Advanced Drug Delivery Reviews

Advanced Drug Delivery Reviews Volume 61, Issue 2, 27 February 2009, Pages 158-171

doi:10.1016/j.addr.2008.11.002 | How to Cite or Link Using DOI Copyright © 2009 Elsevier B.V. All rights reserved. Permissions & Reprints

Mucus-penetrating nanoparticles for drug and gene delivery to mucosal tissues
Samuel K. Laia, b, , , Ying-Ying Wangc and Justin Hanesa, b, c, d, , aDepartment of Chemical & Biomolecular Engineering (JH Primary Appointment), Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218, USA bInstitute for NanoBioTechnology, Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218, USA cDepartment of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA dDepartment of Oncology, Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA Received 14 November 2008; accepted 21 November 2008. Available online 13 December 2008.

Abstract

Mucus is a viscoelastic and adhesive gel that protects the lung airways, gastrointestinal (GI) tract, vagina, eye and other mucosal surfaces. Most foreign particulates, including conventional particlebased drug delivery systems, are efficiently trapped in human mucus layers by steric obstruction and/or adhesion. Trapped particles are typically removed from the mucosal tissue within seconds to a few hours depending on anatomical location, thereby strongly limiting the duration of sustained drug delivery locally. A number of debilitating diseases could be treated more effectively and with fewer side effects if drugs and genes could be more efficiently delivered to the underlying mucosal tissues in a controlled manner. This review first describes the tenacious mucus barrier properties that have precluded the efficient penetration of therapeutic particles. It then reviews the design and development of new mucus-penetrating particles that may avoid rapid mucus clearance mechanisms, and thereby provide targeted or sustained drug delivery for localized therapies in mucosal tissues. Keywords: Mucus; Mucus barrier properties; Mucus-penetrating particles; Therapeutic particles Article Outline 1. Introduction 2. Mucus as a critical barrier to nanoparticle therapies 2.1. Composition of mucus, mucus thickness, and mucus clearance times 2.2. Mechanism of particulate transport in mucus: Channeling through low viscosity pores 2.3. Mucus as a sticky net 3. Conventional nanoparticles are trapped in mucus 3.1. Typical fate of nanoparticles in mucus-covered tissues and conventional strategies for enhancing residence time 3.2. Ex vivo studies of conventional particle transport in mucus 3.3. In vivo studies of conventional particles in mucus 3.4. Shortcomings of mucoadhesive nanoparticles 4. Lessons from nature: Transport of viruses in mucus 5.

Engineering particles to cross mucus barriers 5.1. Role of surface chemistry 5.2. Coating particles with low molecular weight poly(ethylene glycol) 5.3. Implications of large mucus-penetrating particles 5.4. Disrupting the mucus barrier to enhance particle transport 6. Conclusions Acknowledgements References 1. Introduction Delivery technologies that improve drug pharmacokinetics and facilitate localized delivery to target tissues strongly improve the efficacy of various therapies [1] and [2]. In particular, an increasing number of nanoparticle -based drug delivery systems have been approved for human use or are currently being evaluated in clinical trials [3] and [4]. Nanoparticle systems can be engineered to possess a number of desirable features for therapy, including: (i) sustained and controlled release of drugs locally [5] and [6], (ii) deep tissue penetration due to their nano-metric size [7], [8] and [9], (iii) cellular uptake and sub-cellular trafficking [10] and [11], and (iv) protection of cargo therapeutics at both the extracellular and intracellular levels [12] and [13]. The development of controlled release systems for drug and gene delivery to mucosal surfaces, such as those of the lung airways, GI tract, female reproductive tract, nose and eye, is of widespread interest. However, the viscous, elastic and sticky mucus layer that lines all mucosal tissues has evolved to protect the body by rapidly trapping and removing foreign particles and hydrophobic molecules [14] and [15]. The limited permeability of drug delivery particles and many hydrophobic drugs through the mucus barrier leads to their rapid clearance from the delivery site, often precluding effective drug therapies at non-toxic dosages. In order to avoid rapid mucus clearance and/or reach the underlying epithelia, nanoparticles must quickly traverse at least the outermost layers of the mucus barrier (which are cleared most rapidly) [15] and [16]. Until recently, nanoparticles were thought incapable of efficiently penetrating mucus layers [17]. The need for new strategies to increase particle transport rates is underscored by: (i) observations of the immobilization of conventional nanoparticles in mucus ex vivo [18] and [19]; (ii) the very slow transport rate of herpes simplex virus (diameter 180 nm) in mucus ex vivo [19]; and (iii) animal studies showing that mucus immobilizes a range of particle types [20], [21], [22] and [23].

In order to penetrate mucus, synthetic nanoparticles must avoid adhesion to mucin fibers and be small enough to avoid significant steric inhibition by the dense fiber mesh. Recently, we demonstrated that nanoparticles as large as 500 nm, if sufficiently coated with a muco-inert polymer, can rapidly traverse physiological human mucus with diffusivities as high as only 4-fold reduced compared to their rates in pure water [17]. This finding suggests that it is possible to engineer nanoparticles that overcome the mucus barrier. Combined with a suitably tailored drug release profile, these mucuspenetrating particle (MPP) systems offer the prospect of sustained drug delivery at mucosal surfaces and, thus, provide hope for improved efficacy and reduced side effects for a wide range of therapeutics. The generation of MPP loaded with nucleic acids may also greatly enhance the efficacy of this critical family of therapeutic agents [24] and [25]. This article reviews the mechanisms by which mucus hinders or prevents particle penetration, including a discussion of previous work on nanoparticle transport in mucus both ex vivo and in vivo. Subsequently, the recent development of mucus penetrating nanoparticles is described. 2. Mucus as a critical barrier to nanoparticle therapies 2.1. Composition of mucus, mucus thickness, and mucus clearance times Mucus is a viscoelastic gel layer that protects tissues that would otherwise be exposed to the external environment. Mucus is composed primarily of crosslinked and entangled mucin fibers secreted by goblet cells and submucosal glands [26], [27] and [28]. Mucins are large molecules, typically 0.5 40 MDa in size [15] and [16] formed by the linking of numerous mucin monomers, each about 0.3 0.5 MDa [29] and [30], and are coated with a complex and highly diverse array of proteoglycans [15] and [31]. At least twenty mucin glycoproteins have been assigned to the MUC gene family [31] and [32], with several mucin types expressed at each mucosal surface [32], [33], [34] and [35]. Mucins can be generally separated into two families: cell-associated mucins ranging between 100 and 500 nm in length that contain a transmembrane domain, and secreted mucins that are up to several microns long [31], [36] and [37]. Individual mucin fibers are roughly 3 10 nm in diameter, as determined by biochemical and electron microscopy studies [38] and [39]. They are highly flexible molecules, with a persistence length of roughly 15 nm [15]. With the exception of specific disease states, such as chronic obstruction pulmonary disease (COPD) and cystic fibrosis (CF), the mucin content ranges between 2 and 5% by weight for cervical, nasal, and lung mucus [40], [41], [42], [43] and [44], with glycosylated oligosaccharides representing 40 80% of the mucin mass [15]. The water content in most mucus types (i.e., lung, gastric, cervicovaginal) commonly falls within the 90 98% range [35], [43], [44], [45] and [46]. In addition to mucins, mucus gels are loaded with cells, bacteria, lipids, salts, proteins, macromolecules, and cellular debris [15] and [17]. The various components work together to form a nanoscopically heterogeneous environment for particle transport [47], with a shear-dependent bulk viscosity that is typically 1,000 10,000 times higher than the viscosity of water at low shear rates [15], [17] and [47]. Mucus viscoelasticity is tightly regulated in healthy subjects by controlling the mucin to water secretion ratio [48], as well as by varying lipid [49], protein [50], and ion content [51]. Readers are referred to a review on mucus rheology in this issue for further details [47].

Mucus pH can vary greatly depending on the mucosal surface, with highly acidic environments capable of aggregating mucin fibers and greatly increasing the mucus viscoelasticity [52] and [53]. Lung and nasal mucus are in general pH neutral [54] and [55], and eye mucus is slightly basic with pH 7.8 [56]. In contrast, gastric mucus is exposed to a wide range of pH: a large pH gradient exists within the same mucus layer cross-section, with pH rising from the luminal pH of 1 2 to 7 at the epithelial surface [57]. Vaginal secretions typically exhibit a pH in the range of 3.5 to 4.5 due to acidification from lactic acid produced by lactobacilli under anaerobic conditions [15], [58] and [59]. Beyond biochemical differences, the thickness of the mucus blanket also varies for different mucosal surfaces. Estimates of human mucus layer thickness in different tissues are given in Table 1. In the lung, the thickness of the luminal mucus gel is generally thought to be 5 10 µm [60] and [61]. The underlying, less viscoelastic sol layer, also known as the periciliary liquid (PCL), is occupied by cilia and constitutes an additional 5 10 µm in thickness [60], [62] and [63]. However, more recent studies based on confocal fluorescence microscopy suggest that airway mucus may range in thickness from 5 to 55 µm [64], [65], [66] and [67]. Under particular disease conditions, such as CF, the average thickness of airway mucus can markedly increase, likely reflecting accumulation of the mucus resulting from impaired mucociliary clearance [67]. The nasal tract, which has a mucus layer of limited thickness, is readily accessible and considered highly permeable compared to other mucosal surfaces [1]. In the human GI tract, the mucus layer is thickest in the stomach and the colon, but exhibits significant variation [68], [69], [70], [71], [72], [73] and [74]. The relative thickness of the luminal mucus gel and the cell-associated mucus layer in the human GI tract remains under-characterized. The thickness of GI mucus may also vary greatly depending on digestive activity. For instance, dietary fiber deficiency can lead to a significantly lower mucus thickness in rats [75]. In the female reproductive tract, the mucus layer is expected to be at least tens of microns thick in humans, although precise estimates are unavailable. In the eye, the precorneal tear film covering the conjunctiva was previously thought to be 5 µm [76] and [77], but more recently suggested to be as thick as 30 40 µm [56] and [78]. Table 1. Thickness of various types of normal human mucus Type of mucus Aver Reported thicknesses ( m) age thic knes s ( m )

Respiratory Airway 15 7 [64] and [66]; 30 [65] 55 ± 5 [67]

Bronchial 55

Type of mucus

Aver Reported thicknesses ( m) age thic knes s ( m )

Gastrointestinala Gastric Ileal Cecal Colonicb 170 10 37 100 144 ± 52 [208]; 192 ± 7 [209] 10 [74] 36.7 ± 7.2 [72] 39.1 ± 9.9 (A), 57.5 ± 14.5 (T), and 69.6 ± 32.1 (D) [72]; 79 ± 40 [74]; 100 150 [68]; 107 ± 48 (A T) and 134 ± 68 (T D) [73]; 110 160 [69] 101.5 ± 80.3 [72]; 155 ± 54 [73]

Rectal Ocular

125

Mucus 0.035 0.02 0.05 [77] layer (only) Tear film 5 40 Full-size table a Gastrointestinal mucus generally exists as two layers, a basal unstirred or firmly adherent layer and a luminal stirred or sloppy layer [210], [211] and [212]; however, most of the cited references report only one value for mucus layer thickness. Matsuo et al. [72] report the thickness of both an inner layer and outer layer of mucus; in the cecum, the inner layer was measured to be 5.6 +/ 0.2 µm; in the colon, 4.7 +/ 1.4 µm for the ascending colon, 7.0 +/ 3.7 µm for the transverse colon, and 7.6 +/ 3.4 µm for the descending colon; and in the rectum, 12.7 +/ 6.0 µm. b A = ascending colon; T = transverse colon; D = descending colon. 3 [77]; 6 7 [76] 34 45 [78]

View Within Article

Mucus is continuously secreted, then shed and discarded or digested and recycled. Its lifetime is short, often measured in minutes to hours (Table 2). The fastest turnover is typically observed at surfaces with the thinnest mucus layer, perhaps as an evolutionary response to improve the protective features of the mucosa. In the nasal tract, the ciliary motion transports mucus from the anterior towards the posterior region of the nasal cavity, moving entrapped particles and potentially toxic materials toward the nasopharynx. The mucus flow rate is about 5 mm per min, and the mucus layer is renewed approximately every 20 min [32], [79] and [80]. Likewise, the luminal gel layer of respiratory tract mucus is replaced every 10 to 20 min, leading to efficient clearance of inhaled particulates [32]. In contrast, the sol phase of respiratory mucus is thought to be cleared much less rapidly than the more solid-like luminal gel layer [55], [56], [57] and [58]. Continuous secretion and clearance by peristaltic forces leads to quick turnover times, on the order of 4 6 h, in the rat gastrointestinal tract [12], [81] and [82] and likely similar values in humans. In the vagina, mucus is cleared (expelled through the introitus) by intra-abdominal pressure as well as abdominal motions, which squeeze the walls of the vagina together [83] and [84]. The typical clearance time in the human CV tract remains unclear, but is likely on the order of a few hours. In the eye, the tear turnover rate under normal physiological conditions is in the range of 13 20% per minute, leading to nearly complete clearance of most molecules and particulates from the eye within minutes [56].

Table 2. Clearance time and rate for various types of normal human mucus Type of mucus Clearan ce time (min) Clearance rate (mm/min) Ref.

Respiratory General Maxillary sinus Nasal 10 20 20 30 a a [32] [32]

8.8 a 20

a 5 11 5 4.1 ± 1.9

[213] [80] [79], [214] and [215] [216]

Tracheal

a

Type of mucus

Clearan ce time (min) a

Clearance rate (mm/min)

Ref.

4.7 (range: 3.5 6) [216], [217], [218], [219], [220], [221] and [222] 15.5 (range: 4.5 30)b 2.4 ± 0.5 1 [223], [224], [225], [226], [227], [228], [229] and [230] [217] [231]

a

Bronchial Small airways Ocular Full-size table

a a

5 7.7c

a

[56]

a Not available. b Clearance rates were measured using a bronchoscopic or roentgenographic method, which generally results in higher measured rates of clearance, most likely due to the invasiveness of the procedure [229] and [230]. c Clearance time was estimated based on the clearance of various molecules from the ocular surface.

View Within Article

The understanding of mucus layer thickness and clearance times at various mucosal surfaces is important to the development of particles designed to overcome mucosal clearance mechanisms, since they must penetrate mucus at rates markedly faster than mucus renewal and clearance in order to overcome the barrier. 2.2. Mechanism of particulate transport in mucus: Channeling through low viscosity pores The bulk viscosity of healthy human mucus is typically 1000 10,000 times higher than the viscosity of water (at low shear rates). The Stokes Einstein equation for diffusion in homogeneous fluids predicts that neither viruses nor many hydrophilic macromolecules would be capable of penetrating a fluid layer of this viscosity. Indeed, the diffusion of many macromolecules over extended distances in mucus, or at rates faster than mucociliary clearance, was long assumed to be an improbable event [85]. That view began to change with increasing evidence that a number of macromolecules rapidly diffuse through mucus. Important work by the Saltzman and Cone groups showed that the diffusion of various proteins, including antibodies, in human midcycle cervical mucus was largely unhindered

relative to their rates in water [86]. The study also suggested that viruses as large as 30 nm should diffuse nearly unhindered in cervical mucus, which was subsequently proven by Olmsted et al. [19]. In particular, they found that Norwalk (38 nm) and human papilloma virus (HPV; 55 nm) penetrated mucus at rates roughly equivalent to those in water. These results firmly establish that particulates are capable of diffusing through low viscosity pores within the highly elastic mucin fiber matrix. In general, if the viscosity of the fluid that fills the pores in mucus is equivalent to that of water, the diffusion rates of particulates significantly smaller than the average mucus mesh pore size, assuming they do not adhere to mucus, are expected to be similar to their rates in water. Olmsted et al. also showed that the diffusion of herpes simplex virus (HSV), a virus with a diameter of 180 nm, was nearly completely inhibited [19]. Even the very small fraction of virus that was mobile moved at rates 100- to 1000-fold slower than their expected rates in water. The slow transport of HSV agrees well with an obstruction scaling model [87] and [88] that assumes the average pore size to be 100 nm (Fig. 1). This pore size estimate agreed well with the size of viruses capable of efficient infection of mucosal tissues, including polio (28 nm), hepatitis B (43 nm), adenoviruses (60 90 nm), and rotavirus (75 nm) [15]. Based on these lines of evidence, the development of mucus-penetrating particles appeared to be limited to the dimensions of small viruses, raising significant formulation and practical concerns.

Full-size image (36K) High-quality image (122K) Fig. 1. Transmission electron micrograph of human cervicovaginal mucus. The interfiber spacing ranges from 10 nm to 200 nm. Individual fibers can be observed with fiber diameters 10 nm. Scale bar = 200 nm. Figure obtained from [19].

View Within Article

2.3. Mucus as a sticky net Airborne particles typically adhere to respiratory mucus before reaching the alveoli. Filtered particles are removed from the airways by mucociliary clearance and become sterilized by gastric acid [14] and

[89]. Similarly, intestinal mucus prevents the transport of toxic contents from the lumen of the GI tract to the underlying interstitial and vascular fluids [15]. The primary mechanism by which mucus gels efficiently trap foreign particulates is the formation of polyvalent adhesive interactions [15] and [90]. In addition to the negative charges imparted by the presence of carboxyl or sulfate groups on the mucin proteoglycans, there exist periodic hydrophobic naked globular regions along mucin strands, stabilized by multiple internal disulfide bonds [91]. The high density of hydrophobic domains, coupled with the flexible nature of mucin fibers, allows efficient formation of multiple low-affinity adhesive interactions with hydrophobic regions on the surfaces of foreign particulates. Although each low-affinity hydrophobic interaction can be readily disrupted by thermal energy, a large number of low-affinity adhesive interactions with the mucus mesh effectively immobilizes particles with permanent high viscidity [15]. The formation of polyvalent low-affinity interactions is critical for the removal of foreign particles smaller than the average mucus mesh pore size. An excellent example of this phenomenon is the strong immobilization of virus-sized carboxyl-modified polystyrene nanoparticles, including those as small as 59 nm, in human cervical mucus [18] and [19]. Despite possessing a negative surface charge that might be expected to repel negatively charged mucins, polystyrene nanoparticles form multiple hydrophobic adhesive interactions with mucin fibers that are strong enough to bundle mucin strands into thick cables (Fig. 2). Similar immobilization of albumin- or casein-coated particles underscores the efficiency of the hydrophobic adhesive interactions [15]. Hydrophobic interactions also represent an important mechanism by which mucus limits the transport of large entities such as bacteria. Potentially harmful pathogens, including Escherichia coli, Pseudomonas aeruginosa, and Yersinia enterocolitica, are all immobilized by mucus via hydrophobic interactions [92], [93] and [94].

Fig. 2. Fluorescently labeled 59 nm polystyrene particles formed thick cables with mucin fibers in human cervicovaginal mucus. (A) Fluorescent image. (B) Phase image. Scale bar = 500 µm. Figure obtained from [19].

View Within Article

Polyvalent adhesive interactions with mucus can be achieved via electrostatic interactions as well. Since the glycosylated regions of mucin fibers are densely coated with negative charges, mucus can also bind positively charged particles with high avidity. The dual capacity to form polyvalent adhesive interactions via both hydrophobic and anionic forces represents a particularly challenging problem for polymeric nanoparticles designed to deliver drugs and genes, since many commonly used biomaterials are either hydrophobic, such as poly(lactide co-glycolide) and polyanhydrides [95] and [96], or cationic, such as polyethylenimine, chitosan and polylysine [59], [97] and [98]. However, despite the great avidity of these polyvalent adhesive interactions, the capacity for mucus to immobilize foreign particles may become saturated. This was demonstrated by the work of Durrer and coworkers, who found that small particles can be adsorbed to mucin fibers only until the internal area available for adsorption is saturated [99], [100] and [101].

The ubiquitous presence of hydrophobic domains in the mucus mesh also helps to explain why many therapeutic molecules diffuse more slowly through mucus than other molecules of similar or larger size. The diffusion of hydrophobic molecules is often significantly hindered in mucus compared to in water [102], [103] and [104]. Diffusivity is typically reduced as the nonpolar/polar partition coefficient increases, an expected consequence of adhesive interactions between hydrophobic molecules and the hydrophobic domains of mucus. Several investigators have also shown that mucus significantly decreases the diffusion of a number of both small and large molecules, including tertiary amines [105], bovine serum albumin [106], lysozyme [106], and others [104], [106], [107] and [108], presumably due to hydrophobic interactions. Nonetheless, some hydrophobic molecules are capable of diffusing through mucus gels, perhaps because small molecules are not capable of forming polyvalent adhesive bonds with the mucus gel. 3. Conventional nanoparticles are trapped in mucus 3.1. Typical fate of nanoparticles in mucus -covered tissues and conventional strategies for enhancing residence time When administered to various mucosal tissues, conventional nanoparticles are likely to be trapped by mucus via steric or adhesive forces and rapidly eliminated via mucociliary clearance. In the GI tract, for example, nanoparticles delivered orally may undergo: (1) association with chyme, causing direct transit through the GI tract and fecal elimination; (2) adhesion to mucin fibers, followed by mucus clearance and fecal elimination; or (3) transport across the mucus mesh for possible entry to the underlying epithelia or prolonged residence time in the unstirred layer of mucus gel [81] and [109]. Typically, the majority of administered particles do not adhere or transport through the GI mucus layer, but rather undergo direct transit through the GI tract [81] and [110]. The consequent short transit time in the GI tract can be inadequate for particles to release a significant fraction of encapsulated drugs, thereby precluding in many cases the realization of a high local drug concentration over extended periods of time, hence leading to low bioavailability and poor efficacy. To overcome the short transit time, research has largely centered on minimizing the fraction of therapeutics undergoing direct transit and fecal elimination by improving their association to mucus. This phenomenon, known as mucoadhesion, is widely defined as the ability of polymeric systems to adhere to the mucus layer [81], [109] and [111]. Mucoadhesion slows the particle transit time through the GI tract to the time scale of mucus renewal, thereby enhancing drug absorption. The design of mucoadhesive drug delivery systems is currently the predominant approach to improving mucosal delivery of therapeutics. 3.2. Ex vivo studies of conventional particle transport in mucus That mucus slows the diffusion of macromolecules has long been inferred using indirect techniques, or by measuring rates of permeation through a thin layer of mucus in a diffusion chamber [104] and [108]. The same experimental approach has been used extensively to study the diffusion of polymeric particles in mucus. Among the first to quantitatively measure the 1D diffusivity of particles in mucus, Norris and Sinko studied the diffusion of variously sized polystyrene particles in a reconstituted porcine

gastric mucin gel, sandwiched between the two chambers of a Transwell Snapwell diffusion chamber (Fig. 3), and observed a sharp decrease in translocation permeability when particle sizes approached 300 nm [90]. Sanders et al. reported that an extremely low percentage of carboxylated polystyrene nanospheres moved through a 220 µm-thick CF sputum layer after 150 min, with the largest nanospheres studied (560 nm) almost completely blocked by the sputum [112]. Interestingly, nanospheres diffused more efficiently through more viscoelastic sputum samples, which the authors attributed to a more macroporous structure as viscoelasticity increased. More recently, a similar diffusion-chamber method was used to investigate actin's role in limiting the degree to which common mucolytics enhance the diffusion of particles in mucus [113]. Although the diffusion-chamber method is conceptually straightforward, it may be sensitive to parameters that are difficult to control, such as the precise thickness of the mucus layer, the uniformity of the mucus distribution across the face of the filters, blockage of the filter pores by mucus, and alterations in mucus properties during preparation [86]. As an example of the latter, a significant amount of mucin was found to have diffused through the membrane when the pore size of the diffusion chamber filter was greater than 5 µm, thereby reducing the barrier properties of mucus [90].

Fig. 3. Schematic drawing of a Transwell Snapwell diffusion chamber (A). The donor and acceptor chambers are indicated by I and II, respectively. Mucus or mucin gel is sandwiched in between polycarbonate filters attached to the Snapwell ring (E) located between the two chambers. Figure obtained from [112].

View Within Article

To circumvent potential artifacts stemming from diffusion chamber studies, a number of studies measured the dynamic transport of nanoparticles using fluorescent microscopy techniques, including fluorescence recovery after photobleaching (FRAP), and more recently, high resolution multiple particle

tracking (MPT). For example, pioneering work on protein and virus diffusion in human midcycle cervical mucus in two laboratories (Saltzman and Cone) relied on FRAP to elucidate rapid local diffusion in the low viscosity pores present in the mucus mesh [19] and [86]. More recently, Shen et al. applied FRAP to investigate the diffusion of plasmid DNA, with and without association to liposomes, in reconstituted bovine cervical mucus [114]. They showed that bovine mucus hinders the diffusion of linear DNA, especially when the plasmid size exceeds 5 kb, compared to supercoiled DNA. Fluorescence recovery in 3D, via standard confocal scanning laser microscopes, was developed to test the diffusion of macromolecules in CF sputum by Braeckmans et al. in 2003 [18]. Although FRAP has contributed valuable insight into the diffusion of macromolecules and particulates in mucus, FRAP provides only ensemble-averaged diffusion rates and cannot be used to quantify the transport rates of individual particles. A lack of information at the individual particle level may limit insight into the complex phenomena that control nanoparticle transport in the heterogeneous mucus network, where the transport rates of identical particles can vary by several orders of magnitude in the same sample [115]. To this end, Hanes and coworkers have pioneered the use of MPT for drug and gene delivery applications [10], [115], [116] and [117], including the transport of nanoparticles in mucus [17], [118] and [119]. MPT allows for unparalleled insight into the fractions of particles that are adherent to mucus at any given time, and yields quantitative measures of the extent of particle hindrance to free diffusion at the individual particle or ensemble-average level, important feedback in the development of non-mucoadhesive particle surfaces. Coupled with Monte Carlo simulations of Brownian behavior, MPT can also be used to determine in a quantitative and high-throughput manner the transport mechanism (immobile, hindered-diffusion, diffusive, or active/convective transport) governing particle dynamics in mucus and in other complex biological fluids [17], [117] and [120]. Dawson et al. first applied MPT to measure the transport rates of nanoparticles in mucus. Using a set up that achieved 5 nm and 33 ms spatiotemporal resolution, they studied the movements of carboxylate- and amine-modified polystyrene nanoparticles sized 100 500 nm in fresh human sputum obtained from patients with CF [118]. The diffusion of nanoparticles was more than an order of magnitude faster than the diffusivity predicted from the macrorheology of the mucus, but at least 300-fold slower than the same particles in water. Neutrally charged hydrophobic particles 200 nm in diameter appeared to undergo more rapid transport in CF sputum than anionic but otherwise similar particles, suggesting surface charge may be an important parameter in governing transport rates of nanoparticles in mucus. In a separate work, Dawson et al. used MPT to study the diffusion of biodegradable hydrophobic PLGA nanoparticles surface coated with anionic DNA via the cationic surfactant dimethyl dioctadecyl ammonium bromide (DDAB). Coating with hydrophilic DNA improved the average nanoparticle transport rates by 10-fold in reconstituted pig gastric mucus compared to slightly smaller, hydrophobic polystyrene particles (Fig. 4) [119]. In addition, a significant fraction of PLGA-DDAB/DNA particles appeared to undergo unrestricted diffusive transport, supported indirectly by the finding that a considerably higher percentage of PLGA-DDAB/DNA nanoparticles (8.5%) exhibited a MSD > 5-fold higher than the ensemble average MSD, compared to corresponding polystyrene particles. More recently, Lai et al. applied MPT to quantify the transport of various sized particles in fresh undiluted human cervicovaginal mucus [17].

Full-size image (29K) High-quality image (109K) Fig. 4. Ensemble-averaged mean square displacement (MSD) as a function of time scale for PLGADDAB/DNA and COOH-modified polystyrene nanoparticles in reconstituted pig gastric mucus. Figure obtained from [119].

View Within Article

3.3. In vivo studies of conventional particles in mucus The fate of particulates in mucosal tissues of living systems has been studied for over a century, with much of the early work focusing on particle uptake into the intestinal epithelia following oral administration. As early as 1907, Basset and Carné suggested that the intestinal epithelium is impermeable to particles [121]. The lack of particle entry into mucosal epithelia has long been attributed to inefficient penetration through the mucus layer. Florey noted that intestinal mucus in cats surrounds particles of India Ink so that they do not come in direct contact with epithelial cells [122]. Arbos et al. observed that poly(methylvinylether-co-maleic anhydride) nanoparticles were not able to reach the enterocytes lining the small intestine and colon in rats and were largely localized within the mucus layer [20]. Gruber concluded that, irrespective of the size, density, or composition of particles, GI mucus in dogs formed them into mucus -covered slugs [21]; such particle agglomeration is more recently highlighted by Tirosh and Rubinstein in their studies of particles in rat intestines (Fig. 5) [123]. Lamprecht et al., in an investigation of particles targeting inflamed colonic mucosa in rats, noted that a large number of the particles were associated with the thick mucus layer rather than attached to or internalized by intestinal cells [22]. O'Mullane et al. stated that the transport of intact carriers across the gastro-intestinal tract is restricted to exceptional and unusual circumstances [23]. Even the

unusually thin mucus secreted in the tear film wraps particles in a mucus coat, as can be noticed on removing an irritating particle from the eyelid [15].

Full-size image (34K) High-quality image (109K) Fig. 5. Agglomerated non-adhesive particles coated with a mucus plug collected immediately following discharge from the proximal jejunum of the perfused rat (magnification: × 60). Figure obtained from [123].

View Within Article

Many researchers have sought to enhance the mucoadhesion of particles in order to improve their retention at mucosal surfaces. To maximize association with mucus, a variety of mucoadhesive drug delivery systems have been engineered, driven by various interaction forces between mucus and nanoparticles, including hydrogen bonding, van der Waals interactions, polymer chain interpenetration, hydrophobic forces, and electrostatic/ionic interactions [109] and [111]. Electrostatic interaction is one of the most exploited forms of mucoadhesion, as exemplified by chitosan, a cationic polymer obtained from deacetylation of chitin, for a variety of oral and nasal drug delivery applications [124] and [125]. Jubeh et al. similarly concluded that cationic liposomes adhered to healthy colonic rat mucosa at rates 3fold greater than neutral or anionic liposomes [126]. Particles synthesized from common biomaterials, such as poly(ethylene glycol) [127], [128] and [129], polycarbophil and carbopol (derivatives of poly(acrylic acid)) [111] and [130], polymethacrylates [131] and [132], and poly(sebacic acid) [133] and [134] may achieve mucoadhesion via hydrogen bonding, polymer entanglements with mucins, hydrophobic interactions, or a combination of these mechanisms. The association of the aforementioned biomaterials with mucus is typically considered non-specific, as they lack biological molecules that afford recognition and binding to specific target chemical structures on mucus or the surfaces of mucosal epithelia [81] and [109]. Alternatively, mucoadhesion can also be achieved via ligands that recognize particular mucin glycoproteins. For example, tomato lectins,

which specifically recognize and bind to N-acetyl glucosamine-containing complexes on cell surfaces, were initially tested for specific interactions with intestinal tissues [111]. However, there was little improvement in intestinal transit time of tomato lectin compared to other mucoadhesive systems [135], an observation later justified by the strong association of lectins to the mucus gel instead of the epithelia [136], [137] and [138]. Lectins with different sugar specificities have also been investigated, including those derived from Canavalia ensiformis [139] and Lotus tetragonolobus [140]. Readers are referred to excellent reviews for further information on non-specific and specific mucoadhesive materials [109] and [111]. Mucoadhesive particles usually improve the pharmacokinetics of therapeutics in the GI tract compared to free drug alone. For example, mucoadhesive polyanhydride copolymers of fumaric and sebacic acids were used to encapsulate insulin and dicumarol (an anticoagulant drug with poor water solubility and erratic intestinal adsorption) and, in both cases, improved the bioavailability of both loaded drugs compared to their native form [133] and [134]. Chitosan-based or chitosan-coated nanoparticles represent another common group of mucoadhesive systems [124], [125] and [141], and several groups have studied nanoparticles made of these polymers to deliver proteins [142] and [143], peptides [144], and DNA [145] to mucosal tissues. In addition to prolonged residence times, improved oral bioavailability of drugs observed in vivo may be partially attributed to improved drug stability in encapsulated form in the GI tract, which is rich in proteolytic enzymes such as pepsin, trypsin, and chymotrypsin [81]. For example, the stability of insulin and calcitonin exposed to protease degradation is significantly improved when loaded in polymeric nanoparticles [12], [146], [147] and [148]. A range of reports have suggested that nanoparticles are capable of entry into intestinal epithelia via M cells on the domes of the Peyer's patches in small animals. Not surprisingly, the Peyer's patches are the intestinal surfaces least protected by mucus and most exposed to chyme [15]. Since no mucus is secreted in the region surrounding these cells, which protrude relatively unprotected into the lumen, the mucus barrier is minimal [149]. Indeed, M cells are positioned as sensory outposts for cellular immune functions, transcytosing particles that impinge on their surface into the interior of the patch. The dominant opinion is that particulate uptake in GI tract of mammals is principally via the M cells of Peyer's patches, and that uptake by enterocytes plays a minor role [81] and [150]. Polystyrene nanopsheres and microspheres were found in the Peyer's patches of the rat colon 12 h following oral administration [151]. Similarly, Jani et al. found microspheres up to a diameter of 3 µm in Peyer's patches and in the gut associated lymphoid tissue 12 h after oral administration [152]. An ex vivo model using a diffusion chamber with pig intestinal tissue show that radiolabeled poly(isobutyl cyanoacrylate) nanoparticles (211 nm) cross pig Peyer's patches, whereas tissue deprived of Peyer's patches was impermeable to particles [153]. The extent of nanoparticle uptake in rabbits appeared at least an order of magnitude greater than that in mice, a finding attributed to the significantly greater abundance of M cells in rabbit Peyer's patches [150] and [154]. Although the above reports widely support the notion that M cells account for the bulk of particles crossing the intestinal epithelium, M cells occupy a relatively small portion of the total surface area of the GI tract. The potential of drug delivery to the systemic circulation via M cells instead of enterocytes has thus remained debatable. Furthermore, the prospects for using Peyer's patches as a

means of delivering peptides and proteins has been said to be severely compromised by: (a) the limited efficiency and capacity of the absorption pathway, (b) the time to onset of pharmacological response, due to both the kinetics of processing the particles and the flow rate of the lymph; and (c) the potential loss of drug to local lymphocytes and macrophages [155]. A number of studies have suggested that uptake of nanoparticles may occur through enterocytes as well as M cells [156], [157] and [158]; however, as recently pointed out, no satisfying explanation has been advanced to elucidate conflicting reports [159]. A direct consequence of greater uptake into intestinal Peyer's patches is the application of nanoparticles to enhance oral immunization [160], because the low intestinal permeability and high presystemic clearance of vaccines from the gastrointestinal tract requires the use of prohibitively large and repetitive doses to elicit even a modest immune response [161]. In general, antigens incorporated into particles are more effective for oral immunization than soluble antigens [81] and [162]. For example, ovalbumin incorporated into PLGA particles induced potent serum and mucosal immune response, including both IgA and IgG antibodies [163] and [164]. Similar results have been achieved with a variety of toxins, including staphylococcal enterotoxin B toxoid [165], tetanus toxoid [166], as well as DNA vaccines [134] and [145]. For further details, readers are referred to several excellent reviews on the topic of nanoparticle -mediated oral vaccination [81], [156] and [167]. 3.4. Shortcomings of mucoadhesive nanoparticles Although mucoadhesion is a promising approach to increase the bioavailability of drugs delivered via mucosal tissues, important fundamental limitations of this approach exist. Since mucoadhesive systems are bound to the mucus layer through interactions with mucin fibers, the transit time of these systems is determined by the physiological turnover time of the mucus layer [81] and [168]. For oral delivery, considering the intestinal mucin turnover time is between 50 and 270 min [82], mucoadhesive particles are not expected to adhere to mucus for more than 4 5 h. Furthermore, as mucoadhesive systems efficiently adhere to mucus, they are largely incapable of penetrating across the mucus layer and entering the underlying epithelia. Thus, mucoadhesive systems are especially unsuitable for delivery of drug and gene molecules that require intracellular delivery. To overcome these issues, various attempts have been made to engineer particles that adhere specifically to intestinal cells, as exemplified by the conjugation of tomato lectins to nanoparticles. However, these ligand-bound particles appear to have a limited capacity to diffuse through the mucus layer and, instead, undergo premature adsorption to mucus [136], [137] and [138]. Thus, these systems reach the enterocyte surface inefficiently at best, and are instead bound to mucus, transported through the GI tract, and eliminated in the feces in a similar fashion to other mucoadhesive systems. To overcome this problem and achieve longer residence time of particles at mucosal surfaces, the foremost requirement is to engineer particles that can efficiently cross the mucus barrier (Fig. 6).

Full-size image (93K) High-quality image (773K) Fig. 6. Summary schematic illustrating the fate of mucus-penetrating particles (MPP) and conventional mucoadhesive particles (CP) administered to a mucosal surface. MPP readily penetrate the luminal mucus layer (LML) and enter the underlying adherent mucus layer (AML). In contrast, CP are largely immobilized in the LML. Because MPP can enter the AML and thus are in closer proximity to the cells, cells will be exposed to a greater dose of drug released from MPP compared to drug released from CP. As the LML layer is cleared, CP are removed along with the LML whereas MPP in the AML are retained, leading to prolonged residence time for MPP at the mucosal surface. Thus, at long times, there is almost no drug dosing to cells with CP, whereas MPP, because they are retained longer, will continue to release drugs to cells. Since MPP can penetrate both the LML and AML, a fraction may reach and bind to the underlying epithelia and further improve drug delivery. While this schematic reflects the mucosal physiology of the gastrointestinal and cervicovaginal tracts, the same behavior is expected for the respiratory airways. In the respiratory airways, CP are mostly immobilized in the luminal stirred mucus gel layer, whereas MPP penetrate the mucus gel and enter the underlying periciliary layer. Upon mucociliary clearance, a significant fraction of MPP remains in the periciliary layer, resulting in prolonged retention. This schematic does not depict the glycocalyx adjacent to the epithelial surface, which may contribute an additional steric barrier to cellular entry of MPP.

View Within Article

4. Lessons from nature: Transport of viruses in mucus To gain mechanistic insight and rationally engineer particles to cross mucus, we took an alternative approach and looked to nature for guidance. Specifically, we focused on understanding the physicochemical properties (i.e. size and surface chemistry) that govern the rapid transport of specific viruses, which have evolved over thousands of years to infect mucosal tissues. An indirect implication of the highly immobilized nature of virus-sized polystyrene beads in mucus is that viral particles possess a sufficiently large surface area to undergo polyvalent adhesive interactions

with mucus [15]. Nevertheless, many viruses are capable of diffusing in mucus as fast as in water (Fig. 7) [19]. This finding by Cone and coworkers motivated us to attempt to develop mucuspenetrating particles (MPP) by mimicking the essential surface properties of viruses that allow them to avoid mucoadhesion. A hypothesis first proposed by Cone and coworkers noted that many viruses, including poliovirus, Norwalk virus and human papilloma virus (HPV), are densely coated with both positively and negatively charged groups, leading to a densely charged yet net neutral surface [15]. Indeed, the average distance between charge groups on the surface of poliovirus is only 5 Å. This feature is also found in many soluble proteins [169], including antibodies, such as IgG and IgA, which experience minimal hindrance to diffusion by mucus [19] and [86]. We rationalized that the high density of surface charge creates a hydrophilic surface that minimizes hydrophobic entrapment to mucus. Furthermore, an equal density of positive and negative charges may facilitate efficient mucus transport by allowing the viruses to avoid electrostatic adhesive interactions. Some viruses thus appear well designed to penetrate mucus by simultaneously possessing a muco-inert surface to avoid mucoadhesion and a sufficiently small geometry to diffuse through the low viscosity pores in mucus without significant steric obstruction.

Full-size image (27K) High-quality image (99K) Fig. 7. Normalized diffusion coefficients for proteins and viruses in mucus. If a particle diffuses in mucus as fast as it diffuses in saline, Dmuc / Dpbs = 1. The lines drawn on the graph are the ratios predicted by Amsden's obstruction-scaling model using a mucin fiber radius of 3.5 nm and mesh fiber spacing of 100 nm (solid line) or 110 nm (dotted line), which accounts for a 10% increase in the mesh fiber spacing due to 20% dilution of the mucus sample. Figure obtained from [19].

View Within Article

5. Engineering particles to cross mucus barriers 5.1. Role of surface chemistry

Polystyrene nanoparticles as small as 59 nm, covalently modified with carboxyl groups on the surface, were found to be completely immobilized in human cervical mucus [19]. By reducing the concentrations, Dawson et al. found that some particles with surfaces modified with either carboxyl or amine groups were mobile in CF sputum, but the ensemble-averaged transport rates were still slowed at least 300-fold compared to the same particles in water [118]. Norris et al. similarly observed slow transport of nanoparticles surface-modified with carboxyl, sulfate, or amidine groups, although the fastest transport was found with the most hydrophilic (amidine) surface [90]. In contrast, Sakuma et al. studied the in vivo behavior of nanoparticles having surface poly(N-isopropyl acrylamide), poly(N-vinyl acetamide), poly(vinyl amine), or poly(methacrylic acid) chains, and observed increased mucoadhesion for all chains except for poly(N-vinyl acetamide) [170] and [171]. The increased mucoadhesion by the various hydrophilic chains was attributed to the presence of tertiary amines which may interact with anionic components of mucus. Thus, no clear correlation existed between the surface chemistry of synthetic nanoparticles and their transport rates in mucus. 5.2. Coating particles with low molecular weight poly(ethylene glycol) A faithful mimic of muco-inert viruses would involve developing synthetic particles with a similarly high density of both cationic and anionic surface charge groups. However, although techniques have been advanced to generate multiple alternating layers of oppositely charged coatings with thicknesses on the order of angstroms [172] and [173], the engineering of such densely charged yet neutral coatings onto the surfaces of synthetic particles remains exceedingly difficult. Furthermore, even if such chemistry is possible and scalable, densely charged synthetic particles may exhibit immunogenicity similar to viruses. To circumvent the difficult chemistry and immunological concerns, we abstracted the features of the highly charged yet net neutral surfaces of viruses to encompass high hydrophilicity and neutral charge. In particular, instead of a high density of charge groups, we hypothesized an uncharged surface may be as muco-inert as viral capsids, provided the surface is sufficiently hydrophilic and low hydrogen bonding. We thus screened through GRAS (generally regarded as safe) polymers that may satisfy these dual requirements: (i) strong hydrophilicity and (ii) neutral charge. Poly(ethylene glycol) (PEG) is an uncharged hydrophilic polymer routinely used in pharmaceutics to improve systemic circulation and minimize opsonization [4] and [174]. However, it was not obvious, a priori, that PEG may reduce the association of particles with mucus, since PEG was widely used as a mucoadhesive agent [175] and [176]. Some authors have reported that PEG chains may establish adhesive interactions due to their ability to inter-diffuse with the mucus network [128], [175], [176], [177] and [178] and undergo hydrogen bonding [127]. It should be noted that a large fraction of the early work depicting PEG as mucoadhesive was based on PEG coating of hydrogel systems [179], where the characteristic size of the gel is far greater than the average mucus pore spacing. A number of reports over the past decade have further suggested PEG to enhance the mucoadhesive properties of nanoparticles [177], [180], [181] and [182], although the density of the PEG coating was typically under- characterized. In light of these findings, we hypothesized that coating particles with a high density of PEG may reduce particle mucus adhesive interactions, provided: (i) the M.W. of PEG was too low to support adhesion via polymer interpenetration (Fig. 8) and (ii) the PEG density is sufficient to effectively shield the hydrophobic core common to many biodegradable polymers.

Full-size image (71K) High-quality image (640K) Fig. 8. Schematic representation of (A) the interdiffusion of polymers upon contact and (B) the interpenetration between tethered chains on a particle surface and mucin fibers of the mucus gel layer. Figure obtained from [175].

View Within Article

To test our hypothesis, we covalently modified the surfaces of polystyrene nanoparticles with a high density of low M.W. (2 kDa) PEG, and studied their diffusion in fresh undiluted human mucus [17]. Contrary to reports of PEG being mucoadhesive, a dense PEG coating, as characterized by a near-neutral surface charge and negligible protein adsorption, improved the transport of nanoparticles by up to 3 orders of magnitude for particles in the range of 100 500 nm in size (Fig. 9). Large nanoparticles (200 and 500 nm in diameter) densely coated with PEG diffused through mucus with an effective diffusion coefficient (Deff) only 6- and 4-fold lower than that for the same particles in water. The greatly improved nanoparticle transport rates upon PEGylation correlated with a sharp decrease in the fraction of immobile particles for all sizes tested; less than 0.5% of 200 and 500 nm PEGylated particles were immobilized by mucus compared to 32 and 45% for uncoated 200 and 500 nm particles, respectively. To our knowledge, this is the first report to demonstrate that synthetically engineered nanoparticles can rapidly penetrate fresh undiluted human mucus barriers.

Full-size image (57K) High-quality image (467K) Fig. 9. Transport of COOH- and PEG-modified polystyrene particles in fresh, human cervicovaginal mucus (CVM). Ensemble-averaged geometric mean square displacements (<MSD>) as a function of time scale for (A) COOH- and (B) PEG-modified polystyrene particles in human CVM. Comparison of average Deff at a time scale of 1 s in water (W) vs. CVM of subfractions of (C) COOH- and (D) PEGmodified particles, from fastest to slowest. Theoretical Deff for same-sized particles in water is shown as W. The dashed black line at <Deff> = 1 × 10 4 signifies the microscope's resolution. Data represent ensemble average of three experiments, with n 120 particles for each experiment. Figure obtained from [17].

View Within Article

Despite the marked improvement of nanoparticle transport upon PEG modification, a significant fraction of 100 nm PEG-modified particles was found to be immobile or strongly hindered in mucus [17]. One possible explanation for the slower transport of 100 nm (vs. 200 and 500 nm) particles may be inadequate PEGylation of 100 nm particles. To better understand how the density of surface PEG coverage may affect particle transport, and to resolve the paradox between the mucoadhesive properties of PEG widely observed prior to our work and the muco-inert properties of PEG we observed, we recently tested the effect of coating density and PEG M.W. on the transport of PEG-coated particles in mucus [183]. We found that a 40% lower surface coverage of 2 kDa PEG, compared to particles well-coated with 2 kDa PEG, caused a 700-fold decrease in the average transport rate. Particles similarly well-coated with 5 kDa PEG diffused as rapidly through mucus as those with 2 kDa PEG, but a further increase of PEG M.W. to 10 kDa resulted in 1000-fold slower transport. These results suggest that both dense surface coverage and low M.W. PEG are required for PEG-coated particles to rapidly penetrate mucus. Using this design principle, we were able to engineer 100 nm particles that can rapidly penetrate mucus (unpublished observations). By comparing our results to previously published studies of the interaction of PEG-coated particles with mucus, we estimated the range of PEG M.W. and particle surface charge (an indirect measure of PEG surface coverage on non-neutral core particles) that likely provides muco-inert PEG coatings (Fig. 10) [183].

Full-size image (18K) High-quality image (144K) Fig. 10. Phase diagram correlating muco-inert versus mucoadhesive particle behavior to surface charge and PEG MW for various PEG-coated nanoparticles ( 200 500 nm in size) reported in literature. PEG-coated nanoparticles reported to be non-mucoadhesive compared to control (uncoated) particles are indicated by open symbols, and those reported to be mucoadhesive compared to control particles are indicated by filled symbols. The shaded region represents the confirmed range of PEG MW and particle -potential (i.e., PEG surface coverage), and the hatched region an additional predicted range, that provides a muco-inert coating. Letters represent published results from various studies [17], [183], [232], [233], [234] and [235]. Mucoadhesion was not observed based on adhesion to an in vitro mucinsecreting cell line, which is unlikely to produce mucus gels with the complex mesh structure and adhesivity of physiological mucus [234]. Figure obtained from [183].

View Within Article

In addition to improving transport, PEGylation of nanoparticles may also enhance their stability in mucus. Stability is particularly important when particles must diffuse through a thick mucus layer in order to reach underlying cells. Various studies have demonstrated that liposomes are not stable in the GI tract, particularly in the presence of bile salts [184] and [185]. Low gene transfer to lung epithelia by lipoplexes administered via intratracheal or intranasal instillation has been attributed to the low stability and substantial aggregation of liposomes in mucus [186]. Upon PEGylation, Sanders et al. found that mucus components did not interact significantly with PEG-GL67 lipoplexes, and prior exposure to the mucus components did not reduce their gene transfer [187] and [188]. Similarly, in vitro incubation of PEG-coated polylactide nanoparticles in simulated gastric fluid demonstrated 9-fold higher chemical stability than polylactide particles alone [129]. 5.3. Implications of large mucus-penetrating particles

Much to our surprise, we were able to engineer 200 and 500 nm particles to rapidly traverse fresh human cervicovaginal mucus. This directly implies that the average effective mesh spacing in human cervicovaginal secretions must be significantly larger than earlier reports [17] and must include a large number of pores with effective spacings substantially larger than 500 nm. It is unlikely that the rapid

transport of PEGylated particles is due to alterations of the mucus structure, because they do not interact significantly with mucus. Instead, particles likely move in low-viscosity channels or pores within the mucus, as suggested in our earlier work on particle transport in cystic fibrosis mucus [118]. Furthermore, since our results are obtained with fresh, undiluted cervicovaginal mucus that is rheologically and compositionally comparable to mucus found in the airways of the lungs and gastrointestinal tract [15] and [17], it is possible that large PEGylated nanoparticles may also transport quickly in mucus coating other entry sites into the body. Small changes in the mucus hydration may not markedly alter the diffusion rates of nanoparticles from their rates in fresh cervicovaginal mucus, since the interfiber spacing may not strongly depend on hydration [86]. The rapid mucosal transport of large (200 and 500 nm) PEG-modified particles has important implications for the development of therapeutic and imaging applications in vivo. Larger nanoparticles afford substantially higher drug encapsulation as well as reduced aggregation upon freeze drying. In addition, as the size of drug -loaded particles increases, the drug -release kinetics are usually greatly improved as well, allowing sustained release of therapeutics over days and even months along with enhanced therapeutic efficacy [5], [189] and [190]. In contrast, smaller nanoparticles suffer from large burst release typically within hours upon in vitro or in vivo application [191] and [192]. These findings should strongly encourage the commercial development of new nanoparticle -based drug delivery systems for applications at mucosal surfaces. Although large nanoparticles are preferred for improved drug loading and release kinetics, an optimal size likely exists for mucosal applications. Nanoparticles that are too large, even if they are well-coated so as to avoid mucoadhesion, may not diffuse at rates sufficient for overcoming mucociliary clearance due to the elevated friction forces predicted by the Stokes Einstein equation. Since mucus layer thickness and mucus clearance time differ among the mucosal tissues, different nanoparticle diffusion rates may be required in order to overcome different mucus barriers. Furthermore, for cellular uptake into underlying epithelia, an increase in particle diameter is expected to reduce the rate of endocytosis [10] and [193]. The size of MPP must be optimized with all of these considerations in mind, and will likely vary based on the requirements of different mucosal applications. 5.4. Disrupting the mucus barrier to enhance particle transport Treatment of mucus with mucolytics agents [47] may improve the penetration rates of drug and gene carrier particles. The use of mucolytics as an adjuvant to particle transport may be particularly important for diseases where mucus is abnormally viscoelastic, such as CF and COPD. For example, in CF sputum, water content is reduced and cellular debris is increased, leading to an increase in physical entanglements, a decrease in the average mesh pore spacing, and a pronounced increase in the viscoelastic nature of the sputum [194] and [195]. Consequently, CF mucus is suggested to be a confounding barrier even to viruses [196], as well as a major cause for low gene transfer efficiencies in vivo [89] and [197]. Pulmozyme®, a recombinant human DNase (rhDNase) also known as Dornase alfa, is the most commonly used mucolytic in CF and a potential candidate as an adjuvant to particle transport.

Pulmozyme hydrolyzes the DNA that forms dense entanglements with mucin glycoproteins and other mucus constituents, thus reducing the number of crosslinks and the viscoelasticity of mucus [198] and [199]. However, in a diffusion chamber study, Sanders et al. found only a moderate improvement in the transport of uncoated polystyrene nanospheres through CF sputum treated with rhDNase [112]. Dawson et al. observed that, while rhDNase treatment of undiluted CF sputum reduced its macroviscoelastic properties by up to 50%, it did not improve particle diffusion rates [118]. Multiple particle tracking analysis revealed that rhDNase treatment dramatically narrowed the distribution of individual particle diffusion rates in CF sputum, effectively removing the fast-moving outlier particle population. Dawson et al. suggested that soluble DNA fragments created by DNase treatment may increase the microviscosity within mucus pores, thereby eliminating fast-moving particles. This finding suggests that rhDNase may reduce, rather than enhance, the efficacy of nanoparticle therapeutics such as non-viral polymeric gene vectors, since the treatment eliminates the fast-moving fraction that may be critical for effective CF gene therapy [118]. Mucinex® (N-acetyl-L-cysteine; NAC) is another common mucolytic previously tested for its ability to improve the penetration of nanoparticles across CF sputum. NAC reduces the cross-linking of mucin fibers by cleaving disulfide bonds and thus reduces the bulk mucus rheology [200] and [201]. Alton et al. showed that the mucus barrier to non-viral gene vectors may be at least partially overcome by treatment with NAC in an ex vivo model of sheep tracheal epithelium [202]. NAC treatment has also been shown to increase reporter gene expression in the mouse lung; however, no correction of the nasal transepithelial potential difference was observed in CF-null mice, suggesting NAC alone may be inadequate to induce clinically relevant gene therapy via mucoadhesive cationic liposomes and polymers. Although not widely investigated, other promising mucolytic agents are well established for reducing the bulk rheological properties of mucus and may potentially improve nanoparticle transport through mucus. These include Nacystelyn [203] and [204], as well as Gelsolin [205] and thymosin 4 [206], agents that depolymerize F-actin. The varying effects of mucolytics, as in the cases of NAC and rhDNase, suggest that any potential adjuvant use of mucolytics must be carefully investigated for each agent. Furthermore, patients may respond differently to various mucolytics. A more thorough understanding of failures in mucolytic response in different patients may alleviate these concerns [207]. 6. Conclusions The possibility of using polymer nanoparticles for controlled drug or gene delivery at mucosal sites over many hours to days is expected to lead to effective new therapeutics. However, no such product currently exists since conventional therapeutic particles cannot penetrate the human mucosal barrier, which rapidly clears trapped pathogens and particulates. The development of mucus-penetrating particles (MPP), by rendering the surfaces of particles non-mucoadhesive via lessons learned from viruses, provides a powerful strategy for overcoming the mucus barrier. In particular, the efficient transport of large MPP (200 to 500 nm) through human mucus should strongly encourage the commercial development of new nanoparticle -based drug delivery systems for use in various mucosal surfaces, since drug delivery kinetics and loading efficiency are vastly improved as particle

diameter increases. As such, MPP offer the prospect of sustained delivery of a variety of potentially important drugs to treat diseases at mucosal tissues. Acknowledgments This work was funded in part by the NIH (R01EB003558, R21HL089816, and R21EB008515), the Cystic Fibrosis Foundation (HANES08G0) and fellowships from the NSF (Y.-Y.W.) and the Croucher Foundation (S.K.L