Extracellular Silica Nanomatrices Promote In Vitro Maturation of Anti‐tumor Dendritic Cells via Activation of Focal Adhesion Kinase

Abstract

The efficacy of dendritic cell (DC)‐based cancer vaccines is critically determined by the functionalities of in vitro maturated DCs. The maturation of DCs typically relies on chemicals that are cytotoxic or hinder the ability of DCs to efficiently activate the antigen‐specific cytotoxic T‐lymphocytes (CTLs) against tumors. Herein, the maturation chemicals are replaced with extracellular silica nanomatrices, fabricated by glancing angle deposition, to promote in vitro maturation of murine bone marrow‐derived DCs (mBMDCs). The extracellular nanomatrices composed of silica nanozigzags (NZs) enable the generation of mature mBMDCs with upregulated levels of co‐stimulatory molecules, C‐C chemokine receptor type‐7, X‐C motif chemokine recetpor‐1, DC‐specific ICAM‐3 grabbing nonintegrin, and enhanced endocytic capacity. The in vitro maturation is partially governed by focal adhesion kinase (FAK) that is mechanically activated in the curved cell adhesions formed at the DC‐NZ interfaces. The NZ‐maturated mBMDCs can prime the antigen‐specific CTLs into programmed cell death protein‐1 (PD‐1)^lowCD44^high memory phenotypes in vitro and suppress the growth of tumors in vivo. Meanwhile, the NZ‐mediated beneficial effects are also observed in human monocyte‐derived DCs. This work demonstrates that the silica NZs promote the anti‐tumor capacity of in vitro maturated DCs via the mechanoactivation of FAK, supporting the potential of silica NZs being a promising biomaterial for cancer immunotherapy.

Silica nanozigzags (NZs) facilitate the in vitro maturation of dendritic cells (DCs) via mechanoactivation of focal adhesion kinase. The NZ‐maturated DCs in vitro prime the cytotoxic T‐lymphocytes (CTLs) into the programmed cell death protein‐1 (PD‐1)^lowCD44^high effector memory CTLs, and in vivo suppress the growth of tumors, indicating a promising new strategy for cancer immunotherapy.

1. Introduction

Dendritic cell (DC) cancer vaccine is a kind of cellular immunotherapy, which utilizes the autologous DCs to induce anti‐tumor immune responses for specific elimination of targeted cancer cells. To prepare the DC cancer vaccines, the immature DCs are invitro differentiated from the monocytes or hematopoietic stem cells isolated from the patient's blood. Afterward, the immature DCs are stimulated to adopt the maturation phenotypes and pulsed with tumor antigens, before being administered back to the patient's body to initiate the adaptive immune response against cancer cells. To maximize the efficacy of the DC vaccine, the in vitro maturated DCs shall have the optimized ability to migrate to the secondary lymphoid organs, produce cytokines, and induce the effector functions and memory formations of antigen‐specific cytotoxic T‐lymphocytes (CTLs). The maturation phenotypes of DCs can be manipulated by applying different factors that typically involve cytokines and toll‐like receptor (TLR) agonists. However, these maturation agents often have limitations. The maturation cocktail consisting of tumor necrosis factor‐alpha (TNFα), interleukin‐1β, interleukin‐6, and prostaglandin E2 resulted in a high yield of maturated DCs with enhanced migratory capacity but suppressed production of interleukin‐12 (IL12) which is an important cytokine for anti‐tumor CTL function.^{[ 1} , ^{2 ]} The TLR ligands enhanced the expression of co‐stimulatory molecules and cytokines and promoted cross‐presentation in DCs, thereby increasing the capacity of DCs to elicit the CTL‐mediated immune response.^{[ 3} , ^{4 ]} However, some of the TLR ligands such as lipopolysaccharide (LPS) and polyinosinic‐polycytidylic acid (poly I:C) are potentially toxic to animals.^{[ 5} , ^{6 ]} The TLR‐maturated DCs would have a G‐protein expression profile associated with limited migratory capacity.^{[ 7 ]} Therefore, these maturation agents were used in combination to complement each other's deficits. For instance, a combination of TLR agonist (poly I:C with polylysine and carboxymethylcellulose) and multiple cytokines, including TNFα, interleukin‐1β, interferon‐alpha, and interferon‐gamma (IFNγ), were applied to generate the clinical‐grade DCs with migratory capacity, high expression level of IL12 and ability to initiate a more prolonged CTL‐mediated anti‐tumor response.^{[ 8 ]} Given the limitations and complexities of the current strategies, it would be beneficial to have an alternative method for in vitro maturation of DCs that are equipped with the ability to activate CTLs in a simple and biocompatible manner.

Besides the above biochemical signals, the role of mechanical stimuli in modulating the maturation of DCs has also caught attention recently. Biophysical cues such as shear stress and stiffness were reported to influence the in vivo functions and invitromaturation of DCs.^{[ 9 ]} This brings forth the possibility of using biomaterials to deliver biophysical signals for manipulating the properties of in vitro maturated DCs. Indeed, modifications of cell culture substrates with poly(lactic‐co‐glycolic acid) or inorganic topographies (such as titanium disks)^{[ 10} , ^{11 ]} were reported to facilitate invitro maturation of DCs. Although these modified biomaterials could upregulate surface maturation markers in DCs, they would either cause apoptosis or fail to upregulate the production of proinflammatory cytokines, so that the CTLs could not be efficiently activated to an extent comparable to the conventional methods.^{[ 10} , ^{11 ]} The micro‐ or nano‐patterned 2D biomaterials were proposed to influence cell behavior via cell adhesions at the cell‐substrate interfaces.^{[ 12 ]} The physical changes in cell adhesions could be sensed by mechanosensitive proteins, to initiate the intracellular signaling and result in some changes in the expression of genes or proteins. We previously reported that extracellular silica nanomatrices consisting of an array of nanopillars with different shapes and topographies could regulate invitro differentiation of neural stem cells via the cell adhesion‐related pathways without affecting cell viability.^{[ 13 ]} These findings illustrate the biocompatibility and potential for silica extracellular nanomatrices serving as an alternative biomaterial to mediate in vitro modulation of DCs via the cell‐adhesion‐related mechanisms. Therefore, we propose to apply silica nanomatrices to in vitro maturation of DCs into a functional phenotype that will promote the priming of antigen‐specific CTLs against tumors.

In this work, we performed glancing angle deposition (GLAD)^{[ 14 ]} to fabricate the extracellular silica nanomatrices for in vitro maturation of murine bone marrow‐derived DCs (mBMDCs). The silica nanomatrix‐induced maturation was optimized with the engineering of nanostructures in terms of the shapes, the pitch (P), and the number of pitches (N). Among the various silica nanostructures, silica nanozigzags (NZs) with a P of 245 nm and N of 3 were found to optimally promote in vitro maturation of mBMDCs via the mechanoactivation of focal adhesion kinase (FAK). In the functional aspects, we used ovalbumin (OVA) as the model antigen to demonstrate that the NZ‐maturated mBMDCs enable in vitro activation of antigen‐specific CTLs and lead to the formation of CD44^high effector memory CTLs without significant increase of the immune checkpoint molecule, programmed cell death protein‐1 (PD‐1). Furthermore, the injection of antigen‐pulsed NZ‐maturated mBMDCs suppressed the tumor growth and enhanced the survival rate of mice inoculated with the OVA‐expressing B16 melanoma cells, supporting the beneficial use of silica NZs in the preparation of maturated DCs for anti‐tumor treatment. Moreover, the advantageous effects of silica NZs on the maturation of DCs were also observed in human monocyte‐derived DCs (hu‐moDCs), suggesting that the silica NZs can be generally adapted to activate diverse DCs (Figure  1 ). Here, we manifest the beneficial FAK‐mediated effects of silica NZs on in vitro maturation of DCs for anti‐tumor immunotherapy, paving the way to improving in vivo efficacy of DC cancer vaccines.

Figure 1.

Schematic illustration of the effects of silica nanozigzags (NZs) on enhancing the in vitro maturation of anti‐tumor dendritic cells (DCs) via the activation of focal adhesion kinase (FAK). Silica NZs (with a pitch (P) of 245 nm and the number pitch (N) of 3, i.e., P245‐N3) promote the maturation of DCs into a phenotype that can potentially enhance their uptake of antigens, activation of cytotoxic T‐lymphocytes (CTLs) and homing to lymph nodes, via the mechanoactivation of FAK initiated by the formation of curved cell adhesions at the DC‐NZ interfaces. The NZ‐maturated murine bone marrow‐derived DCs (mBMDCs) can activate the antigen‐specific CTLs and program them into the PD‐1^lowCD44^high effector memory CTLs in vitro. Injection of the NZ‐maturated mBMDCs suppresses the growth of tumors in vivo. The silica NZs are also applicable to the in vitro maturation of human monocyte‐derived DCs (Hu‐moDCs).

2. Results and Discussion

2.1. Effects of the Silica Nanostructures on the Maturation of mBMDCs

The mBMDCs were applied to the silica nanomatrices for 20 h to induce their maturation, in the absence of cytokines or maturation chemicals (Figure  2A). The immature mBMDCs adhered to tissue culture polystyrene (TCPS) without any maturation treatment were served as negative control group. To optimize the silica nanomatrix‐mediated maturation of mBMDCs, the shapes (nanohelices or NHs with left and right handedness, vertical nanopillars or vNPs, tilted nanopillars or tNPs, and NZs) and nanostructures (including P and N) of silica nanomatrices were tuned by GLAD. For DCs, CD86 is an important co‐stimulatory molecule for activation of T cells, of which expression level is well‐reported to be upregulated upon the maturation of DCs.^{[ 15 ]} The DC‐specific ICAM‐3 grabbing nonintegrin (DC‐SIGN) is a C‐type leptin receptor (CLR) on the DCs to facilitate the uptake of antigens and the initial contact between DCs and T cells for the priming of T cells.^{[ 16} , ¹⁷ , ^{18 ]} The X‐C motif chemokine receptor‐1 (XCR1) is a receptor for the X‐C motif chemokine ligand‐1 (XCL1) released by the activated CTLs, of which expression on the DCs could sustain the interaction between DCs and CTLs and augment the CTL‐mediated response against cancer cells.^{[ 19} , ^{20 ]} Therefore, the upregulated expressions of these proteins on the maturated DCs indicate the increased capacity of DCs in potentiating the anti‐tumor CTLs, and thus, were picked as the markers to choose the optimal silica nanomatrices for in vitro maturation of DCs.

Figure 2.

The NZ‐mediated maturation of mBMDCs is modulated with an engineering of NZ nanostructures. A) Schematics of experimental timeline to characterize the properties of mBMDCs maturated with the silica nanomatrices composed of different nanostructures. The mBMDCs attached to the tissue culture polystyrene (TCPS), without any maturation treatment, served as the negative control group. B) Schematic illustration of three‐pitch silica NZs composed of controllable pitches (with a P of 165 and 245 nm, i.e., NZs‐P165‐N3 and NZs‐P245‐N3). C) Representative histograms, and the geometric mean fluorescence intensity (MFI) of D) CD86, E) DC‐SIGN, and F) XCR1 in the CD11c⁺ MHC‐II⁺ mBMDCs maturated with the three‐pitch silica NZs made of different P (with the number of independent trials (n) of 4). G) Schematic illustration of silica NZs composed of P of 245 nm and the controllable number of pitch (N of 2, 3, 4, and 5). H) Representative histograms, and I) the geometric MFI of CD86, J) DC‐SIGN, and K) XCR1 in the CD11c⁺ MHC‐II⁺ mBMDCs maturated with the silica NZs (P of 245 nm) with N of 2–5 (n = 4). The lines on histograms indicate the peaks in histograms of TCPS‐attached mBMDCs. The error bars represent the standard error of the mean. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, one‐way ANOVA with Tukey's HSD post‐hoc test.

First, to study the effect of the shape of silica nanomatrices, the mBMDCs were maturated on the silica nanomatrices (with a height H of 490–495 nm) sculpted in a shape of vNPs (i.e., vNPs‐H490), left‐handed and right‐handed NHs (with a P of 245 nm and N of 2, i.e., NHs‐P245‐N2‐LH and NHs‐P245‐N2‐RH), tNPs (i.e., tNPs‐H495) and NZs (with a P of 165 nm and N of 3, i.e., NZs‐P165‐N3) (Figures S1A–E and S2A, Supporting Information). All the silica nanomatrices, except for the tNPs, significantly increased the percentage of CD86⁺ mBMDCs, wherein the highest increase was induced by the NHs and NZs (Figure S2B, Supporting Information). The NHs and NZs, instead of vNPs and tNPs, enhanced the percentage of DC‐SIGN⁺ and XCR1⁺ mBMDCs (Figure S2C,D, Supporting Information). The handedness of NHs showed a negligible effect (Figure S2B–D, Supporting Information). Thus, the effect of the nano‐shape on the upregulation of these proteins could be summarized as an ascending order of tNPs < vNPs < NHs ≈ NZs. As demonstrated by the data distribution and the higher statistical significance in the pair‐wise comparison with the TCPS control group, the NZs tended to result in lower variation in the increase of DC‐SIGN⁺ and XCR1⁺ mBMDCs (Figure S2C,D, Supporting Information). This observation suggests that, compared with the NHs, the NZs could induce a more consistent enhancement in the expressions of DC‐SIGN and XCR1 in mBMDCs, implying that the shape of NZ has an advantage in the in vitro maturation of mBMDCs.

Next, the NZ‐induced maturation of mBMDCs were further optimized by engineering the nanostructures of silica NZs, in terms of P and N. To study the effect of P, P was raised from 165 nm to 245 nm (i.e., NZs‐P165‐N3 versus NZs‐P245‐N3; Figure 2B; Figure S1E,F, Supporting Information) at a given N of 3. The results showed that the NZs‐P245‐N3 led to further enhancement in the expression levels of CD86 and DC‐SIGN, while no significant improvement was observed in the expression of XCR1 (Figure 2C–F). As Young's modulus of silica NZs is reported to increase with the rise of P,^{[ 13 ]} these findings demonstrate that the stiffness of silica NZs possibly plays a role in modulating the NZ‐induced expressions of CD86 and DC‐SIGN, but not XCR1, in mBMDCs. An additional increase of P to 325 nm had no further effect on the upregulation of CD86, DC‐SIGN, and XCR1 (Figures S1F,G and S3, Supporting Information). These findings indicate that larger P is more favorable for the maturation of mBMDCs, while P > 245 nm will not promote the maturation to a greater extent. Then, the effect of N in a range of 2–5 on the NZ‐induced maturation of mBMDCs was studied at the P of 245 nm (Figure 2G; Figure S1F,H–J, Supporting Information). All the NZs‐P245 with diverse N values led to a significant increase in the expression levels of CD86, DC‐SIGN, and XCR1, compared with the TCPS control (Figure 2H‐K). Compared with the N of 2, the N of 3‐5 induced a notable enhancement in the expression levels of CD86 and DC‐SIGN (Figure 2I–J). No significant difference in the expression levels of CD86 and DC‐SIGN was detected among the N of 3‐5. These results illustrate that the NZ‐induced maturation of mBMDCs is promoted by an increase of N and reaches the optimum at N ≥ 3. Overall, the findings demonstrate that engineering the NZ nanostructures could modulate the expression levels of CD86 and DC‐SIGN, but not XCR1, in the mBMDCs. Furthermore, silica NZs with a P of 245 nm and an N of 3 are the optimal nanostructures for the maturation of mBMDCs. Therefore, the silica NZs‐P245‐N3 (Figure  3A) was chosen for the subsequent experiments reported in Section 2.2–2.6.

Figure 3.

Silica NZs‐P245‐N3 promotes the maturation of XCR1⁺DC‐SIGN⁺ mBMDCs, with an enhancement of endocytic capacity and production of pro‐inflammatory cytokines. A) Scanning electron microscopy (SEM) cross‐sectional image of the silica NZs (with a P of 245 nm and three pitches, i.e., NZs‐P245‐N3) perpendicularly protruding on a supporting substrate. Insets: (upper left) SEM image of an individual NZ (scale bar: 250 nm); (upper right) SEM top‐down image of the NZs. B) Schematics of experimental timeline to characterize the properties of mBMDCs maturated with the silica NZs or LPS. The mBMDCs attached to the TCPS, without any maturation treatment, served as the negative control group. C) Representative histograms, and D) the percentage of CD86⁺, E) CCR7⁺, F) XCR1⁺, G) DC‐SIGN⁺, and H) CD206⁺ cells in the CD11c⁺ MHC‐II⁺ mBMDC subset after the 20 h maturation period (D: n = 6, E‐G: n = 4, H: n = 3). I–L) Representative confocal images (I, K, yellow lines indicate the outlines of cells defined by the staining of F‐actin; scale bars: I, 20 µm; K, 10 µm), and J) the quantifications of fluorescence levels of intracellular dextran or L) microspheres in the maturated mBMDCs after a 15 min treatment of AF647‐conjugated dextran or dark‐red fluorescent microspheres. At least 200 cells were measured in each group per trial, and at least 1300 cells were measured in each group for five independent trials (n = 5). The error bars represent the standard error of the mean. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, (D,E,G,H,J,L) one‐way ANOVA with Tukey's HSD post‐hoc test, (F) Welch's ANOVA with Dunnett's T3 post‐hoc test.

2.2. Characterization of the Silica NZ‐maturated mBMDCs

To understand how the silica NZ‐induced maturation of mBMDCs differs from the conventional methods, the NZs‐P245‐N3‐maturated mBMDCs were compared with the TCPS‐attached mBMDCs maturated via a 20 h treatment of 100 ng mL⁻¹ LPS (Figure 3B). The silica NZs did not have a cytotoxic effect on the mBMDCs (Figure S4A, Supporting Information). For the expressions of surface proteins associated with the maturation of DCs, both the silica NZs and LPS significantly increased the percentage of mBMDCs expressing the co‐stimulatory molecule CD86 to a comparable extent (Figure 3C,D; Figure S4B,C, Supporting Information). Compared with the LPS‐maturated mBMDCs, the NZ‐maturated mBMDCs did not increase the expressions of major histocompatibility complex class I and II (MHC‐I and MHC‐II; Figure S4D–F, Supporting Information) but upregulated the level of lymph‐node‐homing factor C‐C chemokine receptor type‐7 (CCR7, Figure 3C,E). Moreover, for the expressions of XCR1 and DC‐SIGN, which were previously found to be upregulated by silica NZs (Figure 2E,F,J,K), LPS also increased the expression of XCR1 in the mBMDCs, but to a lesser extent, compared with the silica NZs (Figure 3C,F). However, LPS did not enhance the expression of DC‐SIGN as the silica NZs did (Figure 3C,G). Besides being an adhesion molecule for promoting the interaction of DCs and T cells, the DC‐SIGN is also a CLR for the internalization of exogenous antigens.^{[ 16 ]} The increased level of DC‐SIGN in the NZ‐maturated mBMDCs illustrates the possible enhancement of antigen‐uptake capacity. Hence, we then examined the expression of another CLR related to antigen uptake, mannose receptor (CD206), and studied the endocytic capacity of mBMDCs via the uptake assay of dextran and microspheres. The expression level of CD206 was diminished in the LPS‐maturated mBMDCs but not in the NZ‐maturated mBMDCs (Figure 3C,H). Consistently with the expression patterns of these CLRs, the significant rise of the uptake of CLR ligand, dextran, was only observed in the NZ‐maturated mBMDCs but not in the LPS‐maturated ones (Figure 3I,J). The number of internalized microspheres in the NZ‐maturated mBMDCs was also larger than those in the LPS‐maturated mBMDCs (Figure 3K,L), indicating a higher pinocytic or phagocytic activity in the NZ‐maturated mBMDCs. These findings demonstrate that silica NZs promote the maturation of mBMDCs and enhance the expressions of CD86, CCR7, XCR1, and DC‐SIGN, as well as the endocytic capacity, in the mBMDCs. Compared with the LPS‐maturated mBMDCs, the NZ‐maturated mBMDCs have a higher level of XCR1, CCR7, DC‐SIGN, and endocytic capacity.

To induce a specific antigen‐targeted immune response, the in vitro maturated DCs shall first be loaded with antigens. In contrast to the low expression level of CLRs and modest endocytic capacity in the LPS‐maturated mBMDCs, the NZ‐maturated mBMDCs retained an upregulated expression level of CLRs and high endocytic capacity after the 20 h maturation period. Previous studies have shown that the LPS stimulation could only initiate a few‐hour upregulation of antigen uptake at the early phase of maturation. The expression of CLR, CD206, and antigen‐uptake capacity were then downregulated in the LPS‐maturated DCs.^{[ 21 ]} These might lead to limited loading of tumor antigens and difficulties in optimizing the timing for antigen loading in the development of the DC vaccine. Therefore, the persistently high endocytic capacity in the NZ‐maturated mBMDCs might overcome this problem and result in a more efficient loading of antigens. Besides, the NZ‐maturated mBMDCs express enhanced levels of DC‐SIGN which is proposed as the target to facilitate the delivery of antigens into the DCs.^{[ 22 ]} The antigens conjugated with the antibodies targeting the extracellular neck domain of DC‐SIGN could be internalized into the DCs via the DC‐SIGN, and be retained in the early endosome to be loaded into the MHC‐I complex for the cross‐presentation to CTLs.^{[ 22 ]} Thus, it is possible to combine the NZ‐induced DC maturation with the DC‐SIGN‐targeted delivery of tumor antigens to facilitate the generation of maturated DCs with cross‐presentation of tumor antigens for the priming of antigen‐specific CTLs. After the uptake of antigens, the mBMDCs should present the antigens in the MHC‐I complexes that specifically bind to the T‐cell receptor (TCR) on the antigen‐specific CTLs.^{[ 23 ]} Then, the other co‐stimulatory signals, in the form of direct interaction with surface proteins or paracrine signaling, are required to stimulate the survival, proliferation, and production of cytokines in the antigen‐specific CTLs.^{[ 24 ]} Although not have a significant increase in the expression of MHC‐I, the NZ‐maturated mBMDCs expressed the baseline level of MHC‐I for presenting the antigens to the CTLs. Both the silica NZs and LPS resulted in a distinctive increase of mBMDCs expressing the co‐stimulatory molecules, CD86, which could bind to the CD28 receptors on the CTLs to provide the crucial co‐stimulatory signals for the activation of CTLs.^{[ 24 ]} Therefore, both the NZ‐maturated and LPS‐maturated mBMDCs should be capable of priming the CTLs. Nevertheless, DC‐SIGN is a high‐affinity receptor for the intercellular adhesion molecule‐3 expressed on the resting T cells, which promotes the initial contact of mBMDCs with the CTLs.^[17] Besides, after the initial recognition of antigen presented on the DCs, the activated CTLs secrete XCL1 that binds with the XCR1 on the DCs to stabilize the DC‐CTL contacts.^{[ 19 ]} As a result, the upregulated expressions of DC‐SIGN and XCR1 on the NZ‐maturated mBMDCs might promote and sustain the DC‐CTL interactions, which could lengthen the duration of TCR activation and co‐stimulation and result in a more efficient activation of CTLs. Furthermore, we also observed that, compared with the LPS‐maturated mBMDCs, the NZ‐maturated mBMDCs expressed a higher level of proinflammatory cytokine TNFα and IFNγ and a comparable level of IL12 after the 20 h maturation period (Figure S5, Supporting Information). As previously reported, the cytokine TNFα can bind to the TNF receptor type II on the CTLs to lower the threshold of TCR activation and serve as the co‐stimulatory signals, resulting in the enhanced proliferation and production of IFNγ in CTLs.^{[ 25 ]} The IFNγ can act on the CTLs to promote the clonal expansion, production of IFNγ, and differentiation in the CTLs.^{[ 26 ]} Therefore, the observation that the NZ‐maturated mBMDCs could produce enhanced levels of TNFα and IFNγ after the maturation period, i.e., at the time of interacting with the targeted CTLs, implies their possibly higher capacity to activate the CTLs and their potential in modulating the effector functions and immunological memory of CTLs. Moreover, the capacity of lymphoid homing is another crucial factor for determining the efficacy of DCs in eliciting the anti‐tumor CTL response in vivo. CCR7 is a chemokine receptor that recognizes its chemokine ligands, CCL21 and CCL19, expressed by the lymphatic endothelial cells and stromal cells in the lymph nodes, respectively, to guide the entry and promote the homing of DCs to the lymph nodes.^{[ 27 ]} Thus, the enhanced level of CCR7 on the NZ‐maturated mBMDCs illustrates their higher possibility to migrate to the draining lymph nodes and come into contact with the antigen‐specific CTLs, resulting in a promotion of anti‐tumor effect induced by the injected DC vaccines. Overall, these findings illustrate that the NZ‐maturated mBMDCs acquire a phenotype that potentially benefits the activation of CTLs and the induction of in vivo anti‐tumor response.

2.3. Mechanisms of the Silica NZ‐promoted Maturation of mBMDCs: Activation of FAK via the Formation of Curved Cell Adhesions at the DC‐NZ Interfaces

To understand how the silica NZs maturate and enhance the mBMDCs, we first visualized the interaction of silica NZs with mBMDCs using scanning electron microscopy (SEM) and found that silica NZs were in contact with the basal plasma membranes of mBMDCs (Figure  4A,B). To further visualize how the silica NZs affect the basal plasma membranes of mBMDCs, we stained the plasma membranes of mBMDCs and the silica NZs with CF488A‐conjugated wheat germ agglutinin and CellTracker Deep Red dye, respectively. The usage of CellTracker Deep Red dye in the staining of silica NZs was validated by the comparable topographies observed in the top‐down SEM images and the confocal images of CellTracker Deep Red dye‐stained silica NZs (Figure S6, Supporting Information). The orthogonal views of the stained NZ‐maturated mBMDCs showed that the basal plasma membranes of mBMDCs were deformed by the silica NZs (Figure 4C,D). As cell adhesions are typically formed at the site of contact between the basal plasma membranes and the underlying substrate, the deformation of basal plasma membranes indicates the possible changes in the pattern of cell adhesions in the NZ‐maturated mBMDCs. To visualize the cell adhesions in the NZ‐maturated mBMDCs, the major cell adhesion molecule, vinculin, lying at the cytoplasmic face of cell adhesions were immunostained. The actin filament (F‐actin) located in the cortical actin under the cell membranes or bundled at the cell adhesions were also stained for visualizing the cell boundaries of mBMDCs. The confocal images of the basal plasma membranes of mBMDCs and the intensity profiles showed that the vinculin⁺ cell adhesions and F‐actin were located at the inter‐NZ spaces (Figure 4E,F), illustrating that the deformed basal plasma membranes of NZ‐maturated mBMDCs project into the inter‐NZ spaces and the cell adhesions are formed at the DC‐NZ interfaces.

Figure 4.

Characterization of cell adhesions at the DC‐NZ interfaces, with an activation of FAK. After being seeded on the silica NZs‐P245‐N3 for 20 h, the mBMDCs were fixed and visualized using SEM and confocal fluorescence microscopy. A) Representative SEM images show an attachment of mBMDCs on the silica NZs. B) An enlarged view of the area indicated with the black box in (A), where the cell membrane is shaded in a pale blue color. The white arrows show the locations where the silica NZs are in contact with the basal plasma membrane of mBMDC. The red dotted lines indicate the outline of a silica NZ. C) Representative maximum‐intensity projection (MIP) of deconvoluted confocal z‐stack shows the CellTracker Deep Red dye‐stained silica NZs (red) and wheat germ agglutinin‐stained plasma membrane (green) of NZ‐maturated mBMDCs. D) Orthogonal views in the XZ plane labeled with the dotted line in (C). E) Representative MIPs of deconvoluted confocal z‐stacks show the CellTracker Deep Red dye‐stained silica NZs (green), immunostaining of vinculin (magenta), and phalloidin‐stained F‐actin (blue) in the basal plasma membrane of NZ‐maturated mBMDCs. Inset: a schematic illustrates the formation of cell adhesions (magenta) at the interfaces between the DC (blue) and silica NZs (green). F) Fluorescence intensity profiles of the silica NZs, vinculin, and F‐actin across the white arrow labeled on the enlarged merge view in (E). G) Individual slices in the z‐stack of an enlarged view indicated by the orange box in (E). The white arrowheads mark the depth of a vertical cell adhesion formed at the DC‐NZ interfaces. H) Representative confocal images show the immunostaining results of vinculin (magenta) and p‐FAK (Y397, green) at the basal plasma membranes of mBMDCs attached to TCPS and silica NZs. The yellow arrows and arrowheads indicate the podosomes and focal adhesions, respectively, in the TCPS‐attached mBMDCs. I) Quantifications of the cell spreading area and J) the percentage of cell area covered with vinculin⁺ cell adhesions in the mBMDCs attached to TCPS or silica NZs. K) Fluorescence intensity profiles of vinculin and p‐FAK across the white arrows in (H). L) Quantification of fluorescence intensity of p‐FAK in the vinculin⁺ cell adhesions of mBMDCs attached to TCPS or silica NZs. The error bars represent the standard error of the mean. ****p < 0.0001, two‐tailed Welch's t‐test, n = 305 cells for the TCPS group and n = 313 cells for the NZ group, pooled from five independent experimental trials.

Further analysis on the individual slices of confocal z‐stack captured at the DC‐NZ interfaces revealed the depth of vinculin⁺ cell adhesions in the NZ‐maturated mBMDCs (Figure 4G), indicating that the NZ‐maturated mBMDCs form vertically curved cell adhesions at the lateral surfaces of silica NZs. Previous studies reported that the deformed cell membranes could induce the clustering of membrane curvature‐sensitive protein, phosphatidylinositol 4,5‐bisphosphate (PIP2), which in turn facilitated the recruitment of vinculin and another mechanosensitive protein FAK to the cell membranes.^{[ 28} , ^{29 ]} The binding of FAK to PIP2 can induce conformational change of FAK to expose its Tyr‐397 autophosphorylation site, resulting in the activation of FAK.^{[ 28 ]} Since the expression of PIP2 was indeed observed in the curved vinculin⁺ cell adhesions in the NZ‐maturated mBMDCs (Figure S7, Supporting Information), we sought to determine whether the activation level of FAK would be upregulated in the cell adhesions of the NZ‐maturated mBMDCs. To examine this question, we compared the immunostaining results of vinculin⁺ cell adhesions and Tyr‐397 phosphorylated FAK (p‐FAK) in the NZ‐maturated mBMDCs with those in the TCPS‐attached mBMDCs. The TCPS‐attached mBMDCs had larger cell spreading areas and had podosomes and focal adhesions occupying a smaller area of basal plasma membranes as the predominant types of cell adhesions (Figure 4H–J). In contrast, the NZ‐maturated mBMDCs had smaller cell spreading areas, and formed clusters of cell adhesions occupying a larger area of basal plasma membranes (Figure 4H–J; Figure S8, Supporting Information). These observations demonstrate that the silica NZs facilitate the formation of cell adhesions in mBMDCs. Moreover, the confocal images and intensity profiles illustrated that p‐FAK was localized in the vinculin⁺ cell adhesions of both NZ‐maturated mBMDCs and TCPS‐attached mBMDCs (Figure 4H,K). Compared with the TCPS‐attached mBMDCs, the cell adhesions of NZ‐maturated mBMDCs had significantly upregulated levels of p‐FAK (Figure 4L). This observation illustrates that the vertically curved cell adhesions formed at the DC‐NZ interfaces contain a higher level of active FAK, compared with the podosomes or focal adhesions formed on the flat TCPS. Therefore, these findings indicate that the silica NZs facilitate the mechanoactivation of FAK by inducing the formation of curved cell adhesions at the DC‐NZ interfaces.

In accordance with the immunostaining results, the upregulation of active p‐FAK in the NZ‐maturated mBMDCs was confirmed by Western blot assay (Figure 5A–C). To further determine the crucial role of FAK in the NZ‐mediated maturation of mBMDCs, the activation of FAK in the NZ‐maturated mBMDCs was inhibited by the treatment with 5 µm FAK inhibitor, PF‐573228 (Figure 5A–C). The inhibition of FAK hindered the NZ‐induced increase of CD86⁺, XCR1⁺, and DC‐SIGN⁺ mBMDCs, but not the CCR7⁺ mBMDCs (Figure 5D–H; Figure S9A,B, Supporting Information). However, the enhanced uptake of dextran and microspheres in the NZ‐maturated mBMDCs were not affected by the inhibition of FAK (Figure 5I–L). Meanwhile, inhibiting the FAK also suppressed the upregulated expressions of TNFα and IL12, but not IFNγ, in the NZ‐maturated mBMDCs after the 20 h maturation period (Figure S10, Supporting Information). Moreover, compared with the silica NZs with smaller P or N which led to inferior upregulation of CD86, XCR1, and DC‐SIGN in mBMDCs (Figure 2D–F,I,J), the optimal silica NZs‐P245‐N3 resulted in the mBMDCs with more abundant cell adhesions that contain high level of active FAK (Figure S11, Supporting Information), demonstrating the possible correlation between the enhanced expression of these surface proteins and the mechanoactivation of FAK in mBMDCs. Therefore, these findings demonstrate that the silica NZs promote the maturation of mBMDCs and upregulate the expressions of CD86, XCR1, and DC‐SIGN via FAK activation. FAK is more frequently associated with the motility or formation of cell adhesions in the DCs,^{[ 30} , ³¹ , ^{32 ]} while limited studies have reported the roles of FAK in regulating the maturation or protein expression in the DCs. As a non‐receptor tyrosine kinase, FAK could possibly phosphorylate multiple downstream targets such as Src kinase and extracellular signal‐regulated kinase to initiate the signaling pathway for regulating the protein expression in DCs.^{[ 33 ]} Besides, FAK also contains the nuclear localization sequences, allowing it to translocate into the nucleus to regulate the gene transcription or modify the chromatin condensation.^{[ 34} , ^{35 ]} It was reported that the nuclear shuttling of FAK was inversely correlated with the kinase activity and membrane recruitment of FAK.^{[ 34 ]} Thus, the NZ‐induced activation of FAK might regulate the protein expression in mBMDCs by preventing the nuclear‐FAK‐related suppressions of gene transcription and chromatin modification. On the other hand, the NZ‐induced upregulations of CCR7 and IFNγ were found to be independent of the FAK activation. Previous studies have proposed that the nano‐patterned substrates could modify the morphologies of cells and nuclei.^{[ 12 ]} The morphological changes could then regulate the nuclear translocation of transcription factors or chromatin modifiers, resulting in changes in gene expression.^{[ 36} , ^{37 ]} The distinctive morphological changes, for example, the decrease of cell spreading area, were still observed in the NZ‐maturated mBMDCs after the treatment with FAK inhibitor (Figure 5I,K). This raises a possibility that the silica NZs might also modulate the maturation phenotypes of mBMDCs via the mechanical regulation of transcriptional activity without the involvement of FAK. Moreover, the NZ‐induced uptake of dextran and microsphere also did not depend on the FAK activation. The expression level of CD206 was comparable between the TCPS‐attached mBMDCs and the NZ‐maturated mBMDCs with or without FAK inhibition (Figure S9C,D, Supporting Information). Although the FAK inhibition abolished the NZ‐induced upregulation of DC‐SIGN, the enhanced internalization of dextran, a ligand of DC‐SIGN and CD206, in the NZ‐maturated mBMDCs were not notably affected. This implies that the dextran could be internalized into the NZ‐maturated mBMDCs via both receptor‐mediated endocytosis and non‐receptor‐mediated pinocytosis or phagocytosis. Previous studies have shown that the nanotopography could enhance the clathrin‐mediated endocytosis at the plasma membranes with curvatures induced by the cell‐substrate interaction.^{[ 38 ]} Thus, the silica NZs might also enhance the endocytic capacity of mBMDCs via the deformation of basal plasma membranes. In short, these findings illustrate that the silica NZs upregulate the expressions of CD86, XCR1, DC‐SIGN, TNFα, and IL12 via the FAK activation; and they enhance the expressions of CCR7, IFNγ, and the endocytic capacity possibly by inducing the morphology‐regulated gene expressions and the deformation of basal plasma membranes.

Figure 5.

Silica NZs promote the maturation of mBMDCs via the activation of FAK. A) Schematics of the experimental timeline to examine the role of FAK in the NZs(‐P245‐N3)‐induced maturation of mBMDCs via the application of FAK inhibitor PF‐573228 (PF). The mBMDCs attached to the TCPS, with the treatment of 0.02% DMSO, served as the vehicle control group. B) Representative Western blots and C) quantification result of the level of Tyr‐397(Y397)‐phosphorylated FAK in the mBMDCs after the 20 h maturation period (n = 5). D) Representative histograms, and E) the percentage of CD86⁺, F) CCR7⁺, G) XCR1⁺, H) DC‐SIGN⁺ cells in the CD11c⁺ MHC‐II⁺ mBMDC subset after the 20 h maturation period (n = 4). I‐L) Representative confocal images (I, K, yellow lines indicate the outlines of cells defined by the staining of F‐actin; scale bars: 10 µm), and J) the quantifications of fluorescence levels of intracellular dextran or L) microspheres in the maturated mBMDCs after a 15 min treatment of AF647‐conjugated dextran or dark‐red fluorescent microspheres. At least 200 cells were measured in each group per trial, and at least 1000 cells were measured in each group for five independent trials (n = 5). The error bars represent the standard error of the mean. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, one‐way ANOVA with Tukey's HSD post‐hoc test.

2.4. In Vitro Effect of the NZ‐Maturated mBMDCs on the Activation of Antigen‐Specific CTLs

In Section 2.2, the NZ‐maturated mBMDCs were found to have a maturation phenotype different from the LPS‐maturated mBMDCs. To examine whether the deviated phenotypes of NZ‐maturated mBMDCs would lead to differences in the activation and memory formation of CTLs, we used the MHC‐I (H‐2k^b)‐restricted OVA_257‐264 peptide as the model antigen to study the different effects of NZ‐maturated mBMDCs and LPS‐maturated mBMDCs on the priming of antigen‐specific CTLs (Figure  6A). After two days of the DC‐CTL co‐culture, surprisingly, a higher expression level of IFNγ was detected in the H‐2k^b‐OVA_257‐264‐tetramer⁺ antigen‐specific CTLs primed by the OVA_257‐264‐pulsed NZ‐maturated mBMDCs, but not in those primed by the OVA_257‐264‐pulsed LPS‐maturated mBMDCs (Figure 6B,C). Meanwhile, according to the results of carboxyfluorescein succinimidyl ester (CFSE)‐based CTL proliferation assay obtained after four days of the DC‐CTL co‐culture, both the OVA_257‐264‐pulsed NZ‐maturated and LPS‐maturated mBMDCs promoted the proliferation of antigen‐specific CTLs, wherein the proliferation induced by the OVA_257‐264‐pulsed LPS‐maturated mBMDCs was more distinctive than those induced by the OVA_257‐264‐pulsed NZ‐maturated mBMDCs (Figure 6D,E; Figure S12, Supporting Information). Therefore, the CFSE results illustrated that both the NZ‐maturated mBMDCs and LPS‐maturated mBMDCs pulsed with OVA_257‐264 were able to activate the antigen‐specific CTLs, while the NZ‐maturated mBMDCs enhanced the production of IFNγ and the LPS‐maturated mBMDCs induced a greater proliferation in the antigen‐specific CTLs. The LPS‐maturated mBMDCs expressed a higher level of MHC‐I (Figure S4E, Supporting Information) and OVA_257‐264‐bound MHC‐I (H‐2k^b‐OVA_257‐264) after the antigen pulsing (Figure S13A–C, Supporting Information). Therefore, the higher availability of H‐2k^b‐OVA_257‐264 on the LPS‐maturated mBMDCs could increase their chance to ligate with the TCRs on the antigen‐specific CTLs, and thus result in a better effect on promoting the proliferation of CTLs. Although the NZ‐maturated mBMDCs only had the baseline expression level of H‐2k^b‐OVA_257‐264 (Figure S13B,C, Supporting Information), their H‐2k^b‐OVA_257‐264 ⁺ subset had a high expression level of the co‐stimulatory molecule CD86 (Figure S13D,E, Supporting Information), which was comparable to those of the LPS‐maturated mBMDCs. This ensures the capacity of the OVA_257‐264‐pulsed NZ‐maturated mBMDCs in inducing the activation and proliferation of the antigen‐specific CTLs. Moreover, the production of IFNγ in the activated CTLs is known to be upregulated after the TCR stimulation,^{[ 39 ]} while its sustained production is positively regulated by the duration of DC‐CTL contact.^{[ 40 ]} Thus, no upregulation of IFNγ in the CTLs primed by the OVA_257‐264‐pulsed LPS‐maturated mBMDCs might be ascribed to that the peak time of IFNγ production had passed. In contrast, in the NZ‐maturated mBMDCs, the higher expression of XCR1 can possibly stabilize and prolong the DC‐CTL interaction,^{[ 19 ]} facilitating the production of IFNγ in the CTLs. Besides, the OVA_257‐264‐pulsed NZ‐maturated mBMDCs had upregulated expressions of TNFα and IFNγ (Figure S13F,G, Supporting Information) which could directly act on the CTLs to promote the production of IFNγ.^{[ 25} , ^{26 ]} Therefore, these properties allow the OVA_257‐264‐pulsed NZ‐maturated mBMDCs to induce the production of IFNγ and proliferation of the antigen‐specific CTLs even without distinctive upregulation of H‐2k^b‐OVA_257‐264.

Figure 6.

The NZ‐maturated mBMDCs promote in vitro activation of antigen‐specific CTLs into the PD‐1^lowCD44^high memory phenotypes. A) Schematics of the experimental timeline to analyze the properties of antigen‐specific CTLs primed by the mBMDCs maturated with the silica NZs(‐P245‐N3) or LPS. The CTLs primed by the TCPS‐attached mBMDCs served as the negative control group. B) Representative histograms and C) the geometric MFI of IFNγ in the antigen‐specific H‐2k^b‐OVA_257‐264‐tetramer⁺ CD8⁺ CTL subsets after 4 days of DC‐CTL co‐culture and the treatment with GolgiPlug on the day of harvest (n = 4). D,F,H) Representative histograms, E) the normalized percentage of CFSE^low divided cells, G) the percentage of CD44^high cells, and I) the geometric MFI of PD‐1 in the antigen‐specific H‐2k^b‐OVA_257‐264‐tetramer⁺ CD8⁺ CTL subsets after 4 days of DC‐CTL co‐culture (E: n = 5, G, I: n = 6). The percentage of CFSE^low CTLs is normalized by mean. The lines on histograms indicate the gating for CFSE^low (D) and CD44^high (F) cells. The error bars represent the standard error of the mean. *p<0.05, **p<0.01, ***p<0.001, two‐way ANOVA with Tukey's HSD post‐hoc test.

To facilitate the establishment of a long‐term anti‐tumor immune response, it is desirable for the DC cancer vaccines to promote the differentiation of the antigen‐specific CTLs into long‐lived memory CTLs with low expression levels of immunoinhibitory molecules.^{[ 41 ]} Therefore, the expression levels of memory marker CD44 and immunoinhibitory molecule PD‐1 in the antigen‐specific CTLs primed by the mBMDCs were also examined in a four‐day period. The OVA_257‐264‐pulsed NZ‐maturated mBMDCs resulted in a higher percentage of the CD44^high antigen‐specific CTLs without a significant increase of PD‐1 (Figure 6F–I). In contrast, the antigen‐specific CTLs primed by the OVA_257‐264‐pulsed LPS‐maturated mBMDCs had a lower level of CD44 but a significant increase in the expression of PD‐1 (Figure 6F–I). As previously reported, when the CTLs are differentiated into the effector memory cells, the CD44 expression will increase while the PD‐1 expression may vary; when the CTLs are differentiated into the terminal effector cells, the PD‐1 expression will increase while the CD44 expression will decrease.^{[ 42} , ^{43 ]} According to these findings, we believe that the NZ‐maturated mBMDCs and LPS‐maturated mBMDCs, with pulsing of OVA_257‐264, tend to induce the differentiation of CTLs into the PD‐1^lowCD44^high effector memory CTLs and the PD‐1^highCD44^low/int terminal effector CTLs, respectively. The expressions of CD44 and PD‐1 in CTLs are oppositely regulated by a master transcription factor, T‐box transcription factor TBX21 (T‐bet), which promotes the expression of CD44 but suppresses that of PD‐1.^{[ 44} , ^{45 ]} The OVA_257‐264‐pulsed NZ‐maturated mBMDCs resulted in a higher proportion of T‐bet^high antigen‐specific CTLs which had upregulated levels of CD44 (Figure S14A–D, Supporting Information). In contrast, the OVA_257‐264‐pulsed LPS‐maturated BMDCs gave rise to antigen‐specific CTLs with intermediate levels of T‐bet and higher levels of PD‐1 (Figure S14A,B,E,F, Supporting Information). These results indicate that the deviated expressions of CD44 and PD‐1 in the primed CTLs are regulated by the T‐bet. The cytokine IFNγ is one of the major cytokines to amplify the expression of T‐bet in the CTLs.^{[ 46 ]} Our results showed that the OVA_257‐264‐pulsed NZ‐maturated mBMDCs had higher expression levels of TNFα and IFNγ (Figure S13F–G, Supporting Information), which could possibly act on the CTLs in a paracrine manner to increase the production of IFNγ in the CTLs.^{[ 25} , ^{26 ]} The CTLs primed by the OVA_257‐264‐pulsed NZ‐maturated mBMDCs indeed had a higher level of IFNγ (Figure 6B,C), which could possibly function in both autocrine and paracrine manners to increase the expression of T‐bet in the CTLs. Therefore, it is demonstrated from our findings that the antigen‐pulsed NZ‐maturated mBMDCs are capable of activating the antigen‐specific CTLs and promoting the differentiation of CTLs into PD‐1^lowCD44^high effector memory CTLs possibly via the TNFα‐IFNγ‐T‐bet pathway.

2.5. In vivo Effect of the NZ‐maturated mBMDCs on Suppression of Tumor Growth

To understand whether the deviated phenotypes of the NZ‐maturated mBMDCs would benefit the elimination of cancer cells in vivo, we compared the effects of OVA_257‐264‐pulsed NZ‐maturated mBMDCs and OVA_257‐264‐pulsed LPS‐maturated mBMDCs on suppressing the tumor growth in the mice bearing the melanoma B16‐OVA tumors (Figure  7A). Compared with the sham control group which was injected with the vehicle (HBSS) without mBMDCs, a single injection of the OVA_257‐264‐pulsed NZ‐maturated mBMDCs enhanced the survival rate of tumor‐bearing mice over the course of 14 days (Figure 7B), which was not observed for the OVA_257‐264‐pulsed LPS‐maturated mBMDCs. Compared with the sham control group, on day 8 after the mBMDC injection, both the OVA_257‐264‐pulsed NZ‐maturated and LPS‐maturated mBMDCs made the tumor volume decrease. However, from day 10 after the mBMDC injection, only the OVA_257‐264‐pulsed NZ‐maturated mBMDCs significantly suppressed the growth of tumor (Figure 7C,D). These findings indicate that the anti‐tumor effects induced by the NZ‐maturated mBMDCs are more long‐lasting than those induced by the LPS‐maturated mBMDCs. These deviated anti‐tumor effects are likely due to that the OVA_257‐264‐pulsed NZ‐maturated mBMDCs promoted the generation of relatively long‐lived PD‐1^lowCD44^high memory effector CTLs in mice, while the OVA_257‐264‐pulsed LPS‐maturated mBMDCs led to the short‐lived PD‐1^highCD44^low terminal effector CTLs. In addition to this hypothesized mechanism, further studies shall also explore the timeframe of CTL differentiation and the characteristics of antigen‐specific CTLs in the mice injected with the antigen‐pulsed NZ‐maturated mBMDCs. The ultimate aim is to optimize the treatment strategy of NZ‐maturated DCs, including the quantity of injected cells and the frequency of injections, to achieve a more robust and sustained anti‐tumor response. Moreover, the PD‐1^lowCD44^high CTLs were reported to be rescued from apoptosis and to restore their expansion and effector functions via the application of immune checkpoint inhibitors,^{[ 47 ]} raising the potential to further improve the anti‐tumor effect via a combinational treatment of NZ‐maturated DCs and immune checkpoint inhibitors. In short, these findings verify that the NZ‐maturated mBMDCs are advantageous over the LPS‐maturated mBMDCs in the suppression of tumor growth in vivo.

Figure 7.

The NZ‐maturated mBMDCs suppress the growth rate of tumors in vivo. A) Schematics of the experimental timeline to examine the in vivo effect of the OVA_257‐264‐pulsed mBMDCs maturated with silica NZs(‐P245‐N3) or LPS on the mice inoculated with B16‐OVA melanoma cells. The tumor‐bearing mice injected with the vehicle (HBSS) without any mBMDCs served as the sham control group. B) The Kaplan‐Meier survival curves of tumor‐bearing mice after the injection of mBMDCs (log‐rank test, p = 0.0493, sham control versus NZ‐maturated mBMDCs). C) The growth rate of tumors in individual mice. D) The summarized tumor volume of each experimental group (^*sham control versus NZ‐maturated mBMDCs, day 8: p = 0.0507, day 10: p = 0.0004, day 12: p<0.0001; ^#LPS‐maturated mBMDCs versus NZ‐maturated mBMDCs, day 10: p = 0.0673, day 12: p = 0.0123; ^†sham control versus LPS‐maturated mBMDCs, day 8: p = 0.0626, two‐way ANOVA with Tukey's HSD post‐hoc test). The error bars represent the standard error of the mean. n = 9 for the sham control group, n = 10 for the NZ‐maturated mBMDCs and LPS‐maturated mBMDCs, pooled from three independent trials.

2.6. Silica NZ‐promoted Maturation of Human Monocyte‐Derived Dendritic Cells

To explore the possibility of clinical translation, we examined the effects of silica NZs on the maturation of hu‐moDCs differentiated via a rapid and serum‐free protocol.^{[ 48 ]} The peptide pool of colon cancer‐related antigen, carcinoembryonic antigen (CEA), was used as the model tumor‐associated antigen (Figure  8A). Unlike the results on the mBMDCs, the silica NZs significantly increased the expression levels of MHC‐I and MHC‐II in the hu‐moDCs with or without pulsing with CEA (Figure 8B–D), indicating that the NZ‐induced expressions of MHC molecules are obscured in the mBMDCs or in the serum‐supplemented culture system. Consistently with the observations on the mBMDCs, the silica NZs also upregulated the expressions of the co‐stimulatory molecule CD80 and CD86 in the hu‐moDCs with or without pulsing with CEA (Figure 8E–G). However, the distinctively enhanced expressions of CCR7, XCR1, and DC‐SIGN were only observed in the NZ‐maturated hu‐moDCs pulsed with CEA, but not in those without antigen pulsing (Figure 8E,H–J). The pulsing with CEA did not cause significant changes in the TCPS‐attached hu‐moDCs (Figure 8H–J), illustrating that CEA might initiate another signaling to amplify the NZ‐mediated upregulation of CCR7, XCR1, and DC‐SIGN. It has been reported that the glycosylated CEA is a ligand for DC‐SIGN,^{[ 49 ]} and the activation of DC‐SIGN can modulate the maturation of DCs.^{[ 50 ]} The oligopeptides of CEA could possibly be glycosylated by the reducing sugars such as glucose in the culture medium via the Maillard reaction.^{[ 51 ]} The silica NZs had slightly upregulated the expression of DC‐SIGN in the unpulsed hu‐moDCs (p = 0.0922, Figure 8E,J). Hence, it is possible that the glycosylated oligopeptides of CEA ligate with the DC‐SIGN on the NZ‐maturated hu‐moDCs to augment the expressions of CCR7, XCR1, and DC‐SIGN. This hypothesis, as well as the possible use of agonistic antibodies against DC‐SIGN to boost the NZ‐induced protein expressions in hu‐moDCs, will be investigated in future studies. Moreover, enhanced degranulation and production of IFNγ were observed in the CTLs co‐cultured with the NZ‐maturated hu‐moDCs (Figure S15, Supporting Information), demonstrating the capacity of NZ‐maturated hu‐moDCs to activate the CTLs. Overall, these findings demonstrate the feasible application of silica NZs to maturate the anti‐tumor hu‐moDCs in vitro.

Figure 8.

Silica NZs promote the maturation of human monocyte‐derived dendritic cells (hu‐moDCs). A) Schematics of the experimental timeline to examine the properties of hu‐moDCs maturated with silica NZs(‐P245‐N3). The hu‐moDCs attached to the TCPS, without any maturation treatment, served as the negative control group. The TCPS‐attached hu‐moDCs maturated with 100 ng mL⁻¹ LPS for 20 h were used as the positive control for the validation of the assay method. B–J) Representative histograms (B,E) and C) the geometric MFI of MHC‐I, D) MHC‐II, F) CD80, G) CD86, H) CCR7, I) XCR1 and J) DC‐SIGN in the CD11c⁺ MHC‐II⁺ hu‐moDC subset after the 20 h maturation period (C, F: n = 6, D: n = 8, G: n = 5, H‐J: n = 4). The lines on histograms mark the peaks in histograms of the unpulsed TCPS‐attached hu‐moDCs. The error bars represent the standard error of the mean. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, (TCPS‐attached versus NZ‐maturated hu‐moDCs, with or without CEA pulsing) two‐way ANOVA with Šídák's multiple comparisons test, (comparison among the hu‐moDCs without CEA pulsing) Welch's ANOVA with Dunnett's T3 post‐hoc test (C‐D, F, H), one‐way ANOVA with Tukey's HSD post‐hoc test (G, I‐J).

3. Conclusion

We use the mBMDCs as a cell model to manifest the capacity of silica nanomatrices in promoting in vitro maturation of DCs. The outcome of silica nanomatrix‐mediated maturation can be modulated via the engineering of nanostructures in terms of shapes, P, and N. The optimal silica nanomatrices, NZs‐P245‐N3, generate a maturation phenotype consisting of upregulated expressions of CD86, CCR7, XCR1, and DC‐SIGN, and elevate the endocytic capacity of DCs. Compared with the conventionally LPS‐maturated mBMDCs, the silica NZs‐P245‐N3 maturate the mBMDCs to acquire higher expression levels of CCR7, XCR1, and DC‐SIGN, as well as to retain the enhanced endocytic capacity after the maturation period. The NZs‐P245‐N3‐induced upregulations of CD86, XCR1, and DC‐SIGN are mediated by FAK that is mechanically activated at the curved cell adhesions formed in the deformed plasma membranes at the DC‐NZ interfaces. The distinctive maturation phenotypes of NZs‐P245‐N3‐maturated mBMDCs enable the activation of antigen‐specific CTLs into the PD‐1^lowCD44^high effector memory CTLs in vitro and a relatively prolonged suppression of tumor growth in vivo. Moreover, the NZs‐P245‐N3‐induced surface phenotypes of maturated DCs are reproducible in the hu‐moDCs. Whereas the NZs‐P245‐N3‐induced increase of MHC molecules is observed in the hu‐moDCs maintained in the serum‐free condition but concealed in the mBMDCs cultured in the serum‐supplemented system. In conclusion, our work demonstrates the superior effects of silica NZs on in vitro maturation of anti‐tumor DCs in both murine and human cells, implying their potential to be developed into a new strategy of cellular immunotherapy for cancer treatment.

4. Experimental Section

Fabrication and Characterization of Extracellular Silica Nanomatrices

Glancing angle deposition (GLAD) was performed to produce silica nanomatrices vertically protruding on a supporting substrate.^{[ 52} , ^{53 ]} Using a custom‐built physical vapor deposition system equipped with the high vacuum chamber (10^‐7–10^‐6 Torr, JunSun Tech Co. Ltd., Taiwan), silica (SiO₂) was evaporated at a rate of ≈0.4 nm s⁻¹ with electron‐beam accelerating voltage of 8.0 kV and emission current of 83–87 mA. Silica was deposited on indium tin oxide (ITO)‐coated glasses or silicon wafers, at a deposition angle (α) of 87° with respect to the substrate's normal direction. The deposition process was monitored using a quartz crystal microbalance located in the vicinity of the sample. An ethanol–water cooling system was used to maintain the substrate temperature at room temperature during GLAD. To sculpt silica in the shape of NZs, the substrate was rotated back and forth by 180° at a given time interval, with which the P of silica NZs was controlled. To produce the tilted nanorods, silica was deposited at α of 87° without rotating the substrate.^{[ 54 ]} To produce the left‐handed and right‐handed nanohelices (NHs), the substrate was rotated counterclockwise and clockwise, respectively, at the rotation rate (R _r, in the unit of degree per second) calculated using the following equation,^{[ 55} , ^{56 ]}

$$R_{\mathrm{r}}=360R_{\mathrm{d}}/P_{\mathrm{NH}}$$ (1)

where R _d was the calibrated deposition rate of SiO₂ on the substrate surface (0.28 nm s⁻¹) at α of 87° and P _NH was the designated pitch of NHs. To produce vertical nanorods, the substrate was rotated at R _r of 6 ° s⁻¹. The silica NZs deposited on silicon wafers were mechanically spilled, and the freshly exposed cross‐sections were characterized with SEM (LEO 1530, Zeiss).

Animals

The C57BL/6 mice were purchased from the Laboratory Animal Services Centre of the Chinese University of Hong Kong. The mice were housed in the animal room with a 12 h/12 h light‐dark cycle with food and water ad libitum. All experimental procedures on mice were approved by the Department of Health of the HKSAR government.

Primary mBMDC Culture

The culture media and supplements were purchased from ThermoFisher unless otherwise specified. The culture method of mBMDCs was optimized based on the previously published protocol.^{[ 57} , ^{58 ]} The bone marrows were isolated from the femur and tibia of the 6‐ to 9‐week‐old C57BL/6 mice. The red blood cells in the bone marrow were lysed by applying a brief hypotonic shock with the autoclaved MilliQ water, which were quenched by adding the corresponding volume of 10 × PBS. The isolated mononuclear cell suspension was seeded at a density of 1 × 10⁶ cells per mL (3 mL per well on the 6‐well plates) in the complete RPMI medium (RPMI 1640 medium supplemented with 15 mm HEPES and 0.5% (v/v) penicillin‐streptomycin‐neomycin (PSN) mixture) supplemented with 10% (v/v) heat‐inactivated fetal bovine serum (HI‐FBS), and 20 ng mL⁻¹ recombinant murine granulocyte‐macrophage colony‐stimulating factor (rmGM‐CSF, PeproTech, Cat# 315‐03) and interleukin‐4 (rmIL‐4, PeproTech, Cat# 214‐14). On day 3, half of the culture media were replaced with the complete mBMDC medium containing rmGM‐CSF and rmIL‐4. On day 6, the non‐adherent and semi‐adherent immature mBMDCs (im‐mBMDCs) were collected. For the mBMDC maturation, the im‐mBMDCs were seeded on TCPS plates with or without 100 ng mL⁻¹ LPS (ThermoFisher, Cat# 00‐4976‐93), or seeded on the nanomatrices in the complete RPMI medium supplemented with 10% (v/v) HI‐FBS for 20 h. The im‐mBMDCs were seeded at a density of 0.7 × 10⁵ cells per well on 48‐well plates with or without 1 cm‐diameter nanomatrices for the co‐culture with CTLs, 1.2 × 10⁵ cells per well on 12‐mm‐diameter polystyrene coverslips (SPL) or 1‐cm‐diameter nanomatrices on 4‐well plates for confocal imaging, 3.5–7 × 10⁵ cells per well on 12‐well plates with or without 2‐cm‐diameter nanomatrices for flow cytometry and Western blot, and 25 × 10⁵ cells per well on 6‐well plates with or without 3‐cm‐diameter nanomatrices for in vivo mBMDC injection. To examine the role of FAK in the NZ‐mediated effects, the mBMDCs were treated with 5 µm FAK inhibitor PF‐573228 (Selleckchem, Cat# S2013) or 0.02% (v/v) DMSO (Sigma, Cat# D2650) at 1 h after cell seeding. To pulse the mBMDCs with antigen OVA_257‐264, the mBMDCs were pre‐incubated in the complete RPMI medium supplemented with 1% (v/v) HI‐FBS for 30 min, and subsequently treated with 1 nm OVA_257‐264 peptide (InvivoGen, Cat# vac‐sin) in the complete RPMI medium supplemented with 1% (v/v) HI‐FBS for 2 h. Afterward, the OVA_257‐264 peptide was removed. The antigen‐pulsed mBMDCs were then used for co‐culture with the CTLs or were detached via the 15‐min incubation in an Accutase cell detachment medium (ThermoFisher, Cat# 00‐4555‐56) at 37 °C for the in vivo injection.

Co‐culture of mBMDCs and CTLs

The 6‐ to 8‐week‐old C57BL/6 mice were immunized via the subcutaneous injection of 100 µL mixture of 10 µg OVA_257‐264 peptide and 50% (v/v) complete Freud's adjuvant (InvivoGen, Cat# vac‐cfa‐10) at the scruff. After 7 days of immunization, the spleen and major lymph nodes were isolated from the immunized mice and dissociated mechanically by smashing the tissues on the cell strainer using the syringe plunger. The red blood cells in splenic cell suspension were lysed via a brief hypotonic shock with sterile MilliQ water which was quenched by adding the corresponding volume of 10 × PBS. The CD8⁺ CTLs were purified from the splenic and lymph node cell suspension via the negative cell isolation using the CD8a⁺ T cell isolation kit (Miltenyi Biotec, Cat# 130‐104‐075) according to the manufacturer's instruction. The CTLs were then co‐cultured with the OVA_257‐264‐pulsed or unpulsed mBMDCs in the complete RPMI medium supplemented with 5% (v/v) HI‐FBS in a DC:CTL ratio of 1:10.

Primary hu‐moDC Culture

The peripheral blood mononuclear cells (PBMCs) were isolated from the buffy coat of the healthy donors (obtained from and approved by the Hong Kong Red Cross Blood Transfusion Service; REC reference number: HASC/17‐18/0795) using the density gradient centrifuge (Lymphoprep, Stemcell Technologies, Cat# 07851). The monocytes were isolated from the PBMCs via positive cell isolation using the human CD14 microbeads (Miltenyi Biotec, Cat# 130‐050‐201) according to the manufacturer's instruction, and seeded at a density of 1.4 × 10⁶ cells per mL on 6‐well plates in serum‐free AIM‐V medium supplemented with 25 ng mL⁻¹ recombinant human GM‐CSF and IL‐4 (Miltenyi Biotec, Cat# 130‐093‐866 and 130‐093‐922). After 24 h of culture, the monocytes were differentiated into the immature hu‐moDCs, in which at least 80% of cells had lost the surface expression of CD14. For the maturation of hu‐moDCs, the immature hu‐moDCs were seeded in fresh AIM‐V medium at a density of 0.6 × 10⁵ cells per well on 48‐well plates with or without the 1‐cm‐diameter NZs for the co‐culture with CTLs, or 2 × 10⁵ cells per well on 12‐well plates with or without 2‐cm‐diameter NZs for flow cytometry. The hu‐moDCs were pulsed with 4 µg mL⁻¹ PepMix human CEA peptide pool (JPT Peptide Technologies, Cat# PM‐CEA) right after being seeded on the 12‐well or 24‐well plates. After 20 h of maturation, the hu‐moDCs were used for the subsequent experiments.

Murine Tumor Models and In vivo mBMDC Injection

The murine B16 melanoma cells expressing the membrane‐bound OVA (B16‐OVA, Cat #CE20513, Crisprbio, Beijing, China) were cultured in the DMEM supplemented with high glucose, GlutaMAX, 10% (v/v) HI‐FBS, 0.5% (v/v) PSN and 400 µg mL⁻¹ selective agent G418 sulfate (ThermoFisher, Cat# 10131035). To ensure that most of the B16‐OVA cells could enter the exponential growth phase at the time of injection, the cells used for in vivo inoculation had passage number ≤ 15 and were expanded in the complete DMEM medium without selective agent G418. The B16‐OVA cells were detached using the TrypLE Express enzyme (ThermoFisher, Cat# 12604021) and passed through the 100‐µm cell strainer. The 9‐ to 11‐week‐old C57BL/6 female mice (with an average weight of 20 g) were subcutaneously injected with 2.5 × 10⁵ B16‐OVA cells in 50 µL HBSS into their shaved right flanks, under ketamine‐xylazine anesthesia. On day 8 after the inoculation of cancer cells, the visible tumors were formed. The tumor‐bearing mice were randomly divided into three groups. Under ketamine‐xylazine anesthesia, 1.75 × 10⁶ OVA_257‐264‐pulsed mBMDCs maturated with silica NZs or LPS in 100 µL HBSS were subcutaneously injected into the tumor‐bearing mice at the area next to the tumor. The sham control group was subcutaneously injected with 100 µL HBSS without any mBMDCs. Tumor size was monitored every two days using the calipers for a total of 14 days. The tumor volume was calculated using the following formula:

$$\left(\mathrm{Length}\times \mathrm{Widt}{\mathrm{h}}^2\right)/2$$ (2)

The mice were sacrificed and considered as death when their tumor volume reached 1000 mm³.

Statistical Analysis

The statistical analyses were conducted in the Prism 9 software (GraphPad), except that the homogeneity of variance between groups was analyzed using Levene's test based on median in the SPSS Statistics version 26. According to the Levene's test results, two‐tailed Student's t‐test or two‐tailed Welch's t‐test was used for the mean comparison between two groups; and one‐way ANOVA followed by Tukey's HSD post‐hoc test or Welch's ANOVA followed with Dunnett's T3 post‐hoc test was used for the mean comparison of more than two groups. Two‐way ANOVA followed by Tukey's HSD post‐hoc test or Šídák's multiple comparisons test was used for the mean comparison between groups with two independent variables. The p‐values <0.05 were considered as statistically significant. Numerical data were presented as mean ± standard error of the mean.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

S.W.T., A.K.L.C., and Q.P. contributed equally to this work. S.W.T. performed biological experiments, analyzed data, and drafted the manuscript. A.K.L.C. secured the source of PBMCs and advised the experiments. Q.P. fabricated silica nanomatrices and performed material characterization. S.W.T., A.K.L.C., S.Q.Z., Z.F.H., and K.K.L.Y. designed this study. S.Q.Z., Z.F.H., and K.K.L.Y. supervised this study. All authors contributed to discussion of results and editing of the manuscript.

Supporting information

Supporting Information

Tam S. W., Cheung A. K. L., Qin P., Zhang S., Huang Z., Yung K. K. L., Extracellular Silica Nanomatrices Promote In Vitro Maturation of Anti‐tumor Dendritic Cells via Activation of Focal Adhesion Kinase. Adv. Mater. 2025, 37, 2314358. 10.1002/adma.202314358

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.