Advantages and Limitations of Integrated Flagellin Adjuvants for HIV-Based Nanoparticle B-Cell Vaccines

The great advantage of virus-like particle (VLP) nano-vaccines is their structural identity to wild-type viruses, ensuring that antigen-specific B-cells encounter viral proteins in their natural conformation. “Wild-type” viral nanoparticles can be further genetically or biochemically functionalized with biomolecules (antigens and adjuvants). Flagellin is a potent inducer of innate immunity and it has demonstrated adjuvant effectiveness due to its affinity for toll-like receptor 5 (TLR5). In contrast to most TLR ligands, flagellin is a protein and can induce an immune response against itself. To avoid side-effects, we incorporated a less inflammatory and less immunogenic form of flagellin as an adjuvant into HIV-based nanoparticle B-cell-targeting vaccines that display either the HIV-1 envelope protein (Env) or a model antigen, hen egg lysozyme (HEL). While flagellin significantly enhanced HEL-specific IgG responses, anti-Env antibody responses were suppressed. We demonstrated that flagellin did not activate B-cells directly in vitro, but might compete for CD4+ T-cell help in vivo. Therefore, we hypothesize that in the context of VLP-based B-cell nano-vaccines, flagellin serves as an antigen itself and may outcompete a less immunogenic antigen with its antibody response. In contrast, in combination with a strong immunogen, the adjuvant activity of flagellin may dominate over its immunogenicity.


Introduction
Virus-like particle (VLP)-based nano-vaccines are a promising tool for HIV-1 vaccine development [1][2][3]. On the one hand, non-infectious VLPs show a higher overall safety profile than life attenuated/inactivated viral vaccines. On the other hand, their structural identity to the prototype viruses provides advantages over synthetically created nano-vaccines in immune system recognition [3,4]. HIV-1-based VLPs can directly target and activate antigen-specific B-cells in vitro [5] and efficiently deliver B-and T-cell antigens into secondary lymphoid organs in vivo [6]. Additionally, VLPs efficiently initiate and modulate B-and T-cell crosstalk both in vitro [7,8] and in vivo [9,10]. Thus, HIV-based VLPs can be considered as efficient B-cell-targeting nano-vaccines. Table 1. HIV-based virus-like particle preparations used in the study.

Isolation and Purification of Splenic Cells, In Vitro Culture Experiments
A single-cell suspension of splenic cells was prepared as described before [5]. Naïve untouched B-cells were isolated from single-cell suspension of splenic cells with the B-Cell Isolation Kit (#130-90-862, Miltenyi Biotec, Bergisch Gladbach, Germany). DCs were enriched by positive selection with anti-CD11c magnetic beads (#130-52-001, Miltenyi Biotec). All isolations were performed according to the manufacturer's instructions. The resulting cells were routinely >98% pure.

Immunization Experiments, Collection of Blood Samples
Bilateral intramuscular (.) VLP injections were performed in the upper leg under isoflurane anesthesia. Mice received VLPs at the final concentration of 400 ng target antigen per mouse (either HIV-Env for Balb/c mice or HEL for C3H mice). Routinely, mice were immunized twice: At day 0 and at day 35. For serological follow-up, mice were immunized three times: At days 0, 35, and 56.
Mice were bled by puncture of the retroorbital sinus with a heparinized 10 µL hematocrit capillary (Hirschmann Laborgerate, Germany) under isoflurane anesthesia. The sera were obtained after 5 min of centrifugation at 8000 rpm and stored at −20 • C until further use. Routinely, mice were bled at day 0 before immunization (pre-immune sera), at day 28 (3 weeks after the first immunization), and at day 49 (2 weeks after the second immunization). For serological follow-up, additional bleedings were performed at days 63 and 77.

Analyses of Humoral Immune Responses
Antibody responses in 1:100 diluted sera were determined by antigen-specific ELISA as previously described for HIV-Env [7] and HEL [13]. For ELISA measuring of KF-specific antibody responses, recombinant KF protein was produced and purified as described elsewhere [16]. 96 F Maxisorp white microwell plates (Thermo Scientific) were coated with 100 µL of 1 µg/mL of recombinant KF protein in 0.1 M bicarbonate buffer (pH 9.6) overnight at 4 • C. After washing with PBS containing 0.05% Tween-20 (PBS-T), wells were blocked with 5% skimmed milk powder in PBS-T (blocking buffer) for one hour at room temperature (RT). Following another washing step, sera diluted 1:100 in blocking buffer were added to wells for one hour at RT. After washing, the wells were incubated with HRP-conjugated anti-mouse IgG (Bethyl), IgG1, or IgG2a (both from BD Biosciences) diluted 1:1000 in blocking buffer for 1 h at RT. Bound HRP-conjugated antibodies were detected with an enhanced chemiluminescence solution composed of 5 mL Luminol solution (3-aminophtolhydrazide, Sigma-Aldrich), 50 µL solution B (p-coumaric acid, Sigma-Aldrich), and 1.6 µL 30% H2O2 (Merck, Darmstadt, Germany). Humoral immune responses were measured with Berthold Detection Systems Orion Microplate Luminometer (Berthold Technologies) and are expressed as log10-transformed relative light units per second (RLU/s log10).

Characterization of Cellular Immune Responses
Two weeks after the second immunization (day 49), Balb/c mice were sacrificed, draining inguinal lymph nodes were removed, and single-cell suspensions were prepared. Cells from non-immunized animals served as a negative control (contr.). Cells were re-stimulated with MHC-II-restricted HIV-Env peptide GVPVWKEATTTLFCASDAKA in the presence of 2 µg/mL anti-CD28 (37.51; eBioscience) and 2 µmol monensin as described elsewhere [9]. After 6 h of stimulation, cells were surface-stained with anti-CD4-FITC and intracellular-stained with anti-IFN-γ-PE antibodies (all from eBioscience) as described [9]. Data were acquired on BD FACSCanto II (BD Biosciences) and analyzed with FlowJo (Tree Star).

Statistical Analysis
Calculations were performed as indicated in the figure legends using GraphPad Prism 7 software (GraphPad, San Diego, CA, USA).

Generation of a Membrane-Bound Form of Truncated KF for HIV-VLP Functionalization
HIV-derived enveloped VLPs can efficiently serve as surface antigen displays consisting of a common HIV-Gag protein backbone and the cellular lipid membrane in which the antigen of interest can be expressed. Previously, we functionalized HIV-Gag viral nanoparticles with model antigens that originally are soluble non-viral proteins [5]. Now, we used this genetic approach to integrate a truncated form of soluble bacterial flagellin of a nonpathogenic E. coli strain that lacks the domains D2 and D3 into HIV-Gag-based viral nanoparticles. Deletion of hypervariable domains ND2-D3-CD3 reduced the immunogenicity of the protein and the systemic inflammatory response against it, but retained the TLR5 agonist activity [16,20]. We used sequences of original plasmids [5,16] (Figure 1A) to insert N-terminal D0-D1 domains (ND0-ND1) connected via a linker with C-terminal D1-D0 domains (CD1-CD0) between the sequences coding for the leader peptide and the transmembrane, as well as the cytoplasmic domains of the vesicular stomatitis virus G-protein (VSV-G). Figure 1A represents the resulting pKF∆-TM construct. The rationale behind this design was: (i) To create HIV-based nanoparticles displaying multiple flagellin molecules on the surface with an orientation optimized for TLR5 recognition [21], and (ii) to enhance molecule flexibility and achieve cis-dimerization of the N-terminal and C-terminal D0-D1 domains by introducing a flexible linker in-between ( Figure 1B). recognition [21], and ii) to enhance molecule flexibility and achieve cis-dimerization of the N-terminal and C-terminal D0-D1 domains by introducing a flexible linker in-between ( Figure 1B).
The flow cytometry analysis of HEK293T cells transiently transfected with pKFΔ-TM or with soluble KFD-p24 3D confirmed the presence of KFΔ protein on the cell surface in the expected orientation as a type 1 membrane protein exposing its N-terminus to the extracellular space ( Figure  1C). After co-transfection of 293T cells with pKFΔ-TM and Hgpsyn (a codon-optimized expression plasmid that encodes HIV-1 GagPol proteins), VLPs could be pelleted by ultracentrifugation through a sucrose cushion. Western blot analyses demonstrated the presence of both HIV-Gag (p55, Gag; p24, capsid) and KFΔ in the pelleted nanoparticles ( Figure 1D). HIV-Gag-mediated budding does not change the orientation of the protein, and therefore KFΔ is presented at the outer surface of the nanoparticles as an ordinary envelope protein.  As reported by Yuan Lu et al., when D0-stabilized flagellin was chemically attached to nonenveloped Hepatitis B core protein VLPs with the D0 domain facing outward, the tendency of flagellin to polymerize caused the VLPs to precipitate. However, attaching the D0 domain to the VLP surface produced a stable nanoparticle adjuvant [21]. According to our pKFΔ-TM design, KFΔ flagellin domains have capacities for cis-dimerization during the particle production by the cell, and released nanoparticles have D0 domains facing the particle surface ( Figure 1A,B). To control transdimerization and nanoparticle conjugation, we measured HIV-Gag protein backbone VLPs and VLPs functionalized with KFΔ (VLP-KFΔ) by dynamic light scattering. No abnormal VLP-KFΔ precipitates  The flow cytometry analysis of HEK293T cells transiently transfected with pKF∆-TM or with soluble KFD-p24 3D confirmed the presence of KF∆ protein on the cell surface in the expected orientation as a type 1 membrane protein exposing its N-terminus to the extracellular space ( Figure 1C). After co-transfection of 293T cells with pKF∆-TM and Hgpsyn (a codon-optimized expression plasmid that encodes HIV-1 GagPol proteins), VLPs could be pelleted by ultracentrifugation through a sucrose cushion. Western blot analyses demonstrated the presence of both HIV-Gag (p55, Gag; p24, capsid) and KF∆ in the pelleted nanoparticles ( Figure 1D). HIV-Gag-mediated budding does not change the orientation of the protein, and therefore KF∆ is presented at the outer surface of the nanoparticles as an ordinary envelope protein.
As reported by Yuan Lu et al., when D0-stabilized flagellin was chemically attached to non-enveloped Hepatitis B core protein VLPs with the D0 domain facing outward, the tendency of flagellin to polymerize caused the VLPs to precipitate. However, attaching the D0 domain to the VLP surface produced a stable nanoparticle adjuvant [21]. According to our pKF∆-TM design, KF∆ flagellin domains have capacities for cis-dimerization during the particle production by the cell, and released nanoparticles have D0 domains facing the particle surface ( Figure 1A,B). To control trans-dimerization and nanoparticle conjugation, we measured HIV-Gag protein backbone VLPs and VLPs functionalized with KF∆ (VLP-KF∆) by dynamic light scattering. No abnormal VLP-KF∆ precipitates were observed ( Figure 1E).
Thus, we developed an effective genetic method for functionalization of HIV-based viral nanoparticle membranes with the truncated form of flagellin.

Activation of Antigen-Presenting Cells in Vitro
To analyze the immunomodulatory effects of KF∆ functionalization on the activation of antigen-specific (cognate) and non-specific (non-cognate) B-cells, we produced native HIV-1 nanoparticles, containing GagPol and Env proteins of HIV-1 (Env-VLP), and Env-VLPs functionalized with KF∆ (Env-VLP-KF∆). The amount of Env (the antigen of interest) in VLP preparations was routinely measured with Env ELISA, and the presence of KF∆ was confirmed with WB.
As it was shown before, soluble KFD-p24 3D activates DCs from wt mice, but not from TLR5-knockout mice in vitro [14]. Therefore, we first proved the bioactivity of the KF∆-functionalized nanoparticles on freshly isolated splenic DCs from wt Bl6 mice. DCs were cultured in the presence of either Env-VLP or Env-VLP-KF∆. The results clearly indicate the bioactivity effects of the KF∆-functionalized nanoparticles on the primary innate immune cells (Figure 2A). Previously, we demonstrated that HIV-based nanoparticle vaccines can directly target and activate naive cognate B-cells in vitro [5]. To investigate whether KFΔ functionalization is able to: i) further improve the activation capacities of Env-VLP on cognate HIV-Env-specific B-cells, and ii) induce polyclonal activation of non-cognate B-cells, we incubated spleen cell suspensions from wildtype Bl6 and BCR-transgenic b12 mice with either Env-VLP or Env-VLP-KFΔ. LPS stimulation was Previously, we demonstrated that HIV-based nanoparticle vaccines can directly target and activate naive cognate B-cells in vitro [5]. To investigate whether KF∆ functionalization is able to: (i) further improve the activation capacities of Env-VLP on cognate HIV-Env-specific B-cells, and (ii) induce polyclonal activation of non-cognate B-cells, we incubated spleen cell suspensions from wild-type Bl6 and BCR-transgenic b12 mice with either Env-VLP or Env-VLP-KF∆. LPS stimulation was used as an internal positive control for the cell activation inducibility (all cells used could be activated with LPS, data not shown). B-cells from b12 mice are able to recognize HIV-Env with their BCRs, which results in B-cell activation. Functionalization of Env-VLPs with KF∆ significantly increased the activation capacities of the Env-based nanoparticles on the cognate B-cells, but had no influence on non-cognate Bl6 B-cells ( Figure 2B).
The observed adjuvantive effect on cognate B-cells might be either due to: (i) a synergistic mode of BCR activation and the direct sensing of KF∆ by activated B-cells, or (ii) a synergistic effect of BCR activation and paracrine influence of KF∆-activated innate immune cells from the spleen suspension. To evaluate the nature of the adjuvantive effect (direct vs. indirect), we isolated primary b12 B-cells out of the spleen cell suspensions and directly stimulated them with Env-VLP or Env-VLP-KF∆. In contrast to the spleen cell suspensions, no adjuvantive effect of KF∆ functionalization was observed ( Figure 2C). These data are consistent with results of Gururjan et al., who demonstrated that mouse naive follicular B-cells do not express TLR5 and do not respond to flagellin stimulation in vitro [22]. Taken together, HIV-1-based viral nanoparticles demonstrated a direct stimulatory effect on dendritic cells and an indirect adjuvantive effect on VLP-stimulated cognate B-cells in co-culture systems after functionalization with KF∆.

Modulation of Env-Specific Antibody and CD4+ T-Cell Responses In Vivo
In order to analyze the adjuvantive effects of KF∆ functionalization on the induction of anti-Env-specific antibody responses, we immunized wt Balb/c mice with Env-VLP or Env-VLP-KF∆ containing the same amounts of HIV-Env antigen. Nanoparticles were applied intramuscularly (a clinically relevant administration route) two times over the four-week interval. Two weeks after the second immunization, total anti-Env IgG antibody responses ( Figure 3A) as well as anti-Env-specific IgG1 ( Figure 3B) and IgG2a ( Figure 3C) subclasses in sera of immunized mice were measured and compared with those in sera before immunizations (pre-immune).
Unpredictably, the KF∆ functionalization significantly decreased anti-Env Ab levels of total IgG and IgG1 subclass ( Figure 3A,B) levels and totally abrogated induction of IgG2a ( Figure 3C), the predominant antiviral IgG antibody subclass in the mouse [23]. To exclude that the KF∆ functionalization only transiently decreases antibody responses, mice were immunized three times and a serological follow-up over 11 weeks was performed. Anti-HIV-Env-specific IgG antibody levels induced by Env-VLP-KF∆ immunization continued to be lower than in Env-VLP immunized mice ( Figure A2). Then, we measured the immunogenicity of KF∆ and observed strong induction of anti-KF-specific antibodies with equally strong production of both IgG1 and IgG2a antibodies ( Figure 3D-F).
In the splenocyte co-cultures, we observed the adjuvantive effect of KF∆ on Env-specific B-cell activation ( Figure 2B). However, in vivo compartmentalization of the immune system components might prevent such indirect effects. After i.m. HIV-based VLP injection, DCs take up the nanoparticles, become activated, and migrate into the T-zones of the secondary lymphoid organs to further activate and instruct cognate CD4+ T-cells. Simultaneously, a direct contact with VLPs facilitates B-cell activation in B-cell zones [6,24]. Activated cognate T-and B-cells migrate to contact each other, because the T-/B-cell collaboration is essential for the generation of antibody-producing plasma cells [24].
Since (i) primary in vivo activation of antigen-specific B-cells requires a direct cognate VLP triggering rather than interactions with VLP-loaded DCs [6] and (ii) Env-VLP-KF∆ demonstrated no direct adjuvantive effect on Env-specific B-cells ( Figure 2C), in vivo induction of cognate CD4+ T-cells by DCs might play a role in the regulation of anti-Env antibody production [25,26].
We analyzed Env-specific CD4+ T-cell responses after immunization with Env-VLP vs. Env-VLP-KF∆. IFN-γ production by CD4+ T-cells after re-stimulation with HIV-Env MHC-class-II restricted immunodominant peptide was significantly impaired in mice immunized with Env-VLP-KF∆ ( Figure 3G), which is consistent with the strong decline in IgG2a antibodies [27].
Taken together, immunization of Balb/c mice with HIV-1-based viral nanoparticles functionalized with KF∆ decreased the Env-specific CD4+ T-cell activation and anti-Env antibody production, while anti-KF antibodies were prominently produced. might prevent such indirect effects. After i.m. HIV-based VLP injection, DCs take up the nanoparticles, become activated, and migrate into the T-zones of the secondary lymphoid organs to further activate and instruct cognate CD4+ T-cells. Simultaneously, a direct contact with VLPs facilitates B-cell activation in B-cell zones [6,24]. Activated cognate T-and B-cells migrate to contact each other, because the T-/B-cell collaboration is essential for the generation of antibody-producing plasma cells [24].
Since i) primary in vivo activation of antigen-specific B-cells requires a direct cognate VLP triggering rather than interactions with VLP-loaded DCs [6] and ii) Env-VLP-KFΔ demonstrated no direct adjuvantive effect on Env-specific B-cells ( Figure 2C), in vivo induction of cognate CD4+ Tcells by DCs might play a role in the regulation of anti-Env antibody production [25,26].
We analyzed Env-specific CD4+ T-cell responses after immunization with Env-VLP vs. Env-VLP-KFΔ. IFN-γ production by CD4+ T-cells after re-stimulation with HIV-Env MHC-class-II restricted immunodominant peptide was significantly impaired in mice immunized with Env-VLP-KFΔ ( Figure 3G), which is consistent with the strong decline in IgG2a antibodies [27].
Taken together, immunization of Balb/c mice with HIV-1-based viral nanoparticles functionalized with KFΔ decreased the Env-specific CD4+ T-cell activation and anti-Env antibody production, while anti-KF antibodies were prominently produced.

Adjuvantive Effect on HEL-Specific Antibody Responses in Vivo
The results obtained after Env-VLP-KFΔ immunization of Balb/c mice, however, were in contradiction with data previously published by us [13] and others [28]. Using B-cell targeting calcium-phosphate (CaP) nanoparticles functionalized with hen egg lysozyme as a model antigen, we demonstrated that additional functionalization with the full length flagellin from Salmonella enterica significantly improved anti-HEL antibody responses in C3H mice [13].

Adjuvantive Effect on HEL-Specific Antibody Responses In Vivo
The results obtained after Env-VLP-KF∆ immunization of Balb/c mice, however, were in contradiction with data previously published by us [13] and others [28]. Using B-cell targeting calcium-phosphate (CaP) nanoparticles functionalized with hen egg lysozyme as a model antigen, we demonstrated that additional functionalization with the full length flagellin from Salmonella enterica significantly improved anti-HEL antibody responses in C3H mice [13].
To verify these results for KF∆ functionalization in the context of HIV-based viral nanoparticles, we produced HEL-VLP (VLPs carrying HEL as a surface antigen [5]) and HEL-VLP functionalized with KF∆ (HEL-VLP-KF∆) (see Table 1). After immunization of C3H mice with either HEL-VLP or HEL-VLP-KF∆, we observed positive adjuvantive effects of KF∆ functionalization on the induced anti-HEL antibody responses ( Figure 4A-C). The anti-HEL total IgG ( Figure 4A) and IgG2a ( Figure 4C) subclass levels were significantly higher in mice immunized with HEL-VLP-KF∆, while induced anti-KF antibody responses remained relatively moderate ( Figure 4D-F vs. Figure 3D-F). Thus, functionalization of B-cell-targeting HEL-VLP vaccines with KF∆ demonstrated an adjuvantive effect and increased anti-HEL IgG antibody responses in C3H mice.

Immunogenicity Balances between the Target Antigens and KFΔ
C3H mice are genetically predisposed to strongly react to HEL protein with antibody production [29]. We compared induction of the total IgG humoral immune response ( Figure 5A) and the potent antiviral IgG2a subtype response ( Figure 5B) in C3H mice against HEL proteins with those in Balb/c mice against HIV-Env proteins four weeks after a single immunization with corresponding VLPs. Based on the differences in the magnitude of antigen-specific antibody induction on the same amount of antigens between Balb/c and C3H mice, one might consider that HIV-Env is a rather weak immunogen in Balb/c mice. . Induction of humoral immune responses in C3H mice with HEL-VLP nano-vaccines functionalized with KF∆. Wt C3H mice were immunized i.m. twice (at day 0 and day 35) with either HEL-VLP or HEL-VLP-KF∆ containing 400 ng of HEL/mouse. The sera samples were obtained at day 0 before immunization (pre-immune sera) and at day 49 (2 weeks after the second immunization). Humoral immune responses against HEL (A-C) and KF (D-F) were measured in 1:100 diluted sera samples and expressed as log10-transformed relative light units per second (RLU/s log10). Each experimental group included six mice. the pre-immune sera group contains samples of 12 mice correspondingly. The columns represent the mean values ± SEM. * p = 0.0169; *** p = 0.0009; **** p < 0.0001; one-way ANOVA with Tukey multiple comparison post-hoc test.

Immunogenicity Balances between the Target Antigens and KF∆
C3H mice are genetically predisposed to strongly react to HEL protein with antibody production [29]. We compared induction of the total IgG humoral immune response ( Figure 5A) and the potent antiviral IgG2a subtype response ( Figure 5B) in C3H mice against HEL proteins with those in Balb/c mice against HIV-Env proteins four weeks after a single immunization with corresponding VLPs. Based on the differences in the magnitude of antigen-specific antibody induction on the same amount of antigens between Balb/c and C3H mice, one might consider that HIV-Env is a rather weak immunogen in Balb/c mice. Although at least one Env MHC-class-II-restricted immunodominant peptide for Balb/c mice is known, after immunization with Env-VLP-KFΔ, there was no induction of Env-specific IFN-γ producing CD4+ T-cells observed. This lack of T-cells was accomplished with diminished anti-Env IgG2a antibody responses. At the same time, significant levels of anti-KF IgG2a antibodies were detected, suggesting that KFΔ has no negative influence on IgG2a class-switch per se (Figure 3). One Although at least one Env MHC-class-II-restricted immunodominant peptide for Balb/c mice is known, after immunization with Env-VLP-KF∆, there was no induction of Env-specific IFN-γ producing CD4+ T-cells observed. This lack of T-cells was accomplished with diminished anti-Env IgG2a antibody responses. At the same time, significant levels of anti-KF IgG2a antibodies were detected, suggesting that KF∆ has no negative influence on IgG2a class-switch per se (Figure 3). One can speculate that in the context of VLP-based B-cell nano-vaccines, KF∆ serves as a protein antigen itself and may outcompete a less immunogenic antigen with its antibody response via the CD4+ T-cell-dependent mechanism, which was recently described as a model of B-cell competition for T-cell help [30]. In contrast, in combination with a strong immunogen (such as HEL in C3H mice) the adjuvant activity of flagellin may dominate over its immunogenicity. This might explain the contradictory data previously published by Vassilieva et al., in which flagellin functionalization enhanced the Env-specific humoral responses to Env-VLP immunization in guinea pigs [28]. Guinea pigs are phylogenetically distant from mice and have higher immunoglobulin combinatorial diversity [31]. These animals are broadly used in HIV studies due to their ability to induce HIV-Env neutralizing antibodies [32], which might indicate a stronger immunogenicity of HIV-Env in guinea pigs than in mice.
To consolidate our hypothesis, we analyzed the immunogenicity balances between the VLP envelope antigens and the KF∆ adjuvant for individual Balb/c and C3H mice. The results summarized in Figure 5C-D provide an empirical support of the antigen competition idea and suggest that by functionalization of viral nanoparticle vaccines, the immunogenicity of flagellin itself has to be taken into consideration. The adjuvantive potency of flagellin might be dependent on both the immunogenicity of each particular antigen used and the genetic background (e.g., MHC-class-II variants) of the vaccine.
In contrast to MHC-class-II mediated presentation to CD4+ T-cells, cross-presentation of VLP-derived antigens to CD8+ T-cells is usually restricted to a subset of CD8+ DCs [33]. However, this CD8+ DC subset that has the ability to cross-present [34] demonstrates the lowest level of relative TLR5 expression in comparison to the other conventional DCs [35]. This implies the need for comprehensive studies on the cross-presentation of flagellin-functionalized VLPs by different DC subsets. Although induction of cytotoxic T-lymphocyte (CTL) responses for prophylactic HIV-1 vaccines was formally concluded in 2007 with the unexpected lack of efficacy in the STEP trial [36], the genetic functionalization of cell-derived nanoparticles (enveloped VLPs and exosomes [37]) with truncated flagellin might be of interest for the further development of CTL-inducing anti-viral and anti-cancer vaccines.

Conclusions
In summary, we developed an effective genetic method for functionalization of HIV-1-based viral nanoparticle membranes with a truncated form of flagellin. KF∆ is presented at the outer surface of the nanoparticles as an ordinary envelope protein. Rational design of the KF∆ molecule prevents precipitation of nanoparticles and exposes the TLR-5 binding site outwards. KF∆-functionalized HIV-based nanoparticles demonstrated bioactivity in vitro. Functionalization of B-cell-targeting viral nanoparticle vaccines with KF∆ demonstrated both inhibitory and adjuvantive immunostimulatory effects on humoral immune responses against target antigens. The overall outcome of immunizations was based on the immunogenicity balance between the antigen and KF∆. Therefore, in contrast to other TLR-ligands, the antigenic property of flagellin in comparison to the immunogenicity of the target antigen has to be considered for the functionalization of viral nanoparticle vaccines. Acknowledgments: We thank Bastian Grewe (Ruhr University Bochum, Germany) for his contribution to the generation of the pKF∆-TM expression plasmid, Harald Unterweger (Friedrich-Alexander University Erlangen-Nürnberg, Germany) for his assistance with DLS measuring, and David Nemazee (The Scripps Research Institute, La Jolla, CA, USA) for b12 mice donation. We acknowledge support by Deutsche Forschungsgemeinschaft and Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) within the funding programme Open Access Publishing.

Conflicts of Interest:
The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Appendix A
Pharmaceutics 2019, 11, x FOR PEER REVIEW 13 of 16

Conflicts of Interest:
The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Figure A1. Cloning strategy for the generation of the pKFΔ-TM expression plasmid. The pKF∆-TM expression plasmid consisted of the pC-HEL-TM backbone as well as sequences coding for domains CD1-CD0 of flagellin and the ND0-ND1-linker, which additionally encoded a linker composed of two repeats of 11 amino acids of the human IgG3 hinge region [14]. Because of the VSV-G leader peptide and the VSV-G TM domain, pC-HEL-TM was used as backbone for pKF∆-TM after removal of the HEL sequence by digesting with BamHI. The fragments of the ND0-ND1-linker and CD1-CD0 were separately amplified from the pKFD-p24 3D expression plasmid in PCR1: Primer sequences: ND0-ND1-linker fwd: 5'-CGTGAACTGCGGATCCGCACAAGTCATTAATACCAACAGCCT-3' and ND0-ND1-linker rev: 5'-ATCCGTGGTTTTACCTCCTGATGTGTGTGTAGTATCACCAAGAGGTGTGC-3' and from the pKF expression plasmid in PCR2: Primer sequences: CD1-CD0 fwd: 5'-GGTAAAACCACGGATCCGCTGA-3′ and CD1-CD0 rev: 5′-TGTCGCCGAAGGATCCACCCTGCAGCAGAGACAGAACC-3′. The insertion of both fragments into the pC-HEL-TM backbone was performed using the In-Fusion HD Eco Dry Kit from Clontech according to the manufacturer's protocol.