Vaccination prevents IL-1β-mediated cognitive deficits after COVID-19

Up to 25% of SARS-CoV-2 patients exhibit post-acute cognitive sequelae. Although millions of cases of COVID-19-mediated memory dysfunction are accumulating worldwide, the underlying mechanisms and how vaccination lowers risk are unknown. Interleukin-1, a key component of innate immune defense against SARS-CoV-2 infection, is elevated in the hippocampi of COVID-19 patients. Here we show that intranasal infection of C57BL/6J mice with SARS-CoV-2 beta variant, leads to CNS infiltration of Ly6Chi monocytes and microglial activation. Accordingly, SARS-CoV-2, but not H1N1 influenza virus, increases levels of brain IL-1β and induces persistent IL-1R1-mediated loss of hippocampal neurogenesis, which promotes post-acute cognitive deficits. Breakthrough infection after vaccination with a low dose of adenoviral vectored Spike protein prevents hippocampal production of IL-1β during breakthrough SARS-CoV-2 infection, loss of neurogenesis, and subsequent memory deficits. Our study identifies IL-1β as one potential mechanism driving SARS-CoV-2-induced cognitive impairment in a new murine model that is prevented by vaccination.


6
Together, these data show that B.1.351 infects the respiratory tract, but not the CNS, of wild 107 type C57Bl/6J mice.

108
To determine if B.1.351 infection of C57Bl/6J mice led to alterations in behavior after 109 recovery, we performed open field (OFT) and novel object recognition (NOR) testing at 30 dpi 110 (Fig. 1d). OFT, which assesses general motor function and anxiety, revealed a small, but 111 statistically significant, decrease in mean movement speed (Fig. 1e). However, the number of 112 lines crossed, total rotations, and time spent immobile were not different between mock-and 113 B.1.351-infected groups (Fig. 1e, Suppl. Fig. 1f). Similarly, the amount of time spent in corner vs 114 center zones did not reveal any differences (Fig. 1e). NOR testing, which investigates brain  121 1g). Preference testing confirmed there was no innate bias for either of the two objects used in 122 the NOR test (Suppl. Fig. 1g). Analysis of OFT and NOR test data did not show any significant baseline frequencies by 30 dpi (Suppl. Fig. 2a-d). Next, we examined leukocyte infiltration into 133 the cortex and hippocampus. Myeloid cell populations (CD45+, Ly6G-, CD3-, CD19-) were 134 identified via CD45 and CD11b expression (Fig. 2a, Suppl. Fig. 2a). CD45 mid CD11b+ cell 135 numbers were similar between mock-and B.1.351-infected animals at 6 dpi in the cortex and 136 hippocampus but were significantly increased in the hippocampus at 30 dpi (Fig. 2b).

137
Importantly, CD45 high CD11b+ cell numbers were significantly increased in the cortex and

152
However, at 30 dpi, Ly6C high cell numbers were the same as in mock-infected mice, while the 153 number of Ly6C low/neg cells was still higher in B.1.351-infected animals (Fig. 2d). Further analysis 154 of the CD45 High CD11b+ Ly6C low/neg population showed that at 6 dpi, ~20% of the cells were 155 P2RY12+ and significantly increased in number compared to mock-infected animals (Fig. 2e).

156
To examine myeloid cell location within the hippocampus, we performed immunohistochemical 1+ area significantly increased in B.1.351-infected mice at 6 dpi compared to mock-infected 159 animals (Suppl. Fig. 3b). IHC detection of IBA-1 and the microglial marker Tmem119 in 160 hippocampi from mock-infected mice revealed that 90% of IBA-1+ cells were also Tmem119+, 161 compared with 80-90% of IBA-1+ cells of hippocampi from B.1.351-infected mice at 6 dpi. Few 162 cells were IBA-1+Tmem119-in our analyses (Fig. 2f). These data indicate that a small, but 163 significant number of Ly6C high inflammatory monocytes infiltrate the forebrain at acute timepoints 164 and contract over time, while microglial activation persists long-term.  and TNF at 6 dpi. While cytokine levels were increased in the cortex, this did not reach 171 significance (Fig. 3a). To determine if this was a generalizable effect of severe respiratory 172 infections, or specific to SARS-CoV-2, we i.n. infected mice with a high dose of the mouse 173 adapted H1N1 influenza A virus (IAV) (strain A/Puerto Rico/8/1934; 2000 TCID50). Despite up to 174 20% loss of body weight, PR8 does not infect the CNS (Suppl. Fig. 4a-b). PR8 infection induced 175 cytokine expression in the lung, but no significant differences in IL-1b, IL-1a, IFNg, or IFNb 176 mRNAs were detectable in the forebrain at 3 or 6 dpi compared with mock-infected animals protein were undetectable in all CNS tissues derived from H1N1-infected mice (Suppl. Fig.   184 4c,e). In B.1.351-infected mice, IL-1b was not detected within GFAP+ or NeuN+ cells (Fig. 3e).

185
However, approximately 80-90% of IL-1b within hippocampi at 6 dpi was detected within IBA-1+ 186 and Tmem119+ cells (Fig. 3e-f). Despite the lack of IL-1b expression, there was a significant 187 increase in the percentage of IBA-1+ area after H1N1 IAV infection in the DG and CA3 (Suppl.

202
Next, we quantitated synaptic puncta within the DG, CA3 and CA1 regions via co-203 localization of the pre-synaptic marker, Synaptophysin, and the post-synaptic marker, Homer1.

204
Synapse loss was observed throughout the hippocampus, beginning at 8 dpi and was 205 significantly decreased by 15 dpi (Fig. 4d-e). Synapses partially recovered in the CA3 by 30 dpi 206 but remained decreased in the CA1 and DG compared to mock-infected mice (Fig. 4e). Analysis 207 of individual pre-synaptic/post-synaptic termini indicate that in the DG, synapse loss is primarily driven by decreased pre-synaptic termini, while in the CA3, post-synaptic termini are lost (Suppl.

384
Interestingly, we find that hippocampal myeloid cell activation does not always lead to IL-1b 385 production, as activated microglia/macrophages were observed in the hippocampus of H1N1

623
for the training day and test day (n=20). Data is represented as mean with standard error of significance was determined using a one-way ANOVA, two-way ANOVA, student's t-test, or 626 paired two-way ANOVA (for H). *=p<0.05, **=p<0.01, ***=p<0.001.