Nanoscale

View Article Online FEATURE ARTICLE View Journal | View Issue

The particle in the spider’s web: transport through biological hydrogels Cite this: Nanoscale, 2017, 9, 8080 Jacob Witten a,b and Katharina Ribbeck*a

Biological hydrogels such as mucus, extracellular matrix, biofilms, and the nuclear pore have diverse func- tions and compositions, but all act as selectively permeable barriers to the diffusion of particles. Each barrier has a crosslinked polymeric mesh that blocks penetration of large particles such as pathogens, nanotherapeutics, or macromolecules. These polymeric meshes also employ interactive filtering, in which affinity between solutes and the gel matrix controls permeability. Interactive filtering affects the transport Received 18th December 2016, of particles of all sizes including peptides, antibiotics, and nanoparticles and in many cases this filtering Accepted 12th May 2017 can be described in terms of the effects of charge and hydrophobicity. The concepts described in this DOI: 10.1039/c6nr09736g review can guide strategies to exploit or overcome gel barriers, particularly for applications in diagnostics, rsc.li/nanoscale pharmacology, biomaterials, and drug delivery.

1. Introduction: hydrogels are example, both archaea1 and bacteria can form biofilms, or microbial aggregates surrounded by secreted extracellular poly- ubiquitous selective barriers in biology meric substances (EPS) that act as a protective barrier and Biological hydrogels, which are composed of hydrated polymer create microenvironments within which microbes thrive and 2,3 networks, are found throughout every domain of life. For adapt to harsh conditions. In eukaryotes, hydrogels have evolved to fulfill an astonishingly diverse set of functions. For example, the nuclear pore is a barrier formed from crosslinked aDepartment of Biological Engineering, Massachusetts Institute of Technology, intrinsically disordered proteins called nucleoporins, which Cambridge, MA 02139, USA. E-mail: [email protected]; Tel: +1 (617) 715-4575 b control the passage of macromolecules between the nucleus Computational and Systems Biology Initiative, Massachusetts Institute of 4,5 Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. and the cytoplasm. One of the largest hydrogels in the body Technology, Cambridge, MA 02139, USA

Jacob Witten received his BA in Katharina Ribbeck is Eugene Bell Mathematics and Biophysics Career Development Professor of from Amherst College in 2014, Tissue Engineering in the where he researched compu- Department of Biological tational methods for hydrogen Engineering at the Massachusetts exchange mass spectrometry. In Institute of Technology. She 2015, he joined Professor received her Ph.D. in Biology at Katharina Ribbeck’s group in the the University of Heidelberg in department of Biological Germany. Her lab studies the Engineering at Massachusetts basic mechanisms by which bio- Institute of Technology as a doc- logical hydrogels barriers toral student. His current exclude, or allow passage of Jacob Witten research combines compu- Katharina Ribbeck different molecules and patho- tational and experimental tech- gens, and the mechanisms patho- niques to study molecular transport in mucus, focusing on appli- gens have evolved to penetrate mucus barriers. Her goal is to cations to drug delivery for treating diseases such as cystic develop the foundation for a theoretical framework that captures fibrosis. general principles governing selectivity biological hydrogels, and for the reconstitution of synthetic gels that mimic the basic selective properties of biological gels.

8080 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

ponents. However, biofilm EPS commonly contains extracellu- lar DNA and various types of polysaccharides.2 One well- studied biofilm is that formed by Pseudomonas aeruginosa,an opportunistic pathogen that can cause biofilm-associated infection in wounds27 or in the lungs of patients with cystic fibrosis (CF).28 Interestingly, even this single species pro- duces anionic (alginate), cationic (Pel),29 and neutral (Psl)30 polysaccharides in differing amounts depending on context and bacterial strain.31 Fibrous structural proteins such as collagens, elastin, fibro- nectin, and laminins, together provide much of the structural integrity of ECM. ECM also contains high levels of polyanionic glycosaminoglycans (GAGs) and proteoglycans (proteins with densely grafted GAGs) that give ECM a high net negative charge density. Structural components of ECM are covalently crosslinked to varying degrees, regulating matrix stiffness.11 In human mucus, the main gel-forming components consist of a family of large, intrinsically disordered secreted glycoproteins called mucins. The main secreted mucins are MUC2, MUC5B, and MUC5AC, although other secreted mucins can be present in smaller amounts as well.7 Mucins have alter- nating regions of small, globular hydrophobic domains and highly glycosylated, primarily anionic (due to sialic acid and 7,32,33 Fig. 1 Scanning electron microscopy of biological hydrogels. (a) sulfation) unstructured regions. Other components of Cervical mucus samples from pregnant patients at low or high risk for mucus include lipids, soluble proteins and peptides, and preterm birth. Scale bar: 200 nm. Reprinted with permission from nucleic acids.32 Mucus composition can vary depending on 196 Critchfield et al. (2013). Copyright (2013) PLOS. Published under CC disease state, and in this review we highlight CF, a genetic BY license. https://creativecommons.org/licenses/by/4.0/legalcode. (b) μ disease in which improper ion balance results in pathologi- Reconstituted basal lamina ECM gels. Scale bar: 25 m. Reprinted with 34 permission from Arends et al. (2015).197 Copyright (2015) PLOS. cally thick, dehydrated mucus. The presence of necrotic Published under CC BY license. https://creativecommons.org/licenses/ neutrophils in CF airways results in high levels of free DNA by/4.0/legalcode. and actin filaments in lung mucus,35 increasing the mucus’ viscoelasticity32 and playing important roles in binding to cations. More detailed reviews can be found on composition of is the mucus gel (Fig. 1a) that lines all wet epithelia and pro- biofilms,2,36 of the ECM,11,24,37 and of mucus.6,7,32,38,39 Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. tects the underlying cells against toxins, pollutants, and invad- While biological hydrogels are distinct in terms of their – ing pathogens.6 10 Another example of a hydrogel-based locations and molecular compositions, they share certain barrier is the extracellular matrix (ECM) surrounding cells common principles that govern selective filtration. The goal of within tissues. ECM provides mechanical stability but also this review is to address the general principles that apply to forms a selective barrier that regulates transport of signaling the permeability of many types of biological hydrogels. – molecules secreted by the cells (Fig. 1b).11 16 Broadly speaking, the transport of a solute through a gel is The permeability of biological hydrogels poses a challenge controlled by the solute’s size, its interactions with the com- for biomedical development. For example, biofilms are ponents of the gel, or a combination of the two (Fig. 2).40 Size, involved in the majority of infections in developed countries; or steric, filtering is a universal feature of biological gels, they form on medical implants and wounds, cause middle-ear which have a polymeric mesh size that constrains the infections and gingivitis, and more.17 One driver of antibiotic diffusion of large particles (Fig. 2a). Pure steric filtering is an resistance in biofilms is binding of antibiotics to EPS, which important component of hydrogel selectivity, but it is also sequesters them and/or reduces their penetration into the crude because it only selects based on one parameter. Filtering biofilm. In humans, the ECM is an obstacle to drug delivery, based on chemical interactions between solutes and gel com- as its mesh prevents large nanoparticles from penetrating ponents is also a common feature of biological gels. – deep into tissue or tumors.18 24 Mucus similarly controls the Depending on their chemistry, the gel components interact permeability of nanoparticles,25,26 but can also limit diffusion with solutes across gradients of chemical properties including of small molecules such as antibiotics. charge and hydrophobicity, thereby differentially affecting Biological hydrogels are complex molecular assemblies diffusion.40 A lack of interactions enables unhindered with context-dependent properties. For biofilm EPS, the struc- diffusion (Fig. 2b), while certain weak and diffuse solute–gel ture of the matrix varies widely across species, strains, and interactions can facilitate penetration of the solute into the gel environments, complicating the characterization of EPS com- without slowing transport (Fig. 2c). Binding to hydrogels, on

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8081 View Article Online

Feature article Nanoscale

Fig. 2 Effects of steric hindrance and chemical interactions on gel penetration. (a) Particles above the mesh size are unable to penetrate the gel, even if they do not interact with the gel. (b) Small inert particles penetrate gels. (c) Under some conditions (see Section 3.4 for details), weak inter- actions with gel polymers can enhance partitioning into the gel and subsequent penetration. The schematic assumes that the bath is fixed at a con- stant solute concentration. (d) Binding to the gel causes enrichment of solute at the bath-gel interface but slowed gel penetration.

the other hand, reduces effective solute diffusivity and hinders disordered proteins called nucleoporins that form a mesh with penetration (Fig. 2d). In drug delivery applications, facilitated a mesh size ∼2.5–5 nm in size that excludes proteins and transport and reduced effective diffusivity from binding can be protein complexes larger than 30–100 kDa, unless they are cha- combined to optimize drug delivery.21,41 peroned by a nuclear transport receptor (NTR).48,49 Note that biological gels are often heterogeneous, with a distribution of mesh sizes. For example, some biofilms 2. Size-dependent filtration contain channels that allow efficient passage of large particles to cells deep in the biofilm.46,50 In mucus, nanoparticles often 2.1 Length scales and hydrogel mesh show a broad diffusivity distribution in particle tracking Size effects are important for the transport of viruses, bacteria, experiments, suggesting that some nanoparticles are trapped ff 32,45,51 eukaryotic cells, particulate pollutants, nanoparticles, and any while others di use nearly freely. One important ques- ff other particle on the same length scale as the gel mesh. In this tion is whether the regions that allow free di usion are con- ffi section, we discuss size filtering while assuming that all par- nected enough to allow e cient penetration through macro- ticles are inert to (do not interact with) gel components. To a scopic mucus layers, or whether these regions are simply large first approximation, steric interactions are quite simple, water-filled pores surrounded by impassable polymeric bar- corresponding in the macroscopic world to the fact that an ele- riers. Answers to this question are somewhat conflicting, but Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. phant, but not a fly, can be stopped by a chain-link fence, overall it appears that mucoinert particles pass through macro- 52–54 while a tennis ball can pass through with some prodding. scopic mucus barriers. In biological hydrogels, the microscopic mesh is formed by On a related note, collagen fibrils in ECM are often aligned entangled and crosslinked polymers. The distance between with each other; this directionality of the mesh drives an- ff adjacent links in the chain-link fence corresponds to the mesh isotropic di usion behavior that may be important for transport 55,56 ff size; objects much smaller than the mesh size diffuse at a rate in ECM. The importance of di usion anisotropy is unclear corresponding to the viscosity of the interstitial fluid (the fluid in other gels, although filamentous bacteriophage can drive between fibers; Fig. 3a), objects on the order of the mesh size liquid crystallization of P. aeruginosa biofilms, suggesting the 57 are obstructed but not completely stopped (Fig. 3b), and potential for anisotropic transport in that context. Finally, in objects much larger than the mesh size are trapped (Fig. 3c). contrast to a chain-link fence, in gels thermal motion and 58 This steric barrier is important for mucus because it blocks structural dynamics play a role in controlling transport. Gel and/or traps large pollutants and potential pathogens, thus polymers and physical crosslinks constantly rearrange, with allowing some mechanism, such as mucociliary clearance, to local deformations allowing large particles to escape steric bar- – 8 ff 59 63 clear the invading particle before it can reach the epithelium. riers and caging e ects. Mesh size is also an important consideration for nanoparticle design, as it sets a maximum possible size for a nanoparticle 2.2 Modulation of gel structure by transporting particles that must penetrate a hydrogel barrier. The mesh sizes of The transport of a particle through a gel may also be promoted mucus, biofilm EPS, and ECM vary and are on the order of by the solute directly interfering with the gel’s component – 10–1000 nm.19,20,23,24,42 46 Note that measurements of mesh polymers or crosslinks, thus changing the mesh size. For size may be incorrect if the particles used to probe the mesh example, attaching the mucolytic protein papain to nano- size interact with the gel;47 we discuss adhesive interactions in particle surfaces somewhat enhanced penetration of the nano- detail in section 4. The nuclear pore is filled with intrinsically particles through intestinal mucus (Fig. 3d).64,65 Treatment of

8082 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

Fig. 3 Steric effects on diffusion in gels (a–c) and mechanisms to modulate steric hindrance (d, e). Thin lines represent thermal motion of the par- ticle. (a) Particles smaller than the mesh size diffuse freely in interstitial fluid. (b) Particles on the order of the mesh size have significant steric hin- drance but eventually penetrate gels. (c) Large particles are trapped. (d) Particles that cleave gel polymers may diffuse more quickly. Notched green circles attached to particle are lytic enzymes; green segments of gel polymer are substrates for the enzymes. (e) Particles may reversibly disrupt gel crosslinks (blue–blue contacts), allowing enhanced diffusion without irreversibly degrading the gel.

CF lung mucus with N-acetyl cysteine, a disulfide bonding Bacterial motion disrupts the gel structure,72 while lytic reducer, likewise facilitated transport by increasing the mesh enzymes such as alginate lyase are both used by bacteria and Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. size,66 although surprisingly treatment with recombinant have potential as anti-biofilm treatments.73,74 The stomach DNase (used in CF to degrade DNA in mucus) did not improve pathogen Helicobacter pylori has also been shown to facilitate transport.67 An elegant example of crosslink interference may its motion by secreting high levels of ammonia into stomach occur in the nuclear pore (discussed in detail in Section 5), in mucus; the local increase in pH deprotonates carboxylic acid which transporters reversibly disrupt physical crosslinks, groups in mucin, which reduces intermolecular hydrogen allowing fast transport and rapid self-healing of the pore struc- bonding and hydrophobic interactions while increasing ture (Fig. 3e).68 This mechanism of reversibly disrupting physi- electrostatic repulsion between mucin strands. H. pylori is able cal crosslinks has not, to our knowledge, been found in any to easily swim through this locally weakened gel.75 other biological system, but it presents an intriguing mechan- ism for biology and engineering applications. Finally, mucin- 2.3 Techniques to analyze particle transport binding particles may sequester mucin strands, reducing the Experimental assays of particle transport generally require the concentration of gel strands elsewhere in the gel. This particles to be fluorescently labeled or radiolabeled. From reduction increases the effective mesh size and thus enhances there, one common technique to measure transport is direct transport of other, non-mucin binding nanoparticles.69,70 visualization of bulk particle transport into or through a gel, Living cells also modulate gel structure as they move. Many followed by qualitative analysis76 or quantitative fitting of con- types of human cells degrade and/or secrete various ECM centration timecourses or spatial profiles as the particles pene- elements in tightly regulated ways, with degradation some- trate the gel.77,78 Bulk transport visualization methods are times necessary for cell motility.71 Improper regulation of ECM simple and easily visualized, but they are generally unable to structure, including overexpression of ECM-degrading matrix measure heterogeneity or other structural details. Two related metalloproteinases or the collagen crosslinking factor LOX, techniques for analyzing particle transport in gels are fluo- has been associated with numerous types of cancer.37 rescence recovery after photobleaching (FRAP) and fluo- Similarly, biofilm EPS is constantly remodeled by bacteria.42 rescence correlation spectroscopy (FCS).55,79 In FRAP and FCS,

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8083 View Article Online

Feature article Nanoscale

the gel is pre-incubated with fluorescently labeled particles. In small solutes such as drugs and proteins.46,91,92 SPT is more FRAP, a region of the gel is photobleached and the recovery of difficult for smaller solutes because of the difficulty inherent fluorescence in the bleached spot is fit to models that calculate in resolving individual molecules with fluorescence diffusion parameters.80 In FCS, the fluorescence autocorrela- microscopy; this technique is rarely if ever used to study tion function within a small gel region is measured and fit.79 diffusion of proteins or smaller molecules in gels. With FRAP and FCS, observing and fitting parameters for sub- To clarify what exactly is meant by slowing solute pene- diffusion may be possible using appropriate fitting tration and improving retention, we briefly present the stan- – procedures.81 84 Subdiffusion refers to diffusion behavior in dard model for diffusion of a solute with first-order reversible which the mean squared displacement of individual particles binding to a gel. Under the assumptions that (1) binding and α scales with t for α < 1, rather than with t1 as in standard unbinding are fast, such that the solute reaches local equili- diffusion; subdiffusion or transient subdiffusion is commonly brium between bound and free states at each point, (2) bound observed in gel transport studies and can impact important solute is immobile, (3) binding sites within the gel are far parameters such as passage time through a gel.85,86 from saturated, and (4) binding to the gel does not disrupt gel ff ff An alternative to bulk transport analysis is single particle structure, we can define the e ective di usivity Deff as: tracking (SPT), in which the thermal motion of individual par- 87 ¼ ð þ = Þ 1 ð Þ ticles is recorded and analyzed. SPT easily identifies both Deff DF 1 NT K D 1 subdiffusion and heterogeneity in gel samples, which is useful where D is the diffusion coefficient of the free solute in the for identifying subpopulations of particles that diffuse rela- F gel (which may be lower than the diffusivity in water, due to tively unhindered.32 This technique’s main weakness is that steric constrains or increased interstitial fluid viscosity), N is diffusion out of a microscope’s focal plane renders it difficult T the total binding site density in the gel, and K is the dis- to track individual particles over tens of microns, which D sociation constant. means that unlike in bulk diffusion or FRAP experiments, The timescale for a solute to penetrate a gel (or equiva- direct measurements of penetration through physiologically lently, the timescale of escape from a gel) τ is related to the sized gels is not feasible.54 lag D ff by: Finally, nanoparticle dispersal has been imaged in vivo for e a variety of mucus systems as well as brain ECM. These assays 2 τlag L =Deff ð2Þ do not provide quantitative measurements of diffusion para- meters, but nonetheless serve as a valuable bridge between where L is the length scale of the gel. These two equations in vitro studies and translation to the clinic.47,52,88,89 show how increasing solute–gel binding strength or gel

binding site density increases τlag, thus slowing penetration or improving retention.77 Fig. 4a shows this effect in action, as 3. Interactive filters the penetration of tobramycin (an aminoglycoside antibiotic) into a P. aeruginosa biofilm is hindered by binding of tobramy- ff 3.1 Overview: e ects of chemical interactions on gel cin to biofilm components.76 Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. penetration and retention Given these results, it is no surprise that specific protein– Biological hydrogels selectively interact with foreign particles ligand interactions are commonly exploited to regulate trans- based on the solutes’ chemical properties. These interactions port and retention within a gel, often for protective purposes often lead to reversible binding, which may be desirable, un- or to control signaling.12,93 For example, sialic acid-terminated desirable, or neutral depending on context. Since reversible O-linked oligosaccharide chains on secreted mucin molecules, binding improves retention within a gel, it may enable longer particularly Muc5AC, act as decoy receptors for sialic acid tar- drug action, as in mucoadhesive drug delivery systems.9,90 geting bacteria and viruses, including the influenza virus.94 Similarly, binding of native immune factors to cervical mucus This interaction slows penetration of sialic acid targeting bac- may promote their retention at high concentration and prevent teria and viruses through the mucus layer to the vulnerable microbial penetration through the cervical mucus plug during epithelium95,96 and competitively inhibits binding to cell- pregnancy, which could lead to intrauterine infection and associated sialic acid.94 A mouse model with lung Muc5AC preterm labor.39 On the other hand, reduced diffusivity may overexpression has increased resistance to influenza, illustrat- slow the penetration of a gel (Fig. 2d), impacting the oral bio- ing this protective function of lung mucus.97 In the ECM, availability of drugs or microbial killing by antibiotics. binding of insulin-like growth factor (IGF) to IGF binding pro- Similarly, binding can sequester a molecule and decrease its teins limits IGF transport, thus helping to regulate intercellu- effective concentration. Finally, diffusivity could simply be a lar signaling.13 Sequestration of transforming growth factor-β tunable regulatory parameter, neither intrinsically “good” nor in the ECM is another well-studied case showing the impor- “bad.” Whatever the effect of gel binding, unlocking the forces tance of binding to the gel matrix.98,99 In addition to non- underlying it is crucial to understanding the function of native covalent protein–ligand interactions, covalent linkage to bio- systems and optimizing drug function. logical hydrogels is also possible; for example, certain muco- As with nanoparticles, bulk diffusion assays, FRAP, and FCS adhesive formulations use free thiol groups to form disulfide are commonly used to measure the diffusion coefficients of bonds to mucin.100

8084 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

Fig. 4 Distinction between “binding” and “partitioning.” (a) Penetration of Cy5-labeled (fluorescent) tobramycin and ciprofloxacin into P. aeruginosa biofilms. Plots represent quantified timecourses of penetration, with representative images shown above plots. Cy5-tobramycin penetrates the gel less extensively than Cy5-ciprofloxacin. Adapted with permission from Tseng et al. (2013).76 Copyright (2013) Society for Applied Microbiology and John Wiley and Sons Ltd. (b) Penetration of Avidin and a neutral Avidin variant (NeutrAvidin), both fluorescently labeled with fluorescein isocyanate (shown in green), into bovine articular cartilage. Positively charged Avidin penetrates more than NeutrAvidin. Reprinted with permission from Bajpayee et al. (2014).21 Copyright (2014) Elsevier. (c, d) Schematic of difference between a and b in terms of nanoscale free energy landscape. (c) Cy5-tobramycin binds specific targets and must escape binding energy wells to continue diffusing, meaning that the energy landscape is not con- stant on the nanoscale. (d) Avidin diffuses in nearly uniformly charged cartilage; thus, binding energy wells are minimal and a Donnan treatment of electrostatics is appropriate. Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. While eqn (1) and (2) describe the simplest effects of gel to act as a nanocarrier for osteoarthritis drugs.21,41 We empha- binding, interesting cases arise when the assumptions are vio- size the dramatic difference between Fig. 4a and b: while lated. For example, saturation of gel binding sites results in tobramycin and Avidin both electrostatically interact with their enhanced penetration because there are fewer binding sites respective gels, tobramycin penetration is inhibited available per solute molecule to arrest diffusion.76 In another with respect to a neutral counterpart, whereas Avidin pene- example, Braga et al. showed that if bound solute is mobile tration is facilitated. The difference arises because Avidin-ECM

with diffusivity DB but the other assumptions hold, (we have electrostatic interactions do not hinder Avidin diffusion, changed the notation slightly to stay consistent with this and in fact these interactions do not constitute binding at all review):101 in some sense; Avidin transport is best described using Donnan partitioning analysis.77,103 Mechanisms of mobility D ¼ð1 f ÞD þ fD ð3Þ eff F B for interacting solutes such as nuclear pore transporters and where f is the fraction of time a solute is bound (or equi- Avidin are related to the nanoscale energy landscape encoun- valently, the equilibrium bound fraction of solute at equili- tered by the solutes (Fig. 4c and d), and are described in

brium). It can be shown that f = NT/(NT + KD), so we recover Section 3.4.

eqn (1) in the limit of immobile bound solute (DB = 0). In the Violation of criterion 4, that binding to the gel does not ff opposite limit of DB = DF, penetration is not slowed because disrupt gel structure, also has interesting e ects. For example, Deff = DF. In the nuclear pore, the mobility of bound solute high concentrations of tobramycin induce precipitation of algi- may help explain why weak binding of NTRs to nuclear pore nate gels; this disruption of barrier integrity facilitates tobra- components does not arrest passage.102 On the applied side, mycin transport at high concentrations.104 Also, reversible Fig. 4b shows that penetration of the positively charged binding of particles to physical gel crosslinks could disrupt protein Avidin into articular cartilage ECM is facilited by existing crosslinking (Fig. 3e), which could help overcome a electrostatic interactions (Fig. 4b), potentially allowing Avidin steric barrier as discussed in Sections 2.2 and 5.

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8085 View Article Online

Feature article Nanoscale

3.2 Biochemistry of reversible binding to the gel matrix The importance of hydrophobic interactions in small mole- cule transport has mainly been studied in the context of oral Reversible binding to biological hydrogels is important for drug bioavailability, which is partially dependent on the ability molecules throughout the size spectrum;92 even proton trans- to diffuse through intestinal mucus to the epithelium. Many port can be slowed by gastric mucus, which protects epithelial hydrophobic drugs such as paclitaxel, testosterone, and cin- tissue from stomach acid.105 Here, we discuss representative narizine bind to exposed hydrophobic sites on mucin, lipids examples that illustrate the importance of various inter- present in mucus, or both; binding slows penetration and molecular forces, as well as several cases showing how modu- renders intestinal mucus a barrier to absorption into the lating gel binding could deliver novel approaches to fighting bloodstream.115 infections. However, for many molecules it is not immediately clear The biochemistry underlying reversible binding to gels is whether hydrophobic or electrostatic interactions drive inter- difficult to study for several reasons. First, crystal structures of active properties. For example, cationic antimicrobial peptides gel components do not usually exist since gel polymers are (CAMPs), an important class of antibiotics and immune mole- generally intrinsically disordered. It is also often not obvious cules, contain both cationic and hydrophobic residues; the to which component(s) of the gel a solute binds, since biologi- charge, or the hydrophobicity, or both, could lead to gel cal gels contain a complex mix of components.76 Despite the binding. EPS–CAMP interactions are particularly important, as lack of structural information, existing literature on gel many antimicrobial peptides are under investigation as poss- binding and the composition of biological hydrogels yields ible anti-biofilm treatments, but binding to EPS components – useful predictions and conclusions about transport in gels; often inhibits their effectiveness.116 118 For example, the Deber these predictions usually center on how charge and hydro- group designed a set of synthetic anti-P. aeruginosa CAMPs – phobicity influence diffusion. Other intermolecular forces and investigated their interaction with alginate.119 122 While such as van der Waals forces and hydrogen bonding are impor- alginate is anionic, highly soluble, and contains no large tant, but their effect on binding is not easily predictable. In hydrophobic domains, alginate blocked the penetration of syn- one case, research has shown that hydrogen bonding is impor- thetic anti-P. aeruginosa CAMPs when a hydrophobicity tant for mucoadhesion of hydrophilic polymers such as poly threshold was exceeded.121 Furthermore, they showed that this (acrylic acid).90 In summary, analysis of a molecule’s overall interaction was mediated by a combination of electrostatic and chemical properties may predict that the molecule will bind to hydrophobic interactions with alginate.119 something within a gel. Similarly, both synthetic and natural CAMPs have been An instructive instance in which charge interactions domi- shown to bind mucins, DNA, and F-actin, and stronger nate is P. aeruginosa lung infections in CF patients, which binding correlates with greater inhibition of CAMP activity – involve two biological gels: mucus and biofilm EPS. against pathogenic bacteria in saliva and CF sputum.123 126 One primary class of antibiotics used to treat P. aeruginosa CAMPs are also investigated as anti-cancer treatments, and infections, aminoglycosides, are highly positively charged and unsurprisingly can be inhibited by ECM: the GAG heparan thus bind to polyanions such as mucins, DNA, and actin, sulfate inhibits antitumor CAMPs.127 Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. which are present at pathologically high levels in CF respirat- ory mucus. This binding impedes antibiotic penetration and ff – inhibits activity.74,106 The P. aeruginosa EPS acts as a second 3.3 The e ect of nanoscale heterogeneity on solute gel protective barrier for biofilm-associated cells. The EPS also interactions contains DNA and, in the case of mucoid strains, high levels of While knowing the composition of a gel gives an overall sense the polyanionic aminoglycoside-binding alginate, although of the gel’s net charge and helps predict how solutes behave, it alginate surprisingly seems to not significantly hinder is also important to understand how nanoscale molecular – penetration.107 109 heterogeneity, or spatially varying chemical properties, tunes Although charge-mediated binding to polyanions is impor- solute–gel interactions. Heterogeneous surfaces are commonly tant for aminoglycosides and other cationic antibiotics such as found in gel-related molecules: protein surfaces and anti- colistin,110 reversible gel binding does not seem to be a major microbial peptides contain charged, neutral hydrophilic, and factor in many other antibiotic treatments of biofilms;111 for hydrophobic residues, and larger objects such as viruses have example, Fig. 4a shows that ciprofloxacin easily penetrates a complex surfaces (Fig. 5a).128 In addition, the gels themselves model P. aeruginosa biofilm. In addition to slowing antibiotic often have locally varying properties that render simplistic penetration, biofilms have other resistance mechanisms models of gel charge problematic. For example, despite the net including slow metabolism and hypoxia in the biofilm interior negative charge of ECM, both positively and negatively charged (which are related to reaction-diffusion dynamics of nutrients nanoparticles and liposomes have reduced diffusivity in re- and oxygen, emphasizing the broad importance of diffusion in constituted ECM gels compared to neutral hydrophilic gels), and slow-dividing “persister cells.”17,112,113 The various particles. The observation that anionic particles are slowed resistance mechanisms may be cooperative; slowed pene- just like cationic particles is probably due to alternate patches tration could allow more time for bacterial adaptations to anti- of negative and positive charge in the gel that bind positive biotic treatment.114 and negative charges, respectively, on the particles.129,130

8086 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

that charge to reduce gel binding. However, the improvement in activity of the PEGylated tobramycin against biofilms was modest, because the PEG chain compromised the biological activity of the molecule.132 One possible strategy to avoid reduced activity is to make the PEGylation reversible via hydro- lysis, a strategy that has been investigated for gentamicin (another aminoglycoside) due to the potentially improved pharmacokinetic properties offered by PEGylation, but not studied in a biofilm context.133 Reversible PEGylation may allow for gel penetration, as well as full activity of the hydro- lyzed form. An alternate strategy would be to minimally change the structure of tobramycin in order to reduce gel binding without compromising activity. Recent experiments have suggested that modifying hydrophobicity could achieve this aim, because hydrophobic interfaces can affect electrostatic binding.134 We also note that immobilized charged groups on hydrophobic surfaces can tune hydrophobic interactions (Fig. 5c), which could be useful for changing the gel binding of partially hydro- phobic solutes such as CAMPs.135 For a more fully developed example of the utility of prevent- ing sequestration by gels, we consider the activity of lysozyme, Fig. 5 Nanoscale heterogeneity of solutes and effect on gel diffusion. a cationic antimicrobial protein, against P. aeruginosa in CF. 136 (a) Surfaces of (i) human rhinovirus (PDB: 2rm2) and (ii) human albumin Lysozyme is inhibited by F-actin, mucin, alginate, and DNA. (PDB: 2bxi), with positive charges in blue and negative charges in red. As with tobramycin, some positive charge is required for lyso- 128 Reprinted with permission from Cone (2009), (Copyright (2009) zyme’s function; however, in a screen of charge-reduced vari- Elsevier). Rhinovirus and albumin are densely coated with opposing ants, at least one variant reduced its inhibition by all of these charges. (b) Two fluorescently labeled peptides with the same net 137,138 charge but different spatial arrangement. The “block” peptide at left polyanions while retaining full antimicrobial activity. interacts weakly with mucin while the “alternating” peptide at right does This mutant lysozyme also displayed superior antibiotic not;131 schematic shows potential mechanism for this difference. (c) activity to wild type in a murine P. aeruginosa lung infection ff E ect of immobilized charges on nearby hydrophobic interactions, model.139,140 There was substantial variation in P. aeruginosa probed by measuring the adhesion of a hydrophobically functionalized killing by lysozyme both with and without the presence of gold (Au) atomic force microscopy tip to surface monolayers. Amine + polyanions throughout the screen, emphasizing that (as with (NH3 ) groups strengthen hydrophobic interactions between these 131 hydrophobic surfaces, while guanidinium (Gdm+) groups weaken or the charged peptide assay in Fig. 5b ) net charge is not Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. eliminate them. Adapted by permission from Macmillan Publisher Ltd: sufficient to determine the strength of electrostatic binding; 135 Nature (Ma et al., 2015). Copyright (2015). context and charge placement are important as well.137,138

3.4 Mechanisms for mobility of interacting solute As an example of the effect of nanoscale heterogeneity on We mentioned in Section 3.1 that gel interactions need not the solute side, when Li et al. examined the diffusion of two reduce effective diffusivity if the bound particles remain ff ∼ peptides, with the same near-neutral net charge but a di erent mobile (DB DF). We now present two mechanisms for fast arrangement of these charges, these peptides diffused differ- diffusivity of interacting particles: mobility of the bound ently into a reconstituted mucin gel. The block-charge peptide complex and high density of interaction sites. In the first interacted weakly with mucin, while the alternating-charge mechanism, if a solute is bound to a site which itself is 131 peptide did not (Fig. 5b). This result is indicative of how diffusing, then the solute is taken along for the ride with DB =

variations in molecular structure beyond simple net charge Dbound complex. Experimentally, Sprakel et al. (2007) observed affect transport properties in gels, which could have appli- transient subdiffusion of a silica particle bound to a polymer cations for drugs whose penetration is inhibited by gel network due to Rouse dynamics.60 binding. Bound complex mobility would be most significant in For example, the Smyth group covalently attached a 5 kDa weakly physically crosslinked gels comprised of flexible poly- polyethylene glycol (PEG) chain to tobramycin and found that mers because these properties allow fast polymer dynamics, the PEGylated variant was more effective in vitro against and for larger solutes because small molecules diffuse more ≫ P. aeruginosa biofilms than the unmodified antibiotic, likely quickly than polymers (DF Dbound complex). Also, bound due to reduced EPS binding. While the PEGylated tobramycin complex diffusion is faster on short timescales because cross- remained highly positively charged as is required for its mech- links and entanglements limit range of motion over long time- anism of action, PEGylation sufficiently changed the context of scales.60 Binding events with fast off rates will therefore take

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8087 View Article Online

Feature article Nanoscale

greater advantage of bound complex mobility. Transport of 4. Selective transport of objects on NTRs through the nuclear pore neatly checks off each of these constraints: NTRs transiently bind weakly crosslinked and the order of mesh size highly mobile nucleoporins. Thus, bound complex mobility 4.1 Adhesive interactions and polyvalent trapping helps explain why NTR binding to nucleoporins does not When large particles are not inert with respect to the gel result in slow transport, as one would naively expect given eqn matrix, even low-affinity binding tends to dramatically slow (1), and accurately predicts that strengthening NTR-nucleo- particle diffusion because the particles bind multivalently to 102,141,142 porin binding should inhibit transport. the gel (Fig. 6a).128 For example, while hydrophobic inter- To understand the high interaction site density mechan- actions moderately reduce the diffusivity of hydrophobic drugs ism, we consider an energy landscape interpretation of diffusion, in which interactions are represented by electro- chemical potential wells (Fig. 4c and d). Escape of a solute from potential wells into interstitial fluid is slow, thus imped- ing diffusion (Fig. 4c), as with tobramycin penetration into bio- films. However, if the density of binding sites is high enough, then it is possible for a solute to transition from one free energy minimum to another without transitioning to bulk interstitial fluid (Fig. 4d). The classic example of this is sliding diffusion on DNA: certain DNA-binding proteins transition from free diffusion in the nucleus or cytoplasm to one- dimensional sliding along DNA upon nonspecific DNA – binding.143 145 A similar effect likely takes place for the diffusion of Avidin (a positively charged 60 kDa protein) in articular cartilage ECM (Fig. 4b), in which proteoglycans allow full three-dimensional mobility. High densities of proteoglycans and GAG chains combined with weak nonspecific interactions enable cations such as Avidin to freely move along and between GAGs (Fig. 4d). In this case, since Avidin moves between interaction sites it does not make sense to refer to “binding” to any par- ticular site (Avidin actually also has true diffusion-arresting binding sites within ECM, for which the term “binding” does make sense).21 The partition coefficient between the gel and a bath in this limit can be quantitatively predicted via Donnan Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. equilibrium analysis, which assumes a constant potential throughout the gel (green dotted line in Fig. 4c and d).77 Note a subtlety of language found in the literature: partition coeffi- cient of a solute between two phases is usually defined as the concentration ratio of the solute in each phase at equilibrium, but in studies of diffusion in biological hydrogels, the par- tition coefficient often refers only to the partitioning of freely diffusing, unbound solute between the phases.77 We define partitioning here in the latter sense. Finally, differential partitioning into hydrogels can be driven by forces other than Donnan equilibrium. Steric inter- actions are an obvious case, as they drive partitioning of large solutes toward phases with larger mesh sizes. Aqueous two- phase systems (ATPSs) consisting of phase-separated dextran and PEG gels differentiate biomolecules based partly on their hydrophobicity.146,147 Similarly, partitioning into membrane- less organelles such as Cajal bodies, P bodies, and P granules —which are likely formed via phase separation, sometimes Fig. 6 Mechanisms of polyvalent trapping in gels. (a) Non-specific including a sol–gel transition—is a key feature of cellular interactions (hydrophobic, electrostatic, etc.) with gel can trap a particle, 148,149 even if individual interactions are weak. (b) Binding to decoy receptors organization. Finally, transport through the nuclear pore (such as sialic acid) that are present on gel polymers can trap a particle. depends in part on a combination of hydrophobic and electro- (c) Gel-binding antibody bound to an otherwise inert particle mediates static partitioning effects.102,150 trapping.

8088 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

in mucus, polystyrene beads are almost completely trapped The mucosal immune system exploits polyvalency by because they form many hydrophobic bonds simultaneously. employing antibody-based trapping of viruses. For example, Similarly, aminated beads bind to polyanions in mucus, gener- anti-HSV-1 immunoglobulin G (IgG) located in cervical mucus ally reducing their diffusivity relative to carboxylated poly- can prevent HSV virions from penetrating the mucus and styrene beads.43,44 Carboxylated beads in turn diffuse far more infecting cells. HSV trapping occurs because the Fc region of slowly in mucus than do PEGylated beads, due either to hydro- IgG binds mucin with low affinity; while this binding does not phobic interactions resulting from incomplete coverage by car- significantly hinder free IgG transport, when many IgG mole- boxyl groups or to interactions with patches of positive charge cules coat one HSV virion, the weak polyvalent binding to on mucin molecules (similarly to the description of ECM, as mucins adds up to an interaction that is strong enough to described in Section 3.3).51,129,151 PEGylated beads also diffuse nearly completely trap the virion (Fig. 6c).160 Recent results more quickly than cationic and anionic nanoparticles in bio- suggest antibody trapping may even be more important than films; the relative diffusivities of the charged nanoparticles sialic acid-mediated trapping for preventing influenza pene- depend on the species.44,46,152 tration.163 It should be noted that strong antibody binding to Low molecular weight (2–5 kDa) PEG coatings are particu- mucus and/or membrane-bound mucins may or may not – larly effective at reducing adhesive trapping for two reasons. occur;164 166 however, theoretical treatments have shown that First, PEG is neutral yet hydrophilic, thus precluding strong weak antibody-mucin binding is sufficient for trapping.167,168 electrostatic or hydrophobic interactions.51 Second, high- Finally, exogenously added anti-PEG IgG and IgM appear to density surface-grafted PEG forms a brush that physically be able to trap PEGylated nanoparticles in mouse cervicovagi- resists the adsorption of gel components,153 while the low PEG nal mucus via the same polyvalent trapping mechanism as for molecular weight precludes adhesion to gel components due viruses.169 As mucosal anti-PEG antibodies may be common in to chain entanglement.51 Neither of these reasons is gel- the human population but are probably not present in animal specific, which explains why PEGylation seems to be fairly uni- mucus models, antibody-mediated trapping (and immuno- versally applicable to promoting diffusion in biological hydro- genicity of PEG in general170) is a potential hurdle for clinical gels: PEGylated particles effectively penetrate many types of applications of PEGylated nanoparticles.169 – mucus,51 53,89,154,155 biofilms,44 and ECM.47,156,157 In addition, recently Maisel et al. (2016) recently showed that PEG mole- 4.2 Implications of gel penetration studies for nanoparticle cular weights as high as 40 kDa can be used for mucus pene- design tration, with higher molecular weights requiring extremely Nanoparticles are used for a myriad of functions in medi- high PEGylation density to reduce entanglements.158 cine171 including gene therapy,53 antimicrobial172 and anti- In addition to uniform surface coatings such as PEGylation cancer treatments,173 insulin delivery for diabetes,174 and sus- or carboxylation, heterogeneous particle surfaces have been tained drug release,175 and many of these contexts require investigated in the context of viral diffusion in mucus. Viruses diffusion through a gel layer discussed in this review. Proper such as human papilloma virus (HPV), Norwalk virus, herpes surface functionalization is therefore key to developing simples virus 1 (HSV-1), and human immunodeficiency virus effective therapeutics. Two major lessons emerge from the Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. (HIV) quickly penetrate cervical or cervicovaginal mucus under studies of PEGylation, uniformly charged nanoparticles, and pH-neutral conditions in the absence of antibodies,91,159,160 viruses described above. First, even weak adhesive interactions likely because these viruses are densely coated with both posi- can trap a particle if there is a high density of binding sites. Of tive and negative charges (Fig. 4a), yielding overall hydrophili- particular note is the ability of the immune system to generate city without having any uniformly charged domains large antibodies against essentially all foreign substances, which enough for electrostatic interactions.128 This explanation is means that antibody trapping should be considered wherever corroborated by the charged peptide diffusion assay, in which an immune response could be present. the alternating-charge peptide had weaker interactions with Second, a good strategy across gel types to avoid both hydro- mucin than the block-charge peptide (Fig. 4b).131 Also, in con- phobic and electrostatic binding is to have a dense hydrophilic trast to influenza, these viruses do not specifically bind sialic but overall neutral coating, which has been implemented in acid; as mentioned in Section 3.1, sialic acid binding hinders two different ways: neutral hydrophilicity and dense opposite influenza transport and this may be compounded by multi- charges. In addition to PEGylation, N-(2-hydroxypropyl) meth- valency (Fig. 6b). At any rate, some non-sialic acid dependent acrylamide polymer (pHPMA)174,176,177 and poly(vinyl alcohol) viruses are unable to penetrate the mucus barrier: adeno- (PVA)178 were recently tested as alternative mucoinert coatings, viruses and adeno-associated viruses (AAVs) used as gene with some success. A dense coating of opposite charges has therapy vectors diffuse thousands of times more slowly in CF shown effectiveness in multiple different contexts, including sputum than in water,161 potentially explaining why the many viruses (as discussed above), self-assembled nanoparticles gene therapy trials for CF employing adenoviruses or AAVs comprised of distinct ratios of polycationic chitosan and poly- have all failed. Interestingly, mutating an AAV to reduce anionic poly(acrylic acid) (also in mucus),26,179 zwitterionic heparin binding enhanced penetration through CF sputum, dilauroylphosphatidylcholine monolayer-coated nanoparticles highlighting the potential for rational design of gene delivery (in intestinal mucus),180 and liposomes with zwitterionic vectors to enhance transport.162 dipalmitoyl phosphatidylcholine membranes (in biofilms and

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8089 View Article Online

Feature article Nanoscale

respiratory mucus as an inhaled treatment carrying amikacin, penetrate the hydrogel.102 NTRs locally dissolve the gel by an aminoglycoside; currently in clinical trials).172 Not all of binding FG repeats and disrupting the crosslinks, allowing for these coatings provide ideal protection against gel interactions, penetration of large objects (Fig. 3e).68 While individual FG and in particular polymer brushes form a useful physical crosslinks are weak and highly dynamic,185,186 even transient adsorption barrier in addition to a biochemical one. Yu et al. crosslinks may contribute to barrier function, and these transi- found that while a zwitterionic liposome diffused faster than ent crosslinks may be dissolved by hydrophobic NTR regions. polystyrene nanoparticles in cervical mucus, PEGylating the Brownian dynamics simulations suggest that while FG cross- liposome still significantly increased its diffusivity.181 links are dynamic, at each point in time the crosslinks perco- The utility of promoting gel transport is application depen- late through the center of the pore, forming a hydrogel in dent, and gel adhesion may be unavoidable or even desirable.9 some technical senses.187 Purified FG-rich nucleoporin Mucoadhesive drug delivery systems are useful for promoting domains self-assemble into hydrogels with transport selectivity long contact times with mucosal surfaces, thus allowing for similar to that of the nuclear pore.68 sustained drug release, particularly on surfaces with slow turn- The other main family of models, the virtual gate and/or over.9 Cationic liposomes may have increased contact time entropic barrier models, notes that there is a free energy with biofilms and promote sustained antibiotic release, and barrier to entering the nuclear pore because nuclear pores are their affinity with bacterial membranes gives them increased extremely small relative to the size of the nucleus or cyto- antibacterial efficacy.175 Furthermore, even slowly diffusing plasm.188 Thus, entering the pore imposes a loss of transla- nanoparticles will eventually penetrate a gel depending on the tional entropy and corresponding increase in free energy; timescale of treatment, and even if the particles themselves do entropic exclusion by nucleoporins causes an even greater not penetrate, released free drug likely will. increase in free energy.185,186 In these models, NTR binding to Several groups have designed nanoparticles that change FG repeats functions like Donnan partitioning: it locally properties over time in order to exploit both fast initial increases the concentrations of NTR near and inside the diffusion and the favorable properties of cationic particles (gel nuclear pore, without immobilizing the NTRs due to fast adhesion and/or cell penetration). Two approaches have been unbinding from FG repeats.141,142 This local enrichment of used: (1) a dissociable mucoinert coating with a cationic NTRs overcomes the barrier to nuclear pore entry and thus core,174 and (2) alkaline phosphatase-mediated phosphate clea- enables effective transport of NTR-cargo complexes.189 Note vage that increases overall surface charge over time, thus switch- that the virtual gate and selective phase models are not ing from mucopenetration to charge-mediated trapping.182,183 mutually exclusive; Donnan-like transport facilitation effects, entropic exclusion, and dissolution of FG repeat crosslinks may all be important for nuclear pore transport in vivo. 5. The nuclear pore: a case study in Finally, the fact that hydrophobic interactions with FG transport at the molecular level repeats are required for transport through the nuclear pore does not mean that charge is not important as well. Our group Having illustrated the main factors that control transport behav- has found that nucleoporins and NTRs tend to have comp- Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. ior in biological polymeric matrices, we now discuss the nuclear lementary net charges (positive and negative respectively), pore in detail as a case study. Nuclear pore translocation encom- thus implying that charge assists in the recruitment and trans- passes every concept discussed above and has been studied in location of NTRs;150 this hypothesis is supported by theore- molecular detail, which is not possible for complex hetero- tical189 and structural141 analysis. geneous macroscopic gels such as mucus, biofilms, and ECM. The nuclear pore complex has a diameter of approximately 100 nm, and the central channel is filled with intrinsically dis- 6. Outlook ordered proteins called nucleoporins. These nucleoporins contain many instances of short hydrophobic repeats called We have taken a whirlwind tour through the forces and con- FG repeats, which likely act as crosslinkers. The nuclear pore cepts that guide diffusion through ECM, mucus, biofilm EPS, inhibits nuclear translocation of proteins larger than roughly and the nuclear pore. In all of these systems, size filtration pre- 30–100 kDa and even some smaller proteins such as his- vents the penetration of overly large particles via a steric tones.48,184 However, NTRs efficiently carry large cargo through barrier; in turn, size filtration can be modulated by gel clea- the nuclear pore.49 This transport ability of NTRs derives in vage or reversible crosslink disruption. Interaction filtering large part from exposed hydrophobic patches on NTR surfaces, enables more precise tuning of gel selectivity. For particles which have hydrophobic and/or π–π interactions with the FG below the mesh size, the nuclear pore displays little selectivity; repeats on nucleoporins.68 The mechanism by which these diffusion in the ECM is mainly determined by electrostatics; interactions mediate selective transport is still under debate, both hydrophobic and electrostatic interactions are important and multiple models exist. in mucus; and electrostatic interactions are certainly impor- In the selective phase model, the nuclear pore is rep- tant in biofilms, while hydrophobic interactions may or may resented as an FG repeat-crosslinked hydrogel whose pore size not be important depending on the species and environment. determines the size cutoff above which inert particles cannot Electrostatic interactions can have multiple effects that facili-

8090 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

tate or repress particle penetration depending on whether the 0038, NIH R01-EB017755, NSF Career PHY-1454673, the interactions constitute binding or Donnan partitioning; these MRSEC Program of the National Science Foundation under interactions are useful for optimizing drug delivery. Similar award DMR – 0819762, and the Burroughs Wellcome Fund diffusion rules apply to biological hydrogels that have not under award 1012566. J. W. was supported in part by the been discussed in great detail here. For example, in the National Science Foundation Graduate Research Fellowship vitreous humor, cationic nanoparticles diffuse more slowly under Grant No. 1122374. The content of the information does than anionic nanoparticles and PEGylation enhances not necessarily reflect the position or the policy of the federal – transport.190 192 government, and no official endorsement should be inferred. Chemical selectivity acts on the nanoscale, where small We wish to thank Prof. Alan Grodzinsky at MIT for his valuable variations in the spatial arrangement of charge tune gel feedback and helpful discussions, and Erica Shapiro, Wes binding affinity.131 Furthermore, the relationship between Chen, Tahoura Samad, Caroline Wagner, Kathryn Smith- charge–charge and hydrophobic interactions is not fully Dupont, and Lisa Walker for careful reading of the manuscript. understood, and the chemical heterogeneity of biological gels lends the problem an additional layer of practical uncertainty. These difficulties notwithstanding, the cases of PEGylated – tobramycin132 and charge-reduced lysozyme137 140 suggest that References engineering drugs with hydrogel interactions in mind can improve efficacy. Engineering drug interactions with biological 1 A. Orell, S. Fröls and S.-V. Albers, Annu. Rev. Microbiol., gels may also lead to new functional biocompatible hydrogels, 2013, 67, 337–354. sidestepping the need for complex synthetic formulations. For 2 H.-C. Flemming and J. Wingender, Nat. Rev. Microbiol., example, complexing LL-37 (a CAMP) with a calcium alginate 2010, 8, 623–633. hydrogel reduced the toxicity of LL-37 toward mammalian cells 3 H.-C. Flemming, T. R. Neu and D. J. Wozniak, J. Bacteriol., and allowed for sustained LL-37 release kinetics.193 Covalently 2007, 189, 7945–7947. crosslinked methacrylated bovine submaxillary mucin hydro- 4 C. Strambio-De-Castillia, M. Niepel and M. P. Rout, Nat. gels also showed sustained release kinetics of both polymyxin Rev. Mol. Cell Biol., 2010, 11, 490–501. B (a cationic antibiotic) and paclitaxel (a hydrophobic cancer 5 H. B. Schmidt and D. Görlich, Trends Biochem. Sci., 2016, drug).194 41,46–61. For large particles, polyvalent weak interactions with the gel 6 M. E. V. Johansson, H. Sjövall and G. C. Hansson, Nat. matrix can lead to adhesion and trapping. Synthetic nano- Rev. Gastroenterol. Hepatol., 2013, 10, 352–361. particles, and probably some viruses as well, use net-neutral 7 D. J. Thornton, K. Rousseau and M. A. McGuckin, Annu. hydrophilic coatings to avoid trapping and promote mucus Rev. Physiol., 2008, 70, 459–486. penetration.128 On the other hand, weak interactions may 8 M. R. Knowles and R. C. Boucher, J. Clin. Invest., 2002, facilitate gel penetration through transient crosslink dis- 109, 571–577. solution. This mechanism may occur in the nuclear pore and 9 K. Netsomboon and A. Bernkop-Schnürch, Eur. J. Pharm. Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. reconstituted gels formed from nucleoporin.68 Studies of the Biopharm., 2016, 98,76–89. nuclear pore’s unique transport properties may lead to new 10 S. K. Lai, Y.-Y. Wang and J. Hanes, Adv. Drug Delivery Rev., innovations; for example, nuclear pore-mimicking nanopores 2009, 61, 158–171. have been developed with novel potential biotechnology 11 A. D. Theocharis, S. S. Skandalis, C. Gialeli and applications.195 N. K. Karamanos, Adv. Drug Delivery Rev., 2016, 97,4–27. Given the importance of gel permeability in living systems, 12 C. J. Dowd, C. L. Cooney and M. A. Nugent, J. Biol. Chem., it may serve as a biomarker for disease states, particularly for 1999, 274, 5236–5244. mucosal diseases in which mucus is easily accessible. For 13 A. M. Garcia, N. Szasz, S. B. Trippel, T. I. Morales, example, we showed that cervical mucus stratifies reproductive A. J. Grodzinsky and E. H. Frank, Arch. Biochem. Biophys., states, distinguishing between pregnant women at high and low 2003, 415,69–79. risk for preterm birth via both rheological and permeability- 14 J. Taipale and J. Keski-Oja, FASEB J., 1997, 11,51–59. based assays (ref. 196 and submitted; Fig. 1a depicts the visual 15 L. Zhang, B. S. Gardiner, D. W. Smith, P. Pivonka and difference between mucus from high- and low- risk patients). A. J. Grodzinsky, J. Theor. Biol., 2010, 263,20–29. Overall, understanding particle transport in biological gels will 16 R. G. Thorne, A. Lakkaraju, E. Rodriguez-Boulan and lead to powerful applications and innovations in diagnostics, C. Nicholson, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, biomimetic gels, pharmacology, and nanomedicine. 8416–8421. 17 K. Lewis, Nat. Rev. Microbiol., 2007, 5,48–56. 18 G. Alexandrakis, E. B. Brown, R. T. Tong, T. D. McKee, Acknowledgements R. B. Campbell, Y. Boucher and R. K. Jain, Nat. Med., 2004, 10, 203–207. This work was supported by the Department of Defense, 19 P. A. Netti, D. A. Berk, M. A. Swartz, A. J. Grodzinsky and Defense Threat Reduction Agency, under award HDTRA-13-1- R. K. Jain, Cancer Res., 2000, 60, 2497–2503.

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8091 View Article Online

Feature article Nanoscale

20 V. P. Chauhan, T. Stylianopoulos, Y. Boucher and J. Demeester, T. Coenye and K. Braeckmans, Nanomed., R. K. Jain, Annu. Rev. Chem. Biomol. Eng., 2011, 2,281–298. 2013, 8, 935–949. 21 A. G. Bajpayee, C. R. Wong, M. G. Bawendi, E. H. Frank 45 S. K. Lai, Y.-Y. Wang, K. Hida, R. Cone and J. Hanes, Proc. and A. J. Grodzinsky, Biomaterials, 2014, 35, 538–549. Natl. Acad. Sci. U. S. A., 2010, 107, 598–603. 22 V. P. Chauhan, R. M. Lanning, B. Diop-Frimpong, 46 A. Birjiniuk, N. Billings, E. Nance, J. Hanes, K. Ribbeck W. Mok, E. B. Brown, T. P. Padera, Y. Boucher and and P. S. Doyle, New J. Phys., 2014, 16, 85014. R. K. Jain, Biophys. J., 2009, 97, 330–336. 47 E. A. Nance, G. F. Woodworth, K. A. Sailor, T.-Y. Shih, 23 M. J. Ernsting, M. Murakami, A. Roy and S.-D. Li, Q. Xu, G. Swaminathan, D. Xiang, C. Eberhart and J. Controlled Release, 2013, 172, 782–794. J. Hanes, Sci. Transl. Med., 2012, 4, 149ra119. 24 A. Jackson and W. Gu, Curr. Rheumatol. Rev., 2009, 5,40– 48 M. Wühr, T. Güttler, L. Peshkin, G. C. McAlister, 50. M. Sonnett, K. Ishihara, A. C. Groen, M. Presler, 25 B. S. Schuster, J. S. Suk, G. F. Woodworth and J. Hanes, B. K. Erickson, T. J. Mitchison, M. W. Kirschner and Biomaterials, 2013, 34, 3439–3446. S. P. Gygi, Curr. Biol., 2015, 25, 2663–2671. 26 M. Abdulkarim, N. Agulló, B. Cattoz, P. Griffiths, 49 D. Mohr, S. Frey, T. Fischer, T. Güttler and D. Görlich, A. Bernkop-Schnürch, S. G. Borros and M. Gumbleton, EMBO J., 2009, 28, 2541–2553. Eur. J. Pharm. Biopharm., 2015, 97, 230–238. 50 J. N. Wilking, V. Zaburdaev, M. De Volder, R. Losick, 27 G. A. James, E. Swogger, R. Wolcott, P. Secor, J. Sestrich, M. P. Brenner and D. A. Weitz, Proc. Natl. Acad. Sci. U. S. J. W. Costerton, P. S. Stewart and others, Wound Repair A., 2013, 110, 848–852. Regen., 2008, 16,37–44. 51 S. K. Lai, D. E. O’Hanlon, S. Harrold, S. T. Man, 28 S. Moreau-Marquis, B. A. Stanton and G. A. O’Toole, Pulm. Y.-Y. Wang, R. Cone and J. Hanes, Proc. Natl. Acad. Pharmacol. Ther., 2008, 21, 595–599. Sci. U. S. A., 2007, 104, 1482–1487. 29 L. K. Jennings, K. M. Storek, H. E. Ledvina, C. Coulon, 52 K. Maisel, L. Ensign, M. Reddy, R. Cone and J. Hanes, L. S. Marmont, I. Sadovskaya, P. R. Secor, B. S. Tseng, J. Controlled Release, 2015, 197,48–57. M. Scian, A. Filloux, D. J. Wozniak, P. L. Howell and 53 J. S. Suk, A. J. Kim, K. Trehan, C. S. Schneider, M. R. Parsek, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, L. Cebotaru, O. M. Woodward, N. J. Boylan, M. P. Boyle, 11353–11358. S. K. Lai, W. B. Guggino and J. Hanes, J. Controlled 30 M. S. Byrd, I. Sadovskaya, E. Vinogradov, H. Lu, Release, 2014, 178,8–17. A. B. Sprinkle, S. H. Richardson, L. Ma, B. Ralston, 54 J. Kirch, A. Schneider, B. Abou, A. Hopf, U. F. Schaefer, M. R. Parsek, E. M. Anderson, J. S. Lam and M. Schneider, C. Schall, C. Wagner and C.-M. Lehr, Proc. D. J. Wozniak, Mol. Microbiol., 2009, 73, 622–638. Natl. Acad. Sci. U. S. A., 2012, 109, 18355–18360. 31 A. Ghafoor, I. D. Hay and B. H. A. Rehm, Appl. Environ. 55 T. Stylianopoulos, B. Diop-Frimpong, L. L. Munn and Microbiol., 2011, 77, 5238–5246. R. K. Jain, Biophys. J., 2010, 99, 3119–3128. 32 S. K. Lai, Y.-Y. Wang, D. Wirtz and J. Hanes, Adv. Drug 56 F. Travascio and W. Y. Gu, Ann. Biomed. Eng., 2007, 35, Delivery Rev., 2009, 61,86–100. 1739–1748. Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. 33 R. Bansil and B. S. Turner, Curr. Opin. Colloid Interface 57 P. R. Secor, J. M. Sweere, L. A. Michaels, A. V. Malkovskiy, Sci., 2006, 11, 164–170. D. Lazzareschi, E. Katznelson, J. Rajadas, M. E. Birnbaum, 34 R. C. Boucher, Trends Mol. Med., 2007, 13, 231–240. A. Arrigoni, K. R. Braun, S. P. Evanko, D. A. Stevens, 35 A. Kater, M. O. Henke and B. K. Rubin, Ann. N. Y. Acad. W. Kaminsky, P. K. Singh, W. C. Parks and P. L. Bollyky, Sci., 2007, 1112, 140–153. Cell Host Microbe, 2015, 18, 549–559. 36 I. W. Sutherland, Microbiology, 2001, 147,3–9. 58 A. Godec, M. Bauer and R. Metzler, New J. Phys., 2014, 16, 37 P. Lu, V. M. Weaver and Z. Werb, J. Cell Biol., 2012, 196, 92002. 395–406. 59 P. Georgiades, P. D. A. Pudney, D. J. Thornton and 38 D. J. Thornton, Proc. Am. Thorac. Soc., 2004, 1,54–61. T. A. Waigh, Biopolymers, 2014, 101, 366–377. 39 N. Becher, K. Adams Waldorf, M. Hein and N. Uldbjerg, 60 J. Sprakel, J. van der Gucht, M. A. Cohen Stuart and Acta Obstet. Gynecol. Scand., 2009, 88, 502–513. N. A. M. Besseling, Phys. Rev. Lett., 2007, 99, 208301. 40 O. Lieleg and K. Ribbeck, Trends Cell Biol., 2011, 21, 543– 61 E. B. Stukalin, L.-H. Cai, N. A. Kumar, L. Leibler and 551. M. Rubinstein, Macromolecules, 2013, 46, 7525–7541. 41 A. G. Bajpayee, M. A. Quadir, P. T. Hammond and 62 L.-H. Cai, S. Panyukov and M. Rubinstein, Macromolecules, A. J. Grodzinsky, Osteoarthr. Cartil., 2016, 24,71–81. 2011, 44, 7853–7863. 42 S. C. Chew, B. Kundukad, T. Seviour, J. R. C. van der 63 L.-H. Cai, S. Panyukov and M. Rubinstein, Macromolecules, Maarel, L. Yang, S. A. Rice, P. Doyle and S. Kjelleberg, 2015, 48, 847–862. mBio, 2014, 5, e01536–e01514. 64 C. Müller, G. Perera, V. König and A. Bernkop-Schnürch, 43 J. S. Crater and R. L. Carrier, Macromol. Biosci., 2010, 10, Eur. J. Pharm. Biopharm., 2014, 87, 125–131. 1473–1483. 65 C. Müller, K. Leithner, S. Hauptstein, F. Hintzen, 44 K. Forier, A.-S. Messiaen, K. Raemdonck, H. Deschout, W. Salvenmoser and A. Bernkop-Schnürch, J. Nanopart. J. Rejman, F. De Baets, H. Nelis, S. C. De Smedt, Res., 2012, 15, 1353–1313.

8092 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

66 J. S. Suk, S. K. Lai, N. J. Boylan, M. R. Dawson, M. P. Boyle 91 S. S. Olmsted, J. L. Padgett, A. I. Yudin, K. J. Whaley, and J. Hanes, Nanomed., 2011, 6, 365–375. T. R. Moench and R. A. Cone, Biophys. J., 2001, 81, 1930– 67 M. Dawson, D. Wirtz and J. Hanes, J. Biol. Chem., 2003, 1937. 278, 50393–50401. 92 K. Khanvilkar, M. D. Donovan and D. R. Flanagan, Adv. 68 S. Frey and D. Görlich, Cell, 2007, 130, 512–523. Drug Delivery Rev., 2001, 48, 173–193. 69 S. L. McGill and H. D. C. Smyth, Mol. Pharm., 2010, 7, 93 S. Byun, Y. L. Sinskey, Y. C. S. Lu, E. H. Frank and 2280–2288. A. J. Grodzinsky, Arch. Biochem. Biophys., 2013, 540,1–8. 70 Y.-Y. Wang, S. K. Lai, C. So, C. Schneider, R. Cone and 94 M. Mammen, S.-K. Choi and G. M. Whitesides, Angew. J. Hanes, PLoS One, 2011, 6, e21547. Chem., Int. Ed., 1998, 37, 2754–2794. 71 D. Alfandari, H. Cousin, A. Gaultier, K. Smith, 95 M. Cohen, X.-Q. Zhang, H. P. Senaati, H.-W. Chen, J. M. White, T. Darribère and D. W. DeSimone, Curr. Biol., N. M. Varki, R. T. Schooley and P. Gagneux, Virol. J., 2013, 2001, 11, 918–930. 10, 321. 72 A. Houry, M. Gohar, J. Deschamps, E. Tischenko, 96X.Yang,L.Steukers,K.Forier,R.Xiong,K.Braeckmans, S. Aymerich, A. Gruss and R. Briandet, Proc. Natl. Acad. K. Van Reeth and H. Nauwynck, PLoS One, 2014, 9, e110026. Sci. U. S. A., 2012, 109, 13088–13093. 97 C. Ehre, E. N. Worthington, R. M. Liesman, B. R. Grubb, 73 R. A. Hatch and N. L. Schiller, Antimicrob. Agents D. Barbier, W. K. O’Neal, J.-M. Sallenave, R. J. Pickles and Chemother., 1998, 42, 974–977. R. C. Boucher, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 74 M. Alipour, Z. E. Suntres and A. Omri, J. Antimicrob. 16528–16533. Chemother., 2009, 64, 317–325. 98 R. O. Hynes, Science, 2009, 326, 1216–1219. 75 J. P. Celli, B. S. Turner, N. H. Afdhal, S. Keates, I. Ghiran, 99 T. Rozario and D. W. DeSimone, Dev. Biol., 2010, 341, C. P. Kelly, R. H. Ewoldt, G. H. McKinley, P. So, 126–140. S. Erramilli and R. Bansil, Proc. Natl. Acad. Sci. U. S. A., 100 A. Bernkop-Schnürch, Adv. Drug Delivery Rev., 2005, 57, 2009, 106, 14321–14326. 1569–1582. 76 B. S. Tseng, W. Zhang, J. J. Harrison, T. P. Quach, 101 J. Braga, J. G. McNally and M. Carmo-Fonseca, Biophys. J., J. L. Song, J. Penterman, P. K. Singh, D. L. Chopp, 2007, 92, 2694–2703. A. I. Packman and M. R. Parsek, Environ. Microbiol., 2013, 102 K. Ribbeck and D. Görlich, EMBO J., 2001, 20, 1320–1330. 15, 2865–2878. 103 S. Byun, M. D. Tortorella, A.-M. Malfait, K. Fok, 77 A. Grodzinsky, Forces, Fields and Flows in Biological E. H. Frank and A. J. Grodzinsky, Arch. Biochem. Biophys., Systems, Garland Science, 2011. 2010, 499,32–39. 78 Y. Cu and W. M. Saltzman, Adv. Drug Delivery Rev., 2009, 104 C. A. Gordon, N. A. Hodges and C. Marriott, J. Antimicrob. 61, 101–114. Chemother., 1988, 22, 667–674. 79 N. Billings, A. Birjiniuk, T. S. Samad, P. S. Doyle and 105 L. Li, O. Lieleg, S. Jang, K. Ribbeck and J. Han, Lab Chip, K. Ribbeck, Rep. Prog. Phys., 2015, 78, 36601. 2012, 12, 4071–4079. 80 D. Axelrod, D. E. Koppel, J. Schlessinger, E. Elson and 106 B. E. Hunt, A. Weber, A. Berger, B. Ramsey and A. L. Smith, Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. W. W. Webb, Biophys. J., 1976, 16, 1055–1069. Antimicrob. Agents Chemother., 1995, 39,34–39. 81 A. Lubelski and J. Klafter, Biophys. J., 2009, 96, 2055– 107 B. Cao, L. Christophersen, K. Thomsen, M. Sønderholm, 2063. T. Bjarnsholt, P. Ø. Jensen, N. Høiby and C. Moser, 82 M. J. Saxton, Biophys. J., 2001, 81, 2226–2240. J. Antimicrob. Chemother., 2015, 70, 2057–2063. 83 A. Lubelski and J. Klafter, Biophys. J., 2008, 94, 4646–4653. 108 W. W. Nichols, S. M. Dorrington, M. P. Slack and 84 O. Krichevsky and G. Bonnet, Rep. Prog. Phys., 2002, 65, H. L. Walmsley, Antimicrob. Agents Chemother., 1988, 32, 251–297. 518–523. 85 S. A. McKinley, L. Yao and M. G. Forest, J. Rheol., 2009, 53, 109 W. W. Nichols, M. J. Evans, M. P. Slack and 1487–1506. H. L. Walmsley, J. Gen. Microbiol., 1989, 135, 1291–1303. 86 A. M. Erickson, B. I. Henry, J. M. Murray, P. J. Klasse and 110 J. X. Huang, M. A. Blaskovich, R. Pelingon, S. Ramu, C. N. Angstmann, Biophys. J., 2015, 109, 164–172. A. Kavanagh, A. G. Elliott, M. S. Butler, A. B. Montgomery 87 T. A. Waigh, Rep. Prog. Phys., 2005, 68, 685–742. and M. A. Cooper, Antimicrob. Agents Chemother., 2015, 59, 88 L. M. Ensign, B. C. Tang, Y.-Y. Wang, T. A. Tse, T. Hoen, 5925–5931. R. Cone and J. Hanes, Sci. Transl. Med., 2012, 4, 138ra79. 111 J. N. Anderl, M. J. Franklin and P. S. Stewart, Antimicrob. 89 C. S. Schneider, Q. Xu, N. J. Boylan, J. Chisholm, Agents Chemother., 2000, 44, 1818–1824. B. C. Tang, B. S. Schuster, A. Henning, L. M. Ensign, 112 O. Ciofu, T. Tolker-Nielsen, P. Ø. Jensen, H. Wang and E. Lee, P. Adstamongkonkul, B. W. Simons, S.-Y. S. Wang, N. Høiby, Adv. Drug Delivery Rev., 2015, 85,7–23. X. Gong, T. Yu, M. P. Boyle, J. S. Suk and J. Hanes, Sci. 113 M. C. Walters, F. Roe, A. Bugnicourt, M. J. Franklin and Adv., 2017, 3, e1601556. P. S. Stewart, Antimicrob. Agents Chemother., 2003, 47, 317– 90 R. Shaikh, T. R. R. Singh, M. J. Garland, A. D. Woolfson 323. and R. F. Donnelly, J. Pharm. BioAllied Sci., 2011, 3,89– 114 N. Billings, M. Millan, M. Caldara, R. Rusconi, Y. Tarasova, 100. R. Stocker and K. Ribbeck, PLoS Pathog., 2013, 9, e1003526.

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8093 View Article Online

Feature article Nanoscale

115 H. H. Sigurdsson, J. Kirch and C.-M. Lehr, Int. J. Pharm., 140 C. C. Teneback, T. C. Scanlon, M. J. Wargo, J. L. Bement, 2013, 453,56–64. K. E. Griswold and L. W. Leclair, Antimicrob. Agents 116 M. Benincasa, M. Mattiuzzo, Y. Herasimenka, P. Cescutti, Chemother., 2013, 57, 5559–5564. R. Rizzo and R. Gennaro, J. Pept. Sci., 2009, 15, 595–600. 141 B. Raveh, J. M. Karp, S. Sparks, K. Dutta, M. P. Rout, 117 G. Batoni, G. Maisetta and S. Esin, Biochim. Biophys. Acta, A. Sali and D. Cowburn, Proc. Natl. Acad. Sci. U. S. A., 2016, 1858, 1044–1060. 2016, 113, E2489–E2497. 118 Y. Herasimenka, M. Benincasa, M. Mattiuzzo, P. Cescutti, 142 B. B. Quimby, S. W. Leung, R. Bayliss, M. T. Harreman, R. Gennaro and R. Rizzo, Peptides, 2005, 26, 1127–1132. G. Thirumala, M. Stewart and A. H. Corbett, J. Biol. Chem., 119 C. Chan, L. L. Burrows and C. M. Deber, J. Pept. Res., 2001, 276, 38820–38829. 2005, 65, 343–351. 143 B. Bagchi, P. C. Blainey and X. S. Xie, J. Phys. Chem. B, 120 L. M. Yin, S. Lee, J. S. Mak, A. S. Helmy and C. M. Deber, 2008, 112, 6282–6284. Protein Pept. Lett., 2013, 20, 843–847. 144 P. C. Blainey, A. M. van Oijen, A. Banerjee, G. L. Verdine 121 C. Chan, L. L. Burrows and C. M. Deber, J. Biol. Chem., and X. S. Xie, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 2004, 279, 38749–38754. 5752–5757. 122 H. H. Kuo, C. Chan, L. L. Burrows and C. M. Deber, Chem. 145 P. C. Blainey, G. Luo, S. C. Kou, W. F. Mangel, Biol. Amp Drug Des., 2007, 69, 405–412. G. L. Verdine, B. Bagchi and X. S. Xie, Nat. Struct. Mol. 123 R. Bucki, A. G. Sostarecz, F. J. Byfield, P. B. Savage and Biol., 2009, 16, 1224–1229. P. A. Janmey, J. Antimicrob. Chemother., 2007, 60, 535– 146 A. D. Diamond and J. T. Hsu, Biotechnol. Bioeng., 1989, 34, 545. 1000–1014. 124 R. Bucki, D. B. Namiot, Z. Namiot, P. B. Savage and 147 C. A. Strulson, R. C. Molden, C. D. Keating and P. A. Janmey, J. Antimicrob. Chemother., 2008, 62, 329– P. C. Bevilacqua, Nat. Chem., 2012, 4, 941–946. 335. 148 C. P. Brangwynne, J. Cell Biol., 2013, 203, 875–881. 125 R. Bucki, F. J. Byfield and P. A. Janmey, Eur. Respir. J., 149 P. Li, S. Banjade, H.-C. Cheng, S. Kim, B. Chen, L. Guo, 2007, 29, 624–632. M. Llaguno, J. V. Hollingsworth, D. S. King, S. F. Banani, 126 D. J. Weiner, R. Bucki and P. A. Janmey, Am. J. Respir. Cell P. S. Russo, Q.-X. Jiang, B. T. Nixon and M. K. Rosen, Mol. Biol., 2003, 28, 738–745. Nature, 2012, 483, 336–340. 127 B. Fadnes, O. Rekdal and L. Uhlin-Hansen, BMC Cancer, 150 L. J. Colwell, M. P. Brenner and K. Ribbeck, PLoS Comput. 2009, 9, 183. Biol., 2010, 6, e1000747. 128 R. A. Cone, Adv. Drug Delivery Rev., 2009, 61,75–85. 151 O. Lieleg, I. Vladescu and K. Ribbeck, Biophys. J., 2010, 98, 129 O. Lieleg, R. M. Baumgärtel and A. R. Bausch, Biophys. J., 1782–1789. 2009, 97, 1569–1577. 152 E. Guiot, P. Georges, A. Brun, M. P. Fontaine Aupart, 130 F. Arends, R. Baumgärtel and O. Lieleg, Langmuir ACS M. N. Bellon Fontaine and R. Briandet, Photochem. J. Surf. Colloids, 2013, 29, 15965–15973. Photobiol., 2002, 75, 570–578. 131 L. D. Li, T. Crouzier, A. Sarkar, L. Dunphy, J. Han and 153 Q. Xu, L. M. Ensign, N. J. Boylan, A. Schön, X. Gong, Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. K. Ribbeck, Biophys. J., 2013, 105, 1357–1365. J.-C. Yang, N. W. Lamb, S. Cai, T. Yu, E. Freire and 132 J. Du, H. M. H. N. Bandara, P. Du, H. Huang, K. Hoang, J. Hanes, ACS Nano, 2015, 9, 9217–9227. D. Nguyen, S. V. Mogarala and H. D. C. Smyth, Mol. 154 X. Li, D. Chen, C. Le, C. Zhu, Y. Gan, L. Hovgaard and Pharm., 2015, 12, 1544–1553. M. Yang, Int. J. Nanomed., 2011, 6, 3151–3162. 133 Y. Marcus, K. Sasson, M. Fridkin and Y. Shechter, J. Med. 155 P. Mastorakos, A. L. da Silva, J. Chisholm, E. Song, Chem., 2008, 51, 4300–4305. W. K. Choi, M. P. Boyle, M. M. Morales, J. Hanes and 134 S. Chen, Y. Itoh, T. Masuda, S. Shimizu, J. Zhao, J. Ma, J. S. Suk, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, 8720– S. Nakamura, K. Okuro, H. Noguchi, K. Uosaki and 8725. T. Aida, Science, 2015, 348, 555–559. 156 P. Mastorakos, C. Zhang, S. Berry, Y. Oh, S. Lee, 135 C. D. Ma, C. Wang, C. Acevedo-Vélez, S. H. Gellman and C. G. Eberhart, G. F. Woodworth, J. S. Suk and J. Hanes, N. L. Abbott, Nature, 2015, 517, 347–350. Adv. Healthcare Mater., 2015, 4, 1023–1033. 136 L. K. Sanders, W. Xian, C. Guáqueta, M. J. Strohman, 157 J. S. Suk, Q. Xu, N. Kim, J. Hanes and L. M. Ensign, Adv. C. R. Vrasich, E. Luijten and G. C. L. Wong, Proc. Natl. Drug Delivery Rev., 2016, 99,28–51. Acad. Sci. U. S. A., 2007, 104, 15994–15999. 158 K. Maisel, M. Reddy, Q. Xu, S. Chattopadhyay, R. Cone, 137 T. C. Scanlon, C. C. Teneback, A. Gill, J. L. Bement, L. M. Ensign and J. Hanes, Nanomed., 2016, 11, 1337– J. A. Weiner, J. W. Lamppa, L. W. Leclair and 1343. K. E. Griswold, ACS Chem. Biol., 2010, 5, 809–818. 159 S. K. Lai, K. Hida, S. Shukair, Y.-Y. Wang, A. Figueiredo, 138 A. Gill, T. C. Scanlon, D. C. Osipovitch, D. R. Madden and R. Cone, T. J. Hope and J. Hanes, J. Virol., 2009, 83, K. E. Griswold, PLoS One, 2011, 6, e16788. 11196–11200. 139 K. E. Griswold, J. L. Bement, C. C. Teneback, 160 Y. Y. Wang, A. Kannan, K. L. Nunn, M. A. Murphy, T. C. Scanlon, M. J. Wargo and L. W. Leclair, D. B. Subramani, T. Moench, R. Cone and S. K. Lai, Bioengineered, 2014, 5, 143–147. Mucosal Immunol., 2014, 7, 1036–1044.

8094 | Nanoscale,2017,9,8080–8095 This journal is © The Royal Society of Chemistry 2017 View Article Online

Nanoscale Feature article

161 K. Hida, S. K. Lai, J. S. Suk, S. Y. Won, M. P. Boyle and 178 A. Popov, E. Enlow, J. Bourassa and H. Chen, Nanomed., J. Hanes, PLoS One, 2011, 6, e19919. 2016, 12, 1863–1871. 162 B. S. Schuster, A. J. Kim, J. C. Kays, M. M. Kanzawa, 179 I. Pereira de Sousa, T. Moser, C. Steiner, B. Fichtl and W. B. Guggino, M. P. Boyle, S. M. Rowe, N. Muzyczka, A. Bernkop-Schnürch, Int. J. Pharm., 2016, 500, 236–244. J. S. Suk and J. Hanes, Mol. Ther., 2014, 22, 1484– 180 W. Shan, X. Zhu, W. Tao, Y. Cui, M. Liu, L. Wu, L. Li, 1493. Y. Zheng and Y. Huang, ACS Appl. Mater. Interfaces, 2016, 163 Y.-Y. Wang, D. Harit, D. B. Subramani, H. Arora, 8, 25444–25453. P. A. Kumar and S. K. Lai, Eur. Respir. J., 2016, 1601709. 181 T. Yu, K. W. Y. Chan, A. Anonuevo, X. Song, B. S. Schuster, 164 K. M. Fahrbach, O. Malykhina, D. J. Stieh and T. J. Hope, S. Chattopadhyay, Q. Xu, N. Oskolkov, H. Patel, PLoS One, 2013, 8, e76176. L. M. Ensign, P. C. M. van Zjil, M. T. McMahon and 165 B. M. Gunn, J. R. Schneider, M. Shansab, A. R. Bastian, J. Hanes, Nanomed., 2015, 11, 401–405. K. M. Fahrbach, A. D. Smith, A. E. Mahan, M. M. Karim, 182 S. Bonengel, F. Prüfert, G. Perera, J. Schauer and A. F. Licht, I. Zvonar, J. Tedesco, M. R. Anderson, A. Bernkop-Schnürch, Int. J. Pharm., 2015, 483,19–25. A. Chapel, T. J. Suscovich, D. C. Malaspina, H. Streeck, 183 W. Suchaoin, I. Pereira de Sousa, K. Netsomboon, B. D. Walker, A. Kim, G. Lauer, M. Altfeld, S. Pillai, H. T. Lam, F. Laffleur and A. Bernkop-Schnürch, I. Szleifer, N. L. Kelleher, P. F. Kiser, T. J. Hope and Int. J. Pharm., 2016, 510, 255–262. G. Alter, Mucosal Immunol., 2016, 9, 1549–1558. 184 M. Breeuwer and D. S. Goldfarb, Cell, 1990, 60, 999–1008. 166 Y.-Y. Wang, H. A. Schroeder, K. L. Nunn, K. Woods, 185 Y. Sakiyama, A. Mazur, L. E. Kapinos and R. Y. H. Lim, D. J. Anderson, S. K. Lai and R. A. Cone, PLoS One, 2016, Nat. Nanotechnol., 2016, 11, 719–723. 11, e0158338. 186 R. Y. H. Lim, N.-P. Huang, J. Köser, J. Deng, K. H. A. Lau, 167 A. Chen, S. A. McKinley, S. Wang, F. Shi, P. J. Mucha, K. Schwarz-Herion, B. Fahrenkrog and U. Aebi, Proc. Natl. M. G. Forest and S. K. Lai, Biophys. J., 2014, 106, 2028– Acad. Sci. U. S. A., 2006, 103, 9512–9517. 2036. 187 R. Moussavi-Baygi and M. R. K. Mofrad, Sci. Rep., 2016, 6, 168 T. Wessler, A. Chen, S. A. McKinley, R. Cone, M. G. Forest 29991. and S. K. Lai, ACS Infect. Dis., 2015, 2,82–92. 188 M. P. Rout, J. D. Aitchison, M. O. Magnasco and 169 C. E. Henry, Y.-Y. Wang, Q. Yang, T. Hoang, B. T. Chait, Trends Cell Biol., 2003, 13, 622–628. S. Chattopadhyay, T. Hoen, L. M. Ensign, K. L. Nunn, 189 M. Tagliazucchi, O. Peleg, M. Kröger, Y. Rabin and H. Schroeder, J. McCallen, T. Moench, R. Cone, I. Szleifer, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 3363– S. R. Roffler and S. K. Lai, Acta Biomater., 2016, 43,61–70. 3368. 170 J. J. F. Verhoef, J. F. Carpenter, T. J. Anchordoquy and 190 B. T. Käsdorf, F. Arends and O. Lieleg, Biophys. J., 2015, H. Schellekens, Drug Discovery Today, 2014, 19, 1945– 109, 2171–2181. 1952. 191 Q. Xu, N. J. Boylan, J. S. Suk, Y.-Y. Wang, E. A. Nance, 171 W. Lin, Chem. Rev., 2015, 115, 10407–10409. J.-C. Yang, P. J. McDonnell, R. A. Cone, E. J. Duh and 172 P. Meers, M. Neville, V. Malinin, A. W. Scotto, J. Hanes, J. Controlled Release, 2013, 167,76–84. Published on 16 May 2017. Downloaded by MIT Library 10/07/2017 16:04:33. G. Sardaryan, R. Kurumunda, C. Mackinson, G. James, 192 T. F. Martens, D. Vercauteren, K. Forier, H. Deschout, S. Fisher and W. R. Perkins, J. Antimicrob. Chemother., K. Remaut, R. Paesen, M. Ameloot, J. F. Engbersen, 2008, 61, 859–868. J. Demeester, S. C. De Smedt and K. Braeckmans, 173 C. Wong, T. Stylianopoulos, J. Cui, J. Martin, Nanomed., 2013, 8, 1955–1968. V. P. Chauhan, W. Jiang, Z. Popović, R. K. Jain, 193 M. Toppazzini, A. Coslovi, M. Boschelle, E. Marsich, M. G. Bawendi and D. Fukumura, Proc. Natl. Acad. M. Benincasa, R. Gennaro and S. Paoletti, Carbohydr. Sci. U. S. A., 2011, 108, 2426–2431. Polym., 2011, 83, 578–585. 174 M. Liu, J. Zhang, X. Zhu, W. Shan, L. Li, J. Zhong, 194 C. V. Duffy, L. David and T. Crouzier, Acta Biomater., 2015, Z. Zhang and Y. Huang, J. Controlled Release, 2016, 222, 20,51–59. 67–77. 195 T. Jovanovic-Talisman, J. Tetenbaum-Novatt, 175 K. Forier, K. Raemdonck, S. C. De Smedt, J. Demeester, A. S. McKenney, A. Zilman, R. Peters, M. P. Rout and T. Coenye and K. Braeckmans, J. Controlled Release, 2014, B. T. Chait, Nature, 2009, 457, 1023–1027. 190, 607–623. 196 A. S. Critchfield, G. Yao, A. Jaishankar, R. S. Friedlander, 176 M. Liu, L. Wu, X. Zhu, W. Shan, L. Li, Y. Cui and O. Lieleg, P. S. Doyle, G. McKinley, M. House and Y. Huang, J. Mater. Chem. B, 2016, 4, 5831–5841. K. Ribbeck, PLoS One, 2013, 8, e69528. 177 W. Shan, X. Zhu, M. Liu, L. Li, J. Zhong, W. Sun, Z. Zhang 197 F. Arends, C. Nowald, K. Pflieger, K. Boettcher, S. Zahler and Y. Huang, ACS Nano, 2015, 9, 2345–2356. and O. Lieleg, PLoS One, 2015, 10, e0118090.

This journal is © The Royal Society of Chemistry 2017 Nanoscale,2017,9,8080–8095 | 8095