Adjuvant formulation structure and composition are critical for the development of an effective vaccine against tuberculosis

2.1. Adjuvant manufacturing

Glucopyranosyl lipid adjuvant (GLA), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1,2-Dipalmitoyl-sn-glycero-3-phosphoglycerol (DPPG), 1,2-dipalmitoyl-3-trimethylammonium-propane (DPTAP), 1,2 dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(succinyl) (succinyl-functionalized DPPE), 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), and egg yolk phosphatidylcholine (Egg PC) were purchased from Avanti Polar Lipids Inc. (Alabaster, AL). DPPG was also purchased from Corden Pharma (Liestal, Switzerland). Squalene was purchased from Sigma-Aldrich (St. Louis, MO). Grapeseed oil by Napa Valley Naturals (San Francisco, CA) was obtained from a local grocer. Miglyol 810 was obtained courtesy of Sasol (Johannesburg, South Africa). Poloxamer 188 and glycerol were purchased from Spectrum Chemical (Gardena, CA). Cholesterol, ammonium phosphate mono and dibasic were purchased from J.T. Baker (San Francisco, CA). Phosphate buffered saline (PBS) of 1 × concentration at pH 7.2 was purchased from Invitrogen (Grand Island, NY). Alhydrogel® 85 (aluminum oxyhydroxide gel) was purchased from E.M. Sergeant Pulp & Chemical Co. (Kalamazoo, MI). Cross linking reagents Sulfo-NHS N-Hydroxyssulfosuccinimide and EDC 1-Ethyl-3-(3 dimethylaminopropyl) carbodimide HCL were both purchased from Thermo Scientific.

Aqueous GLA was manufactured by first combining DPPC and GLA (2:1 molar ratio of DPPC:GLA) in minimal amount of chloroform needed to dissolve all solutes. The GLA/DPPC/Chloroform mixture was then evaporated using a Buchi Rotovapor R-114 (Flawil, Switzerland) or Genevac EZ-2 centrifugal evaporator (Stone Ridge, NY) to remove the chloroform. The dried DPPC and GLA were then rehydrated in ultrapure water at a concentration of 0.25 mg/ml GLA and 0.21 mg/ml DPPC, and sonicated at ~60 °C with either a VWR 75D (West Chester, PA) or Crest Powersonic CP230D (Trenton, NJ) waterbath sonicator for approximately 1–2 h or until it appeared to be a single-phase, translucent formulation. Aqueous GLA was diluted 5× with antigen and buffer prior to immunization studies. Aluminum oxyhydroxide formulation with GLA was prepared by mixing aqueous GLA with Alhydrogel® 85 for a GLA concentration of 0.1 mg/ml and an aluminum concentration of 2 mg/ml. The alum formulations were diluted 2× with antigen and buffer prior to immunization.

GLA-SE emulsions were manufactured by mixing a buffered aqueous phase and an oil phase with a Silverson Heavy Duty Laboratory Mixer Emulsifier (3/4 in. tubular square hole high shear screen attachment; East Longmeadow, MA) at ~7000–10,000 rpm for ~10 min, then microfluidizing the mixture using the Microfluidics M110P (Newton, MA) for 12 passes at 30,000 psi. The buffered aqueous phase was prepared by combining poloxamer 188 and glycerol in ammonium phosphate buffer (pH 5.1). The oil phase was prepared by dispersing DMPC or Egg PC and GLA into squalene, grapeseed oil, or Miglyol 810. The oil phase was then sonicated at ~70 °C for 1–3 h or until evenly dispersed. Component concentrations in the emulsions consisted of 10% v/v oil, 1.9% w/v phosphatidylcholine, 0.1% w/v poloxamer 188, 2.3% w/v glycerol, and 25 mM ammonium phosphate buffer. Prior to mixing with antigen for immunization, the formulations were diluted 5× with phosphate buffered saline and antigen.

GLA-liposome formulations were manufactured by first combining cholesterol and DPPC with either DPPG (anionic), DPTAP (cationic), or succinyl-functionalized DPPE (covalent) and with or without GLA in minimal amount of chloroform or chloroform:methanol:water, which was then evaporated off. The dried components were rehydrated in PBS, then sonicated at ~60 °C until all the lipids were suspended in the buffer at a concentration of 18 mg/ml DPPC, 5 mg/ml cholesterol, and 2 mg/ml DPPG, DPTAP, or succinyl-functionalized DPPE. For small volume batches (<20 ml), the liposomes were sonicated until they appeared to be single phase, translucent formulation. Larger batch sizes (100 ml) were prepared by transferring the crude suspension to the Microfluidics M110P (Newton, MA) for high-pressure homogenization for ~12 passes at 10,000 psi. Post sonication, the covalent liposomes were crosslinked to ID93 with the aid of two reagents, Sulfo-NHS N-Hydroxyssulfosuccinimide prepared in phosphate saline at a final concentration of 1 mM in liposome solution and EDC 1-Ethyl-3-(3 dimethylaminopropyl) carbodiimde HCL prepared in 50 mM MES at a final concentration of 0.5 mM in solution. Mixture of both intermediary liposome solutions and ID93 were allowed to incubate for 2 h at room temperature. Finally, any unbound ID93 was separated from the resulting ID93-bound covalent liposomes by passing it through an SEC column manually packed with 20 ml Sepharose™ CL-4B, an agarose-based gel filtration matrix obtained from GE Healthcare Bio-Sciences AB (Uppsala, Sweden). The liposome formulations were diluted 5× prior to immunization studies. All formulations were stored at 5 °C.

2.2. Adjuvant characterization

With the exception of Alum, GLA-Alum, and covalent liposomes with and without GLA, all other formulations were monitored by HPLC with charged aerosol detection (CAD) to determine GLA concentration and dynamic light scattering (DLS) to determine particle size. Particle size was determined using the Malvern Instruments (Worcestershire, UK) Zetasizer Nano-S or Nano-ZS. For oil-in water emulsions and liposome formulations, samples was prepared at 1:100 dilution by combining 5 μl of each formulation with 500 μl of ultrapure water in a 1.5 ml polystyrene disposable cuvette. For aqueous GLA, a 1:10 dilution was accomplished by combining 50 μl of formulation with 450 μl ultrapure water in a 1.5 ml polystyrene disposable cuvette. In general, three separate cuvettes were prepared. All DLS measurements were then made three times with each of the three separate cuvettes. Zeta potential measurements were also obtained using the Malvern Instruments (Worcestershire, UK) Zetasizer Nano-ZS. Samples were prepared for a 1:20 dilution by combining 50 μl of each formulation with 950 μl of ultrapure water in a 1.5 ml eppitube before being transferred to a disposable capillary cell for analysis.

For HPLC-CAD, samples were prepared for analysis in a 1.5 ml capacity HPLC glass vials obtained from Agilent Technologies (Santa Clara, CA), by combining 50 μl of formulation with 950 μl mobile phase B (1:1 [v:v] methanol:chloroform, 20 mM ammonium acetate, 1% acetic acid) for GLA-emulsion and GLA-liposome samples or mobile phase A (75:15:10 [v:v:v] methanol:chloroform:water, 20 mM ammonium acetate, 1% acetic acid) for aqueous GLA samples. Three separate vials were prepared for each formulation. Samples were then placed on que for injection into a Waters Co. (Milford, MA) Atlantis T3 column attached to an Agilent Model 1100 HPLC (Santa Clara, CA). A gradient consisting of mobile phases A and B was used spanning a 25-minute time period. Detection was performed by a Charged Aerosol Detector obtained from ESA Biosciences (Chelmosford, MA). An 11-point nonlinear quadratic standard curve fit was obtained of GLA standards ranging from 1 to 100 μg/ml GLA.

2.3. Adjuvant–antigen association

With the exception of covalent liposome formulations which already incorporated this step in the manufacturing process, ID93 association with all other formulation platforms was evaluated by incubating 0.1 mg/mL of ID93 with each formulation and PBS for 30 min in 4 ml capacity BD Falcon tubes obtained from BD Biosciences (San Jose, CA) and placed in a beaker filled with ice to mimic post-mixing/pre-immunization conditions. Each ID93-formulation mixture was then introduced into a size exclusion gel chromatography column packed with Sepharose CL-4B and eluted with PBS. Eluted fractions were collected from ID93 control (no adjuvant) and ID93-adjuvant mixtures. Fractions eluting between 11 and 16 ml were pooled as representative of unassociated ID93. A 20 μl sample from each of these pooled fractions was then combined in an eppitube with 20 μl LDS Sample Buffer (4×) and 40 μl of 20% SDS solution obtained from Invitrogen (Grand Island, NY). The sample mixture was then very briefly vortexed, heated at 85 °C for 5 min, and assayed by SDS-PAGE, followed by transfer to a PVDF membrane and finally stained with Colloidal Gold Total Protein Stain from BioRad (Hercules, CA). Stained membranes were scanned on a Konica Minolta bizhub C552 printer/scanner. Pixel intensity histograms of the ID93 bands were collected using ImageJ (NIH), and the area of the ID93 band peak was integrated using a baseline approximation of the background pixel intensity gradient and normalized to the observed intensity of the ID93 control in each membrane in GRAMS/AI (Galactic Software). The association percentage was calculated as a ratio of band intensity of eluted ID93 from adjuvant formulation mixtures compared to ID93 control.

2.4. Animals and immunizations

6–8-week-old female C57Bl/6 mice were purchased from Charles River and maintained in Specific Pathogen Free conditions. After infection the animals were maintained in BL3 containment according to the regulations and guidelines of the IDRI Institutional Animal Care and Use Committee. Mice were immunized three times three weeks apart by intramuscular injection. Each immunization contained 0.5 μg of ID93 recombinant protein (either by admixing ID93 to the preformed adjuvant or for the covalent liposome including sufficient ID93-liposome to obtain ~0.5 μg of ID93) with or without 5 μg of GLA. For BCG immunization 5 × 10⁴ CFU (Pasteur strain, Sanofi Pasteur) were injected intradermally once at the time of the first subunit immunization.

2.5. Antibody titers

ID93-specific endpoint titers for IgG1 and IgG2c were determined one week after the final immunization. Nunc Polysorp plates were coated with ID93 (2 μg/ml) in 0.1 M bicarbonate and blocked overnight at 4 °C with 0.05% PBS-Tween 20 + 1% BSA. Serial dilutions of mouse serum were incubated for 2 h at room temperature, washed and incubated for 1 h with anti-mouse IgG1 or IgG2c conjugated to streptavidin (VWR). Plates were washed and developed using SureBlue tetramethylbenzidine substrate (Kirkegaard & Perry Laboratories). The enzymatic reaction was stopped with 1N H2SO4. Plates were read within 30 min at 450 nm with a reference filter set at 650 nm using a microplate ELISA reader (Molecular Devices) and Soft Max Pro5 software with a cutoff of 0.1 absorbance unit.

2.6. ELISpots

Four weeks after the final immunization splenocytes were isolated from three animals per group. Red blood cells were lysed using Red Blood Cell Lysis Buffer (eBioscience) and resuspended in RPMI 1640 and 10% FBS. A MultiScreen 96-well filtration plate (Millipore) was coated with 10 μg/ml rat anti-mouse IL-5 capture antibody (eBioscience) and incubated overnight at 4 °C. Plates were washed with PBS, blocked with RPMI 1640 and 10% FBS for at least 1 h at room temperature, and washed again. Splenocytes were plated in triplicate at 2 × 10⁵ cells/well and stimulated with media or ID93 (10 μg/ml) for 48 h at 37 °C. The plates were then washed with 0.1% PBS-Tween 20 and incubated overnight with a biotin-conjugated rat anti-mouse IL-5 secondary Ab (eBioscience) diluted 1:250 in 0.1% PBS-Tween 20/0.5% BSA. The filters were developed using the VectaStain ABC avidin peroxidase conjugate and Vectastain AEC substrate kits (Vector Laboratories, Burlingame, CA) according to the manufacturer’s protocol. The reaction was stopped by washing the plates with deionized water. Plates were dried in the dark, and spots were counted on an automated ELISPOT reader (C.T.L. Series 3A Analyzer; Cellular Technology) and analyzed with ImmunSpot software (CTL Analyzer).

2.7. Intracellular cytokine staining

One to four weeks after the final immunization splenocytes were isolated from three animals per group. Red blood cells were lysed using Red Blood Cell Lysis Buffer (eBioscience) and resuspended in RPMI 1640 and 10% FBS. To isolate lymphocytes from the lungs immunized mice were perfused with PBS. Lungs were digested for one hour with Liberase TM (Roche) and dissociated using a GentleMACS system (Miltenyi). Cells were plated at 2 × 10⁶ cells/well in 96-well plates and were stimulated for 1 h with media or ID93 (10 μg/mL) at 37 °C. GolgiPlug (BD Biosciences) was added and the cells were incubated for an additional 7 h at 37 °C. Cells were washed with PBS and stained with LIVE/DEAD Fixable Stain (Invitrogen) for 30 min at 4 °C. Cells were washed and surface stained with fluorochrome labeled antibodies to CD4 (clone GK1.5) and CD8 (clone 53-6. 7) (BioLegend and eBioscience) in the presence of 20% normal mouse serum for 20 min at 4 °C. Cells were washed and permeabilized with Cytofix/Cytoperm (BD Biosciences) for 20 min at room temperature. Cells were washed twice with Perm/Wash (BD Biosciences) and stained intracellularly with fluorochrome labeled antibodies to IFN-γ (clone XMG-1.2), IL-2 (JES6-5H4), TNF (MP6-XT22), and IL-17A (clone TC11-18H10.1) BioLegend and eBioscience) for 20 min at room temperature. Cells were washed and resuspended in PBS. Up to 10⁶ events were collected on a four laser LSRFortessa flow cytometer (BD Biosciences). Data were analyzed with FlowJo. Cells were gated as singlets > live > lymphocytes > CD4 + CD8− > cytokine positive. Analysis and presentation of distributions were performed using SPICE version 5.2, downloaded from <http://exon.niaid.nih.gov/spice.

2.8. MHC class II tetramer production and staining

ID93-specific I-A^b tetramers with the immunodominant epitope from Rv3619 (VIYEQANAHGQ) were produced using methods previously described [26,27]. One week after the final immunization splenocytes were stained for one hour at room temperature with 10 μM tetramer. Cells were washed and stained for surface CD4, CD8, CD44 (clone IM7), Ly6C (clone HK1.4), PD-1 (clone 29F.1A12), KLRG1 (clone 2F1), CD27 (clone LG.3A10), and CD62L (clone MEL14) (BioLegend and eBioscience). Cells were then washed, fixed, permeabilized and stained for T-bet (clone 4B10) using the FoxP3 Fixation/Permeabilization buffer and protocol (eBioscience).

2.9. Mtb aerosol challenge and enumeration

Four weeks after the last immunization, mice (n = 7/group) were aerogenically infected with M. tuberculosis H37Rv (ATCC No. 35718; American Type Culture Collection) using a GlasCol aerosol generator calibrated to deliver 50–100 bacteria into the lungs. To confirm the amount of bacteria delivered an additional three unimmunized animals per infection were euthanized one day later and bacterial burden in the lungs were enumerated. Protection was determined three to four weeks after challenge by harvesting the lungs and spleens from the infected mice, homogenizing the tissue in 0.1% PBS-Tween 80, and plating 5-fold serial dilutions on7H10 agar plates (Molecular Toxicology) for bacterial growth. Bacterial colonies were counted after incubation at 37 °C with 5% CO₂ for 14–21 days.

2.10. Statistical methods

Bacterial burdens were normalized by log₁₀ transformation. Statistical significance of differences in ELISpot values and bacterial burden were determined using one-way analysis of variance. Differences relative to ID93 immunized animals were calculated using Dunnett’s post-test using Prism 5 (GraphPad Software). Differences between immunized animals were calculated with the Bonferroni Multiple Comparison Test using Prism 5 (GraphPad Software).

3.1. Physicochemical characteristics of adjuvant formulations

To explore the formulation space for effective ID93/GLA vaccines we developed multiple adjuvant formulations based on distinct platforms in order to produce a library of formulations with different size, charge, and compositional properties (Table 1 and Fig. S1). We produced oil-in-water stable emulsions, liposomes, and aluminum oxyhydroxide (Alum)-based formulations. Oil-in-water emulsions are comprised of homogeneous ~75 to 85 nm average diameter droplets consisting of an oily core emulsified by phospholipid and nonionic block copolymer. The oily core was comprised of squalene, long chain triglyceride (grapeseed oil), or medium chain triglyceride (Miglyol 810), although the emulsifiers remained constant for each type of oil. We also produced anionic and cationic liposome formulations depending on whether DPPG or DPTAP was included. Alternatively, we produced liposomes covalently linked to ID93 through a succinyl-functionalized DPPE. The liposomes demonstrated average diameter in a similar range as the emulsions, that is from ~76 to 93 nm. Another lipid-based formulation consisting of an aqueous nanosuspension of GLA with DPPC was manufactured at 90 nm and termed aqueous GLA. This same suspension was completely adsorbed to aluminum oxyhydroxide particles to create GLA-Alum, a formulation with a heterogeneous size distribution of large crystalline particles with a highly positive surface charge and ranging from ~1 to 10 μm in size according to the literature [28]. In general, emulsion, liposome, and aqueous formulation platforms demonstrated little change in particle size or GLA concentration over several months at 5 °C, although we note that some particle precipitation was visually apparent over time in one cationic liposome batch and one aqueous batch (data not shown).

Due to the wide variation in charge, size, and composition of the formulations developed, we examined the interaction of each formulation with ID93 to determine the extent of association (Table 1). Each formulation was mixed with ID93 followed by size exclusion chromatography to separate ID93-associated adjuvant particles from unassociated ID93. The eluted fraction volume representing unassociated ID93 was collected and assayed by SDS-PAGE with gold stain and compared to ID93 control (not containing any adjuvant) to estimate the amount of ID93 unassociated with each adjuvant formulation. As expected, the anionic ID93 (pI ~4.7) adsorbed completely to Alum and GLA-Alum after simple mixing. The cationic liposomes (with or without GLA) also associated efficiently with ID93 (95–99% of the ID93 was bound to the liposome after mixing). Association data is not shown for the liposomes covalently linked to ID93 since these were separated from unassociated protein by size exclusion chromatography in the course of manufacture. In contrast, the anionic liposomes (with or without GLA) and the aqueous GLA showed little association with the protein (although it must be noted here that the latter includes a population of very small particles [data not shown] which could potentially coelute with unassociated ID93). The above patterns are interpretable based on electrostatic interactions between the formulations and the negatively charged ID93. However, the various oil-in-water emulsions showed unexpected differences in association with ID93 depending on the oil component. Thus, GLA-SE showed high association with ID93 (96%), with the GLA-grapeseed emulsion having somewhat lower association (76%), and the Miglyol-based emulsion associating with only a fraction of the total antigen concentration (25%). It should be noted that it was necessary to increase the ID93 concentration in the association assay to 0.1 mg/ml in order to detect the protein, compared to the concentration employed in the in vivo studies (0.005 mg/ml). The disparate physicochemical characteristics of the various adjuvant formulations as well as the extent of their association with ID93 provide a potential context for interpreting the effects of formulation parameters on immune responses in immunized mice.

3.2. Alum: GLA and alum synergize to enhance a protective ID93-specific TH1 response

We have previously found that ID93 in the SE formulation elicits a strong CD4+ T helper 2 (TH2) cell response, characterized by IL-5 and IgG1 antibody, which was not beneficial for control of Mtb. Inclusion of the TLR4 agonist GLA into this vaccine formulation switches the ID93 specific T cell response to a predominantly IFN-γ producing T cell response and balanced IgG1 and IgG2 antibody response [17]. This correlated with an increase in protective efficacy against low dose aerosolized Mtb challenge. The SE adjuvant used in those studies is similar to MF59 adjuvant included in human influenza vaccines approved in the EU since 1997 [29]. Despite a good overall safety profile after millions of doses administered, MF59 and other emulsions induce some local reactogenicity and there are still no approved emulsion-based adjuvants in the US. The only FDA-approved human vaccines containing adjuvants are alum-based formulations or in the case of Cervarix MPL and alum (AS04). In fact, over a billion doses of vaccines containing alum have been administered to people of all ages [30]. Alum-based formulations may also offer stability advantages, since antigens may adsorb to the alum particles, which may enhance the immunogenicity of the antigen.

To determine whether alum-based adjuvant formulations with or without a TLR4 agonist would be beneficial for an ID93-based vaccine we immunized mice with either unadjuvanted ID93, ID93 adjuvanted with alum, ID93 adjuvanted with an aqueous nanosuspension of GLA or ID93 adjuvanted with a combination of GLA and alum (GLA-Alum) where the GLA and ID93 are both adsorbed to the alum (Table 1) [28]. One month after the final immunization all immunized groups had substantial anti-ID93 IgG1 titers (Fig. S2A). Inclusion of alum with or without GLA enhanced this response, whereas aqueous GLA did not boost IgG1 responses. Conversely IgG2c titers were only elicited when ID93 was adjuvanted with aqueous GLA or GLA-Alum (Fig. S2B). Formulation of ID93 in alum alone did not elicit IgG2c responses and alum did not enhance responses to ID93/GLA (Fig. S2B).

Immunization with ID93 or ID93-alum elicited ID93-specific IL-5 producing splenocytes (Fig. 1A). Both the aqueous GLA and GLA-Alum adjuvants inhibited IL-5 responses compared to ID93 alone or ID93-Alum, indicating that GLA can override the TH2 programming elicited by protein alone or protein plus alum immunization. Although neither the alum nor the aqueous GLA adjuvant enhanced IFN-γ or TNF production, the combination of GLA and alum elicited a significant IFN-γ and TNF response by CD4 T cells to ID93 stimulation, indicating that these two adjuvants can synergize to elicit TH1 responses following immunization (Fig. 1B). Alum and aqueous GLA, as well as the combination GLA-alum slightly enhanced the frequency of IL-2 producing cells compared to unadjuvanted ID93 (Fig. 1B). Alum also elicited a low frequency of IL-17-producing CD4 T cells and this was not impacted by inclusion of GLA with the alum. Of the ID93-specific TH1 cells, only ID93/GLA-alum produced multifunctional TH1 cells characterized by co-production of IFN-γ and TNF. More than half of these cells also produced IL-2 (Fig. 1C).

Compared to immunization with ID93 alone, immunization with ID93/GLA-alum significantly reduced the bacterial burden in the lungs and spleens of animals after a low dose aerosol challenge with virulent Mtb (Fig. 1D and E). There was no demonstrable benefit from immunization with ID93-alum, whereas immunization with ID93/GLA provided a non-significant trend towards protection in the lung and spleen. Taken together these data demonstrate that GLA and alum can synergize to promote the generation of multi-functional TH1 cells and provide significant protection against aerosolized Mtb, whereas neither GLA nor alum alone elicited a TH1 response or provided a protective benefit.

3.3. Emulsions: Squalene is essential for the generation of protective TH1 responses by ID93 adjuvanted with GLA in an oil-in-water emulsion

When adjuvanted with a squalene-based SE, ID93 elicited a TH2 response that was either non-protective in mice or detrimental to guinea pigs following Mtb challenge [17]. These studies led us to speculate that, while squalene-based emulsions produce unique adjuvant activity, when used in combination with a TLR ligand such as GLA it may be desirable to substitute squalene with an immunologically inert oil. We hypothesized that a non-squalene emulsion could serve as a delivery vehicle for a TH1 inducing TLR agonist adjuvant while not inducing separate unwanted TH2 adjuvant activity of its own.

To determine whether other oils could be substituted for squalene in an oil-in-water formulation of ID93 and GLA we created stable emulsions using either long-chain triglyceride (grapeseed oil) or a medium-chain triglyceride (Miglyol). These alternate emulsions were very similar to the squalene-based emulsions in terms of physicochemical characteristics (Table 1). Mice were immunized three times with either unadjuvanted ID93, ID93 adjuvanted with aqueous GLA, GLA-SE, GLA-Miglyol, GLA-grapeseed, GLA-Alum or unimmunized. As expected, immunization with ID93 elicited a TH2 response (Fig. 2A). Addition of any of the GLA containing formulations ablated this response, indicating that the presence of the TLR4 agonist was sufficient to inhibit TH2 induction, regardless of the delivery system. However, only the squalene containing GLA-SE adjuvant enhanced robust multi-functional TH1 responses (Fig. 2B). The most common response produced by ID93/GLA-SE was CD4 T cells that co-produced IFN-γ and TNF with or without IL-2 upon restimulation (Fig. 2C). The identity of the oil was critical to TH1 induction as changing the oil from squalene to Miglyol or grapeseed oil greatly limited the induction of TH1 responses after immunization. Additionally, the magnitude of the TH1 response elicited by ID93/GLA-SE was substantially greater than that elicited by the ID93/GLA-alum formulation, although the latter did elicit measurable multi-functional TH1 cells (Fig. 2).

Inclusion of any of the GLA containing adjuvants dramatically enhanced the production of ID93 specific IgG2c antibodies, whereas IgG1 titers were not augmented by the inclusion of the adjuvant (Fig. S3). GLA-SE provided the greatest enhancement of IgG2c. Formulation of ID93/GLA in either the grapeseed or Miglyol emulsion did not alter the responses compared to the oil-free aqueous GLA adjuvant suggesting that these oils are immunologically inert as far as effecting adaptive immune responses and that IgG2c enhancement is dependent on the presence of GLA, but can be augmented when formulated in SE.

One month after the final immunization we challenged cohorts of animals with a low dose of aerosolized Mtb to determine how the oil content impacts vaccine efficacy. There was a trend towards lower bacterial burden in the lungs of mice immunized with unadjuvanted ID93 compared to unimmunized mice, although this was not statistically significant. Compared to the animals that were immunized with unadjuvanted ID93, only the ID93/GLA-SE or ID93/GLA-Alum immunizations significantly lowered the lung bacterial burden (Fig. 2D). There was a slight trend towards lower bacterial burden with the other adjuvanted vaccines, but these differences did not reach statistical significance. The pattern of protection was similar in the spleens of these animals as only ID93/GLA-SE and ID93/GLA-Alum immunizations significantly reduced the Mtb burden (Fig. 2E). These data show that inclusion of squalene in an oil-in-water stable emulsion of ID93/GLA is critical for production of a strong multi-functional TH1 response that correlates with protection against Mtb challenge. ID93/GLA-Alum also elicited multi-functional TH1 cells albeit at a much reduced level. However the immune response elicited by this regimen was substantial enough to limit Mtb infection.

3.4. Liposomes: Lipid content alters the immunogenicity of ID93/GLA liposome vaccines

Liposomes are an attractive class of formulations for vaccine delivery due to their versatility, ease of manufacture and low cost of materials [31]. Liposome-based adjuvants are currently employed in many countries in approved influenza and hepatitis A vaccines. Liposomes do not contain an oil component but rather consist of a phospholipid bilayer surrounding an aqueous core. To assess whether liposomes would be a beneficial delivery platform for an ID93-based vaccine we prepared three different DPPC-based liposome formulations of ID93 with and without GLA. The liposome formulations differed in their minor lipid component, which produced either an anionic or cationic liposome as well as whether they were covalently bound to ID93, electrostatically associated (cationic) or not associated (anionic) (Table 1). We also hypothesized that associating the ID93 antigen to the liposome via covalent binding would enhance the liposome efficacy by increasing antigen uptake.

In the absence of GLA the liposome formulations enhanced the number of ID93-specific IL-5 producing splenocytes, although anionic liposomes provided only a marginal enhancement (Fig. 3A). Addition of GLA to any of the liposome formulations ablated this response. Both IFN-γ and TNF production by CD4 T cells were enhanced by addition of GLA to liposome formulations. Surprisingly the anionic liposomes produced the greatest frequency of TH1 cells when combined with ID93/GLA (Fig. 3B). Only the anionic liposome formulation of GLA substantially augmented the frequency of IL-2 producing CD4 T cells, whereas there was little IL-17 induction detectable over background levels. All three liposome formulations of ID93/GLA elicited multifunctional TH1 cells characterized by IFN-γ and TNF production with or without simultaneous IL-2 production (Fig. 3C). These cells were most frequent following immunization with ID93/GLA-anionic. In the absence of GLA none of the liposome formulations significantly enhanced ID93-specific TH1 responses.

The choice of liposome or the inclusion of GLA had little impact on ID93-specific IgG1 titers following immunization (Fig. S4A). Surprisingly, both the cationic and covalently-associated liposomes enhanced ID93-specific IgG2 titers compared to unadjuvanted ID93, even in the absence of GLA, particularly when the liposome was covalently linked to the antigen (Fig. S4B). Addition of GLA further increased these responses in both the anionic and cationic formulations (P < 0.01 for both comparisons).

To determine whether the differences in the magnitude of the TH1 response impacted the protective efficacy of the different ID93/GLA-liposome formulations we challenged immunized mice with a low dose of Mtb. Compared to unadjuvanted ID93, all three GLA-liposome formulations promoted control of Mtb in the lung and trended towards protection in the spleen (Fig. 3D and E). Among the ID93/GLA-liposome groups there was no clear correlation between the magnitude of the TH1 response and the magnitude of protection. However inclusion of GLA was essential to the protective efficacy as ID93 formulated in liposomes alone provided no protective benefit (data not shown). Taken together, these data show that liposome formulations of ID93/GLA enhance TH1 responses following vaccination, particularly the anionic formulation, which correlated with significant protection against Mtb challenge.

3.5. Squalene SE is a more effective formulation than anionic liposomes for enhancing the immunogenicity of ID93/GLA

In humans the liposomal formulation of MPL and QS21, AS01, has proven to enhance the immune responses to two different antigens more robustly than the oil-in-water formulation of the same immunostimulants, AS02. To determine whether an SE or liposome formulation of ID93/GLA would prove more immunogenic and protective we directly compared CD4 T cell responses in animals immunized with either ID93/GLA-SE or ID93/GLA-liposomes. We focused on the anionic liposome formulation because it showed the strongest enhancement of TH1 responses among the various GLA-liposome formulations and there was no clear difference in protective efficacy against pulmonary Mtb between these formulations (Fig. 3). Additionally the anionic liposome formulation was substantially easier to produce and is more amenable to scale up than the covalent liposomes. Due to the lower TH1 response elicited by ID93/GLA-Alum compared to ID93/GLA-SE the former formulation was not evaluated further (Fig. 2).

One week after the final immunization both ID93/GLA regimens induced multi-functional TH1 responses in the spleen and lung (Fig. 4A and B). There was a greater frequency of response following ID93/GLA-SE immunization in both organs, particularly of IFN-γ and TNF co-expressing cells. To further characterize these cells we stained splenocytes with an MHC class II tetramer bound to the dominant epitope from Rv3619, one of the component proteins of ID93 (Fig. 4C). As expected there was a higher frequency of Rv3619 specific cells in animals immunized with ID93/GLA-SE (Fig. 4D). It has been proposed that the long-term fate of antigen experienced T cells can be predicted based on expression of different constellations of surface markers and transcription factors. Kaech and colleagues have suggested that short-lived effector (SLEC) CD4 T cells express high levels of the TH1 committing transcription factor T-bet and the surface marker Ly6C and do not persist, whereas memory precursor effector cells (MPEC) express lower levels of these two proteins and persist to transition into memory T cells [32]. Using expression of two inhibitory surface molecules Woodland and colleagues proposed that SLEC are distinguished as being PD-1- KLRG1+ whereas MPEC are PD-1+ and KLRG1- [33]. The costimulatory molecule CD27 and CD62L which facilitates lymph node homing have also been suggested to be expressed on memory precursor cells [34,35]. One week following immunization ID93/GLA-SE induced more terminally differentiated SLEC CD4 T cells than ID93/GLA-liposome as defined by any of three phenotypes: CD27⁻ CD62L⁻, KLRG1⁺ PD-1⁻, or Ly6C^hi T-bet^hi, whereas ID93/GLA-liposome immunization induced a greater frequency of MPECs defined as either: CD27⁺ CD62L⁺, KLRG1⁻ PD-1⁺, or Ly6C^lo T-bet^lo (Fig. 4E). Importantly there was a large overlap of these phenotypes indicating that the populations identified by different groups using different markers may largely be the same.

One month after immunization the frequency of multi-functional ID93-specific TH1 cells in the spleen and lung were more similar between the two immunization regimens than at one week post-immunization, although responses were still higher in animals that received ID93/GLA-SE (Fig. 5A and B). This likely reflects the increased induction of long-lived MPEC by ID93/GLA-liposome. Both ID93/GLA-SE and ID93/GLA-liposome provided significant protection in the lung following aerosol challenge with Mtb (Fig. 5C and D). Protective efficacy in the lung was indistinguishable between animals immunized with ID93/GLA-SE and ID93/GLA- liposome. The Mtb burdens in the lungs of immunized mice were not significantly different than in mice receiving BCG, which is the gold standard for protection in the mouse model of Mtb challenge. Compared to ID93/GLA-liposome, immunization with ID93/GLA-SE did provide a more robust protective benefit against Mtb dissemination to the spleen, where it was not significantly different than BCG. Taken together we find that immunization with ID93/GLA-SE elicits a greater frequency of ID93 specific TH1 cells and provides more effective protection against disseminated Mtb at the time points tested. However ID93/GLA-liposome elicits a higher frequency of MPEC CD4 T cells and also is protective against experimental challenge.