Research Models of the Nanoparticle-Mediated Drug Delivery across the Blood–Brain Barrier

Abstract

Brain diseases and damages come in many forms such as neurodegenerative diseases, tumors, and stroke. Millions of people currently suffer from neurological diseases worldwide. While Challenges of current diagnosis and treatment for neurological diseases are the drug delivery to the central nervous system. The Blood–Brain Barrier (BBB) limits the drug from reaching the targeted site thus showing poor effects. Nanoparticles that have advantage of the assembly at the nanoscale of available biomaterials can provide a delivery platform with potential to raising brain levels of either imaging therapeutic drugs or imaging. Therefore, successful modeling of the BBB is another crucial factor for the development of nanodrugs. In this review, we analyze the in vitro and in vivo findings achieved in various models, and outlook future development of nanodrugs for the successful treatment of brain diseases and damages.

Introduction

Hundreds of millions of people worldwide suffer from neurological diseases and damages in central nervous system (CNS) including Alzheimer’s, Parkinson’s, Huntington’s, brain tumor, stroke, etc. However, treatment and diagnosis for CNS diseases remain limited due to the inability of most therapeutic and diagnostic drugs to adequately cross the blood–brain barrier (BBB).

The BBB is highly selective semipermeable membrane barrier in the interface between the blood of the brain microvasculature and the brain. It is made up of capillary endothelial cells and basement membrane, pericytes, and astroglial podocytes. It is crucial for achieving a normal function and ensure a homeostatic environment within the CNS. The BBB permits water and small lipophilic gases to diffuse freely by passive diffusion, and also allows the selective passage of electrolytes and major nutrients through transporter proteins or receptor-mediated endocytosis. Moreover, it prevents the entrance of lipophilic neurotoxins and drugs by means of the P-glycoprotein-mediated transport mechanism [1]. Thus, there is a need to develop an unique approach to enhance CNS drugs delivery to the disease site.

Nanoparticles (NPs) have emerged as an effective drug delivery system for the treatment and diagnosis of neurological diseases and damages [2, 3]. When NPs penetrate BBB, therapeutic merit of NPs that deliver target drugs is explicit. Along with this, NPs localized to the target site can be a probe for diagnosis. With the help of unique inherent properties of NPs, the diagnosis of various diseases becomes easier. The early stage diagnosis of disease is very important which significantly increases the possibility of curing diseases. Recently, some NP-based diagnostic systems, such as upconversion NPs [4], magnetic NPs [5, 6], Au NPs [7], quantum dots (QDs) [8], liposomes [9] have been developed for the brain diseases such as Alzheimer disease (AD) to selectively bind β-amyloid (Aβ) and to detect the status or even to remove the plaque. The multifunctionality of NPs offers additional advantages for specific and more accurate diagnosis of diseases. NP surface is quite often designed with different molecules such as antibodies, proteins, peptides to recognize specific receptors/transporters overexpressed on BBB which can help penetration of the barrier and reach the diseased or damaged sites. Sometimes NPs are also conjugated with dual targeting molecules; one is to help crossing BBB while the other to target disease sites at CNS [10, 11]. By using the optical or magnetic properties of the functional NPs, specific CNS diseases can be diagnosed properly and cured by delivering drugs [12–15].

In this review, we cover the technical details of current in vitro BBB models including static BBB models and dynamic BBB models, and then we analyze the in vivo findings achieved in various models. Finally, we outlook future development of NPs for the successful treatment of brain diseases and damages.

Studies with the in vitro BBB models

The efficacy of developed NPs for crossing BBB has been tested in vitro which provides a validation of the NPs prior to applications in vivo. BBB is composed of tightly attached brain microvascular endothelial cells (BMECs) that separates CNS from blood, basement membrane, and astrocytic foot. Therefore, most in vitro models use the EC layer or its co-cultures with other neural cells that even if not completely replicating the complexity of in vivo system, informs the interaction behaviors of NPs with barrier.

Currently, two types of models are available, namely static and dynamic model [16–18]. In static model, the shear stress on BBB due to a blood flow is absent which is thus simpler than dynamic model. Static models are also divided into monoculture, contact co-culture, non-contact co-culture, and triple co-culture model. Representative in vitro BBB models are summarised in Table 1. Figure 1 depicts the in vitro BBB models developed thus far.

Table 1.

In vitro models of NPs transport across the BBB

In vitro BBB model	EC type	Co-cultured cells	Tpye of NPs	Drug	Evaluation of in vitro performance	Ref.
Static models
Monoculture	rat BMECs	N.G.	CDX-RBCNP	Dox	Increased NPs permeability	[21]
Monoculture	hCMEC/D3	N.G.	PECA; HAS	AG	PECA disrupted BBB integrity; HAS improved AG permeability	[22]
Contact co-culture	rat BMECs	Rat pericytes	Silica	N.G.	The 30 nm silica possessed higher permeability than the larger NPs	[23]
Non-contact co-culture	bEnd.3 cells	N2a cells	PEG-b-PLA	N.G.	Physicochemical parameters of NPs less influenced transcytosis	[24]
Triple co-culture	bEnd.3 cells	Rat astrocytes & neurons	CPP-Tf-lip	pGFP	Iincreased NPs permeability and gene transfection efficiency	[25]
Dynamic models
DIV-BBB model	hBMECs	Human astrocytes	N.G.	Sucrose; phenytoin, diazepam	Increased drugs permeability in physiological conditions	[26]
BBB-on-chip model	hBMECs	hBVP; human astrocytes	HNP	Apolipoprotein A1	Increased NPs permeability in both physiological and pathological conditions	[29]
BBB-on-chip model	Human iPS-derived BMECs	Human brain pericytes & astrocytes	N.G.	Cetuximab	High expressions of TEER and TJ proteins; selective transcytosis of drugs	[30]
BBB-on-chip model	hBMECs	Human brain pericytes & astrocytes	N.G.	Anti-transferrin receptor antibody	High expression of TJ proteins; high transcytosis of drugs	[31]

AG andrographolide; bEnd.3 mouse brain vascular endothelial cells; BMECs brain microvascular endothelial cells; BVP brain vascular pericytes; CDX (FKESWREARGTRIERG) peptide; CPP cell-penetrating peptide; Dox doxorubicin; hCMEC/D3 human cerebral microvascular endothelial cell line; HNP HDL-mimetic nanoparticle; HSA human serum albumin; lip liposome; N2a mouse neuronal cell line; N.G. not given; NPs nanoparticles; PECA poly ethylcyanoacrylate; pGFP plasmid encoding green fluorescent protein; PLGA poly(lactic-co-glycolic acid); RBC red blood cell membranes; Tf transferrin

Fig. 1.

In vitro models of blood brain barrier (BBB); A static and B dynamic co-culture

The BBB in vitro models characterize the morphology, transendothelial electric resistance (TEER) and permeability of BBB [19]. Morphology is evaluated under phase contrast microscopy. The tight junction of BBB is characterized via immunostaining of tight junction proteins such as ZO-1. TEER is the quantitative measurement to determine the permeability of BBB. A tight endothelial barrier accompanies high TEER value and low permeability whereas low TEER value means irregular cellular barrier. The endothelial barrier integrity is measured by epithelial voltammeter to determine the TEER value. Generally, BMECs are co-cultured with astrocytes and pericytes to enhance the TEER value and to lower the permeability which gets close to the value in vivo (1000 Ω.cm²). Therefore, this model is a more reliable in vitro BBB model [20].

Static BBB models

In the monoculture model, BMECs are grown in a monolayer on a transwell membrane made of 0.4 µm pores that can exchange biomolecules, drugs and NPs but block cell migration. While this monoculture model can show the relevant expression of enzymes, receptors and transporters [16], it lacks the cell–cell communication and thus exhibits inadequate BBB functions compared to in vivo system. Using the monolayer culture of rat brain capillary ECs, Chai et al. synthesized red blood cell (RBC)-derived cell membrane coated CDX (a peptide derived from candoxin) functionalized PLGA NP (CDX-RBCNP) to study brain targeted drug delivery [21]. Cells were seeded onto a transwell chamber in a 24-well plate pre-coated with rat tail collagen. After 7 days the TEER value was recorded 300 Ω.cm². The cell monolayers were incubated with 50 µM DiD-loaded RBCNPs at 37 °C and then the solution from the lower compartment was taken out at different time points and the fluorescence intensity was measured to determine the permeability of the NPs through the endothelial barrier. The CDX-RBCNPs were shown to have higher transcytosis efficiency compared with RBCNPs. The BBB monolayers pre-treated with 500 µM CDX reduced the transcytosis of CDX-RBCNPs, confirming that CDX peptide enhanced the penetration of the NPs through BBB. Andrographolide (AG) loaded human serum albumin (HSA) NPs and poly ethyl cyanoacrylate (PECA) NPs were also tested in a monolayer culture model using hCMEC/D3 cell line [22]. AG has the potential for treatment of different neurodegenerative diseases, but low bioavailability of AG limits its efficiency. Therefore, NP-based drug delivery system was used for the AG delivery. After 7 days the TEER value was determined, and the apical to basolateral coefficient (P_app) for both AG-loaded NPs and free AG was determined with respect to negative control compound sodium fluorescein (Na-F). Both NPs showed higher P_app value compared with free AG whereas PECA NPs disrupted the integrity of BBB.

In contact co-culture model, BMECs are generally cultured together with other cells such as astrocytes and pericytes which contribute to barrier properties [20, 23]. In this model, Hanada et al. used a commercial in vitro BBB model to evaluate the permeability of silica NPs (30, 100, and 400 nm). BMECs and pericytes were grown on the inner surface and the outer surface of insert membrane, respectively. The results showed that the P_app of the 30 nm NPs was higher than the other NPs and the model could be a useful tool for testing the permeability of NPs. In non-contact co-culture model, Rabanel et al. [24] investigated the permeability of PEG-b-PLA diblocks with different physicochemical properties across the BBB for improving the therapeutic efficacy of neurodegenerative diseases. bEnd.3 cells were seeded in the apical compartment of transwell and N2a cells were plated at the bottom of the basal compartment. The results revealed that size and surface chemistry of the NPs impacted the endocytosis, however they played little influence for the translocation from endocytosis in bEnd.3 cells to transcytosis and uptake by N2a cells. Although the TEER value attained by this co-culture model was increased, it is still lower than the in vivo value. In triple co-culture model, Dos Santos Rodrigues et al. conducted the study of transport processes of CPP-Tf-liposomes NPs across the BBB for the development of gene therapy of neurological diseases. bEnd.3 cells were seeded on upper surface of insert membrane, while astrocytes were cultured on the lower surface of insert membrane, and neurons were seeded at the bottom of the well. Meanwhile, to enhance the efficiency of gene delivery of the NPs by effectively targeting brain and increasing the BBB permeability, they synthesized CPP-Tf-liposome NPs, which were modified on the surface with transferrin (Tf) and cell-penetrating peptides (CPP). The results showed that the NPs efficiently transposed the BBB model followed by effectively transfecting primary neurons [25]. Therefore, this triple co-culture model is considered as the most reliable model that is used in many experiments to study the permeability of NPs through BBB among different static models.

Dynamic BBB models

The static in vitro BBB models lack shear stress which but always exists in BBB and plays key roles in many BBB functions. Therefore, dynamic BBB models have been developed that increases TEER value and expression of tight junction (TJ) proteins. In a typical cone plate model, a rotating cone generates shear stress which transmits through the medium and ECs receive it [18]. But this model is very simple and produces uneven stress on BBB. Another model is the so-called ‘dynamic in vitro BBB model’ (DIV-BBB). Cucullo et al. developed the humanized DIV-BBB model with human BMECs and astrocytes from normal and epileptic brain tissue. BMECs and astrocytes were seeded inner side and outer side of porous hollow fibres, respectively. Then, a medium was pumped into the system with a controlled flow. This flow produces shear stress on the BMECs comparable to the in vivo conditions. A gas permeable tubing system was also maintained for O₂ and CO₂ flowing. This model produced low permeability and high TEER value, and mimicked the physiologic permeability properties of the BBB in vivo environment and a drug-resistant BBB phenotype [26]. Therefore, the DIV-BBB model represents the most realistic in vitro system to study the permeability of nanodrugs across the BBB.

Recently, microfluidic devices have also endeavoured to design BBB models on a chip. The BBB-on-chip is considered to mimic more closely the in vivo BBB physiological microenvironment with the realistic dimensions, geometries and fluid flow. The typical device consists of polydimethylsiloxane (PDMS) substrates, glass slides, and a porous polycarbonate (PC) membrane. The PDMS parts house two channels that are divided by the PC membrane into the upper channel that acts as the brain side of the barrier and the lower channel that acts as the vascular side. For example, the channels with 2 mm (luminal) or 5 mm (abluminal) width and 200 μm depth at the cell culture interface ensure a laminar flow. The PDMS parts are sandwiched between two glass slides with AgCl thin-film TEER electrodes that can sense the electrical signals (e.g., recorded ~ 180–280 Ω·cm²) [27, 28]. Ahn et al. fabricated a BBB chip using organ-on-a-chip technology. In their BBB chip, BMECs were plated in the vascular channel with a porous membrane under a physiological level of shear stress while BVPs were seeded on the other side of the porous membrane. Human astrocytes were cultured in a hydrogel of the perivascular channel. The model can quantitatively analyze the transport and distribution of NPs in the BBB at cellular and tissue levels. Therefore, the BBB chip will become a reliable tool for better tracking of the permeability of nanodrugs across the BBB in both physiological and pathological conditions [29]. Park et al. made a 2-channel BBB model with microfluidic organ-on-a-chip culture technology composed of induced pluripotent stem cell-derived human BMECs of the basal vascular channel and human brain astrocytes and pericytes of the apical parenchymal channel. The endothelium expressed high values of TEER and TJ proteins, and displayed selective transcytosis of therapeutic antibody cetuximab. It may be a useful in vitro BBB model for effective drug screening in the treatment of CNS diseases [30]. Wevers et al. created also a BBB chip with three-channel platform. The platform was patterned by extracellular matrix gel using surface tension technology. The BMECs was seeded against the extracellular matrix gel. Astrocytes and pericytes were cultured on the other side of the gel. The results showed that the endothelium displayed high expression of TJ proteins and anti-transferrin receptor antibody. The model could support the transcytosis of therapeutic antibody for the treatment of CNS diseases [31]. Often, the devices that use only PDMS and glass slides without PC membrane were also designed [32, 33]. For example, Deosarkar et al. made a BBB-on-chip device comprised of two independent vascular channels in the periphery of the device (for ECs culture) and a tissue compartment at the center of the device (for astrocytes culture) that are separated also with microchannels in both static and dynamic flow conditions [33].

The in vitro models can not only avoid research ethical issues, but also mimic the in vivo BBB with co-cultures of different cell types and thus to interpret the penetration and cellular uptake behaviors by measuring key parameters such as TEER and TJ proteins. However, they cannot replicate the in vivo conditions, limiting the efficacy of administered molecules and NPs in specific neurological diseases and damages. The in vivo models can overcome the disadvantages of the in vitro models and are best-established method for studing drug delivery across the BBB, although they need to follow animal ethical consideration and approximately 50% of data from animal experiments are not suitable for humans [34]. The following section discusses the animal models for different brain diseases and damages and the outcomes of NPs developed for such applications.

Application of NPs in the brain animal models

The NPs developed to cross BBB and to treat brain diseases and damages were investigated in many in vivo models, such as Alzheimer’s, Parkinson’s, Huntington’s, brain tumors and stroke. In vivo BBB models for the NPs-mediated drug delivery across the BBB are summarised in Table 2. Figure 2 describes the delivery of NPs-mediated drug into the brain of animal models of neurological diseases across the BBB. Below are detailed the in vivo findings of the NPs used for different brain diseases and damages.

Table 2.

In vivo models of NPs transport across the BBB

Neurological disease	Model method	Tpye of NPs	Drug	Route of delivery	Evaluation of in vivo performance	Ref.
Alzheimer disease	Aβ1-42-induced model with mice	TQ-PEG-PLA	N.G.	i.v	NPs permeability ↑; binding of amyloid plaques binding ↑	[38]
	Aβ1-42 + IBO- induced model with mice	GM1-rHDL	NAP	i.n	Amyloid plaques deoposition ↓	[40]
	APP/PS1 transgenic mice	USPIO-mannitol	N.G.	i.v	NPs distribution of amyloid plaque ↑	[43]
	APPswe/PS1dE9 transgenic mice	W20/XD4-SPIO	N.G.	i.v	NPs distribution of Aβ ↑;inhibiting Aβ aggregation	[44]
Parkinson disease	6-OHDA-induced model with rats	SLN	Bromocriptine	i.p	Extending drug half-life; attenuated akinesia	[51]
	6-OHDA-induced model with rats	PLGA	DA	i.v	Extending DA release; attenuated akinesia	[52]
	6-OHDA-induced model with rats	Lf-PEG-PAMAM	hGDNF	i.v	Gene delivery efficiency ↑; attenuated akinesia	[53]
	6-OHDA induced model with rats	Lf-PEG-PLGA	UCN	i.v	NPs permeability ↑; attenuated akinesia	[54]
Huntington disease	3-NP-induced model with rats	SLN	Thymoquinone	i.p	Attenuated akinesia	[61]
	R6/2 transgenic mice	G7-PLGA	Chol	i.p	Improved cognitive dysfunction; attenuated akinesia	[62]
	R6/2 transgenic mice	Poly(trehalose)	N.G.	i.p	Suppressing huntingtin protein aggregates	[63]
Brain tumor	Bearing C6 cells tumor xenograft model with nude mice	AsTGN-PEG-PCL	Docetaxel	i.v	NPs’ accumulation of tumor ↑ anti-tumor effect ↑	[10]
	Bearing C6 cells tumor xenograft model with Spraque–Dawley rats	Tf-lip-MAN	DNR	i.v	NPs’ accumulation of tumor ↑ anti-tumor effect ↑	[11]
	Bearing 9L cells tumor xenograft model with nude mice	NP-PEG-CTX	N.G.	r.o	NPs’ accumulation of glioma ↑	[68]
	Bearing C6 cells tumor xenograft model with nude mice	As-PEG-PLGA	PTX	i.v	Inhibiting tumor growth		[69]
	Bearing 9L and F98 cells tumor xenograft model with rat	PAA-PEG	CDDP	i.v	Reducing tumor growth		[70]
	Bearing GL261 cells tumor xenograft model with BalbC mice	QD-PLGA-TAT	N.G.	i.v	NPs’ accumulation of tumor ↑		[71]
	Bearing 73c cells tumor xenograft model with nude mice and BalbC mice	Glutathione- coated AuNPs	N.G.	i.v	NPs permeability and retention ↑		[72]
	Bearing U87 MG cells tumor xenograft model with BalbC mice	ANG-PEG-PCL	PTX	i.v	NPs’ accumulation of glioma ↑		[73]
Stroke	Rat MCAO model	ANG-PEG-CeO2	Edaravone	i.v	NPs permeability ↑; efficiently scavenging ROS		[77]
	Rat MCAO model	Phospholipid-PEG- CeO2	N.G.	i.v	NPs’ accumulation of ischemic hemisphere ↑; infarct volume ↓		[78]
	Mouse MCAO model	Pt NPs	N.G.	i.v	Infarct volume ↓; motor function ↑		[79]
	Rat MCAO model	citrate-capped Au NPs	N.G.	i.p	Infarct volume ↓(20 nm); infarct volume ↑(5 nm)		[80]
	Rat MCAO model	Fe3O4-RGD	N.G.	i.v	NPs’ accumulation of ischemic hemisphere ↑		[81]

3-NP 3-nitropropionic acid; 6-OHDA 6-hydroxydopamine; ANG angiopep-2; As AS1411 aptamer; CDDP cis-Diamminedichloroplatinum; Chol cholesterol; CTX cholorotoxin; DA dopamine; DNR daunorubicin; g7 glycopeptides; GM1-rHDL monosialotetrahexosylganglioside-modified reconstituted high density lipoprotein; hGDNF human glial cell line-derived neurotrophic factor gene; IBO ibotenic acid; i.n. intranasal; i.p. intraperitoneal; i.v. intravenous; Lf lactoferrin; MAN p-aminophenyl-α-D-manno-pyranoside; lip liposome; MCAO middle cerebral artery occlusion; NAP (NAPVSIPQ) peptide; N.G. not given; PAA polyaspartic acid; PAMAM polyamidoamine; PEG-PLA Poly(ethylene glycol)-Poly (lactic acid); PLGA poly(lactic-coglycolic acid); Pt Platinum; PTX paclitaxel; QDs quantum dots; RGD Arg-Gly-Asp; r.o. retro-orbital; ROS reactive oxygen species; SLN solid lipid nanoparticles; TAT (YGRKKRRQRRR) peptide; Tf transferrin; TQ TGN (TGNYKALHPHNGC) peptide and QSH (QSHYRHISPAQVC) peptide; UCN Urocortin; USPIO ultrasmall superparamagnetic iron oxide; W20 Aβ oligomer-specific scFv antibody; XD4 class A scavenger receptor activator

Fig. 2.

The delivery of NPs-mediated drug into the brain of animal models of neurological diseases by crossing the BBB

Findings in AD models

AD model was developed by inducing scopolamine through subcutaneous injection in rat or mice [35, 36]. To test the amnesia, the animals were submitted to passive avoidance reflex test [37], AD model was also created by intracerebroventricular injection of Aβ_1-42 solution [38–40]. Accumulation of aggregated Aβ in AD brain destroys the neurovascular units, resulting in a decreased endothelial transport, pericyte degeneration, and astrocyte depolarization that may contribute to the BBB dysfunction [41]. First, Aβ_1-42 (1 mg/ml) was dissolved in saline and incubated for 7 days at 37 °C. After that the mice were anaesthetized and fixed, and 5 µl solution was injected bilaterally into the hippocampus.

With the AD model, a study of using PEGylated PLA NPs was carried out. In particular, two peptides were conjugated: TGN (TGNYKALHPHNGC) that crosses BBB and QSH (QSHYRHISPAQVC) that has affinity to bind with amyloid plaques [38]. For in vivo distribution the NPs were loaded with Coumarin-6 and injected to the mice intravenously. At different time points (0.083, 0.25, 0.5, 1, 2, 4, 8, 12 and 24 h), the blood and brain tissues were collected, and the coumarin-6 concentration in the blood and brain tissues was measured using HPLC. Results showed that after TGN conjugation the brain penetration of NPs increased whereas the accumulation of TGN and QSH conjugated NPs was maximum in hippocampus region.

Administration of both Aβ_1-42 and ibotenic acid (IBO) can induce significant neuronal loss, working memory deficits which provide a useful model for studying the pathogenetic mechanisms involved in AD. Before co-injection, Aβ_1-42 and IBO were mixed to a final concentration of 1 μg/μL Aβ_1-42 and 0.5 μg/μL IBO [40]. Male ICR mice were anaesthetized with chloral hydrate and then fixed in a stereotaxic frame. The animals were bilaterally injected into the dorsal hippocampus via a microsyringe of the mixture of Aβ_1-42 and IBO to establish the AD model mice. Using this model, reconstituted high density lipoprotein modified with monosialotetrahexosylganglioside (GM1) was nasally treated to the AD mice daily at a dose of 10 µl (5 mg/kg). A model neuroprotective peptide NAP derived from activity dependent neuroprotective protein (ADNP) was loaded onto the NPs. Results showed that GM1 modified high density lipoprotein (GM1-rHDL) bound more efficiently with Aβ_1-42 compared with the group without GM1 modification. Two days after intra-hippocampus injection of Aβ_1-42 and IBO, the AD model mice were intranasally administered with NAP loaded GM1-rHDL daily for 2 weeks. After 2 weeks NAP loaded NPs could reduce Aβ_1-42 deoposition, ameliorate neurologic damages and decrease memory deficits. The difference in efficacy between drug-loaded NP and free drug was also studied for the treatment of AD. For example, rivastigmine is a drug responsible for inhibiting cholinesterase and used for the treatment for Alzheimer’s. However, free drugs are not able to cross the BBB failing to accumulate at significant proportions in brain. Therefore, the poly (n-butylcyanoacrylate) NPs coated with polysorbate 80 were used to load the drug and delivered to the brain. It was found that after polysorbate 80 coating the drug accumulation in liver and spleen was reduced whereas the brain uptake increased to 3.8-fold compared with free drug [42].

For the in vivo diagnostic purpose of AD, ultrasmall superparamagnetic iron oxide (USPIO) NPs were used. Aβ_1-42 peptide was conjugated with USPIO NPs via EDC coupling. Since USPIO NPs cannot penetrate BBB, NPs were co-injected with mannitol in PBS. The NPs/mannitol mixture was injected through right femoral vein of 14 months old APP/PS1 transgenic mouse. After the injection MRI images revealed that NPs along with 15% mannitol produced a large number of intense spots in micro-MRI corresponding to Aβ_1-42 plaques while the Aβ_1-42 conjugated NPs did not show any significant toxicity in animals [43]. SPIO NPs were functionalized with Aβ oligomer-specific scFv antibody W20 and class A scavenger receptor activator XD4 and further detected and treated amyloid deposits. The functionalized NPs were injected via tail vein of transgenic mouse. The results demonstrated that the NPs possessed diagnostic value and therapeutic benefits for AD. Thus, the functionalized NPs present as a promising agent for the diagnosis and treatment of AD [44].

Findings in Parkinson disease models

For the in vivo study of Parkinson disease (PD), animal models established use the injection of 6-hydroxydopamine (6-OHDA) [45], 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) [46] or paraquat [47]. These toxins generate hydrogen peroxide radicals that destroy neurons and give rise to generating PD. MPTP does not disrupt the BBB function in animals, and numerous proinflammatory cytokines from the neurovascular units affect TJ protein expression [48]. To check the symptoms of the disease different tests were performed which include open field test, locomotor activity in the actometer, rigidity through the registration of the alteration of gain dynamic, and tremor through the registration of intensity and duration of tremor [49]. Genetic models were also used for PD [50].

NPs have been developed for the treatment of PD. Solid lipid nanoparticle (SLN) consisting of tristearin–tricaprin (2:1 mixture) was designed to load drug bromocriptine (BK). The drug release was slower compared with free drug; the BK release from SLN was 74% after 48 h. The activity of the drug was further tested in the 6-OHDA hemilesioned rats [51]. The action of drug loaded within NP was more rapid and long-lasting than that of free drug. Encapsulated BK was effective after 30 min of administration and became maximum at 5 h whereas the action of free drug was slower and short-lasting as it disappeared after 5 h. The inherent nature of crossing BBB of this SLN could efficiently deliver the drug to brain compared with free drug. Although the mechanism of BBB crossing is not fully understood it has been suggested that this NP can open the tight junction to create a paracellular pathway for translocation of NP to brain. Other mechanisms such as passive diffusion, transport or endocytosis could also be responsible for the NP translocation to brain. Akinesia was evaluated in hemiparkinsonian rats by the bar test, which measures the ability of the rat to respond to an externally imposed static posture. This is a valuable test for the measurement of catalepsy and has been validated for assessing akinesia under parkinsonian conditions.

Dopamine replacement therapy is currently used for the treatment of PD. To model this, 6-OHDA was treated to animals which have lower levels of dopamine and its metabolite 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanilic acid (HVA) in the striatum [52]. The PLGA NPs were developed to load dopamine (DA NPs). The DA NPs infused to the rat model were shown to maintain the stability of dopamine by inhibiting auto-oxidation, releasing steadily (60% in 7 days). The PLGA NPs labelled with FITC showed accumulation in brain parenchyma, astrocytes and neurons which proved that the NPs could target those cells and release dopamine. Therefore, dopamine released in the brain via NPs was able to reverse the dopamine receptor changes caused by a lack of this neurotransmitter. Amphetamine induced rotation study used to measure the effectiveness of the NPs revealed that the number of ipsilateral rotation increased in 6-OHDA lesion in rat model whereas the infusion of DA NPs significantly decreased the rotation after 7 days, demonstrating the in vivo efficacy of NPs for delivering dopamine in the treatment of PD.

Another exemplar study reported the gene delivery for lactoferrin (Lf) targeted brain in the treatment of PD. A mouse model was prepared by injecting of 6-OHDA solution to each striatal site [53]. Following the 6-OHDA lesions, the rats were then injected with hGDNF gene loaded Lf with the conjugation of PEGylated PAMAM dendrimer. The delivery of NPs loaded with hGDNF gene was shown to improve the behavioral activity when compared with control. After the gene delivery, the rats had more symmetrical swings and lower apomorphine induced rotations than control 6-OHDA lesioned rat models. In another study, the Lf conjugated PEG-PLGA NPs loaded with Urocortin (UCN) were further injected to the PD rat model [54]. Lf is cationic iron binding protein similar to transferrin and its receptors are found in BBB. Therefore, Lf conjugated NPs can target BBB, and here the PEG-PLGA NP was conjugated with Lf to target BBB and deliver UCN. UCN is a corticotrophin releasing hormone related peptide, and has recently been proposed as a cytoprotectant for cultured hippocampal neurons, cerebellar granule cells and GABAergic neurones. Free UCN cannot pass through BBB therefore nanoformulation with PEG-PLGA was designed. Negative surface charge of UCN loaded NPs can reduce protein adsorption and thus nonspecific accumulation in other organs. From in vitro results it was clearly seen that NP uptake was energy- and clathrin-dependent endocytosis in bEnd.3 cells. Adding free Lf could decrease the uptake of NP indicating NP entry into cells through receptor mediated endocytosis. It was observed that Lf-NPs could accumulate in substantia nigra, cortex and striatum region after 1 h of in vivo administration. The accumulation of coumarin-6 used to study the bio-distribution of NPs demonstrated 2.36 times higher amount in brain with Lf-NPs than with bare NPs, and the infusion of UCN loaded NPs significantly elevated the apomorphine based rotation behavior, suggesting the efficacy of functionalized polymeric NPs in targeting PD brain cells and improving the functions.

Findings in Huntington disease models

Huntington disease (HD) is neurodegenerative disease that causes motor disturbance, cognitive loss and psychiatric manifestations. It is generated due to expanding CAG repeats coding for polyglutamine. The repeat length causing the disease is between 35 and 45 consecutive glutamines. In cellular model the expanded polyglutamine repeats lead to programmed cell death, and in human tissue and transgenic animals they undergo protein aggregation forming neuronal intranuclear inclusions (NIIs), which are considered hallmark of polyglutamine diseases [55].

Previously, HD mouse model was prepared by a direct introduction of excitatory agonists into CNS. Lucas and Newhouse observed retinal degeneration in mice after systemic administration of glutamate [56]. After that various groups have demonstrated that administration of glutamate type neurotoxins, kainic acid could produce degeneration in striatal GABAergic projection neurons. The model was further improved by introducing other excitotoxins such as quinolinic acid which is found at elevated levels in HD patients released from activated microglia that leads to neurodegeneration [57]. 3-nitropropionic acid (3-NP) is another agent that can replicate a number of pathological symptoms found in HD models. 3-NP reduces the cellular levels of ATP and causes neuronal damage by excitotoxic mechanism [58]. Genetic mouse models have also been developed that fall into three categories depending on phenotypes: (i) mice that express exon-1 or exon-1 and 2 of the human huntingtin (htt) gene containing polyglutamine mutations, (ii) mice with pathogenic CAG repeats inserted into the existing CAG expansion in murine Hdh (knock-in mice), and iii) mice that express the full-length human HD gene (plus murine Hdh). The R6/2 was the first HD transgenic mouse model generated by overexpressing N-terminal fragment of htt (exon 1) with approximately 144–150 CAG repeats at exon 1. According to neurological phenotype this model is very close to human HD and therefore appropriate for testing therapeutic potential. N171-82Q is another HD mouse model similar to R6/2. These mice models have N terminal fragment of htt incorporating both exon 1 and exon 2 with 82 polyglutamines. The animal models were tested with different phenotypes such as weight loss, diabetes, clasping, tremor and convulsions [59]. Huntingtin protein aggregates are present in the neurovascular unit of R6/2 mice and accompanied by a reduced expression of TJ proteins. The BBB integrity is compromised [60]. Thymoquinone (2-isopropyl-5-methyl-1,2-benzoquinone, TQ) is known to have many beneficial effects including neuroprotection and suppression of oxidative stress induced neuropathy. However, poor solubility and high hydrophobicity hamper pharmaceutical applications. Ramachandran and Thangarajan used solid lipid nanoparticles (SLN) to deliver TQ by encapsulation (TQ-SLN) as the SLN has been reported to penetrate BBB through endocytosis and used also for HD. The HD mouse model was developed by injecting 3-NP (10 mg/kg) intraperitoneally which resulted in significant body weight loss. Systemic 3-NP induction significantly affected the muscle grip performance, as assessed by the rotarod test at days 7 and 14. With the treatment with TQ-SLN (10 and 20 mg/kg) the mouse significantly regained the muscle strength contrasting with 3-NP administered group [61]. A low cholesterol level in HD mouse brains might be detrimental for neuronal function. However, cholesterol cannot cross the BBB. Valenza et al. loaded the cholesterol into glycopeptides (g7)-modified polymeric NPs that easily penetrate the BBB (g7-NPs-Chol). And then the g7‐NPs‐Chol was injected intraperitoneally in R6/2 transgenic mice. Within a few hours, the g7‐NPs‐Chol reached neural cells of different brain regions and ameliorated cognitive defects by positively influencing synaptic protein network after systemic injections in R6/2 mice. The nanoparticle-mediated delivery approach provides new directions for cholesterol delivery in HD mouse models [62]. The accumulation of huntingtin protein cause toxicity of neuronal cells in HD mouse brain. poly(trehalose) NPs can prevent the mutant protein aggregation. Debnath et al. synthesized poly(trehalose) NPs composed of a 6 nm iron oxide core and a zwitterionic polymer shell with ~ 5–12 wt % covalently linked trehalose. 100 μL of zwitterionic Poly(trehalose) NPs (0,4 mg/mL) were intraperitoneally injected in R6/2 transgenic mice. The results showed that the poly(trehalose) NPs with zwitterionic surface charge and a trehalose multivalency of ~ 80–200 are crucial for entering neural cells and inhibition of the huntingtin protein aggregation. Their work also brings a new approach for the treatment strategies of other neurodegenerative diseases [63].

Findings in brain tumor models

The BBB in brain tumors is described as blood-tumor barrier (BTB) and less intact compared with the normal brain one due to its disruption during the tumor development process. However, despite leakage of the BTB to some drugs the core of the tumors has the intact barrier, drug delivery is still a challenge in the treatment of brain tumors [64, 65]. Glioma is the most commonly found brain tumor and remains a significant health problem worldwide. Glioma model was developed via injection of N-ethyl-nitrosourea (ENU) intravenously in pregnant mice [66]. DNA repair mechanism in neural cells is less active than in other cells. This reagent causes DNA mutations in neural cells which pass through their progeny. Another way to develop glioma model is through genetic mutations that are responsible for gliomagenesis. Xenografting glioma cells in immunocompromised animal can also be used to create glioma [67]. Xenograft of intracerebral glioma model is prepared in normal mice by growing subcutaneous implantation of glioma cells. To generate the model, a small hole (1 mm diameter) was drilled through the skull, and 10 µl of tumor cell suspension was injected through the hole which was then filled with bone wax. The BBB in brain tumors is described as blood-tumor barrier (BTB) and less intact compared with the normal brain one.

For effective treatment of glioma, dual targeting ligands were conjugated with the NPs. For an example, PEG-PCL NPs were conjugated with phage displayed TGN peptide and AS1411 aptamer for specific targeting of BBB and brain cancer cells, respectively. C6 glioma cells were injected to the right brain of mouse, and DiR loaded NPs were injected intravenously for in vivo imaging [10]. The fluorescence intensity measured at 24 h showed that unmodified NPs could not penetrate the BBB whereas TGN-modified NPs significantly accumulated with a maximum reached at 12 h. While the conjugation of TGN only could not target glioma showing NPs being distributed throughout the whole brain, the conjugation of both TGN and AS1411 aptamer selectively accumulated the NPs to brain glioma. Because of the specific targeting the dual conjugated NPs showed longer survival rate than single ligand modified NPs. As another exemplar study for dual targeting, daunorubicin liposomes were developed by conjugating transferrin (TF) and p-aminophenyl-α-D-mannopyranoside (MAN) [11]. Transferrin targets the transferrin receptors and helps crossing the BBB whereas MAN enables targeting of brain glioma. The targeting efficiency was investigated in in vitro as well as in c6 glioma bearing rats through systemic administration. At day 8, the daunorubicin liposomes were administered via tail vain at a dose of 5 mg/kg, and the administration was made every 2 days with total 3 doses per rat. Based on the study, the dual targeting liposomes exerted the most significant tumor volume inhibition compared with other liposomes (only TF and MAN conjugated liposomes), demonstrating the efficacy of dual targeting approach in the treatment of brain tumors.

To treat the glioma in the model, amine- and PEG-functionalized iron oxide NPs were designed. In particular, Cholorotoxin (CTX) peptide was conjugated to target glioma [68]. The accumulation of NPs within glioma was then detected via MRI. The CTX conjugation was shown to enhance the uptake of NPs in 9L cells compared with non-conjugated NPs. The PEG grafting on NP surface could also reduce nonspecific adsorption of proteins and increase blood circulation. Tumor cells incubated with CTX-conjugated NPs showed significantly enhanced T2 contrast and thus higher R2 relaxivity compared with non-targeted NPs, demonstrating the CTX-conjugated iron oxide NPs were effective for targeting brain tumor with a simultaneous diagnosis capacity. AS1411 aptamer functionalized PEG-PLGA NPs loaded with paclitaxel drug (PTX) were also used for the treatment of glioma. For this, glioma xenograft mice model was developed by injecting c6 cells (3 × 10⁷ cells suspended in 150 µl medium) subcutaneously. After 10 days the weight of the tumours reached to 200 mg and then the NPs were administered intravenously (PTX dose = 3 mg/kg) [69]. The treatment was repeated every other day for seven consecutive injections. The tumour volume treated with aptamer conjugated NPs was smaller compared with non-targeted NP and free PTX. The survival time of animals also increased to 31 days when treated with aptamer conjugated NPs, whereas animals treated with saline, taxol, PTX-NP showed survival time of 18, 24 and 27 days, respectively.

cis-Diamminedichloroplatinum (CDDP) is a chemotherapeutic drug used to treat several types of cancers. Timbie et al. developed biodegradable polyaspartic acid polymers to deliver CDDP into brain glioma for enhanced therapeutic efficacy. They used MRI guided focused ultrasound (MRgFUS) for reversible BBB disruption to penetrate BBB. Cell suspension (9L and F98) was injected into the brain to create brain tumor. Albumin shelled microbubbles, samples and MRI contrast agent were then administered through a tail vein catheter, and sonication was performed using 1.14 MHz single element focused transducer. Based on T1 weighted MR images the tumor volume increased rapidly between days 21 and 28 for the control and samples without FUS whereas for the samples treated with 0.8 MPa FUS the tumor growth was significantly reduced [70]. For the theranostics of brain tumor, Medina et al. developed QD based ‘barcoding’ NP composed of PLGA and QD. Cell penetrating peptide TAT was conjugated to QD-PLGA NP via avidin–biotin interaction to enhance BBB penetration and accumulation in CNS. GL261 cell suspension was injected to albino C57BL/6 mice. For multispectral analysis both QD₅₈₅ and QD₆₅₅ were administered simultaneously through lateral tail vein. After confirming that no difference in accumulation of QD₅₈₅ and QD₆₅₅ in different organs the QD₅₈₅ functionalized with TAT peptide (TAT-NP) and QD₆₅₅ without TAT functionalization (CTL-NP₆₅₅) were treated. Results showed that the concentration of TAT-NP₅₈₅ in liver (1.18 ± 0.18 mg/g) was higher than that of CTL-NP₆₅₅ (0.73 ± 0.32 mg/), also the accumulation of TAT-NP₅₈₅ (0.0044 ± 0.001 mg/g) in brain was enhanced when compared with that of CTL-NP₆₅₅ (0.0029 ± 0.001 mg/g) [71]. Peng et al. reported that renal clearable glutathione coated zwitterionic Au NPs around 3 nm in size could passively penetrate BBB and target brain glioma. They synthesized two sizes of Au NPs (renal clearable 3 nm and non-renal clearable 18 nm). In vivo glioma model was prepared by injecting 73c glioma cells to the left hemisphere of the mice, and after two weeks the Au NPs were administered via tail vein. The uptake of Au NPs in different organs, as measured by ICP-MS spectroscopy, revealed the accumulation of 18 nm Au NPs in major organs such as liver, spleen, heart and lungs was much higher than that of 3 nm Au NPs, which was ascribed to a lower hydrodynamic diameter of 3 nm particles compared with kidney filtration threshold (KFT). The unique luminescence property of 3 nm Au NPs was also effective for in vivo and ex vivo imaging. After removal of skin, the glioma bearing brain showed intense signals which contrasted with the control glioma without the NP administration [72]. Angiopep-2 (ANG) is a ligand of low-density lipoprotein receptor-related protein (LRP) which is highly expressed on the BBB and glioma cells. For the therapeutics of glioblastoma, Xin et al. first developed dual-targeting NP-based drug delivery system by loading PTX into ANG-conjugated PEG-PCL NPs (ANG-NP-PTX). They then established a brain tumor model by implanting U87 MG cells (5.0 × 10⁵ cells suspended in 5 μl PBS) into the striatum of Balb/c nude mice. The permeability of ANG-NP-PTX was obviously increased and the viability of U87 MG cells was significantly decreased in the in vitro BBB model. After 18 days of the glioma cells implantation, ANG-NP was injected into mice via tail vein. In vivo study showed an obvious accumulation of the NPs in the glioma. The study suggests that ANG-NP-PTX has prospects for the treatment of brain glioma [73].

Findings in stroke models

Ischemic stroke is the major acute disability of brain all over the world. It is divided into hyperacute (< 6 h), acute (6–72 h), subacute (> 72 h), and chronic phases (> 6 weeks). The BBB dysfunction is characterized by its disruption and increase in its permeability. The BBB disruption is attributed to dysfunctional TJ proteins, endothelial cells and astrocyte, resulting in increased permeability of the barrier. The BBB permeability is caused by sudden hypoxia, neuroinflammation, and neoangiogenesis in hyperacute, acute, and subacute stages, respectively. In the chronic stage, however, its permeability starts decreasing slowly [74]. Oxygen free radicals induce cellular oxidative stress and leads to cell death that can damage neuronal networks and neurovascular units and stop brain functions [75]. Therefore, detoxification of reactive oxygen species (ROS) can reduce oxidative stress and maintain cellular activity. Different antioxidants including vitamin C, vitamin E, glutathione get involved in scavenging ROS and reducing the oxidative stress [76].

As an effective ROS scavenging material, ceria NP has recently gained much attention. Recyclable ROS scavenging activity of ceria NP made them an excellent candidate for reducing oxidative stress. Bao et al. prepared core/shell structured ceria NPs with ceria as the core and organic ANG and hydrophilic PEG as the shell [77]. Edaravone was also loaded in the organic part to additionally scavenges ROS. In particular, ANG peptide helps the NP penetrate BBB via receptor mediated transcytosis. After intravenous injection the edaravone-loaded ANG-PEG CeO₂ (E-A/P-CeO₂) showed higher accumulation in the brain region after 24 h when compared to the PEG-modified CeO₂, indicating the role of ANG in receptor mediated transcytosis. As a result, the infarct volume was significantly reduced, highlighting the functional NPs are neuroprotective by enhancing brain accumulation and ROS scavenging activity. Kim et al. also prepared phospholipid-PEG encapsulated ceria NP [78]. They showed the prepared ceria NPs could reduce the infarct volume up to 50% at a dose of 0.5–0.7 mg/kg. While the accumulation of ceria NP was very low in non-ischemic brain the accumulation was significantly enhanced at 24 h after ischemia.

Other NPs have also shown to have antioxidant effects. Platinum (Pt) NP is known to convert H₂O₂ to H₂O and O_2. Takamiya et al. proved the antioxidant effect of Pt NPs for use as a neuroprotective agent against ischemic stroke [79]. At 48 h after intravenous treatment of Pt NPs, the motor function was significantly improved and infarction volume was reduced. Au NPs was also found to reduce oxidative stress. Citrate capped Au NPs of 20 nm in size significantly reduced the neurologic deficits and infarction volumes [80]. Interestingly however, the same Au NPs with 5 nm in size increased the infarction volume. The 20 nm Au NPs reduced the expression of apoptosis-inducing factor, cytochrome c and caspase 3, and at the same time increased the expression of antiapoptotic molecules, reasoning how the NPs could reduce the apoptosis in ipsilateral-ischemic hemisphere after ischemia–reperfusion. Recently, Wang et al. prepared integrin specific iron oxide NPs for visualizing collaterals via MRI [81]. Integrin α_vβ₃ is found to be overexpressed in brain microvascular endothelial cells and plays a critical role in promoting postischemic angiogenesis. The iron oxide NPs were functionalized with Arg-Gly-Asp (RGD) peptide sequence to recognize integrin receptors for imaging cerebral collaterals. The results revealed that α_vβ₃ is an appropriate target of the collaterals in acute ischemic stroke. The Fe₃O₄-RGD NPs will provide a valuable chance for reducing the early complications of postischemic stroke.

Concluding remarks

Successful pharmacotherapy of brain disease specific drug was limited by the BBB. However, NP modifications have shown promising findings from preclinical animal studies. NP encapsulation of drugs have been utilized for improved neurological and functional outcome via passive diffusion for brain diseases and damages. They are usually modified with therapeutic ligands, targeting ligands and other functional excipients including PEG to optimize efficacy and minimize undesired side effects on the CNS. Although preclinical studies have demonstrated the benefits of NP modifications, the neurotoxicity and the safety issues must be considered for moving the research to clinical settings.

As the recent development of NPs for imaging has made a significant progress, the applications in brain diseases and damages are also growing rapidly. Particularly for brain tumor, imaging allows better delineation of tumors, visualization of malignant tissue during surgery, and tracking of responses to chemotherapy and radiotherapy. Several in vivo neuroimaging modalities are also available, which include fluorescence microscopy, MRI, CT imaging, and photoacoustic imaging, and some candidate NPs such as Gd-based NPs, iron oxide NPs, Au NPs, C-dots, and upconversion NPs, have begun to be tested for the brain imaging.

Here, we introduce the static BBB models (e.g., Transwell) and dynamic BBB models including the DIV BBB and the microfluidics BBB model on chip. Despite advances in the development of many in vitro BBB models in a variety of studies, none of them can fully replicate the in vivo BBB physiological microenvironment. The choice of the suitable model is critical for obtaining accurate experimental data and developing of NP-based nanomedicine of brain diseases and damages.

While the therapeutic and/or diagnostic efficacy of NPs developed for the brain diseases and damages has been proven in many in vivo animal models, including AD, PD, HD, brain tumors and stroke, the BBB is still a potential hurdle on the road to clinical applications of the developed NPs, attenuating their primary localization to brain. Given the rapid progress of NP development and the recent findings in the in vivo brain models, the NP-based approaches are envisaged to illuminate the theranostics of many devasting brain diseases and damages.

Compliance with ethical standards

Conflict of interest

The author declares no conflict of interest.

Ethical statement

There are no animal experiments carried out for this article.