Manipulating Zika virus RNA tertiary structure for developing tissue-specific attenuated vaccines

Abstract

Traditional live attenuated vaccines (LAVs) are typically developed through serial passaging or genetic engineering to introduce specific mutations or deletions. While viral RNA secondary or tertiary structures have been well-documented for their multiple functions, including binding with specific host proteins, their potential for LAV design remains largely unexplored. Herein, using Zika virus (ZIKV) as a model, we demonstrate that targeted disruption of the primary sequence or tertiary structure of a specific viral RNA element responsible for Musashi-1 (MSI1) binding leads to a tissue-specific attenuation phenotype in multiple animal models. The engineered MSI1-binding-deficient ZIKV mutants (MBD) maintained full competence in MSI1-deficient tissues but were significantly restricted in ZIKV-vulnerable tissues (brain, testis, eye and placenta) and exhibited substantially reduced vertical transmission in mice. Importantly, a single immunization with MBD ZIKV induced robust immune responses and conferred protection against ZIKV challenge in both mice and non-human primates. Thus, our study demonstrates that manipulating viral RNA structures that interact with host proteins represents a powerful platform for developing the next generation of LAVs against emerging viruses.

Keywords: Live Attenuated Vaccine; Musashi-1; RNA Structure; Zika Virus.

Figure 1. MBD ZIKV is specifically attenuated in tissues vulnerable to ZIKV.

(A) Msi1 mRNA expression levels (nTPM, normalized transcripts per million) in different human tissues. Data were obtained from the HPA and GTEx datasets (https://www.proteinatlas.org/ENSG00000135097-MSI1/tissue). For Retina (n = 1), the value represents nTPM derived from the GTEx dataset. For Rectum (n = 1), the value represents nTPM derived from the HPA dataset. For other tissues (n = 2), values represent nTPM derived from both HPA and GTEx datasets. Error bars represent mean ± SD. (B) Relative Msi1 mRNA levels in different tissues of 4-week-old A129 mice determined by RT-qPCR. Data are the mean ± SD. Two-sided Student’s t test. n = 3. n represents biological replicates. (C) MSI1 protein levels in different organs of mouse determined by Western blotting. The Actin and Tubulin act as loading controls. (D) Schematic illustration of the generation of the MBD ZIKV that specifically attenuated in tissues vulnerable to ZIKV. The secondary structure of the stem-loop containing MBS1 and MBS2 is shown, where the MBS1 and MBS2 motifs are presented by letters. The MBS2 tertiary structures of WT and MBD2 are represented as ribbon diagrams. MBS, MSI1 binding site; MBD, MSI1 binding deficient. (E–H) Viral titers in organs of infected A129 mice. Four-week-old A129 mice were infected with 1 × 10⁴ PFU of WT or MBD ZIKV. Organs and blood samples from infected mice were collected and homogenized on 5 dpi. The viral titers were quantified by plaque forming assay. Data are the mean ± SD. n = 5. n represents biological replicates. Two-sided Student’s t test. **P < 0.01, ***P < 0.001, NS, not significant ((E) MBD1 P = 0.0021, MBD2 P = 0.0010; (F) MBD1 P = 0.0011, MBD2 P = 0.0005; (G) MBD1 P = 0.0008, MBD2 P = 0.0015). (I) 4-week-old A129 mice were subcutaneously infected with 1 × 10⁴ PFU of WT ZIKV, MBD1 or MBD2. Testes were collected on 14 dpi. Testis sections were immunostained with an anti-ZIKV-E antibody. Scale bar, 200 μm. (J) Histopathological examination of testis sections from infected animals. Scale bar, 100 μm. Source data are available online for this figure.

Figure 2. MBD ZIKV exhibits highly attenuated neurovirulence in neonatal mice.

(A) Survival curves of BALB/c neonatal mice intracerebrally injected with WT ZIKV (n = 12), MBD1 (n = 12), and MBD2 (n = 11) viruses at a dose of 1000 PFU. Analysis of survival was performed using log-rank test. **P < 0.01 (MBD1 P = 0.0019, MBD2 P = 0.0029). (B) Representative images of mouse brains from infected animals on day 9 post infection (Left panel). Brain width from infected mice was quantified (Right panel). Data are the mean ± SD. n = 6. n represents biological replicates. Two-sided Student’s t test. ***P < 0.001 (P = 0.0003). Scale bar, 5 mm. (C) RNA from the ZIKV-infected whole brains was extracted and the viral RNA copies were determined by RT-qPCR. Data are the mean ± SD. n = 3. n represents biological replicates. Two-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001 (3 dpi: MBD2 P = 0.0126; 6 dpi: MBD1 P = 0.0136, MBD2 P = 0.0295; 9 dpi: MBD1 P = 0.0005, MBD2 P = 0.0046). (D, E) ZIKV-infected cortex (D) and hippocampus (hippo) (E) sections at indicated dpi were stained with anti-ZIKV-E antibody and quantification of relative intensity of ZIKV staining was analyzed. Data are the mean ± SD. Two-sided Student’s t test. (cortex, n = 3; hippo, n = 3). n represents biological replicates. *P < 0.05, **P < 0.01, ***P < 0.001 ((D), 6 dpi: MBD1 P = 0.0004, MBD2 P = 0.0002; 9 dpi MBD1 P = 0.0077, MBD2 P = 0.0082; (E), 6 dpi: MBD1 P = 0.0088, MBD2 P = 0.0029; 9 dpi MBD1 P = 0.0121, MBD2 P = 0.0045). Scale bar, 100 μm. Source data are available online for this figure.

Figure 3. MBD ZIKV is attenuated in human cerebral organoids.

(A) Human cerebral organoids (hCOs) were infected with WT ZIKV or MBD2, and the Morphology of hCOs was monitored on indicated dpi. Scale bars, 500 μm. (B) Organoid area was measured on indicated time after infection. Data are the mean ± SD. n = 3. n represents biological replicates. Two-way ANOVA, *P < 0.05, ****P < 0.0001 (18 dpi P = 0.0101, 30 dpi P = 0.0002). (C) Viral RNA loads in culture supernatants from hCOs infected with WT or MBD2 ZIKV in (A). Data are the mean ± SD. n = 3. n represents biological replicates. Two-way ANOVA, ***P < 0.001 (5 dpi P = 0.0003, 7 dpi P = 0.0008). (D) The expression of ZIKV-E protein on day 7 after infection from (A) was detected by immunostaining. Scale bar, 40 μm. Source data are available online for this figure.

Figure 4. MBD ZIKV exhibits highly restricted infection of fetal brains during vertical transmission.

(A) Immunostaining of placenta sections from embryonic stage 8.5 (E8.5)-E12.5 mice with anti-MSI1 and anti-CK7 (trophoblast cell marker) antibodies. Scale bar, 100 μm. (B) Placentas from E8.5-E12.5 mice were lysed and MSI1 protein was detected by Western blotting. Tubulin acts as a loading control. (C) Experimental scheme. At E5.5, CD-1 mice were intraperitoneally administered with 2 mg of anti-IFNAR1 antibody. At E6.5, the mice were infected with 1 × 10⁵ PFU of WT ZIKV or MBD2 via the s.c. route. At E13.5, maternal and fetal tissues were harvested and quantified for viral RNA loads using RT-qPCR. (D–H) Viral RNA loads are presented for maternal serum (D) (n = 3), spleen (E) (n = 3), maternal brain (F) (n = 3), placenta (G) (WT: n = 36; MBD2: n = 38), and fetal head (H) (WT: n = 36; MBD2: n = 38). n represents biological replicates. The dashed line indicates the limit of detection (LOD) of the assay. Data are the mean ± SD. Two-sided Student’s t test. *P < 0.05, ****P < 0.0001 ((F) P = 0.0497, (G) P = 1.25e-012, (H) P = 2.91e-013), NS, not significant. Source data are available online for this figure.

Figure 5. MBD ZIKV protects mice from WT ZIKV challenge.

(A) Experimental scheme. Four-week-old A129 mice were immunized through the s.c. route with 1 × 10⁴ PFU of MBD ZIKVs or PBS (negative control) (n = 5/group). On day 28 post-immunization, mice were challenged through i.p. route with 1 ×  10⁵ PFU of VEN/2016 ZIKV. (B) Weight changes on indicated day post-immunization, Data are the mean ± SD. n = 5. n represents biological replicates. Two-way ANOVA, ****P < 0.0001 (P = 1.95e-007). (C) Anti-ZIKV IgG antibody titers in mouse serum on day 28 post-immunization. Data are the mean ± SD, n = 5. n represents biological replicates. (D) Neutralizing antibody titers in mouse serum on day 28 post-immunization. The dashed line indicates the (LOD) of the assay. Data are the mean ± SD. n = 5. n represents biological replicates. (E) Post-challenge viremia on indicated days was quantified by RT-qPCR. The dashed line indicates the (LOD) of the assay. Data are the mean ± SD. n = 5. n represents biological replicates. (F, G) Survival (F) and weight changes (G) were monitored at indicated day post-challenge. (F) n = 5, n represents biological replicates. log-rank test. **P < 0.01 (P = 0.0025). (G) Data are the mean ± SD. n = 5. Two-way ANOVA, ****P < 0.0001 (P = 3.84e-015). (H) T-cell responses in vaccinated mice were measured through ELISPOT assay. A129 mice were i.p. inoculated with 1 × 10⁴ PFU of MBD ZIKVs or equal volume of PBS. ELISPOT assays specific for IFN-γ in splenocytes was performed on day 7 post-immunization. Representative images were shown in left panel and the spot forming units (SFU) were counted (right panel). Data are the mean ± SD. n = 5. n represents biological replicates. Two-sided Student’s t test. ***P < 0.001, ****P < 0.0001 (MBD1 P = 5.92e-007, MBD2 P = 0.0004). Source data are available online for this figure.

Figure 6. Immunogenicity and protection efficacy of MBD ZIKV in non-human primates.

(A) Experimental scheme. Cynomolgus macaques were infected with 1 × 10⁵ PFU of WT ZIKV (n = 2), MBD2 (n = 3) or PBS sham (n = 2) via the s.c. route. n represents biological replicates. (B, C) Viral RNA loads in sera (B) and urine (C) were measured on indicated days post immunization by RT-qPCR. Each line represents data from an individual animal in each group. The dashed line indicates the LOD of the assay. (D) Antibody neutralization titers post immunization. Source data are available online for this figure.

Figure EV1. Characterization of MSI1 expression in tissues vulnerable to ZIKV.

Brain, testis and retina sections of 4-week-old A129 mice were stained with an anti-MSI1 antibody. Scale bar, 100 μm.

Figure EV2. MBD ZIKV exhibits attenuated replication kinetics in a human retina cell line expressing MSI1.

(A) MSI1 expression in human retinoblastoma Y79 cells was detected by Western blotting (left panel) and immunostaining (right panel). Scale bar, 50 μm. (B) Y79 cells were infected with WT ZIKV, MBD1 or MBD2 (MOI = 1), and the culture supernatants were harvested at the indicated time points for detection of viral loads by plaque forming assay (left panel). Data are mean ± SD. n = 3. n represents biological replicates. Two-way ANOVA, **P < 0.01, ***P < 0.001 (48: MBD1 P = 0.0007, MBD2 P = 0.0023; 72: MBD1 P = 0.0058, MBD2 P = 0.0021). The expression of ZIKV-E protein at 48 h and 72 h after infection was detected by Western blotting (right panel).

Figure EV3. The attenuation of MBD ZIKV is independent of innate immune or sfRNA.

(A) Vero cells were infected with WT ZIKV, MBD1 or MBD2 (MOI = 0.1), and the viral RNA loads in culture supernatants at 72 h after infection were detected by RT-qPCR. Data are mean ± SD. n = 3. n represents biological replicates. Two-sided Student’s t test. NS, not significant. (B) The expression of ZIKV-E protein at 72 h after infection was detected by immunostaining. Scale bar, 50 μm. (C) A549 cells were infected with WT ZIKV, MBD1 or MBD2 (MOI = 0.1), and the culture supernatants were harvested at the indicated time points for detection of viral loads by plaque forming assay (left panel). Data are mean ± SD. n = 3. n represents biological replicates. Two-way ANOVA, NS not significant. (D, E) Relative levels of IFN-β (D) and IFIT1 (E) mRNA of infected A549 cells at indicated times after infection determined by RT-qPCR. The value of mock infected at 6 h post infection was set as 1. Data are the mean ± SD. n = 3. n represents biological replicates. Two-sided Student’s t test. NS not significant. (F) The expression of MSI1 protein of SH-SY5Y and A549 cells was detected by Western blotting. (G) The sfRNA expression in SH-SY5Y and A549 cells after infection was detected by Northern blotting.

Figure EV4. Characterization of MSI1 expression in the placenta during mouse pregnancy.

Overview of immunostaining of mouse placenta sections at embryonic stage 8.5 (E8.5), E10.5, and E12.5. Tissue sections were stained with anti-MSI1(red) antibody and anti-CK7 antibody (green), a trophoblast cell marker. Nuclei were stained with DAPI (blue). The boxed areas are shown at higher magnification in Fig. 4A. Scale bar, 500 μm.

Figure EV5. MBD ZIKV exhibits attenuated replication kinetics in a human placental cell line expressing MSI1.

(A) MSI1 expression in human placenta choriocarcinoma Bewo cells was detected by Western blotting (left panel) and immunostaining (right panel). Scale bar, 50 μm. (B) Bewo cells were infected with WT ZIKV, MBD1 or MBD2 (MOI = 1), and the culture supernatants were harvested at the indicated time points for detection of viral loads by plaque forming assay (left panel). Data are mean ± SD. n = 3. n represents biological replicates. Two-way ANOVA, **P < 0.01, ****P < 0.0001 (24: MBD1 P = 0.0017, MBD2 P = 0.0093; 36: MBD1 P = 0.000041, MBD2 P = 0.000052; 48: MBD1 P = 0.0062, MBD2 P = 0.0049). The expression of ZIKV-E protein at 36 h after infection was detected by immunostaining (right panel). Scale bar, 50 μm.