Engineered nanomaterials that exploit blood-brain barrier dysfunction for delivery to the brain

Abstract

The blood-brain barrier (BBB) is a highly regulated physical and functional boundary that tightly controls the transport of materials between the blood and the brain. There is an increasing recognition that the BBB is dysfunctional in a wide range of neurological disorders; this dysfunction can be symptomatic of the disease but can also play a role in disease etiology. BBB dysfunction can be exploited for the delivery of therapeutic nanomaterials. For example, there can be a transient, physical disruption of the BBB in diseases such as brain injury and stroke, which allows temporary access of nanomaterials into the brain. Physical disruption of the BBB through external energy sources is now being clinically pursued to increase therapeutic delivery into the brain. In other diseases, the BBB takes on new properties that can be leveraged by delivery carriers. For instance, neuroinflammation induces the expression of receptors on the BBB that can be targeted by ligand-modified nanomaterials and the endogenous homing of immune cells into the diseased brain can be hijacked for the delivery of nanomaterials. Lastly, BBB transport pathways can be altered to increase nanomaterial transport. In this review, we will describe changes that can occur in the BBB in disease, and how these changes have been exploited by engineered nanomaterials for increased transport into the brain.

Graphical abstract

1. Introduction

The blood-brain barrier (BBB) is a specialized physical and functional structure that tightly regulates the transport of molecules between the blood and the brain. The physical structure of the BBB is a precise spatial organization of multiple cell types. Endothelial cell function is highly regulated by pericytes, astrocytes, and continuous tight junctions between endothelial cells. The BBB allows passive diffusion of gasses and regulates the transport of ions, nutrients, and macromolecules through passive and active transporters. While the tight regulation of transport into the brain is imperative for maintenance of healthy brain homeostasis, the restrictive transport also poses a major challenge for the delivery of drugs and other therapeutic agents to the brain; ~98% of all small molecule drugs and ~100% of large molecule therapeutics cannot cross the BBB [1]. The microvessels of the BBB have a relatively small surface area compared to other barriers in the body – 15–25 m² compared to 300–400 m² of the gut and 100 m² of the lung [2]. This can make it difficult to deliver drugs to the brain in sufficient quantities without increasing the risk of systemic toxicity [2]. The BBB is also located in a region of the body with a high metabolic demand, and as a result, the blood flow to the brain is relatively high with a rate of 800 mL min⁻¹, occupying 15–20% of the total blood flow from the heart and 20% of the body’s oxygen and glucose [3], [4]. Cumulatively, these attributes present a major challenge for the access of drugs to the brain and have spurred the development of several drug delivery strategies, including the use of engineered nanomaterials.

Nanomaterials have been actively developed over the past two decades and have several advantageous properties such as encapsulating challenging hydrophobic small molecule cargo, reducing off-target toxicity, and improving drug blood circulation time and distribution. As supramolecular structures, nanomaterials can package high payloads and can have controlled release profiles. Nanomaterials can be optimized in terms of their physicochemical properties, including size, shape, charge, and surface chemistries. Furthermore, engineered nanomaterials can be designed to exploit endogenous biological processes that occur at the BBB in order to deliver drugs to the brain. For example, some nanomaterials can be designed to bind to receptors on the surface of BBB cells, activating signaling pathways that allow them to infiltrate intact, healthy BBB. Engineered nanomaterials can also be used to target BBB dysfunction, which can occur in distinct pathological conditions, such as brain injury or inflammation. In these situations, the integrity of the BBB is compromised, allowing more molecules to pass through the barrier. This can be exploited as a source of targeted therapy, as drugs that would normally be restricted by the BBB can then access the brain and target diseased cells [5]. In this review, we analyze the unique properties of the BBB in health and disease and describe how we can leverage BBB dysregulation to overcome transport challenges of the BBB. We then conclude with a overview of clinical trials that aim to transiently manipulate the BBB and discuss the potential of combining these approaches with nanomaterials. As our understanding of biology and our ability to engineer nanomaterials advance in concert, their convergence can create nanomaterials that can exploit BBB dysfunction.

1.1. Cells of the neurovascular unit

The neurovascular unit (NVU) describes both the physical structure of the BBB and its cellular components, including endothelial cells, pericytes, and astrocytes (Fig. 1). Endothelial cells (ECs) line the cerebral blood vessels and form the main interface with the blood. Capillaries in the brain may have diameters as small as 7–10 µm and the average intercapillary distance is about 40 µm [2]. ECs of the central nervous system (CNS) are connected by two types of junctions, tight and adherens, that resist the paracellular movement of charged molecules such as ions, resulting in high transendothelial electrical resistance. ECs express receptors and efflux and influx transporters that are present on both the luminal and abluminal surfaces, with the luminal surface displaying low levels of leukocyte adhesion molecules [6]. EC surfaces lack fenestrations for small molecule diffusion and generally maintain a net negative surface charge that repels negatively charged compounds [7]. In vitro models of the BBB are useful tools to study BBB development, pathophysiology, and drug transport. Cell sources for these models have been derived from various species and methodologies, and the choice is dependent on the application and salient features to be studied. For example, while primary cultures of human brain endothelial cells are considered to be closest to “native” tissue, their source is limited and therefore immortalized cell lines such as hCMEC/DC have been used in mechanistic studies of transport and expression of receptors and efflux pumps [8]. However, these cell lines have more permissive tight junctions compared to primary human brain endothelial cells, and therefore paracellular transport studies of small molecule drugs in these cell lines may not be reflective of transport in vivo. For more in-depth discussion of BBB models and their use, we refer the reader to reviews by Helms et al. [8] and Naik and Cucullo [9].

Fig. 1. Structure and transport across the blood-brain barrier (BBB).

Brain capillary vessels are lined by a layer of endothelial cells surrounded by pericytes and astrocytic end feet. The transport and exchange of ions and nutrients between the blood and brain is coordinated by specialized tight junctions and several transport mechanisms. Tight junctions control paracellular permeability and include claudins, occludins, and junctional adhesion molecules (JAMs). Gasses such as oxygen and carbon dioxide and lipophilic molecules (< 400 Da, < 8 H-bonds) can passively diffuse across the BBB. Ion transporters are responsible for the exchange of ions to maintain homeostasis across the BBB. ATP-binding cassette (ABC) transporters are efflux pumps that act on certain drugs and xenobiotics. Protein carriers on the luminal and abluminal endothelial surface facilitate the movement of specific substrates such as glucose, fatty acids, and hormones (carrier-mediated transport - CMT). Macromolecules such as insulin and transferrin are transported via receptor-mediated transcytosis (RMT) by first binding to a receptor, followed by endocytosis within a vesicle, and finally exocytosis of the ligand and molecule. Cationic molecules can adsorb to endothelial cell surfaces, due in part to the negatively-charged, mucosal glycocalyx layer, and transport across the BBB via adsorptive-mediated transcytosis (AMT).

The cellular components of the NVU are anchored by an acellular basement membrane composed of heterogeneous extracellular matrix proteins which are synthesized and deposited by the NVU. Pericytes are perivascular cells embedded in the basement membrane and rest abluminal to ECs. The degree of vascular coverage by pericytes and their contractile nature directly correlates with the tightness of inter-endothelial junctions and paracellular transport, capillary diameter and cerebral blood flow, as well as astrocytic polarity and end feet attachment to ECs [10], [11]. Additionally, through the induction of major facilitator super family domain containing 2a (Mfsd2a), pericytes regulate vesicular transcytosis across the BBB [12]. Astrocytes are the major glial cells of the NVU whose cellular processes play a central role in signaling in the BBB. Astrocytes maintain and modulate BBB integrity by strengthening tight junctions and inhibiting transcytosis through angiotensin I signaling and sonic hedgehog-activated Hedgehog signaling in ECs [13]. Together, these cells physically and functionally interact to modulate the passage of molecules that can enter the brain (Fig. 1).

1.2. Transport limitations of nanomaterials across the BBB

There are several mechanisms for the transport of cargo across the BBB. Gasses such as oxygen and carbon dioxide can passively diffuse across the BBB via a concentration gradient. Small molecule, lipid-soluble molecules can diffuse transcellularly, providing their molecular weight is less than 400 Da and they possess a hydrogen bonding capacity of less than 8 hydrogen bonds [1], whereas small hydrophilic compounds can enter via paracellular routes. Molecules which carry a positive charge have an advantage as their cationic nature can interact with the negatively charged glycocalyx, a mucosal layer of polysaccharide macromolecules that lines the luminal surface of ECs (Fig. 1). An intact glycocalyx acts as a malleable, sieve-like barrier to large molecules — passive transport of 150 kDa dextrans is decreased by almost 50% within the glycocalyx layer [14]. Only a small subset of molecules with low molecular weight and defined physicochemical properties can diffuse across the BBB. For most molecules, two defining features directly impact the uniquely low transport rates from the peripheral vasculature across the BBB: endothelial junctional complexes and low rates of transcytosis.

1.2.1. Endothelial cellular junctions

Tight junctions limit paracellular permeability by sealing the space between ECs while also regulating the lateral diffusion of proteins between the apical and basolateral membranes of ECs [15]. The tight junction, also known as zonula occludens, is composed of a variety of proteins, including claudins, occludins, and junctional adhesion molecules (JAMs) (Fig. 1). Specifically, claudins, such as claudin-5, contribute to the structural backbone of the tight junctions. High levels of occludin correlate with decreased permeability and high TEER, and JAMs establish EC polarity and are involved in regulating leukocyte adhesion and migration [16]. Adherens junctions are found intermingled with tight junctions and interact with actin to stabilize inter-endothelial cell adhesion, thereby regulating paracellular permeability. Adherens junction proteins include vascular endothelial cadherin (VE-cadherin) and platelet endothelial cell adhesion molecule 1 (PECAM1). The permeability of the tight junctions is dynamic, and can be regulated by signaling factors that modulate the expression levels and distribution of these proteins [17]. For example, in inflammation, the tight junctions between ECs may be opened as a response to cytokines and other agents, enabling inflammatory cells to cross the BBB [13].

1.2.2. Low rates of transcytosis

In addition to the cells that make up the BBB, various transport mechanisms regulate the passage of molecules through the barrier. These mechanisms include transporters, pumps, and receptors. Most polar molecules cannot diffuse through cell membranes; these molecules are instead transported via specific transport proteins that have polarized expression on the luminal or abluminal surfaces. The orientation of these transporters on either surface can then lead to preferential transport of solutes into or out of endothelial cells. A number of ATP-binding cassette (ABC) energy-dependent efflux transporters, such as P-glycoprotein (P-gp), multidrug resistance-associated proteins (MRPs), and breast cancer resistance protein (BCRP), actively pump many neurotoxic, lipid-soluble compounds out of the brain (Fig. 1) [18]. As a result, it is important to consider the lipidization balance of drugs; increasing the lipid solubility of a drug may increase membrane solubility but may also increase the likelihood of pharmacological clearance by ABC efflux transporters. The above transporters form the delicate balance between influx and efflux that are essential for brain homeostasis. However, this balance also directly affects therapeutic delivery.

Transcytosis of macromolecules via endocytic mechanisms provides the main pathway in which large molecular weight molecules can enter the CNS. These involve either carrier-mediated transport (CMT), receptor-mediated transcytosis (RMT), or adsorptive-mediated transcytosis (AMT) (Fig. 1). CMT is involved in the transport of smaller cargoes such as glucose, hormones, and fatty acids, and is stereospecific [19], [20]. In RMT, macromolecular ligands bind to specific receptors on the luminal surface of ECs which triggers an endocytic event and subsequent vesicle-based transport into the brain [21]. Larger molecules such as insulin and iron transferrin enter via RMT [22], [23]. AMT requires that the molecule be positively charged prior to its electrostatic interaction with cell surface binding sites [21]. Polycationic proteins such as protamine can both bind to the endothelial cell surface but can also penetrate the BBB via AMT [24]. Similarly, cationic proteins such as cationized albumin can be coupled to liposomes and interact with endothelial cells [25], [26]. In both RMT and AMT, cargo can be degraded through the endolysosomal pathway within the endothelial cell, ultimately leading to low rates of transcytosis into brain parenchyma [4]. Major facilitator super family domain containing 2a (Mfsd2a) is a protein selectively expressed in claudin-5 positive brain endothelial cells and is present during the embryonic development of the BBB. Not only is Mfsd2a expressed ~80 times higher in cortical endothelium compared to lung endothelium, in situ hybridization has shown prominent Mfsd2a mRNA expression in CNS vasculature but none in the lung or liver vasculature outside of the CNS. In Mfsd2a^−/− mice, intravenously injected 10kDa and 70kDa dextran were shown to leak into the cortical parenchyma, establishing Mfsd2a as a key regulator of BBB transcytosis [12]; the expression of Mfsd2a in endothelial cells at the BBB alters the lipid composition on the plasma membrane, suppressing the formation of caveolae vesicles and thus transcytosis [27].

2. Disease-associated BBB disruption

As discussed above, the BBB in health is a highly regulated and selective barrier that prevents the transport of a greater majority of therapeutic materials. In this section, we will examine how the pathophysiology of disease can alter the BBB to create opportunities for engineered nanomaterials. These alterations include physical deterioration of the acellular and cellular components of the NVU and changes in surface receptor expression. We structure our discussion around the dynamics of disease onset and how disease changes structure and function of the BBB. Understanding the pathophysiology of BBB that are distinct and common across neurological disorders creates engineering design rules for nanomaterials.

2.1. Physically disrupted BBB

Traumatic brain injury and ischemic stroke are two of the leading contributors to death and disability globally. Both diseases initiate acute dysregulation of the BBB, caused by a physical injury or blockage of blood supply, respectively. This acute event leads to a range of chronic pathological processes including neuroinflammation, leukocyte infiltration, and degradation of tight junction proteins, all of which contribute to increased BBB permeability. This transient and physical disruption of the BBB can be exploited for delivery of therapeutics. However, temporary opening is not well understood, nor well controlled, and therapeutic delivery may be limited to specific sizes and types of cargo. There are also emerging efforts to initiate BBB disruption temporarily using external energy sources, allowing for control over the location, timing, and extent of opening. This controlled BBB permeability has been achieved through multiple energy modes, including chemical, mechanical, and heat which, will be discussed in detail in Section 3.

2.1.1. Traumatic brain injury

Traumatic brain injury (TBI) affects more than 50 million people a year globally and is initiated by external force to the head [28]. TBI pathophysiology occurs in two phases: the primary injury is the immediate effects of the physical trauma to the brain, and the secondary injury is the chronic neurodegenerative processes that unfold over hours to years after the initial injury [29]. The primary injury results in shear injury of blood vessels and tissue destruction, while the secondary injury is ultimately responsible for the progression of BBB dysfunction through the activation of inflammatory cells and the coagulation cascade. Preclinical studies have demonstrated that experimental TBI results in acute BBB disruption. Dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) is a noninvasive technique from which numerous pharmacokinetic parameters such as K^trans (volume transfer constant between plasma and extravascular extracellular space) and v_e (leakage space) can be derived to represent biomarkers for BBB permeability. In a study of human patients with mild TBI (mTBI), DCE MRI revealed that injured patients had higher K^trans and v_e values in both white matter lesions and in normal-appearing white matter compared to the uninjured control group, indicating widespread BBB permeability [30]. In a swine model of head rotational acceleration, multifocal disruption of the BBB from 6–72 hours after injury was shown to lead to an increased extravasation of serum proteins in the brain regions that suffered the greatest tissue deformation during the injury [31]. In support of these imaging studies, neuropathological examinations of post-mortem human brains obtained from teenage athletes with TBI found extravasation and intraparenchymal accumulation of serum proteins, consistent with BBB disruption [32]. BBB dysregulation is not limited to the acute period directly after injury. Several human studies have provided evidence that long-term TBI survivors may experience disturbance of BBB function anywhere from months to years after injury [33], [34].

The precise mechanisms through which BBB dysfunction occurs and is sustained in TBI are not fully understood. However, there are multiple lines of evidence supporting that neuroinflammation contributes to BBB disruption in TBI specifically through damaged tight junctions, changes in matrix metalloproteinase (MMP) levels and activity, and disturbed NVU [35]. In animal models, BBB opening has been linked to the transient increase of proinflammatory cytokines and microstructural vascular abnormalities, reflected in elevated levels of IL-6, TNF-α, and IL-1β 6 hours post-TBI and deformation of pericytes six hours post-TBI, respectively [36]. While the exact causality between cerebrovascular changes and long-term neurological changes have yet to be elucidated, early hypoxia has also been seen within hours after mTBI, likely due to modified vascular reactivity [37]. After injury, immune cells produce and release molecules, like MMPs, that alter the communication between the various components of the BBB [38]. MMPs are involved in ECM turnover and degradation of its constituents. When present at abnormally high levels, MMPs disrupt BBB integrity through the destruction of tight junction proteins. Changes in MMP expression after TBI, specifically ventricular CSF concentrations of MMP-1, MMP-3, and MMP-10, have been shown to increase in TBI patients and correlate to the extent of damage to the ECM [39]. In a weight drop model of TBI in mice, mast cell activation increased and tight junction-associated protein levels decreased after TBI compared to uninjured control brains after 24 hours [40]. Furthermore, evidence of BBB breakdown was reflected in a decrease in the tight junction proteins claudin-5 and ZO-1 [40]. In a mouse model of penetrating TBI, systemically administered nanoparticle accumulation peaked at six hours after injury followed by a significant decrease at 24 hours [41]. While BBB disruption has been consistently observed in humans and animal models, parameters such as the time of onset, duration, and extent of permeability need to be carefully considered when designing nanomaterials strategies to exploit this transient opportunity to access brain tissue. The far-reaching scope of neuroinflammation after TBI and how it progresses from acute damage to the BBB to chronic neuroinflammation is reviewed by Simon et al. [42].

2.1.2. Stroke

Ischemic stroke accounts for 80% of observed strokes and occurs when blood flow in the brain is obstructed due to a migrated clot or debris [29]. When blood flow to the brain is disrupted, the supply of nutrients, ions, and oxygen to the affected area of the brain is limited, which can result in a neurological outburst characterized by necrosis, leukocyte infiltration, oxidative stress, and BBB disruption [29], [43]. BBB disruption occurs immediately after blood vessel obstruction and can last up to several weeks after stroke onset [44]. This enhanced BBB permeability exacerbates neuroinflammation as it allows for the leakage of water and blood components into the extracellular space, contributing to clinical consequences such as edema [45]. The time window of BBB opening after stroke has been thought to be biphasic, with BBB permeability peaking at two distinct time points. In a transient focal cerebral ischemia (tFCI) and reperfusion model, in which blood flow is blocked for a prescribed amount of time and then reperfused, BBB leakage was detected within 30 minutes after tFCI occlusion, characterized by the extravasation of macromolecules less than 3 kDa in size from the blood to brain parenchyma [46]. A biphasic effect was seen when larger macromolecules above 40 kDa in size did not extravasate until 24 hours after tFCI, indicative of a second window in BBB permeability [46]. Although this has been supported by other studies [47], many studies have reported sustained increases in BBB permeability for several weeks post-stroke [48]. For example, after acute ischemic stroke in 42 human patients, DCE MRI indicated continuous BBB disruption up to 90.7 hours after acute ischemic stroke [49].

While the specific processes underlying BBB dysfunction and repair in stroke are still not fully understood, recent studies have provided evidence that MMP activation leads to the degradation of tight junction proteins which ultimately allows for inflammatory cell infiltration that triggers BBB breakdown [49]. Upregulation of MMPs has also been shown to result in endothelial dysfunction and degradation of tight junction proteins. More specifically, MMP-2 and MMP-9 degrade type IV collagen, laminin, and fibronectin, which deteriorates the basal lamina and tight junction components [50]. In a middle cerebral artery occlusion-induced stroke model in rats, elevating MMP-2/9 secretion with glucose reperfusion resulted in decreased levels of ZO-1 and occludin, and the ectopic redistribution of ZO-1 from the cell junctions to the cytoplasm [51]. During ischemia, tight junctions are degraded, which leads to the infiltration of immune cells. Macrophage migration inhibitory factor (MIF), a regulator of the immune and inflammatory responses, has been shown to increase ZO-1 disruption, increase BBB permeability, and decrease expression of claudin-5, ZO-1 and occludin when applied to rat brain endothelial cells [52]. Reactive oxygen species (ROS) are another contributor to the inflammatory cascades [43], [44]. During ischemia, elevated ROS activate microglia and astrocytes which then release inflammatory cytokines (IL-1β, IL-1α, TNF-α, and IL-6) that increase BBB permeability [53]. In a photothrombosis vascular occlusion rat model, ROS levels in the peri-ischemic brain were increased which was associated with an increase in BBB permeability and cellular damage three hours after photothrombosis [54]. When designing therapeutics that target the endothelium during BBB disruption, it is important to consider these molecular and signaling pathways. Recent advances in targeted stroke therapy are discussed in-depth in a review from D’Souza et al. [55].

2.2. Neurodegenerative diseases

2.2.1. Alzheimer’s disease

Alzheimer’s disease (AD) is the most common neurodegenerative disease and is characterized by the accumulation of amyloid-β plaques and the presence of neurofibrillary tangles consisting of hyperphosphorylated tau [56]. The “vascular hypothesis” has been proposed to describe AD disease etiology – impaired function of cerebral blood vessels and cells of the neurovascular unit occur as a result of cerebral hypoperfusion and lead to neuronal death and cognitive dysfunction [57]. Studies in humans using high-content screening of brain microarrays and brain plasma measurements have observed increased extravasation of plasma proteins that are associated with BBB breakdown and vascular dysregulation occurring in early-stage AD [58]–[60]. The presence of BBB dysfunction in AD has been supported by markedly decreased levels of tight junctions measured in both animal models [61], [62] and in postmortem human studies [63], [64]. This disruption of tight junctions in AD is a result of aggressive pericyte degeneration and structural changes to endothelial cells and astrocytes [65]. The apolipoprotein E4 genotype (APOE4), the strongest known genetic risk factor for late onset AD, has been shown to increase BBB damage and induce pericyte loss [66]. In post-mortem tissue samples, pericyte degeneration in APOE4 carriers correlated to the magnitude of BBB breakdown, measured by levels of fibrin and IgG extravascular accumulations [67]. In addition, the pericyte number and coverage area were reduced by 31% in APOE4 carriers compared to 15% in APOE3 carriers, while the IgG extravasation increased by 2.6-fold [67]. Cerebrovascular changes have also been shown to contribute to pericyte loss, indirectly increasing BBB permeability. In an AD transgenic mouse model, parenchymal microvascular damage was accompanied by pericyte loss and rearrangement over time [68]. Similarly, compared to non-AD controls, reductions in mural vascular cells and a 60% reduction in pericyte population were correlated to an increased extravasation of IgG and fibrin across the leaky BBB [69].

The causal relationship, if any, between BBB dysfunction and pathogenic amyloid-β deposits is still not fully understood. Amyloid-β deposits induce oxidative stress that can cause endothelial dysfunction. Structural changes to endothelial cells and astrocytes from elevated ROS decrease their interaction, leading to increased paracellular and transcellular transport. Brain endothelial cells derived from human-induced pluripotent stem cells (iBECs) have been used to investigate molecular mechanisms for how BBB dysfunction is manifested in AD patients. AD-derived iBECs exhibited reduced levels of the efflux transporter P-glycoprotein (P-gp), of which amyloid-β is a substrate, and increased levels of efflux transporters MRP1, MRP2 and BCRP compared to iBECs derived from healthy controls [70]. Excess amyloid-β deposits also induce inflammation resulting in pericyte dysregulation that contributes to BBB dysfunction, but BBB dysfunction exacerbates amyloid-β accumulation by disrupting its clearance [71]. A recent study determined that BBB breakdown as a biomarker of early AD was independent of the presence of amyloid-β and tau [72]. When comparing individuals with early cognitive dysfunction to those with normal cognitive function, increased levels of CSF fibrinogen but no differences in CSF amyloid-β and tau levels were observed, indicating BBB breakdown precedes amyloid-β and tau changes.

2.2.2. Parkinson’s disease

Parkinson’s disease (PD) is the second most common neurodegenerative disease and the most common movement disorder in aging populations. PD pathology includes accumulation of oligomeric α-synuclein protein, also known as Lewy bodies, that contribute to loss of dopaminergic neurons [73]. Evidence of increased BBB permeability in PD includes observed increases in the extravasation of serum proteins, including albumin, iron, and erythrocytes in the striatum of PD patients [74], [75]. Loss of BBB tight junction proteins ZO-1, occludin, and claudin-5 have also been observed in both PD patients [76] and in animal models [77]. Although the mechanism of BBB breakdown in PD has not been studied as extensively as it has been in AD, it is hypothesized to initiate from vascular dysfunction of the basal ganglia. Regional alterations in BBB permeability were visualized with DCE MRI in PD patients in comparison to two control groups: patients with a known cerebrovascular disorder and an apparently healthy group. BBB leakage was increased in the PD group, indicated by higher K^trans value, with the highest K^trans value found in the posterior cortical regions of PD brains [78]. In PD these regions typically display hypoperfusion, which is associated with changes to vasculature, such as loss of capillaries and twisted vessels, that can impair BBB function [78]. Neuroinflammation, specifically the release of pro-inflammatory cytokines such as TNF-α and activation of microglia, has also been found to impact BBB integrity in PD [73], [79]. Like AD, there are genetic factors that can contribute to BBB dysfunction directly. Many PD patients, especially those exposed to pesticides, have been found to carry an allele of the MDR1 gene that encodes a dysfunctional P-gp efflux transporter [80]. Through positron emission tomography (PET), human PD patients were shown to have an 18% increased accumulation of [C]-verapamil, a P-gp substrate, in comparison to the healthy control group, indicative of reduced P-gp function in the midbrain of PD patients [81]. A pixel-by-pixel t-test on the two groups demonstrated that the location of BBB impairment overlapped the region where neuronal damage has been shown to occur in postmortem PD brains [81]. This was evidence that an impaired BBB from environmental factors, like pesticides, can accelerate PD disease progress if not cause it.

2.3. Brain tumors

Brain tumors can arise from a primary tumor that originates in the brain, or more commonly, form due to metastases from the periphery. Brain metastases are the most common central nervous system (CNS) tumor, with more than 150,000–200,000 new patients diagnosed with brain metastases each year in the US [82], [83]. Estimates for brain metastases range anywhere from 10 times more common to equal to the incidence of primary malignant tumors of the brain, such as glioblastoma multiforme (GBM) [84]. Both primary and metastatic tumors in the brain can modulate local BBB and this modified interface is referred to as the blood-tumor barrier (BTB). In order to meet the metabolic demand of the tumor, brain tumors have multiple strategies to increase blood vessel supply, including through the secretion of angiogenic factors like vascular endothelial growth factor (VEGF) to generate new vasculature. Besides mediating angiogenesis, VEGF can also modulate BBB permeability [85]. The growth of newly formed vessels is rapid and poorly regulated, resulting in heterogeneous and dysfunctional vessels. Approaches have been taken to ‘normalize’ tumor vasculature using anti-angiogenic therapies. The concept of vascular ‘normalization’ refers to reverting the leaky tumor vasculature to a structurally and functionally normal vasculature. While these therapies can transiently modulate hypoxia-induced factor 1α (HIF1α)-induced transcriptional responses and leaky vasculature, anti-angiogenic therapies at high doses may decrease BTB permeability and thus affect access of chemotherapeutics when used in combination therapy [86], [87].

Glioblastoma expansion into the brain parenchyma disrupts BBB integrity, permeability, and cellular composition. While the BTB is dysregulated compared to healthy BBB, efflux transporters such as ABC-b1 and ABC-g2 are still present on the peritumoral vasculature as well as on glioma cells. This leads to efflux of systemically administered chemotherapeutics, which limits their ability to reach therapeutic concentrations in the tumor microenvironment without causing systemic toxicity. Furthermore, downregulation of claudin-5 and occludin accompanied upregulation of VEGF and was shown to correlate with BBB pathological opening in orthotopic GBM mouse models [88], [89]. It is important to note that commonly used xenograft mouse models of GBM are created through the implantation of human cell lines, and features commonly observed in these tumors, such as a homogeneously leaky tumor vasculature and noninvasive growth pattern in the brain, may not reflect human GBM where the BTB can be highly heterogenous [90]. Nonetheless, disrupted BTB has been observed in subsets of human patients through MRI and PET [91]. The heterogeneity of the BTB in GBM contributes to the failure of chemotherapeutics for the treatment of GBM in clinical trials [92], [93]. The BTB is also characterized by tumor-promoting glioma stem cells that reside in perivascular niches and differentiate into aberrant vascular pericytes [94]. This then leads to aberrant astrocytic endfeet displacement and proliferation, the latter of which may induce expression of tumor proliferation factors such as tumor necrosis factor-α (TNF-α), tumor growth factor-β (TGF-β), IL-6, and insulin growth factor-1 (IGF-1) [95]. Additionally, the expression of junctional proteins such as VE-cadherin decreases in BTB endothelial cells, thereby increasing BTB permeability [10]. Permeability to the BTB is further characterized by increased transendothelial fenestrations through which nanoparticles smaller than 11.7–11.9 nm in diameter are observed to cross the BTB [96].

While the BTB in primary brain tumors and brain metastases can share phenotypes as described above, the BTB in metastases can also have unique phenotypes. Often these phenotypes vary based on the originating primary tumors; for example, in human breast cancer patients with brain metastasis, brain metastases that arise from HER2-positive tumors tend to have preserved BTBs whereas those that arise triple-negative and basal tumors often have disrupted BTBs [97]. For a more in-depth discussion of BTB in preclinical and human brain tumors, we refer readers to the following review [98].

3. Engineered nanomaterials that exploit BBB dysfunction

Several commonalities in the pathophysiology of BBB dysregulation across neurological diseases include endothelial cell dysfunction, the increased infiltration of cells, and loss and disorganization of acellular structures such as tight junctions, glycocalyx, or basement membrane. Increased permeability due to damaged vasculature presents an opportunity for nanomaterials introduced in systemic vasculature to passively accumulate into the brain parenchyma. The disease process may also lead to changes in protease activity at the BBB compared to healthy tissue due to their upregulation, changes in spatial distribution, or concentrations of cofactors. Moreover, the inflammatory response can also be modulated by nanomaterials to take advantage of endogenous immune cell homing to deliver therapeutics to damaged tissue. Nanomaterials can be engineered to exploit BBB dysfunction and can be used to improve the temporal and spatial specificity of therapeutics to the diseased brain.

Changes in BBB permeability is temporally dynamic. In TBI, biphasic permeability has been observed, but with conflicting evidence. Evans blue (EB) extravasation in controlled cortical impact (CCI) brain injured rats showed severe brain edema peaking at 24 hours after injury followed by a second opening of the BBB at three days [99]. Similarly, a temporal study of BBB permeability following permanent focal cerebral ischemia in rats found EB extravasation peak at three days with a delayed second opening after seven days [100]. Other studies have also observed increased BBB permeability for up to 30 days after ischemic injury [101]. Variability in permeability may be due to differences in animal models, animal strains, and other biological variables such as age and sex. For example, permeability in TBI models has been shown to be sex-dependent [102]. Nonetheless, while further study is needed to fully elucidate the temporal dynamics of these disease pathologies, increased BBB permeability has consistently been observed within hours after injury in several animal models of brain injury and this transient opening has been leveraged by several groups for nanomaterial transport into the brain.

Passive transport of systemically administered nanomaterials into tumors has been proposed to be a direct result of leaky vasculature. This phenomenon was first described by Matsumura and Maeda in 1986 and is known as the “enhanced permeation and retention” (EPR) effect [103]. The EPR effect has been correlated with increased selectivity in accumulation of nanoparticle-formulated chemotherapeutics such as Doxil^® in tumors compared to off-target organs in clinical settings [104]. An EPR-like effect has been observed for nanomaterials in acute brain injuries such as TBI and stroke. In a one-hour transient middle cerebral artery occlusion model in rats, ~130 nm PEGylated liposomes were intravenously injected 0, 1, 3, 6, or 24 hours after reperfusion and were found to accumulate the most in the ischemic brain at 6 hours [105]. Similarly, in a CCI model of TBI in mice, ~80 nm PEGylated liposomes administered intravenously 3–24 hours after injury showed higher accumulation of liposomes in the ipsilateral side of the injured brain compared to uninjured control tissues (~0.4% ID dose/g tissue) at three and six hours after injury and administration (~0.85% and 0.6% ID dose/g tissue, respectively) [106]. This was also observed with other nanomaterials; an ~80 nm electrostatic siRNA nanocomplex administered intravenously between 5 min and 24 hours accumulated significantly when administered up to six hours after injury [41]. In these examples, nanomaterial accumulation was highly localized to the site of the injured brain tissue and not uninjured brain tissue, allowing for a form of passive targeting to the injury. These studies demonstrate that following acute injury to the brain, intravenously delivered nanomaterials have access to the site of injury and thus provides a strategy to leverage disease pathophysiology to passively target injured brain tissue from a systemic administration.

3.1. Design considerations for passive accumulation of nanomaterials across physically disrupted BBB

In Section 2, we described the pathological conditions in which the BBB becomes physically disrupted. While nanomaterials are promising tools for therapeutic delivery to the brain, there are several design parameters that affect the passive accumulation of these systems. In particular, the impact of physicochemical properties of nanomaterials have been studied, including the size, shape, charge, and chemical composition of nanomaterials. We will discuss these properties and how these properties, when optimized, can improve the passive accumulation of nanomaterials.

3.1.1. Size and shape

Size is a design parameter that directly affects nanomaterial transport into the brain. Nanomaterial size has been demonstrated to impact circulation half-times, extravasation through fenestrated blood vessels, and macrophage uptake [107]. As a general rule, nanomaterials with a size over 200 nm have essentially no observed BBB transport, and renal filtration rapidly clears nanomaterials < 5 nm [108], [109]. For example, when gold nanoparticles with diameters of 10, 50, 100, or 250 nm were intravenously injected into rats, 50 nm and larger nanoparticles were found only in the blood, spleen, and liver, while only the 10 nm size nanoparticles were detected in the brain and showed the most widespread organ distribution [110].

There are several studies evaluating size-dependent nanomaterial accumulation into the brain when the BBB is physically compromised. For example, a comparison of 100, 200, and 800 nm PLGA-PEG nanoparticles administered in a TBI mouse model revealed 100 nm nanoparticles exhibit deeper brain penetration and longer blood circulation time than 200 nm and 800 nm nanoparticles [111]. Similarly, Bharadwaj et al. examined the entry of 20, 40, 100, and 500 nm PEGylated polystyrene nanoparticles into the brain after TBI and showed an inverse relationship between nanoparticle size and accumulation within the injured brain penumbra [112]. In a study of nanoparticle accumulation in ischemic brain injury, NIR organic fluorogens with aggregation-induced emission characteristics were fabricated with controlled sizes of 10, 30, and 60 nm, and systemically administered in both healthy rats and a rat model of photothrombotic ischemia (PTI) to induce BBB damage. In the PTI model, these nanoparticles with diameters greater than 30 nm did not extravasate into the brain, while 10 nm diameter nanoparticles were able to diffuse from non-ischemic regions of the brain, suggesting that 30 nm nanoparticles were the most appropriate to delineate the ischemic injury. No nanoparticle accumulation was observed in healthy brains [113]. Besides size-dependent transport across the damaged BBB, accumulation in brain tissue is also influenced by diffusion in the brain extracellular space (ECS), which occupies 20% of the total brain volume. Through optical imaging of dextrans and water-soluble quantum dots, the ECS has been shown to be composed of fluid-filled pores of 38–64 nm in width in healthy brains [114], < 10 nm after terminal ischemia [114], and 7–100 nm in regions with tumors [115]. Overall, optimal nanomaterial size is dependent on not only the material composition of the nanoparticle but also the physiological conditions of the BBB.

The shape of nanomaterials also plays a critical design role in systemic circulation, cellular uptake, and BBB transport [116]–[118]. Most of the extensive research in this area has focused on spherical polymeric nanoparticles. Methods to generate anisotropic nanomaterials with more complex shapes, for example 1D (e.g., worms), 2D (e.g., elliptical and circular discs), and 3D (e.g., UFO and barrels), generally have lower throughput and are less reproducible, therefore limiting their applications [119]. Shape-dependence of nanomaterial-cell interactions have been studied in in vitro models due to the ability to make quantitative measurements more easily. In general, it has been observed that nanomaterials with rod-like shapes have higher uptake than that of spherical shapes in cells [120]. Nowak et al. used a 3D human BBB microfluidic model comprised of a monolayer of hCMEC/DC cells and demonstrated that by using a lower concentration of spheres such that the total endothelial association of spheres and rods is comparable, the basolateral transport flux rate of carboxylated rod-shaped polystyrene particles (~300 nm and ~120 nm, major and minor axes, respectively) is about twice that of 200 nm carboxylated polystyrene spheres. This shows that per unit particle that associates with the brain endothelium, rod-shaped particles are better transported across the BBB compared to their spherical counterparts [109].

The effect of shape on in vivo accumulation in the brain has been studied with varying results. Kolhar et al. functionalized 200 nm polystyrene nanoparticles and nanorods of equal volume with antibodies against transferrin receptors and observed that targeted nanorods exhibited ~7-fold higher accumulation in the brain compared to targeted nanoparticles [120]. In an in vivo model of local intracerebral LPS-induced brain inflammation, however, brain accumulation of 200 nm polystyrene spheres and nanorods with an aspect ratio of two had more dependence on whether the particles were targeted with anti-VCAM antibody than their shape [121]. The shape of nanoparticles in circulation also has a significant impact on hydrodynamics and interactions with vascular targets. The flow over endothelial cells impart fluid shear stress which works against nanoparticle binding. Kolhar et al. demonstrated that in a microfluidic channel coated with ovalbumin, anti-ovalbumin coated nanorods exhibited higher adhesion propensity than that of spheres at shear rates ranging from 15 to 250 s⁻¹ [120].

3.1.2. Surface charge and chemistry

The nanomaterial surface is what the body effectively “sees” and therefore has a large impact on biological interactions. Surface charge is one aspect of the nanomaterial surface that can impact the circulation time and selective accumulation at specific tissue sites. In general, positively charged nanomaterials are more easily internalized into the cell than neutral and negatively charged nanomaterials due to the net negative charge of the cellular membrane [122], [123]. In particular, the BBB has a high density of negative charges due to the dense layer of proteoglycans in the glycocalyx formed on the surface of endothelial cells. Moscariello et al. showed that a denatured and cationized human serum albumin coating on ~25 nm nanodiamonds facilitated BBB crossing in healthy mice and could be tracked at the single cell level [124]. On the contrary, in a separate study, in situ rat brain perfusions of colloidal NPs (~70–130 nm) with varying surface charges demonstrated that low concentrations of anionic NPs had higher apparent cerebrovascular permeability surface-area product than their neutral control counterparts [125]. It is likely that the known short circulation time of cationic nanomaterials, due to interactions with the blood and the reticuloendothelial system, has an influence on the outcomes of these studies [123]. It is also well-understood that blood proteins adsorb to the nanoparticle surface during circulation, and this protein corona impacts nanoparticle in vivo distribution [126]. Specifically, platelet factor 4 is released at high concentrations by activated platelets near vessel injury sites [127]. In a recent study, it was demonstrated that the surface interactions between nanoparticles and activated platelet factors resulted in direct interaction with heparan sulfate proteoglycans on endothelial cells, and induced uptake in cultured brain endothelial cells [128]. Following an intravenous injection of ~80 nm Cy5-labeled gold NPs in mice and activation of platelets in the exposed skin of mice using two photon lasers, confocal imaging revealed that nanoparticles accumulated at the vessel endothelial wall inside the blood vessels within 200 s after platelet activation and the activation site was four times brighter than a negative control nonactivated site [128]. In the context of a physically compromised BBB, damaged vasculature can allow nanoparticle access to the injured brain tissue. In a study of 100 nm polystyrene nanoparticles modified with peptides of various physicochemical properties administered systemically in a CCI mouse model of TBI, nanoparticles functionalized with basic peptides had elevated off-target organ accumulation and short blood half-lives, whereas NPs modified with acidic, zwitterionic, or neutral peptides demonstrated increased blood circulation and increases in relative accumulation in injured brain compared to uninjured brain [129].

Surface chemistry also plays a role in the diffusion of nanomaterials into the brain parenchyma. Previously, it was known that globular, rigid substances greater than 64 nm would not diffuse uniformly through the brain ECS [114]. However, it was later shown that, if densely coated with low molecular weight poly(ethylene glycol) (PEG), much larger particles could penetrate within normal brain parenchyma. Specifically, 114 nm PEGylated nanoparticles diffused in ex vivo human brain tissue easily, while similarly sized COOH-coated nanoparticles had limited diffusion. This result was further confirmed using real-time imaging in mice. Following direct brain cortex injection of PEG- and COOH-coated NPs, COOH-coated particles of all sizes were uniformly stuck in the tissue, whereas 40 and 100 nm particles with dense PEG coatings penetrated up to 200 μm into the tissue within the imaging period [130]. While PEG coating can improve nanoparticle penetration, PEGylation has also been shown to interfere with function; for example PEG sterically interferes with membrane fusion of lipid-based NPs, resulting in poor endosomal escape [131]. Additional advantages and disadvantages of PEGylation as a strategy for improving nanomaterials for drug and gene delivery is covered in a review by Suk et al. [132].

The effect of surface chemistry on nanomaterial accumulation in the brain has also been studied in the context of injured brains. PLGA DSPE-PEG siRNA nanoparticles with four different surface coatings that were previously shown to enhance active penetration of nanoparticles across BBB in other diseases were applied - polysorbate 80 (PS 80), poloxamer 188 (Pluronic F-68), DSPE-PEG-glutathione (GSH), and DSPE-PEG-transferrin (Tf). Specifically, PS 80 coated NPs bind to endogenous apolipoproteins to promote their BBB penetration [133]. Following weight drop-induced TBI in mice, nanoparticles were intravenously injected and at four hours post-injury, brains of mice injected with nanoparticles with PS 80 coating had a 5-fold and 3-fold higher fluorescence intensity when compared with free siRNA and uncoated nanoparticles, respectively. Nanoparticle accumulation across repaired BBB two weeks after injury was also evaluated. Interestingly, while the absolute signal decreased slightly, mice administered with PS 80 coating nanoparticles still displayed the highest fluorescence signal in the brain again compared to free siRNA (~5-fold) and siRNA-loaded PEG-NPs (~3 fold) [134].

3.2. Physical BBB disruption initiated by external energy

Exploitation of the physical disruption in the BBB for the delivery of nanomaterials to the brain parenchyma is discussed in Section 3.1. While this can be an effective strategy, it is only relevant to specific pathologies (e.g., TBI and stroke). Furthermore, relying on the dynamic evolution of disease may be challenging in a clinical setting. As an alternative strategy that allows control of the timing, extent, and spatial location of physical BBB disruption, approaches have developed to transiently manipulate the BBB using external energy. The goal of these approaches is to reversibly disrupt the BBB for local drug delivery. We will discuss how mechanical energy, heat and electricity, and chemical energy have been used for this purpose.

3.2.1. Mechanical energy

Focused ultrasound (FUS) combined with microbubbles (MBs) is an emerging strategy to physically alter the structural features of the BBB and to transiently improve the permeability of therapeutics across the BBB [29]. This strategy works through a process known as acoustic cavitation, whereby low-intensity FUS induces the oscillation of intravenously administered MBs, resulting in shear stress that temporarily disrupts capillary tight junctions between cells (Fig. 2) [135]. The size of nanomaterials that can infiltrate into the brain after BBB opening has been investigated with gold nanoparticles. When gold nanoparticles 3, 15, and 120 nm in diameter were administered systemically for four hours, brain accumulation increased by 3.1, 18.2 and 5.4 times, respectively, compared to non-FUS assisted delivery [136]. FUS has been tested in combination with nanoparticle-based drug delivery vehicles as a treatment for various CNS disorders. The Price group investigated the use of magnetic resonance imaging (MRI)-guided FUS with polymeric “brain-penetrating” nanoparticle gene vectors in an orthotopic glioma mouse model and observed increased total accumulation and volume of spread in tumors [137]. Upon activation of MBs with FUS, there was a ~2-fold increase in nanoparticle signal in the tumor than without FUS [137]. In addition to improving accumulation, FUS-induced BBB opening enhanced nanoparticle dispersion within the tumor itself, reflected in the ~2-fold increase in transfection volume seen in tumors in the presence of MBs and FUS before delivery of nanoparticles when compared to those receiving solely nanoparticles [137]. FUS has also been used for the delivery of mRNA. After systemic injection of ~95 nm lipid nanoparticles (LNPs) carrying mRNA encoding luciferase immediately followed by MB injection and FUS irradiation, an ~18-fold increase in luciferase expression was observed in the irradiated hemisphere of the brain after six hours [138].

Fig. 2. External sources of energy can be used to initiate BBB permeability.

Focused ultrasound can oscillate circulating microbubbles to produce mechanical energy. The resulting shear stress disrupts tight junctions and increases paracellular transport of nanomaterials. Superparamagnetic nanoparticles activated in a low radiofrequency magnetic field can dissipate heat and transiently induce BBB opening. Pulsed electrical fields can destabilize the electrical potential across the endothelial cell membrane, creating pores in the lipid bilayer that can allow for nanoparticle entry. Hyperosmotic agents can change the osmotic pressure of the endothelial cell, leading to cell shrinkage and vasodilation, expanding the endothelium and leading to tight junction disruption.

While FUS-induced BBB disruption has shown preclinical and early clinical success, it is worth noting that FUS-induced BBB disruption can also induce adverse effects such as hemorrhage, brain damage, and inflammation [139]. To mitigate some of these risks, endothelial cell sonoporation with FUS and MBs has been explored as a way to enhance transcellular transport as an alternative to disruption of paracellular tight junctions [140]. To determine whether altering ultrasonic parameters can lead to endothelial-selective transfection, Gorick et al. modulated the peak-negative pressure (PNP) from a low value of 0.1 MPa to 0.4MPa. MBs were conjugated and immunofluorescent overlap was observed with BS-I lectin and GLUT1, endothelial markers, in the FUS-targeted area of the brain [140]. The fraction of endothelial-selective transfection at 0.1MPa was ~85%, and decreased with higher PNPs, with 0.4 MPa resulting in only ~55% sonoselectivity. At the lowest PNP, BBB disruption was not immediately detected after FUS, as measured by T1-weighted MRI images taken of the FUS-targeted regions, nor was any significant inflammatory response observed at 24 hours, as measured by changes in GFAP, compared to higher PNP regimes [140]. Given these results, low-PNP FUS with MBs could represent an alternative strategy to deliver nanomaterial therapeutics in more sensitive disease states where disruption of the BBB and the acute sterile inflammatory response that follows pose greater risks.

Mechanical energy can also be generated by a magnetic source. To track the encapsulated therapeutics and drive magnetic nanoparticles (MNPs) to their target regions, MNPs can be combined with MRI, a technique known as magnetic resonance targeting [141], [142]. This technique relies on the magnetic properties of MNPs. MNPs are physically forced across the BBB by an external magnetic field [143]. The magnetic force itself can also disrupt the endothelial cell-cell junctions vital to keeping the BBB intact. Qiu et al. used NdFeB magnets in combination with MNPs to create an intracellular magnetic force. In an in vitro BBB model, after being exposed to the magnet for 1 hour, adherens junctions were disrupted at intercellular interfaces [144]. Magnetic targeting of MNPs in the mouse tail after a tail-vein injection was also achieved, where MNPs accumulated specifically in the large veins and small venules. Qiu et al. demonstrated local accumulation of the MNPs and altered permeability of the vasculature by correlating circulating free indocyanine green dye signals with the location of the magnet. These results demonstrated that an external magnetic field can localize delivery of MNPs to the target tissue and increase vascular permeability through the disruption of endothelial cell structures [144]. It is important to note that this study was done in the tail, because the magnetic gradients needed for magnetic capture depend on the amount and size of the nanoparticles, and the flow rate of the vessels. However, unlike ultrasound waves, magnetic fields are not attenuated by biological tissue, presenting an opportunity for targeted delivery to deeper tissues [144]. While targeting and distribution specifically within the brain was not observed, magnetic targeting has implications for controlled, targeted brain delivery and thus, has promise to be developed as a strategy to cross the BBB [145]. Lastly, in a method referred to as OptoBBB, Li et al. demonstrated that transcranial picosecond-laser irradiation of gold nanoparticles can temporarily increase BBB permeability by paracellular diffusion [146]. Pressure generation due to thermoelastic expansion of the nanoparticles generates a mechanical wave that initiates calcium influx through mechanosensitive ion channels, leading to calcium signaling and subsequent increase in paracellular permeability through the reorganization of the actin cytoskeleton [147]. In a subsequent study, targeting the gold nanoparticles to the glycocalyx on the luminal surface of the BBB increased OptoBBB-facilitated BBB opening [147].

3.2.2. Heat and electricity

The idea of using hyperthermia to treat malignancies was first introduced in the 19^th century, but it was not until the 1980s that it was explored as a way to treat brain disorders and tumors, such as glioblastoma (GBM) [148]. Hyperthermia represents an opportunity to disrupt the BBB, as it has been shown that high temperatures over 40°C can damage cells, including those vital to the integrity of the BBB [149]. One of the first studies to show this, done by Kiyatkin and Sharma, examined the effect of elevated temperature on brain structures in rats and observed that brains warmed to the highest temperature (38–42°C) had the greatest BBB permeability. This temperature range led to the largest number of leaked blood cells and resulted in brain edema [149]. In these early studies, hyperthermia was typically achieved through warming pads and thermal probes, causing global temperature elevation throughout the whole brain. Since then, technology has greatly progressed to physically localize heating to decrease risk of damage to nearby tissues. This thermal energy can be generated through several sources, such as nanoparticles, magnetic resonance imaging (MRI), or interstitial lasers [150]. Superparamagnetic nanoparticles activated in a gradient or low radiofrequency magnetic field allows them to dissipate heat locally to their surroundings (Fig. 2) [139]. In a proof-of-concept study, Tabatabaei et al. investigated this strategy for brain drug delivery by administering 3–18 nm poly(maleic-acid-co-olefin)-coated Fe₃O₄ magnetic particles via a microcatheter to the middle cerebral artery, immediately followed by 30 minutes of exposure to a radiofrequency field. Evans blue dye extravasation was observed in the presence of magnetic heating of the injected MNPs, resulting in an increase in dye diffusion of ~4-fold compared to MNPs in hypothermia recovery and normothermia environments. This induced BBB opening was transient, as the dye could no longer cross the BBB after two hours [151].

Electroporation (EP) uses pulsed electrical fields to induce transient, localized BBB disruption. These high-amplitude, short-duration electrical fields destabilize the electrical potential across the cell membrane which creates aqueous pores in the lipid bilayer and lead to increased BBB permeability (Fig. 2) [152]. The opening can be reversible or irreversible, depending on the magnitude of electric treatment and whether the pores are able to be resealed. This strategy has been used over the last decade in combination with chemotherapy, known as electrochemotherapy, notably in brain tumors where point-source EP can be used. Point-source EP involves a single intracranial needle electrode placed directly inside the tumor tissue and an external surface electrode on the skin [153]. In a study investigating the effect of point-source EP as a function of time, there was increased BBB disruption in rats 4–24 hours post-EP that correlated with the extent of hypoperfusion and vasoconstriction when measured by MRI [153]. The extent of EP can be modified depending on the frequency of waveforms. Lorenzo et al. developed a second-generation strategy of EP, where intracranial electrodes deliver high frequency irreversible EP (H-FIRE) that applies alternating polarity pulses to increase penetration in target cells and mitigates differences in electrical conductivity in complex tissues [154]. H-FIRE was able to disrupt the BBB for up to 3 days post-treatment, thereby increasing the therapeutic window for drug delivery [154]. Furthermore, low pulsed electrical fields, where two electrodes are pressed against both sides of the skull, have been investigated as a non-invasive method to induce BBB disruption. Unlike traditional EP, which induces transcellular BBB disruption, this method induces paracellular BBB disruption. In a feasibility study, EP with pulses at 100–300 volts could induce subtle BBB opening for up to 30 minutes after treatment in mice The extent of disruption was able to be modulated by changing voltage or number of pulses administered [155].

3.2.3. Chemical energy

Osmotic disruption of the BBB by intracarotid infusion of hyperosmotic agents is a strategy that has been explored since the early 1970s and has been used for the enhanced delivery of antibodies, nanoparticles, viral vectors, water-soluble drugs, and chemotherapeutic agents [156]. Hyperosmotic agents, most commonly mannitol, glycerol, or arabinose, can reversibly open the BBB and allow paracellular transport into the brain via various mechanisms that place stress on the tight junctions [156]. The change in osmotic pressure results in a net flow of water both out of the endothelial cells and out of the brain. This causes shrinkage of the endothelial cells as the water moves into the blood vessel lumen, and vasodilation that expands the cerebrovascular endothelium, respectively, both contributing to the net effect of tight junction disruption (Fig. 2) [157]. On the cellular level, the mechanism underlying osmotic-induced BBB disruption is poorly understood. To address this knowledge gap, Linville et al. developed a tissue-engineered microvessel model of the BBB composed of stem cell-derived human brain microvascular endothelial cells to enable live-cell imaging. After perfusing the engineered microvessels with mannitol and a fluorescent tracer, focal leaks emerged within 10 minutes, indicating both that BBB opening occurs almost immediately after mannitol dosing and that the increase in paracellular permeability is spatially heterogeneous [158]. Mannitol perfusion also triggered morphological changes to the endothelial cells themselves, including shrinkage, endothelium thinning, and the formation of vacuoles [158].

The primary method used in preclinical models to induce osmotic opening of the BBB involves intra-arterial injection of hyperosmotic mannitol followed by infusion of therapeutic agents. Although osmotic BBB disruption presents an opportunity to increase drug exposure and delivery to the brain, there are several limitations associated with the strategy. If not controlled properly, the delivery of the osmotic agent and resulting BBB opening can result in the uncontrolled influx of brain fluid into the central nervous system, leading to edema, epilepsy, and other neurodegeneration [159]. Various factors can affect the delivery efficiency, such as the concentration of mannitol, injection rate, and retention time. To investigate the effects of these various parameters on BBB disruption, Foley et al. tested five different volumes and four flow rates of injection to reach the same final dose of mannitol in mice and found that while varying the flow rate did not affect the degree or location of disruption, the volume had a significant effect, with the highest volume of mannitol resulting in the greatest BBB disruption [160]. One minute following an optimized mannitol delivery, intra-arterially injected adeno-associated virus (AAV) in the intra-carotid artery had a 6-fold increase in transgene expression in the brain tissue compared to when delivery was not preceded by mannitol [160]. To increase reproducibility and consistency of this approach, osmotic BBB opening has been visualized with two-photon microscopy to monitor the movement of osmotic agent across the BBB [161]. Following intra-arterial injection of mannitol, 0.58 kDa rhodamine, and 153 kDa bevacizumab-FITC, Chu et al. used two-photon microscopy to visualize cerebral vasculature at 100 µm depth into the cortex. Osmotic opening in the parenchyma was observed, as measured by the extravasation of rhodamine and fluorescently-labeled antibody, with a half-time of 526.2 s and 618.6 s, respectively [161].

In order to investigate the widespread inflammation induced by whole-brain osmotic-induced BBB permeability, rats treated with intracarotid artery hyperosmolar mannitol were evaluated for changes in protein expression relative to an untreated control group. Within 48 hours after mannitol delivery, there was an upregulation of proinflammatory factors, including IL-1β and TNF-α, with a simultaneous downregulation of occludins, claudin-5, and ZO-1, tight junction proteins necessary for BBB integrity [162]. These apparent induced changes subsided 96 hours after mannitol delivery, however there are possibilities of long-term consequences. While transient immune response induced by osmotic opening has the potential to be exploited in the treatment of CNS disorders, such as AD, and for infiltrative cancers, further research is necessary to determine the extent of clinical applications. As a way to mitigate some of these limitations, Chu et al. combined intra-arterial injection of mannitol with dynamic susceptibility contrast MRI which allowed them to modulate the mannitol speed and infusion rate to estimate the perfusion area and time delay of the brain injury with no long-term consequences [163]. When this MRI-guided approach was implemented on a patient with a brain tumor with intra-arterial infusion of bevacizumab following mannitol-induced osmotic BBB opening, there was an apparent decrease in the size of the enhancing mass, reduced edema, and an improvement in cognitive abilities after the patient was discharged [164].

3.3. Targeting changes in dysregulated BBB

For systemically administered nanomaterials, the gateway from the blood to target tissues is the molecular surface accessible to them. The concept of ‘molecular zip codes’ refers to molecules that are expressed on cell surfaces or on the ECM in distinct pathological states that can be used to localize ligands for specific delivery [165]. Applying this concept to the strategy of nanomaterial targeting can improve cellular- and barrier-specific drug delivery. For example, phage display, a technique for selection of an unbiased library of peptides presented on the surface of phage, a biological nanomaterial, was used to identify a peptide ligand that binds to brain endothelium but not to retinal endothelium [166]. Nanomaterials that target endogenous receptors are often referred to as “molecular Trojan horses,” as they harness existing properties of the BBB to deliver therapies into the brain. Receptors that are uniquely expressed on brain endothelium and not on peripheral endothelium provide a means for specific targeting. An in-depth overview of intact BBB receptor targeting can be found in reviews by Lajoie and Shusta, Anthony et al., and Pardridge [167]–[169]. We will focus our discussion on leveraging dysregulation and changes at the BBB to target nanomaterials. In designing ligand targeted nanomaterials, the affinity and density of targeting molecules that contribute to the overall avidity must be considered; high avidity interactions may lead to nanomaterials adhering to, rather than transporting across, endothelium [170].

3.3.1. Targeting receptors on dysregulated BBB

Several transcriptional changes occur in the BBB across neurological disorders. A transcriptional analysis of stroke, multiple sclerosis, traumatic brain injury, and seizure mouse models revealed that these conditions share similarities in gene expression changes in several pathways, such as inflammation response and leukocyte migration [171]. Although disease etiology varies, neuroinflammation is a hallmark in multiple neurological disorders [172], including amyotrophic lateral sclerosis (ALS) [173], stroke [174], and TBI [175]. Neuroinflammation causes the upregulation of multiple well-studied receptors present on the BBB for immune trafficking. Engineered nanomaterials can leverage these altered surface receptors on dysregulated BBB to increase specificity and transport into the brain parenchyma.

3.3.1.1. Inflammation-associated receptor expression

One well-studied characteristic of inflamed endothelium is the increased expression of leukocyte adhesion molecules such as selectins, VCAM-1 and ICAM-1. P-selectin and E-selectin initiate the leukocyte recruitment process [176], [177], while VCAM-1 and ICAM-1 engage in a later stage of leukocyte recruitment [177]–[179]. These proteins are expressed at low levels on healthy BBB but are highly upregulated during inflammation and can be used to target inflamed tissue.

P-selectin is upregulated on cells in the inflamed GBM microenvironment and on GBM cells directly [180]. One study used dendritic polyglycerol sulfate nanocarriers, which bind to and inhibit P- and L-selectins. These nanocarriers were conjugated to the chemotherapeutic paclitaxel (PTX) via pH-cleavable hydrazone bonds and demonstrated tumor-specific uptake compared to a dendritic polyglycerol untargeted control, as measured by intravital fluorescence imaging and subsequent immunohistochemistry and confocal imaging [180]. ICAM-1 is also upregulated in inflammation. Anti-ICAM-1 antibody targeted polystyrene nanoparticles administered intravenously had 5-fold greater accumulation than control IgG nanoparticles in brains from a TNFα-induced inflammation model [181]. However, an increased accumulation of anti-ICAM-1 nanoparticles was also observed in the lungs, likely due to secondary inflammation in the lungs following the unilateral striatal injection of TNF-α into the brain [181]. ICAM-1 targeting has also been studied in BBB culture models using ~250 nm polystyrene nanoparticles conjugated with control IgG or anti-ICAM-1 antibody. After TNF-α stimulation, ICAM-1 was upregulated in cultures of NVU cell types, resulting in ICAM-1 targeted nanoparticles internalizing 4.1-fold more in endothelial cells, 2.3-fold more in astrocytes, and 1.7-fold more in pericytes compared to when no inflammation was induced. This study also examined the route of endocytosis using small molecule inhibitors: amiloride, a CAM-mediated endocytosis inhibitor, filipin, a caveolae-mediated endocytosis inhibitor, and monodansylcadaverine, a clathrin-coated pit inhibitor. Amiloride treatment induced a 30–40% reduction of uptake nanoparticles in astrocytes, endothelial cells, and pericytes compared to a control medium, indicating that CAM-mediated endocytosis may be the main route of uptake [182]. In a comparison of multiple receptors upregulated in inflammation, Marcos-Contreras et al. compared liposomes conjugated to antibodies against VCAM-1, TfR-1, and ICAM-1. In healthy mice, VCAM-1 targeted liposomes accumulated in the brain 4-fold more than TfR-1 targeted liposomes and one order of magnitude more than ICAM-1 and control IgG targeted liposomes. In a TNF-α-induced mouse model of inflammation, VCAM-1 targeted liposomes accumulated in the inflamed brain up to 6.0% ID/g, compared to 1.2% ID/g for ICAM-1 targeted liposomes [183]. The relative selectivity of VCAM-1 nanomaterial targeting in the context of inflammation shown in this study demonstrates how both ligand targeting and the underlying pathophysiology can be leveraged to improve nanomaterial transport and specificity.

3.3.1.2. Receptors upregulated in cancer

While the BTB retains some aspects of drug barrier function, it often undergoes changes in expression of cellular receptors that can be leveraged for targeting [184]. Transferrin receptor, which is expressed on healthy BBB for the transport of iron, is upregulated on GBM cell surfaces. Transferrin has been extensively investigated as a ligand for drug delivery and nanomaterial applications in healthy BBB [185], [186]. Magnetic silica PLGA nanoparticles modified with transferrin administered to an orthotopic mouse model of glioblastoma increased tumor accumulation ~4–5-fold compared to untargeted nanoparticles. With the administration of doxorubicin and paclitaxel dual-loaded targeted particles, up to a 47.5-fold reduction in tumor size was achieved [187]. LRP1 is another receptor that has been used to increase transport on healthy BBB as it is upregulated in glioblastoma cells [188]. Paclitaxel is a chemotherapeutic drug that is actively transported out of endothelial cells by P-gp, preventing the accumulation of therapeutic relevant concentration in the brain. To increase paclitaxel uptake in the brain, three paclitaxel molecules were conjugated to the peptide Ang-2, a peptide ligand for LRP1. Following in situ brain perfusion via the carotid artery in mice, this nanomaterial conjugate, ANG1005, had a ~2.5-fold increased uptake into the brain parenchyma compared to free paclitaxel; while free paclitaxel is a substrate for Pg-p, ANG1005 was shown to not interact with P-gp [189]. In a mouse model of glioblastoma, ANG1005 increased the median survival time of mice by 15% compared to the vehicle control when administered intravenously [189]. In this example, the modification of paclitaxel with a targeting moiety also prevented small molecule drug efflux from the BBB. Several clinical trials have supported the promise of ANG1005 as a treatment for breast cancer brain metastases [190], [191]. Preliminary data from a Phase I clinical trial demonstrated tolerability although limited efficacy for its use in glioblastoma, with further results pending Phase II efficacy studies [192], [193]. Another study targeted Cas9/sgRNA ribonucleoprotein encapsulated in disulfide cross-linked polymeric shells with Ang-2, designed to target and release cargo inside GBM cells [194]. These ~30 nm targeted nanoparticles demonstrated 2.5-fold more tumor accumulation when compared to untargeted nanoparticles and 15.3-fold more tumor accumulation than Cas9/sgRNA alone, resulting in a ~3-fold increase in median survival time, 33.8% indel efficiency in the target gene, and significant inhibition of GBM tumor growth [194].

3.3.2. BBB targets made accessible by BBB dysfunction

While the tight junction complex is typically sequestered from larger nanomaterials in healthy BBB, BBB dysregulation can expose these paracellular molecules for targeting. Claudins are a family of proteins that are an important component of tight junctions that regulate ion permeability across the BBB. Claudins are located in epithelial and endothelial cells in tissues with tight junctions [195]. The incorporation of claudin family members into tight junctions, and therefore tight junction permeability, is known to change with different contexts. Claudin-5 is responsible for impermeable tight junctions at the BBB [196]. Claudin-1 is upregulated with age and is associated with increased BBB permeability; its overexpression results in its interaction with other tight junction proteins such as ZO-1, which is thought to result in the replacement of claudin-5 in the junctions [196], [197]. In stroke, there is upregulation of claudin-1 and downregulation of claudin-5. Based on this, the peptide C1C2 has been used to bind claudin-1 and interfere with its incorporation into tight junctions in order to maintain tight junction structure and prevent leakiness of the BBB [196]. This was demonstrated in BMEC monolayers in vitro and in a middle cerebral artery occlusion stroke model by comparing inulin leakage over a period of 28 days [196]. Another study observed increased uptake of C1C2-targeted PEGylated gadolinium nanoparticles into the brains of aged mice (12 months) that had upregulated claudin-1 compared to younger mice (2 months), particularly in the corpus callosum. Using MRI, the permeability constant for C1C2-targeted particles revealed that permeability was 3-fold higher in aged mice compared to younger mice. Furthermore, colocalization of nanoparticles and claudin-1 was observed via microscopy of brain tissue two hours post-injection. One caveat was that increased nanoparticle accumulation was also observed in other organs that have increased claudin-1 expression in aged mice [197].

3.4. Biological modification of BBB

3.4.1. Altering BBB transport rates

Low rates of transcytosis are a defining feature of the BBB that is central to its ability to tightly regulate the molecules that are exposed to the brain; however, this low rate of transcytosis is also a major obstacle to the transport of nanoscale materials. New discoveries in biology have precipitated emerging strategies to influence basal transport rates across the BBB. For example, sphingosine 1-phosphate receptor-1 (S1PR1) is an important receptor for BBB integrity. Knockdown of S1PR1 results in BBB leakage for small tracers (< 3 kDa), but not for larger ones (> 10 kDa) via the alteration in the subcellular tight junction protein distribution [198]. Consequently, one method of opening the blood-brain barrier is to inhibit S1PR1, although the applications would only be relevant to small molecule drugs or the smaller range of nanomaterials. Two drugs that inhibit S1P1 function (fingolimod and NIBR-0213) were shown to increase accumulation of a 1 kDa tracer. The effects of the drugs were shown to be reversible and did not show major signs of CNS inflammation [198].

A molecular target to modify transport at the BBB relevant to nanomaterials is Mfsd2a. Mfsd2a is highly enriched on the cell membranes of brain microvasculature over peripheral vasculature [199] and suppresses caveolae-mediated transcytosis. The function of Mfsd2a is to transport lysophosphatidylcholine-docosahexaenoic acid (LPC-DHA) and other LPC-coupled derivatives [200]. By regulation of the lipid composition of BBB endothelial cells, Mfsd2a suppresses caveolae vesicle formation and transcytosis, likely due to the disruption of caveolae domains and displacement of caveolin-1 by DHA [27]. Temporarily inhibiting Msfd2a increases rates of transcytosis in the BBB and several studies have examined the potential for Mfsd2a inhibition in drug delivery. Yang et al. showed that in mice, downregulation of Mfsd2a with siRNA and Mfsd2a genetic knockout both resulted in increased BBB permeability as measured by Evans blue dye extravasation. Conversely, AAV-mediated overexpression of Mfsd2a decreased Evans blue extravasation, potentiating a strategy to prevent BBB leakiness after intracerebral hemorrhage [201]. This manipulation of transport can be exploited by nanomaterials. In one study, priming nanoparticles loaded with tunicamycin, an Mfsd2a inhibitor, were delivered first to increase transcytosis rates at the BBB. Doxorubicin-loaded nanoparticles targeted with hyaluronic acid to bind to CD44 on breast cancer metastases as well as transcytosis targeting peptides (LRQRRRLYC) that bind to heparin sulfate proteoglycans on the BBB were subsequently injected. Accumulation into brain metastases was improved by the priming Mfsd2a-inhibiting nanoparticle pre-treatment, with a 4.3-fold increase in brain accumulation of doxorubicin-loaded nanoparticles compared to when no priming pre-treatment was used. This increased accumulation translated to an improved median survival of mice from 27 days to 46 days with and without priming pre-treatment, respectively [200]. Alteration of BBB transport rates may have unintended negative consequences for brain health, and the safety of this approach remains to be evaluated. However, the transient manipulation of basal transport rates of the BBB could prove to be an effective strategy for delivering drugs into the brain parenchyma.

3.4.2. Vasoactive agents

BBB permeability has been altered using vasoactive compounds. Vasoactive compounds, most commonly vasoactive peptides (VAPs), modulate vascular activity through their vasodilating or vasoconstricting properties and have been shown in literature to increase BBB permeability preclinically [202]. The VAPs that have shown the most promising evidence to support BBB opening are regadenoson and bradykinin, in addition to their various analogs [202]. Regadenoson is an FDA approved adenosine 2A receptor agonist typically prescribed as a cardiac stress agent, although previous studies have demonstrated its ability to increase BBB permeability in both mice and rats [203]. In a more recent study, temozolomide, a chemotherapeutic, was administered with and without regadenoson in rats with an intact BBB. After 120 minutes, they observed that the temozolomide concentration in the brain of rats was 60% higher when regadenoson was also administered compared to temozolomide alone [204]. Bradykinin, a peripheral vasodilator, was investigated in 1986 in combination with other pharmacologics to try to control and modulate BBB permeability. Bradykinin infusion resulted in significant BBB breakdown, evidenced by the number of vessels showing leakage of a horseradish peroxidase (HRP) tracer into the parenchyma [205]. In subsequent studies, the use of bradykinin analogs, like cereport, a B2 bradykinin receptor agonist, have been adopted due to the short-half life and narrow therapeutic index of bradykinin [206]. The ability of cereport, also known as RMP-7, to increase BBB permeability was investigated in a rat glioma model in which they administered a range of intravenous doses of RMP. They found that there was a dose-dependent effect on BBB permeability, with the highest permeability seen in the brain tumor and surrounding tissue [207]. Although regadenoson and cereport have both shown increased BBB permeability in animal studies, both have failed to show a benefit in humans when co-administered with contrast agents or chemotherapeutics, respectively [204], [208].

3.5. Leveraging cells and cellular products for delivery across the blood-brain barrier

Cells have a natural ability to home to diseased tissue that can be leveraged for drug delivery. Using cells as delivery systems capitalizes on complex natural biological components and their structures and has been shown to enable long circulation, targeted biodistribution, and biocompatibility [209]. In particular, cells have evolved mechanisms to infiltrate into the BBB in pathological states. Both cells and cell-derived products, such as their membranes or exosomes, have been used to increase transport of therapeutics into the brain [210].

3.5.1. Cellular hitchhiking for nanomaterial delivery

3.5.1.1. Immune cell hitchhiking

During normal physiological conditions, low levels of circulating immune cells infiltrate across the BBB. However, in many diseases of the brain, neuroinflammation disrupts this homeostasis, stimulating the release of cytokines and chemokines that activate brain endothelial cells to express endothelial adhesion proteins such as VCAM-1 and ICAM-1. This then directs the widespread recruitment and infiltration of peripheral inflammatory cells [184], [211]–[213]. The infiltration of immune cells into the brain parenchyma during inflammation has been leveraged to deliver therapeutics [214] and has been termed as cellular “hitchhiking”. To employ infiltrating immune cells as delivery vehicles, therapeutic cargo can either be internalized or functionalized to the surface of these cells. These strategies have leveraged several sources of immune cells for hitchhiking, predominantly leukocytes such as macrophages and neutrophils.

The Mitragotri group introduced the concept of a cell “backpack” – a synthetic drug-laden disc that has dimensions that are too large to be phagocytosed by macrophages, and thus are carried by the cell [215]. In an application of this concept to the inflamed brain, Klyachko et al. constructed polymer discs 7–10 µm in diameter and a few hundred nanometers in thickness through photolithography and layer-by-layer assembly. The multiple layers of the disc include a release region, payload region encapsulating dye, and a cell attachment region comprised of anti-CD11b antibodies for macrophage conjugation. Using the natural infiltration of the macrophage into the inflamed brain in an LPS-induced encephalitis mouse model, co-localization of DiO-labeled macrophages and DyLight 550-labeled ‘backpacks’ was visualized by confocal microscopy in the inflamed brain, whereas no fluorescence was found in the inflamed brain when backpacks were injected alone without their cell carriers. Importantly, while the attachment of the polymeric disc backpacks onto macrophages increased their size, backpacks did not alter cellular functions important for brain infiltration, including cell adhesion [216].

Nanoparticle-formulated drugs can also be loaded inside of leukocytes taking advantage of the natural phagocytic activity of macrophages and neutrophils (Fig. 3). M1 polarized macrophages derived from the bone marrow were loaded with doxorubicin/poly(lactide-co-glycolic acid) (PLGA) nanoparticles. When administered systemically in an orthotopic U87 glioma mouse model, macrophage-mediated transport of ~160 nm DiR-loaded PLGA nanoparticles led to an approximately 2-fold increase in average fluorescence intensity in the brain compared to PLGA nanoparticles alone. Macrophage-mediated doxorubicin-loaded PLGA nanoparticle transport significantly prolonged mouse survival with median survival of 38.5 days compared to 21, 26.5, and 25 days for free doxorubicin, doxorubicin-loaded PLGA nanoparticles, and macrophages, respectively [217]. Neutrophils have a natural ability to penetrate the BBB and BTB. In one study, neutrophils encapsulating liposomes that contained anticancer therapeutic paclitaxel were used to treat GBM. Neutrophils were isolated from mouse bone marrow and loaded with ~100 nm paclitaxel-loaded cationic liposomes. In a glioma surgical resection model implanted with G422 and C6 cells, tumor resection in the brain induced an inflammatory reaction releasing inflammatory factors in the blood for neutrophil recruitment. Intravenously administered paclitaxel-liposome loaded neutrophils then localized to the inflamed brain ~4-fold more than in normal brains at 48 hours post-injection. The longest survival in tumor-bearing mice was observed following treatment with paxlitaxel-liposome neutrophils at a dosage half that of Taxol and paclitaxel-liposome controls; the 50% survival rate was extended 61 days compared with 29 days for Taxol and 38 days for paclitaxel-liposomes [218].

Fig. 3. Nanomaterials can leverage the natural homing ability of cells and cellular products to infiltrate the BBB in disease.

Nanoparticles (NPs) can be loaded inside of leukocytes to take advantage of their natural ability to infiltrate inflamed BBB. Nanoparticle drug delivery systems can also target specific immune cell through ligand targeting in situ. Particles can be surface-functionalized onto RBCs and sheared off from the RBC surface onto the endothelium as RBCs squeeze through capillaries. Nanomaterials can be coated with RBC membranes to increase their circulation time and can also be conjugated with ligands that target upregulated receptors on diseased endothelium. Exosomes are natural nanomaterials secreted by cells. Macrophage-derived exosomes can display surface peptides to specifically target endothelial receptors. Exosomes can then carry therapeutic cargo, such as siRNA and small-molecule drugs, across the diseased BBB via an immune cell-mediated endocytosis pathway.

Rather than loading of immune cells ex vivo, nanoparticle-formulated systems can also target specific immune cells in situ following systemic administration in vivo (Fig. 3). Multiple sclerosis patients often present with a massive influx of leukocytes, specifically monocytes, from the peripheral circulation into the CNS. In one study, uptake efficiency of high-density lipoprotein-mimicking peptide-phospholipid scaffold (HPPS) nanoparticles was compared in immune cells in the peripheral blood in an experimental autoimmune encephalomyelitis mouse model. More than 98% of blood monocytes phagocytosed HPPS compared to only 3% of the neutrophils at 24 hours post-injection [219]. In images of frozen sections of the spinal cord from an experimental autoimmune encephalomyelitis mouse model, a large number of HPPS particles was shown to pass through the BBB and co-localized with GFP-expressing inflammatory monocytes in the spinal cord [219]. The inflammatory response after cerebral ischemia involves the recruitment of leukocytes to the site of stroke and injured penumbra. One study exploited the migration of these immune cells to deliver nanoparticles to the brain. Nanoparticles were modified with cyclic RGD (cRGD) peptide to bind to integrin αvβ1 receptors highly expressed on leukocytes. Following intravenous administration of ~150 nm cRGD-liposomes in a rat model of ischemia and reperfusion injury, the percentage of monocytes with liposomes measured in the blood over 12 hours was much higher compared to when untargeted liposomes were administered, regardless of sampling time. This strategy also demonstrated a therapeutic effect. ER, a potent scavenger of hydroxyl radicals, was loaded into cRGD liposomes and administered systemically after reperfusion. The infarct volume from coronal sections of ER-cRGD-liposome treated group after three hours perfusion was the lowest of all the groups at any time point [220].

3.5.1.2. Red blood cell hitchhiking

While immune cells can naturally infiltrate inflamed tissues, they are highly phagocytic which can lead to degradation of cargo. As an alternative cell vehicle, red blood cells (RBCs) are long circulating and highly abundant. Attaching nanomaterials to the surface of RBCs has been shown to decrease the rapid clearance of nanoparticles by the liver and spleen, and instead lead to transient accumulation in the lungs [221]. It has been observed that when RBCs with surface nanoparticles pass through narrow diameter capillaries such as those found in the lung, nanoparticles can be sheared off from the RBC and transferred to capillary endothelium as the RBCs squeeze through capillaries (Fig. 3) [222]. To circumvent the lung as the first narrow diameter capillary bed after tail-vein injection, Brenner et al. extended this RBC-hitchhiking strategy for brain accumulation by administration through an intra-carotid artery (IA) injection. This was applied in the context of stroke and heart attack, since patients already have an IA catheter in place as standard of care. With this approach of delivery of RBC-adsorbed ~100 nm nanoparticles via IA catheter, a high 12% ID in the brain was achieved. This approach achieved high specificity: the ratio of nanoparticles present in brain vs. liver was 14.3-fold higher and brain vs. blood was 27-fold higher than that for free nanoparticles [223].

3.5.2. Red blood cell-based delivery across the BBB

Due to their long life-span and high surface to volume ratio, RBC membranes have been used to coat nanomaterials, combining the biological advantages of native RBC membranes and the payload capacity of nanomaterials (Fig. 3) [224]. RBCs are anuclear which make them particularly useful as a source of membranes. When compared with traditional polyethylene glycol coating methods to create a “stealth” effect, this strategy suppressed internalization by J774A.1 macrophage cells by a further 20% after 24 hours of incubation [225]. To combine the long circulation times of RBC-wrapped nanoparticles with specific targeting to the BBB, PLGA nanoparticles were coated with an RBC membrane and conjugated with a the D-peptide fragment of candotoxin, ^DCDX; this peptide has high affinity for nicotinic acetylcholine receptors and has been shown to migrate across the BBB [226]. Following intravenous administration, an approximately 2-fold higher distribution of nanoparticles was found in the glioma of mice treated with ^DCDX targeted nanoparticles compared to untargeted nanoparticles, which primarily colocalized with tumor blood vessels. Targeted 75 nm RBC-wrapped PLGA nanoparticles carrying doxorubicin lengthened the survival of glioma-bearing mice 4.3-fold compared to untargeted nanoparticles [227].

3.5.3. Exosome-based delivery across the BBB

Exosomes are natural membrane-coated vesicles produced and secreted by cells that range in size from 30–100 nm [228]. Exosomes have the capacity to package complex compositions of cargo involved in cell-cell communication such as proteins, nucleic acids, and lipids. In addition, exosomes can have long circulation half-lives and the potential to have high biocompatibility, as they share similar physical and biological characteristics to their cellular origin [229]. Recent work has demonstrated that exosomes play a key part in intercellular transmission of genetic information between cells [230], [231], representing a natural delivery system within the body. Exosomes have the capacity to cross the BBB through a variety of mechanisms, although transport rates are influenced by the donor cell and physiological state of the BBB [232]. Cellular engineering efforts have been made to reprogram the contents and surface of exosomes, and recent studies have applied this for drug delivery, including small molecules, RNA, protein, and plasmids. For surface engineering, targeting ligands can be fused with endogenous exosome membrane proteins for display on the exosomal membrane. Due to their high biocompatibility, stability, and low immunogenicity, exosomes have been utilized to deliver exogenous payloads across the BBB in preclinical models and in human patients across a variety of neurological disorders [233]–[235].

Macrophages are resident innate immune cells of the CNS. They continuously monitor the CNS surroundings to sense for disturbances and naturally home to states of inflammation in the brain. Macrophage-derived exosomes retain this intrinsic homing property and have shown promise as a delivery vehicle for many diseases where neuroinflammation is present (Fig. 3). For example, in a LPS-induced mouse model of inflammation, systemically administered naïve macrophage-derived exosomes localized to the brain parenchyma 3.1-fold faster and accumulated 5.8-fold more in the inflamed brain than in the healthy brain [236]. A brain perfusion study was also performed by perfusing the mouse brain with exosomes in physiological buffer in both healthy and inflamed mice in order to exclude the potential involvement of an immune cell-mediated endocytosis pathway. Exosomes showed a significantly (~1.75 fold) higher brain/perfusion ratio than a co-administered vascular marker BSA in both healthy mice and the brain inflammation model, suggesting exosomes can enter the brain independently of infiltrating immune cells in healthy and inflamed brain [236].

There have been major efforts to modify the surfaces of exosomes to influence their cell tropism. Rabies virus has natural tropism to neurons and brain endothelial cells. A peptide taken from the rabies virus glycoprotein (RVG) is a ligand for the nicotinic acetylcholine receptor [237] and has been used to target the brain in multiple nanoparticle systems [238], including exosomes [239]. In order to create RVG-targeted exosomes, RVG was fused with Lamp2g, an abundant exosome membrane protein, and this construct was transfected into dendritic cells to create RVG-targeted exosomes that were then loaded with exogenous siRNA [240]. These exosomes were intravenously injected in healthy mice and delivered siRNA to neurons, microglia, and oligodendrocytes in the brain, leading to knockdown of BACE1, a therapeutic target in AD, by 60% (mRNA) and 62% (protein) as determined by qPCR and western blot from cortical sections of the brain. In another example of exosome targeting, targeting peptide Angiopep-2 and cell-penetrating peptide Tat were fused with Lamp2b proteins and stably expressed in HEK293T cells. These dual-targeted, cell-penetrating exosomes loaded with doxorubicin had a 6-fold increased accumulation into tumors relative to untargeted exosomes in an orthotopic glioma mouse model. Furthermore, tumor growth was reduced by 6-fold relative to free doxorubicin group and 2-fold to untargeted exosomes [241].

The exosome surface can also be directly modified with chemical manipulation. Neuropilin-1 (NRP-1) is a transmembrane glycoprotein that is overexpressed in glioma cells and the BTB. Using click chemistry, a peptide ligand of NRP-1, RGERPPR (RGE), was conjugated onto the surface of exosomes derived from the mouse macrophage cell line RAW 264.7 [242]. DiI-labeled RGE-modified exosomes encapsulating curcumin were intravenously administered in an orthotopic xenograft tumor model in mice and exosome accumulation was observed over 24 hours. Strong fluorescence was measured in the tumors after RGE-targeted exosome administration whereas no fluorescence was detected after free dye administration [242].

4. Clinical directions of external energy-mediated opening of the BBB

Modulation of BBB permeability with external energy is a promising direction to enhance nanomaterial-mediated drug delivery into the brain. While chemical opening has been explored in the past, focused ultrasound (FUS)-mediated brain delivery has been the most prevalent in recent clinical trials. Although a vast majority of these trials were intended for brain cancer applications, this technology has adapted to now include Phase I clinical trials for neurodegenerative disorders like AD, ALS, and PD.

4.1. Chemical opening

While declining in prominence in the last decade, investigational therapeutic strategies such as osmotic opening of the BBB in combination with chemotherapies, have been used in cancer patients due to the lack of therapeutic options in brain cancers. In one such study, patients with malignant brain tumors were treated with intra-arterial chemotherapy 5 minutes after being administered mannitol to induce BBB opening. This procedure was repeated across 5 different universities and CT scans across all centers revealed that 75% of patients with primary central nervous system lymphoma saw no tumor enhancement [243]. More recent studies have investigated the safety and efficacy of osmotic BBB disruption specifically for recurrent malignant glioma [244]–[246]. After a single dose of intra-arterial administration of mannitol followed by bevacizumab, the Boockvar group saw a 46.9% median reduction in the volume of tumor enhancement and a median MR perfusion reduction of 32.14% [244]. In a follow-up prospective study, the group repeated the procedure but analyzed overall survival and progression-free survival of recurrent glioblastoma patients. After receiving a single dose of intra-arterial bevacizumab after mannitol-induced BBB disruption, the median progression-free survival was 10 months, an improvement from the 5.6 months seen with intravenous administration of bevacizumab alone [245].

4.2. FUS

To this date, FUS has been tested in various clinical trials to evaluate the safety, tolerability, and feasibility of transiently opening the BBB to allow for targeted delivery of therapeutics to the CNS. Clinical applications of FUS were initially investigated in cancer, as there is a lack of therapeutic options for brain cancers that opens the doors to investigational therapeutic strategies. In 2016, one of the first clinical trials (NCT02253212) utilizing this strategy as a mechanism to circumvent the BBB used an implantable ultrasound device system known as the SonoCloud in combination with systemically injected microbubbles. Patients with recurrent glioblastoma underwent BBB disruption monthly via the implanted SonoCloud followed by intravenous treatment with carboplatin. Patients underwent dose escalations in the range of 0.5 MPa to 1.1Mpa, for a maximum of 41 sonications, and contrast-enhanced MRI images indicated BBB disruption for patients receiving acoustic pressures of at least 0.65 Mpa, with the depth of BBB opening increasing with higher pressures. Regarding safety, no patients presented with clinical symptoms such as hemorrhage, ischemia, or edema and the treatment was regarded as well-tolerated [247]. In another study, MR-guided focused ultrasound (MRgFUS) in combination with intravenously injected microbubbles was used to administer chemotherapy to 5 patients with glioma in a Phase I clinical trial (NCT02343991). This device, the ExAblate Neuro, in comparison to the SonoCloud, is transcranial, noninvasive, and provides the added benefits of spatial control and a more uniform BBB opening [248]. In this study, patients were administered a systemic sub-therapeutic dose of chemotherapy 1 hour prior to receiving sonications. Contrast-enhanced MRI images confirmed BBB opening in all patients and revealed a 15–50% enhancement in signal intensity in the sonicated tissue versus the non-sonicated tissue 20 hours post-treatment [248]. Although both studies saw established preliminary safety and BBB opening, later-phase trials are necessary to establish efficacy.

Outside of tumor-targeted therapies, FUS has been tested clinically for the treatment of AD and ALS. Insightec, the first and only company to bring a focused ultrasound system to market, created the ExAblate Neuro system which it has been used in many recent clinical trials, which can be read here [249]–[251]. Notably, the ExAblate Neuro system was used in a Phase I clinical trial (NCT02986932) where MRgFUS was used to open the BBB in 5 patients suffering from early to moderate stage AD. Contrast extravasation within the sonicated volume was seen immediately after MRgFUS in all patients, confirming that the BBB was successfully opened [252]. The transient nature of BBB opening was demonstrated when, after 24 hours, no extravasation was observed, indicating closure of the BBB [252]. This is the first human study to report non-invasive and reversible BBB disruption and supports continued investigation in the ExAblate Neuro system as the procedure was well tolerated by all patients with no serious clinical events reported. In the case of ALS, the MRgFUS was also performed using the ExAblate Neuro system (NCT03321487). This study investigated the transient permeability of the BBB following focused ultrasound guided by functional MRI to target the primary motor cortex in 4 patients with ALS. Gadolinium leakage at the area of sonication confirmed that MRgFUS-mediated opening of the BBB occurred specifically at the primary motor cortex, and no serious adverse events related to the procedure or device were reported up to 60 days post-procedure [253]. Although the safety and efficacy of FUS studied in recent clinical trials did not reveal any immediate concerns, it is important to note that FUS-BBB opening likely has significant secondary effects including, but not limited to, initiation of an inflammatory response, clearance of amyloid-β and Tau proteins, alteration of cerebral blood flow, and suppression of neuronal activity [254]. These have been demonstrated in literature preclinically, and are factors to consider for future clinical studies involving FUS [255]–[258].

4.3. Future clinical directions

There are many other clinical trials that are still in their infancy and actively recruiting volunteers for additional applications. For example, clinical trials are being initiated to investigate controlled BBB opening to treat malignant brain tumors in pediatric patients (NCT05293197, NCT05630209). In particular, these strategies are now being investigated for macromolecular biologic cargo. The ExAblate system combined with monoclonal antibody Aduhelm infusion therapy is being tested in patients with mild AD or mild cognitive impairment, with an estimated completion date of July 2029 (NCT05469009, NCT03671889). A new Phase I/II clinical trial has recently been approved to use MRgFUS for intracerebral delivery of the enzyme GCase to patients with PD (NCT05565443). Positive outcomes from these trials to deliver protein cargo with FUS-mediated BBB opening combined with continued preclinical research with nanomaterials may provide a promising avenue for the delivery of nanomaterials with FUS in the future.

5. Conclusion

Advancements in the design and understanding of nanomaterials have provided new technologies for disease prevention and treatment. However, delivery of nanoscale materials to the brain has remained a challenge. Our understanding of the BBB as a dynamic structure and the identification of molecular players involved in its regulation continues to evolve. As biological understanding of the BBB and engineering advances converge, we will be able to create technologies that exploit changes in the BBB in disease. Recently, single-cell sequencing has been used to delineate specific cell types in healthy and diseased brain endothelium [259]–[261]. These single-cell datasets could be used to identify specific cell types in dysregulated BBB to target therapeutically. Additionally, the dysregulated BBB in diseased states is highly dynamic and yields changes in permeability that can be exploited by nanomaterials. In injuries such as TBI and stroke, there is a dramatic physical damage to the BBB that can serve as an opportunity for passive accumulation of nanomaterials, and engineering efforts have focused on nanomaterial physicochemical properties that can be optimized to maximize accumulation. Extending this approach to diseases that have limited physical disruption of BBB, avenues of research are investigating how to control the temporospatial and reversible physical disruption of the BBB while minimizing the potential impact on brain homeostasis using external energy. For example, focused ultrasound is increasingly being investigated in human patients to increase the delivery of therapeutic compounds in cancers and neurodegenerative diseases. There is increasing appreciation of the role neuroinflammation plays across every neurological disorder, including an upregulation of surface receptors which may serve as new therapeutic targets. There have also been multiple strategies to mimic the natural homing of immune cells with nanomaterials for improved BBB transport – for example through binding of specific immune cell receptors upregulated on the BBB, camouflaging nanomaterials with immune cell membranes, or hitchhiking on or inside living immune cells. Future nanomedicines for the brain are likely to emerge from a convergence of new biological insights and finely tuned engineered materials.

Abbreviations

BBB

blood-brain barrier

ECs

endothelial cells

ECS

extracellular space

CNS

central nervous system

NVU

neurovascular unit

PEG

poly(ethylene) glycol

NPs

nanoparticles

P-gp

P-glycoprotein

mRNA

messenger RNA

TJ

tight junction

AJ

adherens junction

RMT

receptor-mediate transcytosis

ABC transporter

ATP-binding cassette transporter

TfR

transferrin receptor

GBM

glioblastoma

TBI

traumatic brain injury

FUS

focused ultrasound

RBC

red blood cell