Enhancement of Peptide Vaccine Immunogenicity by Increasing Lymphatic Drainage and Boosting Serum Stability

The capacity of antitumor T-cell responses to be curative in cancer patients is known from clinical studies of adoptive cell therapy, but the induction of potent endogenous T-cell responses through therapeutic vaccination remains a challenge (1, 2). Trials where patients received personalized vaccines targeting predicted neoantigens have generated significant enthusiasm in the field because they demonstrate the feasibility of immunizing against antigens that should not be restrained by tolerance (3, 4). Yet the immunogenicity of the vaccines reported in these trials remains modest. For example, Ott and colleagues used “long” neoantigen-containing peptides as immunogens, which are attractive due to their defined nature, relative ease of manufacturing, favorable safety profile, and requirement of professional antigen presenting cells (APCs) for presentation (5). However, as observed in many other peptide vaccine trials, CD8⁺ T-cell responses were detectable only after extended peptide stimulation ex vivo (4). Much room remains for improvement in the immunogenicity of these vaccines.

To generate more potent peptide vaccines, a variety of approaches based on optimized formulation and immunization regimens have been demonstrated. Vaccine responses have been boosted by the utilization of altered peptide ligands, or high-affinity, non-self epitopes that prime cross-reactive T-cell responses against tumor antigens (6). Significant effort has been put into developing adjuvants to enhance T-cell responses following peptide vaccination, including the use of TLR agonists (7), type I interferons (8), and Treg-disrupting monoclonal antibodies (7, 9). Vaccine formulation can also contribute to the persistence and quality of antitumor immunity. One of the most common emulsion formulations, Incomplete Freund's Adjuvant (IFA), has been found to depot peptide antigens at the injection site and sequester T cells distal from the tumor, which may harm antitumor immunity, whereas formulations that are nonpersisting at the site of vaccination may prove to be more effective (10).

The poor lymphatic drainage and short in vivo half-life of linear peptides likely contribute to the low potency of peptide vaccines (11–14). We previously reported that enhanced lymph node (LN)-targeting of peptides can be achieved by covalently modifying peptides and molecular adjuvants with diacyl lipids, with the peptides linked through a PEG spacer, and this trafficking enhancement results in a corresponding increase in T-cell priming (14). The mechanism of action is thought to involve binding of the lipid tails to endogenous albumin, allowing peptides and molecular adjuvants to “hitchhike” on albumin to the draining LN, similar to the mechanism of action of sentinel LN mapping dyes in clinical use (15, 16).

Here, we explored in detail mechanisms of action of these molecular vaccines and evaluated the role of distinct components of the amphiphile (amph)-vaccine platform. We tested the relative efficacy of several different albumin-binding moieties, the stability of amph-vaccine conjugates in the presence of serum proteases, the effect of using different adjuvants with amph-peptides, and the importance of the neonatal Fc receptor and cross-presenting dendritic cells (DCs) in amph-vaccine immunogenicity. We also assessed the duration and location of antigen presentation following amph-vaccine administration in vivo. Together, our data support the use of albumin-binding as a general design criterion for improving peptide vaccines by improving stability and lymphatic distribution, resulting in enhanced and prolonged antigen presentation in the LNs for superior CD8⁺ T-cell priming.

Solid-phase DNA synthesis, purification, and characterization of lipo-CpG (ODN 1826) were carried out as previously described (14). Synthesis of the vitamin E-CpG conjugate was performed as described in ref. 14 using α-Tocopherol-TEG Phosphoramidite (Glen Research) with the following changes: after conjugation in solid phage, cleavage and ultramild deprotection was performed using 3:1 tertbutylamine in water for 6 hours at 65°C and the 5′ dimethoxytrityl (DMT) group was removed using 80% acetic acid at room temperature for 2 hours, followed by an ethanol precipitation. Following cleavage and deprotection, oligos were purified via RP-HPLC and quantified using UV-VIS. For synthesis of TAMRA-labeled vitamin E-CpG, a 3′ TAMRA bead was utilized (TAMRA lcaa CpG 500Å, ChemGenes), followed by the ultra-mild cleavage and deprotection conditions described above. Synthesis of vitamin B3 CpG was performed using 5′-Niacin-CE Phosphoramidite (Link Technologies). All CpG oligonucleotide conjugates used an ODN class B sequence 1826 with two guanine spacers.

Reaction partners and conjugate structures are shown in Supplementary Fig. S1. The peptide antigen was first synthesized using 9-fluorenyl methoxycarbonyl (Fmoc) solid-phase peptide synthesis with preloaded Wang resin (EMD Millipore). Synthesis was performed on a CEM Liberty Blue automated microwave peptide synthesizer. Coupling reactions were performed using 4 eq. of Fmoc-protected amino acid, 4 eq. of N,N'-diisopropylcarbodiimide (DIC)) and 8 eq. of ethyl(hydroxyimino)cyanoacetate (Oxyma pure) and removal of Fmoc groups on resin-attached amino acids was achieved with 20% 4-methylpiperidine in DMF. After synthesis, peptides were cleaved off the resin in a 95:2.5:2.5 trifluoroacetic acid (TFA)/triisopropylsilane (TIPS)/H₂O mixture for 3 hours and washed with dichloromethane. Rotary evaporation and precipitation in cold diethyl ether yielded the crude product. Crude peptides were purified by HPLC (Waters Prep 150) on a C18 Phenomenex Kinetex column in a water-acetonitrile gradient containing 0.1% v/v TFA. Pure fractions were collected and identified using ESI-MS. The combined fractions were subjected to rotary evaporation to remove acetonitrile, frozen in liquid nitrogen, and lyophilized to dryness. Purity was verified to be >95% using LC-MS on an Agilent QTOF 6520 LC-MS. Example chromatograms and ESI-MS spectra are shown in Supplementary Fig. S2.

For the DSPE-PEG₂₀₀₀-AVGALEGPRNQDWLGVPRQL-COOH, purified H₂N-AVGALEGPRNQDWLGVPRQL-COOH was dissolved in DMF along with excess diisoproplyethylamine (DIEA). This mixture was then added to approximately 2 molar equivalents of DSPE-PEG₂₀₀₀-NHS (Nanosoft Polymers) also dissolved in DMF. Reaction progress was monitored by ESI-MS until the peptide peak was consumed (approximately 18 hours) and only the polydisperse product peak, and some DSPE-PEG₂₀₀₀ starting material were seen. The reaction mix was then diluted with water, and product was purified by HPLC using a Phenomenex Jupiter C12 column. After lyophilization to dryness, LC-MS confirmed purity to be >95% and the product mass was a polydisperse mixture centered around 5,162 m/z.

To synthesize DSPE-PEG₂₀₀₀-CAVGALEGPRNQDWLGVPRQL-COOH, purified H₂N-CAVGALEGPRNQDWLGVPRQL-COOH was dissolved in DMF and added to approximately 1.25 molar equivalents of DSPE-PEG₂₀₀₀-maleimide (Laysan Bio) also dissolved in DMF. Reaction progress was monitored by ESI-MS until the peptide peak was consumed (approximately 24 hours), and only the polydisperse product peak and some DSPE-PEG₂₀₀₀-maleimide were seen. The reaction mix was then diluted with water and purified as above. After lyophilization to dryness, LC-MS confirmed purity to be >95% and the product mass was a polydisperse mixture centered around 5,349 m/z.

For synthesis of DSPE-PEG₂₀₀₀-(click)-AVGALEGPRNQDWLGVPRQL-COOH, with peptide H₂N-AVGALEGPRNQDWLGVPRQL still on resin, 2 molar equivalents of 5-azidopentanoic acid, 1.9 molar equivalents of HBTU and 8 molar equivalents of DIEA in DMF were added. This mixture was allowed to react overnight before standard cleavage from resin and purification as described above. The purified peptide (Az)-AVGALEGPRNQDWLGVPRQL-COOH was then dissolved in DMSO along with DSPE-PEG₂₀₀₀-DBCO (Avanti) in approximately a 1:1 molar ratio. The reaction was monitored until the peptide peak was nearly consumed and the product was then purified as above. After lyophilization to dryness, LC-MS confirmed purity to be >95% and the product mass was a polydisperse mixture centered around 5,405 m/z.

For the ABP-antigen constructs, following synthesis of the antigen component, the PEG₂₀₀₀ linker (Creative PEGworks) was coupled manually onto peptidyl resin containing the antigen peptide; the resin was then transferred back to the synthesizer to complete the construct (17). The crude peptides were cleaved overnight at room temperature (by volume, 82.5% trifluoroacetic acid, 5% water, 5% thioanisole, 5% phenol, and 2.5% ethane dithiol), triturated with −80°C diethyl ether, and purified via RP-HPLC. The ABP moieties were cyclized by incubating the purified constructs (0.1 mg/mL) in 0.1 mol/L NaHCO₃ (Sigma), pH 8.0 for 24 hours to facilitate disulfide formation. The cyclized constructs were desalted via solid-phase extraction, lyophilized, and analyzed via LC-MS to verify identity and purity. An example chromatogram and corresponding mass spectrum is shown in Supplementary Fig. S3A and S3B. Conditions for cyclization were verified on a monodisperse construct with G4 spacer (Supplementary Fig. S3C and S3D).

For the labeled constructs, EGP long and ABP-PEG₂₀₀₀-EGP were synthesized as described above, incorporating a C-terminal propargylglycine to serve as a chemical handle. Prior to disulfide cyclization, purified EGP long and ABP-PEG₂₀₀₀-EGP long peptides were coupled to TAMRA-azide (tetramethylrhodamine, 5-isomer, Lumiprobe) via Cu(I)-catalyzed click chemistry. Briefly, ∼0.5 μmol of each peptide was dissolved in 720 μL of 50:50 b.v. solution of t-butanol (Sigma) and water. The following reagents were then added (final concentration in parentheses): TAMRA-azide (500 μmol/L), Tris (pH 8.0, final concentration 50 mmol/L; Roche), ascorbic acid (100 mmol/L, Sigma), TCEP (1 mmol/L, Hampton Research), TBTA (100 μmol/L; Alfa Aeser), and copper sulfate (5 mmol/L; Sigma) to a final volume of 1.00 mL. The reactions were mixed thoroughly and incubated at room temperature for one hour. The reactions were quenched in 20 mL 95:5 water:acetonitrile with 0.1% TFA additive, purified via RP-HPLC, and lyophilized. The TAMRA-labeled constructs were cyclized and desalted as described above.

Peptide antigen sequences used are as follows: gp100 EGP long, EGP₂₀ (gp100_20-39): AVGALEGPRNQDWLGVPRQL (18); gp100 EGP short, EGP₉: EGPRNQDWL; E7 from HPV: RAHYNIVTF; SIV-gag AL11: AAVKNWMTQTL; the MHC II-restricted neoantigen MUT30: VDWENVSPELNSTDQ (3); OVA_251-270: GLEQLESIINFEKLTEWTSS; and HIV gag p17_64-93: LQPSLQTGSEELRSLYNTVATLYCVHQRIE. The ABP sequence chosen was the minimal core sequence identified in a phage display library of albumin binders: DICLPRWGCLW (19).

All animal studies were carried out under an institute-approved IACUC protocol following federal, state, and local guidelines for the care and use of animals. C57BL6/J mice were procured from The Jackson Laboratory. Neonatal Fc receptor knockout mice (FcRn^−/−, B6.129 × 1-Fcgrt^tm1Dcr/DcrJ), Batf3^−/− mice (B6.129S(C)-Batf3^tm1Kmm/J), and pmel-1 mice (B6.Cg-Thy1^a/Cy Tg(TcraTcrb)8Rest/J) were obtained from The Jackson Laboratory and bred in-house. Six- to 12-week-old female mice were used for these studies.

Immunizations comprised of peptides or peptide conjugates mixed with adjuvant were administered subcutaneously at the base of the tail in 100 μL total, 50 μL on each side, with sterile saline diluent. The same vaccine preparation and injection technique was used for both prime (day 0) and boost (day 14). Whole protein OVA (LowEndo, Worthington Biochemical) was dosed at 10 μg. Peptides were dosed at 10 μg and conjugates were dosed at the molar equivalent. Adjuvants c-di-GMP (InvivoGen) and polyI:C (HMW, InvivoGen) were dosed at 25 ug, and lipo-CpG and vitamin E-CpG were dosed at 1.24 nmol (equivalent to 7.9 mg CpG). For LN trafficking studies using the Spectrum In Vivo Imaging System (IVIS, PerkinElmer), the CpG dose was 3.3 nmol, and the peptide dose was 15 nmol.

For trafficking studies and immunization studies using IFA (Montanide ISA 51 VG, SEPPIC), aqueous peptide solutions were formulated at a 1:1 ratio with IFA and emulsified using a syringe adapter emulsifier provided by the manufacturer until a stable emulsion was formed. In this experiment only, mice were boosted twice, on days 14 and 28, and intracellular cytokine staining (ICS) was performed on day 35.

Pmel-1 T cells were isolated from spleens of pmel-1 mice with an EasySep CD8⁺ T-cell Isolation Kit (Miltenyi), according to the manufacturer's protocol. Purified pmel-1 T cells were labeled by incubating for 20 minutes at 37°C in 5 μmol/L CFSE or CellTrace Violet (Thermo Fisher Scientific) in PBS plus 0.1% BSA, quenched with RPMI plus 20% FBS, and washed. T cells were resuspended in sterile saline, and 5 × 10⁵ pmel-1 T cells were transferred retro-orbitally into recipient wild-type (WT) mice. Inguinal, axillary, brachial, lumbar, cervical, and mesenteric LNs were collected, mechanically dissociated, stained, and analyzed by flow cytometry on a BD LSR Fortessa with the antibodies indicated in the “Evaluation of murine immune responses and flow cytometry” section. Proliferation index (PI) was quantified using PI = log(FI_nd/MFI_all)/log(2), where MFI_all = median fluorescence intensity of live Thy1.1⁺ pmel-1 T cells and FI_nd = peak fluorescence intensity of viable nondivided pmel-1 T cells.

To assess the serum stability of peptides, four EGP₂₀ constructs [EGP₂₀, amph-EGP₂₀ (N), amph-EGP₂₀ (C), and PEG₂₀₀₀-EGP₂₀] were subjected to overnight incubation at 37°C in RPMI with 10% fresh mouse serum over concentrations ranging from 10 ng/mL to 2.4 μg/mL. Serum-treated or control constructs were then incubated with pooled splenocytes from 3 previously DSPE-PEG-EGP₂₀ immunized mice (50,000 splenocytes per condition) for 24 hours at 37°C (in RPMI with 10% heat inactivated FBS), with brefeldin A added at 18 hours and stained and run on the flow cytometer as described below.

Antibodies to IFNγ (clone XMG1.2), TNFα (clone MP6-XT22), CD69 (clone H1.2F3), PD-1 (clone 29F.1A12), Tim-3 (clone B8.2C12), CD44 (clone IM7), CD62L (clone MEL-14), IL4 (clone 11B11), and CD8α (clone 53-6.7) were purchased from BioLegend, and CD4 (clone RM4-5) and IL17A (clone eBio17B7) were purchased from eBioscience. Gp100 tetramer (iTAg Tetramer/PE H-2D^b gp100), E7 tetramer (iTAg Tetramer/PE H-2D^b E7), and OVA tetramer (iTAg Tetramer/PE H-2K^b OVA) were purchased from MBL. For all experiments where murine blood was used, blood was collected retro-orbitally using VWR Microhematocrit Capillary Tubes into MiniCollect tubes (0.5 mL K2E K2EDTA, Greiner Bio-One GmbH) 6 days following boost unless otherwise noted. Antibody staining was performed at a dilution of 1:100 for 25 minutes at 4°C in the presence of mouse Fc block (Mouse BD Fc Block) in PBS with 1% BSA and 5 mmol/L EDTA unless otherwise indicated. Tetramer staining was performed in buffer containing 50 nmol/L dasatinib (Sigma-Aldrich) at room temperature for 30 minutes, followed by surface staining at 4°C for 15 minutes. Viability was assessed by LIVE/DEAD Fixable Aqua (Life Technologies) or DAPI. Cells were analyzed using BD LSR Fortessa and BD FACSCanto flow cytometers. Data was analyzed using FlowJo v10. ICS was performed as previously described (14), with fixation and permeabilization using BD Cytofix/Cytoperm according to the manufacturer's protocol. Peptides for restimulation were used at a concentration of 10 μg/mL with sequences as follows—for gp100 immunizations: gp100_{25-33 native}, EGSRNQDWL; SIV-gag AL11: AAVKNWMTQTL; OVA: SIINFEKL; HIV gag p17: RSLYNTVAT.

The human HLA-B57–restricted KF11-specific CD8⁺ T-cell clone (recognizing KAFSPEVIPMF; ref. 20) was maintained in the presence of IL2 (50 U/mL, R&D Systems), using CD3-specific monoclonal antibody 12F6 (0.1 μg/mL, Enzo Life Sciences), and irradiated allogeneic feeder peripheral blood mononuclear cells (PBMC). Human PBMCs were freshly isolated from HLA-B57⁺ blood donors with Ficoll-Hypaque (Sigma-Aldrich). PBMCs were isolated from buffy coats collected from HLA-typed blood donors after written informed consent and approval by the Partners Human Research Committee (protocol #2005P001218). Monocytes were enriched using CD14⁺ magnetic isolation, according to the manufacturer's instructions (EasySep Human Monocyte Isolation Kit, StemCell). DCs were differentiated during a 6-day culture at 10⁶ cells/mL in AIM-V medium (Thermo Fisher), supplemented with 1% HEPES, 1% human AB serum, IL4 (20 ng/mL; CellGenix), and GM-CSF (10 ng/mL; CellGenix). On days 2 and 4, fresh IL4 and GM-CSF were added. DCs were harvested on day 6 and incubated with free HIV p24_28-60 peptide (VEEKAFSPEVIPMFSALSEGATPQDLNTMLNTV) or DSPE-PEG-p24_28-60 for 4 hours at 37°C and then washed. The DCs were then plated with the KF11 CD8⁺ T-cell clone in a 2:1 effector-to-target ratio in 96-well plates (Millipore) coated with anti-IFNγ (clone 1-D1K, Mabtech, 2 μg/mL). DCs pulsed with equivalent molar concentrations of the optimal KF11 peptide were used as controls. Cells then were cocultured overnight wrapped in foil at 37°C in AIM-V medium supplemented with 1% HEPES and 10% human AB serum on human IFNγ ELISPOT plates (EMD Millipore). The ELISPOT plate was developed according to the manufacturer's protocol. Plates were scanned using a CTL ImmunoSpot Plate Reader, and data were analyzed using CTL ImmunoSpot Software. The magnitude of the epitope-specific response was reported as the number of spot-forming units (SFU) per 10⁴ T cells.

Mass spectrometry (MS) analysis of peptide fragments generated after incubation in cytosolic, endosomal, or lysosomal degradation conditions was carried out as previously described (21). Briefly, 2 nmol/L of p24_28-60 peptide or DSPE-PEG-p24_28-60 were incubated with cell extracts from each cellular compartment that were isolated from APCs derived from primary human monocytes at 37°C in 50 μL of degradation buffer (50 mmol/L Tris–HCl, 1 mmol/L MgCl₂, 1 mmol/L ATP, and 137 mmol/L potassium acetate at pH 7.4, pH 5.5, or pH 4.0). The reaction was stopped with formic acid, and peptides were purified via trichloroacetic acid precipitation. The degradation mix was injected into a Nano-HPLC (Eksigent) at a 400 nL/minute flow rate. A Nano cHiPLC trap column (200 μm × 0.5 mm ChromXP c18-CL 5 μm 120 Å; Eksigent) was used to remove salts in the sample buffer. Peptides were separated in a Nano cHiPLC column (75 μm × 15 cm ChromXP c18-CL 5 μm 300Å; Eksigent) with 0.1% formic acid in water and 0.1% formic acid in acetonitrile. Mass spectra were recorded in the range of 370 to 2,000 daltons (Da) on a LTQ Orbitrap Discovery (Thermo). In tandem MS mode, the eight most intense ions were selected with a 1 Da window for fragmentation in a collision cell using Helium with 35 V as collision voltage. MS fragments were searched against source peptide databases with Protein Discoverer 1.4 (Thermo Scientific). For semiquantification, the integrated area under a peak generated by a given peptide is proportional to the abundance of that peptide. Each degradation time point was run on the MS at least twice. The MHC class I and II epitopes from gag p24 were identified using the from HIV Los Alamos Immunology database (http://www.hiv.lanl.gov/).

Statistical methods were not used to determine sample size, but sample numbers were chosen based on estimates from pilot studies and published results, such that appropriate statistical tests may yield statistically significant results. Unless otherwise stated, analysis of tetramer and ICS results were analyzed by using one-way ANOVA with a Tukey posttest using GraphPad Prism software. Where ANOVA was used, variance between groups was found to be similar by Bartlett test. Where comparisons were predetermined, as indicated in figure legends, ANOVA with Bonferroni correction was used. Pmel-1 proliferation data were analyzed using Student t test within a given LN to compare EGP₂₀ peptide and DSPE-PEG-EGP₂₀. No samples were excluded from analysis.

We previously reported that peptide antigens conjugated to a DSPE phospholipid tail through a poly(ethylene glycol) spacer (Fig. 1A) exhibit increased lymphatic trafficking following immunization, and, thereby, a 20- to 30-fold increased immunogenicity in mice (14). This enhanced lymphatic trafficking was associated with the ability of the conjugates to bind serum albumin, suggesting that albumin serves as an in situ molecular chaperone delivering amph-vaccines to LNs. If this proposed mechanism of action is correct, other albumin-binding moieties might be equally well-suited to target peptide vaccines to lymphoid tissues. To test this idea, we prepared cyclized albumin-binding peptides linked via a PEG₂₀₀₀ spacer to the melanoma gp100 antigen EGP₂₀ (ABP-PEG-EGP₂₀; Fig. 1B and Supplementary Fig. S3; ref. 19). When mice were immunized with TAMRA-labeled EGP₂₀ peptide or ABP-PEG-EGP₂₀, the ABP-conjugated peptide resulted in 13.6-fold and 18.2-fold higher fluorescence in inguinal and axillary LNs 24 hours later, respectively, compared with unmodified EGP₂₀ (Fig. 1C and D). Vaccination with ABP-PEG-EGP₂₀ elicited a 5-fold increase in antigen-specific T cells compared with unmodified peptide (Fig. 1E).

As a second test of the generality of albumin binding as a mechanism for targeting antigen/adjuvants to LNs, we also evaluated vaccine conjugates of the most biologically active form of vitamin E, α-tocopherol, which has been shown to bind to serum albumin with an affinity of K_D ∼ 7 μmol/L (22). Constructs utilizing vitamin E as the albumin-binding domain for EGP₂₀ peptide or the TLR9 agonist CpG ODN 1826 were synthesized (Fig. 2A and B). Fluorescence imaging of excised LNs after immunization with fluorophore-tagged Vit. E-CpG showed 2.9-fold and 3.1-fold higher CpG levels in the draining inguinal and axillary nodes, respectively, compared with unmodified oligonucleotides (Fig. 2C). Consistent with this enhanced lymphatic delivery, mice immunized with whole protein ovalbumin (OVA) adjuvanted with Vit. E-CpG showed 10-fold higher frequencies of OVA-specific CD8⁺ T cells compared with vaccination with unmodified CpG or CpG conjugated with vitamin B3, which is not known to bind to serum albumin (Fig. 2D and E). Vit. E-CpG performed comparably with the diacyl lipid-CpG conjugate lipo-CpG, which was previously demonstrated to exhibit enhanced adjuvanticity by virtue of its improved LN trafficking (14). To see if α-tocopherol offered similar enhancements to peptide antigens, α-tocopherol-PEG-maleimide was used to conjugate vitamin E-PEG to the gp100 antigen EGP₂₀ long via an N-terminal cysteine (Vit. E-PEG-EGP₂₀; Fig. 2B). Mice immunized with Vit. E-PEG-EGP₂₀ showed T-cell responses approximately 10-fold higher than unmodified peptides (Fig. 2F and G). Vitamin E conjugation, thus, improves the adjuvant activity of CpG and immunogenicity of peptide antigens, and together with the ABP conjugate data, suggests that directing molecular vaccines to bind albumin may be a general design principle to increase LN accumulation and immunogenicity.

We next explored factors controlling the fate of amph-vaccines in LNs. We previously showed that amph-vaccines accumulate in F4/80⁺ macrophages and CD11c⁺ DCs (14), but it is unclear how albumin-binding vaccines are acquired by LN APCs. The neonatal Fc receptor (FcRn) is the best characterized receptor for albumin, and studies have implicated FcRn in the cross-presentation of immune complexes (23, 24). However, immunization of WT or FcRn^−/− animals with lipo-CpG and DSPE-PEG-E7 peptides showed no role for FcRn, as FcRn^−/− mice generated T-cell responses indistinguishable from WT mice (Fig. 3A and B).

Although amph-vaccines were found to accumulate in many APC populations in LNs, we expected that any DSPE-PEG peptide conjugate (even optimal nonamer epitopes) may require processing to cleave the DSPE-PEG linkage for antigen presentation, and, thus, might require cross-presenting DCs for activity. To test this hypothesis, we immunized Batf3^−/− mice, which lack the transcription factor necessary to develop cross-presenting DCs (25). Compared with WT mice, priming of T-cell responses to the optimal EGP 9-mer peptide (DSPE-PEG-EGP₉) were severely compromised in Batf3^−/− mice (Fig. 3C and D), indicating that even “short” DSPE-PEG peptides may be reliant upon cross-presenting DCs for effective priming.

We previously focused on forming amphiphile conjugates by linking DSPE-PEG-maleimide to the free thiol of an N-terminal cysteine introduced on peptide antigens. However, this conjugation chemistry would be problematic if an antigen sequence of interest had an internal cysteine. To determine whether the immunogenicity of amph-vaccines is sensitive to the linkage chemistry, conjugates were synthesized where DSPE-PEG was either linked using cysteine-maleimide chemistry, copper-free click chemistry, or via amine-succinimidyl ester chemistry to form an amide bond directly to the N-terminus of the peptide antigen (Supplementary Fig. S1). All of these conjugate-peptides demonstrated superior immunogenicity relative to unmodified peptides, and conjugates formed with the different linkages were not statistically different from another (Fig. 3E), suggesting that immunogenicity of peptide constructs is insensitive to the PEG-lipid linkage. When mice were immunized with either N- or C-terminal DSPE-PEG or ABP-PEG antigen conjugates, both showed a trend toward N-terminal conjugation being superior to C-terminal conjugation (Fig. 3F).

PEG conjugation has been shown to enhance the serum stability of peptides by protecting from serum proteases (12, 26). To test whether amph-peptides show a similar increase in resistance to proteolytic degradation, unmodified EGP₂₀ peptide, DSPE-PEG linked to EGP₂₀ either through an N-terminal or C-terminal cysteine, or PEG-EGP₂₀ without the diacyl lipid (linked N-terminally) were incubated with fresh mouse serum. Serum-treated or fresh peptides were then added to splenocytes from gp100-primed mice, and the IFNγ production of the stimulated cells was quantified by ICS. As shown in Fig. 4A, the percentage of IFNγ⁺ CD8⁺ T cells in cultures stimulated by unmodified EGP₂₀ peptide (normalized to the maximal response of the untreated antigen) showed a substantial drop after incubation of the peptide overnight with serum. By contrast, each of the PEG and DSPE-PEG peptide conjugates retained nearly maximal stimulatory activity after the same serum treatment (Fig. 4B–D).

To evaluate the relative contribution of enhanced serum stability versus lymphatic trafficking in the immunogenicity of amph-vaccines, mice were vaccinated with DSPE-PEG-EGP₂₀ or the identical construct lacking the diacyl lipid (PEG-EGP₂₀). PEG-EGP₂₀ retained approximately 40% of the activity of the albumin-binding DSPE-PEG-EGP₂₀ (Fig. 4E), indicating that protection from serum proteases is an important but not dominant contributing mechanism in amph-vaccine immunogenicity. Given the observation that albumin-binding conjugates both trafficked to LNs more effectively and showed enhanced stability in vivo, we directly assessed whether any dose of a soluble long peptide vaccine could elicit T-cell responses comparable with the equivalent amph-vaccine. With EGP₂₀ antigen, a 20× higher dose of unmodified antigen was required to elicit comparable priming (Fig. 4F). We also carried out comparative immunizations of free peptides and amphiphile conjugates of four additional “long” and optimal 9-mer peptide antigens, including both CD4 and CD8 epitopes, to assess the generality of the increased immunogenicity of amph-vaccines, and found consistent increases in T-cell responses elicited by each of these antigens (Supplementary Fig. S4). These data suggest that amph-vaccines promote an increase in vaccine potency that would be difficult to achieve in humans simply by increased free peptide dosing.

Enhancement of presentation of an antigen in vivo may result from a number of distinct processes: quantitatively more peptide can be taken up in a given LN, presentation can occur in more nodes throughout the lymphatic chain, or presentation might occur for an extended duration of time. To define more completely the location and longevity of peptide presentation, we utilized pmel-1 TCR-transgenic CD8⁺ T cells, which recognize EGP₉ in the context of H-2D^b, as reporters of antigen presentation. Mice were immunized with DSPE-PEG-EGP₂₀ plus adjuvant, EGP₂₀ peptide plus adjuvant, or adjuvant alone, and CFSE-labeled pmel-1 CD8⁺ T cells were transferred intravenously 24 hours later, followed by transfer of CellTrace Violet-labeled pmel-1 CD8⁺ T cells an additional 24 hours after that (Fig. 5A and B). Under these conditions, the CFSE-labeled pmel-1 T cells encountering antigen will proliferate in the node in which they encounter antigen and serve as an effective proxy for antigen presence after immunization. The violet-labeled pmel-1 T cells (which do not have sufficient time to divide) upregulate CD69 in response to antigen encounter and inform about antigen levels 48 hours after immunization. Substantially higher proliferation of CFSE-labeled reporter cells was seen with DSPE-PEG-EGP₂₀ compared with EGP₂₀ peptide in all nodes examined, except the direct draining inguinal node (Fig. 5C and D). Similarly, CD69 was upregulated among pmel-1 T cells transferred in 48 hours after immunization to a much higher extent in all nodes with DSPE-PEG-EGP₂₀ compared with unconjugated peptide (Fig. 5E and F), and presentation of DSPE-PEG-EGP₂₀ was seen even in the distal mesenteric nodes. Next, mice were immunized and CFSE-labeled pmel-1 T cells were transferred 7 days later. In LNs that were resected 48 hours after transfer, no proliferation in any node was seen with unmodified EGP₂₀ peptide, but DSPE-PEG-EGP₂₀ resulted in proliferation in the inguinal, axillary, and lumbar nodes, indicating that DSPE-PEG peptide is presented for at least one week in vivo (Fig. 5G and H). Thus, DSPE-PEG peptides achieved broad lymphatic distribution and extended in vivo presentation.

To assess the phenotype of T cells primed by this sustained and systemic antigen presentation, we measured surface markers and cytokine production capabilities of antigen-specific peripheral blood T cells. At 6 days post boost, we observed that approximately 90% of antigen-specific cells primed expressed markers consistent with an effector memory phenotype (CD44^highCD62L^low), with <3% PD-1⁺Tim3⁺ exhausted cells (Supplementary Fig. S5A and S5B). We observed that with the adjuvants used here, the CD8⁺ T-cell response exhibited a Tc1 rather than a Tc2 or Tc17 phenotype, as determined by production of IFNγ but not IL4 or IL17 within tetramer⁺CD8⁺ T cells (Supplementary Fig. S5C).

Although DSPE-PEG peptides showed clear superiority over unmodified antigens when administered with molecular adjuvants, peptide vaccines are often clinically administered in combination with IFA, a water-in-oil emulsion that induces local inflammation and forms a depot, protecting antigens from degradation and releasing them over time (27). In mouse models, peptides in IFA alone have been shown to create a sink for effector T cells, but addition of TLR agonists to peptides in IFA elicits potent immune responses (10). We, thus, sought to examine the LN trafficking of peptides in IFA vs. DSPE-PEG-conjugates and to compare the immunogenicity of IFA versus molecular adjuvants for peptides or DSPE-PEG peptides. First, FAM-labeled DSPE-PEG-EGP₂₀ was formulated with lipo-CpG, c-di-GMP (STING agonist), polyI:C (TLR3 agonist), or IFA, and injected subcutaneously. Drainage to the inguinal and axillary nodes was quantified by IVIS (Fig. 6A–C). Formulation of antigen with lipo-CpG, c-di-GMP, or polyI:C did not substantially alter LN accumulation, but formulation of DSPE-PEG peptide in IFA prevented efficient lymphatic drainage. To evaluate the impact on vaccine immunogenicity, mice were primed and boosted with EGP₂₀ or DSPE-PEG-EGP₂₀ adjuvanted with c-di-GMP or polyI:C, either injected in saline or formulated in IFA, and responses were quantified with ICS (Fig. 6D). Responses to free peptides adjuvanted with either polyI:C or CDN trended toward improvement with IFA, but responses to DSPE-PEG peptide were hindered by formulation in IFA, decreasing approximately 50% in magnitude compared with formulation in saline. These data suggest DSPE-PEG peptides exhibit superior immunogenicity compared with unmodified peptides when combined with a number of molecular adjuvants, including the most potent clinically used adjuvant combinations.

To assess if processing and presentation of amph-peptides is performed efficiently by human APCs in vitro, we performed an IFNγ ELISPOT with primary human monocyte-derived DCs. The KF11 CD8⁺ T-cell clone (recognizing HIV gag p24_30-40 in the context of HLA-B57; ref. 20) was used to probe for presentation in a coculture. DCs generated from an HLA-B57⁺ donor were pulsed with the HIV-derived gag p24_28-60 peptide or DSPE-PEG-p24_28-60 at three concentrations and cocultured with KF11-specific CD8⁺ T cells, and IFNγ SFU were quantified. As shown in Fig. 7A, similar responses were observed for DSPE-PEG-p24 and p24 peptide, indicating that the functional integration of uptake, processing, and presentation of DSPE-PEG peptide is equivalent to unmodified peptide in vitro for an immunodominant MHC class I–restricted epitope.

The repertoire of peptides generated by antigen processing in APCs is one of many factors that influence T-cell responses (28). To uncover if conjugation of peptide antigens with DSPE-PEG alters the cleavage patterns and resultant peptides generated, HIV gag p24_28-60 peptide or DSPE-PEG-p24_28-60 were subjected to degradation in cellular extracts from endosomal, lysosomal, or cytosolic compartments derived from human monocyte-derived DCs and analyzed by MS as previously described (21). The MHC class I and II epitopes and the lengths of all fragments generated by degradation of p24_28-60 peptide and DSPE-PEG-p24_28-60 were virtually identical at both 60 and 120 minutes in each of the cellular compartments analyzed (Fig. 7B and C; Supplementary Table S1), suggesting that conjugation with DSPE-PEG does not interfere with antigen-processing machinery by human APCs.

We have previously shown that conjugation of CpG oligonucleotide adjuvants or peptide antigens with diacyl lipids (through a PEG spacer in the peptide case) significantly enhances LN accumulation of these vaccine agents, and, thereby, augments T-cell priming (14). Here, we extended those results to show that substitution of the diacyl lipid with α-tocopherol as the albumin-binding moiety similarly enhanced both antigen and adjuvant activity. We also showed that nonlipid modifications, namely, fusion of antigens with albumin-binding peptides, also allowed for enhancement of LN delivery and T-cell priming. The former result was noteworthy, given that α-tocopherol has a low affinity for albumin (K_D ∼ 7 μmol/L; ref. 22). Effective lymphatic delivery, even with low-affinity binders, is likely facilitated by the very high concentrations of albumin in interstitial fluid (29). Our data suggest that amphiphilic peptides utilizing lipid albumin-binding domains, DSPE-PEG-EGP20 and Vit. E-PEG-EGP20, are superior to compositions incorporating an ABP. However, we cannot rule out that further optimization of the affinity of these ABPs for albumin or their linkages to the peptide antigens might further increase activity of ABP-PEG peptide conjugates (19).

Although our prior work provided a reasonable basis for understanding how albumin could mediate vaccine trafficking, the mechanisms underlying antigen capture and processing in the LN remained unclear. The best characterized receptor for albumin, FcRn, does not play a role in the immune response to these albumin-binding conjugates, as FcRn^−/− and WT mice responded to amph-vaccines similarly. Although other receptors for albumin have been described, including albondin (gp60) and several scavenger receptors that bind conformationally altered albumin to redirect it for degradation, these other receptors are not known to be expressed on APCs (30). Albumin is readily internalized by macropinocytosis (31), and this may represent a main mode of uptake of albumin-bound vaccines.

Conjugation of optimal MHC I epitopes with DSPE-PEG has enhanced activity for a number of antigens (14), but it was unclear if this linkage changes the antigen presentation requirements of short peptides. Namely, short soluble peptides may undergo peptide exchange, loading directly onto MHC on the surface of nonprofessional APCs, including T cells and B cells, which may promote tolerance in some circumstances (32–34). By contrast, DSPE-PEG peptides appear to rely on cross-presenting DCs, even for optimal nonamer antigens, which may allow optimal epitopes to be used in vaccination without concern for presentation by nonprofessional APCs.

The duration of antigen presentation has been shown to impact the magnitude of T-cell effector and memory responses, with longer antigen presentation resulting in more effective CD8⁺ T-cell memory formation compared with brief exposures (35, 36), and priming in distal nodes was identified as one of the characteristics of DEC-205–targeted antigens, which promote potent CD8⁺ T-cell immunity (37). Performing vaccination at multiple injection sites in order to increase the number of LNs utilized for priming has also been shown to enhance the magnitude of T-cell responses (38). We showed here that DSPE-PEG peptides embodied these important principles, as they are presented in multiple nodes systemically and for at least one week in vivo. We observed that with the adjuvants used here, the T-cell response exhibited a Tc1 phenotype, and we did not see evidence of exhaustion by PD-1 and Tim-3 staining. These results suggest that immunization with DSPE-PEG peptides recapitulates some traits of systemic viral infection: presentation of antigen in many lymphoid organs for a prolonged duration.

Peptide antigen modification with PEG significantly enhanced T-cell priming, even without including an albumin-binding domain. This was likely mediated through an increase in serum stability and in vivo half-life, as has been demonstrated for PEG modification of erythropoiesis stimulating peptides and antiviral peptides previously (12, 26). This finding indicates that some of the activity of DSPE-PEG peptides derives from protection of the antigen from serum proteases, although further significant enhancements in immunogenicity are seen with the addition of an albumin-binding domain.

IFA is widely used to enhance immunogenicity of unmodified peptides, but our data suggest that vaccines with lipophilic domains that rely on efficient lymphatic drainage for their mechanism of action may be hindered in activity with formulation in IFA. The mechanism of action of IFA is thought to involve extending in vivo presentation of antigen by forming an antigen depot (27). This may be unnecessary in the case of DSPE-PEG peptides, which showed an extended in vivo half-life. Although IFA increased priming to free peptides, IFA decreased the T-cell response approximately 2-fold for DSPE-PEG peptides, indicating that IFA should be avoided for translation of DSPE-PEG peptides clinically. This would have the dual benefit of avoiding T-cell sequestration and induction of apoptosis in IFA depots, as has been observed in mice (10).

Detailed MS analysis of fragments produced by degradation in cellular extracts from primary human APCs demonstrated that DSPE-PEG conjugation did not alter the repertoire of class I or II epitopes generated or the overall characteristic lengths of peptide fragments in the pool. Functional antigen presentation assays using HIV gag p24 specific CD8⁺ T cells confirmed that presentation of the HLA-B57–restricted KF11 epitope was equivalent in vitro. A lack of enhancement of presentation in vitro with DSPE-PEG peptide over unmodified peptide is consistent with the hypothesized mechanism of action of amph-peptides. Enhancements in trafficking and in vivo serum stability did not play an appreciable role in the in vitro ELISPOT.

In summary, simple covalent modification of molecular vaccines with albumin-binding domains was capable of ferrying these molecules throughout the lymphatics, enhancing serum stability and duration of presentation in vivo and increasing T-cell responses by an order of magnitude over unmodified vaccines and may be of interest for promoting potent antitumor CD8⁺ T cells clinically.

K.D. Moynihan, R.L. Holden, B.L. Pentelute, and D.J. Irvine are co-inventors on a pending patent application based on the concept of antigens conjugated to albumin binding peptides. D.J. Irvine has ownership interest in and is a consultant/advisory board member for Vedantra Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of General Medical Sciences or the National Institutes of Health.

Conception and design: K.D. Moynihan, R.L. Holden, B.L. Pentelute, D.J. Irvine

Development of methodology: K.D. Moynihan, B.L. Pentelute

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.D. Moynihan, R.L. Holden, N.K. Mehta, C. Wang, J. Dinter, S. Liang, W. Abraham, M.B. Melo, A.Q. Zhang, N. Li

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.D. Moynihan, N.K. Mehta, C. Wang, J. Dinter, M.B. Melo, A.Q. Zhang, S. Le Gall, D.J. Irvine

Writing, review, and/or revision of the manuscript: K.D. Moynihan, R.L. Holden, N.K. Mehta, M.R. Karver, D.J. Irvine

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.R. Karver, S. Liang, B.L. Pentelute

Study supervision: K.D. Moynihan, S. Le Gall, B.L. Pentelute, D.J. Irvine

K.D. Moynihan is supported by the Fannie & John Hertz Foundation Fellowship, the NSF Graduate Research Fellowships, and the Siebel Scholarship. D.J. Irvine is an investigator of the Howard Hughes Medical Institute.

This work was supported in part by the Ragon Institute of MGH, MIT, and Harvard, the Mayo Clinic-Koch Institute Collaboration, the Koch Institute Support (core) grant P30-CA14051 from the National Cancer Institute, the Marble Center for Cancer Nanomedicine, the Bridge Project partnership between the Koch Institute for Integrative Cancer Research and the Dana Farber/Harvard Cancer Center the V Foundation, and the NIH (awards EB022433, CA174795, CA206218, and CA172164). The project described was supported by award Number T32GM007753 from the National Institute of General Medical Sciences.

We thank the Koch Institute Swanson Biotechnology Center for technical support, specifically the applied therapeutics and whole-animal imaging facility, the histology facility, and the flow cytometry facility. We thank Mark Karver and the Peptide Synthesis Core Facility of the Simpson Querrey Institute at Northwestern University for peptide synthesis. The U.S. Army Research Office, the U.S. Army Medical Research and Materiel Command, and Northwestern University provided funding to develop this core facility, and ongoing support is being received from the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource (NSF ECCS-1542205).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.