Combination Chemical and Mechanical Tumor Immunomodulation Using Cavitating Mesoporous Silica Nanoparticles

Abstract

Combinatorial methods to repolarize tumor-associated macrophages from anti-inflammatory to pro-inflammatory phenotypes offers a promising route for cancer immunotherapy. However, most studies examine biochemical combinations alone. Therefore, we studied simultaneous chemical and mechanical stimuli as orthogonal cues for enhanced immunomodulation. We engineered the surfaces of hydrophobically functionalized mesoporous silica nanoparticles (F108-hMSNs) to encapsulate the immunomodulator resiquimod and kill cancer cells through high-intensity focused ultrasound (HIFU)-mediated inertial cavitation, releasing damage-associated molecular patterns (DAMPs) for prolonged macrophage stimulation. The HIFU doses alone did not affect cells, but in combination with F108-hMSNs, achieved significantly higher cancer cell death and DAMP generation. Inflammatory markers (CD86, MHC II, iNOS) were upregulated in tumor-associated-like macrophages treated with F108-hMSNs in the presence of HIFU and experienced the greatest inflammatory phenotypic shift of all conditions tested. This work suggests that chemical and mechanical activation facilitated by engineered nanoparticles offer a promising treatment against immunologically cold tumors.

Graphical Abstract

Introduction

Interactions between the immune system and tumors can promote or arrest cancer cell proliferation. Tumor cells that successfully evade the immune system can delegate immune cells to a tissue regenerative state that inhibits effector cell functions and promotes angiogenesis.^1,2 These tumors are considered immunologically “cold” and are more resistant to immunotherapy, contributing to the failure of several clinical trials.³ Therefore, initiating an innate immune response to counteract the suppressive tumor microenvironment is a promising treatment strategy to render solid tumors more responsive to immunotherapy. Several strategies to reprogram the tumor microenvironment have been proposed, including cellular backpacks^4–6 and tumor lysate-loaded hydrogels.⁷ However, the majority of these systems stimulate a single pathway, limiting the magnitude and durability of immune cell activation within the tumor microenvironment. Stimulation of multiple, orthogonal pathways should enable better immune activation than a single pathway stimulated in isolation. Others have used physical methods like photodynamic therapy in combination with agonists for Toll-like receptors (TLRs) or stimulator of interferon genes (STING).^8,9 Another team developed chemotherapeutic prodrug nanostructures loaded with a STING agonist and decorated with immune checkpoint inhibitors for enhanced targeting.¹⁰ These combination therapies have achieved higher degrees of inflammation, inhibited growth of tumors, and prolonged survival of tumor-bearing mice in comparison to each constituent therapy in isolation.

Inspired by these results, we have developed a class of mesoporous silica nanoparticles (MSNs) capable of initiating two distinct immunomodulatory pathways for enhanced macrophage activation in tumor-like environments (Fig. 1). The first pathway is through the induction of damage-associated molecular pattern (DAMP) receptors, which recognize damage to biological tissue. DAMPs are a category of endogenous biomacromolecules that promote inflammatory gene expression. These factors can be fragments of extracellular matrix, signaling proteins secreted in extracellular vesicles, or other macromolecules that are ubiquitously expressed intracellularly and released from necrotic cells. In the reported work, the engineered MSNs are capable of facilitating cavitation under stimulation with high intensity focused ultrasound (HIFU),^11,12 which causes mechanical damage to nearby tissue through the formation of streaming flows. These engineered nanoparticles enable local generation of DAMPs deep within the body at sites where other modalities may be unsuccessful.

Figure 1.

(a) Schematic illustration of bubble growth and collapse on an hMSN surface. Illustration is not drawn to scale. (b) Application of HIFU stimulates particle propulsion through cancer cells to create DAMPs and release R848, inducing pro-inflammatory phenotypes of tumor-associated macrophages.

The second pathway is TLR activation by release of the small molecule immunomodulator resiquimod (R848). R848 is a TLR 7/8 agonist that is known to prompt macrophages toward an inflammatory phenotype,¹³ which renders the tumor more susceptible to immune intervention through excretion of nitric oxide and increased phagocytosis of malignant cells.^14,15 Along with DAMPs generation, TLR agonism offers a second modality for immune activation through two orthogonal pathways, offering a stronger and more sustained inflammatory phenotype in the tumor microenvironment than existing therapies. Extensive previous work has shown that MSNs can serve as effective reservoirs for immunomodulatory agents like R848.^16,17

To our knowledge, the combined effects of chemical and mechanical activation of macrophages from modified MSNs have yet to be investigated. Here, we sought to understand the relative contributions of DAMP generation and TLR agonism on the final phenotype of macrophages in the interest of inducing a strong and durable shift of tumor-associated macrophages toward an inflammatory phenotype. Such an approach could increase the likelihood of initiating an adaptive immune response in cold tumors and serve as a standalone treatment or as a complement to existing immunotherapies.

Synthesis and Characterization of MSNs and F108-hMSNs

MSNs were synthesized using a modified Stöber process, using tetraethyl orthosilicate (TEOS) as the silica precursor and the cationic surfactant cetyltrimethylammonium chloride (CTAC) as a structure-directing agent (Fig. 2a).^18,19 The surfaces of the MSNs were modified with the twelve-carbon molecule dodecyltrichlorosilane (DDTS) to yield hydrophobically modified MSNs (hMSNs), as confirmed by FTIR (Fig. S1). Hydrophobic modification of particle surfaces prevents water from wetting the strongly hydrophobic pores, establishing a Cassie-Baxter state.²⁰ Then, particles were then stabilized with the surfactant Pluronic F108 (F108-hMSNs) and loaded with R848 (R848@F108-hMSNs) simultaneously by co-precipitation of the hydrophobic particles and R848 from isopropanol to water. By transmission electron microscopy (TEM, Fig. 2b inset), the MSNs had a diameter of 45.4 ± 0.2 nm, whereas their hydrodynamic diameters were measured at 143 ± 27 nm by dynamic light scattering (DLS, Fig. 2b). The difference in these measurements results from the hydrodynamic shell around the particles, particle aggregation, and the propensity for DLS to detect larger particles and thus skew the average.²¹ Diameters of F108-hMSNs and R848@ F108-hMSNs were also measured by DLS to be 232 ± 50 nm and 152 ± 28 nm, respectively (Fig. 2b). Zeta potentials of MSNs, F108-hMSNs, and R848@F108-hMSNs were −15.5, −3.0, and −2.3 mV in PBS, respectively (Fig. 2c); the decrease in magnitude of these zeta potential measurements may signify a reduction in colloidal stability, which can lead to aggregation of particles and explain the increase in size of F108-hMSNs measured by DLS. However, the decrease in magnitude of the F108-hMSN zeta potential was expected because alkylation of the particle surfaces diminished the number of negatively charged silanol groups, therefore resulting in a less negative charge than bare MSNs.

Figure 2.

(a) Schematic illustration of particle synthesis and preparation (not to scale). (b) Size distribution of MSNs, F108-hMSNs, and R848@F108-hMSNs. Inset: TEM image of MSNs. (c) Zeta potential of MSNs, F108-hMSNs, and R848@F108-hMSNs. **p < 0.01, ***p < 0.001 using ANOVA and Tukey’s multiple comparison test. N = 3 ± SE.

Next, the acoustic properties of F108-hMSNs were tested. In short, a 1.1 MHz focused ultrasound transducer was submerged in a water bath and fitted with a coupling cone to guide the sample to the focal zone of the transducer (Fig. 3a). To quantify the activity of particles in the presence of HIFU, solutions of PBS alone, MSNs in PBS, or F108-hMSNs in PBS were added to an inverted transfer pipette so that the particle solution was held in the pipette bulb. The pipette bulb was then secured on top of the coupling cone and exposed to HIFU. Explanation of the analysis of the acoustic data can be found in the SI (Fig. S2). We found that F108-hMSNs treated with HIFU displayed significantly greater sensitivity compared to MSNs and PBS using pulsed ultrasound with spatial-average-time-average fluences of 0.74 W/cm² (Fig. 3b). However, F108-hMSNs and MSNs both showed the greatest response to HIFU at 1.39 W/cm². Therefore, we used a fluence of 1.39 W/cm² in all subsequent studies.

Figure 3.

(a) Schematic illustration of HIFU setup. Not to scale. (b) Ultrasound response of PBS, MSNs, and F108-hMSNs over increasing ultrasound fluences. *Significance marker between F108-hMSNs and PBS; ^#significance marker between F108-hMSNs and MSNs. Significance determined using ANOVA and Tukey’s multiple comparison test, *p < 0.05, **p < 0.01, ^#p < 0.05. (c) Size distribution of F108-hMSNs before (yellow) and after (red) HIFU. Average diameters before and after HIFU are 232 ± 51 and 202 ± 61 nm, respectively (mean ± SD). (d) R848 release from F108-hMSNs with (red) and without (yellow) HIFU. Inset: molecular structure of R848. (e) HIFU-mediated cytotoxicity of 4T1 cells without F108-hMSNs measured via MTT assay. N = 3 ± SE. (f) F108-hMSN-mediated cytotoxicity of 4T1 cells without HIFU measured via MTT assay. N = 6 ± SE. (g) F108-hMSN- and HIFU-mediated cytotoxicity of 4T1 cells measured via MTT assay. N = 3 ± SE. Significance determined using ANOVA and Tukey’s multiple comparison test, *p < 0.05, **p < 0.01. Significance between viabilities of the same F108-hMSN concentration (f) without and (g) with HIFU determined using a Student’s t test, ^#p < 0.05, ^##p < 0.01.

R848 release from the F108-hMSNs was tracked by UV-vis spectroscopy. We found that R848@F108-hMSNs treated with HIFU released R848 faster, and to a greater extent, than unstimulated particles (Fig. 3d). Combining the drug release kinetics and the precise spatiotemporal control of energy deposition achievable with HIFU, this method offers a potential strategy for drug delivery that imposes a burst release at the specific place and time that it is most optimal.²² While we measured a mass of released drug slightly greater than what was loaded, we posit that our method of normalizing UV-vis spectra to account for nanoparticle scattering effects in the ultraviolet range may be a cause (Fig. S3).²³ We hypothesize that HIFU-treated particles release drug faster than non-HIFU-treated particles because cavitation-mediated fluid flow strips some fraction of the Pluronic coating on the surfaces of the particles, thus removing a transport barrier for drug release. In fact, DLS measurements found that particles had smaller diameters of 201 ± 61 nm after treatment with HIFU (Fig. 3c). Thus, we hypothesize the rate of drug release could be tuned in future work by using other surfactants, such as phospholipids, to reduce premature release.²⁴

Next, we investigated the cytotoxic effects of HIFU-treated F108-hMSNs on 4T1 murine mammary carcinoma cells. We found that, in the absence of F108-hMSNs, HIFU at fluences up to 1.39 W/cm² had no significant effect on 4T1 viability compared to cells receiving no HIFU treatment (Fig. 3e). This was expected, as malignancies are known to respond to HIFU treatments only in the range of 100–1600 W/cm² in the absence of contrast agents or other types of responsive nanoparticles.^25,26 Next, we tested the viability of 4T1 cells after treatment with F108-hMSNs in the absence of HIFU. We found that cell viability decreases significantly at concentrations of 890 ng/mL and above (Fig. 3f). Combining F108-hMSNs and HIFU, we found 4T1 cells treated with at least 2800 ng/mL F108-hMSNs at fluences of 1.39 W/cm² resulted in significant cell death compared to cells treated with HIFU only (Fig. 3g). We also found that F108-hMSNs were significantly more damaging to 4T1 cells with HIFU than without HIFU at concentrations of at least 280 ng/mL. Achieving an effect of this magnitude in vivo would require an intravenous nanoparticle dose of about 35 mg/kg, assuming a delivery efficiency of 0.1%;²⁷ this dose is one order of magnitude less than the LD50 of amorphous silica (Calculation S1).^28,29 The results from this study informed our decision to use 8900 ng/mL of F108-hMSNs to generate as many DAMPs from 4T1 cells as possible.

R848@F108-hMSN Effects on Macrophage Phenotype

To test our central hypothesis that the combination of chemical and mechanical stimulation conferred by R848@F108-hMSNs can produce a strong and durable pro-inflammatory phenotype in macrophages, we performed flow cytometry on bone marrow-derived macrophages (BMDMs) challenged with combinations of R848 and DAMPs released in the presence of 4T1 cells. To create a representative pro-tumoral immune microenvironment in vitro, we cultured BMDMs in hypoxia (1% O₂) and tumor-conditioned media (comprising 10 vol.% media from a 4T1 cell culture in BMDM media) for the duration of all studies.³⁰ Separately, HIFU was applied to a 12-well plate containing 4T1 cells and, where appropriate, R848@F108-hMSNs. After HIFU insonation, the 4T1 cell media supernatant was transferred to BMDMs in culture and incubated in hypoxia for one or three days to compare the longevity of cellular responses, after which they were removed from the plate and phenotyped using flow cytometry (Fig. 4a–f).

Figure 4.

Relative expression of (a) CD86, (b) MHC II, (c) iNOS, (d) RAGE, (e) CD206, and (f) HIF-1α as measured by flow cytometry and quantified by mean fluorescence intensity (MFI) on or in fixed and permeabilized BMDMs compared to unstimulated controls. (g) Scoring system for the degree of inflammation for each marker and condition tested. Inflammatory score was calculated as log₂(fold change) for CD86, MHC II, and iNOS and −log₂(fold change) for CD206 and HIF-1α. Total score is the sum of inflammatory scores for each marker. (h) IL-12 released from BMDMs as measured by ELISA. Data where no IL-12 was detected is marked ND. *Significance of a condition compared to the same timepoint of the Complete Therapy condition. *p < 0.05, **p < 0.01, ***p < 0.001 using ANOVA and Tukey’s multiple comparison test. ^#Significance of Day 3 timepoint compared to Day 1 timepoint. ^#p < 0.05, ^##p < 0.01, ^###p < 0.001 using a Student’s t test. N = 4 biological replicates ± SE. A complete table of p values from Tukey’s multiple comparison test can be found in Table S3.

First, 4T1 cells were insonated with HIFU in the absence of R848@F108-hMSNs as a negative control (“HIFU Only”). Phenotypic markers on BMDMs did not change more than two-fold after the addition of HIFU-treated 4T1 supernatants compared to unstimulated controls. Expression of the cell surface markers major histocompatibility complex (MHC, Fig. 4b) II and CD206 (Fig. 4e) were both downregulated by a factor of two by Day 3, which suggests that applying HIFU to 4T1 cells at the fluences described in this study elicits neither an inflammatory nor anti-inflammatory response.

Next, a material control was used to reveal the effects of F108-hMSNs without HIFU stimulation on the phenotype of BMDMs (“Particles Only”). We found that the particles had an inherent inflammatory effect, illustrated by upregulation of inducible nitric oxide synthase (iNOS, Fig. 4c) and CD86 (Fig. 4a), which was the highest upregulation of CD86 of all conditions tested. This inflammatory effect was slightly dampened through upregulation of hypoxia inducible factor 1α (HIF-1α, Fig. 4f), which transcribes genes for anti-inflammatory proteins such as VEGF.³¹ In addition to the previous inflammatory markers, expression of the receptor for advanced glycation end products (RAGE, Fig. 4d) doubled in the presence of F108-hMSNs, which is known to interact in other inflammatory pathways.³²

Next, the “Probe Lysis” condition was used to determine the effects of DAMPs generated from 4T1 cells alone on BMDMs activation. 4T1 cells were lysed using a probe sonicator. We confirmed that probe sonication of 4T1 cells led to their complete destruction (Fig. S4), as compared to partial cell killing by F108-hMSNs and HIFU (Fig. 3e). Therefore, the lysate from the probe sonicator was diluted by a factor of 2.25 to approximate the concentration of DAMPs released in the MTT assay. The diluted lysate was added to BMDMs, and induced inflammatory protein expression in BMDMs as expected, illustrated by increased expression of MHC II and iNOS on both days.

To isolate the effects of R848 alone on BMDM phenotypes (“Free R848”), BMDMs were incubated with R848 at a concentration of 140 nM, 10x greater than its EC50.¹³ We chose this concentration to promote robust inflammatory activation of the BMDMs and to match the concentration of R848 released from R848@F108-hMSNs in the presence of HIFU. Interestingly, free R848 elicited both inflammatory effects, like upregulation of MHC II and iNOS, and anti-inflammatory effects, such as upregulation of CD206 and downregulation of CD86. R848-dependent elicitation of pro- and anti-inflammatory markers is consistent with previous studies.³³

Lastly, phenotypic markers of BMDMs were measured after incubation with the supernatant of 4T1 cells treated with R848@F108-hMSNs and HIFU (“Complete Therapy”). The Complete Therapy condition evoked inflammatory responses from macrophages, especially by way of iNOS. The iNOS marker was upregulated on Days 1 and 3, showing that F108-hMSNs in the presence of HIFU can lead to durable immune activation over at least three days. Additionally, we observed a two-fold upregulation of RAGE on Day 3. The magnitude of change in RAGE was similar to that of the Particles Only condition, offering evidence that the RAGE receptor recognizes and binds to F108-hMSNs, since it is known that RAGE expression is upregulated after binding to its ligand.³⁴ The data also showed an increased expression of HIF-1α. Interestingly, the expression profiles of RAGE and HIF-1α on BMDMs for each condition were strikingly similar, suggesting these two proteins may be implicated in the same pathway.

Bimodal distributions were observed in some conditions for the iNOS marker, which was not observed in other markers. To corroborate our conclusions, we employed a secondary method of analysis: rather than comparing the median fluorescence intensity across conditions, we enumerated the number of BMDMs in a population that fluoresced brighter than an arbitrary threshold of 10⁴ relative fluorescence units (RFU; Fig. S5). We found that the same trend held in both analyses: the Complete Therapy condition produced the highest iNOS expression in BMDMs, followed by Probe Lysis and Free R848 conditions (Fig. S6).

The expression profiles of CD86, iNOS, and CD206 in the Complete Therapy condition appear to include the effects of particles, DAMPs, and R848. For CD86, we observed upregulation from the particles, but downregulation from DAMPs (Probe Lysis condition) and R848, so the Complete Therapy condition displayed a slight downregulation on Day 1 and moderate upregulation on Day 3. iNOS was upregulated from the particles, DAMPs, and R848, which led to a substantial upregulation for the Complete Therapy condition. Particles downregulated CD206, R848 upregulated CD206, and DAMPs seem to have little effect, which led to a slight reduction of CD206 on Day 1 and little change on Day 3 in the Complete Therapy condition. This data suggests that the particles, R848, and generation of DAMPs are all critical for the shift of BMDMs to a pro-inflammatory phenotype.

To compile changes detected in the flow cytometry markers, we created a scoring system for the degree of inflammation that a population of BMDMs will attain with each treatment (Fig. 4g). We created this scoring system because it is important to understand how the data collected in Figs. 4a–f contribute to the total phenotype of the macrophage population. To do this, we took the log₂ transform of the fold changes in marker expression from each condition, added the transforms for CD86, MHC II, and iNOS (classical inflammatory markers) and subtracted the transforms for CD206 and HIF-1α (anti-inflammatory markers). Scores from Day 1 and Day 3 were added together. In our system, a negative score correlates to a net reduction in inflammatory markers and/or an upregulation in anti-inflammatory markers. In contrast, a positive score signifies a net upregulation in inflammatory markers and/or a reduction in anti-inflammatory markers. We observed that HIFU alone has little effect on the phenotype of BMDMs, whereas particles, DAMPs, and R848 all play a critical role in the large degree of inflammation observed in the Complete Therapy condition, generating a score of 10.2. These data show that HIFU-mediated drug release and DAMP generation from F108-hMSNs is a promising strategy for shifting tumor-associated macrophages to an inflammatory, anti-tumor phenotype.

In addition to measuring changes in phenotype, we performed ELISA on BMDM media to detect interleukin 12 (IL-12), a known inflammatory mediator that is crucial for promoting T cell-mediated immunity (Fig. 4h).³⁵ BMDMs treated with DAMPs from 4T1 lysates or with free R848 secreted around 20 and 65 ng/mL IL-12, respectively. However, BMDMs treated with F108-hMSNs, in both the Particles Only and Complete Therapy conditions, secreted negligible levels of IL-12 for all timepoints. Even though the Complete Therapy condition induced an inflammatory phenotypic shift in BMDMs, the secretion of IL-12 was inhibited. This is likely because F108-hMSNs have been shown to suppress IL-12 secretion from its interaction with scavenger receptor A (CD204), which is reported to recognize amorphous silica.³⁶ IL-12 is specifically suppressed with ligation of scavenger receptors, such as CD204.³⁷ Suppression of IL-12 transcription is calcium-dependent. Therefore, CD204 binding to F108-hMSNs is expected to induce calcium influx into the cell and prevent transcription of IL-12.

Taken together, this data illustrates the strengths of this new combination approach to increase the inflammatory properties of immunological cold tumors, i.e.: the ability of R848@F108-hMSNs to mechanically destroy tumor cells in the presence of HIFU, activate macrophages within a tumor-like microenvironment (and release NO for additional tumor cell killing, as indicated by an increased expression of iNOS), and promote an adaptive immune response through upregulation of CD86 and MHC II. However, one limitation of this treatment is its inability to stimulate IL-12 release, which is known to activate natural killer cells and T cells.^38–40 Although this treatment suppresses IL-12 release from BMDMs, Orr et al. claims that silica binding to CD204 stimulates secretion of other inflammatory cytokines such as TNF-α.³⁶

Conclusion

We showed a combination of chemical and mechanical stimuli using R848-loaded, HIFU-enhancing mesoporous silica nanoparticles offers a promising method of immunotherapy by promoting inflammatory phenotypic expression of macrophages within immunologically cold environments. MSNs were functionalized with an alkyl moiety, which gave rise to both their acoustic activity and ability to load the hydrophobic immunomodulator R848. We determined that F108-hMSNs release R848 faster in the presence of HIFU than when unstimulated. We tested our combination therapy on BMDMs and found that the combination of R848 and DAMP release as a product of F108-hMSN stimulation by HIFU resulted in the most potent anti-inflammatory to pro-inflammatory phenotype shift of all conditions tested. Of the markers investigated, iNOS, MHC II, and CD86 were the primary drivers of the phenotypic change. For this reason, the combination therapy outlined here, and potentially other versions that combine chemical and mechanical agonism, has the potential to initiate an immune response through education of T cells through the T cell receptor complex and cytotoxicity from production of reactive oxygen species through iNOS.