We scalably assembled spherical BCN with and without loaded MA using flash nanoprecipitation (FNP) as previously described [29]. Dynamic light scattering (DLS) showed that both MA-BCN and BCN were consistent in size (364 ±19 nm and 354 ± 14 nm, respectively), and were monodisperse based on their polydispersity indices (PDIs) of 0.24 ± 0.04 and 0.21 ±0.05, respectively (Fig 1A). We next verified that the BCN formulation maintained its characteristic interconnected aqueous channels using cryogenic transmission electronic microscopy (cryo-TEM) (Fig 1B) and negative staining transmission electron microscopy (TEM) (Fig 1C). Using small angle X-ray scattering (SAXS) studies, we confirmed the internal cubic organization of BCN. Bragg peaks at the √2, √4, and √6 ratios show that the primitive type of cubic internal organization was preserved between MA-BCNs and BCNs (Fig 1D). Thus, MA encapsulation did not disturb the BCN architecture. We also manufactured a PLGA nanocarrier formulation (PLGA-NP) with and without MA encapsulation. PLGA-NP morphology were confirmed using cryoTEM (Suppl Fig 1A) and SAXS (Suppl Fig 1B) and the size of blank PLGA-NP and MA-loaded PLGA-NP (MA-PLGA) were found to be 169 ± 4 and 163 ± 5 nm, respectively, with PDIs of 0.18 ± 0.04 and 0.19 ± 0.12, respectively (Fig 1E). Using the coumarin derivatization method [21], we found that BCNs could more efficiently encapsulate MA than PLGA-NPs (Fig 1F). We previously noted that the highly stable cubic architecture of BCN can lead to the retention of cargo within the endolysosomal compartment [22]. Indeed, Texas Red-Dextran loaded BCN were mostly found in lysosomes while Texas Red-Dextran loaded PLGA-NP could be found both within the lysosome and cytosol of cells (Supp Fig 2A, B).

As PLGA has served as a component of a wide range of vaccine formulations, including Mtb vaccines [30, 31], we benchmarked PEG-b-PPS BCN against PLGA-NP as a nanocarrier platform for lipid antigens. Both MA-BCN and MA-PLGA were highly effective in activating both mouse (DN1) and human (M11) MA-specific T cells in vitro and inducing IFN-γ production (Fig 2A-D). In particular, MA-BCN were significantly better at activating MA-specific T cells compared to free MA at equivalent concentrations (Fig 2A, B and D). In comparison to MA-BCN, MA-PLGA was significantly more effective at stimulating mouse MA-specific T cells in vitro at lower concentrations. Interestingly, even in the absence of MA, PLGA-NP showed strong stimulation particularly for M11 T cells. These results demonstrate that MA encapsulation within BCN enhances its ability to stimulate antigen-specific T cells and reveal a considerable background, and thus difficult to control, stimulatory effect of blank PLGA-NP. In contrast, blank BCN showed little background immunostimulation [23], allowing a more controlled dose-dependent increase in T cell responses as the loaded MA concentration increased.

We next tested the ability of MA-loaded and unloaded BCN and PLGA to stimulate DN1 T cells in vivo. Intranasal (IN) vaccination of hCD1Tg mice was followed by the adoptive transfer of DN1 T cells. Surprisingly, we found that vaccination with MA-BCN induced a significantly higher percent of proliferation and activation in DN1 T cells than MA-PLGA in the draining lymph nodes (LN) and lungs at 1 week post-vaccination (Fig 2E-G). The extent of cell proliferation induced by MA-PLGA was not significantly different from that of blank PLGA (Fig 2F). Since MA-BCN effectively stimulated MA-specific T cells in vivo while MA-PLGA did not, we focused on further characterizing and assessing vaccination via solely MA-BCN formulations.

Given the lack of knowledge regarding BCN and lipid antigen kinetics, it would be of interest to test whether delivered antigen could persist in the mouse weeks post-vaccination (Fig 3A). We switched from intranasal vaccination to intratracheal (IT) vaccination, as it allowed for the reliable delivery of a larger volume of the vaccine. We found that a significant proportion of the adoptively transferred DN1 T cells could proliferate and become activated in the LNs, lung, and spleen of MA-BCN-vaccinated mice 6 weeks post vaccination (Fig 3B-D). This activation and proliferation were MA-specific, as no activation and proliferation of Ag85B-specific p25 Tg (p25) T cells were observed[34].

Since BCN allow for the co-loading of lipid and protein antigens, we assessed the ability of dual loaded Ag85B-MA-BCN to activate antigen-specific T cells as well as lead to antigen persistence. We co-loaded Ag85B and MA in BCN and found Ag85B-MA-BCN to have a diameter of 380 nm and PDI of 0.223, comparable to MA-BCN (Fig 4A). Encapsulation efficiency was also relatively high at 70%.Vaccination of hCD1Tg mice with Ag85B-MA-BCN activated and induced proliferation of both p25 and DN1 T cells in the LN, lung, and spleen (Fig 4B-D) at one-week post-vaccination. In addition, DN1 T cells were still able to be activated 6 weeks post-vaccination in the context of an Ag85B-MA-BCN vaccination, while p25 T cells did not show increased activation or proliferation compared to blank BCN control at this time point (Fig 4E-H). Therefore, MA, but not Ag85B, appears to persist in the context of an Ag85B-MA-BCN vaccination.

To determine whether BCN encapsulation contributed to MA persistence, we designed an experiment that would allow direct comparison between MA-BCN and free MA. To account for their differential efficiency in stimulating DN1 T cells, we pulsed bone marrow-derived dendritic cells (BMDCs) from hCD1Tg mice with MA or MA-BCN at 10 μg/mL and 5 μg/mL MA concentration, respectively (Fig 5A). hCD1Tg mice were immunized with MA or MA-BCN-pulsed BMDCs at either 1 week or 6 weeks before the adoptive transfer of DN1 T cells, allowing comparison of short-term vs. long-term vaccination conditions. We found that while there were no significant differences in DN1 T cell activation and proliferation between mice vaccinated with free MA and MA-BCN pulsed BMDCs at 1-week post-vaccination, after 6 weeks, DN1 T cell response could only be detected in the MA-BCN vaccination condition (Fig 5B, C). Thus, BCN encapsulation contributed to the persistence of MA. We then tested whether the BCN structure could contribute to antigen persistence by comparing PEG-b-PPS BCNs to PEG-b-PPS MCs using a similar experimental set up (Suppl Fig 3). We found that MA-BCN and MA-MC overall had similar ability to activate DN1 T cells in both short-term and long-term vaccination time points (Suppl Fig 3B, C). Thus, the structure of the encapsulating NP did not play a significant role in antigen persistence of MA.

To investigate if the IT route of vaccination played a role in the development of antigen persistence, we tested intravenous (IV) and subcutaneous (SC) vaccination routes with MA-BCN. DN1 T cell activation and proliferation were detected at 6-weeks post-vaccination through both routes of vaccination (Fig 5D-F), suggesting that the antigen persistence is not unique to IT vaccination. To assess whether Mtb lipid antigen persistence could also be observed in mice vaccinated with an attenuated strain of Mtb, we vaccinated hCD1Tg mice SC with Mtb mc²6206 strain and tested for the persistence of both Ag85B and MA after 6 weeks. The Mtb mc²6206 strain displays less virulence in mouse studies than BCG but still confers long-term protection against virulent Mtb challenge equivalent to BCG[35]. We confirmed that no bacterial burden remained at the 6 weeks time point (Suppl Fig 4A). Similar to MA-BCN-vaccinated mice, DN1 T cells were activated and proliferated in the draining lymph nodes and spleen of mice vaccinated with the attenuated Mtb strain (Fig 5D-F). In line with our findings with BCN, no proliferation was seen in p25 T cells (Suppl Fig 4B-D). These results suggest that incorporation of MA in a nanocarrier vaccine allows for the mimicry of the lipid antigen persistence induced by vaccination with an attenuated Mtb vaccine.

To identify the location of antigen persistence, we determined the biodistribution of BCN after IT vaccination using BCN loaded with the hydrophobic dye DiD (DiD-BCN). Since DiD is a lipophilic dye which, like MA, stably loads into the BCN bilayers, we could track the localization of BCN without having to perform additional chemical modifications. We found that IT-administered DiD-BCN were located almost exclusively within the lung at all time points tested (4 h, 24 h, 48 h, 6 days, and 6 weeks) (Fig 6A). We were able to detect fluorescence through both in vivo imaging system (IVIS) and flow cytometry at 6-weeks post-vaccination that was significantly above unvaccinated control (Fig 6A, B), despite DiD quickly losing fluorescence upon mixture into aqueous environments [36], suggesting that some DiD-BCN structure remained intact at this time point. Flow cytometric analysis (Suppl Fig 5) revealed that DiD BCN was primarily found within alveolar macrophages (AMs) at both early and late time points with neutrophils and DCs showing some fluorescence at early time points (4 and 24 h) (Fig 6C).

We next tested whether MA likewise persisted primarily within AMs. 6-weeks after vaccination, we enriched for AMs using a column-based magnetic cell isolation system containing SiglecF antibody, which is highly expressed on AMs and a small population of eosinophils in the lung (Fig 6D). We co-cultured enriched AMs or the flow through with or without hCD1Tg-expressing BMDCs and DN1 T cells (Suppl Fig 6). While both enriched AM and flow through fractions from MA-BCN vaccinated mice co-culture with BMDCs could activate DN1 T cells (Fig 6E), co-culture with enriched AMs led to significantly higher DN1 T cell activation, suggesting AMs are the primary location of MA persistence. The ability of flow through to likewise lead to DN1 T cell activation suggests MA may also persist in other cell types or could be attributed to residual AMs in flow through. Furthermore, AMs could not themselves present the MA to DN1 T cells, noted by the lack of T cell activation in the absence of BMDCs, suggesting that MA may be transferred from AMs to DCs for presentation. In fact, CD1b is not expressed on AMs [19, 37] and therefore the presence of CD1b-expressing DCs may be necessary both in vitro and in vivo for T cell activation to occur. These data indicate that MA-BCN mainly persists within AMs in the lung after IT vaccination and CD1b-expressing DCs is required for antigen presentation and activation of MA-specific T cells.

To study the memory phenotype of DN1 T cells after MA-BCN vaccination, we constructed a mixed DN1 bone marrow (BM) chimera mouse model (DN1-hCD1Tg), since adoptively transferred DN1 T cells in unvaccinated mice cannot survive long term (Fig 7A). After vaccination, the percent of CD44⁺ DN1 T cells in the blood increased quickly and remained high until the last time point of 40 days (Fig 7B). Various classical memory T cell subsets have been characterized 1 month after initial antigen-encounter, we therefore determined the DN1 T cell phenotype 6-weeks post vaccination. We found that the most prevalent memory population within DN1 T cells were CD44⁺CD62L⁺, markers used to define central memory T cells, particularly in the LN (Fig 7C and Supp Fig 7A). To characterize the memory DN1 T cells, we performed RNA-seq analysis on sorted CD44⁺CD62L⁺ (memory) and CD44⁻CD62L⁺ (naïve) DN1 T cells from LNs of MA-BCN vaccinated DN1-hCD1Tg BM chimeric mice. We found that memory and naïve DN1 T cells clustered separately after principal component analysis (PCA), despite these samples coming from the same animals (Suppl Fig 7B). A total of 995 differentially expressed genes (DEGs) were identified of which 542 upregulated and 453 downregulated in the memory subset (Suppl Fig 7C). Next, we determined which T cell population memory DN1 T cells most resembled. Toward this end, we obtained data from ImmGen database (including terminally differentiated effector (Te), memory precursor (Tmp), central memory (Tcm), effector memory (Tem), regulatory (Treg))⁴², and two additional publications (follicular helper (Tfh)⁴³, exhausted (Texh)⁴⁴) and compared the respective DEG lists. Using PCA, we found that memory DN1 T cells clustered most closely to Tfh, Treg, and Texh (Fig 7D) and when this subset is isolated, most closely to Tfh cells (Fig 7E).

Within the DEGs, we noted upregulation of key Tfh cell transcription factors BCL6 and TCF1 (TCF7), chemokines CXCR5 and CXCR4 and surface receptor ICOS, important for Tfh cell migration into the germinal center, and adaptor protein SAP (SH2D1A), important for T cell-dependent B cell immunity (Fig 7F)⁴⁵. Inhibitory receptors PD-1 (PDCD1), TIGIT, CTLA4, and transcription factor EOMES, which are known to be expressed on Tfh cells were also upregulated. In addition, we observed downregulation of transcription factor KLF2, which can inhibit Tfh cell differentiation, and CCR7 and IL-7R, receptors known to have decreased expression in T_FH cells. Consistent with our transcriptome analysis, we found increased percentage of CXCR5⁺PD-1⁺ DN1 T cells in MA-BCN vaccinated mice 6 weeks post-immunization (Fig 7G, H). Both our transcriptomic and experimental data highlighted that memory, but not naïve DN1 T cells, were proliferating, given the enrichment for the mitotic cell cycle process (Fig 7I) and increased KI67⁺ percentage within this population (Fig 7J). MA-BCN vaccination thus led to the differentiation of DN1 T cells into a proliferating Tfh-like T cell population.

This study was carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of Northwestern University (Protocol number: IS000011717).

Poly (ethylene glycol)₁₇-block-poly-(propylene sulfide)₇₅ (PEG₁₇-b-PPS₇₅) copolymer was synthesized as previously reported [72, 73]. Briefly, monomethoxy PEG (molecular weight=750 Da) was functionalized with mesyl chloride and substituted with thioacetate to produce a PEG acetate initiator for the polymerization for PPS.

BCN were produced using FNP technique as previously described[72]. In brief, PEG₁₇-b-PPS₇₅ BCN polymer and any hydrophobic cargo was dissolved in THF and loaded onto a syringe. The hydrophilic cargo was loaded onto a separate syringe, and both impinged against one another through the confined impingement mixer into scintillation vial. The vial was placed in a vacuum desiccator overnight to remove the organic solvent. NPs were filtered through LH-20 column with PBS as an eluent and suspended in PBS for later use. MA encapsulation efficiency was analyzed using coumarin derivatization as previously described [21]. MA was loaded at a concentration of 100 µg/mL of total BCN formulation and DiD was loaded at 5 µg/mL total BCN formulation. MA loading capacity in BCNs was 2%. To measure the encapsulation efficiency of Ag85B, Ag85B-MA-BCN were centrifuged at 10,000 g for 10 minutes. The supernatant was collected and concentrated using Amicon® Ultra-4 10K centrifugal filter devices. To determine the unencapsulated protein concentration, the supernatant was incubated with Pierce™ BCA Protein assay reagent following the manufacturer protocol and the absorbance was measured at 660 nm using a SpectraMax M3 multi-mode microplate reader (Molecular Devices, LLC). The percentage encapsulation efficiency was calculated as a percent of encapsulated protein to the total amount of protein added. Ag85B was encapsulated at a concentration of 140 µg/mL and loading capacity was found to be 2.88%.

PLGA-NP were prepared by the Oil-in-Water single emulsion method. Briefly, PLGA (20 mg in 800 µL dichloromethane) organic phase was emulsified using an ultrasonic processor with aqueous phase containing 2 mL of polyvinyl alcohol (PVA) solution (2.5% w/v) to form an emulsion. The emulsion was then added dropwise into 4 mL of stirring 0.5% w/v PVA solution at room temperature to evaporate the organic solvent. NPs were collected after 4 hours of stirring followed by centrifugation at 10,000 x g for 10 minutes. After centrifugation, NPs were washed twice with cold water to remove residual PVA and redispersed in PBS. Hydrophilic (Texas red-Dextran) or hydrophobic (MA) loading was performed by adding the payloads to the aqueous and organic phases, respectively.

The nanocarrier size (z-average diameter) and polydispersity was measured using dynamic light scattering (DLS) using a Nano 300 ZS Zetasizer (Malvern Panalytical, Malvern, U.K.). BCN and PLGA morphology was visualized with cryogenic transmission electron microscopy (cryoTEM) as previously detailed [21]. Transmission electron microscopy (TEM) was performed to image BCN using uranyl acetate as a negative stain as previously reported [22]. BCN aggregate structure and internal morphology was characterized with small angle X-ray scattering (SAXS). These studies were performed at the DuPont-Northwestern-Dow Collaborative Access Team beamline at Argonne National Laboratory’s Advanced Photon Source with 10 keV (wavelength λ = 1.24 Å) collimated X-rays, as described previously[72].

Confocal images of RAW 264.7 macrophages stained with lysosomal dye Lysotracker™ (green) and NucBlue™ stain (blue) following incubation with Texas Red-labeled PLGA-NP and BCN (red) for 8 hr. The cells were then imaged within a humidified chamber using a 63X oil-immersion objective on a SP5 Leica Confocal Microscope using HyD detectors and lasers. Data analysis was performed using ImageJ software.

Human CD1 transgenic mice (hCD1Tg) in C57BL/6 background or MHC II-deficient background [19] and CD1b-restricted MA-specific TCR transgenic mice on Rag^−/− background (DN1Tg/hCD1Tg/Rag^−/−) [20] were generated in our lab as previously described. P25-specific TCR transgenic mice (Jackson lab) were crossed onto Rag^−/− background. Both males and females were used.

For cell surface staining, cells were pre-incubated with 2.4G2 Fcγ RII/RIII blocking mAb for 15 minutes and then stained with the appropriate combinations of mAbs diluted in HBSS + 2% FBS for 30 minutes at 4°C. Cells were analyzed on a BD FACS CantoII, or Cytek Aurora Spectral Cytometer and data were processed with FlowJo software (TreeStar). The complete list of antibodies use is as follows: anti-human Vβ5.1 (LC4, Invitrogen), anti-TCRβ (H57-597, BioLegend), anti-CD44 (IM7, BioLegend), anti-Vβ11 (KT11, BioLegend), anti-CD45.2 (104, BioLegend), anti-CD45.1 (A20, BioLegend), anti-SiglecF (E50-2440, BD Bioscience), anti-CD45 (30-F11, BioLegend), anti-CD11c (N418, BioLegend), anti-CD11b (M1/70, BioLegend), anti-Ly6G (1A8, BioLegend), anti-CD19 (6D5, BioLegend), anti-NK1.1 (PK136, BioLegend), anti-CD3 (17A2, BioLegend), Zombie Red (BioLegend), anti-mouse CD185/CXCR5 (L138D7, Biolegend), anti-mouse CD279/PD-1 (29F.1A12, Biolegend), and anti-Ki67 (SolA15, Invitrogen).

Nanoparticle and lipid titrations were performed as previously described [21]. In brief, hCD1Tg MHC II^−/− bone marrow-derived dendritic cells (BMDCs) were pulsed with varying quantities of nanoparticle of sonicated MA (Sigma-Aldrich, St. Louis, MO) and then co-cultured for 48 hours with DN1 T cells isolated from lymph nodes (axial, brachial, submandibular, inguinal) of DN1Tg/hCD1Tg/Rag^−/− mice. The supernatant was used for IFN-γ enzyme-linked immunosorbent assay (ELISA) performed as described previously [21] and cells were stained for activation.

Hydrophobic polyvinylidene fluoride 96-well plates (Millipore, Bedford, MA) were coated with anti-human IFN-γ mAb 1-D1K (Mabtech, Cincinnati, OH) diluted 1:400 in PBS (Gibco, Waltham, MA) and incubated at 4°C overnight. The following day, mAb was removed and each plate was washed and incubated at room temperature for 2 hours in the presence of media. Human monocyte-derived dendritic cells and MA-specific M11 T cells were seeded 50,000 cells and 2,000 cells per well, respectively, together with serially diluted MA and NPs. After 16-hour incubation, plates were washed and incubated with biotinylated anti-human IFN-g (7-B6-1), diluted 1:3000 in PBS + 0.5% FBS, for 4 hours. Plates were then washed with PBS 5 times and incubated for 2 hours with ExtrAvidin-Alkaline phosphatase diluted 1:1000 in PBS + 0.5% FBS. Spots were visualized after incubation with the BCIP/NBT substrate for up to 20 minutes and then imaged with ImmunoSpot reader v.2.0.

For both intranasal (IN) and intratracheal (IT) vaccination, mice were anesthetized using inhaled isoflurane. For IN vaccination, 25 µL of liquid was administered at the nostrils of the mice for a total of 1 μg of MA per mouse. For IT vaccination, mice were administered 25 or 50 μL of liquid for a total of 2.5 μg of MA with or without 2.0 μg of Ag85B. For BMDC vaccinations, hCD1Tg BMDCs were cultured in GM-CSF and IL-4 for 5 days[74] and pulsed with MA-BCN (5 μg/mL of MA) or free MA (10 μg/mL of MA) for 24 hours. Cells were washed and resuspended in PBS and IT vaccinated at a dose of 1×10⁶ BMDCs/mouse. Intravenous vaccination was administered by tail vein injection. Subcutaneous vaccination was administered between the shoulders over the neck portion of the mouse.

Attenuated Mtb strain (H37Rv ΔpanCD ΔleuCD referred to as mc² 6206) was kindly provided by Jacobs’ Laboratory at Albert Einstein College of Medicine. Bacteria were grown as previously described in supplemented Middlebrook 7H9 media [35, 75]. Mice were vaccinated subcutaneously with 1×10⁶ CFU in 100 μL of PBS. Mycobacterial burden quantification was performed by plating serial dilutions of lung or spleen homogenate on Middlebrook 7H11 agar plates [35, 75].

DN1 or p25 T cells were isolated from lymph nodes of DN1Tg/hCD1Tg/Rag^−/− or P25Tg/Rag^−/− mice, respectively. Cells were labeled with CellTrace™ Violet reagent (Invitrogen™) as per manufacturer’s instructions and 3-5×10⁶ cells were injected into hCD1Tg mice either 1 day before vaccination (short term) or 6 weeks post-vaccination (long term). On day 7 post-adoptive transfer, draining lymph nodes (axial, brachial, mediastinal), lung, and spleen were obtained. Single-cell suspensions were prepared by mechanical disruption in HBSS/2% FBS. The lung was digested with collagenase IV (1mg/ml; Sigma-Aldrich) and DNase I (30µg/ml; Sigma-Aldrich) for 30 minutes at 37°C before the disruption.

BCN were loaded with DiD dye (Thermo Fisher) and administered to mice IT at 4 hours, 24 hours, 48 hours, 6 days, and 6 weeks prior to the final time point, at which all mice were sacrificed and analyzed together. Lung, spleen, and draining lymph nodes were harvested and imaged using a near-IR in vivo Imaging System (IVIS; Center for Advanced Molecular Imaging, Northwestern University). The single-cell suspension was then prepared from indicated organs and cells were stained for flow cytometric analysis.

hCD1Tg mice were immunized IT with either BCN or MA-BCN (2.5 μg of MA/mouse). 6 weeks later, lungs were isolated and single-cell suspension was stained with anti-SiglecF-PE (E50-2440, BD Bioscience). SiglecF-positive cells were enriched using anti-PE MultiSort Kit (Miltenyi Biotec). 1×10⁵ flow through and AM enriched fractions were co-cultured with 1×10⁵ hCD1Tg BMDCs and 3×10⁵ DN1 T cells for 48 hours and cells were stained for the expression of activation markers.

Bone marrow cells ^from DN1Tg/hCD1Tg/Rag^−/− (CD45.2) and hCD1Tg mice (CD45.1) were depleted of mature T cells using anti-Thy-1.2 (AT83.A-6) plus rabbit complement (Cedarlane Laboratories). Equal numbers of cells (5×10⁶) were adoptively transferred into hCD1Tg (CD45.1) mice irradiated with 900 rads. After 6 weeks, BM chimeric mice were intratracheally vaccinated with either 2.5 μg of MA in MA-BCN or equivalent volume of PBS and the activation status of DN1 T cells were monitored in the blood 1-2 times per week until experiment completion.

CD44⁺CD62L⁺ and CD44⁻CD62L⁺ DN1 T cells were sorted from LNs of DN1-hCD1Tg BM chimera mice vaccinated with MA-BCN at 6 weeks post vaccination by BD FACS Ariax with purity >98%. RNA was extracted with RNAeasy® mini kit (Qiagen). Libraries were generated by using the Illumina Truseq preparation kit and sequenced on HiSeq4000. Reads were analyzed with Ceto pipeline (https://github.com/ebartom/NGSbartom) using STAR and HTseq for alignment on mm10 mouse genome and reading counting, as described previously [76]. Paired differential expression analysis was performed using DESeq2 to account for original mouse sourcing[77]. All downstream analysis was performed in R and figured generated using ggplot. Gene enrichment analysis was performed using Metascape [78]. Comparisons to other T cell profiles were performed using raw transcriptome data from ImmGen database[79], and two additional publications for follicular helper (T_FH)[80], exhausted (Texh)[81]. Analysis was performed as described above to obtain gene expression fold change values relative to internal naïve condition.

The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE226075 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE226075).

Statistical analyses were performed using Prism software 9 (GraphPad software, Inc.). Multi-group comparisons were done using two-way analysis of variance (ANOVA) with significance determined by Tukey’s multiple comparison. Where appropriate, outliers were identified through Grubbs method with alpha=0.05. Statistically significance denoted as ns = not significant, *p<0.05, **p<0.01, ***p<0.001, and ****p< 0.0001.