Male mice exhibit less protection after prime-and-trap vaccination

To assess the influence of sex on the efficacy on liver stage malaria vaccines, we utilized a two-dose heterologous P&T regimen¹⁶ in male and female BALB/cJ mice. This P&T regimen involves epidermal DNA plasmid vaccination by gene gun on days 0 and 2. The DNA vaccine encodes a P. yoelii circumsporozoite protein (CSP) minigene designed to maximize CD8⁺ T cell activation and minimize antibody responses¹⁶. Twenty-eight days after DNA vaccination, a single dose of 2 × 10⁴P. yoelii RAS (Py-RAS) was injected IV to trap circulating CSP-specific CD8⁺ T cells in the liver. Mice were challenged 28 days later with IV administered 0.5–1.0 × 10³ infectious wild-type P. yoelii spz (Py-WT) (Fig. 1a) and monitored for blood stage patency to assess protection. All female mice were protected (19/19), whereas only 15% of male mice (3/20) were protected (Fig. 1b). Regardless of sex, all naïve mice were susceptible to infection.

a Scheme of male (M) and female (F) BALB/cJ mice prime-and-trap (P&T) vaccination experiments. b Percentages of mice protected or not protected against challenge with 0.5 − 1.0 × 10³ Py-WT sporozoites, as measured by blood parasitemia up to day 14 post-challenge. Numbers above bars indicate numbers of protected mice out of the total mice challenged, derived from 4 independent experiments. c Representative rainbow images of luminescence in livers 44 h post infection (hpi) with 1 × 10⁴ Py–Luc sporozoites. Rainbow scales are expressed in radiance (p/s/cm²/sr). d Quantification of bioluminescent signal in log-transformed total flux (p/s) from mice at 44 hpi (2 independent replicates, n = 8–10/group). Groups in b were compared using a two-sided Fisher’s exact test, and groups in d were compared by a two-sided Wilcoxon test; only relevant comparisons are depicted. Error bars represent mean ± s.e.m; ns = p > 0.05, ***p < 0.001. Source data are provided as a Source Data file.

To determine if vaccinated male mice could achieve partial liver stage protection, we next monitored liver burden by in vivo bioluminescent imaging in P&T vaccinated mice of both sexes after virulent spz challenge. Mice were challenged 28 days after P&T using 1 x 10⁴ P. yoelii spz that constitutively express a GFP-luciferase fusion protein (Py–Luc)²⁹ and evaluated at 44 h post-injection (hpi). Naïve male and female mice were likewise challenged. Vaccinated female mice showed a 45-fold reduction in luminescent signal compared to naïve female mice, whereas vaccinated male mice showed only an 11-fold reduction compared to naïve males (Fig. 1c, d). Notably, there was no difference in luminescence signal at 44 hpi between naïve female and male mice (Fig. 1c, d), suggesting that liver stage replication of the parasite was similar for both sexes. Additionally, augmenting P&T with a more potent prime or with a co-administered Py-RAS with the glycolipid adjuvant 7DW8-5³⁰ did not overcome the low rate of protection observed in male mice (Fig. S1).

To confirm that sex-bias in protection was not unique to this mouse and parasite strain, we conducted similar P&T studies using C57BL/6 mice and P. berghei parasites for trapping (Pb-RAS) and challenge (Pb-WT). DNA encoding PbTRAP and PbRPL6 was used for priming, since these two antigens can be recognized in C57BL/6^17,31 (Fig. S2a). As expected, P&T vaccinated female C57BL/6 mice showed increased protection over vaccinated male mice (Fig. S2b, c), confirming that P&T vaccination induced a sex-bias in protection across two mouse malaria models.

Female mice generate higher density of memory CD8⁺ T cells in the liver

Given the importance of T cell responses to spz vaccines^13,32, we first sought to phenotype memory CD8⁺ T cells in male and female mice following P&T vaccination (Fig. 2a). Twenty-eight days after vaccination, we isolated liver and spleen lymphocytes and ex vivo stimulated with PyCSP^280-288 (SYVPSAEQI) peptide prior to intracellular cytokine staining (ICS) to evaluate CD8⁺ T cells at this memory timepoint. PyCSP^280-288 is a known immunodominant epitope in the Py-RAS/BALB/cJ mouse model, and responsive cells represent a major population of antigen-specific CD8⁺ T cells after P&T vaccination¹⁶. A higher total CD4/CD8 ratio was observed in vaccinated male livers compared to female livers, but no significant difference in the CD4/CD8 ratio was found in the spleen (Fig. 2b, c, S5a). Male livers weighed significantly more than female livers in both naïve and vaccinated mice (Fig. 2e); thus, liver weights were incorporated in the analysis (results expressed per gram of liver) to account for the potential impact of sex differences in physical size on T cell numbers.

a Male (M) and female (F) BALB/cJ mice were vaccinated with the prime-and-trap (P&T) regimen. 28 days later, liver cells were analyzed by flow cytometry. b Representative flow plots of CD4 and CD8 surface markers on CD3⁺ cells isolated from the liver of mice. c Ratio of CD4-to-CD8 T cells in the liver of P&T vaccinated (+) and unvaccinated mice (-) female (F) and male (M) BALB/cJ mice. d Representative flow plots of IFN-γ (top), CD107a (middle), and TNF (bottom) on CD8⁺ T cells isolated from the liver and ex vivo stimulated with PyCSP^280-288 peptide. e Liver weights of unvaccinated (open circle) or P&T vaccinated (closed circle) mice. f, g Quantification of CD8⁺ T cells responding to PyCSP^280-288 (SYVPSAEQI) peptide defined by total number of IFN-γ⁺ and/or TNF⁺ and/or CD107a⁺ producing CD8⁺ T cells per gram of liver (f) and as proportion of CD3⁺ T cells (g). h, i Quantification of proportion of IFN-γ⁺ (left), TNF⁺ (middle), CD107a⁺ (right) (h), and IFN-γ⁺ TNF⁺ CD107a⁺ (i) producing CD8⁺ T cells in the liver. Detailed gating information is included in Supplementary Fig. S3. Data for c–i are shown from 3 independent experiments (n = 14 P&T groups, n = 5–7 unvaccinated group per sex). j Representative flow cytometry plot depicting CSP-tetramer⁺CD44^hi CD8⁺ T cells Trm, Tem, and Trm phenotypes by markers CD69 (left) and CXCR3 (right). See supplementary Fig. S4 for the gating tree. k Frequency of CSP-tetramer⁺CD44^hi Tem (CD62L^–CD69^–), Tcm (CD62L⁺CD69^–), Trm (CD62L^–CD69⁺) subsets of CD8⁺ T cells in the liver. Individual p-values for significant differences are shown, colored to correspond to memory T cell subsets where relevant. l. Frequency of CSP-tetramer⁺CXCR3⁺CD62L^–CD69⁺CD44^hi cells of CD8⁺ T cells in the liver. Data for j–l are shown from 2 independent experiments (n = 8–9). Statistical significance for data in c, f–i, k–I, was determined by Kruskal-Wallis test with Dunn’s multiple comparison and data in e was evaluated with a two-sided Wilcoxon Test; error bars represent mean ± s.e.m; ***p < 0.001, **p < 0.01, *p < 0.05, ns p > 0.05. Source data are provided as a Source Data file.

With differences in liver weight by sex acknowledged, we then evaluated the vaccine-induced memory CD8⁺ T cell populations. Upon peptide stimulation, liver Trm cells generated by P&T vaccination are known to express IFN-γ, TNF, and the degranulation marker CD107a¹³. We thus evaluated the total cell number, frequency, and effector function of CSP-specific CD8⁺ T cells isolated from the liver and spleen of male and female P&T-immunized mice. First, we examined the number of total cytokine-producing CSP-specific CD8⁺ T cells defined as any cell expressing IFN-γ, TNF, CD107a, or any combination of the three (Fig. 2d). Females showed higher cell counts per gram of liver compared to males (Fig. 2f). However, when evaluating total CSP-specific liver CD8⁺ T cells as a percent of CD3⁺ T cells, cytokine-producing CD8⁺ T cell frequencies did not significantly differ between male and female mice (Fig. 2g). Second, we evaluated the relative production of IFN-γ, TNF, and CD107a within the respective CD8⁺ T cell population. Vaccinated male and female mice had similar frequencies of IFN-γ, TNF, and CD107a producing CD8⁺ T cells in the liver (Fig. 2h). Third, we evaluated a putative cytotoxic ‘functional’ memory population defined as CD8⁺IFN-γ⁺TNF⁺CD107a^{+ 33,34}. Within the liver CD8⁺ T cells repertoire, cells expressing all three markers were detected at similar frequencies in both male and female mice (Fig. 2i). In the spleen, there were also similar frequencies of IFN-γ⁺TNF⁺CD107a⁺ producing CD8⁺ T cells in male and female mice (Fig. S5e).

We also evaluated antigen-specific memory CD8⁺ T cells induced by P&T vaccination in another experiment using the H-2k^d tetramer specific for MHC class I peptide K^d-PyCSP^280-288. We defined putative Trm cells as  CSP-tetramer⁺CD62L^–CD69⁺CD44^hiCD8⁺or CSP-tetramer⁺CXCR3⁺CD62L^–CD69⁺CD44^hiCD8⁺ (Fig. 2j). P&T vaccinated females and males showed an increased frequency of CSP-specific Trm cells in the liver at 28 days post-immunization, though males did not reach significance. Using either definition of CSP-specific Trm cells in the liver, there was no significant difference in Trm cell frequency between vaccinated male and female mice as reported as a proportion of CD8⁺ T cells (Fig. 2k, l, S6d, e). In the spleen, most cells had an effector memory phenotype (Tem: CSP-tetramer⁺CD62L^–CD69^–CD44^hi), and there were no significant differences in the proportion of CSP-specific Tem cells between vaccinated male and female mice (Fig. S6d, e).

Taken together, P&T vaccination induces more antigen-specific memory CD8⁺ T cells per gram of liver by ICS in female mice compared to male mice. However, the frequency of these cells as a proportion of CD3⁺ T cells did not differ between male and female mice. Also, there were no detectable difference in effector function between male and female mice based on expression of IFN-γ, TNF, and/or CD107a. Finally, there was no detectable difference in the frequency of putative Trm cells. Thus, sex divergent outcomes in P&T vaccinated mice are potentially associated with differences in density, but not the quality, of vaccine-induced memory CD8⁺ T cells.

Sex-bias in hepatic inflammatory response to RAS immunization

We examined gene expression profiles following RAS immunization to contextualize the sex-specific microenvironment that traps and guides memory CD8⁺ T cells in the liver. The acute inflammatory response of the liver following attenuated spz immunization mediates the formation of functional hepatic spz-specific CD8⁺ T cell responses³⁵. Thus, to capture the transcriptional profile during initial innate phase and peak CD8⁺ T cell production phase, bulk mRNA sequencing was performed on whole livers isolated 44 h or 6 days post-RAS immunization. Mice were either primed with DNA encoding PyCSP or were naïve, then immunized with Py-RAS (Fig. 3a). Principal component analysis (PCA) identified sex as a major driver of variation in the transcriptome, mapping to the first principal component (PC1) that explained 35.6% of variation (Fig. 3b), even when omitting sex chromosomal-linked contributions (Fig. S7a) and stratifying by vaccination status (Fig. 3c). Differential expression analysis of mock-infected male and female mice confirmed differentially expressed genes (DEGs) in the Cytochrome P450 (Cyp) enzyme superfamily (Fig. S7b), aligning with known sex differences in the liver³⁶. DEGs were examined relative to mock-infected controls within sex to account for this baseline variability between sexes (Fig. S7d).

a Experimental schematic. Livers from mock vaccinated, prime-and-trap vaccinated (P&T), or RAS-only (RAS) vaccinated male (M) and female (F) BALB/cJ mice were collected 44 h or 6 days after the indicated vaccine for RNAseq (n = 4 per group). b Principal Component (PC) analysis of male and female transcriptomes following vaccination (containing all vaccinated and mock-vaccinated mice). c Hierarchical clustering on differentially expressed genes (FDR < 0.05 and |LogFC | > 1.5, Benjamini-Hochberg adjusted) and unique to response to vaccination (not shared with baseline differences between the mock-vaccinated male and female mice) with column annotations for sex, vaccine status, and timepoint. Euclidean distance metric was used for sample clustering. d GSEA of selected Hallmark (top) and Gene Ontology Biological Processes pathways (bottom) with genes ranked by fold-change values relative to mock injected mice by sex. NES = normalized enrichment score. Column annotations depict sex and vaccine status as designated in panel (c). e Heatmap and hierarchal clustering of genes that contain at least one differentially expressed gene in either timepoint (44 h or 6 days) (FDR < 0.01, |LogFC | > 2, Benjamini-Hochberg adjusted) and appear in the Gene Ontology Biological Processes pathway ‘T cell activation’. Column annotations depict sex, timepoint, and vaccine status. f Calculated pathway score of aggregate samples by T cell activation pathways for the 44 h and 6-day timepoint (n = 8/group) relative to mock samples (n = 4). Detailed explanation of calculation in Methods section. g Heatmap and hierarchal clustering of genes that contain at least one differentially expressed gene at the 6 days timepoint (FDR < 0.05, |LogFC | > 1, Benjamini-Hochberg adjusted) and appear in the Hallmark ‘Interferon alpha response’, or Gene Ontology Biological Processes pathways ‘Phagocytosis’ and ‘Response to chemokine’. Column annotations depict sex and vaccine status (c). h Calculated pathway score of aggregate samples by ‘Interferon alpha response’, ‘Phagocytosis’, and ‘Response to chemokine’ pathways at day 6 post-immunization (n = 8/group) relative to mock samples (n = 4). Data produced from one independent replicate. The figure f and h represent the mean value with 95% confidence intervals; **p < 0.01, *p < 0.05, ns p > 0.05.

Prior studies comparing early and late-arresting whole spz vaccines in female mice have determined that type I IFN gene signature at the time of immunization impairs hepatic CD8⁺ T cell memory via a CD8⁺ T cell-extrinsic mechanism³⁵. IFN signaling pathways have also been a documented response to wild-type^37,38 and attenuated spz^{39 in the liver}. Yet, despite a well-established female bias toward heightened interferon signaling responses upon inflammation²¹, a direct comparison between sexes following attenuated spz delivery remains to be reported. Thus, we performed gene sequence enrichment analysis (GSEA) on previously defined transcriptional patterns^40,41. During the initial innate immune response phase at 44 h post-RAS immunization, male and female mice responded similarly for several pathways, including IFN-α response(Fig. S7e, Fig. 3d). Further evaluation of differentially expressed genes identified a gene signature shared between male and female mice related to liver inflammation (Saa1, Saa3) (Fig. S7c).

The day 6 timepoint captures the transcriptome during peak T cell recruitment, proliferation, and ongoing innate immune cell recruitment in the liver following clearance of RAS from infected hepatocytes^4,32. Consistent with this, the gene ontology pathway for T cell activation was significantly induced in both vaccinated male and female mice at this timepoint (Fig. 3d), though females experienced a significantly higher response than male mice (Fig. 3e, f). The T cell activation pathway includes genes indicative of a CD8⁺ T cell response, including key genes encoding CD8, CD3, and CD28 surface markers. It is notable that there was a lack of detectable Plasmodium 18S rRNA at 6 days post-immunization, indicating that transcriptomes were driven by inflammation resulting from RAS clearance and the onset of adaptive cellular responses (Fig. S7f). Indeed, unsupervised clustering revealed that liver gene expression signatures clustered first by sex, then by timepoint and vaccine type (Fig. 3c).

Finally, further evaluation of the inflammatory environment during this T cell response at day 6 was conducted. Networks associated with responses to IFN-α were upregulated in female mice, but not in male mice at this timepoint (Fig. 3d). Further investigation found little to no induction of genes related to proteins associated with a type I IFN response (Gbp2, Gbp3, ifi44, ifit2) and CXCR3 ligands (Cxcl9, Cxcl10) in male mice (Fig. 3g). Female mice also experienced an increased IFN-γ response compared to male mice (Fig. 3d, S7g, h). Several other pathways were evaluated due to their potential to influence hepatic CD8⁺ T cell memory responses³⁸: phagocytosis was upregulated in females but not males, whereas both male and female mice induced responses to chemokines at this timepoint (Fig. 3g, h). Overall, this suggests that while male mice can form chemokine cues important for CD8⁺ T cells, there is a diminished response to inflammatory signals, like interferons, when compared to female mice.

Androgens, not estrogens, alter protection from malaria challenge

Given the known effects of sex hormones on altering adaptive immune responses, we sought to evaluate the impact of sex hormones on protection from P&T vaccination. To evaluate whether estrogens in females or androgens in males contributed to the discordance in protection by sex following vaccination, we removed the gonads in adult mice at least 14 days prior to P&T vaccination and then later quantified liver burden 44 h post-spz challenge by Plasmodium 18S rRNA RT-PCR to assess protection (Fig. 4a). For females, ovariectomy (OVX, removing estrogens and progestins) had no effect on protection outcome, with OVX females maintaining protection (Fig. 4b, S8a). For males, orchiectomy (ORX, removing androgens) led to high levels of protection, a reversal of the outcome in intact males (Fig. 4c, S8b). IFN-γ ELISPOT on splenocytes obtained at post-challenge timepoints did not demonstrate significant differences in IFN-γ producing cells between intact and OVX females or intact and ORX males (Fig. S8d). Taken together, this data illustrates that among female mice, OVX to ablate circulating estrogens does not affect protection outcomes, whereas ORX to ablate androgens prior to vaccination significantly improves protection in male mice.

a Experimental regimen with orchiectomy (ORX) in males (M), ovariectomy (OVX) in females (F), or sham (SHAM) surgery followed by prime-and-trap (P&T) vaccination. 28 days later, mice were either challenged or lymphocytes were analyzed by flow cytometry. b, c Protection following 1 × 10³ Py-WT sporozoite challenge defined as no detectable pan-Plasmodium 18S rRNA copies 44 h post-challenge. Data are shown from 2 independent experiments (n = 8–10/group). Only relevant comparisons are shown. d Liver weight of ORX and SHAM operated male BALB/cJ mice vaccinated with P&T regimen and harvest 28 days later. e Quantification of total responsive CD8⁺ T cells to PyCSP^280-288 peptide defined by total number of IFN-γ⁺, TNF⁺, and/or CD107a⁺ producing CD8⁺ T cells corrected by weight of liver. f Ratio of CD4-to-CD8 T cells in the liver. g Total number of IFN-γ⁺, TNF⁺, and/or CD107a⁺ producing CD8⁺ T cells as proportion of CD3⁺ T cells in the liver. h, i Quantification of proportion of IFN-γ⁺ TNF⁺ CD107a⁺ (h) and IFN-γ⁺ (left), TNF⁺ (middle), and CD107a⁺ (right) producing CD8⁺ T cells in the liver. Data for d–i are shown from 2 independent experiments (n = 7–9/group). j Frequency of CSP-tetramer⁺CD44^hi Tem (CD62L^–CD69^–), Tcm (CD62L⁺CD69^–), Trm (CD62L^–CD69⁺) subsets of total CD8⁺ T cell. k Quantitative evaluation of TCF1 MFI on CSP-tetramer⁺ Trm cells in the liver (CSP-tetramer⁺CD62L^–CD69⁺CD44^hiCD8⁺ T cells). l, m Proportion of PD-1⁺ (l) and LAG-3⁺ (m) on liver Trm cells. Data for j–m are shown from 2 independent experiments (n = 7–9/group). Statistical significance for data in b and c was determine by two-sided Fisher exact test; data in e–j was determined by Kruskal–Wallis test with Dunn’s multiple comparison and data in d and k–m was evaluated with the two-sided Wilcoxon Test. In j, individual adjusted p-values for significant differences are shown, colored to correspond to memory T cell subsets where relevant. Error bars represent mean ± s.e.m; box plot depicts median, interquartile range, and whiskers extending to maximum and minimum values within 1.5x of the interquartile range; ***p < 0.001, **p < 0.01, *p < 0.05, ns p > 0.05. Source data are provided as a Source Data file.

Androgens do not alter hepatic CD8⁺ T cell memory populations

Next, we characterized the impact of androgens on the memory CD8⁺ T cell repertoire in male mice after P&T immunization by ICS. Male mice underwent ORX or equivalent sham surgeries (SHAM), were P&T vaccinated, and then the vaccine response at 28 days post-immunization was evaluated as above. As previously reported⁴², we observed that livers from vaccinated ORX mice were significantly smaller compared to the vaccinated SHAM mice (Fig. 4d). We analyzed the number of CSP-specific CD8⁺ T cells per gram of tissue, and did not find a significant difference between ORX and SHAM mice (Fig. 4e). There was no difference in the CD4/CD8 T cell ratio in the livers of ORX and SHAM mice (Fig. 4f). Additionally, there was no significant difference in the frequency of CSP-specific CD8⁺ T cells as a percent of CD3⁺ T cells in the liver (Fig. 4g) and the spleen (Fig. S8e, f). Next, we assessed differences in cytokine-producing CD8⁺ T cells in ORX P&T immunized mice upon re-stimulation compared to SHAM mice. Both P&T vaccinated groups had increased levels of cytokine producing CD8⁺ T cells as compared to the naïve controls, but there was no significant difference in the frequency of IFN-γ⁺, TNF⁺, or CD107a⁺ T cells, or T cells expressing all three markers in the liver between the ORX and SHAM groups (Fig. 4h, i). A similar pattern was seen when comparing the frequency of IFN-γ⁺TNF⁺CD107a⁺ T cells in the spleen (Fig. S8g). Thus, androgens do not appear to alter the total number, frequency, or function of CSP-specific CD8⁺ T cells memory cells in the liver as measured by ICS.

In a separate experiment, we characterized the impact of androgens on memory CD8⁺ T cell subsets in ORX versus SHAM male mice after P&T vaccination by tetramer staining. ORX and SHAM male mice showed similar frequencies of total memory CSP-tetramer⁺CD44^hiCD8⁺ T cells in the liver and spleen (S9a,d). Although Trm and Tem phenotypes were present in both ORX and SHAM vaccinated mice livers and spleen, there were no differences in Trm or Tem composition (Fig. 4j, S9b, e). Taken together, androgens do not appear to alter the frequency of CSP-specific CD8⁺ T cells memory subsets in the liver.

Since there was no difference in the number or proportion of CSP-specific CD8⁺ T cells in the livers of immunized ORX and SHAM male mice 28 days post-immunization, other factors must mediate the improved protection that occurs after withdrawal of androgens. We next evaluated the quality of CSP-tetramer⁺ Trm cells for markers of dysfunction. Androgen receptor signaling has been linked to a more dysfunctional T cell state via the transcription factor TCF1^43,44,45. Thus, we hypothesized that the presence of androgens may alter TCF1 expression and increase expression of co-inhibitory molecules like programmed cell death protein (PD-1) and lymphocyte activation gene 3 (LAG-3) that have also been connected to dysfunctional states⁴⁶. As expected, higher expression of TCF1 was observed on central memory CD8⁺ T cells (Tcm) in the liver than on Tem and Trm cells (Fig. S9g). However, importantly, there was no difference in TCF1, PD-1, and LAG-3 expression on tissue resident CD8⁺ T cells (defined as CSP-tetramer⁺CD62L^–CD69⁺CD44^hi CD8⁺ T cells) in ORX compared to SHAM vaccinated mice (Fig. 4k–m).

In all, there was no detectable difference between the frequency, density, functionality profiles, or exhaustion/dysfunction status, of antigen-specific CD8⁺ T cells in male mouse livers at 28 days post-immunization following ORX or SHAM surgeries.

Hormone environment at time of challenge impacts protection outcomes

The question remained: why are intact male mice not protected by vaccination? Theoretically, androgens could influence the efficacy of P&T vaccination during two potential phases: 1) the production phase of immune memory at the time of vaccination, or 2) the recall phase of immune memory at the time of challenge. To identify at what stage androgens alter protection outcomes, ORX was performed before or after P&T vaccination and then mice were challenged three weeks after the post-P&T ORX procedures to allow the hormone environment in the liver to re-equilibrate (Fig. 5a). SHAM mice likewise received surgery at the post-P&T timepoint. As in previous experiments, male mice were protected by P&T when ORX was performed prior to vaccination. Remarkably, when ORX was performed after P&T vaccination, the mice were also protected (Fig. 5b, S10). However, since ORX is irreversible, this method of hormone manipulation did not allow for evaluation of the effect of androgens uniquely at the time of vaccination.

a Schedule of orchiectomy (ORX) in male (M) mice before or after prime-and-trap (P&T) vaccination as shown. After the completion of surgery at day 56, mice were rested 21 days to establish a new hormone equilibrium and then challenged with 1 × 10³ Py-WT sporozoites. b Protection after Py-WT sporozoite challenge by pan-Plasmodium 18S rRNA RT-PCR from livers collected 44 h post-challenge. Data are shown from 2 independent experiments (n = 8–10/group). c Schedule of administration of acyline (Acy) in male mice at day 5 and 4 prior to either P&T vaccination (prior to gene gun and RAS) or prior to challenge, or both. Testosterone (T) add-back group received 3 injections of testosterone on day 3 and 1 prior to challenge and 1 day following challenge. Mice were challenged 28 – 42 days post P&T vaccination. d Protection after Py-WT sporozoite challenge by pan-Plasmodium 18S rRNA RT-PCR from livers collected 44 h post-challenge. Data are shown from 2 – 3 independent experiments (n = 10–14/group). Statistical significance for data in b and d was determined by a two-sided Fisher exact test. Only relevant comparisons are shown. Error bars represent mean ± s.e.m; ***p < 0.001, **p < 0.01, *p < 0.05, ns p > 0.05. Source data are provided as a Source Data file.

To further interrogate at which phase of vaccination and challenge androgens were exerting immune effects, we developed an acute and reversible model to suppress testosterone production in male mice. Acyline is a gonadotropin-releasing hormone (GnRH) antagonist that suppresses downstream production of testosterone in a dose-dependent manner and has previously been applied in mice⁴⁷, and humans^48,49. We optimized the dose of acyline for the BALB/cJ mouse model by selecting the dose that most consistently retained suppressed testosterone levels for over 7 days but under 14 days to ensure that hormone levels were maintained during peak immune responses of the P&T regimen (Fig. S11a, b). Next, we depleted testosterone prior to both steps of P&T vaccination, prior to spz challenge, or both (Fig. 5c). We found reducing testosterone at vaccination or challenge was sufficient to increase protection outcomes in male mice (Fig. 5d). To further confirm that testosterone at time of challenge altered protection outcomes, we added back testosterone and once again saw a reduction in protection (Fig. 5d). We further confirmed that administration of acyline, testosterone, and ORX surgery did not significantly alter liver burden in unvaccinated mice after a challenge dose of 1000 Py-WT sporozoites (Fig. S11f, g). Together with the above liver CD8⁺ T cell characterization data, these findings lead us to conclude that the hormone environment at the time of challenge is a driving factor behind limited vaccine-induced protection in intact male mice.

Androgens inhibit CD8⁺ T cell recruitment response

The mechanisms by which vaccine-induced CD8⁺ T cells enact their effector mechanisms at the time of challenge involve three mechanisms: cytolytic pathways, cytokine pathways³³, and “sensing and alarm” to facilitate recruitment of other immune cells⁵⁰. To study the direct impact of androgens on CD8⁺ T cells cytolytic enzyme and cytokine production, we used an ex vivo stimulation assay where we exogenously supplied dihydrotestosterone (DHT), the metabolically active form of testosterone, and evaluated the impact on CD8⁺ T cells in an otherwise hormone-free media environment. Intact male and female mice, and ORX male mice were P&T vaccinated (Fig. 6a), and lymphocytes were isolated from the liver and stimulated with PyCSP^280-288 peptide in the presence or in the absence of DHT (Fig. 6b). CD8⁺ T cells from ORX male mice showed a reduction in IFN-γ and Granzyme B, but not TNF and CD107a, in the presence of DHT (Fig. 6c, d). In the presence of DHT, there was no reduction in proportion of a putative functional CD8⁺ T cells expressing IFN-γ, TNF and CD107a (Fig. 6e). However, in female mice Granzyme B⁺ functional CD8⁺ T cells were reduced (Fig. 6f). Thus, overall, the presence of DHT reduced the activity of CD8⁺ T cells via inhibition of IFN-γ⁺ and Granzyme B⁺, however, this fold reduction was quite slight.

a Female (F), male (M), and orchiectomized male (ORX) BALB/cJ mice were vaccinated with the prime-and-trap (P&T) regimen. 35 days later, liver cells were analyzed by flow cytometry. b Representative flow plots of IFN-γ (top), TNF (middle top), CD107a (middle bottom), and Granzyme B (bottom) on CD8⁺ T cells isolated from the liver and ex vivo stimulated with PyCSP^280-288 peptide in hormone free condition with dihydrotestosterone added to media ( + DHT) or with ethanol control (- DHT). c, d Quantification of proportion of cytokines IFN-γ (c – left) and TNF (c -right) and cytolytic markers Granzyme B (d – left) and CD107a (d – right) producing CD8⁺ T cells in the liver. e Quantification of proportion of IFN-γ⁺TNF⁺CD107a⁺ producing CD8⁺ T cells in the liver. f Quantification of proportion of Granzyme B⁺IFN-γ⁺TNF⁺CD107a⁺ producing CD8⁺ T cells in the liver. Data are shown from 2 independent experiments (n = 10-12/group, naïve composed of n = 6 per sex). Statistical significance for data in (c–f) was determined by a paired two-sided Wilcoxon Test. Box plot depicts median, interquartile range, and whiskers extending to maximum and minimum values within 1.5x of the interquartile range; ***p < 0.001, **p < 0.01, *p < 0.05, ns p > 0.05. Source data are provided as a Source Data file.

Next, we evaluated the impact of androgens on the “sensing and alarm” function of Trm cells, which involves antigen-specific Trm cells serving as a local sensor and initiating an innate-like alarm signal to recruit more CD8⁺ T cells and other immune cells. Previous work identified a positive association between the number of CD8⁺ T cells per infected hepatocyte and P. yoelii parasite clearance⁵¹. Thus, we used immunohistochemistry to quantify inflammatory foci and CD8⁺ cells⁵² in P&T vaccinated and unvaccinated female, male, and ORX male mice challenged with Py-WT spz. For this experiment, we intentionally selected a high challenge dose of 1 x 10⁵ spz to ensure sufficient immune foci for spatial quantification (Fig. 7a, b). As expected, we observed a higher number of inflammatory foci and CD8⁺ cells at 38 – 44hpi in P&T vaccinated mice compared to unvaccinated mice challenged with Py-WT spz and mock-injected control mice (Fig. 7c). Within the vaccinated mice, female and ORX male mice had more and larger inflammatory foci compared to male mice (Fig. 7d, e). Furthermore, within each inflammatory focus, there was a higher total number and density of CD8⁺ cells in female and ORX male mice compared to intact male mice (Fig. 7f–h). Thus, the presence of androgens reduced the number and size of inflammatory foci and impaired the recruitment of CD8⁺ cells to these foci in vaccinated male mice compared to female or ORX male mice.

a Female (F), male (M), and orchiectomized (ORX) male BALB/cJ mice were vaccinated with the prime-and-trap regimen (P&T) or left unvaccinated (-). 28 days later, mice were challenged with 1×10⁵ Py-WT spz and liver was split for fixation for immunohistochemistry and RT-PCR of host genes between 38 and 44 h. b Representative inflammatory foci as determined with immunohistochemistry (IHC) in liver tissue captured under light microscope in vaccinated (P&T + Py-WT), unvaccinated (Py-WT), and mock-challenge mice where red demarks CD8⁺ cells and blue marks nuclear stain. One slice of left lateral lobe was analyzed per mouse and all inflammatory foci were counted within each lobe. c Average number of inflammatory foci per mouse corrected to tissue size evaluated (mm²) in vaccinated, unvaccinated, and unvaccinated mock-challenged mice. d, e Average number of inflammatory foci (d) and size of foci (μm²) (e) in female (F), male (M), and orchiectomized male (M ORX) mice that were vaccinated (+) and challenged with spz. f The average number of CD8⁺ cells within foci per each vaccinated, unvaccinated, and unvaccinated mock-challenged mouse. g, h. Within vaccinated and challenged F, M, and M ORX mice, the number of CD8⁺ cells in each foci and the density of CD8⁺ cells scaled to size of respective foci (CD8⁺ cells per foci size (mm²)). i Gene expression of Cxcl9, Cxcl10, Ifng in vaccinated and challenged female, male, and ORX male mice calculated as log₂ fold change relative to pooled mock-challenged mice. Data are shown from 2 independent experiments (n = 5-8/group), except for the male P&T + Py-WT ORX group that was included in one experiment (n = 5 mice). Average values were calculated per mouse (c, d, f). Statistical significance for data was determined with Kruskal-Wallis test with Dunn’s multiple comparison. Error bars represent mean ± s.e.m; box plot depicts median, interquartile range, and whiskers extending to maximum and minimum values within 1.5x of the interquartile range; ***p < 0.001, **p < 0.01, *p < 0.05, ns p > 0.05. Source data are provided as a Source Data file.

Next, having determined that androgens impact CD8⁺ cell recruitment to immune foci, we sought to confirm that there was reduced inflammation in male mice. From the same set of mice, we evaluated the gene expression of key pro-inflammatory markers Cxcl9, Cxcl10, and Ifng in the liver at the time of challenge by gene-specific RT-PCR. Previous studies have shown that production of chemokines, CXCL9 and CXCL10, and the cytokine IFN-γ correlate with attracting and enhancing T cell responses to inflammatory foci^{33,53,54,55,56,57}. We found significantly higher expression of Cxcl9, but not Cxcl10 and Ifng, in female and ORX male mice compared to male mice at the time of challenge (Fig. 7i). Taken together, this data demonstrates that the presence of androgens at time of challenge slightly reduces memory CD8⁺ T cell cytolytic capacity and cytokine production and reduces inflammation connected with Cxcl9 expression, linking to subsequent impaired recruitment of additional immune cells in male mice.