Tumor-Mesoporous Silica Nanoparticle Interactions Following Intraperitoneal Delivery for Targeting Peritoneal Metastasis

Abstract

The use of intraperitoneal administration of nanoparticles has been reported to facilitate higher concentrations of nanoparticles in metastatic peritoneal tumors. While this strategy is appealing for limiting systemic exposure of nanocarrier delivered toxic cargoes and increasing nanoparticle concentrations in avascular peritoneal tumors, little is known about the mechanism of nanoparticle accumulation on tumor tissues and currently, no nanoparticle-based product has been approved for intraperitoneal delivery. Here, we investigated the nanoparticle-specific characteristics that led to increased peritoneal tumor accumulation using MCM-41 type mesoporous silica nanoparticles as our model system. We also investigated the components of the peritoneal tumor stroma that facilitated nanoparticle-tumor interaction. The tumor extracellular matrix is the main factor driving these interactions, specifically the interaction of nanoparticles with collagen. Upon disruption of the collagen matrix, nanoparticle accumulation was reduced by 50%. It is also notable that the incorporation of targeting ligands did not increase overall tumor accumulation in vivo while it significantly increased nanoparticle accumulation in vitro. The use of other particle chemistries did not grossly affect the tumor targetability, but additional concerns arose when those tested particles exhibited significant systemic exposure. Mesoporous silica nanoparticles are advantageous for intraperitoneal administration for the treatment of peritoneal metastasis due to their physical stability, tumor targetability, strong interaction with the collagen matrix, and extended peritoneal residence time. Maximizing nanoparticle interaction with the tumor extracellular matrix is critical for developing strategies to deliver emerging therapeutics for peritoneal cancer treatment using nanocarriers.

INTRODUCTION

Nanocarrier technologies have been employed to deliver therapeutic agents to solid tumors in order to decrease toxicity while also increasing the efficiency of delivery^1–3. The mechanism by which nanoparticles deliver therapeutic payloads to solid tumors is still not fully understood. Recent attempts have been made to reevaluate the ideas surrounding the enhanced permeation and retention (EPR) effect following intravenous (IV) administration of nanoparticles⁴. The revelation that nanoparticle accumulation in solid tumors following IV administration is an active trans-endothelial process instead of a passive process of extravasation through large gaps in the blood vessels greatly affects how nanocarriers should be developed in the future⁴. While our mechanistic understanding of nanoparticle-based drug delivery is still developing, the optimism within the research community is still apparent due to the constant growth witnessed in recent years⁵. Revealing the underlying mechanism of nanoparticle accumulation in solid tumors is still critical for developing effective delivery systems for tumor-specific treatments. This is especially true for the poorly studied intraperitoneal (IP) delivery of nanoparticles for the treatment of peritoneal metastasis, in which nanoparticles are targeted to distribute within the peritoneal fluid, improve the retention in the peritoneal cavity and improve the accumulation on the surface of tumor tissues.

Intraperitoneal delivery has shown many benefits when compared to intravenous delivery when treating tumors that have metastasized to the peritoneal cavity^6–9. Attempts have been made to develop nanoparticle-based drug delivery systems to supplant the use of various therapeutic agents for the local treatment of peritoneal metastasis by increasing peritoneal retention time of chemotherapeutics and radiotherapeutics, reducing systemic leakage, increasing tumor specificity, and limiting the immune response. Expansile polymeric nanoparticles were shown to accumulate on peritoneal tumors and, following a biological trigger, release paclitaxel specifically in the solid tumor^10–13. In addition to modifying the release of the therapeutic payload, attempts have been made to increase peritoneal tumor affinity through the modification of the nanoparticle surface with targeting ligands.^14–19 Folate is the most commonly used targeting ligand when developing nanoparticle systems for the treatment of peritoneal metastasis of ovarian cancers^{17, 18, 20}. This has been met with varying levels of success, including the near doubling of the overall accumulation of various nanoparticle compositions. We previously reported high tumor selectivity of 80–100 nm mesoporous silica nanoparticles (MSNs) following intraperitoneal administration without the use of extensive nanoparticle modification^{7, 21}. The tumor-specific accumulation of nanoparticles upon IP administration could reach 82% of the injected dose per gram of tissue compared to the less than 1% of injected dose delivered to tumors through the IV route²². Along with the observed peritoneal tumor selectivity, our previous report²¹ detailed limited systemic exposure and long peritoneal residence times of MSNs following intraperitoneal administration. This was evidenced by no noticeable MSN accumulation in the heart and lungs, while the majority of the injected dose was found in peritoneal tumors one week after intraperitoneal administration. This data is corroborated by the reported slow degradation of MSNs with small pore sizes at the lower end of the mesopore range²³ and limited systemic exposure of particles greater than 50 nm in diameter⁸. However, the mechanism of tumor selectivity was unclear, which prompted a subsequent investigation into the nanoparticle-specific properties and tumor microenvironment factors that led to the observed, specific tumor targeting.

Current knowledge suggests that the surface accumulation of nanoparticles on tumor tissues is primarily passive in nature, but the potential of an active process involving the use of tumor associated macrophages (TAM) in the peritoneal fluid trafficking the particles to the tumor site has not been thoroughly studied. Once the nanoparticles reach the tumor surface, favorable interactions with the tumor extracellular matrix (ECM) could increase total accumulation of the nanocarrier in the tumor tissue. The ECM is primarily comprised of collagen and fibronectin which form a scaffolding network that supports the tumor structure²⁴. Integrins are a large family of adhesion molecules that anchor tumor cells to the ECM and are important contributors to cancer metastasis²⁵. Understanding how nanoparticles interact with these ECM components following IP administration will dictate the formulation process for these therapeutics in the future.

The goal of this work was to understand the mechanistic and nanoparticle-specific factors facilitating the selective accumulation of nanoparticles on the surface of peritoneal tumors following IP administration. A MCM-41 type mesoporous silica nanoparticle system was used to study the specific interactions with nanoparticles and the surface of peritoneal tumors. Particles of varying chemistries were used to uncover other differences in tumor accumulation. Athymic nude mice bearing peritoneal OVCAR-8 tumors were used to understand nanoparticle accumulation and the potential TAM trafficking of MSNs. Athymic nude mice bearing MIA-PaCa-2 tumors were used to investigate nanoparticle accumulation changes based on a different tumor model. A 3D tumor spheroid model was used to further investigate differences between in vitro and in vivo nanoparticle accumulation. The controlled tumor microenvironment using 3D tumor spheroids was altered to determine what characteristics of the tumor microenvironment may influence this selective, passive accumulation in vivo. This investigation included the effects of extracellular collagen disruption on MSN accumulation and integrin-nanoparticle co-localization of MSNs in the spheroid.

MATERIALS AND METHODS

Materials:

MCM-41 type Mesoporous Silica Nanoparticles and polystyrene nanoparticles were purchased from Sigma-Aldrich (St. Louis, MO, USA). (3-Aminopropyl)triethoxysilane (APTES), Dicyclohexylcarbodiimide (DCCI), triethylamine (TEA), dimethylsulfoxide (DMSO), dioxane and dimethyl ether were purchased from Sigma-Aldrich (St. Louis, MO, USA). Athymic Nude-Foxn1nu mice were purchased from Envigo (South, Easton, MA, USA). OVCAR-8 cell line was purchased from the National Cancer Institute (Frederick, MD, USA). 1.5% (w/v) agarose, RPMI 1640 cell culture media, and 0.25% (w/v) trypsin 0.03% (w/v)-EDTA solution were purchased from American Type Culture Collection (Rockville, MD, USA). Fetal bovine serum (FBS) was purchased from Atlanta Biologicals (Norcross, GA, USA). Bovine serum albumin (BSA), Goat serum, Triton X-100, and Clostridium Histolyticum (Collagenase) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Anti-Integrin alpha V beta 3 antibody, Anti-Collagen I antibody, Rabbit anti-mouse CD68 polyclonal antibody, Goat anti-rabbit polyclonal antibody Alexa 647, Gold antifade mounting medium and Recombinant Rabbit monoclonal Anti-Human IgG H&L (Alexa Fluor® 647) secondary antibody were purchased from Abcam (Cambridge, Massachusetts, USA). Iron-doped hydroxyapatite (HA-Fe) nanoparticles were gifted from Dr. Mei Wei at the University of Connecticut (Department of Materials Science, Storrs, CT, USA). Fluorescein isothiocyanate (FITC) and paraformaldehyde (PFA) was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Cyanine5.5 NHS (Cy5.5) ester was purchased from Lumiprobe (Hunt Valley, MD, USA).

Preparation of Functionalized MSN and Polymeric Nanoparticles

MSN-FITC and MSN-Cy5.5:

MCM-41 type mesoporous silica nanoparticles (MSN) (~250 nm) were purchased from Sigma-Aldrich. Both the FITC and Cy5.5 dyes were loaded similarly. Four milligrams of the selected dye was weighed and added to a solution of 1.50 mL of APTS and 3 mL of ethanol. The mixture was stirred for 24 hours and subsequently added to 700 mg of MSNs suspended in 2.5 mL of ethanol. The whole suspension was then stirred for 48 hours, washed with water and vacuum dried to obtain MSN-FITC and MSN-Cy5.5.

MSN-Cy5.5-FA:

(3-Aminopropyl)triethoxysilane (APTES) (1.5 mL) was dissolved in 3 mL of ethanol at room temperature. One hundred and twenty milligrams of folic acid was dissolved in 3 mL of DMSO and 0.06 mL of triethylamine. This folic acid solution was added to 400 mg of MSN-Cy5.5 along with 0.12 mL of Dicyclohexylcarbodiimide (DCCI) and stirred for 24 hrs at 40°C. The resulting suspension was washed with dimethylsulfoxide (DMSO), dioxane, dimethyl ether and water. The resultant particles were then dried using vacuum. To confirm the presence of folic acid, we tracked the color change of the dried particles from white to yellow and measured absorbance at 280 nm using an UV-vis spectrophotometer (Shiamadzu, Japan).

MSN-Cy5.5-PEG:

Eight milligrams of the dried MSN-Cy5.5 described in the previous section was resuspended in 3 mL of absolute ethanol. To this suspension, 16 mg of mPEG-NHS was added and spun slowly overnight under nitrogen. The resultant MSN-Cy5.5-PEG was washed with ethanol and water. The particles were dried with vacuum. PEG conjugation was determined using a Nicolet Magna 560 Fourier Transform Spectrometer (Nicolet Instrument Corporation, Wisconsin, USA) to represent the stretching vibration (2939 and 2879 cm⁻¹) and bending (1560 and 1664 cm⁻¹) of the CH₂ and CH₃ bonds. PEG coverage was also confirmed with a decrease in MSN surface area as measured using an Autosorb IQ Gas Sorption apparatus (Quantachrome Instruments, Florida, USA) using nitrogen as the analysis gas with a degassing condition of 150°C for 6 hours.

Polymer nanoparticles (Poly-Cy5.5) were prepared according to our previous publication²⁶. Briefly, norbornene-functionalized monomers P(NBCh9-b-NBPEG) and Cy5.5 were dissolved in DMF. The resulting solution was dialyzed against distilled water for 48 h and then filtered through a 0.45 μm membrane.

Mesoporous silica nanoparticles characterization:

Particle size was measured using the Nanosight Nanoparticle Tracking Analysis (NTA) system NS500 (Malvern, Worcestrshire, UK) using a detection threshold of 5 and particle concentration of 3.06×10⁹ particles/mL. TEM images were taken with the FEI Tecnai T12 S (FEI, Oregan, USA) instrument at a voltage of 120 kV to confirm the nanoparticle morphology and pore structure. All particle sizing techniques were evaluated using deionized water as the suspension media and probe sonication (20% amplitude, 1 minute) to fully suspend the powdered material. To understand the potential impact of protein adhesion when injecting the MSNs to the peritoneal cavity, the particle sizing and zeta potential measurements were also carried out on particles that were incubated in RPMI supplemented with 10% fetal bovine serum for one hour prior to characterization.

3D Tumor Spheroid Culture:

An OVCAR-8 ovarian adenocarcinoma cell line was cultured into multicellular spheroids using a liquid overlay method described by Friedrich et al. ²⁷ Two thousand cells were seeded in each well of a 96 well-plate coated with 50 μL of 1.5% agarose and 200 μL of RPMI 1640 with 10% FBS. Media was changed every 2 days by removing 100 μL of the 200 μL of media in each well and replacing it twice with fresh RPMI. After 12 days of growth, the spheroids were then transferred to an agarose coated plate for interaction experiments.

Collagenase Treatment for ECM Disruption

Collagen distribution on the spheroids was determined by incubating 12-day-old spheroids with a Collagen I antibody at a concentration of 1:1500 for 2 hours after blocking with 2% BSA in PBS. A fluorescent secondary antibody incubated at a concentration of 1:500 in blocking solution was used to visualize the 3D distribution of collagen on spheroids by taking optical sections of the spheroid tumor using confocal microscopy and subsequently creating a max projection image using the 3D slicer plugin in ImageJ. To determine the effects of collagenase treatment on nanoparticle adhesion, spheroids were incubated with a solution of 0.5 mg/mL collagenase in 1x PBS and filtered with a 0.2 μm syringe filter for 12 hours prior to incubation with nanoparticles. This concentration and time allowed for a slight damage to the spheroid periphery without disrupting the spheroid’s full integrity. Mesoporous silica nanoparticles functionalized with FITC were added to each well at a final concentration of 0.1 mg/mL in RPMI without FBS. Spheroids were incubated with the nanoparticles at 37°C and 5% CO₂ for 2 hours. Five spheroids from each test group were removed from the plates, washed twice with 1x PBS and imaged using the IVIS Lumina LT Series III pre-clinical in vivo imaging system (Perkin Elmer, Waltham, MA, USA) at field of view A in a welled glass slide. The amount of accumulation was quantified using the total background subtracted radiant efficiency values obtained from equal sized regions of interest (ROI).

Alteration of αvβ3 integrin ECM interactions

To determine the effects of αvβ3 integrins on the adhesion properties on nanoparticles to the tumor ECM, the six 12-day-old spheroids were selected and placed into the wells of a 96-well plate. Two-hundred μL of 4% formaldehyde were added to each well and incubated for 15 minutes. The fixed spheroids were then added to chamber slides where they were blocked with 200 μL of 2% BSA in 1x PBS for 1 hour at room temperature. The primary αvβ3 integrin antibody was incubated at 1:100 dilution for 2 hours and then washed. The secondary antibody was incubated with the spheroids for 2 hours at room temperature at a dilution of 1:500. These spheroids were then subsequently incubated with the FITC-MSNs at a concentration of 0.1 mg/mL in serum-free media for 2 hours. The spheroids were washed and imaged using confocal microscopy to visualize any differences in nanoparticle adhesion on the tumor surface before and after integrin interaction antibody interaction. The total fluorescent signal was then determined using ImageJ on thresholded images by generating a z-projection and measuring the integrated density within the spheroid’s region of interest.

Modified nanoparticle-tumor interaction in 3D tumor spheroids

To test nanoparticle-specific interactions with 3D tumor spheroids, dye-loaded MSN, MSN-PEG, and MSN-FA were each incubated for 2 hours with 12-day-old OVCAR-8 spheroids at a concentration of 0.1 mg/mL suspended in serum-free media. Analysis were carried out as described above using the IVIS Lumina LT to measure bulk fluorescence changes on background-subtracted images and images of each spheroid were taken using confocal microscopy and displayed as max z-projections.

Uptake of MSNs by peritoneal macrophages

Ascites was collected 24 hours after IP treatment with MSNs using a 1 mL syringe. The ascites fluid was then centrifuged for 5 minutes at 1500 rpm. The supernatant was removed and 1 mL of ultrapure water along with 250 μL of 4.25% NaCl was added to the tube along and centrifuged for another 5 minutes at 1500 rpm. The cells were resuspended in 1 mL of complete medium and counted. The cell suspension was added to chamber slides and incubated for 2 hours 37°C and 5% CO₂. The cells were subsequently washed 3 times with PBS and fixed with 4% paraformaldehyde. The cells were permeabilized with 0.25% Triton X-100 in PBS and the blocking solution used for staining was 10% goat serum and 1% bovine serum albumin in PBS. The primary antibody was used at a concentration of 1:500, while the secondary antibody was at 1:1000. Gold antifade media with DAPI was used to mount the slides and stain the nucleus. The slides were then imaged using confocal microscopy to determine if the fluorescent signal from the MSNs was found inside the stained peritoneal macrophages. The MSN-specific fluorescence was then measured using ImageJ on YEN thresholded images and represented as integrated density.

Ex vivo tissue incubation

All animal studies were performed according to an approved protocol by Institutional Animal Care and Use Committee (IACUC) at the University of Connecticut.

Peritoneal tissues (tumor, liver, intestines, kidneys, spleen, pancreas) were carefully excised from mouse cadavers that were bearing OVCAR-8 peritoneal tumors. Each tissue was washed with warm 1x PBS 3 times and then placed in a 50 mL centrifuge tube filled with 10 mL of either MSN-Cy5.5, MSN-Cy5.5-FA, or MSN-Cy5.5-PEG suspensions at a concentration of 0.5 mg/mL in serum-free RPMI. The organs were placed in a water bath at 37°C and shaken for 4 hours. After the 4-hour incubation period, the organs were washed with 1x PBS and placed on a glass dish to be imaged using the IVIS Lumina LT. The fluorescence values for each organ were normalized by the ROI used to measure the fluorescence signal of each organ.

Mouse Peritoneal OVCAR-8 and MIA PaCa-2 Tumor Model and Biodistribution:

All animal studies were performed according to an approved protocol by Institutional Animal Care and Use Committee (IACUC) at the University of Connecticut.

For the establishment of a peritoneal ovarian cancer xenograft model, 6 to 8-week-old female Athymic Nude-Foxn1nu mice were injected with 1 × 10⁷ luciferase positive OVCAR-8 cells suspended in sterile 1x PBS directly to the peritoneal cavity. The MIA PaCa-2 cells were injected at the same concentration and developed for the same amount of time as the OVCAR-8 IP tumors. The weight of each mouse and general state of health were monitored every 2 days until noticeable ascites was detected, which appeared at approximately 4 weeks. Mice exhibiting a 20% change in weight within 2 days or visible signs of serious health concerns were sacrificed and excluded from the MSN biodistribution studies. After notable ascites was detected, MSN-Cy5.5, MSN-Cy5.5-PEG, MSN-Cy5.5-FA, Fe-HA-Cy5.5, Poly-Cy5.5 in saline were injected into the peritoneal cavity. Fluorescence distribution was monitored using the IVIS Lumina LT system in real-time. After 24 hours the mice were sacrificed, tissues were excised, and nanoparticle-specific, background-subtracted fluorescence signal was measured in each organ. The background-subtracted fluorescence signal was then normalized by the region of interest and represented as a percentage of signal with respect to the fluorescence signal found in every organ analyzed in each mouse.

All experiments were conducted with at least three replicates and p-values < 0.05 were considered significant using one-way ANOVA.

RESULTS AND DISCUSSION

MSN surface properties impact tumor accumulation

Our first step in understanding the nanoparticle specific interactions with peritoneal tumors following IP administration began with the study of the formulation-specific characteristics that could potentially influence nanoparticle accumulation using MCM-41 type MSNs. To maintain uniformity in our studies we performed our modifications on a single batch of purchased MSNs. The surface of the MSNs was modified with folate acid (FA), a targeting ligand to actively target tumor cells. Ovarian cancers are known to overexpress folate receptors and many groups have modified their nanoparticle systems with FA to improve targetability of their nano drug delivery systems following IP administration^16–19. In addition, the interaction of the MSNs with the tumors was decreased by PEGylation of the MSN surface. With the addition of polyethylene glycol (PEG) on the surface of the MSNs, a decrease in nanoparticle interaction with the native tumor extracellular matrix was anticipated based on literature reports highlighting increased PEG surface coverage that resulted in decreased interaction of the particle with collagen, a major component of tumor ECM^{28, 29}. The surface charge of the MSNs was also a consideration, knowing that the ECM exhibits a net negative charge and could potentially affect interaction with the MSNs²⁹. The mean particle size and zeta potential were measured using Nanoparticle Tacking Analysis (Nanosight NS500). The system is able to provide a number based particle size distribution representing the particle dispersion status in water. TEM was used to confirm the size distribution. The particles were polydisperse and displayed considerable aggregation under these imaging conditions, with mean sizes of 250–290nm (Figure 1 and Table 1). PEGylation of the MSNs was confirmed using FTIR and BET analysis (Figure 2).

Figure 1:

TEM images of the unmodified and modified MSNs on carbon film, 400 mesh copper grids.

Table 1:

Particle size and zeta potential data of MSN formulations determined using Nanoparticle Tracking Analysis (Nanosight NS500).

MSN Type	Particle Size [Mean ± SD]	Zeta Potential [Mean ± SD]	Particle Size (+FBS) [Mean ± SD]	Zeta Potential (+FBS) [Mean ± SD]
MSN-Cy5.5	289 ± 18 nm	8.0 ± 0.1 mV	260 ± 50 nm	−7.3 ± 0.6 mV
MSN-Cy5.5-FA	270 ± 28 nm	−2.3 ± 0.3 mV	290 ± 45 nm	−10 ± 0.7 mV
MSN-Cy5.5-PEG	252 ± 40 nm	8.6 ± 0.4 mV	284 ± 30 nm	−9.8 ± 0.6 mV

Figure 2:

IR Spectra and BET data of various MSN and MSN-PEG formulations. The peak at 1050 cm⁻¹ represents Si-O-Si stretching, the 2939 cm⁻¹ and 2879 cm⁻¹ peaks represent CH₂ and CH₃ stretching vibration of PEG, and the 1560 cm⁻¹and 1664 cm⁻¹peaks represent bending of CH₂ and CH₃ bonds in PEG. An increase in average pore diameter along with a significant decrease in surface area was observed following PEGylation.

Our unmodified MSNs exhibited a net negative surface charge of −47 mV and showed predominant tumor accumulation in mice²¹. To load Cy5.5 to the MSNs, the surface was amidated, bringing the zeta potential to a neutral 8 mV. The PEGylated and folic acid formulations exhibited similarly neutral zeta potentials (Table 1). After incubating with FBS, these particles remained particle sizes with slightly larger standard deviations. Zeta potential appeared to be slightly negative but with similar values. Since all the tested formulations exhibited similar sizes and surface charge and others also reported shifts in charge in the presence of proteins^{30, 31}, we decided to focus primarily on the effects of PEG and FA on tumor interaction.

We first investigated the differences in interaction using the OVCAR-8 3D tumor spheroids. It was found that MSN-Cy5.5-FA particles exhibited a statistically significant increase in accumulation compared to the other two modified MSN formulations which exhibited accumulation characteristics similar to one another (Figure 3).

Figure 3:

(A) Fluorescence signal analysis using IVIS Lumina LT of recovered OVCAR-8 spheroids after a 2-hour incubation with various MSN formulations. (n=7, p<0.001, one-way ANOVA) (B) Representative thresholded fluorescence images of tumor spheroids after a 2-hour incubation with modified MSN formulations.

These results are similar to other published reports and follow the general logic that the use of targeting ligands on the surface of nanocarriers can significantly enhance the interaction compared to unmodified particles.

Using these results, we expected that a similar trend would be observed in MSN accumulation in vivo. As we discussed in the introduction, the MSNs do not undergo considerable degradation within the first 24 hours^21,23 and our previously published data²¹ showed the majority of the accumulation of MSNs takes place within the first day, so we decided to investigate accumulation at 24 hours. Regardless of the tested MSN formulation, the particles were able to distribute throughout the peritoneal cavity evenly enough to significantly favor tumors found in a specific location in the cavity (Figure 4). The accumulation of MSNs in tumors, while significantly greater than in the other major organs, is heterogeneously distributed among various tumor tissues collected throughout the peritoneal cavity with no evident correlation between accumulation and tumor tissue location in the peritoneal cavity (Figure 4A). Within a group of similarly treated mice, the tumors with the highest tumor signal were not limited to the same specific region of the peritoneal cavity. In our tumor model, the tumors typically develop in 4 separate sections: the diaphragm/top of liver, the pancreas/stomach, the intestines/kidneys, and on the ovaries. We would expect, that if there are highly aggregated systems that cannot effectively distributed throughout the cavity, the particles should be primarily found on the tumors removed from the intestines/ovaries. We could not consistently find one of those regions where the particles tended to accumulate the most. The biodistribution data presented in Figure 4B is expressed as percent of MSNs found on each tissue compared to the total fluorescence recovered on all of the tissues analyzed. For the purposes of this analysis, the sum of the fluorescence signal on all of the tumor tissues was used to represent the total tumor accumulation. Based on the quantified fluorescence signals in excised tissues, MSN formulations exhibited statistically significant accumulation on tumors compared to other tissues within the peritoneal cavity, while exhibiting nominal systemic exposure (Figure 4B).

Figure 4:

(A) Representative images of excised tumors 24 hours post IP injection of MSN-Cy5.5. Tumors were numbered depending on what organ they were nearest to in the peritoneal cavity when they were excised. Tumor collection varied in number, size and location between all mice tested. (B) Quantification of the percent of total organ fluorescence signal (native units are: [p/s/cm²/sr]/[μW/cm²]/cm²) normalized by the area of the region of interest. Tumor accumulation of all of the nanoparticle formulations was significantly greater than the nanoparticle accumulation in all other organs. (n=3, mean ± SD, p<0.001, one-way ANOVA)

Fluorescence quantification in this study was used to determine trends that were not easily determined visually. Due to the heterogenous nature of the distribution of MSNs on tumor tissues and the heterogeneity of the number, size and location of the tumor tissues, we investigated the optimal way to represent the trends in tumor accumulation between the three MSN formulations. The addition of all of the background-subtracted tumor signal normalized by the ROI size produced the largest overall tumor signal (Figure 5). The tumors were mostly smaller than all of the other organs and a normalization of the signal by the region of interest was used to better compare the total fluorescence values observed in large organs like the intestines and liver with the smaller tumors. When comparing 4 different measures of tumor accumulation of MSNs between the formulations, no differences were found in in vivo tumor accumulation between all three formulations, confirming our findings (p>0.05, one-way ANOVA).

Figure 5:

Comparison of tumor fluorescence between 3 MSN formulations using various quantification methods. (A) Summation of tumor fluorescence (all tumor tissues summed together per mouse). (B) Average tumor fluorescence (average tumor signal from all the excised tumors per mouse). (C) Average of max tumor fluorescence (the 3 excised tumor tissues with the highest signal averaged together per mouse). (D) Summation of max tumor fluorescence (the 3 excised tumor tissues with the highest signal summed together per mouse). (n=3, mean ± SD)

The addition of folate did not further improve accumulation of the MSNs on the tumors 24 hours after administration (Figure 4B). This result did not corroborate the results of the in vitro testing with the spheroids. Many researchers have found that proteins in body fluids tend to attach on the surface of nanoparticles and limit the exposure of these ligands to cell membrane receptors^{30, 32, 33}. The introduction of a PEG layer has been shown to decrease protein adsorption upon suspension in fluids containing proteins³⁴ and decrease interaction of nanoparticles with collagen²⁸. Serum-free RPMI was used for the incubation of the spheroids in order to determine if interaction between the MSN surface and ECM components was due to interference by the adsorption of proteins on the surface of the particles. While this is not the most accurate representation of peritoneal tumors in vivo, as evidenced by the discrepancies in MSN-Cy5.5-FA accumulation between in vitro and in vivo data, this allowed for a better mechanistic understanding of the bare interaction of the particles with the tumor surface. The microenvironment of the spheroids is not expected to be as developed as in the in vivo situation since the OVCAR-8 spheroids develop without the need of exogenous signals or synthetic matrices in a well-controlled environment. This discrepancy in findings between our in vitro and in vivo results are primarily goverened by the increased level of complexity when moving from uniformly shaped, well-controlled environments to the highly unpredicatbale nature of tumor growth following IP implantation of cells. The variables that can affect the ability of an IP-administered nanoparticle to freely distribute throughout the peritoneal cavity, such as size and polydispersity, have a greater impact on tumor accumulation than targeting ligands and the poor interaction of PEG with ECM. To better understand whether differences between the formulations without the need for uniform distribution of the MSNs could be observed, an ex vivo model was used to incubate the tumors in a uniform suspension of MSNs for 4 hours in an attempt to ascertain differences in MSN accumulation.

It was found that there remained no observable differences in nanoparticle accumulation between the three formulations once the fully formed mouse OVCAR-8 IP tumors (Figure 6) were used. Interestingly, the formulation that had the tumors with the highest tumor signal came from the MSN-Cy5.5 group, but the disparity between the signal of those tumors and the tumors with lower signals in that group was the greatest. Care was taken to divide the tumors between IP regions evenly between the groups to limit the differences in the smoothness of the tumors found on the liver surface versus the tumors with found on the pancreas. When working with the mouse tumors in serum-free media ex vivo we can see that the added tumor complexity overshadowed the active targeting advantage observed with the spheroid model. Also, within one in vivo experiment, the variability in number of tumors, tumor location, tumor size, tumor shape and tumor exposure to nanoparticles adds more variables that are controlled in vitro than the protein corona could potentially affect. All in all, the differences observed between our in vivo, ex vivo, and in vitro studies highlight that the contribution of the passive, physical entrapment of the particles within the tumor peripheral extracellular matrix far outweighs any other biological interactions between the tumor components and nanoparticles following intraperitoneal administration.

Figure 6:

Fluorescence signal analysis using IVIS Lumina LT of excised OVCAR-8 tumor tissues after a 4-hour ex vivo incubation with 3 MSN formulations (n=5, not significant, one-way ANOVA).

Nanoparticle chemistry and its relation to IP biodistribution

It is apparent that, due to the nature of the development of metastatic peritoneal tumors and the robust nature of their tumor microenvironment, nanoparticle technologies are well suited for the IP delivery of therapeutics. The question is whether this nanoparticle-tumor interaction changes with significant alterations in nanoparticle chemistry. For further investigation into intraperitoneal delivery of nanoparticle tumor-nanoparticle interactions, the biodistribution of MSNs after IP administration was compared to a polymeric micelle formulation, polystyrene nanoparticles, and iron doped hydroxyapatite (Fe-HA) nanoparticles. The polymer nanoparticles used were brush-like in nature with a PEG hydrophilic segment, cholesterol hydrophobic segment and a polynorbornene backbone with a particle size of around 150 nm²⁶. The Fe-HA nanoparticles were rod-like, magnetic, and prone to fast settling and aggregation in PBS. The NTA measured particle size was around 200 nm and differed from the TEM measurements mentioned in the publication³⁵. The purchased polystyrene nanoparticles were monodisperse and had a measured particle size of 190 nm, which agreed with the advertised 200 nm. All of the particles were similar in size to normalize the effects of size on intraperitoneal distribution. The Fe-HA particles were provided by a collaborator and represented particles with a similar tendency to aggregate and settle as MSNs due to their higher material densities. The polystyrene nanoparticles were used to represent a particle system with a lighter density that has very limited biodegradability³⁶ that we expect to have a very high intraperitoneal residence time like our MSN system. The brush-like polymeric micelles were extensively developed in our lab²⁶ and were representative of a particle formulation of similar size, lower density, and higher degradability than MSN. In this case we expected these particles to exhibit high systemic circulation and low peritoneal residence time in comparison to the MSNs. It was observed that all three nanoparticle formulations were able to accumulate on tumor tissues at a greater percentage per tumor surface area compared to other tissues, except for the Fe-HA particles in which the greatest accumulation was observed in the liver (Figure 7). It was noted that the polymeric micelle and polystyrene particles did accumulate in the heart, showing that these particles were capable of escaping the peritoneal cavity 24 hours after IP administration. It was reported that particles with sizes smaller than 50 nm will be removed via the lymphatic system after i.p. injection and enter the circulation ⁸. However, if the size of the MSNs is too large, it may affect their ability to distribute throughout the peritoneal cavity.

Figure 7:

Percent of total organ signal normalized by the area of the region of interest. Tumor accumulation for all nanoparticle formulations was significantly greater than the nanoparticle accumulation on all other organs except for the high accumulation of Fe-HA particles in liver (n=3, mean ± SD, p<0.001). Significant accumulation of polymeric micelles and polystyrene particles was observed in the heart and lungs.

It is apparent that even with the IP administration of particles of varying surface chemistries, the tumors within the peritoneal cavity typically have the highest accumulation of nanoparticles. Due to polydispersity and degradation of the administered nanoparticles, some of the smaller particles and degradation fragments may eventually reach the systemic circulation via absorption through the lymphatic system⁸. This can prove disadvantageous when administering high doses of chemotherapeutics and radiotherapeutics to specifically treat tumor tissues within the peritoneal cavity. The degradability of the tested polymer nanoparticles, and the polydispersity of the unmodified iron-doped hydroxyapatite nanoparticles, led to heart and lung exposure that was not observed with the MSNs. These observations, however, do not clearly point to the nanoparticle-based properties that could be specifically affecting their interaction with the tumor.

Peritoneal metastasis tumor models from different tumor cell origins do not affect the accumulation of mesoporous silica nanoparticles

Following our previous reports about the ability of MSNs to selectively accumulate on tumor tissues in peritoneal metastasis models with SKOV-3 and OVCAR-8 tumors, we investigated whether the specific tumors from different tumor cell origins would impact nanoparticle accumulation. In this study we focused attention on MIA PaCa-2 and OVCAR-8 tumors. Three weeks after the initial injection of MIA PaCa-2 and OVCAR-8 cells to mice, ascites became evident, signaling significant peritoneal tumor development. The tumor growth within the peritoneal cavity for both MIA PaCa-2 and OVCAR-8 mouse models was similar to other proposed peritoneal metastasis mechanisms, i.e., by exfoliation and peritoneal fluid dissemination³⁷. MIA PaCa-2 cells tended to develop large solid tumors around the pancreas region of the peritoneal cavity. Some MIA PaCa-2 tumors were found embedded within the diaphragm adjacent side of the liver with minimal invasion into the mesentery. OVCAR-8 cells developed more diffuse tumors that not only appeared around the pancreas region, but also observed to a significant degree within the mesentery. There was also some noticeable infiltration of OVCAR-8 tumors within the diaphragm. Regardless of the tumor cell origin, MSN-Cy5.5 with a mean particle size of 250 nm primarily accumulated on tumor surfaces 24 hours after IP administration with some accumulation on the intestines (Figure 8).

Figure 8:

Representative thresholded images of excised organs 24 hours and 6 days after Cy5.5 loaded MSNs were administered intraperitoneally to athymic nude mice bearing either MIA PaCa-2 or OVCAR-8 peritoneal tumors. Red arrows denote embedded tumor tissues in the liver.

There was no detectable systemic exposure of the IP injected MSNs even 6 days after administration as evidenced by the lack of a signal in the heart and lungs. Six days after IP injection, the MSNs remained primarily on tumor tissues. When compared to the 24-hour biodistribution for MIA PaCa-2 tumors, there was no evidence of nanoparticle accumulation in healthy organs six days after IP administration. There was observed heterogeneity in nanoparticle distribution within the excised tumor tissues and in cases like MIA PaCa-2 tumor 2b after 24 hours of MSN incubation, there was a concentration of MSNs on one side of the tumor tissue. This heterogeneity is to be expected³⁸ and can be overcome in the future with a larger injection volume and massaging the cavity after injection to make sure the organs are not touching the parietal peritoneum and obstructing flow of particles. Poor MSN flow within the cavity is not a concern in this case as evidenced by the distribution of MSNs not being limited to tumors near the injection site. Using the OVCAR-8, 6-day image as an example, tumor 2a was found on the pancreas and tumor 2b was a tumor found on the diaphragm. The overall nanoparticle distribution on the excised tumors did not follow a significant pattern based on where the tumors were located. The overall percentage of injected MSNs found on tumors did not vary significantly between the two tumor models. These data show that the major driving force for MSN accumulation within the tumor microenvironment of both cancer types is not reliant on the primary tumor origin. Most of the nanoparticle-tumor interaction is dependent on the extracellular matrix developed around the solid tumor and lack of interaction of MSNs with the mesothelium surrounding the healthy peritoneal organs. High specific accumulation of MSNs on tumor tissues of different cell origins, tumor size and cavity distribution following IP injection, along with the extended peritoneal residence time, shows the promising nature of this drug delivery system to deliver high concentrations of therapeutic payloads for the treatment of peritoneal metastasis.

Peritoneal macrophages do not significantly alter nanoparticle accumulation in peritoneal tumors

After investigating the impact of nanoparticle chemistry and tumor cell lines on peritoneal tumor accumulation, we explored whether the peritoneal and peritoneal tumor microenvironments would facilitate the selective accumulation of nanoparticles. Before the MSNs are able to accumulate on the tumor surface following IP administration, the particles must distribute within the peritoneal fluid. Tumor associated macrophages make up a significant portion of the peritoneal fluid and the periphery of the tumor microenvironment in a patient with late stage ovarian cancer, and are known to promote tumor progression³⁹ in addition to being associated with the uptake of IP administered nanoparticles within the tumor tissue^{14, 15}. TAMs were isolated from the peritoneal fluid of athymic nude mice bearing OVCAR-8 peritoneal tumors. Twenty-four hours after the initial injection of MSNs, the amount of isolated TAMs that engulfed the MSNs within the peritoneal fluid was limited (Figure 9A). Six days after the initial injection we were able to identify MSNs engulfed by the the isolated macrophages, showing that the peritoneal residence time of the MSNs following IP administration is quite long (Figure 9B). The lack of nanoparticle uptake by macrophages in the peritoneal fluid within the first 24 hours does not align with the significant tumor accumulation of MSNs observed within the first hour after injection²¹.

Figure 9:

(A) Isolated peritoneal macrophages (red, Anti CD68) 24 hours after IP injection of MSNs (green, FITC). Nucleus stained with DAPI (blue). (B) Isolated peritoneal macrophages 6 days after IP injection of MSNs. (C) Cryo-sectioned OVCAR-8 tumors from mice treated with an IP injection of MSN incubated (green, FITC) for 24 hours. Tumors were stained for peritoneal macrophages (red, Anti CD68) and cell nuclei were stained with DAPI (blue).

Tumor tissues were stained to indentify any co-localization of macrophage with fluorescent MSNs. The MSNs that accumulated on the tumor periphery did not directly overlap with the macrophage rich regions of the tumor (Figure 9C). The green fluorescence associated with MSNs was analyzed using ImageJ on YEN thresholded images and quantified as integrated density (intden). Within 24 hours, much greater fluorescense (228204 intden) was detected in tumor tissues compared with that in macrophages (90270 intden). These discrepancies indicate that the primary avenue for tumor accumulation of the MSNs is not associated with macrophage trafficking. The passive accumulation of nanoparticles on tumors in the peritoneal cavity is governed by the distribution of the nanoparticles in the cavity and the physical surface interactions with tumor tissues. The fact that MSNs were still being taken up by macrophages in the peritoneal fluid 6 days after the initial injection shows that the MSNs have a long peritoneal residence time, increasing their chances of interacting with tumor surfaces.

Tumor ECM, particularly the collagen matrix, drives tumor-nanoparticle interaction

Once the nanoparticles reach the tumor surface from the peritoneal fluid, the interaction between the nanoparticle and the tumor surface will dictate the subsequent fate of the nanoparticle. It is apparent that the most significant factor dictating this interaction is the tumor surface. The major components of the tumor ECM are collagen (primarily subtype I) and fibronectin, which commonly binds integrins, the cell adhesion molecules^{40, 41}. To investigate the interaction between MSNs and collagen, 3D tumor spheroids with OVCAR-8 cells were used as a platform for ECM manipulation that closely resembles in vivo peritoneal tumors. The growth of these 3D spheroids mimics the avascular tumors found in most cancers that metastasize to the peritoneal cavity⁴².

The collagen distribution in the OVCAR-8 3D tumor spheroids was visualized using antibody staining to determine how uniformly the matrix was distributed on the surface of the spheroids. The collagen matrix formed a well-distributed capsule around the tumor surface and was present at a higher density on the surface of the spheroids with a noticeable decrease in density 40 μm into the tumor tissue (Figure 10).

Figure 10:

(A) Maximum intensity projection of a non-fixed 12-day old OVCAR-8 spheroid stained with a Collagen I antibody and imaged using confocal microscopy. The blue staining represents Collagen I location. The spheroid ECM is uniformly distributed and forms a dense capsule around the spheroid (B) A XZ reconstruction of OVCAR-8 spheroid showing a higher collagen density on the surface of the spheroid.

Since the nanoparticle-tumor interactions are initiated via an interaction with the ECM capsule, we hypothesized that a disruption of this capsule would decrease the overall accumulation of the MSNs. Following incubation of OVCAR-8 3D tumor spheroids with collagenase for 12 hours at a concentration of 0.5 mg/mL, it was found that the overall structure of the tumor tissue remained intact. There was a slight change at the periphery of the spheroid surface, showing disruption of the outer collagen capsule. When the collagenase concentration was increased, the spheroids lost their spheroidal structure followed by a statistically significant decrease in overall accumulation of the MSNs on the spheroid surface within the first 2 hours of incubation (Figure 11). The effect of the ECM disruption on nanoparticle accumulation was not affected by the collagenase concentrations above 0.5 mg/mL that did not completely disrupt the spheroidal shape.

Figure 11:

(A) Representative images of OVCAR-8 spheroids incubated with MSN-FITC obtained using a confocal microscope with a 10x objective. The spheroids treated with collagenase prior to MSN incubation showed significantly less spheroid accumulation of the MSNs. (B) Thresholded IVIS Lumina LT fluorescent images of OVCAR-8 spheroids collected in a welled microscope slide. Collagenase treated spheroids were treated with 0.5 mg/mL collagenase for 12 hours prior to the 2-hour MSN-FITC incubation (n=7). (C) The MSN accumulation on the surface of collagenase-treated OVCAR-8 spheroids at various collagenase concentrations. (n=5, p <0.01, one-way ANOVA)

These results suggest that the physical entrapment of the nanoparticles within the collagen matrix of the tumor tissue is a major driving force for the selective accumulation of MSNs following IP administration. The estimated average pore size of the collagen-rich ECM is approximately 100 nm; particles larger than 100 nm tend to be entrapped and cannot penetrate deeper into the matrix⁴³. The collagen matrix in this case acts as a porous sieve that, upon digestion with collagenase, increases in average pore size. This would subsequently facilitate MSN mobility eventually leading to fewer entrapped particles on the surface of the tumor tissue. This explanation, however, only accounts for a single portion of the peritoneal tumor microenvironment that may facilitate nanoparticle accumulation.

The partial digestion of the collagen matrix using collagenase provided a gross understanding of how changes to the ECM structure could dramatically change nanoparticle accumulation. Further investigations of other cell-cell interactions within the ECM that could potentially modify the nanoparticle accumulation were undertaken. Integrins are a large family of adhesion molecules that anchor tumor cells to the ECM and are important factors in facilitating cancer metastasis²⁵. The OVCAR-8 cell line expresses αvβ3 integrins that commonly bind tumor cells to fibronectin, a major component of the tumor ECM⁴⁴. This integrin is also associated as potential target for various anticancer therapies based on its role in angiogenesis⁴⁵. By irreversibly binding an antibody to this integrin and blocking its signaling, we sought to interfere with the production of collagen by decreasing fibronectin-integrin. Ultimately we wanted to investigate whether targeting the molecular machinery as opposed to the bulk digestion of collagen could affect MSN tumor accumulation. Interestingly, this did not affect the ability of the MSNs to accumulate on the surface of the spheroids, thus demonstrating that the disruption of the integrins did not disrupt the interaction between fibronectin and collagen enough to cause drastic remodeling of the ECM (Figure 12).

Figure 12:

(A) Confocal image (10x) of a spheroid incubated with MSN-FITC (green) particles without a pre-incubation with integrin antibodies. (B) Confocal image (10x) of spheroid that was pre-incubated with anti-αvβ3 integrin antibodies for 2 hours and subsequently incubated with MSN-FITC particles (green). (C) Representative confocal image (40x) of the surface of a spheroid incubated with both anti-αvβ3 integrin antibodies (red) and MSN particles (green). (D) Average fluorescence intensity of z-projections generated in imagej comparing MSN accumulation with and without anti-αvβ3 integrin antibody pre-incubation. (n=4, paired t test, no significance)

These results strongly suggest that MSN accumulation on the surface of OVCAR-8 tumor spheroids primarily involves physical entrapment within the porous collagen matrix of the spheroid and, without a major disruption of the collagen structure, a noticeable change in accumulation would not be expected. Studies⁴⁶ have shown that disrupting the tumor extracellular matrix allows for better distribution of chemotherapies deeper into the tumor tissue. This ECM disruption was studied by priming the tumor microenvironment with therapeutics to degrade ECM or inhibit ECM synthesis^47–49. ECM disruption was also studied in some groups with the use of physical modulation of the tumor microenvironment with ultrasound, radiation or hypothermia ^50–52. The goal of these strategies is to disrupt the internal ECM so the particles may freely penetrate deeper into the tumor tissue, effectively depositing therapeutics throughout the entire solid tumor mass. Without these ECM modulating strategies, only the cells close to the particles that are releasing the therapeutic will die, causing partial treatment and ultimately relapse. The accumulation and penetration of nanoparticles in peritoneal tumors directly affects one another, and the optimal balancing of both increasing accumulation and subsequently increasing nanoparticle penetration would come from the timing of the doses. If the nanoparticle is carrying a radionuclide or a toxic agent that disrupts this ECM capsule, multiple dosing may become greatly affected if each subsequent dose accumulates half as much. Future studies into how the remodeling of the collagen matrix can affect nanoparticle accumulation can provide valuable insight into how patients with peritoneal tumors respond to nanoparticle treatments if they have previously received extensive radiation and IP chemotherapy treatments.

CONCLUSIONS

The intraperitoneal delivery of nanoparticles remains a viable approach for increasing concentrations of therapeutic agents in metastatic peritoneal tumors with limited systemic toxicity. The distribution and accumulation of nanoparticles delivered to the peritoneal cavity is primarily passive in nature and is enhanced by a formulation that is stable, thus maximizing the probability of all tumors in the cavity being exposed to nanoparticles. The interaction of the particles with tumor ECM, which is primarily comprised of a collagen matrix, promotes the selective accumulation of the particles on the tumor surface, and modifications to this structure can greatly decrease the ability of MSNs to accumulate on the tumor tissue. MCM-41 type mesoporous silica nanoparticles with a size of 250 nm were able to selectively accumulate on tumor tissues following intraperitoneal delivery without the need of targeting ligands, and have a long peritoneal residence time, making them promising nanocarriers for the delivery of therapeutics for the treatment of peritoneal metastasis.

Highlights.

• Nanocarriers dosed intraperitoneally selectively accumulate on peritoneal tumors

• Collagen matrix on tumor surface drives the specific interaction with nanoparticles

• Targeting ligands does not affect nanoparticle targetability

• Nanoparticle type determines systemic exposure which affects biodistribution

ABBREVIATIONS

EPR

enhanced permeability and retention effect

IV

intravenous

IP

intraperitoneal

MSN

Mesoporous Silica Nanoparticles

TAM

tumor associated macrophages

ECM

extracellular matrix

MCM-41

Mobil Composition of Matter Number 41

FA

Folic Acid

Fe HA

Iron Doped Hydroxyapatite nanoparticles

PEG

Polyethylene glycol

MIA PaCa-2

Human epithelial pancreatic carcinoma

OVCAR-8

Human epithelial ovarian adenocarcinoma

APTES

(3-Aminopropyl)triethoxysilane

DCCI

Dicyclohexylycarbodiimide

TEA

triethylamine

DMSO

dimethylsulfoxide

FBS

fetal bovine serum

BSA

bovine serum albumin

FITC

fluorescein isothiocyanate

Cy5.5

cyanine 5.5