and near-infrared wavelengths offer many probing mechanisms and highly specific contrast approaches not available for other modalities. These can be used for variety of interrogations, from intrinsic functional information on blood oxygenation to molecular sensing 188 . The light radiation is non-ionizing, and therefore reasonable doses can be repeatedly employed without harm to the animal or patient. Optical contrast methods offer the potential to differentiate between soft tissues, due to their distinct light absorption spectra otherwise indistinguishable using other modalities. Also, specific absorption by natural chromophores (such as oxy-haemoglobin) allows functional information to be obtained. The use of extrinsically-administered "switchable" and "tumour-selective" fluorescent optical agents further advances the application possibilities by allowing visualization of otherwise invisible cellular and sub-cellular processes 189 190 191 .

During the last decade, a large number of commercially available fluorescent probes and markers are increasingly being offered, from non-specific fluorescent dyes and fluorescent proteins to targeted or activatable photoproteins and fluorogenic-substrate-sensitive fluorochromes to enable a highly potent field for biological imaging. So far, these contrast mechanisms were proven efficient in a number of clinical and small-animal applications, including probing of tissue hemodynamics 192 193 , gene expression profiling 194 , detecting protease up-regulation associated with cancer growth and inflammation 195 196 continuous monitoring of the efficacy of anti-cancer treatments and other therapeutic drugs 197 . Since many of the probes are developed to fluoresce in the near-infrared (NIR) optical window, where optical absorption is very low so that light can penetrate deeply, fluorescence imaging has been successfully translated from a microscopy level to whole body small animal imaging and clinics 198 199 . The combination of such probes with optical imaging may yield a unique, highly sensitive technology for in vivo and real-time imaging of the expression patterns for various enzymes, which are crucially involved in tumour formation and metastasis. A good example are various breast cancer cell lines that have been identified to over-express specific enzymes such as matrix metalloproteinases 200 , which are not over expressed in normal cells.

Despite these advantages, optical imaging is severely limited by scattering: thick tissues diffuse and absorb light and significantly reduce the resolution, penetration capabilities and the overall image fidelity 201 . Even state-of-the-art multiphoton microscopy 202 is usually limited to superficial imaging up to a depth of 0.5-1 mm in most living tissues. Recent efforts to image entire embryos for example required naturally transparent specimen or special chemical treatment to clear them from scattering, which is only suitable for post-mortem imaging. Some other macroscopic photographic approaches like epi-fluorescence suffer from similar light diffusion limitations and therefore have low penetration depth, lack quantification abilities, and overall cannot accurately provide depth and size information 203 . Yet, some diffuse optical tomography (DOT) methods were developed that can provide volumetric images of optical contrast in entire human brain 204 with applications ranging from real-time functional neuro-imaging to the detection of hematomas.

It its more advanced form, fluorescence-mediated molecular tomography (FMT) illuminates the sample under investigation at multiple projections and utilizes mathematical models of photon propagation in tissues to reconstruct the underlying imaging contrast in three dimensions, based on distribution of fluorescent molecular probes or fluorescent proteins 195 196 197 205 . Several different implementations, developed over the past years, have been successfully used to three-dimensionally image bio-distribution of fluorochromes in entire animals, and determine molecular pathways of cancer, neurodegenerative and cardiovascular disease, offering quantitative imaging. Whole-body fluorescence tomography of small animals works optimally in the near-IR region where the lower tissue attenuation allows the penetration of photons over several centimeters 206 , but provides low spatial resolution (e.g. on the order of 1 mm in case of whole-body imaging of mice). Figure 7a gives a general schematic of state-of-the-art free-space FMT scanner for in vivo tomographic imaging of small animals 207 .

FMT systems were so far successfully used in molecular imaging studies of brain disease. In one of the studies 24 , using near-infrared fluorescent molecular beacons and inversion techniques that take into account the diffuse nature of photon propagation in tissue, three-dimensional in vivo images of protease activity in orthopic gliomas were obtained. In this study, 2 × 10⁵ cells (9L or HT1080) were stereotactically implanted into unilateral brain hemispheres of nude mice. Animals were then intravenously injected with the cathepsin-B imaging probe (2 nmol Cy 5.5 per animal). The experiments presented the ability of FMT to three-dimensionally and quantitatively resolve fluorochromes in deep tissues and follow their response over time (Figures 7b and 7c).

Some of optical imaging complications associated with poor spatial resolution and lack of anatomical reference can possibly be mitigated by a marriage between non-invasive optical molecular imaging and other high resolution anatomical imaging modalities such as MRI, or X-Ray CT. The latter combination was recently employed to study the progression of Alzheimer's disease in vivo using a fluorescent oxazine dye to quantify amyloid- [beta] plaques in a transgenic murine model 208 . The authors reported very accurate signal localization and correlation of in vivo results to ex vivo images of excised brain (Figure 7d), thereby emphasizing that FMT is not only a potential tool to study in vivo molecular functions, but it can also provide precise mapping of those functions onto high resolution animal anatomy, simultaneously provided by X-Ray CT (Figure 7g). Furthermore, the CT information was used to build a more precise forward model in the FMT image reconstruction process, which also improved spatial resolution and quantification performance of FMT, as can be seen in Figure 7f that was reconstructed using prior structural information (priors) from CT as compared to Figure 7e made without image priors. Another multimodal imaging study was demonstrated by McCann et al. where FMT and MRI were combined to study structure and function of small rodents 209 . Three-dimensional multimodal images were fused to provide a volumetric model of living mouse brains. Interestingly, this approach allows continuous monitoring of tumour morphology, progression and protease activity.

The main challenge for optical imaging of diffuse tissues is degradation of the spatial resolution, which is always exchanged for penetration. As the size of the imaged object grows, imaging resolution quickly deteriorates 210 . It is therefore possible to perform optical tomography, e.g. FMT, through entire mice with high sensitivity, but low resolution of about 1 mm or worse 188 211 . Optoacoustic (or photoacoustic) tomography is an alternative hybrid imaging modality that has recently demonstrated unprecedented high-resolution visualization of optical contrast deep in tissues of small animals 25 212 213 . Optoacoustic imaging relies on detection of ultrasonic signals induced by absorption of pulsed light, thus, high optical absorption contrast can be simultaneously combined with good spatial resolution of ultrasound, not limited by light scattering in tissue. The amplitude of the generated broadband ultrasound waves reflects local optical absorption properties of tissue. Unlike classical optical imaging, the spatial resolution here is not determined nor limited by light diffusion; therefore such performances cannot be achieved by any other optical imaging technology developed so far. Originally, optoacoustic imaging of tissues targeted endogenous tissue contrast, primarily resolving oxy- and deoxy-hemoglobin and different vascular structures. Wang et al. demonstrated high resolution imaging of vascular anatomy in the mouse brain with capability to visualize, with high spatial resolution, functional parameters, e.g. blood oxygenation levels, deep in an intact living mouse brain 212 . Much like the ultrasound, optoacoustics can form images in real time, currently in 2D but potentially also in 3D 214 . In this way, it can be used for real time tracking of dynamic phenomena, such as fast hemodynamic changes 215 , bio-distribution of diagnostic agents or pharmacokinetics. However, recently good contrast was also obtained from other biological tissues that do not contain haemoglobin, like fat, bones, and other internal structures 216 . The method was so far used for high-resolution whole-body visualization of several optically diffusive model organisms whose sizes may vary from sub-millimeter up to a centimeter range, e.g. insects, worms, fishes, and small mammals 25 216 217 . However, since optoacoustics was already successfully applied to brain imaging in primates 218 and whole breast imaging in humans 219 , selected clinical implementations are also foreseen. Advantageously, spatial resolution in optoacoustics can be kept relatively high (between 20-200 μm) for the entire penetration range of several millimeters to centimeters of tissue.

In addition to offering rich intrinsic tissue contrast, optoacoustic imaging can also be used to visualize exogenous molecular and functional markers. Naturally, almost all materials in nature absorb light therefore can become potential candidates for providing contrast in optoacoustic imaging. For high contrast imaging, of special interest are compounds having high molar extinction (absorption). Several dedicated agents were so far exploited for enhancing contrast in optoacoustics. Gold NPs of various shapes (nanorods, nanocages, nanoshells) 220 , were shown to increase optoacoustic signals in vivo. Single-walled carbon nanotubes (SWNT) provide an excellent contrast for optoacoustics and, when conjugated with peptides or other specific targeting compounds, can be used as molecular contrast agent 221 .

Clearly, many other dedicated contrast agents could potentially be developed for optoacoustic imaging applications. However, additional studies are required to address a variety of efficiency, BBB penetration capabilities, dosing, safety and toxicity concerns associated with those new contrast agents. Instead, many widely adopted optical contrast agents, such as fluorochromes, can be readily used by applying multispectral optoacoustic tomography (MSOT) 25 . It uses pulsed illumination at multiple wavelengths in order to spectrally identify reporter molecules with distinct spectral signatures, such as common fluorochromes or other chromophores within the background tissue absorption. In this way, various additional molecularly-relevant information contained in the optical spectrum can potentially be resolved such as fluorogenic or chromogenic bio-markers associated with gene expression, morphogenesis or decease progression. The method is capable of high resolution 3D visualization of molecular probes, such as common optical molecular probes and fluorescent proteins, located deep in scattering living tissues 25 222 . It can therefore simultaneously deliver anatomical, functional and molecular information with both high resolution and penetration capabilities.

In conclusion, even though optoacoustic imaging methods like MSOT are in their infancy from both technical and application standpoints, it is a rapidly emerging field in the imaging sciences that can overcome major limitations of optical imaging while retaining its contrast and sensitivity advantages 223 . It is therefore expected to drastically expand the capabilities of photonic imaging in the field of in vivo imaging of drug delivery markers.

Naturally, every imaging modality comes with its own pros and cons and no method can fulfill the complete range of requirements for every application. Table 3 summarizes the main performance characteristics of different imaging modalities, related to their potential use in real-time tracking of nanocarriers in the brain.