Treatment of SARS-CoV-2 induced pneumonia with NAD+ in a mouse model

The global COVID-19 epidemic has spread rapidly around the world and has already caused the death of more than one million people. As there is yet no vaccine, it is urgent to develop effective strategies to treat COVID-19 patients. Here, we used a mouse-adapted SARS-CoV-2 infection model to explore potential therapeutic targets for SARS-CoV-2 pneumonia. Global gene expression analysis of infected mouse lungs revealed dysregulation of genes associated with NAD+ metabolism, immune response and cell death, similar to that in COVID-19 patients. We therefore investigated the effect of treatment with NAD+ and found that the pneumonia phenotypes, including excessive inammatory cell inltration and embolization in SARS-CoV-2 infected lungs were signicantly rescued by boosting NAD+ levels. Most notable, cell death was suppressed substantially (>65%) by NAD+ supplementation. Thus, our in vivo mouse study supports trials for treating COVID-19 patients with NAD+ or its precursors.

critical COVID-19 develop severe respiratory distress, and death (Pan et al., 2020). The ongoing COVID-19 pandemic in the absence of an effective vaccine has prioritized the development of drugs for COVID-19 treatment.
To explore potential therapeutic targets for COVID-19, we inspected the dysregulated genes in a mouseadapted SARS-CoV-2 (MASCp6) infection mouse model (Gu et al., 2020). We performed global gene expression analysis of SARS-CoV-2 infected lungs of 8-month old mice 3.5 days post infection (dpi). GO and KEGG gene set enrichment analysis of signi cantly up-regulated genes revealed the strong activation of immune related biological processes and pathways, including cytokine-cytokine receptor interaction, NOD-like receptor signaling, chemokine signaling, IL-17 signaling, TNF signaling, IL-6 production and other virus defense pathways ( Figure S1A, B). Infection led to robust induction of chemokine and interleukin genes, including Il1b, Il6, Ccl2, Ccl3, Ccl4, Ccl7, Cxcl1, Cxcl2, Cxcl5, Cxcl9, and Cxcl10, as well as many related receptor genes, including Ccr1, Ccr4, Ccr5, Ccr5, Ccr7, Cxcr2, Cxcr5, Il1r2, and Il1rn (Figure S1B, C; Figure S2A). In addition, 20 genes associated with necroptosis and apoptosis, such as Tnf, Tlr3, Daxx, Stat1 and Stat2 were signi cantly up-regulated ( Figure S2B). Moreover, neutrophils, macrophages, dendritic cells, B cells, and CD4 + T cells related genes were all up-regulated in infected mice as was also observed in COVID-19 patients ( Figure S2C) (Bost et al., 2020). These results indicated that SARS-CoV-2 infection induces very similar immune processes in both mice and humans (Bost et al., 2020;Chua et al., 2020;Liao et al., 2020;Zhu et al., 2020) Nicotinamide adenine dinucleotide (NAD + ), the cell hydrogen carrier for redox enzymes, plays important roles in multiple metabolic pathways (Miller et al., 2020;Nasi et al., 2020;Omran and Almaliki, 2020). We recently found that PARPs, one of main NAD + -responsive signaling family, were signi cantly up-regulated while NAD + levels were substantially reduced in Zika virus infected mouse brains (Pang et al., manuscript submitted). In SARS-CoV-2 infected lungs, Parp9, Parp10, and Parp14, the three PARPs members which regulate immune and in ammatory responses (Rajman et al., 2018), were also signi cantly up-regulated ( Figure 1A). Meanwhile, NAD + -responsive genes, the Sirtuin family members including Sirt2 and Sirt4, were both down-regulated ( Figure S2D). We therefore explored the expression of genes involved in NAD + biosynthetic process and found some of those required for NAD + de novo synthesis and the kynurenine pathway from tryptophan, were up-regulated ( Figure 1B, C). In addition, the expression of Nampt, one of the key enzymes for NAD + salvage was induced ( Figure 1B). The above results suggested that over-consumption of NAD + by up-regulation of PARPs led to decreased NAD + levels and this stimulated compensatory de novo synthesis and salvaging of NAD + in infected lungs.
Based on the observations that both SARS-CoV-2 and ZIKV infection led to dysregulation of NAD + associated genes (Figure1 and S2) and reduction of NAD + levels in both mice and human sera (our study and (Heer et al., 2020), we investigated whether NAD + supplementation could rescue SARS-CoV-2-induced pathological effects. We rst characterized in more detail the pathological features (pneumonia) of infected mouse lungs. Immunostaining with SARS-CoV-2 spike protein and nucleocapsid protein antibodies indicated that SARS-CoV-2 mainly infected airway epithelial cells and some alveolar epithelial cells closed to airways ( Figure S3). The infection pattern is similar to another mouse-adapted model of SARS-CoV-2, in which receptor binding domain in the spike protein was remodeled to facilitate its e cient binding to mouse ACE2 (Dinnon et al., 2020). We con rmed the presence of in ammatory cell in ltration, alveolar septal thickening, embolization, epithelial damage, and cell death in MASCp6 infected lungs, similar to the effects in the hACE2 mouse model (Figure 2A; Figure S4A, B) (Jiang et al., 2020;. Notably, we detected embolization in some airways with many macrophages (CD68 + ) phagocytizing infected cells ( Figure S4A). We treated SARS-CoV-2 infected mice with NAD + (1mg/g/day i.p. for 3 days), and sacri ced them on 3.5 dpi ( Figure S5A). Compared with severe in ammatory cell in ltration and embolization of airway detected by H&E staining in saline group, the number and size of in ammatory cell aggregations were signi cantly reduced obviously in the NAD + treatment group ( Figure   2A-C). In addition, the ratio of embolization airway and the relative embolism area were both reduced very signi cantly (P<0.001) after treatment ( Figure 2D, E; Figure S5B). However, comparable levels of SARS-CoV-2 intensity and infected area were detected in saline and NAD + treatment groups ( Figure S5B-E) suggesting the NAD + treatment did not reduce the rate of infection but relieved its symptoms.
Previous studies have shown that SARS-CoV-2 infection induce the in ltration of neutrophils and macrophages in lungs of patients (Bao et al., 2020;. We found there was more severe aggregation of macrophages in the small airways, especially in those that were close to alveoli, than the big airways in mouse lungs ( Figure 3A). The numbers of macrophages and epithelial cells positive for the activated form of caspase 3 was reduced very signi cantly (>65%) in NAD + treated mice ( Figure 3; Figure   S6D, E). Meanwhile, there were much fewer infected alveolar epithelial cells close to the small airways in NAD + treated mice ( Figure S5B). These results indicate that NAD + supplementation can protect the lung from in ammatory injury, including cell death, caused by SARS-Cov-2 infection in both old and young mice ( Figure S6).
In this study, we demonstrated that very similar immune processes and pathways, including cytokine-cytokine receptor interaction, TNF signaling, and IL-6 production, were activated in SARS-CoV-2 infected mice as those observed in SARS-CoV-2 infected patients and organoids model (Bost et al., 2020;Liao et al., 2020;Yang et al., 2020;Zhu et al., 2020). A previous clinical study indicated that IL-6 monoclonal antibody, tocilizumab, could improve the outcome in severe and critical COVID-19 patients . Together with the presence of typical pneumonia phenotypes in infected lungs, we can conclude that the mouse model is suitable for drug development before clinical trials in COVID-19 patients.
Consistent with our work, a previous metabolomics study showed that the level of tryptophan, the major precursor for the NAD + de novo synthesis pathway, was decreased, and the levels of intermediate substrates, kynurenine and quinolinate, were increased in the sera from severe COVID-19 patients . We have revealed here that these changes likely re ect dysregulated the NAD + related gene set by SARS-Cov-2 infection and induced expression of a set of PARP family members, similar to those in COVID-19 patients (Heer et al., 2020).
However, our results indicate that NAD + supplementation leads to limited area of infection but does not inhibit multiplication of SARS-CoV-2 in vivo. We found instead that NAD + could effectively prevent the cell death caused by SARS-CoV-2 and ZIKV in mouse lung and brain, respectively (this study and other manuscript submitted). Based on the similar dysregulation of NAD + metabolism in both ZIKV and SARS-CoV-2 infected mouse models, the mechanism underlying treatment with NAD + for cell death may be similar. NAD + de ciency induced by up-regulated PARPs could affect the function of many related metabolic pathways, such as oxidative phosphorylation, TCA cycle, and tryptophan metabolism. NAD + supplementation helps to maintain the balance of metabolic processes in virus infected cells and prevent the death of those cells, as well as suppress the subsequent in ammatory response. This may explain why NAD + precursors, including NMN, NR and NAM, represent anti-in ammatory effects in different animal models including aging, autoimmune encephalomyelitis, ischemia, Ataxia Telangiectasia (Fang et al., 2016;Gomes et al., 2013;Kaneko et al., 2006;Katsyuba et al., 2020;Rajman et al., 2018). Therefore, our study is in strong support of initiating a trial for treating COVID-19 patients with NAD + or its precursors.
Declarations AUTHOR CONTRIBUTIONS Z.X. and C.-F.Q. conceived of the research and Z.X. designed the study. Z.X. and Y.J. wrote the manuscript. Y.J. and Y.D. designed and performed most of the experiments. Y.J. did RNA sequencing analysis and immunostaining. Y.D. did inoculation of the virus and took care of the mice.

Method Details
Infection and administration of BALB/c mice.
Anesthetized 6-7 weeks and 8 months old female BALB/c mice were intranasally (i.n.) treated with a dose of 7.2×10 5 plaque forming unit (PFU) of SARS-CoV-2 in a total volume of 40 μL in each group. The infected mice were administered with either NAD + dissolved in 0.9% saline or the equivalent volume of 0.9% saline for three consecutive days and one time per day by intraperitoneal injection starting at half a day after inoculation the virus. These procedures were operated in Biosafety Level 3 laboratory (BSL-3).

Collection of mice samples
The mice were euthanized on 3.5 dpi. The lungs and tracheas of each mouse were removed and xed in 4% paraformaldehyde (PFA) at 4°C two days for frozen section and para n section or homogenized with TRIzol reagent for RNA extraction.

Measurement of viral RNA
Lung and trachea tissue homogenates were clari ed by centrifugation at 6,000 rpm for 6 min, and the supernatants were transferred to a new EP tube. Viral RNA (vRNA) was extracted using the QIAamp Viral RNA Mini Kit (Qiagen, Cat#52906) according to the manufacturer's protocol. vRNA quanti cation in each sample was performed by Q-PCR targeting the S gene of SARS-CoV-2 as described previously (Gu et al., 2020).

Histopathological staining
Fixed lung tissues for histological examination were subjected to alcohol gradient dehydration and embedded in para n. Each embedded tissue was sectioned at 5μm thickness sections. At least three sections of each tissue were used for staining with hematoxylin and eosin (H&E) and examined under a light microscopy. The tissue handling and H&E staining were accorded to standard protocol.

RNA-seq analysis
The signi cantly differentially expressed mRNAs were selected with |Log 2 (fold change)| > 1 and p value < 0.05 by R package DESeq2 (Pertea et al., 2015). GO enrichment analysis and KEGG enrichment analysis of signi cantly differently expressed genes were performed using the OmicStudio tools at https://www.omicstudio.cn/tool. Gene number and p value of GO and KEGG terms were visualized using the R package ggplot2. The network of the genes among different GO and KEGG terms were visualized using Cytoscape (version 3.7.2). The heat maps and bar plots of signi cantly differently expressed genes were drawn with the R package ggplot2 and GraphPad (Prism 7) respectively.

Quanti cation And Statistical Analysis
For quanti cation of in ammatory cell aggregation number and area, embolism airway ratio and relative area, and alveolar septum relative area, three lung sections for each mouse (3 mice for saline group and 4 mice for NAD + group) were included. In ammatory cell aggregation was de ned by dense cell cluster and dark blue staining with hematoxylin. The density of aggregation was calculated as the number of aggregation divided by the area of minimal convex polygon which can cover all of aggregation in each lung section. The area of aggregation was the total area added by area of each individual aggregation. Embolism airway ratio was calculated as the number of airways having embolism divided by the number of total airways in each section. The relative area of embolism was calculated as the area of each embolism divided by the area of corresponding airway. The relative area of alveolar septum was calculated as the total area of the substances divided by the area of whole lung section. For quanti cation of the density of CD68 + cells, ve regions for each lung section, two sections for each mouse were included. The density was calculated as the number of CD68 + cells divided by the area of corresponding region. For quanti cation of the ratio of Cas3 + cells to infected cells, three to four infected airways for each mouse were included. The ratio was calculated as the number of Cas3 + cells divided by the number of infected cells in the same location of the adjacent section. For quanti cation of SARS-CoV-2 relative uorescence intensity and infected area, two sections for each mouse were included. Relative uorescence intensity was calculated as the total uorescence intensity of SARS-CoV-2 signal divided by the area of whole lung section. The relative infected area was calculated as the area of minimal convex polygon which can cover all of virus signal divided as the area of whole lung section. Statistical analyses were carried out with unpaired t test or Welch's t test. Data are presented as mean ± SEM. Figure 1 SARS-CoV-2 infection leads to dysregulation of NAD+ associated genes in lungs. (A-B) RNA-seq analysis of SARS-CoV-2 infected lungs of 8-month old mice on 3.5 days post infection (dpi). Bar plots indicate the expression levels of PARPs (A) and NAD+ biosynthetic process related genes (B). All data showed as means ± SEM, *p<0.05, **p<0.01, ***p<0.001. (C) Schema of NAD+ metabolism pathway (Refer to Rajman et al., 2018). NA: nicotinic acid; NAMN: nicotinic acid mononucleotide; NAAD: nicotinic acid adenine dinucleotide; NR: nicotinamide riboside; NAM: nicotinamide; NMN: nicotinamide mononuclotide; Trp: tryptophan; FK: formyl-kynurenine; Qa: quinolinate. Red letters: up-regulated (fold change>1 and p value<0.06); Blue letters: down-regulated (fold change<1 and p value<0.06).

Figure 2
NAD+ supplementation relieves pneumonia caused by SARS-CoV-2 infection. (A) H&E staining shows that in ammatory cell in ltration (green arrow head) and embolism in airway (blue arrow head) induced by infection were rescued by NAD+ supplementation. (B-E) Quanti cation of the density (B) and area (C) of in ammatory cell aggregation, the ratio of embolism airway (D) and relative area of embolism (E). All data showed as means ± SEM, t test. ***p<0.001.

Figure 3
NAD+ supplementation prevents excessive aggregation of macrophages and cell death in infected lungs.
(A) Immunostaining of SARS-CoV-2 nucleocapsid protein (top panels, green) and marker of macrophage, CD68 (bottom panels, red) in two adjacent lung sections. Yellow border: excessive aggregation of macrophages. (B) Immunostaining of SARS-CoV-2 nucleocapsid protein (top panels, green) and the