8The current leading Zika vaccine candidates in clinical testing are based on live or killed virus 9 platforms, which have safety issues, especially in pregnant women. Zika subunit vaccines, 10 however, have shown poor performance in preclinical studies. We hypothesized that Zika 11Envelope (E) protein subunit vaccines have performed poorly because the antigens tested have 12been recombinant E monomers that do not display critical quaternary structure epitopes present 13on Zika E protein homodimers that cover the surface of the virus. To test this hypothesis, we 14 engineered and produced stable recombinant E protein homodimers. Unlike the E monomer, the 15 dimer was recognized by strongly neutralizing monoclonal antibodies isolated from Zika-immune 16 individuals. In a mouse model of vaccination, the dimeric antigen stimulated strongly neutralizing 17 antibodies that targeted epitopes that were similar to epitopes recognized by human antibodies 18 following natural Zika virus infection. In contrast, the monomer antigen stimulated lower levels of 19 neutralizing antibodies directed to simple epitopes on domain III of E protein. In a mouse model 20 of ZIKV challenge, only E dimer antigen stimulated protective antibodies, not the 21 monomer. These results highlight the importance of mimicking the highly structured flavivirus 22 surface when designing subunit vaccines. The flavivirus field has a long history of using E 23 monomers as vaccine antigens with limited success. These results are applicable to developing 24 second generation subunit vaccines against Zika as well as other medically important flaviviruses 25 such as dengue and yellow fever viruses. 26 130 Figure 3: ZIKV rE D stimulates antibodies that target complex epitopes on the virion. A) Method for 131 depleting EDIII binding antibodies from mouse immune sera. Recombinant ZIKV EDIII (His-tagged) was 132 coupled to nickel beads i then incubated with immune sera ii . Magnetic pull down removes nickel beads and 133 EDIII binding antibodies iii , leaving leavind sera depleted from EDIII binding antibodies iv . B) The level of 134 EDIII-binding antibodies in the serum of mice immunized with the indicated antigens is shown as a 135 percentage of the total level of ZIKV specific IgG, as measured by ELISA of EDIII-depleted and control-136 depleted serum. C) The neutralizing activity of EDIII-depleted and undepleted sera was determined and 137 expressed as the dilution at which 50% of the virus was neutralized (Neut50). D) A blockade of binding 138 (BOB) assay was used to evaluate if mice immunized with rE M or rE D developed antibodies that blocked 139 the binding of A9E, G9E and EDE C10 human mAbs. Data points represent individual mice. Statistical140differences were determined by one-way ANOVA followed by a Tukey's test (p<0.05).