Ultra-low volume intradermal administration of radiation-attenuated sporozoites with the glycolipid adjuvant 7DW8-5 completely protects mice against malaria

Malaria is caused by Plasmodium parasites and was responsible for over 247 million infections and 619,000 deaths in 2021. Radiation-attenuated sporozoite (RAS) vaccines can completely prevent blood stage infection by inducing protective liver-resident memory CD8+ T cells. Such T cells can be induced by ‘prime-and-trap’ vaccination, which here combines DNA priming against the P. yoelii circumsporozoite protein (CSP) with a subsequent intravenous (IV) dose of liver-homing RAS to “trap” the activated and expanding T cells in the liver. Prime-and-trap confers durable protection in mice, and efforts are underway to translate this vaccine strategy to the clinic. However, it is unclear whether the RAS trapping dose must be strictly administered by the IV route. Here we show that intradermal (ID) RAS administration can be as effective as IV administration if RAS are co-administrated with the glycolipid adjuvant 7DW8-5 in an ultra-low inoculation volume. In mice, the co-administration of RAS and 7DW8-5 in ultra-low ID volumes (2.5 μL) was completely protective and dose sparing compared to standard volumes (10–50 μL) and induced protective levels of CSP-specific CD8+ T cells in the liver. Our finding that adjuvants and ultra-low volumes are required for ID RAS efficacy may explain why prior reports about higher volumes of unadjuvanted ID RAS proved less effective. The ID route may offer significant translational advantages over the IV route and could improve sporozoite vaccine development.


Introduction
The global burden of malaria remains unacceptably high with an estimated 247 million infections and 619,000 deaths in 2021 [1]. Many clinical malaria cases are concentrated in sub-Saharan Africa and are caused by P. falciparum (Pf), which is transmitted through the bites of infectious female Anopheles mosquitoes. Several pre-erythrocytic and erythrocytic vaccines target Pf and can provide varying degrees of protection against infection, clinical disease, and death (reviewed in [2,3]). However, the only vialed vaccines to routinely induce sterile protection against Plasmodium challenge in humans are liveattenuated whole sporozoite (spz) vaccines (i.e., Sanaria PfSPZ Vaccine and PfSPZ-CVac [4][5][6][7][8][9][10][11][12]). These are aseptic, puri ed, cryopreserved spz vaccines that induce both humoral and cellular immune responses [11]. Antibodies are mainly induced by the immunodominant circumsporozoite protein (CSP) antigen, and these antibodies can bind to spz to block hepatocyte invasion [4,13,14]. Although high titers of CSP-binding antibodies alone can confer high levels of protection [15,16], induction of CD8 + T cells, speci cally liver-resident memory CD8 + T (Trm) cells, appears to be critical for reliable and durable sterile protection [17,18].
To simplify and improve whole spz vaccination, we developed a two-step heterologous vaccine strategy called prime-and-trap [19,20]. Prime-and-trap combines priming with a nucleic acid-based vaccine in the periphery (e.g., skin) followed by expression of the cognate antigen in the liver through spz-or other vehicle-mediated delivery. In its rst generation, prime-and-trap was based on skin priming using plasmid DNA encoding the P. yoelii (Py) rodent malaria CSP antigen followed by a single intravenous (IV) dose of cryopreserved radiation attenuated spz (cryo-RAS) to direct and "trap" the activated and expanded CD8 + T cells in the liver. This strategy induced robust CSP-speci c CD8 + Trm responses in the liver and conferred durable sterile protection in this rodent malaria model for at least four months [20]. However, it was unclear whether the RAS dose must be strictly administered IV. This question is of substantial interest, since success with non-IV administration routes could simplify the translational feasibility of spz vaccines, including prime-and-trap.
Intradermal (ID) administration of RAS is an attractive alternative to IV administration since it attempts to mimic the natural route of exposure via mosquito bite. Moreover, the skin is accessible, patrolled by antigen presenting cells (APCs), and compared to other routes, can be dose-sparing [21][22][23]. Unfortunately, previous attempts at ID RAS administration (ID-RAS) in mice or humans were ineffectivethere was substantially higher vaccine e cacy following IV RAS administration (IV-RAS) than after ID [11,[24][25][26][27]. In prior studies, the amount of vaccine spz delivered to the liver as measured by total liver parasite burden was reduced after ID as compared to IV administration, and this difference was implicated as a primary reason for the failure of the ID route [28,29]. Other studies have also suggested that the reason for ID spz administration failure may be due to the spz inducing a more tolerogenic skin environment, which could ultimately lead to more regulatory immune responses in the liver [25,30]. However, most of these studies used standard ID injection volumes (10-50 µL), which do not mimic the ultra-low volumes delivered by probing mosquitoes [31,32], nor facilitate e cient exit of spz from the skin, since spz must move by contact-dependent motility [33,34]. Based on the available data, and recognition of this unique biology and motility requirements of spz, we hypothesized that two key aspects of RAS administration are critical for effective ID vaccination: 1) the injection volume must be compatible with the contact-dependent motility of the spz, and 2) the tolerogenic skin immune environment must be overcome.
In this study, we used the Py rodent malaria model to determine if ID-RAS can replace IV-RAS as the trapping component of the prime-and-trap vaccine. As ID-RAS are known to be less immunogenic and protective than IV-RAS, we investigated if we could improve the e cacy of ID-RAS trapping by reducing the volume and/or co-administering RAS with the glycolipid adjuvant, 7DW8-5. 7DW8-5 is a synthetic glycolipid adjuvant that was selected for this vaccine approach because it potently activates iNKT cells to preferentially induce Th1 cytokines (e.g., IFN-γ), inducing a cascade of immune cell activation including CD8 + T cells (reviewed in [35]). We showed that mice primed with DNA encoding the PyCSP antigen administered via gene gun followed by trapping with 7DW8-5-adjuvanted ID-RAS (ID-RAS + 7DW8-5) are highly protected against Py spz challenge. We also showed that reducing the volume used for ID-RAS to an ultra-low volume of 2.5 µL is dose-sparing and provides sterile protection for at least four months.
Furthermore, we showed that these modi cations have the potential to improve RAS-only vaccination in addition to prime-and-trap. Overall, we demonstrate that in mice, ID-RAS is as protective as IV-RAS when co-administered with a potent adjuvant in an ultra-low volume and may provide an alternative non-IV route for spz vaccination.

Materials and Methods
Mice Female 4-6 week-old BALB/cJ mice were purchased from Jackson Laboratories (Bar Harbor, ME) and housed at the University of Washington in an Institutional Animal Care and Use Committee (IACUC)approved animal facility. All mice were used under an approved IACUC protocol (4317-01 to SCM) and in accordance with relevant guidelines and regulations. All methods are reported in accordance with ARRIVE guidelines.

DNA vaccination by gene gun
The Py circumsporozoite protein (CSP) DNA vaccine plasmids were constructed in the pUb.3 vector and co-administered with Escherichia coli heat-labile toxin (LT)-encoding plasmid adjuvant as described [19,[36][37][38]. The PyCSP-minigene encodes the SYVPSAEQI epitope and the PyCSP plasmid encodes the fulllength CSP protein without the major repeat region. Supplementary Fig. 1 details amino acid sequences and agarose gel restriction digest plasmid validation for all PyCSP vaccines. All plasmid stocks were Sanger sequenced (GeneWiz Inc.) before use. Gene gun DNA vaccine cartridges were constructed as previously described [20,37]. Mice were vaccinated on a shaved abdomen using a PowderJect-style gene gun by priming using two cartridges per day on Days 0 and 2 (0.5 µg DNA per cartridge). This method of priming with PyCSP/LT-encoding plasmids via gene gun is referred to as ggCSP.

Cryopreserved irradiated spz vaccination
Cryopreserved Py wild type (WT) 17XNL (cryo-RAS) were radiation-attenuated (100 Gy by C0-60), puri ed, vialed, and produced by Sanaria Inc. (Rockville, MD) [11,39]. The vials were shipped to Seattle and stored in vapor phase liquid nitrogen per manufacturer recommendations. Cryo-RAS were thawed in a 37°C water bath for 30 seconds, diluted in Schneider's insect media (Gibco, Thermo Fisher Scienti c), and administered within 30 minutes of thawing. Spz counts were con rmed on a hemocytometer within one hour of injection. Figure legends specify the dose, volume, route, and number of injections for each experiment.
Freshly-dissected spz production and challenge Female Anopheles stephensi mosquitoes infected with wild-type P. yoelii 17XNL (Py WT) were reared at Seattle Children's Research Institute (Seattle, WA). Fresh spz were obtained by salivary gland dissection 14-18 days post-infection followed by Accudenz gradient puri cation as described [40]. Heat-killed spz (HK-spz) were generated by incubating Py WT spz in a 55°C water bath for 30 minutes. All spz were diluted in Schneider's insect media for administration. Figure legends specify the dose, volume, route, and number of injections for each experiment. For all spz challenge administrations, 1x10 3 freshly dissected Py WT spz in 100 µL were injected retro-orbitally (RO) IV. Blood stage protection after spz challenge was assessed by Giemsa (Sigma-Aldrich) stained thin blood smear microcopy on Days 3-14 post-challenge. Mice were deemed protected if blood smears remained negative for parasites up to Day 14. Intradermal and intravenous spz injections ID injections in standard volumes (STV) of 10-50 µL were administered with a BD Veo Insulin Syringe with Ultra-Fine needle 6mm x 31G 3/10 mL/cc (#324909). STV injections were administered in two ID injections per dose on the lower back near the base of the tail. Ultra-low volume (ULV) ID injections of 2.5 µL were administered with a 10 µl Sub-microliter injection syringe (World Precision Instruments, Inc #NANOFIL) and a 36G Beveled needle (World Precision Instruments, Inc #NF36BV). ULV injections were administered in two ID injections per dose on the left rear footpad. IV injections were all administered RO in 100 µL with an Exel International Insulin Syringes with a 29G permanently attached needle.
Supplementary Fig. 2 diagrams the locations of all ID and IV injections.
Glycolipid adjuvant preparation 7DW8-5 powder previously made under Good Manufacturing Practice (GMP) conditions was reconstituted in DMSO and prepared for injection as described [20]. 7DW8-5 or DMSO vehicle control was mixed with the cryo-RAS vaccines immediately before administration. All mice received 2 µg of 7DW8-5 adjuvant per immunization.
ELISA: Interferon-γ (IFN-γ) or IL-4 cytokine levels were determined by commercial ELISA kit according to manufacturer's instructions (BioLegend, San Diego, CA; #430801 and #431104). Blood was collected into tubes containing EDTA and then plasma was isolated and frozen. For liver tissue, half of the liver was excised, weighed, and pulverized by bead beating in 3 mL lysis buffer (phosphate-buffered saline (PBS), 1:100 Pierce protease inhibitor (Thermo Fisher Scienti c, # A32953), 0.05% Triton X-100). Homogenized samples were centrifuged at 16,000 x g for 10 min at 4°C. Supernatant was collected and frozen. All samples were diluted in the kit assay diluent, and absorbance was read on the CLARIOstar Plus plate reader (BMG Labtech, Germany) according to kit instructions. Standard curves and cytokine concentrations were calculated in Microsoft Excel.
PyCSP binding antibodies in mouse serum were determined by direct ELISA as previously described [41]. Blood was collected via submental bleed; serum was isolated and frozen. All serum samples were heat inactivated for 30 min at 56°C and centrifuged at 17,000 x g for 10 minutes prior to ELISA analysis. 50 ng per well recombinant PyCSP was plated in in 0.1M NaHCO 3 , pH 9.5, and incubated overnight at room temperature. Serum was diluted over a range of 1:50 to 1:109,350, and binding was detected with goat anti-mouse IgG Fc-HRP (Southern Biotech, #1013-05). Absorbance at 450 nm was determined with the BioTek ELx800 reader.
For CD8 depletion con rmation by ow cytometry, blood was collected via submental bleed into tubes containing EDTA 24 hours post CD8 depletion antibody or isotype injection. Whole blood was then resuspended in ammonium-chloride-potassium lysis buffer for 2-3 min to lyse red cells. The reaction was quenched with MACS buffer (PBS, 1 mM EDTA, 0.5% fetal bovine serum (FBS)). The nal cell pellet containing whole blood leukocytes was resuspended in MACS buffer, blocked, stained, and xed for ow cytometry as described below. The following Abs were used to assess CD8 cell depletion validation: live/dead dye-NIR, CD3e-BUV395, B220-BV711, CD4-Alexa Fluor, CD8a-BV421. Detailed information on ow reagents in Supplementary Table 1. Cell count per 100 µL blood was calculated based on known starting volume of mouse blood to normalize data. Flow cytometry was conducted on the LSRII instrument (BD Biosciences), and data were analyzed with FlowJo version 10.7.1 (BD Biosciences). For CD1d blocking con rmation, IFN-γ induced by 7DW8-5 was measured by ELISA. At 24 hours post CD1d or isotype depletion, 7DW8-5 was injected by the IV route. Six hours later, blood was collected (as described above), plasma was isolated, and IFN-γ cytokine levels were analyzed by ELISA was described above.
RAM2 spz-invasion blocking antibodies: RAM2 monoclonal antibodies were kindly provided by Noah Sather at Seattle Children's Research Institute. RAM2 antibodies were produced and puri ed as described [41]. For spz-invasion studies, mice were injected IP with 150 µg of RAM2 or matched isotype control 24 hours before RAS immunization. Two hours post immunization, blood was collected via submental bleed and serum was isolated to quantify the amount of antibody circulating via ELISA, using RAM2 as a standard curve as previously described [42]. Serum was serially diluted over a range of 1:25 to 1:1,476,225 and binding was determined as described above with goat anti-mouse IgG-HRP (Southern Biotech, #1015-05). Standard curves for RAM2 were generated by nonlinear regression (log[agonist] vs response[three parameters]) in GraphPad Prism (San Diego, CA). Serum antibody concentrations were quanti ed by interpolating the average values from three different dilutions along the sample binding curve to the corresponding standard curves and multiplying by the dilution factor to determine the nal concentration.

Parasite burden reverse transcription polymerase chain reaction (RT-PCR)
To quantify liver burden, half of the liver was excised, pulverized by bead beating into NucliSENS lysis buffer (bioMérieux), and nucleic acid was extracted as previously described [20,43]. RNA was subjected to RT-PCR with the SensiFAST™ Probe Lo-ROX Kit (Bioline, London, UK) using a mouse GAPDH RT-PCR assay (IDT Inc, Coralville, IA) multiplexed with a Pan-Plasmodium 18S rRNA assay on a QuantStudio 5 real-time PCR machine (Thermo Fisher Scienti c) as described [44]. Plasmodium 18S rRNA copy numbers per reaction were determined using a custom lot of quanti ed Armored RNA encoding full-length Plasmodium 18S rRNA (Asuragen, Austin, TX). To quantify popliteal draining lymph node (PO dLN) burden, the left PO dLNs were excised and pooled with alike PO dLN from the same group. Pooled PO dLNs were pulverized by bead beating in NucliSENS lysis buffer and processed for RT-PCR as described above.
Liver lymphocyte Isolation and ow cytometry Liver lymphocytes were isolated by mechanical dissociation and Percoll density gradient as previously described [19,45]. Brie y, livers were excised, mashed into a single cell suspension, and intrahepatic lymphocytes were isolated. Final liver lymphocyte pellets were transferred to a V-bottom 96-well plate for blocking, staining, and xing for ow cytometry. All antibodies and staining conditions were as previously described [19,20] and reagents are listed in Supplementary Table 1. Representative gating strategy is shown in Supplementary Fig. 6. Flow cytometry was conducted on the LSRII instrument (BD Biosciences), and data were analyzed with FlowJO version 10.7.1 (BD Biosciences).

Ex vivo IFN-γ ELISPOT
PyCSP peptide (SYVPSAEQI) was synthesized by Genemed Synthesis and reconstituted in DMSO. Mouse IFN-γ ELISPOT (eBioscience) was conducted by stimulating 5×10 5 splenocytes with CSP peptide (or DMSO vehicle control) at 1 µg/ml for 18 hr at 37°C and developed following manufacturer guidelines as reported previously [19,46]. The number of spot-forming units (SFU) in each well was calculated using an ImmunoSpot 5.1 Analyzer (Cellular Technology Limited, OH). SFU were normalized to DMSO control wells and SFU per million splenocytes were reported. nCounter® gene expression Gene expression analysis was performed using the NanoString nCounter® Mouse Host Response Panel. Liver samples were prepared as described above for RT-PCR with n = 3 mice per group. Total RNA was extracted on the EasyMag system (bioMérieux) and the concentration was estimated with Nanodrop (Thermo Fisher Scienti c). RNA (100 ng) was prepared for gene expression analysis at the Fred Hutchinson Cancer Research Center Genomics & Bioinformatics Core (Seattle, WA). Brie y, RNA samples were mixed with biotinylated capture and orescent reporter probes that were hybridized at 65°C for 12-16 hours. Hybridized samples were run on the NanoString nCounter® Mouse Host Response Panel using the recommended manufacturer protocol. After data collection, the nCounter® .RCC les were imported into nSolver Analysis Software 4.0 for review of quality control metrics, and the panel of housekeeping genes and positive controls was used to compute the normalization factor. Further data analysis was performed in RStudio version 2022.02.01 + 461 with R version 4.1.3. The normalized count matrix was evaluated for outliers using principal component analysis and no outliers were identi ed. Log 2 transformed normalized counts per million were assessed for differential expression for ~ 0 + vaccine using limma version 3.50.3 [47]. Pairwise contrasts were performed for each vaccine group (IV-RAS, ID-RAS, ID-RAS + 7DW8-5) and control (ggCSP only). Signi cant genes were de ned at FDR < 0.05 with BH correction and an absolute log 2 fold change > 1 (Supplementary File 1). Selected pathways from MSigDB hallmark and KEGG collections [48, 49] were utilized to visualize differentially expressed genes.

Statistics
Comparisons of parasite burden RT-PCR, ow cytometry, and ELISPOT groups were done using nonparametric Kruskal-Wallis one-way analysis of variance with Dunn's multiple comparisons test. ELISA data was analyzed with non-parametric Mann-Whitney test unless otherwise speci ed in the gure legend. Protection data was evaluated using Fisher's exact test. All groups were compared against the ggCSP prime and 2x10 4 IV RAS trap positive control as a benchmark. Error bars in gures are reported as standard deviation (SD) of the mean with individual mouse samples shown if applicable. All p-values and individual experiment statistics are listed in corresponding gure legends. Statistical signi cance was de ned as p < 0.05. Prism GraphPad Prism 9.1.2 Software (San Diego, CA) was used for all calculations, unless noted otherwise.

Results
Glycolipid adjuvant 7DW8-5 potentiates prime-and-ID RAS trap vaccination: Consistent with previous reports [11], we found that ggCSP prime-and-ID RAS trap using standard ID injection volumes (STV) did not protect BALB/cJ mice against Py spz challenge ( Fig. 1A-B). Although 7DW8-5 appears to be dispensable for IV-RAS in prime-and-trap [20], we hypothesized that the adjuvant could improve the e cacy of ID-RAS by helping to overcome the tolerogenic environment of the skin [25,30,50]. To investigate this, mice were trapped with 2x10 4 ID-RAS +/-7DW8-5 and then challenged four weeks later with 1x10 3 IV-administered Py spz (IV-spz). We found that protection induced by ID-RAS was signi cantly improved from 10-50% by the addition of 7DW8-5 (Fig. 1B). Additionally, protection was further improved to 80% by decreasing the administration volume from 50 µL to 10 µL, which was not signi cantly different from the 100% protection achieved by IV-RAS trap (Fig. 1B). This suggested that ID-RAS trapping could be effective in prime-and-trap when combined with the potent adjuvant 7DW8-5. Next, we sought to determine if the ID-RAS dose could be de-escalated while maintaining high levels of sterile protection, as was observed for IV-RAS [20]. However, reducing the dose of ID-RAS to 5x10 3 or 5x10 2 in 10 µL completely abrogated protection, despite the presence of the adjuvant (Fig. 1C). Taken together, this data demonstrates that prime-and-ID-trap is signi cantly improved by 7DW8-5 and by decreasing the ID injection volume to 10 µL, but that these changes are insu cient to de-escalate the ID-RAS dose.
Prime-and-ultra-low volume 7DW8-5-adjuvanted ID-RAS trap completely protects mice against Py spz challenge: Previous studies found that fewer ID-RAS home to the liver compared to IV-RAS and suggest this as a primary reason why ID-RAS was less effective [28,29]. We hypothesized that differential parasite liver burdens after RAS administration could be responsible for the difference in protection observed when trapping with ID-RAS in 50 µL versus 10 µL. Moreover, since spz are known to migrate out of the skin in a process that requires surface contact [33], we reasoned that by further reducing the volume used for ID-RAS, we could improve the motility of the spz to allow them to more effectively migrate out of the skin and home to the liver. To investigate the impact of injection volume on ID-spz liver burden, we coadministered de-escalating doses of the ID-RAS + 7DW8-5 trap in ultra-low volumes (ULV) of 2.5 µL. We found that 100% of the mice trapped with 2x10 4 ULV ID-RAS + 7DW8-5 were protected against spz challenge (Fig. 1D). Additionally, the dose of ID-RAS could be reduced four-fold to 5x10 3 RAS with only a modest loss of protection. However, protection was completely lost when the dose was reduced to 5x10 2 RAS, which suggests that the number of ID-RAS required for protection in this model is between 5x10 2 and 5x10 3 . This data demonstrates that prime-and-ULV ID-RAS + 7DW8-5 trap vaccination is equivalently protective at four weeks to our previously established prime-and-IV-RAS trap strategy [20].
To con rm that 7DW8-5 was not detrimental to spz viability, we examined if the co-administration of ULV ID-RAS and 7DW8-5 impacted the number of spz that reached the liver. Previous ID-spz studies demonstrated that ID-spz travel to the liver via lymphatic and vascular systems, with a signi cant portion detectable in the draining lymph node [51]. To investigate these relevant tissue sites, naïve mice were immunized with 2x10 4 ULV ID-RAS +/-7DW8-5. Four hours later, livers and the ipsilateral popliteal draining lymph nodes (PO dLN) were harvested to quantify parasite liver burden by RT-PCR. The parasite liver burden was found to be similar across all groups, but ULV ID-RAS groups had substantially higher parasite loads in the PO dLN compared to IV-RAS (Fig. 1E). In our model, a completely protective ULV ID-RAS dose was found to be between 5x10 2 and 5x10 3 parasites. To estimate the minimum vaccine liver burden needed to protect animals, we compared parasite liver burdens in mice immunized with 5x10 3 or 5x10 2 ULV ID-RAS and found a minimum protective threshold of ~ 3x10 5 Plasmodium 18S rRNA copies per liver (Supplementary Fig. 3). This data suggests that 7DW8-5 does not impact spz homing or liver invasion and that equivalent high numbers of parasites invade the liver following 2x10 4 IV-RAS or ULV ID-RAS. Next, we compared the parasite liver burden following IV-or ID-spz challenge in a STV or ULV. We found that both IV-spz and ULV ID-spz yielded similar numbers of parasites in the liver, but STV ID-spz parasite load was signi cantly lower (Fig. 1F). Together, this data validates that ID-spz utilize lymphatics and vascular systems to home to the liver and that when injected in an ULV, ID-RAS reach the liver in equivalent numbers as IV-RAS.
Finally, we asked whether active spz motility in the skin and during liver invasion was critical for protection for ULV ID-RAS + 7DW8-5. Non-motile, heat-killed spz (HK-spz) cannot actively migrate, do not invade hepatocytes, and do not achieve sterile protection against IV-spz challenge in mice [20,52]. Similarly, here we found that mice trapped with IV-or ID-HK-spz +/-7DW8-5 did not provide signi cant protection against spz challenge (Supplementary Fig. 4). This data con rms the critical importance of spz motility for prime-and-trap vaccination.
7DW8-5 potentiates ultra-low volume repeated ID-RAS only vaccination: RAS-only vaccines administered by direct venous inoculation 3-5 times are a benchmark experimental malaria vaccination strategy that achieves sterile protection in mice and humans (reviewed in [5,53]). Thus, we investigated if ULV ID-RAS was compatible with repeated RAS-only vaccination. To assess this, mice were immunized with 2x10 4 ULV ID-RAS +/-7DW8-5 three times at one-month intervals. Repeated IV-RAS routinely achieves 100% sterile protection in the BALB/cJ mouse model and was used as the benchmark in this experiment [53]. Here repeated dosing of ULV ID-RAS + 7DW8-5 was as protective as repeated IV-RAS ( Supplementary Fig. 5). Thus, using the same spz dose, ID-RAS is as equivalently protective as IV-RAS in both prime-and-trap and repeated RAS-only vaccination strategies.
The number of CD69 + /KLRG1 lo /CSP-tet + Trm cells were similar in all the immunized groups, but the number of CD69 + /CXCR6 hi /CSP-tet + Trm cells were signi cantly reduced in the ULV ID-RAS group compared to the ULV ID-RAS group with 7DW8-5 ( Fig. 2A-C, Supplementary Fig. 6B). Additionally, we found that the total number of CD44 hi /CD62L lo activated CD8 + T cells in the liver were signi cantly reduced in the ID-RAS group compared to the ID-RAS + 7DW8-5 group (Fig. 2D, Supplementary Fig. 6B). This data suggests that the high parasite burden observed following IV-RAS or ULV ID-RAS +/-7DW8-5 induces a high-frequency of CSP-speci c liver CD8 + T cells. However, since equivalently high numbers of CD69 + /KLRG1 lo /CSP-tet + Trm cells were observed for treatments that differed in protection outcomes in the challenge experiments above, this suggests that CD69 + /KLRG1 lo /CSP-tet + Trm cells de ned by phenotypic surface markers alone may be insu cient to explain protection. Consistent with other malaria vaccination studies in rodents, in our model CD69 + /CXCR6 hi /CSP-tet + de ned Trm cells may be especially critical for protection [18,54]. We also determined if the different administration routes impacted CD8 + T cell responses in the spleen. However, similar with previous reports we found that that unlike CD8 + T cell responses in the liver, the splenic responses did not correlate with protection [27] (Fig. 2E). Taken together, our ndings corroborate previous work suggesting liver CSP-speci c CD8 + Trm cells are induced by RAS vaccination and are likely the most important immune cell populations for protection in mice.
Prime-and-7DW8-5 adjuvanted ID-RAS trap induces in ammatory innate immune responses in the liver: CD8 + T cells but not iNKT cells are critical for protection from spz challenge following RAS vaccination in mice [11,55]. However, the immunostimulatory mechanism by which 7DW8-5 acts is through binding CD1d-expressing APCs and activating iNKT cells, so we investigated if iNKT cells at the time of challenge were required for protection [56]. We depleted or blocked CD8 or CD1d before challenge and found that protection was completely lost when CD8 + cells were depleted but was not impacted by the signi cant reduction of CD1d cells (Fig. 3A-B, Supplementary Fig. 7). Thus, we con rmed that prime-and-ID trap protection is likely driven primarily by CD8 + cells.
Previous studies have shown that IV-administered 7DW8-5 induced a potent and transient spike of systemic IFN-γ (and to a lesser extent IL-4) in mouse blood [20], but intramuscular (IM) administration of 7DW8-5 did not [57]. Consistent with this data, we found that ID administration of 7DW8-5 did not induce systemic IFN-γ or IL-4 ( Fig. 3C-D). However, liver IFN-γ concentrations were signi cantly increased after prime-and-ULV ID-RAS + 7DW8-5 compared to the unadjuvanted IV-RAS or ULV ID-RAS controls (Fig. 3E). This nding suggests that although ID-7DW8-5 does not induce systemic cytokine expression, it likely impacts local tissue cytokine expression. Next, we explored the key factors in the liver responsible for the differential protection outcomes. We hypothesized that 7DW8-5 in uences the innate immune responses in the liver, which subsequently in uences the quality and polyfunctionality of the induced CD8 + memory T cell responses. To evaluate this, livers were harvested vaccinated animals 44 hours post-trapping to explore gene expression changes induced by 7DW8-5 in the liver. Unadjuvanted RAS immunization (IV-RAS or ULV ID-RAS) was the least immunogenic and showed no differentially expressed genes compared to the ggCSP only control animals ( Supplementary Fig. 8). However, in the ULV ID-RAS + 7DW8-5 group, we found 119 and 154 differentially-expressed genes (FDR Adj. P ≤ 0.05 and log2 fold change of ± 1) compared to ggCSP only and ULV ID-RAS groups respectively (Fig. 3F-G). Most notably, genes associated with interferon signaling, natural killer cytotoxicity, and antigen processing were signi cantly upregulated in the 7DW8-5 groups (Fig. 3H). Parasite liver burden was also measured at the time of transcriptomic analysis sampling and showed a signi cant decrease of Plasmodium 18S rRNA copies in the 7DW8-5adjuvated ULV ID-RAS group compared to IV-RAS ( Supplementary Fig. 8). This data suggests that the kinetics of parasite clearance in the liver differ between ULV ID-RAS + 7DW8-5 and IV-RAS, which we propose is driven by the in ammatory effects of the adjuvant. Taken together, this data indicates that coadministration of ID RAS + 7DW8-5 drives the immune environment in the liver toward a pro-in ammatory state that may be more favorable for CD8 + T cell memory formation.
PyCSP antibodies induced by priming against non-repeat regions are not detrimental to ID-RAS trapping: All experiments thus far used the well-characterized and immunogenic CSP epitope (SYVPSAEQI, presented on H2-K d MHC) for ggCSP priming, but this vaccine does not induce anti-CSP IgG antibodies ( Supplementary Fig. 9). Vaccination with full length CSP protein is important for increasing epitope diversity and will likely be required for translation of the prime-and-trap vaccine strategy. However, it was not yet clear if antibodies induced by full-length CSP priming would be detrimental to ID-RAS trap since anti-spz antibodies are known to be active in the dermis [58]. The major repeat region of CSP binds the majority of potent spz neutralizing antibodies [59], so we rst cloned the full-length CSP gene -without the major repeat region -into our plasmid backbone (ggCSP full-length no repeat (FL NR)) ( Supplementary Fig. 1). The intention of this construct was to maximize the antigenic landscape while eliminating the target of the most potent spz neutralizing antibodies. To evaluate antibody responses to priming, we compared anti-CSP antibodies induced by ggCSP (epitope), ggCSP (FL NR), or the pUb.3 plasmid backbone without the CSP insert (control DNA) via ELISA. As expected, only the mice immunized with ggCSP (FL NR) produced anti-CSP antibodies on day 28 ( Supplementary Fig. 9), which due to the design of the ggCSP (FL NR) construct could be attributed to epitopes outside the repeat region.
Next, we investigated if these priming-induced antibodies targeting epitopes outside the major repeat region could impact the number of ID-RAS that reached the liver. We harvested livers and PO dLNs from mice primed mice with ggCSP (FL NR) and trapped with ULV ID-RAS +/-7DW8-5 to compare the parasite burdens and evaluate spz exit from the skin. Although parasite liver burden was signi cantly reduced in the ULV ID-RAS trap groups compared to IV-RAS group, the levels were still relatively high and well above our de ned protective threshold (Fig. 4A-B). This data suggests that the priming did indeed induce antibodies against the non-repeat regions of CSP that could impact ID-RAS homing to the liver, but that this impact was relatively minor. We hypothesized that the minor reduction in liver burden would not impact protection. Indeed, we found that similarly high levels of protection were achieved in ggCSP (FL NR) primed animals as observed in the ggCSP (epitope) primed mice despite trapping in the presence of anti-CSP antibodies and reduced liver burdens (Fig. 4C). Importantly, the trapping dose could still be reduced four-fold without a signi cant loss of protection (Fig. 4C). To evaluate the durability of protection, mice were similarly immunized, and protection was assessed four months post trapping.
Strikingly, all mice were equivalently highly protected from spz challenge in both the high (2x10 4 ) and low (5x10 3 ) dose ULV ID-RAS + 7DW8-5 groups (Fig. 4D). This data demonstrated that antibodies against the non-repeat regions of CSP induced by priming with ggCSP (FL NR) were not detrimental to IV-RAS or ULV ID-RAS +/-7DW8-5 trapping.
High titers of exogenously-administered anti-CSP repeat region spz neutralizing mAb inhibit prime-and-trap vaccination: Attenuated spz vaccines are more effective in malaria-naïve individuals (reviewed in [3]), which may in part be due to pre-existing antibodies in malaria-experienced individuals neutralizing vaccine spz before they can reach the liver. Thus, we sought to evaluate a scenario where high titers of pre-existing anti-spz antibodies were present prior to prime-and-trap vaccination. The major repeat region of CSP is the target of the most potent spz neutralizing antibodies [59], and these antibodies can be found in varying concentrations in naturally-exposed individuals [60, 61]. In a nal set of experiments, we therefore investigated if prime-and-trap would still be effective if RAS were administered in the presence of high titers of potent pre-existing spz neutralizing antibodies. RAM2 is a spz neutralizing monoclonal antibody (mAb) that binds PyCSP with high a nity and induces high rates of sterile protection against mosquito bite challenge in mice [41]. Here, we examined the impact of immunizing in the presence of high titers of RAM2. Mice were primed with ggCSP (FL NR) and trapped with 2x10 4 IV-RAS or ULV ID-RAS + 7DW8-5 (Fig. 5A). Importantly, 24 hours prior to trapping, 150 µg RAM2 or matched isotype control mAb were administered IP. Protection induced by prime-and-trap was completely abrogated by the presence of high titers of RAM2 antibodies regardless of whether the RAS trap was delivered by IV or ULV ID (Fig. 5B). Circulating anti-CSP mAb titers were con rmed to be ~ 40 ng/µL at the time of immunization by ELISA (Fig. 5C). To further elucidate the impact of RAM2 on vaccine spz, we measured the parasite liver burden and found that RAM2 did not signi cantly reduce Plasmodium 18S rRNA copies in the liver of the IV-RAS group, but signi cantly reduced the liver burden of the ULV ID-RAS group (Supplementary Fig. 9). Taken together, this data suggests that high titers of spz invasion blocking antibodies may interfere with primeand-trap or attenuated spz vaccine e cacy, but notably, sterile protection was similarly impacted in both IV-and ULV ID trapping groups.

Discussion
There were over 600,000 malaria deaths in 2021, highlighting the importance of a more effective vaccine that can prevent clinical manifestations and stop further transmission. Decades of pre-clinical and clinical studies of RAS vaccines have demonstrated the safety, feasibility, and e cacy of this vaccine strategy [4,8,9,11], but efforts to simplify and improve administration may further improve the impact of spz vaccines. ID vaccine administration is of growing interest due the increased immunogenicity and dose sparing potential [62]. A systematic review and meta-analysis found that ID immunization is dosesparing for many non-malaria infectious diseases as compared to IM or subcutaneous (SC) administration (reviewed in [23]). However, IV administration of RAS is much more e cient than IM or SC administration and ID-RAS vaccination has previously required ~ 7X higher doses to reach equivalent protection as IV-RAS [11]. Here, we explored two methods to increase the e cacy of prime and ID-RAS vaccination: 1) reduction in the administration injection volume, and 2) use of a glycolipid adjuvant. We demonstrate that prime-and-trap with an equivalent dose of ID-RAS is as effective as IV-RAS when coadministered in an ultra-low volume with the glycolipid adjuvant 7DW8-5. Thus, both microvolumes and adjuvanting were critical for the success of ID-RAS trap vaccination.
In human and mouse studies, ID-RAS vaccine failures were attributed to regulatory cellular responses [25] and low parasite burdens in the liver [28,29]. Our data supports both hypotheses. First, ggCSP priming followed by co-administration of ID-RAS + 7DW8-5 signi cantly improved protection from spz challenge. Glycolipid adjuvants, including 7DW8-5, bind CD1d expressing APCs and are known to induce a cascade of immune cell activation [35]. In our model, the 7DW8-5 adjuvant effects appeared to be necessary to modulate a favorable pro-in ammatory immune environment in the liver. Signi cant levels of protection were never achieved in our hands after ID-RAS immunization without 7DW8-5. This nding is supported by the previous literature also showing that modulation of the immune environment with adjuvants or epidermal disruption improves non-IV RAS administration [50,63,64], and that adjuvants or other proin ammatory modulating factors are likely required to overcome tolerogenic skin responses and/or regulatory liver responses for e cacious ID-RAS vaccination [25,30]. Second, we found that protection achieved from prime and ID-RAS + 7DW8-5 trap vaccination could be further improved by reducing the ID injection volume. Others have also noted that lower volumes may improve the migration capacity of spz out of the dermis [29,65]. In line with prior work, we found that ULV ID-RAS signi cantly increased the number of parasites that reached the liver compared to STV ID-RAS. However, achieving parasite liver burdens equivalent to IV dosing was not su cient for protection. Thus, it seems that the combination of high parasite liver burden and a pro-in ammatory liver immune environment is required for ID-RAS vaccination.
Inducing high levels of malaria-speci c CD8 + Trm cells in the liver is required for RAS vaccine e cacy [18]. Our data suggests that regardless of administration route or inclusion of 7DW8-5, all our tested prime-and-trap vaccine strategies induced KLRG1 lo -de ned CD8 + Trm cells in the liver. Yet, in challenge experiments, the same treatment groups did not show equivalent protection. However, and consistent with other studies, there were reduced numbers of CXCR6 hi -de ned CD8 + Trm cells in the livers of the unadjuvanted ID-RAS mice compared to adjuvanted mice [18,27], which did correlate with protection outcomes. CXCR6 has been implicated as a key liver homing marker that may be critical for memory T cell maintenance in the liver [54]. Thus, CXCR6 hi -de ned CD8 + Trm cells may be more important for sterile protection in our model. It is tempting to speculate that the quality and functionality of the CD8 + Trm cells is driving the protective differences, but the data cannot de nitively address this question at this time. In addition to the number of CD8 + Trm cells shown here, our ndings warrant future exploration into polyfunctionality of vaccine-induced CD8 + Trm cells in the liver.
At the time of challenge, we found that CD8 + cells, but not CD1d-expressing cells, were critical for sterile protection. While a potential limitation of our study is that we did not achieve full CD1d cell neutralization, the data nonetheless agrees with several other studies in CD1d knockout mice that also concluded CD1d was dispensable at the time of challenge for RAS vaccine e cacy [18,66]. We propose that CD1dexpressing cells are critical for ID-RAS + 7DW8-5 trapping to bind 7DW8-5 and induce a strong proin ammatory immune response to activate and form CD8 + Trm cells. Then, if induced correctly, liver CD8 + Trm cells may be su cient for protection. In this model, we propose that CD1d cells are needed for an optimal immune response to the trapping vaccine, but not for sensing parasites or activating CD8 + T cells at the time of challenge. Given the clear importance of CD8 + T cells for conferring protection, we also investigated the events during vaccination to that gave rise to either protective or non-protective responses. Innate immune responses during vaccination are known to be critically important for shaping the subsequent adaptive response, including the quality and the durability of CD8 + T cell responses [66-68]. Our targeted gene expression studies using the Nanostring platform provided helpful insight into the immune response in the liver after trapping. These studies revealed several key ndings. First, despite the high RAS dose used for immunization, Sanaria-produced aseptic, cryo-RAS are highly puri ed and did not induce innate in ammatory responses in the liver at the 44-hour timepoint regardless of administration route. Second, the addition of 7DW8-5 completely altered the innate response to trapping in the liver, with interferon signaling and other pro-in ammatory associated pathways signi cantly upregulated in prime and ULV ID-RAS + 7DW8-5 vaccinated mice in comparison to unadjuvanted groups. Based on these data, we speculate that interferon signaling and pro-in ammatory responses at the time of trapping likely result in a recruitment of leukocytes, an increase in antigen processing and presentation, and enhanced memory CD8 + Trm cell formation. We also note that other groups have studied adaptive regulatory cellular responses to ID-RAS and detected higher CD4 + regulatory immune responses and lower CD8 + T cell activation seven days post-spz administration [25]. Unfortunately, we did not perform gene expression analysis at an equivalent timepoint, and future studies could address whether 7DW8-5 overcomes these regulatory responses at seven days.
Antibodies also play an important role in pre-erythrocytic vaccine protection. Previous studies have suggested that the majority of antibodies act to inhibit spz in the skin [58], but increasingly the importance of anti-spz antibodies in mediating clearance of parasites outside of the skin is being appreciated [69]. We hypothesized that ID-RAS vaccines would be inhibited to a greater extent by anti-spz antibodies compared to IV-RAS, and we found that this was indeed the case. Using ggCSP (FL NR) priming, the liver burden of ID-RAS (but not IV-RAS) was signi cantly reduced by the anti-CSP antibodies induced by priming, but protection was unaffected. However, regardless of the administration route, protection was signi cantly impacted by the presence of high titers of potent anti-CSP repeat region mAb exogenously administered prior to RAS trapping. This observation was unexpected as we hypothesized that protection induced from ID-RAS would be more impacted by high titer mAb than IV-RAS. Nonetheless, our nding is supported by another research group that found mosquito bite (MB) administered spz are more infectious than IV-spz and that high titers of mAb blocked IV-spz but not MB-spz [70]. Therefore, more studies may be warranted to understand the impact of anti-spz antibody responses to both primeand-IV trap and IV-RAS only vaccines. Such studies could provide important information about the levels of circulating pre-existing antibodies that inhibit successful spz vaccination.
Finally, one of the key ndings here is that administration in an extremely low volume is critically important for successful ID-RAS vaccination. We advise that these ultra-low volumes will still be necessary when scaling up ID-RAS to larger animal models or humans. Inoculation in ultra-low volumes improves spz motility in the skin and allows spz to e ciently invade blood vessels and lymph to home to the liver and dLN, respectively [34]. In our report, 2.5 µL was selected as the smallest volume that could be reliably prepared in the research laboratory for pre-clinical mouse injections. This volume is very low compared to standard ID-administered vaccines (50-100 µL), but still higher than the estimated mosquito saliva injection of < 1 µL [71]. Additionally, reducing the injection volume to more closely mimic those occurring during mosquito probing may further improve ID-RAS. Studies with PfSPZ Vaccine and PfSPZ Challenge have shown that direct venous inoculation of 0.3-0.5 mL of PfSPZ through a 25-gauge needle is extremely well tolerated, simple, and reliable when administered by personnel after minimal training [72]. Conceptually, ID administration appears easier, but reproducibly injecting even 50-100 µL ID at an accurate depth and volume with a standard single-needle syringe can be challenging [73]. However, accurate and reliable ID injection may be possible through the development of a microarray needle patch or another as-yet-to-be-developed administration device. Moreover, without a dedicated administration device, ULV ID-RAS injections could further complicate administration for larger scale vaccine implementation. Thus, engineering innovations like microarray patches could revolutionize ID-RAS administration in the eld and allow simple, quick, and pain-free administration of ULV ID-RAS.
In summary, the use of ultra-low volumes for ID-RAS administration signi cantly improves the number of vaccine parasites that home to and invade the liver. For prime-and-trap vaccination, the combination of both 7DW8-5 and ULV ID-RAS at the trapping step is required for complete protection from spz challenge. Taken together, prime-and ULV ID-RAS + 7DW8-5 trap is a highly effective vaccine in mice that has signi cant translational potential. Combined with the recent report of in vitro production of Plasmodium falciparum sporozoites [74], our insights about lower administration volumes and adjuvants provide a potential path forward for advancing pre-erythrocytic malaria vaccines.

Declarations Acknowledgments
We thank Tess Seltzer, Veronica Primavera, Cecilia Kalthoff, and Alexis Kaushansky (Seattle Children's Research Institute) for assistance and support of Py-infected mosquito production and the NIH Tetramer Core Facility (contract number75N93020D00005) for providing PyCSP monomers. We thank the veterinary staff of the UW Department of Comparative Medicine. We also thank Sanaria for the cryopreserved Pysporozoites.   High titers of exogenously-administered spz neutralizing mAb inhibit prime-and-trap vaccination A) Experimental design of prime-and-trap studies. B) Results of protection studies after challenge with 1x10 3 WT puri ed Py spz administered four weeks after trapping with RAS +/-7DW8-5 administered IV or ID ULV (2.5 μL, X2 injections). RAM2 or isotype control mAb was injected IP into mice 24 hours prior to trapping as indicated. Protection data from N=10 mice across two independent experiments and analyzed with Fisher Exact Test, **p<0.01.