Effect of the Hydrophilic-Hydrophobic Balance of Antigen-Loaded Peptide Nanofibers on Their Cellular Uptake, Cellular Toxicity, and Immune Stimulatory Properties

Recently, nanofibers (NFs) formed from antigenic peptides conjugated to β-sheet-forming peptides have attracted much attention as a new generation of vaccines. However, studies describing how the hydrophilic-hydrophobic balance of NF components affects cellular interactions of NFs are limited. In this report, three different NFs were prepared by self-assembly of β-sheet-forming peptides conjugated with model antigenic peptides (SIINFEKL) from ovalbumin and hydrophilic oligo-ethylene glycol (EG) of differing chain lengths (6-, 12- and 24-mer) to investigate the effect of EG length of antigen-loaded NFs on their cellular uptake, cytotoxicity, and dendritic cell (DC)-stimulation ability. We used an immortal DC line, termed JAWS II, derived from bone marrow-derived DCs of a C57BL/6 p53-knockout mouse. The uptake of NFs, consisting of the EG 12-mer by DCs, was the most effective and activated DC without exhibiting significant cytotoxicity. Increasing the EG chain length significantly reduced cellular entry and DC activation by NFs. Conversely, shortening the EG chain enhanced DC activation but increased toxicity and impaired water-dispersibility, resulting in low cellular uptake. These results show that the interaction of antigen-loaded NFs with cells can be tuned by the EG length, which provides useful design guidelines for the development of effective NF-based vaccines.


Introduction
Peptide-based synthetic vaccines have attracted a significant amount of attention as a new generation of vaccines, because of their safety benefits and ease of production when compared with that of conventional whole pathogen-based vaccines [1,2]. However, poor immune responses are induced when only minimal antigenic epitopes are used without combining suitable adjuvants (immune stimulants), which are sometimes toxic. Nanocarrier-based delivery systems are a promising approach to overcome those drawbacks of peptide vaccines. In designing the nanocarrier, it is important to consider the interaction between the nanocarrier surface and cells, such as antigen presenting cells (APCs).
Over the past few decades, various nanocarriers have been developed, including liposomes [3][4][5], polymeric nanoparticles [6][7][8], and polymeric micelles [9]. In many of these systems, building block molecules for the construction of nanocarriers are first synthesized and then combined with antigenic peptides via several procedures, including nanomaterial formation and loading of antigenic peptides (encapsulation, chemical immobilization or physical adsorption), to give a nano-formulation. Recently, interaction of the NFs was found to be significantly dependent on EG length, whereas that of micelles was independent of EG length. Notably, uptake by DC of NFs composed of EG with a moderate length was effective, and the NFs activated DC without exhibiting significant cytotoxicity. The findings provide useful design guidelines for the development of effective nanofiber-based vaccines. DC of NFs composed of EG with a moderate length was effective, and the NFs activated DC without exhibiting significant cytotoxicity. The findings provide useful design guidelines for the development of effective nanofiber-based vaccines. Schematic illustration of the self-assembly process for nanofiber formation and the proposed model of highly antigen-loaded nanofibers based on previous structural study [36]. The schematic illustration was created by modification of Figure 1 and Figure 2 of reference [35].

Self-Assembly Behavior of EGn Peptides
We have reported previously the preparation of NFs in aqueous solution by heat-treatment of the EG12 peptide (EGn, where n is the length of the EG chain), which is composed of a β-sheet-forming sequence, an antigen sequence, and a hydrophilic oligo-ethylene glycol (12-mer). In this study, three building block peptide amphiphiles with different EG lengths (6-mer, 12-mer, and 24-mer, which are termed EG6, EG12, and EG24, respectively) were prepared for assessing the effect of EG length on cellular uptake of NFs, cytotoxicity, and immune stimulatory activities.
Initially, we investigated the effect of EG length on the self-assembly behavior of these peptides by estimating the CFC of each peptide. EGn peptides were incubated in the presence of the thioflavin T (ThT) dye at 300 μM in phosphate buffered saline (PBS) containing 5% dimethylsulfoxide (DMSO) at 37 °C for 24 h, and a time course of change in ThT fluorescence was measured. The ThT assay is often used to monitor the growth of amyloid-like nanofibers from their component peptides or proteins. A remarkable increase in ThT fluorescence was observed, indicating the formation of NFs by each peptide (Figures 2 and S1). After a certain period of time, the intensity of the ThT fluorescence reached a plateau value. The systems were in a dynamic equilibrium between fibrils and the peptide monomer when the ThT fluorescence intensity reached a plateau value. The concentration of the free peptide monomer at equilibrium corresponds to the CFC [38,39]. The peptide concentration in the supernatant following ultracentrifugation of the EGn peptide solutions incubated at 300 μM and 37 °C for 24 h provided estimates of the CFC values, which were 96.0 ± 3.6 μM for EG6, 72.4 ± 3.7 μM for EG12, and 83.8 ± 1.0 μM for EG24. Schematic illustration of the self-assembly process for nanofiber formation and the proposed model of highly antigen-loaded nanofibers based on previous structural study [36]. The schematic illustration was created by modification of Figures 1 and 2 of reference [35].

Self-Assembly Behavior of EG n Peptides
We have reported previously the preparation of NFs in aqueous solution by heat-treatment of the EG 12 peptide (EG n , where n is the length of the EG chain), which is composed of a β-sheet-forming sequence, an antigen sequence, and a hydrophilic oligo-ethylene glycol (12-mer). In this study, three building block peptide amphiphiles with different EG lengths (6-mer, 12-mer, and 24-mer, which are termed EG 6 , EG 12 , and EG 24 , respectively) were prepared for assessing the effect of EG length on cellular uptake of NFs, cytotoxicity, and immune stimulatory activities.
Initially, we investigated the effect of EG length on the self-assembly behavior of these peptides by estimating the CFC of each peptide. EG n peptides were incubated in the presence of the thioflavin T (ThT) dye at 300 µM in phosphate buffered saline (PBS) containing 5% dimethylsulfoxide (DMSO) at 37 • C for 24 h, and a time course of change in ThT fluorescence was measured. The ThT assay is often used to monitor the growth of amyloid-like nanofibers from their component peptides or proteins. A remarkable increase in ThT fluorescence was observed, indicating the formation of NFs by each peptide (Figure 2 and Figure S1). After a certain period of time, the intensity of the ThT fluorescence reached a plateau value. The systems were in a dynamic equilibrium between fibrils and the peptide monomer when the ThT fluorescence intensity reached a plateau value. The concentration of the free peptide monomer at equilibrium corresponds to the CFC [38,39]. The peptide concentration in the supernatant following ultracentrifugation of the EG n peptide solutions incubated at 300 µM and 37 • C for 24 h provided estimates of the CFC values, which were 96.0 ± 3.6 µM for EG 6 , 72.4 ± 3.7 µM for EG 12 , and 83.8 ± 1.0 µM for EG 24 . Below the CFC, EGn peptides could either form spherical micelle-like structures or exist as isolated molecules in water because of their amphiphilic structures [40]. The CAC was determined for EGn peptides in PBS (pH 7.4) using the pyrene 1:3 method to gain information on the association state of EGn peptides at relatively low concentrations. The CAC was estimated to be 16.6 μM for EG6, 21.7 μM for EG12, and 29.9 μM for EG24 ( Figure S2). The CAC value increased as the EG length increased. Because CAC values were smaller than the CFC values, EGn peptides self-assembled into spherical micelle-like structures over the concentration range between the CAC and the CFC, and existed as monomers at concentrations below the CAC.

Structural Characterization of EGn Nanofibers
NFs consisting of EGn peptides were prepared by incubation of the peptide solution in PBS containing 5% DMSO at 60 °C for 24 h. The incubation was carried out at a higher temperature in this experiment when compared with that of the ThT assay to accelerate NF formation. The resulting NFs were characterized by TEM, CD, ς-potential measurements, and the 8-anilino-1-naphtalene sulfonic acid (ANS) assay.
TEM images revealed that all peptides successfully formed NFs with a homogenous distinct width of ca. 6-8 nm and lengths of several micrometers (Figure 3). The values of the ς-potentials were −33.2 ± 10.7 mV for EG6 NFs, −33.2 ± 10.7 mV for EG12 NFs, and −32.4 ± 9.6 mV for EG24 NFs, showing that the surface of the NFs were negatively charged. The ANS assay was performed to obtain information on the surface hydrophobicity of the NFs [41,42]. The peaks shifted toward shorter wavelengths, and the intensities of the signals increased significantly for all NFs ( Figure S3). These observations indicate that there are hydrophobic domains on the surface of the peptide NFs. CD was used to examine the secondary structures adopted by peptides in the NFs. The characteristic negative Below the CFC, EG n peptides could either form spherical micelle-like structures or exist as isolated molecules in water because of their amphiphilic structures [40]. The CAC was determined for EG n peptides in PBS (pH 7.4) using the pyrene 1:3 method to gain information on the association state of EG n peptides at relatively low concentrations. The CAC was estimated to be 16.6 µM for EG 6 , 21.7 µM for EG 12 , and 29.9 µM for EG 24 ( Figure S2). The CAC value increased as the EG length increased. Because CAC values were smaller than the CFC values, EG n peptides self-assembled into spherical micelle-like structures over the concentration range between the CAC and the CFC, and existed as monomers at concentrations below the CAC.

Structural Characterization of EG n Nanofibers
NFs consisting of EG n peptides were prepared by incubation of the peptide solution in PBS containing 5% DMSO at 60 • C for 24 h. The incubation was carried out at a higher temperature in this experiment when compared with that of the ThT assay to accelerate NF formation. The resulting NFs were characterized by TEM, CD, ς-potential measurements, and the 8-anilino-1-naphtalene sulfonic acid (ANS) assay.
TEM images revealed that all peptides successfully formed NFs with a homogenous distinct width of ca. 6-8 nm and lengths of several micrometers (Figure 3). The values of the ς-potentials were −33.2 ± 10.7 mV for EG 6 NFs, −33.2 ± 10.7 mV for EG 12 NFs, and −32.4 ± 9.6 mV for EG 24 NFs, showing that the surface of the NFs were negatively charged. The ANS assay was performed to obtain information on the surface hydrophobicity of the NFs [41,42]. The peaks shifted toward shorter wavelengths, and the intensities of the signals increased significantly for all NFs ( Figure S3). These observations indicate that there are hydrophobic domains on the surface of the peptide NFs. CD was used to examine the secondary structures adopted by peptides in the NFs. The characteristic negative Cotton peak at 217 nm was observed for the three EG n NFs, showing that the peptides adopt a β-sheet conformation ( Figure 4) [43]. These results indicate that the three NFs possess a similar secondary structure regardless of EG length. Cotton peak at 217 nm was observed for the three EGn NFs, showing that the peptides adopt a βsheet conformation ( Figure 4) [43]. These results indicate that the three NFs possess a similar secondary structure regardless of EG length.

Cellular Uptake, Cytotoxicity, and Maturation of DCs
We investigated the effect of the EG length of nanofibers on their cell association, cytotoxicity, and DC stimulatory activities. The peptide NFs obtained by incubation at a concentration of 1.5 mM and 60 °C for 24 h were used. The length of the NFs was controlled to be 230-260 nm by an extrusion Cotton peak at 217 nm was observed for the three EGn NFs, showing that the peptides adopt a βsheet conformation ( Figure 4) [43]. These results indicate that the three NFs possess a similar secondary structure regardless of EG length.

Cellular Uptake, Cytotoxicity, and Maturation of DCs
We investigated the effect of the EG length of nanofibers on their cell association, cytotoxicity, and DC stimulatory activities. The peptide NFs obtained by incubation at a concentration of 1.5 mM and 60 °C for 24 h were used. The length of the NFs was controlled to be 230-260 nm by an extrusion

Cellular Uptake, Cytotoxicity, and Maturation of DCs
We investigated the effect of the EG length of nanofibers on their cell association, cytotoxicity, and DC stimulatory activities. The peptide NFs obtained by incubation at a concentration of 1.5 mM and 60 • C for 24 h were used. The length of the NFs was controlled to be 230-260 nm by an extrusion procedure using a membrane filter with a diameter of 450 nm ( Figures S4 and S5). Information about the dispersion state of NFs in water was determined by dynamic light scattering (DLS) measurements of EG 6 NFs, EG 12 NFs, and EG 24 NFs. The DLS histograms for EG 12 and EG 24 NFs exhibited a unimodal peak, with average diameters of 203.7 ± 119.7 nm for EG 12 NFs, and 116.6 ± 68.2 nm for EG 24 NFs. These values are inconsistent with the size estimates from TEM images and this is possibly because of their non-spherical morphology. Nonetheless, these unimodal histograms clearly indicate that EG 12 and EG 24 NFs exist as isolated NFs without aggregation in aqueous media ( Figure S6). In contrast, the DLS histogram for EG 6 NFs exhibited two peaks with sizes of 135.4 ± 14.5 nm and 3498.3 ± 389.0 nm, indicating that EG 6 NFs formed large aggregates. The secondary aggregation of the NFs may be caused by association of surface-exposed hydrophobic domains on the NFs. Longer EG chains effectively prevent these interactions, yielding highly dispersed, stable EG 12 and EG 24 NFs. Conversely, the 6-mer EG is not sufficiently long to prevent secondary aggregation of the peptide NFs, resulting in the observed large aggregates. To compare the behaviors of NFs, the cellular uptake, cytotoxicity, and DC stimulatory activities of non-heat-treated EG n peptides (non-fiber) were also evaluated. The samples for these experiments were prepared by direct dissolution of the peptides in medium at a given concentration to avoid self-assembly into NFs.

Cellular Association
We evaluated the effect of EG length on the cellular association of EG n NFs.
The fluorescence-labeled EG n NFs were incubated with JAWS II cells for 1 h at 37 • C. JAWS II cells are an immortalized immature DC line that was established from bone marrow cultures of C57BL/6 p53-knockout mice [44,45]. The amount of NFs associated with the cells was evaluated by flow cytometry (FCM). For comparison, cellular association of non-heat-treated EG n peptides was also performed.
As shown in Figure 5a, the intensity of fluorescence signals from cells incubated with EG n peptides (non-fiber) increased as the concentration of the peptides increased. This trend was common to all peptides examined. Comparison of the fluorescence intensity of the three EG n peptides at the same concentration revealed that they were very similar, indicating that EG length had no effect on cellular association of the peptides. In contrast, cellular association of EG n NFs was influenced noticeably by EG length (Figure 5b). The amount of associated EG n NFs was larger as the EG length decreased. This trend was more apparent as the concentration of the peptide increased. This result indicates that longer EG chains may prevent interactions between cells and NFs. procedure using a membrane filter with a diameter of 450 nm ( Figures S4 and S5). Information about the dispersion state of NFs in water was determined by dynamic light scattering (DLS) measurements of EG6 NFs, EG12 NFs, and EG24 NFs. The DLS histograms for EG12 and EG24 NFs exhibited a unimodal peak, with average diameters of 203.7 ± 119.7 nm for EG12 NFs, and 116.6 ± 68.2 nm for EG24 NFs. These values are inconsistent with the size estimates from TEM images and this is possibly because of their non-spherical morphology. Nonetheless, these unimodal histograms clearly indicate that EG12 and EG24 NFs exist as isolated NFs without aggregation in aqueous media ( Figure  S6). In contrast, the DLS histogram for EG6 NFs exhibited two peaks with sizes of 135.4 ± 14.5 nm and 3498.3 ± 389.0 nm, indicating that EG6 NFs formed large aggregates. The secondary aggregation of the NFs may be caused by association of surface-exposed hydrophobic domains on the NFs. Longer EG chains effectively prevent these interactions, yielding highly dispersed, stable EG12 and EG24 NFs. Conversely, the 6-mer EG is not sufficiently long to prevent secondary aggregation of the peptide NFs, resulting in the observed large aggregates. To compare the behaviors of NFs, the cellular uptake, cytotoxicity, and DC stimulatory activities of non-heat-treated EGn peptides (non-fiber) were also evaluated. The samples for these experiments were prepared by direct dissolution of the peptides in medium at a given concentration to avoid self-assembly into NFs.

Cellular Association
We evaluated the effect of EG length on the cellular association of EGn NFs. The fluorescencelabeled EGn NFs were incubated with JAWS II cells for 1 h at 37 °C. JAWS II cells are an immortalized immature DC line that was established from bone marrow cultures of C57BL/6 p53-knockout mice [44,45]. The amount of NFs associated with the cells was evaluated by flow cytometry (FCM). For comparison, cellular association of non-heat-treated EGn peptides was also performed.
As shown in Figure 5a, the intensity of fluorescence signals from cells incubated with EGn peptides (non-fiber) increased as the concentration of the peptides increased. This trend was common to all peptides examined. Comparison of the fluorescence intensity of the three EGn peptides at the same concentration revealed that they were very similar, indicating that EG length had no effect on cellular association of the peptides. In contrast, cellular association of EGn NFs was influenced noticeably by EG length (Figure 5b). The amount of associated EGn NFs was larger as the EG length decreased. This trend was more apparent as the concentration of the peptide increased. This result indicates that longer EG chains may prevent interactions between cells and NFs.  We performed confocal laser scanning microscopic (CLSM) observations of JAWS II cells incubated with various EG n NFs to evaluate the association of NFs in further detail ( Figure 6). The CLSM images of cells incubated with EG 12 NFs clearly show that the NFs were internalized into cells. The fluorescence signals were observed as dot-like images, indicating that EG 12 NFs were internalized via endocytosis. In contrast, EG 24 NFs showed no fluorescence signal, indicating poor cellular uptake of EG 24 NFs. The confocal images of EG 6 NFs-treated cells showed large intensive fluorescence signals on the surface of cells, indicating that some aggregation of NFs were apparently adsorbed onto the surface of cells. This observation indicates that a large proportion of the fluorescence signal from EG 6 NFs-treated cells observed in FCM measurements was derived from NFs that had adhered to the surface of cells. The FCM and CLSM results comprehensively showed the efficient uptake of EG 12 NFs by cells. We performed confocal laser scanning microscopic (CLSM) observations of JAWS II cells incubated with various EGn NFs to evaluate the association of NFs in further detail ( Figure 6). The CLSM images of cells incubated with EG12 NFs clearly show that the NFs were internalized into cells. The fluorescence signals were observed as dot-like images, indicating that EG12 NFs were internalized via endocytosis. In contrast, EG24 NFs showed no fluorescence signal, indicating poor cellular uptake of EG24 NFs. The confocal images of EG6 NFs-treated cells showed large intensive fluorescence signals on the surface of cells, indicating that some aggregation of NFs were apparently adsorbed onto the surface of cells. This observation indicates that a large proportion of the fluorescence signal from EG6 NFs-treated cells observed in FCM measurements was derived from NFs that had adhered to the surface of cells. The FCM and CLSM results comprehensively showed the efficient uptake of EG12 NFs by cells.

Cytotoxicity
The cytotoxicity of EGn NFs and EGn peptides (non-fiber) were evaluated. The JAWS II cells were incubated with the EGn NFs or EGn peptides at 37 °C for 24 h at different concentrations (10-50 μM), and cell activity was evaluated. Interestingly, cell activity after incubation with all EGn peptides increased as the peptide concentration increased (Figure 7a). Relative cell activity was essentially 100%, independent of the EG length when peptides were co-incubated with cells at higher concentrations (40-50 μM). Co-incubation of peptides with cells at lower concentrations (10-20 μM) reduced cell activity to 70-90%. In addition, peptides with shorter EG chains showed relatively higher toxicity.
In contrast, cell activity was observed to decrease after incubation with each NF, and this reduction in cell activity was concentration-dependent ( Figure 7b). In particular, the activity of EG6 Figure 6. CLSM images of JAWS II cells treated with various EG n NFs labeled with fluorescein. JAWS II cells were incubated with EG n NFs at 37 • C for 2 h in serum-free medium. After incubation, the cells were treated with Lyso Tracker Red and Hoechst for staining intracellular acidic compartments and nuclei, respectively. Scale bars represent 20 µM.

Cytotoxicity
The cytotoxicity of EG n NFs and EG n peptides (non-fiber) were evaluated. The JAWS II cells were incubated with the EG n NFs or EG n peptides at 37 • C for 24 h at different concentrations (10-50 µM), and cell activity was evaluated. Interestingly, cell activity after incubation with all EG n peptides increased as the peptide concentration increased (Figure 7a). Relative cell activity was essentially 100%, independent of the EG length when peptides were co-incubated with cells at higher concentrations (40-50 µM). Co-incubation of peptides with cells at lower concentrations (10-20 µM) reduced cell activity to 70-90%. In addition, peptides with shorter EG chains showed relatively higher toxicity.
NFs-treated cells was reduced to 40% at a concentration of 50 μM. For EG12 and EG24 NFs, cell activity was 70% for EG12 NFs and 80% for EG24 NFs even at a NFs concentration of 50 μM. Thus, NFs with shorter EG chains exhibited higher toxicity. It is well known that DC maturation is accompanied by enhanced expression of co-stimulatory molecules (CD40, CD80, and CD86) and by an increase in the secretion of immune-stimulatory cytokines (IL-6, IL-10, IL-12, and TNF-α) [46]. We measured the amount of expressed co-stimulatory molecules (CD86) on JAWS II cells cultured in the presence of EGn NFs or EGn peptides for 24 h to determine the effect of EG length on DC maturation. As a positive control, the expression of CD86 on lipopolysaccharides (LPS)-stimulated JAWS II was also measured. The expression of CD86 on JAWS II cells after incubation with EGn peptides is shown in Figure 8. Even at a peptide concentration of 50 μM, the expression level of CD86 was almost the same as that of untreated JAWS II cells. These results indicate that EGn peptides do not stimulate DC. In contrast, co-incubation of EGn NFs with JAWS II cells significantly enhanced the expression of CD86, and this enhancement was dependent on the concentration of the NFs. In particular, when EG12 NFs and EG6 NFs were co-incubated with JAWS II cells, the amount of expressed CD86 was comparable to or larger than that on LPS-stimulated DC.
We also evaluated secretion of immune-stimulatory cytokines. Using enzyme-linked immunosorbent assay (ELISA) methods, we measured the amount of TNF-α and IL-6 contained in the supernatant after 24 h culturing of JAWS II in the presence of EGn NFs or EGn peptides (Figure 9). Co-incubation with EGn peptides did not alter the secretion levels of TNF-α and IL-6. In contrast, interestingly, the secretion of TNF-α and IL-6 was drastically enhanced by co-incubation with EG6 In contrast, cell activity was observed to decrease after incubation with each NF, and this reduction in cell activity was concentration-dependent ( Figure 7b). In particular, the activity of EG 6 NFs-treated cells was reduced to 40% at a concentration of 50 µM. For EG 12 and EG 24 NFs, cell activity was 70% for EG 12 NFs and 80% for EG 24 NFs even at a NFs concentration of 50 µM. Thus, NFs with shorter EG chains exhibited higher toxicity.

DC Stimulatory Activity.
It is well known that DC maturation is accompanied by enhanced expression of co-stimulatory molecules (CD40, CD80, and CD86) and by an increase in the secretion of immune-stimulatory cytokines (IL-6, IL-10, IL-12, and TNF-α) [46]. We measured the amount of expressed co-stimulatory molecules (CD86) on JAWS II cells cultured in the presence of EG n NFs or EG n peptides for 24 h to determine the effect of EG length on DC maturation. As a positive control, the expression of CD86 on lipopolysaccharides (LPS)-stimulated JAWS II was also measured. The expression of CD86 on JAWS II cells after incubation with EG n peptides is shown in Figure 8. Even at a peptide concentration of 50 µM, the expression level of CD86 was almost the same as that of untreated JAWS II cells. These results indicate that EG n peptides do not stimulate DC. In contrast, co-incubation of EG n NFs with JAWS II cells significantly enhanced the expression of CD86, and this enhancement was dependent on the concentration of the NFs. In particular, when EG 12 NFs and EG 6 NFs were co-incubated with JAWS II cells, the amount of expressed CD86 was comparable to or larger than that on LPS-stimulated DC.
We also evaluated secretion of immune-stimulatory cytokines.
Using enzyme-linked immunosorbent assay (ELISA) methods, we measured the amount of TNF-α and IL-6 contained in the supernatant after 24 h culturing of JAWS II in the presence of EG n NFs or EG n peptides (Figure 9). Co-incubation with EG n peptides did not alter the secretion levels of TNF-α and IL-6. In contrast, interestingly, the secretion of TNF-α and IL-6 was drastically enhanced by co-incubation with EG 6 NFs and EG 12 NFs, but not by EG 24 NFs. These results indicate that NFs with relatively short EG chains have an immune-stimulatory effect as adjuvants for DC maturation. NFs and EG12 NFs, but not by EG24 NFs. These results indicate that NFs with relatively short EG chains have an immune-stimulatory effect as adjuvants for DC maturation.   NFs and EG12 NFs, but not by EG24 NFs. These results indicate that NFs with relatively short EG chains have an immune-stimulatory effect as adjuvants for DC maturation.

Discussion
In this study, we have investigated cellular uptake, cytotoxicity, and DC stimulatory activity of antigen-loaded peptide NFs with different EG lengths and their component peptides. Three building block peptide amphiphiles with different EG lengths (6-mer, 12-mer and 24-mer) were prepared. ThT assay, TEM observation, and CD measurement revealed that all type of peptide amphiphiles are successfully formed β-sheet rich nanofibers with distinct widths (Figures 2-4). The association state of EG n peptides was dependent on sample concentration. EG n peptides self-assembled into NFs above the CFC, formed spherical micelles at concentrations between the CFC and CAC, and existed as monomers in solution below the CAC. Based on these findings, we discuss separately the effect of EG length on cellular uptake, cytotoxicity, and immune stimulation for three peptide states: NFs, micelles, and monomers.

Effect of EG Length of Nanofibers on Their Cellular Uptake, Cytotoxicity, and Immune Stimulation Ability
The EG length of NFs significantly affected their cellular uptake, cytotoxicity, and DC stimulatory activity. Here, we discuss the effect of EG length on these properties of NFs using structural models derived from SAXS, WAXD, CD, and FT-IR data of a previous study [36]. FT-IR, CD, and WAXD results indicate that EG 12 NFs contain β-sheet structures. In addition, synchrotron X-ray scattering profiles of EG 12 NFs revealed that the morphology of the NFs is rectangular, and they do not form cylinder structures like filament micelles, presumably because of the laminated structure of β-sheets. In general, amyloid-like nanofibers have a common characteristic cross-β-sheet structure, where tightly packed β-sheets orientate themselves perpendicularly to the fiber elongation axis [47]. By combining these findings, we propose a model of EG n NFs ( Figure S7). β-sheet structures consisting mainly of hydrophobic amino acids form the framework of NFs with EG chains facing outwards to provide water-dispersibility. The surfaces of NFs possess hydrophobic and hydrophilic domains that consist of EG chains based on this model. The ANS assay results support the notion that there are hydrophobic domains on the surface of NFs.

Cellular Association and Internalization of NFs
The amount of NFs associated with cells increased in the order of EG 24 NFs, EG 12 NFs, and EG 6 NFs (Figure 5b). Surface hydrophobicity of nanomaterials has been well documented to affect cellular association and uptake by phagocytic cells [24,[30][31][32]. Surface hydrophobicity of nanomaterials facilitates interactions between nanomaterial surfaces and cellular membranes. This may lead to higher cell association of nanomaterials and occasionally increase the chance of recognition by particular receptors involved in cellular uptake. Our results show that cellular association of NFs decreased as the EG chain length increased (Figure 5b), although these NFs commonly possess hydrophobic domains on their surface, as evidenced by the ANS fluorescence assay. These results suggest that longer EG chains inhibit hydrophobic interactions between the NF surface and cell membranes, which can be explained using the model structures presented in Figure 9. The NF skeleton region, consisting mainly of hydrophobic amino acids, may facilitate the interaction with the cell membrane and the EG chain located on the lateral face of the NF may inhibit this interaction.
Results from CLSM observation revealed that EG 12 NFs were more efficiently internalized by JAWS cells than EG 6 NFs and EG 24 NFs (Figure 6). Since the surface of EG 12 NFs is negatively charged, the mechanism for internalization of EG 12 NFs would be mainly via phagocytosis by scavenger receptor, which recognizes anion species, although further studies using some inhibitor for phagocytosis are required. Thus, the internalization behavior by non-phagocytic cells would be different from that by JAWS II cells. The internalization of EG 6 NFs was low, whereas their association propensity to cells was high. Because the size of nanomaterials can affect cell internalization [48][49][50][51], the dispersion state of NFs in aqueous media should be considered in addition to interactions between NFs and the cell surface. The results from DLS indicate that EG 12 and EG 24 NFs exist as isolated NFs without aggregation in aqueous media, whereas EG 6 NFs form large aggregates. This observation is consistent with CLSM images showing large aggregates adsorbed onto the cell surfaces. Thus, it is likely that the low efficiency of cellular internalization of EG 6 NFs can be attributed to their apparent size in water. The aggregation of EG 6 NFs is too large for cell uptake. This interpretation is consistent with a previous study that showed that cellular uptake of microparticles with a diameter of a few micrometers or more by phagocytic cells is slow and inefficient [50,51]. Thus, for development of NFs that are efficiently taken up by cells, it is important to design a EG chain length that allows modest interactions with cell membranes while ensuring water-dispersibility.

Cytotoxicity of NFs
Generally, nanomaterials with cationic or hydrophobic surfaces can induce significantly higher toxicities when compared with hydrophilic or anionic nanomaterials [29]. A mechanism of cytotoxicity is cell membrane perturbation, including structural alternation, pore formation, and phase transitions, which cause nonspecific entrance of extracellular components to the cytosol. An increase in hydrophobic interactions between the surface of nanomaterials and cell membranes could perturb the membrane. In the present study, the cytotoxicity of EG n NFs was found to increase in the order of EG 24 , EG 12 , and EG 6 ( Figure 7). These results indicate that longer EG chains inhibit the interaction between NFs and cell membranes, which leads to lower cytotoxicity of NFs with long EG chains. It is also possible that the cytotoxicity of NFs may be related to biological stress, e.g., induction of reactive oxygen species (ROS). The detailed mechanism of cytotoxicity by NFs is the subject of ongoing research.

DC Stimulation Ability of NFs
DC activation ability of EG 6 NFs and EG 12 NFs was much higher than that of EG 24 NFs. Matzinger and colleagues have proposed that hydrophobic portions in various biomolecules may be involved in the activation of the immune system [22]. Hydrophobic components in molecules are usually masked from the external environment by hydrophilic components. However, when protein denaturation or cell disruption occur, these hydrophobic components become exposed and interact with particular surface receptors of immune cells, which activates the immune system. In agreement with the notion proposed by Matzinger, recently, the relationship between the surface hydrophobicity of nanomaterials and their immune stimulatory activities has been reported [23][24][25][26][27][28]. For example, Moyano and colleagues reported that the surface hydrophobicity of ligand-modified gold-nanoparticles was correlated with expression of pro-inflammatory cytokine genes in splenocytes from mice in vitro [23]. Shima and colleagues also reported that the activation ability of nanoparticles was significantly affected by the hydrophobicity of polymers constituting the nanoparticles [24]. In the present study, EG 6 NFs and EG 12 NFs stimulated DC maturation more effectively than EG 24 NFs, as evidenced by the quantitative evaluation of expressed co-stimulatory molecules ( Figure 8) and secreted immune-stimulatory cytokines, IL-6 and TNF-α (Figure 9). Based on these results, it is reasonable to consider that the hydrophobic part of NFs plays an important role in DC activation. Longer EG chains seem to inhibit the recognition of hydrophobic surfaces of NFs by DC surface receptors in a similar manner to that described above. However the mechanisms responsible for DC maturation by EG n NFs remain unclear and further studies are required. In addition, because IL-6 signaling cannot only promote anti-tumor-adaptive immunity, but also drive malignancy [52], the role of IL-6 in this NFs-based vaccine system should be examined further in vivo.

Cellular Uptake, Toxicity, and DC Stimulatory Ability of Micelles
In the concentration range where EG n peptides form micelle-like structures, their cellular association, cytotoxicity, and stimulation ability were not dependent on EG length (Figure 5a, Figure 7a, and Figure 8). These results suggest that the surface components of the micelle-like structures would be almost the same. Poly(ethylene glycol) (PEG) interactions with biological components, including cellular membranes and proteins, are weak because of their nonionic hydrophilicity and high mobility [53]. Thus, a low-level of interaction between the surface of the micelle-like structures and cell membrane components, including receptors involved in cellular uptake and DC maturation, led to lower uptake by DC, lower cytotoxicity, and no DC activation in comparison with NFs.
3.3. Cellular Uptake, Toxicity, and DC Stimulatory Ability of Monomeric Molecules EG n peptides exist as monomeric molecules in aqueous media at concentrations below ca. 15-30 µM. The cellular uptake of monomeric peptides was not dependent on EG length (Figure 5a) and this may be attributed to size. Monomeric peptides are too small for efficient uptake by DC regardless to EG length. In addition, the monomeric peptides exhibited some cytotoxicity but no DC activation ability, which is in sharp contrast to the results with NFs (Figure 7a, Figure 8, and Figure 9). These results suggest that monomeric peptides interact with cell membranes, possibly through an N-terminal hydrophobic region, but are not well recognized by receptors involved in DC activation. Recognition by receptors would be dependent on the size of the hydrophobic portion.

Design of NF-based Vaccines
Recently, various types of NFs for immunotherapy have been reported [54,55]. In particular, NFs formed from antigenic peptides conjugating to β-sheet-forming peptides have been recognized as very promising candidates for next-generation nanoparticle-based vaccines. In the present study, we demonstrated that the hydrophilic-hydrophobic balance of peptide NFs affects their cellular uptake, cytotoxicity, and DC activation ability. NFs consisting of EG with a moderate length (12-mer) showed the most balanced character: Highly efficient cell entry, low cytotoxicity, and high DC activation ability, indicating that the NFs have significant potential as NF-based vaccines, which can be used without additional adjuvants. In general, the relationship between toxicity and DC stimulation ability is a trade-off. It is important to improve the stimulation ability, while simultaneously reducing the cytotoxicity of the NFs. Our results demonstrate that such balance can be simply tuned by the length of the EG. This feature is important for designing safe NF-based vaccines with high immune stimulatory ability. In contrast to NF uptake, the uptake of micelles and monomeric peptides by DC cells inefficient and showed no DC stimulation ability independent of EG length. This result indicates that the assembly style of building block peptide molecules influences the properties of the nanoassembly formed from these building blocks. Finally, to develop NFs with strong immunity-inducing ability, it is necessary to precisely adjust the EG length and introduce intracellular environment-responsive links for efficient release of antigens in cells.
Although we focused on the effect of hydrophilic and hydrophobic balance of nanofibers on their interaction with cells, surface charge of nanofibers is also an important factor in determining the interaction. In general, positively charged nanomaterials are more effectively internalized to cells than negatively charged ones, but they are more toxic. Thus, for design of nanofiber vaccines, it is necessary to address the role of their surface charge. In addition to surface charge, the length of NFs is also an important factor determining their property as a nano-vaccine. In a previous study, we investigated the effect of nanofiber length on their cellular uptake using various NFs with different lengths (40 nm, 120 nm, 280 nm, 800 nm). The study demonstrated that nanofibers with a length of 280 nm were most effectively uptaken by phagocytic cells compared to the others (unpublished data). Based on this finding, we used NFs with a length of 230-260 nm for the cell experiments in the present study. However, other properties-e.g., cytotoxicity and the ability to stimulate immune cells etc.-could exhibit different size-dependencies. Therefore, the optimization of nanofiber length is also required for developing effective nanofiber vaccine.
The important attributes of a vaccine, which are antigen processing, antigen presentation, T-cell stimulation, and successful activation of adaptive immune response against target antigen, should also be evaluated. However, because the NFs used in this study comprised the minimum required block (β-sheet forming peptide, antigenic peptide, oligo(ethylene glycol)) and the antigen could not be released in the cells, the effective antigen presentation via MHC class I pathway and subsequent induction of immunity are not expected. Therefore, we are addressing the development of intracellular environment-responsive NFs for efficient release of antigens in cells and the characterization of their function to induce immunity in vivo.
Peptide synthesis: Coupling reactions were performed using a standard Fmoc protocol. The coupling cycle included 3 repeats of Fmoc deprotection (20% piperidine in DMF, 5 min), a wash in DMF, two repeats of amino acid coupling: L-Fmoc amino acids (4 eq.), HBTU (3.6 eq.), HOBt (4 eq.), and DIPEA (8 eq.) for 30 min, and a final DMF wash. After all coupling reactions, the obtained peptides were cleaved from the resin using a solution of H 2 O/TFA/triisopropylsilane (100:5:2 volume ratio) containing 500 mM phenol for 2 h. The resulting peptides were precipitated in ice cold diethyl ether, filtered, centrifuged, and washed with diethyl ether. The crude peptides were purified by reversed-phase high-performance liquid chromatography (RP-HPLC, SPD-10A and LC-10A, Shimadzu Scientific Instruments, Kyoto, Japan). Other building blocks with different EG lengths were synthesized by a similar procedure.

Preparation of Antigen-Loaded Peptide NFs
EG n peptide was dissolved in HFIP and dried with nitrogen flow to allow film formation. The obtained film was re-dissolved at a concentration of 30 mM in DMSO. The resulting solution (15 µL) was added to PBS (285 µL) to a final concentration of 1.5 mM and incubated at 60 • C for 24 h. Following incubation, the resulting peptide nanofiber dispersion was dialyzed against PBS for 24 h using dialysis membrane (MWCO 8,000, GE Healthcare, Chicago, IL, USA) to remove DMSO and free peptides. For cell-based experiments (cytotoxicity, DC maturation), the length of NFs was controlled by filtration using a syringe filter with a pore size of 0.45 µM (GE Healthcare).

Preparation of Fluorescence-Labeled Antigen-Loaded Peptide NFs
The HFIP-treated mixture of EG n and EG n -FAM peptides was dissolved in DMSO; then, 15 µL of the solution was added to 285 µL PBS to give final concentrations of 1.425 mM for EG n and 0.071 mM for EG n -FAM. The solution was incubated at 60 • C for 24 h, and then the resulting peptide nanofiber dispersion was dialyzed against PBS for 24 h using dialysis membrane (MWCO 8,000, GE Healthcare) to remove DMSO and free peptides. The length of NFs was controlled by filtration using a syringe filter with a pore size of 0.45 µM (GE Healthcare).

Determination of Critical Aggregation Concentration
CAC was determined using the pyrene 1:3 method [56]. First, a saturated solution of pyrene was prepared by mixing an excess of pyrene with PBS, and using the supernatant to dissolve EG n peptides at a concentration of 150 mM (the stock solution). A concentration range of EG n peptides from 2.5 µM to 100 µM was then prepared using serial dilutions of the stock solution with the saturated solution of pyrene. The final concentration of pyrene was equal in each solution. The fluorescence emission of pyrene was monitored using a fluorescence spectrometer (RF5300 PC, Shimadzu Scientific Instruments, Kyoto, Japan) with an excitation wavelength of 335 nm at 37 • C. The ratio of the emission intensities at 376 nm and 392 nm were then plotted as a function of the EG n peptide concentration (log scale). The CAC was determined from an abrupt change in the slope of the plot using the least-squares fitting technique.

ThT Assay
PBS containing 10 µM Thioflavin T (ThT) was dispensed into a 96-well plate. EG n peptides solution (6 mM) was prepared and added to the 96-well plate, giving a final concentration of 300 µM.

Measurement of Surface Hydrophobicity
The surface hydrophobicity of NFs in the solution was determined using an ANS fluorescent probe as previously reported [41,42]. A concentration range of EG n NFs in PBS from 9.4 µM to 200 µM was prepared. The nanofiber dispersion was mixed with the equivalent volume of PBS containing 20 µM ANS. The intensities of ANS fluorescence ranging from 400 nm to 600 nm (excitation; 370 nm) were monitored using a fluorescence spectrometer (RF5300 PC) at 37 • C.

Cell Culture
JAWS II, a DC line derived from mouse bone marrow, was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). The cells were grown in EMEM supplemented with 20% FBS, 5 ng/mL murine GM-CSF, and antibiotics at 37 • C, 5% CO 2 .

Evaluation of Cellular Association of Peptide NFs
JAWS II cells were seeded into 12-well plates (2.5 × 10 5 per well) and cultured for 12 h at 37 • C in a humidified atmosphere (5% CO 2 ). After 12 h, the cells were washed with PBS and serum-free culture medium. The fluorescence-labeled peptide NF dispersion was added gently to the cells followed by incubation for 2 h at 37 • C. Following incubation, the cells were washed with PBS and 0.2% trypan blue aqueous solution, which was used to quench the flourescence from surface-adsorbed NFs [57]. The cells were then detached using trypsin and subsequently analyzed by FCM (Guava EasyCyte Plus, Millipore, Burlington, MA, USA). As a comparison to peptide NFs, the cellular uptake of the building block peptides without heat treatment was investigated under the same conditions. 4.2.9. CLSM Observation of NF-Treated Cells JAWS II cells (1.5 × 10 5 ) were cultured for 12 h in a 35 mm glass-bottom dish and subsequently washed with PBS and serum-free culture medium. Fluorescein-labeled peptide NFs were gently added to the cells, followed by incubation for 2 h at 37 • C with 5% CO 2 . After incubation, the cells were washed with PBS, and then incubated for 5 min with a solution containing LysoTracker Red DND-99 (50 nm) and Hoechst 33342, trihydrochloride, trihydrate (3.24 µM). LysoTracker Red DND-99 and Hoechst 33342 were used to stain the intracellular acidic compartments and nuclei, respectively. After staining, the cells were washed twice with PBS, then observed by CLSM using an FV10i microscope (Olympus, Tokyo, Japan).

Quantitative Expression Analysis of Co-Stimulatory Molecules and Cytokines from NF-Treated Cells
The expression of co-stimulatory molecules and cytokines was evaluated by specific immunostaining, as well as by ELISA. For immunostaining, JAWS II cells (2 × 10 5 ) were cultured for 12 h in a 24-well plate followed by washing with PBS containing 3% FBS and 0.05% NaN 3 , and then with serum-free culture medium. The DCs were pulsed with peptide NFs for 24 h, and then immunostained with a mouse monoclonal antibody for CD86 (a maturation marker), and subsequently analyzed by FCM to estimate their CD86 expression level. For quantitative analysis of TNF-α and IL-6 expression, the supernatants following the 24 h co-incubation of DCs with peptide NFs were collected and analyzed using an ELISA kit. The maturation of JAWS II cells cultured in medium with and without LPS (1 µg/mL) was evaluated as the positive and negative control, respectively. In addition, to compare the DC-activation ability between the NFs and heat-untreated building block peptides, JAWS II cells cultured with the peptides were also evaluated.

Evaluation of Cytotoxicity of Peptide NFs
The cytotoxicity of peptide NFs was evaluated using a Cell Counting Kit-8 (Dojindo Molecular Technologies, Kumamoto, Japan) according to the manufacturer's instructions. Briefly, JAWS II cells were seeded into 96-well plates (1.0 × 10 5 per well) and cultured for 12 h at 37 • C in a humidified atmosphere (5% CO 2 ). After 12 h, the cells were washed with PBS and serum-free culture medium. The nanofiber dispersion was gently added to the cells followed by incubation for 24 h. The cells were washed with PBS three times and the medium was replaced with a solution containing 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt (WST-8), and 1-methoxy-5-methylphenazinium methylsulfate at a 10-fold dilution. After a 2 h incubation, the absorbance was measured at 420 nm using a plate reader (Multiskan JX, Thermo Fisher Scientific, Waltham, MA, USA). The relative cellular activity was calculated using the following equation: % relative cellular activity = A 420 nm (NFs − treated cells) − A 420 nm (blank) A 420 nm (untreated cells) − A 420 nm (blank) × 100 (1) where A 420 nm is the absorbance at 420 nm, A 420 nm (untreated cells) is the absorbance at 420 nm after incubation in the absence of peptide NFs, and A 420 nm (blank) is the absorbance of medium containing WST-8 reagent at 420 nm. As a comparison, the cytotoxicity of building block peptides without heat treatment was investigated in a similar manner.

Other Characterizations
TEM measurements were performed using a JEM-1200EX II (JEOL, Tokyo, Japan) with an acceleration voltage of 85 keV. The samples were negatively stained with 0.1% phosphotungstate. p-potentials of NFs were measured using a Micro-Electrophoresis Zeta Potential Analyzer Model 502 (Nihon Rufuto, Tokyo, Japan). DLS analysis was performed using a particle size analyzer (ELSZ-1000, Otsuka Electronics, Osaka, Japan) at 25 • C. The light source was a He-Ne laser (630 nm) set at an 1ngle of 45 • . Experimental data were analyzed using the marquardt provided by the manufacturer. CD spectra were measured using a J-720 spectropolarimeter (Jasco, Tokyo, Japan) at 25 • C. The data were obtained using a 0.1 cm path length cell at a scan speed of 20 nm/min.

Conclusions
This study showed that the hydrophilic-hydrophobic balance of antigen-loaded NFs significantly impacted on their cellular uptake, cytotoxicity, and DC stimulation ability, which differs noticeably from the results observed for micelles formed from the same components of NFs. Building blocks consisting of β-sheet-forming peptides conjugated with antigenic peptides and hydrophilic EG with different lengths (6-mer, 12-mer and 24-mer) were found to successfully form NFs with homogenous widths. The uptake of NFs consisting of EG with a moderate length (12-mer) by DC was effective, and these NFs activated DC without exhibiting significant cytotoxicity. Increasing the EG chain length significantly reduced the interactions with cells. Conversely, decreasing the EG chain length enhanced DC activation ability but increased toxicity and impaired water-dispersibility, resulting in low cellular uptake. Thus, since cell entry, cytotoxicity, and the immune stimulation ability of antigen-loaded NFs can be tuned by the length of the EG moiety, the antigen-loaded NFs have potential as NF-based vaccines that can be used without additional adjuvants. In order to achieve efficient immune response in vivo, the development of intracellular environment-responsive NFs is now in progress. We believe the findings obtained in this study contribute to the understanding of the interaction between the surface of one-dimensional assemblies and cells, and provide useful design guidelines for development of effective NF-based vaccines.