Nanosac, a Non-Cationic and Soft Polyphenol Nanocapsule, Enables Systemic Delivery of siRNA to Solid Tumors

Abstract

For systemic delivery of siRNA to solid tumors, the carrier must circulate avoiding premature degradation, extravasate and penetrate tumors, enter target cells, traffic to the intracellular destination, and release siRNA for gene silencing. However, existing siRNA carriers, which typically exhibit positive charges, fall short of these requirements by a large margin; thus, systemic delivery of siRNA to tumors remains a significant challenge. To overcome the limitations of existing approaches, we have developed a carrier of siRNA, called “Nanosac,” a non-cationic soft polyphenol nanocapsule. A siRNA-loaded Nanosac is produced by sequential coating of mesoporous silica nanoparticles (MSN) with siRNA and polydopamine, followed by removal of the sacrificial MSN core. The Nanosac recruits serum albumin, co-opts caveolae-mediated endocytosis to enter tumor cells, and efficiently silences target genes. The softness of Nanosac improves extravasation and penetration into tumors compared to its hard counterpart. As a carrier of siRNA targeting PD-L1, Nanosac induces a significant attenuation of CT26 tumor growth by immune checkpoint blockade. These results support the utility of Nanosac in the systemic delivery of siRNA for solid tumor therapy.

Small interfering RNA (siRNA) is a short double-stranded RNA, 20-25 base pairs in length, which downregulates specific gene expression by inducing mRNA degradation. Due to the high efficiency and specificity, siRNA has been actively pursued as a therapeutic agent for cancer, viral infections, and autoimmune diseases.^1-4 However, their instability in circulation and inability to enter cells make it difficult to develop effective siRNA therapeutics.^5-7 For in vivo delivery, siRNA is covalently modified or encapsulated in nanoscale carriers, such as cationic liposomes, polymeric nanocarriers, and inorganic particles.^8-10 Recently, two siRNA products received the approval of the U.S. Food and Drug Administration: patisiran (ONPATTRO™), siRNA encapsulated in a lipid nanoparticle (NP), for the treatment of hereditary transthyretin-mediated amyloidosis,¹¹ and givosiran (Givlaari™), siRNA covalently linked to a ligand targeting hepatocytes, for the treatment of acute hepatic porphyria.¹² These developments demonstrate that therapeutic delivery of siRNA is possible when accompanied by carriers that can protect siRNA and facilitate its cellular uptake. However, most existing carriers have shown limited success in systemic delivery of siRNA to the organs other than the liver or lungs.^{13, 14} Without a reliable carrier for systemic delivery, siRNA will remain sidelined in the therapy of undruggable diseases, which would have significantly benefited from its efficiency and specificity otherwise.

Most synthetic carriers of siRNA are cationic.^{15, 16} The positive charge allows the carrier to form an electrostatic complex with siRNA to protect it from nucleases and facilitate cellular uptake. However, it is also the main factor that makes it difficult to deliver siRNA systemically. First, cationic carriers tend to interact with serum proteins non-specifically to form large aggregates that can cause embolism¹⁷ or attract opsonins that subject them to phagocytic clearance.^{13, 18, 19} Moreover, cationic formulations often show intrinsic pro-inflammatory properties, leading to undesirable side effects, such as pulmonary inflammation.^20-22 A stealth coating such as polyethylene glycol can help reduce non-specific protein interactions and attenuate the pro-inflammatory effects, however, at the expense of efficiency in cellular uptake²³ and endosomal escape of the particles.^{24, 25} Systemic delivery of siRNA requires the carriers to reach target organs as intact particles, without causing toxicity and compromising the ability to load, protect siRNA and bring it to cells.

Another issue in nanocarrier-based siRNA delivery to tumors, which is common to most NPs, is the low tumor distribution.²⁶ With poorly organized vasculature and dysfunctional lymphatics, tumors develop greater interstitial fluid pressure than normal tissues, which hinders transvascular convection and intratumoral penetration of a drug.^{27, 28} The compact extracellular matrix of the tumor microenvironment further limits the interstitial drug diffusion.^{29, 30} To improve the intratumoral delivery, a drug may be loaded in a carrier that is disintegrated in the acidity of tumors³¹ or transported by caveolae-mediated endocytosis and transcytosis in tumors.^{32, 33} Alternatively, a drug can be loaded in a flexible carrier that can be deformed during the paracellular transport.^34-38 Recent studies report that soft extracellular vesicles prove superior to its rigid counterparts in intratumoral delivery of doxorubicin³⁹ and hollow nanocapsules with a flexible polysaccharide shell improve interstitial transport of subcutaneously injected mRNA vaccine to lymph nodes.⁴⁰ However, these approaches are yet to be demonstrated in systemic delivery of siRNA.

Given the challenges above, we aim to develop a siRNA carrier for its systemic delivery to tumors. To avoid the issues related to cationic carriers and exploit the advantages of flexible systems, we produce soft and non-cationic nanocapsules, called Nanosac. siRNA is first coated on a sacrificial mesoporous silica nanoparticle (MSN) and covered with polydopamine (pD), whereupon the MSN core is removed to produce a hollow capsule (Fig. 1a). We hypothesize that Nanosac will protect siRNA in circulation and enhance transvascular and intratumoral delivery of siRNA. As a surface coating of NPs, pD has shown to recruit intact albumin in serum and facilitate transendothelial transport and cell uptake of the particles.⁴¹ Therefore, the pD capsule of Nanosac may provide additional benefits to systemic delivery of siRNA. To test these, we evaluate the stability of encapsulated siRNA, uptake pathway and intracellular trafficking, and extravasation and intratumoral transport of Nanosac. The efficiency of Nanosac as a systemic carrier of siRNA is demonstrated with a siRNA targeting immune checkpoint in a syngeneic mouse model of CT26 colorectal carcinoma.

Fig. 1. Preparation and characterization of Nanosac.

(a) Schematic of O/siRNA/pD (Nanosac) preparation. MSN: mesoporous silica nanoparticles; MSN^a: MSN conjugated with (3-Aminopropyl)triethoxysilane (APTES); MSN^a/siRNA: MSN^a with siRNA loaded on the surface; MSN^a/siRNA/pD: MSN^a/siRNA covered with pD; O/siRNA/pD (Nanosac): siRNA-loaded nanocapsules after MSN core removal. (b) Transmission electron micrographs (TEM) of Nanosac and the precursors. Visualized by negative staining with 1% uranyl acetate. Scale bars: 50 nm. (c) Zeta potential and Z-average of Nanosac and the precursors (n = 3 independently and identically prepared batches, mean ± s.d) (d) Force-distance curves for MSN^a/pD and O/pD (Nanosac) measured by AFM, and Young’s moduli of MSN^a/PD and Nanosac (n = 15 tests of a representative batch, mean ± s.d.). ****: p < 0.0001 vs. MSN^a/pD by unpaired t-test.

Results and Discussion

Production of Nanosac

MSN was chosen as a sacrificial template due to the monodispersity, large surface area for siRNA loading, and the established protocol of removal. First, MSNs were synthesized by the sol/gel procedure.⁴² MSNs were spherical and negatively-charged (−39.0 ± 11.3 mV). The average diameter of MSNs was measured to be 67.2 ± 2.7 nm by transmission electron microscopy (TEM) (Fig. 1b, Supporting Fig. 1) and 106.1 ± 0.3 nm by dynamic light scattering (DLS), where the difference may be explained by a mild degree of MSN aggregation during the DLS measurement. Subsequently, the MSNs were modified with (3-aminopropyl)-triethoxysilane (APTES) to form amine-functionalized MSNs (MSN^a) (Supporting Fig. 2). The zeta potential of MSN^a was +16.2 ± 5.9 mV due to the acquisition of amine groups on the surface. Negatively-charged siRNA was adsorbed to MSN^a by electrostatic interaction. The siRNA-loaded MSN (MSN^a/siRNA) showed a negative charge (−41.6 ± 2.1 mV), indicating the binding of siRNA on the MSN^a surface. After the siRNA binding, there was a slight increase in the hydrodynamic diameter (from 106.1 ± 0.3 nm of MSN^a to 137.6 ± 13.6 nm of MSN^a/siRNA) and the polydispersity index (PDI) (from 0.11 of MSN^a to 0.33 of MSN^a/siRNA), as measured by DLS. The slight size increase may be partly due to the addition of siRNA (1.7 wt% as described later); however, given no apparent change in NP morphology by TEM (Fig. 1b), the size increase may again be ascribed to mild aggregation during the siRNA encapsulation. Next, MSN^a/siRNA was coated with polymerized dopamine (pD) layer by 6h incubation in dopamine solution. The pD-coated siRNA-bound MSN (MSN^a/siRNA/pD) showed brown color when dispersed in deionized water, a rough surface in TEM, a slight reduction of negative charge (from −41.6 ± 2.1 mV of MSN^a/siRNA to −25.9 ± 0.8 mV of MSN^a/siRNA/pD), and the increase of z-average (from 137.6 ± 13.6 nm of MSN^a/siRNA to 152.3 ± 2.7 nm of MSN^a/siRNA/pD), indicating the presence of pD shell. Lastly, siRNA-loaded nanocapsules (O/siRNA/pD, Nanosac) were produced by removing the MSN core from the MSN^a/siRNA/pD by hydrogen fluoride (HF) buffered to pH 5 with ammonium fluoride, in which siRNA was confirmed to be stable (Supporting Fig. 3). The resulting Nanosac showed a hollow nanocapsule structure with a z-average of 176.7 ± 2.7 nm and a negative surface charge (−22.6 ± 7.1 mV). The thickness of pD capsule, now clearly visible in TEM, is estimated to be 8.5 ± 1.3 nm, which roughly corresponds to a half of the size difference (14.7 nm) between MSN^a/siRNA and MSN^a/siRNA/pD. In the absence of significant change in thickness, the increased size of Nanosac relative to MSN^a/siRNA/pD is thought to reflect mild aggregation. Fig. 1c summarizes the zeta potentials and z-averages of all NPs measured by DLS. Both MSN^a/siRNA/pD and Nanosac maintained their sizes in 50% fetal bovine serum (FBS) at least for 24 h (Supporting Fig. 4), indicating that it would circulate as nanoparticles without aggregation.

Of note, Nanosac could be collected and washed through low-speed centrifugation (2,000 rcf), despite the small size. The ease of collection may be attributable to the formation of floccules, reversible aggregates of NPs,^43-45 by charge neutralization of NP surface during the etching process (Supporting Fig. 5). After MSN^a/siRNA/pD was treated with cationic ammonium fluoride (2M HF/8M NH₄F, pH 5) for the removal of MSN, the NPs assumed near neutral charges of 10.3 ± 0.5 mV (in the first wash) and 0.027 ± 0.91 mV (in the second wash), forming floccules with an average size of 1401.7 ± 80.4 nm and 558.5 ± 31.7 nm, respectively. NPs after the first wash seemed to have rather a positive charge, likely due to the residual free NH₄⁺ present at the slipping plane where the zeta potential is measured. With additional washing, the Nanosac floccules recovered the negative surface charge and the z-average of <200 nm (180.9 nm after the third wash; 157.1 nm after the fifth wash; Supporting Fig. 6).

The feasibility of storing Nanosac as a lyophilized product was tested by freeze-drying Nanosac with a varying amount of trehalose as a lyoprotectant. Nanosac lyophilized with at as little as 110 wt% trehalose did not aggregate after lyophilization and maintained the particle size of freshly prepared Nanosac when reconstituted, whereas non-protected Nanosac aggregated to form >1 μm particles (Supporting Fig. 7a). The lyophilized Nanosac did not leak siRNA upon gel electrophoresis, indicating that Nanosac remained stable after lyophilization (Supporting Fig. 7b). While all the experiments in the current study were performed with freshly prepared NPs, these results suggest that Nanosac may be lyophilized and stored for later use.

Nanosac is softer than MSN^a/pD.

Nanosac, without the MSN core, was expected to be more flexible than the NPs with the core. To determine the flexibility of NPs, we examined Young’s moduli of MSN^a/pD and Nanosac by atomic force microscopy (AFM). The force-distance curve, a plot of the force measured by the AFM cantilever versus the distance between AFM tip and sample surface, was used to calculate the elastic moduli following the Hertzian model.^46-48 MSN^a/pD and Nanosac showed a significant difference in the slope of the force−distance curve and the calculated Young’s moduli (Fig. 1d). Nanosac exhibited a modulus of 1.6 ± 0.2 MPa, ~17-fold lower than that of MSN^a/pD (17.5 ± 1.4 MPa), indicating that Nanosac is softer than MSN^a/pD. Of note, AFM was performed on dry NPs. The flexibility difference between the two NPs may be even greater in an aqueous medium, in which the Nanosac remain hydrated and swollen.⁴⁸

Nanosac is non-toxic in vitro.

The cytotoxicity of MSN^a, MSN^a/pD, and Nanosac was examined with mouse colon carcinoma CT26 cells. After 48 h incubation with the cells, MSN^a, MSN^a/pD, and Nanosac showed negligible effects on cell viability at a concentration up to 500 μg/mL (higher concentration was not tested nor used) (Supporting Fig. 8a). In contrast, a common gene carrier polyethyleneimine showed dose-dependent toxicity in CT 26 cells, as expected of cationic polymers (Supporting Fig. 8b). The in vitro transfection of siRNA was carried out at 100 μg/mL, where the NPs were not apparently toxic.

Nanosac encapsulates and protects siRNA.

The siRNA binding capacity of MSN^a was evaluated by the agarose gel retardation assay. GAPDH-siRNA (siGAPDH) was complexed with MSN^a for 5 min, varying the weight ratio of siRNA to MSN^a (Supporting Fig. 9a). After the complexation, the supernatant (containing unbound siRNA) and the pellet (containing MSN^a/siGAPDH) were separated by centrifugation at 16,000 rcf and analyzed by agarose gel electrophoresis. The supernatant did not show free siRNA band at a MSN^a/siGAPDH weight ratio of 30 or higher (Supporting Fig. 9a top), suggesting that the siRNA binding capacity of MSN^a was as high as 3.3 wt%. Unlike typical polyplex or lipoplex, the MSN^a/siGAPDH complexes released siRNA upon gel electrophoresis (Supporting Fig. 9a bottom), indicating that the interaction between siRNA and MSN^a was relatively weak, consistent with an earlier study with aminated MSN and CpG oligodeoxynucleotide.⁴⁹ siRNAs targeting luciferase gene (siLuc) and PD-L1 (siPD-L1) showed a similar pattern (Supporting Fig. 9b, c). On the other hand, MSN^a/siRNA/pD and Nanosac, which had additional pD layer on top of siRNA, showed no free siRNA band upon gel electrophoresis (Fig. 2a). This result demonstrates that the pD layer contributed to the stable encapsulation of siRNA. Consistently, we found no sign of siRNA loss during the pD coating, MSN etching and subsequent washing steps, further supporting the protective effect of the pD layer (Supporting Fig. 10). The quantity of siRNA loaded in Nanosac was directly measured by dissolving the pD layer and quantifying the siRNA eluted in gel electrophoresis. The pD layer was partially hydrolyzed in the acidic PBS (pH 3) for 72 h, the minimal time to achieve the maximum siRNA recovery (Supporting Fig. 11a). siRNA was stable in the acidic PBS for 72 h (Supporting Fig. 11b). The siRNA loading capacity of Nanosac (siRNA/Nanosac wt%) used in most studies was estimated to be 1.7 wt%.

Fig. 2. Stability of Nanosac.

(a) Gel electrophoresis of siLuc, MSN^a/siLuc, MSN^a/siLuc/pD, and Nanosac with/without RNase or SDS challenge; (b) Gene silencing by MSN^a/siLuc, MSN^a/siLuc/pD, and Nanosac with/without RNase challenge (n = 3 test of a representative batch, mean ± s.d.). ***: p < 0.001; ns: not significant by Sidak’s multiple comparisons test following one-way ANOVA. Gene silencing by MSN^a/siRNA, MSN^a/siRNA/pD, and Nanosac, measured after 48 h treatment in complete medium. (c) siGAPDH in CT26 cells, (d) siLuc in 4T1-Luc cells, and (e) siPD-L1 in IFN-γ-activated CT26 cells. (n = 3 test of a representative batch, mean ± s.d.). ***: p < 0.001, ****: p < 0.0001 vs. No siRNA by Dunnett’s multiple comparisons test following two-way ANOVA. (f) Top: Representative western blot of PD-L1 expression in IFN-γ-activated CT26 cells by MSN^a/siPD-L1, MSN^a/siPD-L1/pD, and Nanosac. Bottom: Quantitative presentation of western blotting.

To test if the pD layer protects siRNA from enzymatic challenge and anionic environment in physiological fluids, MSN^a/siRNA, MSN^a/siRNA/pD, and Nanosac were subjected to ribonuclease A (RNase) 100 μg/mL ± 1% SDS (Fig. 2a) or 50% FBS. Free siLuc was completely degraded by RNase (Fig. 2a, Lane 2). MSN^a/siLuc, where siLuc was loosely bound to MSN^a (Supporting Fig. 9b; Fig. 2a, Lane 3), had a fraction of siLuc in the well (as a MSN^a/siLuc), but the loosely-bound siLuc was degraded by RNase treatment (Fig. 2a, Lane 4). Interestingly, 1% SDS-treated MSN^a/siLuc (Fig. 2a, Lanes 5 and 6) showed no additional signs of instability compared to SDS-non-treated MSN^a/siLuc (Fig. 2a, Lanes 3 and 4), indicating that electrostatic forces played only a partial role in the interaction between MSN^a and siLuc. This suggests that other forces less affected by 1% SDS, such as hydrogen bonding,⁵⁰ contributed to the siRNA loading on MSN^a. Hydrogen bonding is attributable to -SiOH and -NH₂ of MSN^a and consistent with the loose binding (Supporting Fig. 9a bottom). On the other hand, MSN^a/siLuc/pD and Nanosac resisted RNase challenge ± 1% SDS treatment, indicating that pD layer prevented the detachment and degradation of siRNA (Fig. 2a, Lanes 7-10). (The MSN^a/siLuc/pD was located in the gel well in the visible image of the gel, Supporting Fig. 12a). Consistently, MSN^a/siPD-L1/pD and Nanosac remained stable in 50% FBS, showing no signs of degradation due to serum nucleases, whereas siPD-L1 that was free or loosely bound to MSN^a/siPD-L1 was completely degraded (Supporting Fig. 12b). To verify the protection of siRNA from RNase by pD layer, we examined the in vitro transfection efficiency of MSN^a/siLuc, MSN^a/siLuc/pD, or Nanosac with and without RNase treatment (Fig. 2b). All NPs showed significant silencing of luciferase expression (14.2 %, 14.8 %, and 27.1% of the respective non-treated control) in 4T1-Luc cells. After RNase treatment, MSN^a/siLuc showed no significant silencing of the luciferase expression, whereas MSN^a/siLuc/pD and Nanosac maintained the silencing effect. This result is consistent with the gel electrophoresis and confirms that pD layer could protect siRNA from nuclease degradation. The protective effect of pD was preserved through the MSN removal process, as indicated by gel electrophoresis and gene silencing of the RNase-treated Nanosac.

Nanosac silences gene expression in vitro.

In vitro gene silencing by Nanosac and its precursors (MSN^a/siRNA, MSN^a/siRNA/pD) was evaluated with three siRNAs (siGAPDH, siLuc, and siPD-L1) with a non-specific siRNA (siCont) as a control. All NPs containing siGAPDH showed significant inhibition of GAPDH expression in CT26 cells, whereas those with siCont showed no difference from blank NPs (Fig. 2c). MSN^a/siGAPDH, MSN^a/siGAPDH/pD, and Nanosac reduced the GAPDH expression to 48.9, 48.9 and 49.4%, respectively, whereas the NPs with no siRNA or siCont induced no GAPDH silencing. A similar trend was observed with siLuc in 4T1-Luc cells (Fig. 2d). All NPs with siLuc exhibited excellent luciferase silencing effects compared to NPs with no siRNA or siCont. These results demonstrate that the additional pD layer and the absence of MSN core did not negatively affect siRNA delivery to the tested cells. In preparation for in vivo studies described later, the gene silencing effect was tested with siPD-L1 in CT26 cells. CT26 cells were incubated with 100 ng/mL of interferon-γ (IFN-γ), a condition that induces immunosuppression in tumor microenvironment,^{51, 52} to induce PD-L1 expression, prior to the NP treatment. MSN^a/siPD-L1, MSN^a/siPD-L1/pD, and Nanosac, at a concentration equivalent to 200 nM siPD-L1, reduced the PD-L1 expression by 50.5, 74.2, and 75.6% compared to blank NPs in the IFN-γ-treated CT26 cells, while those with siCont showed no silencing effect (Fig. 2e). A similar trend was shown in Western blots of the cell lysates (Fig. 2f). The silencing effect was dose-dependent. The cells treated with MSN^a/siPD-L1/pD and Nanosac at a concentration equivalent to 100 nM of siPD-L1 showed the same level of siPD-L1 expression as no-IFN-γ-treated control cells. At 200 nM siPD-L1, MSN^a/siPD-L1/pD and Nanosac further reduced siPD-L1 expression to below the basal level (Supporting Fig. 13). MSN^a/siPD-L1/pD and Nanosac showed greater silencing effects than MSN^a/siPD-L1 at all concentrations, suggesting a positive role of pD coating, including the protective effect shown in Fig. 2a. This difference was not evident with siGAPDH or siLuc. Given the dose-dependence, we speculate that siGAPDH or siLuc may have been more efficient than siPD-L1 and all three particles reached the maximum silencing effect at the dose tested in this study (100 nM for siGAPDH; 150 nM siLuc).

Nanosac delivers siRNA to CT26 cells via caveolae-mediated endocytosis.

Gene silencing is the indirect evidence of efficient intracellular delivery of Nanosac. To verify intracellular delivery and understand the mechanism, we examined the uptake of NPs containing cy3-labeled siRNA (MSN^a/siRNA-cy3, MSN^a/siRNA-cy3/pD, and Nanosac) by CT26 cells by confocal microscopy. Consistent with gene silencing, all three NPs entered CT26 cells in 24 h (Supporting Fig. 14). Since none of the NPs has a specific ligand to facilitate their cellular uptake, we suspected that serum proteins in the medium adsorbed to the NPs forming a protein corona to mediate cellular uptake.

To test this, we incubated CT26 cells with the NPs under conditions blocking specific endocytosis pathways and analyzed by fluorospectrometry. MSN^a was first labeled with Cy5 (Supporting Fig. 15) for detection and quantification of uptake. The dye was confirmed to be stable through the dissolution of pD layer or the removal of MSN (Supporting Fig. 16). CT26 cells were incubated at 4 °C for 30 min to inhibit energy-dependent endocytosis or pretreated with inhibitors of endocytosis pathways, such as chlorpromazine (inhibitor of clathrin-mediated endocytosis), methyl-β-cyclodextrin (inhibitor of caveolae-mediated endocytosis), and amiloride hydrochloride (inhibitor of macropinocytosis) for 30 min at a subtoxic concentration for each compound (Supporting Fig. 17). The cellular uptake of all three NPs in serum-containing medium was blocked at 4 °C (Fig. 3a), which suggests that all NPs used energy-dependent internalization pathways. The uptake of MSN^a-cy5 (without pD) in serum-containing medium was significantly inhibited by chlorpromazine, indicating that the naked MSN^a used a clathrin-mediated endocytosis pathway. In contrast, the uptake of pD-coated NPs (MSN^a-cy5/pD and Nanosac) was significantly inhibited by methyl-β-cyclodextrin, but not by chlorpromazine or amiloride hydrochloride, indicating that the pD-coated NPs were mainly internalized by the cells via caveolae-mediated endocytosis. These results suggest that pD layer recruit proteins in the serum-containing medium in a distinct manner than the naked MSN^a to enable caveolae-mediated endocytosis. In the serum-free medium, cellular uptake of MSN-cy5/pD was not affected by methyl-β-cyclodextrin (nor by other inhibitors) (Supporting Fig. 18), supporting the putative role of serum proteins in endocytosis of the pD-coated NPs. Of note, the NP uptake in serum-containing medium was not completely reduced by the tested endocytosis inhibitors to the same level as that in 4 °C (Fig. 3a), and MSN^a-cy5/pD uptake in serum-free medium was still affected by the low temperature (Supporting Fig. 18). This suggests that the NPs may also enter CT26 cells by additional pathways, such as clathrin-/caveolae-independent endocytosis,⁵³ irrespective of the protein corona.

Fig. 3. Endocytosis pathway of NPs.

(a) Effects of endocytosis inhibitors on NP uptake in serum-supplemented medium. CT26 cells were preincubated for 30 min with each inhibitor at a subtoxic concentration, followed by treatment of NPs in 10% FBS-supplemented medium for 2 h (n = 3 tests of a representative batch, mean ± s.d.). ****: p < 0.0001, ns: not significant vs. 37 °C by Dunnett’s multiple comparisons test following two-way ANOVA. CPZ: chloropromazine. Albuminylation of NPs. (b) Zeta potential of MSN^a, MSN^a/pD, and Nanosac before and after incubation in 50% FBS (n = 3 tests of a representative batch, mean ± s.d.). (c) SDS-PAGE of protein corona composition formed on MSN^a, MSN^a/pD, and Nanosac. NPs (4 mg/mL) were incubated in 50% FBS for 2 h and rinsed with PBS twice. (d) Spectral counts of proteins, analyzed by LC-MS/MS, bound on the MSN^a, MSN^a/pD, and Nanosac after 2 h exposure to 50% FBS. (e) Representative SDS-PAGE gel of albumin after pulse proteolysis. Native albumin (nAlb), denatured albumin (dAlb), MSN^a incubated with albumin (MSN^a+Alb) and MSN^a/pD with albumin (MSN^a/pD+Alb) were treated with thermolysin for 3 min. Lane 1: nAlb; Lane 2: dAlb; Lane 3: MSN^a+Alb; Lane 4: MSN^a/pD+Alb. % digestion albumin was defined as (1-albumin band intensity after proteolysis/albumin band intensity before proteolysis) × 100. n = 3 independently and identically performed experiments (mean ± s.d.). ***: p < 0.001 and ****: p < 0.0001 vs. nAlb by Dunnett’s multiple comparisons test following one-way ANOVA. Intracellular trafficking of NPs. (f) Confocal microscope images locating cy5-labeled MSN^a, MSN^a/pD, and Nanosac relative to lysosomes in CT26 cells. Green: Lysotracker (lysosome); Red: cy5-labeled NPs; Blue: Hoechst 33342 (nuclei). Scale bars: 10 μm. (g) Pearson’s correlation coefficients indicating the degree of NP/lysosome colocalization in confocal images: R=1 (perfect colocalization), R=0 (no colocalization). n = 5 tests of a representative batch (mean ± s.d). ***: p < 0.001 and ****: p < 0.0001 by Tukey's multiple comparisons test following one-way ANOVA. (h) Fluorescence intensity profiles along the white lines in (f). (i) Release kinetics of siRNA from Nanosac, performed in different pHs or H₂O₂ (100 μM) with constant agitation at 37 °C. n=3 tests with representative batches (mean ± s.d.).

Albumin binding to Nanosac mediates its cellular uptake.

To identify the serum proteins responsible for differential cellular uptake profiles, we incubated MSN^a and pD-coated NPs (MSN^a/pD and Nanosac) with 50% FBS, rinsed twice, and analyzed by SDS-PAGE and LC-MS/MS. All NPs showed a reduction of zeta potential upon the incubation with 50% FBS (Fig. 3b), reflecting protein corona formation. MSN^a displayed the most change, suggesting a relatively high protein binding per NP, which was confirmed by SDS-PAGE band intensity (Fig. 3c) and MS/MS counts (Fig. 3d). Albumin (66 kDa) binding was prominent for all NPs, whereas MSN^a showed additional protein binding (100-250 kDa). LC-MS/MS analysis verified that the albumin was the major corona protein for all NPs. The additional proteins bound to MSN^a included complement c3, inter-α-trypsin, vimentin, angiotensinogen, SERPIN domain-containing protein, and GLOBIN domain-containing protein. The differential pattern of protein corona may be one explanation for the difference between MSN^a and the pD-coated NPs in cellular uptake profile. Since our previous study showed that the conformation of surface-bound albumin could also affect the endocytosis pathway,⁴¹ we examined the status of albumin bound on MSN^a and MSN^a/pD by the thermolysin proteolysis, which digests unfolded proteins (Fig. 3e). The band intensity of denatured albumin and the albumin bound to MSN^a-cy5 reduced by 81.5 ± 4.1 % and 50.4 ± 3.0 %, respectively, upon the pulse proteolysis, whereas native albumin underwent 12.1 ± 5.8 % of proteolysis. The albumin bound to MSN^a/pD showed a similar degree of digestion (15.2 ± 8.6 %) as native albumin. This result indicates that albumin conformation was better preserved on MSN^a/pD than on MSN^a.

Nanosac traffics to cytosol and releases siRNA by ROS.

The results so far collectively suggest that the pD-coated NPs enter CT26 cells via the caveolae-mediated pathway due to the surface-bound albumin, at least partly. To compare intracellular trafficking, we located Nanosac and its precursors in the cells relative to a lysosome marker (LysoTracker Green) (Fig. 3f). MSN^a-cy5 was colocalized with LysoTracker Green (Fig. 3h) with a Pearson’s correlation coefficient of 0.83 (Fig. 3g), indicating a high degree of colocalization of MSN^a-cy5 and lysosomes, i.e., that most of the internalized MSN^a-cy5 was transported to lysosomes. This was comparable to siRNA-cy3 complexed with Lipofectamine 2000 (Lipofectamine/siRNA-cy3), a common transfection reagent known to enter cells by the clathrin-mediated endocytosis (Supporting Fig. 19). In contrast, only a few MSN^a-cy5/pD and Nanosac were overlapped with lysosomes (Fig. 3h) with a Pearson’s correlation coefficient of 0.15 – 0.3 (Fig. 3g), showing that the caveolae-mediated endocytosis facilitated cytosolic delivery of the pD-coated NPs avoiding destructive lysosomal sequestration.

The pD layer of Nanosac degrades in acidic conditions releasing siRNA, as shown in the determination of siRNA loading (pH 3, Fig. 3i, Supporting Fig. 20) and release studies performed at pH 5.2 and 6.2. However, when Nanosac directly traffics to cytosol, the release of siRNA from Nanosac may not be explained by the acidic pH of lysosomes. Instead, we suspect that reactive oxygen species (ROS), known to be elevated in malignant cells,⁵⁴ may facilitate intracellular release of siRNA. To test this, we incubated Nanosac in 100 μM H₂O₂ (pH 7.4) solution⁵⁵ at 37 °C with agitation. Nanosac released more siRNA in the presence of H₂O₂ (Fig. 3i, Supporting Fig. 20), suggesting that cytosolic ROS may account, at least partly, for the release of siRNA from the cytosol-trafficked Nanosac and subsequent gene silencing.

Softness of Nanosac helps reduce macrophage uptake without affecting the interactions with tumor cells or endothelial cells.

Systemically injected drug carriers first encounter immune cells or macrophages, which destroy therapeutic siRNA before reaching the target.^56-60 Surviving carriers are translocated across the endothelial layer to enter tumors. Ideal drug carriers should avoid the recognition and internalization by macrophages but efficiently interact with peritumoral endothelium. We examined if MSN^a-cy5/pD and Nanosac, which entered CT26 cells similarly well (Supporting Figs. 14, 21), show any difference in the uptake by macrophages and endothelial cells due to the differential flexibility (Fig. 1d). MSN^a-cy5/pD and Nanosac were incubated with J774A.1 macrophages and HUVECs. Unlike CT26 cells, macrophages took up Nanosac 2-fold less than MSN^a-cy5/pD in 30 min and 2 h (Fig. 4a), indicating that the softness of Nanosac helped avoid macrophage uptake. Confocal microscopy showed a consistent result. In contrast, MSN^a-cy5/pD and Nanosac entered HUVEC, both resting and activated by TNF-α (simulating peritumoral endothelium), with comparable efficiency (Supporting Fig. 21), as they did CT26 cells. We then examined if the NPs can translocate through the endothelial layer. NPs (MSN^a-cy5, MSN^a-cy5/pD and Nanosac) were added to the apical side of the Transwell insert, in which a confluent HUVEC layer was activated by TNF-α, incubated for 6 h, and quantified in the media of both apical and basolateral sides. MSN^a-cy5/pD and Nanosac were comparable in crossing the activated HUVEC layer (MSN^a-cy5/pD: 16.6 ± 2.4% vs. Nanosac: 18.3 ± 2.0%, Supporting Fig. 22). Interestingly, the extents of transendothelial transport of these two NPs were greater than that of MSN^a-cy5 (7.9 ± 0.8%), suggesting that the albumin bound on the pD surface may contribute to the transport of MSN^a-cy5/pD and Nanosac across the HUVEC layer.

Fig. 4. Comparison of MSN^a-cy5/pD and Nanosac in macrophage uptake, extravasation, and tumor spheroid penetration.

(a) Quantitative measurement and confocal microscope images of J774a.1 macrophage taking up MSN^a-cy5/pD and Nanosac. n = 3 tests of a representative batch (mean ± s.d). ***: p < 0.001 and ****: p < 0.0001 by Sidak's multiple comparisons test following two-way ANOVA. Green: Wheat Germ Agglutinin (cell membrane); Red: cy5-labeled NPs; Blue: Hoechst 33342 (nuclei). Scale bar: 50 μm. (b) Time-lapse intravital microscopic images of MSN^a-cy5/pD and Nanosac circulating in CT26 tumor-bearing BALB/c mice. Green: Dextran-FITC (locating blood vessel), Red: cy5-labeled NPs. (c) Z-section images of CT26 tumor spheroids incubated with MSN^a-cy5/pD or Nanosac. Scale bars: 500 μm. (d) Spheroid depth-wise fluorescence intensity profiles. n = 3 tests of a representative batch (mean ± s.d). ****: p < 0.0001 vs. MSN^a-cy5/pD by Sidak’s multiple comparisons test following two-way ANOVA. (e) Horizontal fluorescence intensity profile at 160 μm.

Nanosac extravasates and penetrates into tumors better than hard counterpart.

We examined the real-time intratumoral delivery of MSN^a-cy5/pD and Nanosac via intravital confocal microscopy using CT26 tumor-bearing mice with a dorsal window chamber. When tumors grew to 30 mm³, FITC-dextran (2000 kDa) was injected by intravenous (IV) injection to locate tumor vessels, followed by an IV injection of MSN^a-cy5/pD and Nanosac. The Nanosac was initially seen within vessels and then gradually extravasated and dissipated into the extravascular regions in 1 h (Fig. 4b and Supporting mov. 1). In contrast, MSN^a-cy5/pD continued to be retained in the vessel gradually losing the intensity (i.e., cleared from the system) over 1 h with little extravasation (Supporting mov. 2).

Intravital microscopy also shows that Nanosac traveled further into tumors than MSN^a-cy5/pD. To compare the ability of MSN^a-cy5/pD and Nanosac to travel in tumors quantitatively, we incubated the NPs (1 mg/mL) with CT26 tumor spheroids, 500 μm in diameter, for 4 h. Nanosac showed greater tumor penetration and accumulation than MSN^a-cy5/pD (Fig. 4c, Supporting Fig. 23). The mean fluorescence intensity (MFI) of Nanosac in the spheroid was 2-fold higher than that of MSN^a-cy5/pD (Fig. 4d). Of note, both NPs showed the peak MFI at 160 μm and decreasing MFI thereafter, which is likely due to the attenuation of fluorescence signal with the increasing depth. When compared at the same penetration depth (160 μm), the spheroid incubated with Nanosac showed stronger MFI inside than MSN^a-cy5/pD (Fig. 4e), indicating that the soft Nanosac penetrated into the spheroid better than the hard counterpart.

siPD-L1-loaded Nanosac attenuates CT26 tumor growth via immune checkpoint blockade.

We used Nanosac to deliver siPD-L1 in CT26 colon tumor-bearing Balb/c mice for inhibiting PD-1/PD-L1 immune checkpoint interaction in tumors. The siPD-L1-loaded Nanosac or MSN^a/siPD-L1/pD were administered IV at a dose equivalent to siPD-L1 0.75 mg/kg/time 10 times every 2 days (q2d×10) via tail vein injection. The Nanosac-treated group showed a significant attenuation in tumor growth as compared to the 5% dextrose (D5W)-treated group (p < 0.0001) and MSN^a/siPD-L1/pD-treated group (p < 0.05) (Fig. 5a) with no weight loss (Fig. 5b). The PD-L1 expression in CT26 tumor of the Nanosac-treated group was significantly less than those of D5W- and MSN^a/siPD-L1/pD-treated groups (Fig. 5c, Supporting Fig. 24). To check if the silencing effect of siPD-L1 induced an immune checkpoint blockade, we examined the T-cell populations in tumor-draining lymph nodes (TDLNs). The Nanosac-treated group had more CD8⁺ T-cells and fewer CD4⁺ T-cells in TDLNs than the D5W-treated group (Fig. 5d, Supporting Fig. 25). MSN^a/siPD-L1/pD showed no effect on CD8⁺ T cell population and a slight reduction in CD4⁺ T cells compared to the D5W treatment. Consequently, the CD8⁺/CD4⁺ ratio in TDLNs of the Nanosac group was significantly higher than those of the D5W- and MSN^a/siPD-L1/pD-treated groups (Fig. 5d). This result indicates that siPD-L1-loaded Nanosac enhanced tumor infiltration of CD8⁺ T cells, while MSN^a/siPD-L1/pD with an equivalent dose of siPD-L1 was not as effective. The antitumor effect study was repeated with a higher unit dose of siPD-L1 (1.5 mg/kg/time, q2d×7). In this case, the growth of CT26 tumors was significantly delayed by both Nanosac and MSN^a/siPD-L1/pD compared to the D5W control, with the corresponding reduction in PD-L1 expression in CT26 tumor (Supporting Fig. 31). MSN^a/siCont/pD-treated group showed no significant difference from the D5W-treated group in both tumor growth and PD-L1 expression, indicating that the effect of Nanosac or MSN^a/siPD-L1/pD was not attributable to the carriers but to siPD-L1. In both regimens, none of the animals experienced significant weight loss throughout the study period (Fig. 5b, Supporting Fig. 31). Finally, we compared the anti-tumor efficacy of Nanosac (1.5 mg/kg/time, q2d×5) with the clinically approved checkpoint blockade agent, anti-PD-L1 antibody, administered in a typical preclinical regimen (intraperitoneal injection at a dose of 10 mg/kg/time, q2d×5)^{61, 62} in another set of CT26 tumor-bearing Balb/c mice. The Nanosac-treated group attenuated tumor growth significantly as compared to the D5W-treated group and anti-PD-L1 antibody-treated group with no weight loss (Fig. 6).

Fig. 5. Anti-tumor activity of 5% dextrose (D5W), MSN^a/siPD-L1/pD, and Nanosac in Balb/c mice bearing CT26 tumors (siPD-L1: 0.75 mg/kg/time, q2d×10).

(a) Average tumor size (mm³). (b) Average body weight (g, grams). *: p < 0.05 and **: p < 0.01 between average tumor sizes on day 20 post-first injection by Tukey's multiple comparisons test following two-way repeated measures ANOVA. Arrowheads indicate times of treatment. n = 5 mice per treatment (mean ± s.d). (c) Expression of PD-L1 in CT26 tumors (Supporting Fig. 24). n = 4 mice per group (mean ± s.d). *: p < 0.05; ns: not significant vs. D5W by Dunnett's multiple comparisons test following one-way ANOVA. (d) %CD8⁺ cells, %CD4⁺ cells, and CD8⁺/CD4⁺ ratio in TDLNs of treated animals. n = 5 mice per treatment (mean ± s.d). *: p < 0.05, ***: p < 0.001, ****: p < 0.0001, and ns: not significant vs. D5W by Dunnett's multiple comparisons test following one-way ANOVA. (e) Fluorescence micrographs of tumor sections showing FITC-lectin-stained vessels (green) and MSN^a/siRNA-cy5/pD or Nanosac (red) at 24 h from IV injection. Scale bars: 50 μm. See Supporting Fig. 27 for additional micrographs. (f) quantitative analysis of micrographs in (e): % NPs departing from the lectin-positive endothelial cells was calculated as the area of free NPs (red) divided by the area of the total NP fluorescence (free NPs and NPs overlapping with endothelial cells: red + yellow). Three fields were randomly selected and analyzed by the Nikon A1R confocal microscope analysis software. (g) Photomicrographs of hematoxylin and eosin (H&E)-stained liver and spleen sections. No significant lesions were observed in either organ microscopically examined in all treatment groups. See Supporting Fig. 28 and 29 for high magnification photomicrographs.

Fig. 6. Anti-tumor activity of D5W, anti-PD-L1 antibody, and Nanosac in Balb/c mice bearing CT26 tumors.

(anti-PD-L1 antibody: 200 μg/mouse/time, intraperitoneal injection; siPD-L1: 1.5 mg/kg/time, IV injection; q2d×5). *: p < 0.05, **: p < 0.01, and ***: p < 0.001 between average tumor sizes on day 20 post-first injection by Tukey's multiple comparisons test following two-way repeated measures ANOVA. n = 5 mice per group (mean ± s.d).

Softness benefits tissue-level distribution of NPs.

To understand the mechanism of superior antitumor effect of Nanosac, biodistribution of Nanosac and MSN^a/siRNA/pD was examined in Balb/c mice with CT26 tumors using siRNA-cy5 (siRNA 0.75 mg/kg). When examined 24 h after injection, the two NPs showed no significant difference in siRNA distribution in major organs (Supporting Fig. 26). The lack of difference in liver distribution between the two NPs is intriguing, given the significant difference in macrophage uptake in vitro (Fig. 4a). This may be partly explained by that the dose used in this study corresponds to less than one trillion particles, identified to be the threshold to overwhelm the Kupffer cells⁶³; thus, neither NPs may have reached a critical point to reflect their difference in macrophage uptake on the biodistribution in the liver. Likewise, tumor distribution of the two NPs was comparable. Therefore, we repeated the same experiment in another set of animals to observe tissue-level NP distribution in tumors 24 h, locating siRNA-cy5 signals (red) relative to FITC-lectin-stained blood vessels (green). The fraction of Nanosac departing from the vessels (i.e., entering tumor interstitium), quantified as free NP (red) / total NP (red + yellow), was greater than that of MSN^a/siRNA/pD (Fig. 5e, 5f, Supporting Fig. 27), which is consistent with intravital microscopy (Fig. 4b, Supporting Mov. 1, 2). These results suggest that the softness of Nanosac benefit the NP distribution on the tissue level rather than the organ level at the dose tested in this study.

Since the biodistribution study indicated that NPs accumulated mainly in the liver and spleen (Supporting Fig. 26), we examined representative tissue samples from liver and spleen in animals treated in anti-tumor efficacy studies (Fig. 5a, 5b, Supporting Fig. 31a, 31b) to assess the effects of repeatedly administered Nanosac and MSN^a/siPD-L1/pD. None of the animals showed significant lesions in either organ microscopically examined (Fig. 5g). Nevertheless, MSN^a/siPD-L1/pD and Nanosac induced a subtle difference in tissue architecture that reflected the effect of softness on the NP transport. In the liver sections, MSN^a/siPD-L1/pD (Fig. 5g-2) and Nanosac (Fig. 5g-3) accumulated as multifocal randomly distributed brown pigment clusters, which were found to be NPs laden in Kupffer cells in high magnification photomicrographs (arrowheads, Supporting Fig. 28). In most animals, the NP-laden macrophages in the liver were accompanied by mild to moderate numbers of neutrophils and lymphocytes with no associated hepatocellular lesions. However, in one of the MSN^a/siPD-L1/pD treated animals, a portal region of liver demonstrated a severe lesion composed of abundant clusters of macrophages loaded with NPs admixed with neutrophils and other inflammatory cells with hepatocellular dropout and necrosis. The spleens of animals receiving MSN^a/siPD-L1/pD and Nanosac (Fig. 5e-5, e-6, Supporting Fig. 29c-f) had a less prominent zonal pattern within the white pulp (white boxes, Supporting Fig. 29c,e) due to the expansion of the sinusoidal spaces by the macrophages filled with intracytoplasmic NPs (arrowheads, Supporting Fig. 29d,f) and occasionally mildly-increased apoptosis of lymphocytes. Interestingly, the MSN^a/siPD-L1/pD treated animals had marked infiltration of macrophages filled with the NPs in perihepatic lymph nodes, whereas the Nanosac-treated animals showed no signs of NP accumulation (Supporting Fig. 30). This observation suggests that, even though Nanosac did not completely evade the macrophage uptake in the liver and spleen, it may have had greater agility in passing through these organs, consistent with the literature.^{64, 65} The contrast between the two NPs in liver or spleen distribution is even clearer in the animals treated with a higher dose. Here, MSN^a/siRNA/pD (MSN^a/siCont/pD and MSN^a/siPD-L1/pD) demonstrated noticeable intrahepatic accumulation within large clumps of macrophages (arrowheads, Supporting Fig. 31d-4, 31d-6), and less densely and more dispersed in Nanosac-treated livers (arrowheads, Supporting Fig. 31d-8). Consistently, NP-laden macrophages were often present in the spleen of the MSN^a/siRNA/pD groups but sparsely shown in the Nanosac-treated animals (arrowheads, Supporting Fig. 31d-12, 31d-14, 31d-16). The histological evaluation suggests that, even though the initial organ-level distribution appears similar between MSN^a/siPD-L1/pD and Nanosac, the cumulative effect of multiple injection is more favorable with Nanosac, due to the advantages in tissue level and cell level distribution of soft particles.

The results shown so far support that Nanosac can facilitate systemic delivery of siRNA to solid tumors. Here, Nanosac offered two features, non-cationic surface and softness, which are particularly beneficial for systemic delivery of siRNA to tumors. Nanosac did not carry positive charges, thus avoiding toxicity⁶⁶ and non-specific protein adsorption leading to NP aggregation and capillary entrapment.⁶⁷ This enabled IV administration of siRNA-loaded Nanosac in a dose sufficient to achieve a therapeutic response. Softness enhanced transvascular and interstitial delivery of Nanosac to tumors, consistent with earlier studies.^{37, 39, 68} Given that Nanosac showed no difference from the hard counterpart in the endothelial interaction (Supporting Fig. 21) and transendothelial movement (Supporting Fig. 22) in a static in vitro condition, we speculate that Nanosac may have exploited transient vascular openings occurring in vivo⁶⁹ in extravasation at tumors, although it is unclear how softness may have benefited that process. The improved intratumoral penetration of Nanosac may be explained by the deformability in the tumor interstitium. Supporting this possibility, a study with polymeric NPs of varying rigidities has shown that soft NPs translocate a porous membrane and penetrate into collagen gel more easily than hard NPs.³⁸ Softness also helped reduce macrophage uptake of Nanosac (Fig. 4a). A recent report of silica nanocapsules with controlled softness attributes the reduced phagocytosis of soft particles to their deformation induced by actin-rich cell protrusions, which slows down the particle uptake.⁷⁰ For these reasons, softness has been one of the motivations to pursue cell-derived particles such as extracellular vesicles as a drug carrier.³⁹ While sharing the advantage of soft NPs, Nanosac avoids challenges related to cell-derived vesicles, such as scalability, heterogeneity, control of payload loading, immunogenicity, and potential risk of disease transfer.⁷¹

The pD surface of Nanosac has brought at least three additional benefits toward siRNA delivery to tumors based on its interaction with serum albumin (Fig. 3c, d). First, it allows Nanosac to interact with cells in distinct mechanisms. While Nanosac showed lower macrophage uptake than the hard counterpart in vitro, its uptake by CT26 cells and endothelial cells was unaffected by the softness (Supporting Fig. 21). We speculate that these cells may have interacted with Nanosac via the albumin on the pD surface. We previously reported that pD-coated polymeric NPs selectively bound to albumin preserving its native conformation.⁴¹ With the albumin-bound (albuminylated) surface, the pD-coated polymeric NPs took advantage of albumin-mediated interactions with endothelial cells and tumor cells, to achieve greater drug delivery to tumors than control NPs that had non-specific, denaturing interactions with serum proteins.⁴¹ The surface-bound albumin may have exploited increasing demand of cancer cells for albumin as a source of energy and nutrients^{72, 73} to enhance cancer cell uptake of NPs and interacted with glycoproteins expressed on peritumoral endothelium.^74-77 Nanosac, with the pD surface, was albuminylated in serum (Fig. 3d); therefore, it may have benefited from the surface-bound albumin in the interaction with tumor cells and endothelial cells in the same manner as the pD-coated polymeric NPs.⁴¹

The pD coated NPs (MSN^a-cy5/pD and Nanosac) were transported across the endothelial layer better than MSN^a-cy5 with no pD coating (Supporting Fig. 22). This suggests that albuminylation may also have contributed to the extravasation of Nanosac via the albumin-mediated interaction with endothelial cells, leveraging the effect of softness. In the same vein, a recent study with gold NPs has suggested that, in the absence of the enhanced permeability and retention effect, an active transendothelial pathway, possibly the caveolae-mediated transcytosis via the albumin corona, may account for the NP extravasation into tumors.⁷⁸ On the other hand, it is worth reminding that albumin has long been used as a dysopsonin to reduce the binding of other proteins on NPs,^79-82 thereby extending the circulation time of NPs.^{79, 80} The multiple roles of albumin suggest that the surface-bound albumin may enhance both tumor accumulation and cellular interaction of NPs. This distinguishes the in-situ adoption of albumin from traditional PEGylation, which has shown limitations in pursuing both.²³

Moreover, the albuminylated Nanosac took caveolae-mediated endocytosis to enter CT26 cells (Fig. 3a), consistent with the uptake pathway of albumin⁸³ and albumin-bound drugs.^{84, 85} The uptake pathway has an important implication in the intracellular delivery of siRNA. While the ultimate fate of the internalized caveolae remains controversial⁸⁶ (fuse with endosomes and follow the degradative pathway⁸⁷or take alternative cellular compartments to avoid degradative pathway like viruses do⁸⁸), our microscopic observation indicates that Nanosac avoids the sequestration in acidic lysosomal compartments and traffics to the cytosol (Fig. 3f), where it unloads siRNA likely by intracellular ROS (Fig. 3i) to achieve efficient gene silencing (Fig. 2). This intracellular trafficking pathway allows Nanosac-encapsulated siRNA to avoid lysosomal degradation, a significant advantage for siRNA delivery. Of note, the uptake and trafficking of NPs appear to be influenced by the quality and purity than the quantity of the surface-bound albumin. As shown in SDS-PAGE and proteomics analysis of the protein corona, MSN^a captured more albumin than MSN^a/pD or Nanosac (Figs. 3c, 3d) but trafficked predominantly to lysosomes (Fig. 3f), which may be attributable to the additional proteins interfering with albumin (Fig. 3d) and the denaturation of the surface-bound albumin (Fig. 3e). This observation underscores the advantage of pD as a non-denaturing surface for selective albumin adsorption.

Conclusion

We have developed Nanosac, soft non-cationic nanocapsules, for systemic delivery of siRNA. Nanosac is produced by sequential attachment of siRNA and polydopamine on a sacrificial MSN core, followed by the removal of the MSN. Encapsulating siRNA in the capsules, Nanosac avoids the issues common to cationic gene carriers, such as toxicity and non-specific protein binding while protecting siRNA from RNase. Nanosac entered tumor cells by caveolae-mediated endocytosis, likely via albumin recruited from serum, trafficked to the cytosol, and silenced target genes. Due to the softness, Nanosac showed lower macrophage uptake, greater extravasation and penetration into tumors than the hard counterpart. As a carrier of siPD-L1, Nanosac facilitated CD8⁺ T cell recruitment to tumors and controlled tumor growth significantly better than the hard counterpart. These results support that Nanosac offers enabling features for systemic siRNA delivery to tumors.

Experimental

Materials

All chemicals including tetraethyl orthosilicate (TEOS), cetyltrimethylammonium chloride (CTAC), triethanolamine, 3-aminopropyl)triethoxysilane (APTES), ammonium hydrogen difluoride, ammonium fluoride, 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT), chlorpromazine, methyl-β-cyclodextrin, and amiloride hydrochloride were purchased from Sigma Aldrich (St. Louis, MO, USA), unless specified otherwise. Dopamine hydrochloride was purchased from Alfa Aesar (Ward Hill, MA, USA). Cy3-labeled GAPDH siRNA, luciferase siRNA, RNase A, wheat germ agglutinin-488, Hoechst 33342, Lysotracker green, and Lipofectamine 2000 were purchased from Invitrogen (Eugene, OR, USA). PD-L1 siRNA (sense, 5′-CCCACAUAAAAAACAGUUGTT-3′; antisense, 5′-CAACUGUUUUUUAUGUGGGTT-3′) and negative control siRNA (sense, 5′-UGAAGUUGCACUUGAAGUCdTdT-3′; antisense, 5′-GACUUCAAGUGCAACUUCAdTdT-3′) were purchased from Integrated DNA Technologies (IDT, Coralville, Iowa, USA). Sulfo-cyanine5 NHS ester was purchase from Lumiprobe (Hunt valley, MD, USA). Luciferase assay kit was purchased from Promega (San Luis Obispo, CA, USA). GAPDH ELISA kit was purchased from Abcam (Burlingame, CA, USA). PD-L1 ELISA kit was purchased from Biomatik (Wilmington, DE, USA). Gibco Dulbecco’s Modified Eagle’s medium (DMEM) and Gibco RPMI 1640 medium (RPMI) were purchased from ThermoFisher Scientific (Waltham, MA, USA). Vascular cell basal medium and endothelial cell growth kit-BBE were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). PE anti-mouse CD4, FITC anti-mouse CD3, and APC anti-mouse CD8a antibodies were purchased from BioLegend (San Diego, CA, USA). Anti-PD-L1 antibody (clone 10F.9G2) was purchased from Bio X Cell (Lebanon, NH, USA). FITC-Lectin was purchased from Vector Laboratories (Burlingame, CA).

Preparation of siRNA-loaded nanocapsules

siRNA-loaded nanocapsules (O/siRNA/pD, Nanosac) were prepared by adsorbing siRNA on amine-modified mesoporous silica nanoparticles (MSNs), coating the siRNA-bound MSNs with pD, and removing the sacrificial MSNs. First, MSNs were synthesized according to the sol-gel procedure⁴² with slight modification. CTAC (25%, 5 mL), a cationic surfactant, was added to 15 mL of deionized water, stirred at 300 rpm for 15 min at 75 °C to form micelles serving as mesopore templates, and mixed with 0.8 mL of 10% triethanolamine at 75 °C for additional 15 min. To the CTAC/triethanolamine micelles, TEOS (1.5 mL) was added at a rate of ~30 drops per minute and stirred for 1 h at 300 rpm at 80 °C to form silica layers around the micelle clusters. CTAC was then removed by refluxing the mixture with methanol and HCl (500:19, v/v) at room temperature for 24 h. The MSNs was centrifuged at 20,000 rcf for 20 min and washed three times with methanol. MSNs (50 mg/mL) were mixed with 25 μL of APTES in ethanol at room temperature for 24 h to modify the surface with amine groups. Thereby formed MSN-APTES (MSN^a) particles were centrifuged at 20,000 rcf for 20 min and washed three times with ethanol. The purified MSN-APTES were mixed with siRNA in HEPES-buffered saline (pH 7) at a weight ratio of 50/1 and incubated for 5 min. The siRNA-loaded MSNs (MSN^a/siRNA) were coated with polydopamine (pD) layer by incubation in 1 mL of 1 mg/mL dopamine hydrochloride solution in Tris buffer (10 mM, pH 8.5) for 6 h at room temperature with rotation. Finally, the pD-coated, siRNA-loaded MSNs (MSN^a/siRNA/pD) were dispersed in 50 μL of deionized water and added to 200 μL of buffered oxide etch solution (2M HF/8M NH₄F, pH 5) to remove the sacrificial MSNs. After 5 min, the resulting nanocapsules (O/siRNA/pD, Nanosac) were washed three times by centrifugation (4600 rpm for 5 min). For fluorescent labeling of MSNs, 25 mg of MSN^a was dispersed in anhydrous dimethylformamide (8 mL) containing sulfo-cy5-NHS (1 mg) and triethylamine (80 μL). The reaction solution was stirred in dark for 24 h. The labeled MSN^a (MSN^a-cy5) were washed with ethanol five times and dispersed in deionized water. To test the feasibility of lyophilization, Nanosac was lyophilized by the Labconco FreeZone 4.5 Liter −84C Benchtop Freeze Dryer (Kansas City, MO, USA) with a varying amount of trehalose as a lyoprotectant. The dried Nanosac was reconstituted in deionized water and analyzed by the Malvern Zetasizer Nano ZS90 (Worcestershire, United Kingdom) and gel electrophoresis.

Characterization of NPs

Intermediate NPs (MSN, MSN^a, MSN^a/siRNA, and MSN^a/siRNA/pD) and siRNA-loaded nanocapsules (O/siRNA/pD, Nanosac), which were collectively called ‘NPs,’ were dispersed in phosphate buffer (10 mM, pH 7.4), and their sizes and zeta potentials were determined by the Malvern Zetasizer Nano ZS90. Their morphology was examined by a FEI Tecnai T20 transmission electron microscope (Hillsboro, OR) after negative staining with 1% uranyl acetate or 1% phosphotungstic acid. For AFM analysis, samples were prepared by placing a droplet of NP suspension on a 300-mesh copper grid (Electron Microscopy Sciences, Hatfield, PA, USA). Excess samples were removed by blotting paper and the grid was air-dried prior to measurement. The images and Young’s moduli of the NPs were obtained by an Asylum Cypher (Oxford instruments, Abingdon, United Kingdom). The Young’s modulus of NPs was determined by fitting the force-distance curve by the Hertz equation.^{47, 48}

$$\mathrm{F}=\frac{E}{1-v^2}\frac{tan\beta }{\sqrt{2}}{\delta }^2$$

Gel retardation assay for testing siRNA-loading capacity and siRNA stability

siRNA-loading capacity of MSN^a was evaluated by the agarose gel retardation assay. siRNA-loaded MSNs (MSN^a/siRNA) complexes were prepared varying the MSN^a/siRNA weight ratio from 1/1 to 50/1. The complexes were loaded in 2% agarose gel and run in 0.5× TAE buffer at 80 V for 40 min. The gel was stained with ethidium bromide, and siRNA bands were detected at 302 nm using Azure C300 (Dublin, CA, USA). For stability testing, siRNA or NPs were challenged with 166 U/mL RNase for 15 min ± 8 mg/mL SDS for additional 2.5 h or 50% FBS for 1 h, both at 37 °C, and analyzed by agarose gel electrophoresis. To quantify siRNA loaded in Nanosac (O/siRNA/pD), the NPs were dispersed in dilute HCl solution (pH 3) and incubated for 72 h. The samples were centrifuged at 16,000 rcf for 10 min, and the supernatant and pellet were separately analyzed by agarose gel electrophoresis.

Cell Culture

CT26 mouse colon carcinoma (ATCC), luciferase-expressing 4T1 mouse mammary carcinoma cells (4T1-luc, donation of Prof. Michael Wendt at Purdue University), and J774A.1 macrophages (ATCC) were cultured in DMEM medium, complemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 μg/mL streptomycin. Human umbilical vein endothelial cells (HUVEC, ATCC) were cultured in vascular cell basal medium with endothelial cell growth kit-BBE (ATCC).

Cytotoxicity of NPs

CT26 cells were seeded in a 96 well plate at a density of 10,000 cells per well and cultured at 37 °C in 5% CO₂. After overnight incubation, the culture medium was replaced with fresh medium containing MSN^a, MSN^a/pD, and Nanosac at 5 – 500 μg/ml and incubated for 48 h. Cell proliferation was quantified by the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay, where the cells were treated with 75 μg of MTT and incubated for 4 h. The formazan crystals were dissolved in DMSO and quantified by a SpectraMax M3 microplate reader (Molecular Devices, CA, USA) at the wavelength of 562 nm. The cell viability was defined as the absorbance divided by that of control cells that did not receive any treatment.

Confocal Microscopy

Cells were seeded on a cover glass placed in a 12 well plate and incubated for 24 h. The medium was replaced with a suspension of Cy3 or Cy5-labeled NPs (MSN, MSN^a/pD, and Nanosac) and incubated for a specified time period at 37 °C. The cells were then gently rinsed with PBS twice and fixed with 4% paraformaldehyde in PBS for 15 min. Cell membrane was stained with FITC-labeled wheat germ agglutinin (5 μg/ml) for 10 min. Hoechst 33342 nuclear stain (2 μM) was added 5 min prior to the imaging. Confocal microscopy was performed by the Nikon A1R confocal microscope (Melville, NT, USA).

Gene silencing

Three model siRNAs targeting GAPDH, luciferase, and PD-L1 expression (siGAPDH, siLuc, and siPD-L1) were used to test siRNA delivery via MSN^a, MSN^a/pD, and Nanosac. Non-specific siRNA (siCont) was used as a negative control for each evaluation. siGAPDH was tested with CT26 cells. The cells were seeded in 12-well plates at a density of 2 × 10⁵ cells per well, grown for 24 h, and incubated with no treatment or MSN^a, MSN^a/pD, and Nanosac loaded with siGAPDH or siCont (at a concentration equivalent to 100 nM siRNA) in complete medium for 48 h. The cells were rinsed with PBS twice and treated with 100 μL of lysis buffer for 15 min. The GAPDH level in the cell lysate was quantified by a standard GAPDH assay kit (Abcam, GAPDH ELISA Kit). siLuc was tested with 4T1-luc cells. The cells were seeded in 12-well plates at a density of 10⁵ cells per well, grown for 24 h, and incubated with no treatment or the NPs loaded with siLuc or siCont (at 150 nM siRNA) in complete medium for 48 h. The cells were lysed in passive lysis buffer for 10 min and analyzed for luciferase activity by the Luciferase Glow Assay Kit (Promega). siPD-L1 was tested with CT26 cells. The cells were plated in 6-well plates at a density of 10⁵ cells per well, incubated for 24 h, pretreated with 100 ng/mL of IFN-γ for 12 h to induce PD-L1 expression, and incubated with no treatment or the NPs loaded with siPD-L1 or siCont (at 200 nM siRNA) in complete medium for 48 h. In another set, the IFN-γ-pretreated CT26 cells were incubated with NPs loaded with siPD-L1 at varying concentrations (50-200 nM siPD-L1). The cells were lysed with a cell lysis buffer containing 1% protease inhibitor and analyzed for PD-L1 expression by the PD-L1 ELISA kit (Biomatik).

Western blot

The siPD-L1 treated CT26 cells were lysed with lysis buffer containing 1% protease inhibitor. The protein content in the cell lysates was quantified by the BCA assay. The lysates were boiled in Laemmli buffer for 5 min, resolved by 10% SDS-polyacrylamide gel electrophoresis, and transferred to polyvinylidene fluoride membrane. The membrane was blocked by 5% nonfat dried milk in TBST buffer (pH 7.4, 20 mM Tris, 150 mM NaCl, and 0.05% Tween 20) for 1h at room temperature. The membrane was incubated with anti-PD-L1 and anti-GAPDH antibodies for 24 h at 4 °C. The membrane was washed three times and incubated with secondary IgG-HRP antibody for 1 h at room temperature. The membrane was washed three times, and protein bands were imaged by Azure C300 (Dublin, CA).

NP uptake by J774A.1, HUVEC, and CT26 cells

J774A.1 cells were seeded in 6-well plates at a density of 10⁶ cells per well. HUVEC and CT26 cells were seeded in 12-well plates at a density of 10⁵ cells per well. After overnight, J774A.1 cells were treated with MSN^a-cy5/pD or Nanosac for 30 min or 2 h. HUVEC and CT26 cells were incubated with MSN^a-cy5/pD or Nanosac for 2 h or 6 h. The cells were then rinsed with PBS twice and lysed in dilute HCl (pH 3) solution with 3 cycles of freezing and thawing. Cy5 was retrieved from the cell lysate by 10 sec probe sonication at 30% amplitude, followed by 72 h incubation, and quantified by Synergy Neo2 plate reader (Biotek, Chittenden County, VT, USA).

Transendotheial transport of NPs

HUVEC cells were seeded at a density of 80,000 cells per well in a Transwell insert (3 μm pore) pre-coated with rat-tail collagen type I. Transendothelial electrical resistance (TEER) across the HUVEC layer was monitored daily by EVOM2™ epithelial voltohmmeter (World Precision Instruments, Sarasota, FL, USA). When the TEER value reached a plateau (indicating confluency), the HUVEC layer was incubated with TNF-α (10 ng/mL) for 4 h. After rinsing the cell layer twice with PBS to remove TNF-α, 0.1 mg of MSN^a-cy5, MSN^a-cy5/pD, or Nanosac were added to the apical side of the Transwell and incubated for 6 h. The media in apical and basolateral sides were collected, and the fluorescence intensity of the collected media were measured by Synergy Neo2 plate reader (Biotek, Chittenden County, VT, USA) to quantify the NPs in each side.

Mechanism of NP uptake

First, inhibitors of endocytosis pathways (chlorpromazine, methyl-β-cyclodextrin, and amiloride hydrochloride) were tested on CT26 cells to identify safe concentration ranges. CT26 cells were seeded at a density of 10⁴ cells per well in a 96 well plate and incubated for overnight. At 70 – 80% confluency, the cells were incubated with different endocytosis inhibitors for 30 min and then rinsed three times with PBS. The cell viability was determined by the MTT assay. Next, CT26 cells were seeded at a density of 10⁵ cells in a 12 well plate, incubated to 70 – 80% confluency, treated with chlorpromazine (5 nM), methyl-β-cyclodextrin (5 mM), or amiloride hydrochloride (1 mM) for 30 min, and rinsed with PBS three times. The cells pre-treated with each inhibitor were incubated with MSN^a-cy5, MSN^a-cy5/pD, or Nanosac (0.5 mg/ml) for 6 h. Cy5 levels in the cells were quantified as described in the NP uptake section. To confirm the cellular uptake mechanism, the NPs and lysosomes were located by confocal microscopy. CT26 cells were incubated with MSN^a-cy5, MSN^a-cy5/pD, and Nanosac for 6h. Lysosomes were labeled with LysoTracker Green (200 nM) for 30 min, and the nuclei were stained with Hoechst 33342 (2 μM) for 5 min. NPs, LysoTracker, and Hoechst were detected at λ_Ex/λ_Em of 350 nm/461 nm, 488 nm/520 nm, 646 nm/662 nm, respectively.

siRNA release

Nanosac equivalent to 10 μM of siPD-L1 was suspended in 100 μL of deionized water with pH 5.2, pH 6.2, pH 7.4, or H₂O₂ (100 μM). The Nanosac suspensions were incubated at 37 °C under constant agitation. Nanosac suspension was sampled at predetermined time points, loaded in 2% agarose gel, and run in 0.5× TAE buffer at 80 V for 40 min. The gel was stained with ethidium bromide, and the siRNA bands were detected at 302 nm using Azure C300 (Dublin, CA, USA). The percent siPD-L1 release was calculated as (band intensity of released siPD-L1/band intensity of 10 μM of siPD-L1) × 100.

Protein corona analysis

The composition of protein corona forming on NPs was analyzed by SDS-PAGE. Four milligrams of MSN^a-cy5, MSN^a-cy5/pD, and Nanosac were incubated in 1 mL of 50% FBS for 2 h with rotation. The NPs were washed twice with PBS, boiled in Laemmli buffer for 5 min, and resolved by 12% SDS- polyacrylamide gel electrophoresis. The gel was stained by Coomassie staining and imaged by Azure C300 (Dublin, CA). Protein bands in gel were excised and analyzed by liquid-chromatography mass-spectrometry (LC-MS/MS) as described below.

The status of albumin bound to NP surface was determined by pulse proteolysis.⁸⁹ Four milligrams of MSN^a-cy5 and MSN^a-cy5/pD were incubated in 1 mL of human serum albumin (10 mg/mL) for 2 h with rotation. The NPs were centrifuged at 16000 rcf for 10 min and washed with PBS twice. The albumin-bound NPs were treated with 0.2 mg/mL of thermolysin in HEPES buffer (pH 7.4, 20 mM) containing 100 mM NaCl and 10 mM CaCl₂. After 3 min incubation at room temperature, 5 μL of 50 mM EDTA was added to a 15 μL aliquot to quench proteolysis. For the control, 0.1 mg/mL of native albumin or denatured albumin (boiled at 95 °C for 10 min) were treated in the same manner. The samples were analyzed by SDS-PAGE. The protein bands were detected with Azure C300 to analyze the extent of proteolysis of surface-bound albumin. The percent digestion was calculated as (1 - band intensity of albumin after proteolysis/band intensity of albumin prior to proteolysis) × 100.

LC-MS/MS analysis of protein corona

In-gel protein digestion:

Gel bands were excised and de-stained 3 times with 25 mM ammonium bicarbonate (ABC)/50% acetonitrile (ACN) and once with 80% ACN. The gel pieces were dried in a vacuum centrifuge for 15 min. Reduction of disulfide bond was carried out using 10 mM dithiothreitol in 25 mM ABC at 55°C for 1 h. Alkylation of cysteine was carried out using 55 mM iodoacetamide in 25 mM ABC at room temperature in the dark for 45 min. Gels were then washed twice with 25 mM ABC/50% ACN, dried and transferred to Barocycler tubes and digested in for 2 h in Barocycler at 50°C and 20,000 psi (50 seconds at 20,000 psi, 10 seconds at atmospheric pressure for a total of 120 cycles or 2 h) using LysC/trypsin protease mix (Promega) at an enzyme-to-substrate ratio of 1:25. After digestion, supernatants were transferred to a new tube and remaining peptides were extracted using 60% ACN/5% trifluoroacetic acid (TFA). Pooled peptides were dried in a vacuum centrifuge to prepare for LC-MS/MS analysis.

Mass Spectrometry analysis:

Samples were analyzed in the Dionex UltiMate 3000 RSLC nano System combined with the Q-Exactive High-Field (HF) Hybrid Quadrupole Orbitrap MS (Thermo Fisher Scientific). Peptides were re-suspended in 3% ACN/0.1% Formic Acid (FA)/96.9% MilliQ, and 5 μL was used for LC-MS/MS analysis. Peptides were separated using a trap (300 μm ID × 5 mm packed with 5 μm 100 Å PepMap C18 medium) and the analytical columns (75 μm × 50 cm packed with 2 μm of 100 Å PepMap C18 medium) (Thermo Fisher Scientific) using a 120 min method at a flow rate of 300 nL/min. Mobile phase A consisted of 0.1% FA in water and mobile phase B consisted of 0.1% FA in 80% ACN. The linear gradient started at 5% B and reached 30% B in 80 min, 45% B in 91 min, and 100% B in 93 min. Next, the column was held at 100% B for the next 5 min before bringing back to 5% B and held for 20 min to equilibrate the column. The column temperature was maintained at 37 °C. MS data were acquired with a Top 20 data-dependent MS/MS scan method with a maximum injection time of 100 ms, a resolution of 120,000 at 200 m/z. Fragmentation of precursor ions was performed by high-energy C-trap dissociation (HCD) with the normalized collision energy of 27 eV. MS/MS scans were acquired at a resolution of 15,000 at m/z 200. The dynamic exclusion was set at 20 s to avoid repeated scanning of identical peptides.

Bioinformatics and data analysis:

The raw MS/MS data were processed using MaxQuant (v1.6.3.3)⁹⁰ with the spectra matched against the bovine (Bos Taurus) protein database downloaded from Uniprot (http://www.uniprot.org) on 5/20/2020. Data were searches using trypsin/P and LysC enzyme digestion allowing for up to 2 missed cleavages. MaxQuant search was set to 1% FDR both at the peptide and protein levels. The minimum peptide length required for database search was set to seven amino acids. Precursor mass tolerance of ± 10 ppm, MS/MS fragment ions tolerance of ± 20 ppm, oxidation of methionine protein N-terminal acetylation (K) were set as the variable modifications and carbamidomethylation of cysteine (C) was set as a fixed modification. The “unique plus razor peptides” were used for peptide quantitation. Razor peptides are the non-redundant, non-unique peptides assigned to the protein group with most other peptides. LFQ intensity values were used for relative protein abundance measurement. Proteins detected with at least 1 unique peptide and at least 2 MS/MS counts were only included for the final analysis.

NP penetration into tumor spheroids

3×10³ of CT26 cells were seeded in a round-bottom 96 well plate (Corning), briefly centrifuged at 3,000 rcf for 5 min to aggregate at the bottom of the plate, and incubated for 72 h to form spheroids. The spheroids were treated with MSN^a-cy5/pD or Nanosac (1 mg/mL as NPs) for 4 h, rinsed with PBS, and fixed with 4% paraformaldehyde for 30 min. Z-stack confocal images of the spheroid were obtained with 20 μm intervals from the bottom to the middle of the spheroids, by the Nikon A1R confocal microscope (Melville, NT, USA). Cy5 was detected at λ_Ex/λ_Em of 646 nm/662 nm.

Intravital confocal microscopy

5–6 weeks old female Balb/c mouse were purchased from Envigo (Indianapolis, IN) and acclimatized for 7 days prior to the procedure. All animal procedures were approved by Purdue Animal Care and Use Committee, in conformity with the NIH guidelines for the care and use of laboratory animals. For the real-time observation of NP transport in tumors, a dorsal window chamber was installed in the back of a mouse.^91-93 A Balb/c mouse was anesthetized by 2.5% isoflurane in oxygen flow using an anesthesia machine (Matrx VMS, Midmark). A window chamber was surgically implanted onto the dorsal skinflap, where 10⁶ of CT26 cells suspended in 25 μL of PBS were subsequently injected. The tumor-inoculated skinflap was covered with a coverslip (1 cm diameter) and monitored every other day. When the tumor size reached ~30 mm³, the mouse was given 100 μL of wheat germ agglutinin (WGA) 488 (1 mg/mL) for blood vessel staining, followed by MSN^a-cy5/pD or Nanosac (6 mg per mouse) injection, via a preinstalled mouse tail vein catheter. Intravital imaging was performed under isoflurane anesthesia using the Nikon Intravital MP confocal upright microscope equipped with a 10× objective. WGA 488 and cy5 signals were detected at λ_Ex/λ_Em of 488 nm/520 nm and 646 nm/662 nm, respectively.

Systemic siRNA delivery to subcutaneous tumor

Tumor-bearing mice were prepared by subcutaneous injection of 3×10⁵ CT26 cells suspended in 100 μL of growth medium in the upper flank of the right hind leg of a 5-6 weeks old female Balb/c mouse. When the tumor size reached 50 mm³, animals received IV injection of D5W, MSN^a/siCont/pD, MSN^a/siPD-L1/pD, or Nanosac (all equivalent to siRNA 0.75 mg/kg/time, q2d×10 or 1.5 mg/kg/time, q2d×7) via tail vein. The treatment was repeated seven times with a 2-day interval. In another set of experiment, Nanosac (equivalent to siPD-L1 1.5 mg/kg/time, q2d×5) was compared with anti-PD-L1 antibody (10 mg/kg/time, q2d×5, intraperitoneal injection). Tumor size was monitored every 2 days. The length (L) and width (W) of each tumor were measured by a digital caliper, and the volume (V) was calculated by the modified ellipsoid formula: V = (L × W²)/2.⁹⁴ To evaluate PD-L1 silencing in CT26 tumors by siPD-L1, tumors were sampled at sacrifice and analyzed by Western blotting. Tumors were homogenized by the Omni Tissue Master 125 homogenizer (Kennesaw, GA) and lysed with a lysis buffer containing 1% protease inhibitor. The tumor lysates were analyzed by Western blot as described previously. For histological evaluation, the livers and spleens were harvested and fixed in 10% neutral buffered formalin. The fixed tissues were embedded in paraffin, sectioned, and stained with hematoxylin and eosin. For immunophenotyping, tumor draining lymph nodes were harvested, cut into pieces, and filtered through 100 μm and 40 μm cell strainers. The single cell suspension was incubated with anti-mouse CD16/32 antibody to block non-specific binding and identified by anti-CD3 (FITC), CD4 (PE) and CD8 (APC) antibodies. The stained cells were analyzed by the BD Accuri C6 Flow Cytometer (BD Bioscience, Bedford, MA).

Biodistribution

Tumor-bearing mice were prepared by subcutaneous injection of 3×10⁵ CT26 cells suspended in 100 μL of growth medium in the upper flank of the right hind leg of a 5-6 weeks old female Balb/c mouse. When the tumor size reached 100 mm³, animals received IV injection of MSN^a/siRNA-cy5/pD or Nanosac (all equivalent to siRNA 0.75 mg/kg). After 24 h, tumor, liver, heart, spleen, lung, and kidney were harvested, weighed, and frozen. The frozen tissues were homogenized in dilute HCl (pH 3) by a Qiagen TissueRuptor with disposable probes. Cy5 was retrieved from the tissue lysate by 10 sec probe sonication at 30% amplitude, followed by 72 h incubation, and quantified by Synergy Neo2 plate reader (Biotek, Chittenden County, VT, USA).

Tumor distribution

CT26 tumor-bearing mice received a single IV injection of MSN^a/siRNA-cy5/pD or Nanosac (both equivalent to siRNA 0.75 mg/kg). After 24h, the mice were injected with 100 μL of FITC-lectin (1 mg/mL in sterile saline) via tail vein. After 5 min, animals were sacrificed, and tumors were harvested, fixed in 10% neutral buffered formalin solution, infiltrated with 30% sucrose/PBS solution at 4 °C, and embedded in optimal cutting temperature (OCT) compound (Fisher Scientific, Pittsburgh, PA). Cryostat sections of each tissue were obtained at a thickness of 16 μm and mounted on a glass slide. Images were taken with a Nikon A1R confocal microscope.

Statistical analysis

All statistical analyses were performed with GraphPad Prism 9 (La Jolla, CA). All data were analyzed with one-way or two-way ANOVA test to determine the statistical difference of means among various groups, followed by the recommended multiple comparisons tests. A value of p < 0.05 was considered statistically significant.