In a first stage, the influence of the washing alcohol on carboxylic surface functionalization in acidic conditions was assessed. In previous work, ethanol was used to remove the structure directing agent (Yamada et al., 2012; Seré et al., 2018). Inefficient protein conjugation with these particles suggested Fischer esterification of the carboxyl groups when the template was removed with ethanol (Master Organic Chemistry LLC, 2020). To prevent this reaction, NPs were washed with TBA instead. TBA offers more steric hindrance, inhibiting the electrophilic carbon of the carboxylic acid to attack the alcohol. It is therefore less reactive than ethanol. Functionalization of these particles was first assessed with FTIR represented in Figure 1. Non-functionalized particles (MSNP-OH, blue curve) show absorption bands at 480 cm⁻¹ corresponding to O–Si–O bending, 800 cm⁻¹ for Si-O bending, 950 cm⁻¹ for SiO-H bending, in the range of 1,000–1,200 cm⁻¹ for Si–O–Si stretching and a broad band around 3,500 cm⁻¹ for SiO–H stretching. The band at 1,640 cm⁻¹ is probably due to scissor bending vibrations of molecular water. Coating the particle surface with carboxyl groups, introduces additional bands at 1,720 cm⁻¹ corresponding with the C=O stretch and 1,420 cm⁻¹ which corresponds to the C–O stretch. The latter band is slightly visible in the FTIR spectrum of TBA washed particles (MSNP-COOH, red curve in Figure 1), but not present for ethanol washed particles (MSNP-COOH**, orange). Other infrared absorption bands characteristic for carboxyl groups, such as the O–H and C–H stretching band at 3,500 cm⁻¹ and the O–H bending at 950 cm⁻¹, overlap with the spectrum for non-functionalized particles.

In essence, the FTIR spectra gave a first indication of successful carboxylic functionalization of MSNPs, especially for TBA washed MSNPs. The distinctive absorption bands for carboxyl groups are less abundant for ethanol washed particles, confirming the Fisher esterification hypothesis. It was therefore decided to use TBA during template removal.

SAXS and TEM analysis confirmed the mesoporous structure (Supplementary Figure 1 and Figure 5). Zeta potential and TGA were performed for aforementioned samples and amine functionalized particles (MSNP-NH₂) in order to assess the grafted functional groups. Figure 2A, shows the zeta potential as a function of pH for MSNPs with different surface functionalization. The shape of the curve confirms successful carboxylic surface functionalization. As a reference, the isoelectric point (IEP) is considered. Non-functionalized MSNPs have an IEP at a pH value of 3.5. This value shifts to 6.7 for MSNP-NH₂, due to the change of surface charge. MSNP-COOH show no IEP as the zeta potential of these samples is negative over the complete pH range, confirming efficient carboxylic surface functionalization. Following the same reasoning, the shift of MSNP-COOH toward lower zeta potential values in the higher pH range suggests more carboxyl groups as compared to ethanol washed particles (MSNP-COOH**).

TGA decomposition profiles give a quantitative indication of the functional groups and were normalized to the weight of dry samples. This weight was set at 150°C, after removal of impurities and adsorbed moisture (Musso et al., 2015). Three regions can be observed for samples with functional groups and each can be ascribed to a certain surface chemistry (Gallas et al., 2009; Kozlova and Kirik, 2010; Musso et al., 2015). The first region from 150 to 370°C for carboxylated MSNPs and from 150 to 500°C for MSNP-NH₂, is attributed to the loss of silanol groups. The decomposition profile of carboxylated MSNPs shows a plateau before going into the second region ranging from 440 to 700°C. This steep decline is ascribed to the carboxylic surface functionalization. Amine functionalized MSNPs show a steep slope in the range from 500 to 700°C, representing amine functionalization. The gradual weight loss when increasing the temperature beyond 700°C is ascribed to further dehyroxylation. The TGA curves in Figure 2B demonstrate that the particles are grafted with carboxyl groups. The curves of carboxylated MSNP coincide for both washing methods in the first region, indicating an equal amount of silanol groups. In the second region, which represents the carboxylic surface functionalization, the TBA washed particles (MSNP-COOH) show slightly more weight-loss than ethanol washed particles (MSNP-COOH**). This confirms a higher amount of carboxyl groups, supporting the esterification hypothesis for ethanol washed particles.

The particles were degraded in 0.9% NaCl solution and molybdenum blue chemistry was used to assess the degradation kinetics shown in Figure 3. The results indicate that degradation occurs fastest for amine functionalized particles (green curve), followed by non-functionalized (blue curve) and carboxylated particles (orange and red curve). It has been shown that the negatively charged ions of a medium are the main mediators for particle degradation (Seré et al., 2018). The fast degradation for amine functionalized MSNPs can therefore be ascribed to the attractive electrostatic interactions between the degrading ions and the positively charged particles. Both silanol groups of non-functionalized particles as well as carboxyl groups of carboxy-MSNP, repel the negative ions. Due to the larger negative potential of carboxylated MSNPs, repulsion is higher and degradation therefore slower as compared to non-functionalized particles. As for the particles produced with different template removing alcohols, TBA washed particles (MSNP-COOH, red curve) degrade slower than ethanol washed particles (MSNP-COOH**, orange curve). Data in Figure 2A show that ethanol washed particles have lower absolute zeta potential values, suggesting a decreased ability to repel the degrading ions and resulting in faster degradation.

Based on these results, we conclude that the washing procedure has an impact on the surface functionalization. In addition, TBA washed particles degrade slightly slower, increasing their availability for DCs. We therefore selected TBA as the better washing alcohol for further NP production in this work.

After improving the washing procedure, the optimal amount of CES for carboxylation was determined. Carboxylic surface functionalization of mesoporous silica materials using the CES precursor, has mainly received attention in the catalytic research field where SBA-15 sieves are functionalized (Han et al., 2007; Tsai et al., 2009). In addition, it is also used for carboxylic mesoporous silica coating of superparamagnetic iron oxide NPs (Sharifi et al., 2013). Other groups produced MSNP-COOH by producing 5-(Triethoxysilyl)pentanoic acid and incorporating it in a one-pot synthesis. The silica end of this carboxyl precursor resembles the TEOS structure more closely but requires another synthesis step (Feinle et al., 2017). Keeping the industrial scaling and costs in mind, water soluble CES was used in this work. Traditional co-condensation might cause the formation of additional silica entities in the solution, especially when higher amounts are used. Earlier research could therefore only implement trace amounts of CES (Gu et al., 2013; Sharifi et al., 2013). Since this work improved the functionalization route, the optimal carboxylation was investigated.

MSNP-COOH samples were produced following the standard protocol and using TBA to remove the CTAB template. The samples were denoted as MSNP-COOH-l, MSNP-COOH-m, and MSNP-COOH-h as the amount of CES was varied 0.075, 0.150, and 0.300 mmol, respectively. Figure 4A represents zeta potential measurements. Incorporating more carboxyl groups on the particle surface, increases inter-particle repulsion in the higher pH range (Duffy and Hill, 2011). It results in a higher zeta potential and more stable NPs. MSNP-COOH-h samples in Figure 4A (triangle full line) do not follow this expected trend. The progression of the zeta potential curve deviates, suggesting saturation of the NP surface when higher amounts of CES are used during synthesis.

The amount of grafted carboxyl groups was estimated with TGA. The normalized decomposition profiles are illustrated in Figure 4B and associated results in Table 3. The first temperature region from 150 to 370°C, represents the mass loss due to silanol groups. MSNP-COOH-l show more mass loss in this window as compared to the other two samples. The mass loss due to silanol groups of MSNP-COOH-m and MSNP-COOH-h samples is equal. This suggests an unsaturated particle surface when functionalized with 0.075 mmol CES; while adding 0.15 mmol CES or more, decreases the amount of silanol groups to a constant value, indicating particle saturation. The same trend is observed in the temperature range ascribed to the weight loss of carboxyl groups from 440 to 700°C. The amount of carboxyl groups is lower when a low amount of CES (0.075 mmol) is used. On the other hand, amount of grafted carboxyl groups is the same for both 0.15 and 0.30 mmol CES. An equal amount of carboxyl groups suggests saturation of the NP surface, which occurs due to steric hindrance of the carboxyl groups. Adding more CES results therefore in the same amount of carboxylic surface functionalization.

Particles functionalized with 0.15 mmol CES were considered to be optimal for protein loading, as this amount seems to saturate the particle surface.

Finally, in search of the optimal particle, two different silica precursors were examined. TEOS is a well-known and widely used precursor, a standard in the synthesis of silica NP synthesis (Rahman and Padavettan, 2012; Wu et al., 2013; Liberman et al., 2014; Zulfiqar et al., 2016). TPOS is known to render slightly larger particles and larger size distribution due to a lower hydrolysis rate during synthesis (Yamada et al., 2012). The size of TPOS based NPs would still be suitable for passive targeting of DCs (Yamada et al., 2012). Large particulate systems are mainly taken up by DCs, then transferred toward the lymph nodes where they stimulate the CD8⁺ T-cells. By contrast, smaller particles are taken up by macrophages in the skin which are considered to be tissue resident (Manolova et al., 2008). We therefore aim to produce MSNPs between 20 and 50 nm in this work. In addition, the system's diameter increases with 300–500 nm by coating the MSNPs with tumor proteins to produce the nanovaccine. Therefore, NP monodispersity is not required, as the conjugation of peptides and proteins causes polydispersity of the final product. Our hypothesis is that slower hydrolysis during synthesis could result in more stable particles with regard to biodegradation, arousing interest in TPOS based particles.

TEOS or TPOS silica precursor (2.75 mmol) were used during synthesis and both non-functionalized and carboxy-MSNP were produced for further characterization. Figures 5, 6 show typical TEM images and the size distribution of carboxy-MSNP produced with TEOS (MSNP-COOH) and TPOS (MSNP-COOH*), respectively. As expected, TPOS based particles are larger with an average core diameter of (48 ± 8) nm. The TEOS precursor rendered smaller particles with an average diameter of (29 ± 4) nm. In addition, the size distribution of TPOS based particles is larger.

The MSNPs were characterized using zeta potential, TGA, and biodegradation experiments. Zeta potential and TGA data were comparable for TEOS and TPOS based nanoparticles, for both non-functionalized as well as carboxy-MSNP (data not shown). Thus, the silica precursor does not influence the electrochemical stability of the particles in solution, nor does it affect the surface chemistry. Biodegradation experiments were performed in 0.9% sodium chloride and results are shown in Figure 7. All MSNPs degraded in saline solution. The degradation behavior of non-functionalized and carboxylated MSNPs is in line with aforementioned conclusions. Particles with carboxylic surface chemistry degrade slower due to the repulsive interaction with the degrading ions in the medium. In contrast to our expectations, TEOS based particles degrade slower than particles synthesized with TPOS comparing particles with identical functionalization. The degradation curves follow the same course, but are shifted when TEOS is used for core synthesis. At a degradation of 50%, the curve is shifted for 12 and 18 h in case of non-functionalized and carboxylated MSNPs, respectively. The effect of the silica precursor is in contrast to what is expected from the study of Yamada et al. (2012). Probably, the synthesis route used in this work, affects the pore and particle size and hence the biodegradability. Slower degradation implies more chance of DCs to take up the nanovaccine in the human body.

The optimal particles for this work were MSNP-COOH, synthesized with a TEOS precursor, functionalized with 0.15 mmol CES and washed with TBA.

The first in vitro steps were taken by determining the optimal MSNP dose to treat and analyze DCs. The dose response relationship between MSNPs and the viability of DCs is shown in Figure 8A. The graph was established by culturing DCs and exposing them to MSNPs. After flow cytometry acquisition, the viability was determined using an adjustment of the general gating strategy in Supplementary Figure 2. Single cell selection (Supplementary Figures 1A,B) was followed by extracting the CD11c⁺, FVD⁻ populations in a single gate as represented in Figure 8B. Control conditions were taken into account with iDCs (untreated) and mDCs (LPS stimulated). The test conditions consisted of DCs incubated with: 50, 100, 200, and 400 μg MSNPs per million DCs. Each NP treated DC culture received an LPS stimulation signal 90 min after the particles were introduced. Data were pooled from two experiments containing five cultures per condition.

The main factors affecting NP cytotoxicity are surface functionalization, size, and concentration. Although MSNPs are known to be biocompatible and biodegradable, their cytotoxicity significantly increased when their concentration increased as shown in Figure 8. Anova pairwise comparison was used to determine statistical significance, using iDC and mDC as control conditions. No significant difference was observed when DCs were treated with 50 or 100 μg MSNPs per million DCs. Increasing the quantity of MSNPs significantly decreased DC viability (***p ≤ 0.001). The threshold toward cytotoxicity for the NPs in this work was set at 100 μg per million DCs and was therefore used as standard for loading DCs. In addition, the viability of DCs treated with 50 or 100 μg MSNPs was significantly higher than DCs treated with higher NP doses (not indicated on graph).

The uptake of particles and nanovaccine by DCs was examined using confocal microscopy. First, DCs were incubated with labeled MSNPs (MSNP-FITC), stimulated with LPS and stained for MHC class I molecules. To gain a better insight in the uptake process, half of the cells were incubated at 4°C and the other half at normal physiological temperatures, i.e., 37°C. The results are illustrated in Figure 9. Figures 9A,B show images of MSNP treated DCs, incubated at 37°C. The green signal was attributed to the FITC labeled MSNPs, the red signal to PE-Cy7 labeled MHC class I molecules. Since MHC class I molecules are membrane molecules, they visualize the cells' exterior. The green signal of MSNPs is concentrated inside the red MHC signal. No particles were observed being attached to the cell surface, indicating cellular uptake of the MSNPs. In addition, Figures 9C,D, display the uptake mechanism since these cells were incubated at different temperatures. It is known that cell metabolism shuts down at 4°C (Figure 9C) (De Vleeschouwer et al., 2005; Belmans et al., 2017). The absence of the green MSNP signal, stipulated no particle uptake by DCs. Incubation at 37°C on the other hand, did show particle uptake within the same experiment (Figure 9D). In conclusion this demonstrated that particle internalization is an active process (ATP dependent) and not merely NPs penetrating the membrane of DCs. This was an indication of endocytosis, meaning that particles and transported proteins enter the cross-presentation pathway.

Nanovaccine uptake, or at least the internalization of the transported tumor proteins, is of great importance to induce an anti-tumor immune response. To determine protein uptake by DCs after treatment with the nanovaccine, MSNPs were loaded with OVA, Alexa Fluor 488 conjugates (OVA-Alexa). The same incubation and staining process was followed as for particle uptake. Confocal microscopy images confirmed uptake of OVA-Alexa, as shown in Figure 10. The green OVA signal is observed within the red MHC signal. Moreover, a yellow overlay of the green OVA signal and the red MHC class I signal was observed for several cells. This was a first suggestion of cross-presentation of OVA by DCs. The cross-presentation was quantitatively determined with flow cytometry.

In a first stage of these experiments, the effect of MSNPs on DCs was investigated by incubating the cells with as-synthesized and UV-sterilized particles. The effect of the nanovaccine was tested as well by incubating the cells with MSNP-OVA. The viability and maturation effect on the DCs were examined with flow cytometry. No significant difference was observed between the viability of treated DCs and control conditions iDC and mDC (data not shown). In addition, the sterilized particles were characterized and no changes were observed before or after sterilization. These results imply that the sterilization step can be included for further application as it does not negatively affect DCs, nor does it affect the quality of the NPs.

After demonstrating particle and vaccine uptake, the maturation effect of the treatment on DCs was examined. Figure 11 shows the expression levels of the most important maturation markers of these cells, which include CD86, IA/IE, and CD40. MHC class I and CD80 expression levels were excluded from the data as these levels were high and comparable for all samples. The complete data can be found in the Supplementary Figure 3. The results were obtained by pooling data of six experiments with three cultures per condition per experiment.

iDCs showed low expression levels for the maturation markers, as expected (Figure 11, yellow). The immature state is typically specialized in sampling its environment in vivo. Increased expression of costimulatory molecules and MHC only takes place once pathogens are encountered and stimulation occurs. mDCs thus represented the positive control group for the maturation effect (Figure 11, orange). The expression levels of the shown markers was significantly higher for mDCs (Anova Dunnett's multiple comparisons test (**p ≤ 0.01), not indicated on the graph). DCs treated with particles alone (Figure 11, blue) did not receive a maturation trigger, in order to assess the effect of MSNPs. These DCs showed significantly increased expression levels of CD86 and MHC class II as compared to iDCs (*p ≤ 0.1 for MSNP and **p ≤ 0.01 for UV MSNP). In addition, higher expression of CD40 was observed. However, the expression levels of maturation markers were still lower than those of mDCs. These results were significant for CD40 (*p ≤ 0.1) and CD86 (*p ≤ 0.1 for MSNP and ***p ≤ 0.001 for UV MSNP). We concluded that MSNPs have a moderate adjuvant effect on DCs and thus possibly the adaptive immune system. Incorporating immunomodulatory agents, such as TLR ligands (R848), on the particle surface could increase the stimulating effect further. In addition, DCs treated with MSNPs and UV sterilized MSNPs exhibited comparable expression levels of costimulatory and MHC molecules. As expected, DCs treated with the nanovaccine exhibited similar expression levels of MHC and costimulatory molecules as compared to mDCs (Figure 11, red, no significant difference). Both conditions were stimulated with LPS. The expression levels of DCs treated with the nanovaccine seemed even higher, which was possibly associated with the accumulative effect of LPS and NPs. MSNPs caused moderate maturation, which is further increased by LPS. These results indicated the adjuvant effect of MSNPs on DCs. which was a first suggestion of DCs being able to communicate the protein information delivered by NPs to the T-cells of the immune system.

In order to investigate the cross-presentation of OVA on MHC class I molecules, DCs were loaded with the nanovaccine and cells were stained for flow cytometry with SIINFEKL-PE. This antibody stains the combination of the SIINFEKL peptide and MHC class I, and is therefore a quantitative indication of the cross-presentation of MSNP delivered OVA. The data and statistical analysis can be found in the Supplementary Figures 4, 5 and Supplementary Tables 1, 2, results are shown in Figure 12. Data of four experiments were pooled, each experiment consisted of three to five groups per condition. Tukey HSD all pairwise comparison implicated significant differences between the cross-presentation of all negative control groups (i.e., iDC, mDC, and MSNP treated DCs) and DCs treated with the model nanovaccine (***p ≤ 0.001). In addition, cross-presentation of OVA on MHC class I by iDCs was significantly lower than OVA treated DCs (**p ≤ 0.01). A clear tendency toward higher cross-presentation was perceived when cells were treated with the nanovaccine. This indicated that DCs should be able to stimulate the CTLs of the immune system to specifically attack the tumor.