2.1.

Chemical Modification of Chitosan

Chitosan was chemically modified to improve its solubility and tissue permeation capacity. The conjugation of 4-imidazole acetic acid monohydrochloride (IAA) to the primary amines of chitosan for production of modified chitosan-IAA was performed as previously described.⁵⁰ Briefly, 86% deacetylated chitosan of $130\mathrm{kDa}$ molecular weight (PCL 113, Novamatrix, Norway ) and IAA (Acros Organics, Morris Plains, New Jersey) were dissolved in $0.1\mathrm{M}$ 2-( $N$ -morpholino)ethanesulfonic acid (pH 6.5, Sigma-Aldrich, St. Louis, Missouri ) to a concentration of 1.0% and 2.0% w/v, respectively. A theoretical modification of 50% was targeted for the reaction. IAA and chitosan were combined at a volumetric ratio of 1:5 while on ice. $20\mathrm{M}$ excess (in relation to IAA) of 1-ethyl-3-(3-dimethylaminopropyl carbodiimide hydrochloride (Pierce Biotechnology, Rockford, Illinois ) was added and immediately vortexed to promote addition of IAA to the chitosan backbone. Reactions were allowed to continue overnight with end-over-end mixing. Solutions were dialyzed for $2\text{days}$ against $5\mathrm{mM}$ hydrochloric acid (Acros Organics) thrice and then deionized water thrice using Snakeskin pleated dialysis tubing ( $\mathrm{10,000}\mathrm{MW}$ cut-off; Pierce Biotechnology, Inc.). Following dialysis, the samples were lyophilized for $24\mathrm{h}$ . Degree of substitution was determined with via ${}^1\mathrm{H}$ NMR. Chitosan and chitosan-IAA were resuspended at a stock concentration of 1% w/v in $0.2\mathrm{M}$ sodium acetate (pH 4.5, Sigma Aldrich) immediately prior to dilution for tissue application.

2.2.

Synthesis and Validation of Gold Nanoparticles

Gold nanoparticles were synthesized as untargeted tissue permeability probes. Gold spheres of $9.5\pm 0.35\mathrm{nm}$ and $20.7\pm 0.26\mathrm{nm}$ hydrodynamic diameter were synthesized by a citrate reduction of tetrachloroauric (III) acid ( $\mathrm{H}\mathrm{Au}{\mathrm{Cl}}_4$ , Sigma-Aldrich, St. Louis, Missouri ) under reflux, as previously described.⁵¹ Nanoparticle size was confirmed by dynamic light scattering (DLS) with a ZetaPlus system (Brookhaven Instruments Corporation, Holtsville, New York ). Each set of gold nanoparticles were coated with a fluorescent $5\mathrm{kDa}$ polyethylene glycol (PEG) to reduce their aggregation in media. Briefly, 0.1% w/v fluorescein-PEG-amine (Laysan Bio, Arab, Alabama ) was reacted with $2\mathrm{M}$ excess of 2-iminothiolane (Pierce Biotechnology) for $2\mathrm{h}$ . Upon completion, the solution was filtered at room temperature with Amicon Ultra centrifugation filters (3000 MWCO; Millipore, Billerica, Massachusetts ) at $3220\times \mathrm{g}$ for up to $60\min$ . Gold nanoparticles, suspended at a concentration of $7\times {10}^{10}\text{particles}∕\mathrm{ml}$ (determined by spectral absorbance analysis) were allowed to react with ${10}^{-4}\mathrm{M}$ fluorescein-PEG-thiol at a volumetric ratio of 5:1 for $45\min$ on a shaker for each set of nanoparticles. PEG (Sigma-Aldrich) was subsequently added to a final concentration of 2% w/v. The particles were centrifuged at $1500\times \mathrm{g}$ for $30\min$ and resuspended in deionized water. Gold coating of nanoparticles was confirmed by monitoring the $\sim 494\text{-}\mathrm{nm}$ absorbance peak (absorption maxima of fluorescein on our PEG) using a Synergy HT UV-Vis spectrophotometer (Biotek, Winooski, Vermont ). The final hydrodynamic radius of the gold nanoparticles was measured to be $32.9\pm 1.27\mathrm{nm}$ and $43.9\pm 0.51\mathrm{nm}$ using DLS, demonstrating similar size increase following PEGylation as was previously described.⁵²

2.3.

Synthesis and Validation of Fluorescent Macromolecules

Fluorescent macromolecules were prepared as targeted and untargeted tissue permeability probes. These contrast agents were selected based on size, shape, and charge to provide comparison to commonly known and utilized molecular contrast agents.⁵³ $3\mathrm{kDa}$ rhodamine-dextran, $40\mathrm{kDa}$ fluorescein-dextran, and $150\mathrm{kDa}$ AlexaFluor 647-IgG were purchased from Invitrogen (Carlsbad, California) for use as untargeted probes. For targeted labeling studies, mouse antihuman antibodies specific for epidermal growth factor receptor (EGFR; clone 108; custom synthesized by the Baylor College of Medicine, Houston, Texas ) were reacted with AlexaFluor 647 carboxylic acid succinimidyl esters using commercially available labeling kits (Invitrogen). The purified conjugates were suspended in PBS at a concentration of $1.0\mathrm{mg}∕\mathrm{ml}$ . Dye-labeled isotype controls were synthesized at the same concentrations. Prior to tissue labeling, the bioactivity and specificity of the conjugates was confirmed using live EGFR-positive 1483 cells and EGFR-negative MDA-MB-435S cells as described in Ref. 54, in both the presence and absence of chitosan and chitosan-IAA.

2.4.

Topical Permeation of Fresh Bladder Tissue

Guinea pig bladder mucosa was used as a model tissue to evaluate tissue permeability in the presence and absence of chitosan-based permeation enhancers because of the folding nature of the tissue. The guinea pig bladder allows for effective imaging of both epithelial and stromal layers of the tissue at once. All animals were cared for in accordance with institutional guidelines. The protocols were reviewed and approved by the IACUC at Rice University. This model has been previously validated for the study of topical permeation enhancement.³³ Whole bladders were excised from $3\text{-}\text{to}4\text{-week}$ -old female Hartley guinea pigs (Charles River Laboratories, Wilmington, Massachusetts ) directly following animal sacrifice. The bladders were sectioned into 8 to 10 pieces of uniform surface area $(3\times 3\mathrm{mm})$ using a scalpel and washed once in PBS and then immediately used for permeation studies. Two topical permeation enhancers, chitosan and chitosan-IAA, diluted to 0.01% w/v in DMEM/F12 medium (Invitrogen), were evaluated for their ability to increase tissue permeability. Media alone was used as a negative control. $100\mu \mathrm{L}$ of permeation-enhancing solution was topically applied to the apical surface of each tissue biopsy for $15\min$ .

2.5.

Assessment of Transepithelial Macromolecule Delivery

To determine whether chitosan-treated tissues are selectively permeable to macromolecules of specific sizes, fluorescent macromolecules ranging from $3\mathrm{kDa}\text{to}150\mathrm{kDa}$ in size were topically applied to the tissue surfaces of bladder biopsies. Tissue penetration was then monitored optically. The apical surface of the epithelium was topically treated for $15\min$ with prewarmed $(37°\mathrm{C})$ 0.01% w/v chitosan or chitosan-IAA, washed once in cold media, and covered with a 1:1:1 mixture of $3\mathrm{kDa}$ rhodamine-dextran, $40\mathrm{kDa}$ fluorescein-dextran, and $150\mathrm{kDa}$ AlexaFluor 647-IgG, each diluted to a concentration of $1\mu \mathrm{M}$ in cold media. Tissues were immersed with this solution for $15\min$ at $4°\mathrm{C}$ and then imaged using confocal microscopy at three different excitation wavelengths (described in the following). Images were collected in $2\text{-}\text{to}5\text{-}\mu \mathrm{m}$ steps from the surface into the tissue. Following imaging, the tissues were washed three times in cold media ( $15\min$ total) and reimaged to assess the removal of unbound macromolecules. Experiments were repeated with chitosan-IAA treatment at $25°\mathrm{C}$ or $4°\mathrm{C}$ to determine the influence of temperature on tissue permeation. Each labeling condition was evaluated in six independent experiments.

2.6.

Assessment of Transepithelial Nanoparticle Delivery

To determine whether chitosan-treated tissues permit the transepithelial delivery of nanoparticles, nanoparticles of different sizes were topically applied to the surface of permeabilized bladder biopsies. Nanoparticle delivery was then monitored optically. Briefly, the apical surface of the epithelium was topically treated for $15\min$ with prewarmed $(37°\mathrm{C})$ 0.01% w/v chitosan or chitosan-IAA diluted in DMEM/F12 medium, washed once in cold media, and covered with fluorescein-PEG-gold spheres of 33 or $44\mathrm{nm}$ diameter, each diluted to a concentration of $1\times {10}^6\text{nanoparticles}∕\mathrm{ml}$ in cold media. Tissues were immersed with this solution for $15\min$ at $4°\mathrm{C}$ and then imaged using confocal microscopy. Images were collected in $2\text{-}\text{to}5\text{-}\mu \mathrm{m}$ steps from the surface into the tissue. Following imaging, the tissues were washed three times in cold media ( $15\min$ total) and reimaged to assess the removal of unbound nanoparticles. Each labeling condition was evaluated in six independent experiments.

2.7.

Time-Course Analysis of Tissue Recovery

To monitor the recovery of bladder epithelial barrier function following the removal of chitosan-IAA, fresh bladder biopsies were treated with 0.01% w/v chitosan-IAA or media for $15\min$ at $37°\mathrm{C}$ , washed, and probed with nontargeted contrast agents at regular time intervals during tissue recovery. Briefly, chitosan-IAA and media-treated samples were washed three times with warm or cold media and then allowed to recover in media at $37°\mathrm{C}$ or $4°\mathrm{C}$ . At 0, 15, 30, 60, 90, or $120\min$ after chitosan-IAA removal, the samples were placed on ice and topically labeled with a 1:1:1 mixture of untargeted fluorescent macromolecules ( $3\mathrm{kDA}$ dextran, $40\mathrm{kDa}$ dextran, and IgG antibody) or $44\text{-}\mathrm{nm}$ gold nanoparticles. Samples were allowed to incubate with the macromolecules or nanoparticles for $15\min$ prior to imaging via confocal microscopy. Images were collected $40\mu \mathrm{m}$ below the tissue surface to allow for optical sectioning of both the epithelium and underlying stroma. Tissue samples were assessed in duplicate using a minimum of three independent experiments.

2.8.

Molecular-Specific Labeling of Human Oral Biopsies

To demonstrate the feasibility of delivering molecular-specific optical contrast agents targeted against biomarkers of neoplasia, human oral biopsies were co-treated with fluorescent EGFR antibodies and chitosan-IAA. Briefly, paired clinically normal and abnormal biopsies of the oral mucosa were obtained from five consenting patients at the University of Texas M.D. Anderson Cancer Center (MDACC). Patients gave written informed consent, and the clinical protocols were approved by the Institutional Review Boards at MDACC and Rice University. Biopsies were immediately placed and remained in chilled media until they arrived at Rice University. The biopsies were embedded vertically into 3% w/v ultrapure agarose (Invitrogen) to prevent the influx of permeation enhancers and contrast agents at the tissue margins, as previously described.³³ The apical surface of the epithelium was left free of agarose to facilitate topical labeling. Based on the increased number of epithelial layers in human oral resections, tissues were topically treated with $1\mu \mathrm{M}$ antiEGFR-647 diluted in 0.01% w/v chitosan-IAA for $60\min$ at $37°\mathrm{C}$ , washed three times with cold media, and then sliced transversely into $200\text{-}\mu \mathrm{m}$ -thick slices using a Krumdieck tissue slicer. Tissue slices were counterstained with 0.01% proflavine, a nucleic acid dye, and immediately imaged using confocal microscopy.

2.9.

Confocal Image Acquisition

All images were obtained using a Carl Zeiss LSM 510 confocal microscope (Thornwood, New York) equipped with $488\text{-}\mathrm{nm}$ $\left(30\text{-}\mathrm{mW}\right)$ , $543\text{-}\mathrm{nm}$ $\left(1\text{-}\mathrm{mW}\right)$ , and $633\text{-}\mathrm{nm}$ $\left(5\text{-}\mathrm{mW}\right)$ lasers. Images were collected using photo multiplier tube (PMT) detectors and Zeiss LSM 5 image examiner software. Samples were sequentially excited with each laser line with power settings held constant for each laser. Fluorescence emission was collected using $500\text{-}\text{to}530\text{-}\mathrm{nm}$ ( $40\mathrm{kDa}$ fluorescein-dextran), $565\text{-}\text{to}615\text{-}\mathrm{nm}$ ( $3\mathrm{kDa}$ rhodamine-dextran), and $650\text{-}\text{to}710\text{-}\mathrm{nm}$ ( $150\mathrm{kDa}$ Alexa647-IgG) bandpass filters. Tissue reflectance at $633\mathrm{nm}$ was collected using a $635\text{-}\mathrm{nm}$ dichroic beamsplitter. Images were acquired at $0.5\text{frames}\text{per}\text{second}$ using a $20\times$ objective with a pinhole of 2.56 Airy units. For the untargeted macromolecule and nanoparticle permeation studies, the gain was held constant with excitation at 488, 543, and $633\mathrm{nm}$ . At the gains utilized, the fluorescence of solutions outside the tissue was oversaturated, but the fluorescence of solutions within the tissue was not. In the EGFR targeting studies, the gain was held constant for antiEGFR-647 imaging ( $633\text{-}\mathrm{nm}$ excitation) and proflavine imaging ( $488\text{-}\mathrm{nm}$ excitation).

2.J.

Image Analysis

To quantify the fluorescence intensity of the stroma as a function of recovery time and temperature, representative confocal fluorescence images were analyzed using Image J v1.38 (NIH, Bethesda, Maryland) . The mean fluorescence intensity per unit area was determined by selecting regions of interest (ROIs) within each image and then dividing the measured fluorescence intensity by the area of the ROI. The ROI margins were drawn just inside the stroma border as determined from reflectance overlay images. The entire stroma was included for each image. When images contained more than one region of discontinuous stroma, the mean intensity for the regions was averaged and treated as a single value. Background noise was determined at each time-point from samples pretreated with media in place of chitosan-IAA. The normalized mean fluorescence intensity of the stroma was calculated by subtracting the measured background and normalizing for the fluorescence intensity of the stroma at time of chitosan-IAA removal. The normalized mean fluorescence intensity of the stroma was assessed for each macromolecule using five representative images from three independent experiments (15 images total per macromolecule per time-point and temperature).

The mean reflectance intensity per unit area of the stroma in nanoparticle labeling studies was quantified in the same manner as described for the fluorescence studies, using five representative images from five independent experiments (25 images total per labeling condition).

To quantify the fluorescence intensity of EGFR labeling in transverse images in human oral biopsies, representative confocal fluorescence images were analyzed. In normal samples, ROIs were bounded by the apical and basal surfaces of the epithelium, thereby including any labeling throughout the epithelium. In neoplastic samples, ROIs were drawn to include all squamous cells from the surface of the tissue to the lower margin of EGFR labeling. Regions of fibroblasts, as identified morphologically by proflavine counterstaining, were excluded from the analyses. The mean fluorescence intensity per unit area was assessed using 15 images per biopsy, consisting of three representative images from five separate tissue slices. No background corrections were made because the measured background was $<1\mathrm{\%}$ of the measured mean EGFR labeling intensity for all images evaluated. Differences in mean labeling intensity between normal and neoplastic samples were assessed on a patient-to-patient basis using a two-tailed, unpaired Student’s $t$ -test, with $p$ -values of $⩽0.01$ being considered statistically significant. The relative contrast ratio was determined by dividing the mean fluorescence intensity of each cancer biopsy by that of the contralateral normal.

3.1.

Synthesis and Characterization of Imidazole-Modified Chitosan

A schematic for the synthesis of chitosan-IAA is shown in Fig. 1 . We successfully introduced imidazole acetic acid to the primary amines of chitosan via a carbodiimide-mediated reaction. Following synthesis, ${}^1\mathrm{H}$ NME characterization of the purified polymer provided on average a degree of substitution of about 3.0% of the primary amines modified for each chitosan molecule. Chitosan-IAA polysaccharide prepared through this type of reaction has been previously shown to provide enhanced polymer solubility and buffering capacity.⁵⁰

Fig. 1

Chemical schematic for the synthesis of imidazole-modified chitosan (chitosan-IAA).

3.2.

Transepithelial Macromolecule Delivery through Fresh Bladder Mucosa

Figure 2 shows representative confocal fluorescence images of bladder tissue treated topically with 0.01% w/v chitosan-IAA, chitosan, or media for $15\min$ at $37°\mathrm{C}$ followed by topical application of a 1:1:1 mixture of $3\mathrm{kDa}$ rhodamine-dextran, $40\mathrm{kDa}$ fluorescein-dextran, and $150\mathrm{kDa}$ Alexa647-IgG. The tissue reflectance images are shown on the left, and the corresponding fluorescence images are shown to the right. The yellow/white lines indicate the boundary between the stroma and the epithelium. Due to the three-dimensional (3-D) folding of the resected bladder, it was possible to image the epithelium in cross section using confocal microscopy at a depth of $40\mu \mathrm{m}$ . In the reflectance images, the epithelium was distinguished from the stroma by its darker appearance. Both chitosan and chitosan-IAA were found to facilitate the transepithelial delivery of macromolecules. Macromolecules of all three sizes accumulated in the stroma of permeation-enhanced tissues, causing the stroma to appear brighter than the epithelium in the fluorescence images. Chitosan-IAA treatment facilitated more intense stromal accumulation than nonmodified chitosan for all sizes of macromolecules evaluated. The macromolecules could be removed with several brief washes in media (data not shown), demonstrating that macromolecule accumulation is reversible. Experiments were independently repeated by labeling tissues with one macromolecule at a time (data not shown) to exclude macromolecule interactions. In media-treated controls, the macromolecule penetration was limited to the superficial epithelium. No significant macromolecule penetration was observed following chitosan-IAA or chitosan treatment at $25°\mathrm{C}$ or $4°\mathrm{C}$ (data not shown).

Fig. 2

Stromal accumulation of fluorescent macromolecules following topical permeation treatment. Freshly excised bladder tissue was treated for $15\min$ with 0.01% w/v chitosan-IAA, chitosan, or media and then probed with a 1:1:1 mixture of $3\mathrm{kDa}$ rhodamine-dextran, $40\mathrm{kDa}$ fluorescein-dextran, and Alexa 647-IgG. The yellow/white lines indicate the boundary between the epithelium and stroma. In reflectance images, the epithelium was distinguished from the stroma by its darker appearance. Following permeation treatment, the fluorescent macromolecules accumulated in the stroma. Stromal accumulation was brighter following treatment with imidazole-modified chitosan than with nonmodified chitosan. In media-treated controls, macromolecule penetration was limited to a few cells in the superficial epithelium. The scale bar represents $100\mu \mathrm{m}$ . The epithelium (E) and stroma (S) are labeled in the first image. (Color online only.)

3.3.

Transepithelial Nanoparticle Delivery through Fresh Bladder Mucosa

Figure 3 shows representative confocal fluorescence images of bladder tissue treated topically with 0.01% w/v chitosan-IAA, chitosan, or media for $15\min$ $37°\mathrm{C}$ and then probed with fluorescein-PEG-gold spheres of 44 or $33\mathrm{nm}$ diameter. Both chitosan and chitosan-IAA were found to facilitate the transepithelial delivery of nanoparticles. Following nanoparticle application, fluorescence was primarily observed in the stroma of permeable tissues. Stromal labeling was generally less uniform than observed with macromolecules. Pretreatment with chitosan-IAA resulted in more intense stromal accumulation than nonmodified chitosan. This difference was more obvious following application of larger nanoparticles. The stromal reflectance was visibly enhanced in samples demonstrating nanoparticle uptake. Tissues exposed to nanoparticles showed a stromal reflectance enhancement of 1.5 to 1.7 ( $44\mathrm{nm}$ gold) and 1.7 to 2.1 ( $33\mathrm{nm}$ gold) compared to tissues treated with media in place of chitosan following quantification of reflectance at a fixed gain and regions of the stroma. Minimal to no nanoparticle accumulation was observed in media-treated controls.

Fig. 3

Stromal accumulation of gold nanoparticles following topical permeation treatment. Freshly excised bladder tissue was treated for $15\min$ with 0.01% w/v chitosan-IAA, chitosan, or media, and then fluorescein-PEG-gold spheres of 44- and $33\text{-}\mathrm{nm}$ diameter were applied topically. Transepithelial delivery of both sizes of nanoparticles was observed by monitoring for the localization of nanoparticle-associated fluorescence. The stromal reflectance was visibly enhanced in samples showing accumulation of nanoparticles. No significant change in stromal fluorescence or reflectance was observed in media-treated controls. The scale bar represents $100\mu \mathrm{m}$ .

3.4.

Route of Contrast Agent Delivery Following Chitosan-IAA Treatment

The route of contrast agent delivery through the epithelium was optically interrogated in fresh bladder tissues pretreated with chitosan-IAA. Confocal fluorescence images were collected in $2\text{-}\mu \mathrm{m}$ steps from the surface of the tissue following the topical application of $3\mathrm{kDa}$ rhodamine-dextran or $44\mathrm{nm}$ fluorescein-PEG-gold. Representative videos are available online as Video 1 ( $3\mathrm{kDa}$ dextran) and Video 2 ( $44\mathrm{nm}$ gold). The videos start at $2\mu \mathrm{m}$ below the surface and advance at rate of $2\mu \mathrm{m}∕\mathrm{s}$ . The stills shown for Video 1 and Video 2 represent frames collected at $10\mu \mathrm{m}$ below the tissue surface. $3\mathrm{kDa}$ rhodamine-dextran labeling, shown in Video 1, was characterized by ringlike fluorescence surrounding each cell in the field of view. The labeling appeared extracellular, suggesting a paracellular route of dextran delivery. These rings became progressively smaller in diameter with increasing depth in the epithelium, correlating well with known bladder epithelium morphology. In contrast, the delivery of larger particles, shown in Video 2, appeared to follow both paracellular and transcellular routes. Fluorescein-PEG-gold delivered in a paracellular manner appeared as ringlike labeling that became progressively smaller with increasing depth. Fluorescence associated with nanoparticles was also observed in the cytoplasm of cells across the full thickness of the epithelium, suggestive of transcellular transport. Individual nuclei appeared as a black spot within each cell, suggesting that the transcellular movement of nanoparticles was limited to the cytoplasmic compartment. In both samples, the transition from epithelium to stroma (first visible at $14\text{to}20\mu \mathrm{m}$ ) was characterized by a relative increase in fluorescence. Dextran and nanoparticle accumulation in the stroma was generally uniform. Few morphological features were defined in stroma, other than blood vessels of horizontal and vertical orientation (see Video 1). These vessels were characterized by low-contrast agent accumulation, appearing as dark stripes and circles/ovals, respectively. Similar routes of transport for these two contrast agent sizes were seen following pretreatment of tissues with nonmodified chitosan (data not shown).

Optical sectioning of bladder epithelium and stroma following the topical application of $3\mathrm{kDa}$ rhodamine-dextran. The mucosal surface was pretreated with chitosan-IAA30 for $15\min$ at $37°\mathrm{C}$ . Confocal fluorescence images were acquired parallel to the tissue surface in $2\text{-}\mu \mathrm{m}$ steps with constant laser power and gain. Rhodamine-dextran permeation through the epithelium of the tissue appears to follows a paracellular route, appearing as bright rings around cells and minimal labeling within the cells. As the imaging progresses deeper, shown in the video, the stroma appears, clearly defined by a selective accumulation of contrast agents. Representative images, taken at a depth of $10\mu \mathrm{m}$ , are shown here. The scale bar represents $100\mu \mathrm{m}$ (QuickTime, $3.4\mathrm{MG}$ ). .

Optical sectioning of bladder epithelium and stroma following the topical application of $44\mathrm{nm}$ fluorescein-PEG-gold. The mucosal surface was pretreated with chitosan-IAA30 for $15\min$ at $37°\mathrm{C}$ . Confocal fluorescence images were acquired parallel to the tissue surface in $2\text{-}\mu \mathrm{m}$ steps with constant laser power and gain. Representative images, taken at a depth of $10\mu \mathrm{m}$ , are shown here. Nanoparticle transport in the epithelium appears to follow both paracellular and transcellular routes, appearing as bright rings around cells and diffuse labeling within cells. Individual nuclei appear as a black spot within each cell. Deeper imaging within the tissue samples, seen in the video, demonstrates clear nanoparticle uptake throughout the stromal regions. The scale bar represents $100\mu \mathrm{m}$ (QuickTime, $3.3\mathrm{MG}$ ). .

3.5.

Epithelial Recovery Following Chitosan-IAA Treatment

Figure 4 shows tissue permeability as a function of time and temperature following the removal of chitosan-IAA. Tissue recovery at $37°\mathrm{C}$ is illustrated on the left, and tissue recovery at $4°\mathrm{C}$ is illustrated on the right. Samples were pretreated with 0.01% chitosan-IAA for $15\min$ and washed in warm or cold media, and then macromolecules or nanoparticles were applied topically at different time-points. The mean fluorescence intensity of the stroma was measured at each time-point to assess epithelial permeability. At $37°\mathrm{C}$ , the barrier function of the epithelium was found to be recovered quickly, on the order of minutes to hours depending on the size of the permeability probe. The transepithelial delivery of $44\text{-}\mathrm{nm}$ nanoparticles was reduced by over 90% within $15\min$ of chitosan-IAA removal. By $90\min$ , the epithelium remained permeable ( $\sim 22\mathrm{\%}$ of initial) only to molecules of $3\mathrm{kDa}$ in size. With $2\mathrm{h}$ of recovery time, the transepithelial delivery of $3\mathrm{kDa}$ molecules was reduced by 95%. Little tissue recovery was observed for samples held at $4°\mathrm{C}$ . Macromolecules and nanoparticles of all sizes continued to accumulate in the stroma at all time-points evaluated. The ability of the macromolecules and nanoparticles to reach the stroma was reduced by only 10% over the course of $2\mathrm{h}$ . No significant differences were observed between macromolecules or nanoparticles of different sizes at $4°\mathrm{C}$ . Pretreated tissue that was not washed and that was kept at $37°\mathrm{C}$ still retained macromolecule permeation for all sizes for up to $2\mathrm{h}$ (data not shown).

Fig. 4

Fluorescence intensity as a function of time and temperature. Tissues pretreated with chitosan-IAA were washed and allowed to recover at $37°\mathrm{C}$ (left) or $4°\mathrm{C}$ (right). The recovery of barrier function was probed at regular time intervals using fluorescent macromolecules and nanoparticles. Tissues held at $37°\mathrm{C}$ rapidly recovered barrier function, while tissues held at $4°\mathrm{C}$ displayed continued permeability as evidenced by stromal fluorescence. The ability of the recovering epithelium to block contrast agents was size dependent, with larger molecules excluded more rapidly than smaller molecules.

3.6.

Delivery of Targeted Contrast Agents for Cancer Detection

To demonstrate the clinical potential of chitosan-IAA to aid in delivery of targeted optical contrast agents for the early detection of cancer biomarkers, paired human oral biopsies were topically labeled with fluorescent EGFR-specific antibodies diluted in 0.01% w/v chitosan-IAA. We have previously demonstrated that antibodies targeted to EGFR lack the ability to penetrate normal oral mucosa and squamous carcinomas.³³ Figure 5 shows representative transverse sections of normal (top) and cancerous (bottom) gingiva tissue labeled for EGFR and counterstained with the nucleic acid dye proflavine. Labeling with antiEGFR-647 is displayed in red and proflavine in green. Differential EGFR labeling was observed between normal and cancer samples from all five patients evaluated. EGFR labeling in normal epithelium was limited to cells in the basal layer. This labeling was observed in all normal samples evaluated, and was generally less intense and more diffuse than labeling in cancer samples. Nonspecific antibody labeling was observed in keratinized tissues, appearing as a bright stripe above the epithelium. EGFR labeling in squamous cell carcinoma tissue was characterized by intense, ringlike labeling extending downward from the tissue surface. This labeling appeared highly uniform, labeling all squamous cells within the treatment zone. Regions of fibroblasts, which were identified by the presence of small, widely spaced nuclei using proflavine staining, did not show significant EGFR labeling. With $1\mathrm{h}$ of chitosan-IAA treatment, antibody labeling reached a depth of approximately $150\text{to}250\mu \mathrm{m}$ in cancer tissue and $⩽400\mu \mathrm{m}$ in normal epithelium.

Fig. 5

Representative confocal images of antiEGFR-647 labeling in fresh human oral biopsies co-treated with chitosan-IAA, collected at the same gain. Biopsies were topically treated with antiEGFR-647 diluted in 0.01% w/v chitosan-IAA for $1\mathrm{h}$ at $37°\mathrm{C}$ , washed three times, sliced transversely, and imaged. In normal tissues (top), EGFR labeling is limited to the basal layer of epithelial cells, appearing as a faint region of labeling at the base of the epithelium. Nonspecific labeling is observed at the apical surface of keratinized tissues such as the retromolar trigone tissue shown below. EGFR labeling in squamous cell carcinoma (bottom) is characterized by bright, ring-like extracellular labeling extending downward from the tissue surface. The scale bar represents $100\mu \mathrm{m}$ .

Figure 6 shows the mean fluorescence intensity of paired oral biopsies collected from five patients. The biopsy pairs were labeled and imaged as described earlier. The mean fluorescence intensity of EGFR labeling in normal biopsies was generally low and showed little variation from patient to patient, despite four different anatomic sites. The intensity of EGFR labeling in squamous cell carcinoma biopsies showed a statistically significant $(p⩽0.01)$ difference when compared to their respective contralateral controls. The relative contrast ratio of cancer-to-normal tissue was found to be 8.1, 10.4, 14.0, 13.3, and 9.5 for patients 1 to 5, respectively.

Fig. 6

Measured mean fluorescence intensity of paired oral biopsies collected from patients with cancer of the buccal (1), retromolar trigon (2), maxilla (3, 4), and tongue (5) regions. Biopsies were collected from clinically abnormal and contralateral normal regions. The biopsy pairs were topically labeled for EGFR in the presence of chitosan-IAA and imaged transversely. Statistically significant differences $(P⩽0.01)$ were observed in mean fluorescence between biopsy pairs.