key: cord-0817110-sdbpo3aw
authors: Markus J, Lehtinen; Ritesh, Kumar; Bryan, Zabel; Sanna M, Mäkelä; Derek, Nedveck; Peipei, Tang; Sinikka, Latvala; Sebastien, Guery; Charles R, Budinoff
title: The effect of the probiotic consortia on SARS-CoV-2 infection in ferrets and on human immune cell response in vitro
date: 2022-05-23
journal: iScience
DOI: 10.1016/j.isci.2022.104445
sha: 4bcd69bc580b30659989fd4488a82084f665ddf1
doc_id: 817110
cord_uid: sdbpo3aw

Probiotics have been suggested as one solution to counter detrimental health effects by SARS-CoV-2, however, data so far is scarce. We tested the effect of two probiotic consortia, OL-1 and OL-2, against SARS-CoV-2 in ferrets and assessed their effect on cytokine production and transcriptome in a human monocyte-derived macrophage (Mf) and dendritic cell (DC) model.The results showed that the consortia significantly reduced the viral load, modulated immune response, and regulated viral receptor expression in ferrets compared to placebo. In human Mf and DC model, OL-1 and OL-2 induced cytokine production and genes related to SARS-CoV-2 anti-viral immunity. The study results indicate that probiotic stimulation of the ferret immune system leads to improved anti-viral immunity against SARS-COV-2 and that genes and cytokines associated with anti-SARS-CoV-2 immunity are stimulated in human immune cells in vitro. The effect of the consortia against SARS-CoV-2 warrants further investigations in human clinical trials.

The emergence and fast spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) that causes coronavirus disease 2019 , has resulted in an acute need for effective therapies and vaccination, but also a need for effective dietary strategies for improving immune function. The potential of probiotics, live microorganisms that confer health benefit on the host, in managing SARS-CoV-2 infection has not been extensively studied, despite the well-established connection between gut microorganisms and immune function (Hill et al., 2014) .

In humans, COVID-19 symptoms vary from mild respiratory symptoms such as fever, fatigue, and dry cough, to severe respiratory failure with acute respiratory distress symptom, acute cardiac injury and multiorgan failure. Although most individuals subsequently resolve the infection, the disease may also progress to severe pneumonia in susceptible groups. The incubation time of the virus in human is about 5 days, severe disease typically develops 8 days after symptom onset, and death occurs at about 16 days (Schultze and Aschenbrenner, 2021) .

In February 2020, the World Health Organization (WHO) assembled an international panel to develop animal models for COVID-19 to accelerate the testing of anti-SARS-CoV-2 therapies (Munoz-Fontela et al., 2020) . Ferret model is one of the recommended by the panel to study SARS-CoV-2.

Ferrets (Mustela putorius furo) have been used traditionally for testing the pathogenicity and transmission of human respiratory viruses, including influenza. The studies performed so far, support that experimental SARS-CoV-2 infection in ferrets results in a predominantly upper respiratory tract infection (Munoz-Fontela et al., 2020) . Ferrets are highly susceptible to SARS-CoV-2 and show an increase in viral load, and shed it in nasal washes, saliva, urine, and feces . Ferrets recapitulate the human-to-human transmission but not the severity of the COVID-19 (Kim et al., 2020) .

SARS-CoV-2 uses angiotensin-converting enzyme 2 (ACE2) as the main receptor for cell entry (Hoffmann et al., 2020) , but with several co-factors involved, including neuropilin-1 (Cantuti-Castelvetri et al., 2020) . The domain in ferret ACE2 that binds the spike protein of SARS-CoV-2 differs by two amino acids from its human homologue and binds to SARS-CoV-2 with similar affinity (Munshi et al., 2021) . SARS-CoV-2 primarily infects epithelial cells in the respiratory tract (Alfi et al., 2021) , however, infections of the gastrointestinal tract have also been observed (Stanifer et al., 2020) , but not extensively studied. In innate immune cells, such as monocytes and macrophages, the infection seems to be abortive (Zheng et al., 2021) .

J o u r n a l P r e -p r o o f SARS-CoV-2 is an enveloped virus with an +ssRNA genome. The specific innate immune receptors and signaling pathways triggering the interferon (IFN) response in the case of SARS-CoV-2 are yet to be fully determined, however, based on current findings Toll-like receptor (TLR) 3, TLR7, TLR8, RIG-I, and MDA5 recognize viral RNA and drive MyD88 and NF-κB -dependent pro-inflammatory and type I IFN response that further trigger STAT1, STAT2, and IRF9 -dependent induction of IFN-stimulated genes (ISG) (Schultze and Aschenbrenner, 2021) . On the other hand, SARS-CoV-2 suppresses the innate immune activation and IFN response to enable its replication in the host cells Taefehshokr et al., 2020) that seems to lead to inefficient B cell and T cell responses (Zhou et al., 2020a) . Human studies indicate that delayed innate IFN response results at a higher risk to develop severe COVID-19 (Hadjadj et al., 2020) . In COVID-19 patients with moderate to severe disease, high chemokine (e.g. CXCL2, CXCL8, CXCL9, CCL2) and pro-inflammatory cytokine (e.g. interleukin (IL)-1β, IL-6, IL-1Ra) response was detected in bronchoalveolar lavage fluid (BALF) (Zhou et al., 2020b) , in post-mortem lung biopsies and serum samples (Blanco-Melo et al., 2020) but without significant upregulation of type I or type III IFNs, although ISGs from e.g. IFIT and IFITM families were induced. Accordingly, intranasal IFN-I administered pre-or post-virus challenge reduced disease burden (Hoagland et al., 2021) . The above suggest that stimulation of innate immune function could be beneficial in controlling the viral replication and eradication.

In the past two decades the effect of probiotics on immune function and respiratory viral infections have been studied in numerous clinical studies, but with variable quality. Although meta-analyses indicate efficacy of probiotics against acute upper respiratory tract (URT) infections in general (Hao et al., 2015; King et al., 2014; Shi et al., 2021) , the results per strain vary, and accordingly probiotic effects on immunity should be investigated on a per strain or consortia basis (Hill et al., 2014) .

Clinical studies on specific strains like Bifidobacterium (B.) lactis Bl-04 showed that specific probiotics can reduce the risk of URT illness episodes in healthy adults (West et al., 2014) or reduce rhinovirus load in human nasal wash samples (Turner et al., 2017) . The clinical and pre-clinical evidence suggest that use of probiotics could prime immune system prior to viral infection (Lehtoranta et al., 2020) .

For example, Lactobacillus (L.) acidophilus NCFM induced the upregulation of TLR3, IL-12, and IFN-β in murine DCs (Weiss et al., 2010) and Lacticaseibacillus strains inhibited influenza A replication in human monocyte-derived macrophages (Miettinen et al., 2012) .

One of the hypotheses for the probiotic mode of actions against SARS-CoV-2 respiratory and intestinal infection is via strengthening the gut epithelial barrier and beneficial modulation of the gut microbiota and the immune system (Harper et al., 2021) . Age and comorbidities that make people more susceptible to severe COVID-19 are associated with perturbed gut microbiota, dysbiosis, and decrease in epithelial barrier integrity in the gut (Vignesh et al., 2021) . SARS-CoV-2 infects gut J o u r n a l P r e -p r o o f epithelial cells (Stanifer et al., 2020) which may cause lipopolysaccharide (LPS) and other pathogenassociated molecular patterns (PAMPs) to leak across the epithelial barrier. The resulting innate immune activation via the gut-lung-axis may contribute to the cytokine storm in these individuals (Vignesh et al., 2021) .

To investigate the effect and mechanism of action of probiotics against SARS-CoV-2 we designed two probiotic consortia to stimulate innate immune function against SARS-CoV-2: OL-1 (B. lactis Bl-04, B. longum subsp. infantis Bi-26, Lacticaseibacillus rhamnosus Lr-32, Lacticaseibacillus paracasei Lpc-37, Ligilactobacillus salivarius Ls-33) and OL-2 (B. lactis Bi-07, L. acidophilus NCFM, Limosilactibacillus fermentum SBS-1, Lactococcus lactis Ll-23, Streptococcus thermophilus St-21). We tested the effect of the probiotic consortia in a ferret SARS-CoV-2 challenge model and in human monocyte-derived macrophage (Mf) and dendritic cell (DC) model. The results in ferrets show that OL-1 and OL-2 modulate the duodenal and lung immune response during infection, ACE2 expression in duodenum, and reduce SARS-CoV-2 viral load in nasal washes. We further show that the exposure of OL-1 and OL-2 to human monocyte-derived Mfs and DCs results in upregulation of genes and cytokines critical for early innate immune activation against SARS-CoV-2.

Ferrets have been shown to model SARS-CoV-2 replication and infection in the airways and to express ACE2 (Kim et al., 2020) . We evaluated the effect of the two probiotic consortia OL-1 and OL-2 on the SARS-CoV-2 infection, ACE2 expression, and immune function in ferrets in two placebo-controlled studies: main (Figure 1 ) and pilot ( Figure S1 ). The pilot study was run early after the onset of the COVID-19 pandemic to more understand the model, whereas the main study included changes to the protocol based on the learnings from the pilot study and was aimed to test the efficacy of the OL-1 and OL-2.

In the main study, probiotic consortia were supplemented 7 days (D) prior (D-7) to viral infection (D0) and until D10 post infection. Lung and duodenum samples were gathered for qPCR gene expression analysis (D5 and D10) and tissue staining (D5 and D10); and nasal washes were collected D1-D10 for determining the viral load by qPCR ( Figure 1A) . The SARS-CoV-2 viral load over the infection time (D1-D10) showed statistically significant differences in the area under the curve (AUC) between the OL-1, OL-2, and placebo groups (p=0.0088; Kruskall-Wallis). Note that at day 1 all groups have non-zero values (means of 152, 297, and 201 genome/µL for OL-1, OL-2, and placebo respectively) ( Figure 1B) .

Post-hoc pairwise Wilcoxon tests further demonstrated the significant differences in AUC between the placebo and the OL-1 (p=0.009) and OL-2 (p=0.021) groups, respectively. On average, OL-1 and OL-2 reduced the AUC by 68% and 52%, respectively, as compared to the placebo group, suggesting influence of the probiotic consortia on the anti-viral immunity against SARS-CoV-2.

The time-course analysis of the SARS-CoV-2 nasal wash qPCR showed that the viral loads peaked in general at D3 (average viral titer of 4.9x10 4 genome/µL, 1.8x10 4 genome/µL, and 2.4x10 4 genome/µL in the Placebo, OL-1, and OL-2 groups, respectively) and were mostly resolved by D9 (average viral titer of 319 genome/µL, 394 genome/µL, and 552 genome/µL in the placebo, OL-1, and OL-2 groups, respectively) ( Figure 1C ). Further investigation of the individual line plots ( Figure 1D ) showed that the viral loads of few animals show later peaking (1/9, 5/11, 3/9 animals in the placebo, OL-1, and OL-2 groups, respectively). Most later peaking animals were found in the treatment groups that could reflect the effects observed on the viral replication.

In comparison to the main study, the pilot study had a different probiotic supplementation regime, where OL-1 was administered 21 days prior to infection (D-21) and OL-2 one day post infection (D1) ( Figure S1 ). The viral load analysis showed undetectable to very low viral genomes in the samples ( Figure S1 ), despite the use of the same viral dose to induce the infection and housing dependent J o u r n a l P r e -p r o o f infectivity. The pilot study did not show treatment effect on the viral load. To further understand why the viral load was low or absent in the pilot study or any treatment effects, the lung and duodenum samples at D0 (pre-infection), D5, and D21 were collected for later qPCR analyses, and in addition D0 samples were used later as a comparator for the main study responses.

The body temperature of the ferrets and visual symptoms of viral infection were monitored in both studies. The SARS-CoV-2 infection in ferrets did not affect the body temperature, body weight or induce coughing, sneezing, or dyspnea in the animals during infection, in line with previous literature (Peacock et al., 2021; Ryan et al., 2021) .

To understand further the differences in immune response between the pilot and the main study, ferret immune response to SARS-CoV-2, and the mechanism of action of OL-1 and OL-2 on the lung viral load decrease, lung and duodenal tissues from the main ( Figure 1A ) and the pilot study ( Figure   S1 ) were collected, and gene expression measured using qPCR ( Figure S2, Figure 2 ).

First, we wanted to understand the effect of the virus infection on the immune response. Importantly, pre-infection samples (D0) were taken only for the OL-1 and placebo groups from the pilot study.

Ferrets in the study were outbred, had the placebo from the same batch and the virus was known to cause a strong immune response that is clearly different from the potential small differences in the baseline immunity between the main and the pilot study, we decided to use D0 non-infected placebo group samples from the pilot study as a baseline to normalize not only the pilot ( Figure S2A ), but also the main study ( Figure S2B ) immune response data to the SARS-CoV-2 infection. Statistical analysis indicated only minor changes in the pilot study for lung and duodenal gene expression ( Figure S2A ), in line with the low infectivity in the experiment. Instead, in the main study, IFN (IFNA and IFNL1), cytokine (IL-10), and chemokine (CCL2, IL8) gene expression was increased in responses to the SARS-CoV-2 infection in all the treatment groups (adjusted p<0.1; FDR) ( Figure S2B ). It is noteworthy that IFNG expression was downregulated in the lung tissue and ACE2 expression was downregulated in the duodenum, whereas it was upregulated in the lungs ( Figure S2B ). At D5 both consortia, but not placebo, had a significant increase in CXCL10 in duodenum and lungs (adjusted p<0.1; FDR) ( Figure   S2B ). As there was not a non-infected placebo group in the main study, the results ( Figure S2B ) should be interpreted cautiously. In summary and based on the viral load and qPCR results the main study seems to model better SARS-CoV-2 infection and immune response, whereas in the pilot study the infectivity and immune response is less apparent.

To compare the effect of OL-1 and OL-2 to placebo, we did a within study timepoint comparison of the qPCR data for the pilot (Figure 2A ) and the main study ( Figure 2B ). In the main study, expression of several immune genes was regulated by OL-1 and OL-2 in timepoint comparison to the placebo group ( Figure 2B ). At D5 both consortia induced upregulation of IFNA in the duodenum, and OL-1 further upregulated CXCL10, IL8, and TLR8 in the lungs compared to placebo (adjusted p<0.1; FDR). At D10 both consortia had higher CCL2, IFNL1, IL8, and TLR8 gene expression in duodenum compared to placebo (adjusted p<0.1; FDR). OL-1 had in addition a decrease in IFNG expression and OL-2, on the other hand, increased IFIH1, IL6, and TNFA (adjusted p<0.1; FDR) ( Figure 2B ). In the lungs, there were no significant differences at D10 between the groups. Overall, the results indicate stronger immunomodulation in the duodenum, but also effects in the lungs, in line with the route of administration by gavage.

In the pilot study there were minor differences at D0 and D5. At pre-infection (D0), OL-1 decreased It should be noted that the post-infection gene-expression results regarding the OL-1 and OL-2 effect compared to placebo were different between the pilot (D5 and D21) ( Figure 2A ) and the main study (D5 and D10) ( Figure 2B ). We think that the main factor driving the difference was the lower level of infection in the pilot study that resulted in significantly different profile and lower level of SARS-CoV-2-induced gene expression than in the main study ( Figure S2) . Thus, the anti-SARS-CoV-2 immune responses that the probiotic consortia are modulating are different, as are then the probiotic consortia effects compared to placebo, on that immune response in the pilot (Figure 2A ) and in the main study ( Figure 2B ). In conclusion, the probiotic effects between the pilot and the main study are not directly comparable. Figure S2B ). However, the fecal pellets remained of normal consistency in the ferrets. The treatments with OL-1 or OL-2 did not have an effect on the duodenal inflammation scores (p>0.05;

ANOVA, Dunnett's).

To test how OL-1 and OL-2 could potentially influence anti-SARS-CoV-2 immunity in humans, we used fresh human peripheral blood monocyte-derived macrophages (Mf) and dendritic cells (DC).

The immune cells from four donors without technical replicates were incubated with OL-1, OL-2, controls, or TLR agonist mix Poly I:C (pIC) (TLR3) and R848 (TLR7/8) -that mimics immune response to viral RNA. We also used the combination of pIC+R848 with OL-1 or OL-2 to evaluate modulation of the cell response to pIC+R848 by the consortia (Figure 4) .

Incubation of the OL-1 or OL-2 with Mfs for 24 hours resulted in an increase in all the measured cytokines from supernatant by ELISA of which IL-6 and TNF-α had most pronounced increase, After 48 hours of incubation with DCs both consortia increased secretion of IL-12p70 compared to control ( Figure 4G ) and IFN-γ with or without pIC+R848 ( Figure 4H ), indicating Th1 polarization of the DCs by both consortia directly, but also augmentation of Th1 response upon TLR3 and TLR7/8 stimulation. TNF-α and IL-1β may support Th1 polarization of DCs, however, the effects of the consortia on the cytokine production were mixed as direct incubation increased TNF-α and IL-1β, but in the pIC+R848 response modulation there was either no effect or decrease of IL-1β in OL-2 samples ( Figure 4I and 4J). Both consortia increased IL-10 production ( Figure 4K ) but suppressed TGF-β ( Figure 4L ), supporting polarization towards Th1, although some potentiation of the Th17 polarizing cytokines IL-23 and IL-6 secretion was observed ( Figure 4M and 4N ). In summary, the results suggest that consortia could drive polarization of the DCs towards Th1 cell induction, but with some induction of Treg and Th17 associated cytokines.

To further test the effect of OL-1 and OL-2 on immune cell stimulation and to compare to pIC+R848 response, we conducted transcriptomic analyses. Briefly, RNA was extracted for sequencing from all treated groups of Mfs and DCs from the experiments described above and at the same time points than ELISA samples were taken (Figure 4 ).

To gain an initial broad view of the transcriptome profiles, we conducted a principal component analysis (PCA). In the analysis of the full data set, ie DC and Mf samples together, PC1 captured 70% and PC2 14% of the variance, accounting for the differences in the cell type and the treatment effect, respectively. As the cell type was the largest driver of variation, further PCA was conducted by the cell type (Mfs and DCs were analyzed separately). In the Mf and DC specific PCAs, the donor samples clustered together based on treatment, and did not show any extreme outliers ( Figure 5A and 5B).

The PCA of the Mf data resulted in the separation of OL-1, OL-2, and pIC+R848 from the controls (DMSO and ctrl, PC1 49%), with OL-1 clustering closer to the pIC+R848 group than OL-2 ( Figure 5A ).

The PCA of the DC data showed a similar pattern ( Figure 5B ), however, in the DC data, PC1 mainly captured the spread of treatment, and accounted for a higher amount of variance than in Mf (68% vs 49%), suggesting a larger treatment effect on the transcriptome. In addition, pIC+R848 clustered separately from the control and the probiotic consortia. Overall, in Mf and DC datasets OL-1 and OL-J o u r n a l P r e -p r o o f 2 clustered with pIC+R848 and separate from the controls, indicating a potentially general similarity in immune gene activation. PC loadings for OL-1 Mf samples showed major influence by IL1B, CXCL8, CXCL1, CXCL2, PPBP (CXCL7), and TNFAIP6 on the PC1 ( Figure 5C ). Of these, IL1B, CXCL8, and TNFAIP6 were shared with OL-2, whereas CCL4L2 and CXCL10 were unique ( Figure 5D ). The results suggest that the key effects of the consortia versus control are on early innate and chemokine response. For DCs, OL-1 PCA showed separation of OL-1 from the control on PC1 with IFI6, CLEC2D, SIGLEC6, and CLC driving the effect ( Figure 5E ). Of these IFI6 and CLC were shared with OL-2, whereas IL15RA, HESX1, ISG20, and GBP4 were unique ( Figure 5F ). GBP4, IFI6, ISG20, and CLEC2D genes are inducible by IFNs, and further CLC (Galectin-10) and CLEC2D (LLT1) expression by DCs are associated with modulating B and T cell responses.

In differential gene expression (DEG) analysis, OL-1 and OL-2 were compared to control and pIC+R848 to DMSO. In Mfs OL-1, OL-2, and pIC+R848 treatments resulted in 158, 523, and 1264

DEGs, respectively, whereas in the DCs OL-1, OL-2, and pIC+R848 treatments resulted in 501, 1264, 670 DEGs, respectively (Table S1 Differentially expressed genes of significance, Figure 5 ). The number of shared DEGs between OL-1, OL-2 and pIC+R848 was 1350 for Mfs and 596 for DCs. OL-1 and OL-2 treatments shared 308 DEGs in DCs and 184 in Mfs, with pIC+R848 having 1904 unique DEGs in DCs and 184 in Mfs. Overall, the results support the observed increase in cytokine production and cell stimulation by OL-1 and OL-2, and similarity with pIC+R848 stimulation ( Figure   4 ).

We further investigated the effect of the OL-1 and OL-2 on the cytokine, chemokine, and costimulatory molecule gene expression. Overall, the profiles of OL-1 and OL-2 were similar, except for downregulation (e.g. IL2, IL7, IL27, IL36G) or upregulation (e.g. IL6, IL32, TNFA) of some transcripts in Mfs by OL-2 ( Figure 6 ). Specifically, for Mfs, OL-1 and OL-2 stimulation upregulated most prominently EBI3 (IL27B), IL1A, IL1B, IL12B, and IL23A, necessary for type I immunity. Accordingly, chemokines profiles in Mfs showed broad activation of CCL and CXCL class chemokines, especially CCL1, CCL3, CCL4, CCL15, CXCL1, CXCL2, CXCL8, and CXCL10 that target neutrophils, Mfs, and NK cells. In addition, OL-1 and OL-2 induced HLA and co-stimulatory molecule expression, in line with innate response stimulation, but also upregulated inhibitory CD274 ( Figure 6 ). In DCs, OL-1 and OL-2 induced upregulation of IL2, IL15, IL18, and IL23 that are associated with T cell survival. Furthermore, treatments increased chemokine CX3CL1, CXCL9, CXCL10, CXCL11, CCL17, and CCL22 expression that are associated with attraction and activation of T cells. Probiotic consortia further inhibited CCL2, CCL3, and CCL4 expression that attract CCR2 and CCR5 positive cells such as innate NK cells and monocytes. Co-stimulatory gene expression results showed that DCs upregulated co-stimulatory genes CD80, CD83, CD86 after OL-1 or OL-2 treatment. Interestingly, inhibitory PD-ligands (CD274, J o u r n a l P r e -p r o o f PDCD1LG2) were upregulated but on the other hand tolerogenic response -associated CD31 (Clement et al., 2014) and CD36 (Lee et al., 2021) were downregulated. DCs showed upregulation of HLA class I and downregulation of HLA class II genes ( Figure 6 ).

SARS-CoV-2 is known to suppress host cells anti-viral mechanisms to enable replication. As we had observed immune stimulation by OL-1 and OL-2, we wanted to further understand the effect of the consortia to anti-viral pathways critical for SARS-CoV-2 defense. We conducted pathway analysis using differential expression data which takes into account log2 fold changes and p-values for all genes in our dataset and calculates a probability of a KEGG pathway to be perturbed. Overall, OL-1, OL-2, and pIC+R848 affected 81, 36, and, 87 pathways in the Mfs and 71, 79, and 105 pathways in DCs, respectively (adjusted p-value < 0.1, FDR) (Table S2 Pathway In Mfs and in response to OL-1 and OL-2, interleukin IL12B and IL1B and chemokine CXCL10 and CXCL8 genes were highly expressed compared to low or no induction in DCs (Figure 7) . However, the antiviral genes MX1 and MX2, OAS1, OAS2, OAS3 and ISG15 were strongly induced in DCs and to a somewhat lower extent in Mfs. Interestingly, in DCs the neuropilin 1 receptor gene NRP1 involved in SARS-CoV-2 entry (Cantuti-Castelvetri et al., 2020) was reduced by all the treatments, as well as the gene coding ACE, whereas IFIH1 gene coding for MDA5 receptor was upregulated. Monocytederived innate immune cells seem to express the genes for SARS-CoV-2 receptor ACE2 and TMPRSS2 protease (Yang et al., 2020) that is needed for the activation of the viral spike protein and facilitation J o u r n a l P r e -p r o o f

The potential of probiotics to support immune function against SARS-CoV-2 is yet to be established.

In this study we have shown the effect of two probiotic consortia OL-1 and OL-2 on reducing nasal wash viral load in a ferret SARS-CoV-2 challenge model. The consortia modulated cytokine, chemokine, and IFN gene expression profiles in lungs and duodenum suggesting innate immune stimulation by OL-1 and OL-2. In human monocyte-derived Mf and DC model we further show that both consortia stimulate Th1 type cytokines and activate IFN response transcriptomic pathways, critical for innate defense against SARS-CoV-2. Further, the data suggests that the consortia modulates main receptor ACE2 in ferrets and may modulate co-receptors for viral entry in humans.

We evaluated the effect of OL-1 and OL-2 consortia in a pilot and main study on SARS-CoV-2 challenge.

The results show a decrease in nasal wash viral load in the main study by both OL-1 and OL-2 probiotic consortia (Figure 1 ). In the main study, the time course and the magnitude of the SARS-CoV-2 viral titers in the placebo group was comparable to a recent SARS-CoV-2 dose response study in ferrets (Ryan et al., 2021) , where the authors were using high (5 x 10 6 plaque forming units (pfu)/ml), medium (5 x 10 4 pfu/ml) and low challenge dose (5 x 10 2 pfu/ml), of which the medium dose is comparable to 10 5 TCID50 target dose in our study. Ryan et al showed around 10 7 viral RNA copies/ml in the peak replication at 3 days with the high dose, whereas our lower viral dose resulted in an average of 4.9 x 10 7 genomes/ml (4.9 x 10 4 genomes/uL) in the placebo group ( Figure 1) . They further showed that the initial viral inoculum plays a role in subsequent viral load. This indicates potential issues in the viral inoculum of our pilot study that was hampered by low infectivity of ferrets and low viral titers ( Figure   S1 ). The low infectivity was also apparent in the very mild immune response of the ferrets to the virus ( Figure S2A ) and as such the results of the main study are a better representation of the course of infection and immune response in ferrets, as well as the effect of OL-1 and OL-2.

Although ferrets have been recognized as a good model for coronavirus infections, there are very few studies in ferrets where probiotics have been evaluated. Recently, a study assessed the effect of a probiotic consortia in ferrets and showed results on behavioral phenotypes after maternal pIC immune activation (Dugyala et al., 2020) . However in mice models, probiotics administration has been shown to reduce viral load against influenza and some other respiratory pathogens (Lehtoranta et al., 2020) and in humans B. lactis Bl-04 (in consortia OL-1) supplementation was shown to decrease nasal wash viral titers in a rhinovirus challenge model (Turner et al., 2017) and risk of colds in healthy active J o u r n a l P r e -p r o o f adults (West et al., 2014) . Our study results support the use of ferrets for evaluating the probiotic or microbial therapies efficacy against viral infections.

ACE2 is the main receptor for the SARS CoV-2 and it has been suggested that decreasing expression could limit viral entry into the cells, but studies on intestinal ACE2 are limited. It has been postulated that there is a potential intestinal infection and fecal-oral transmission of SARS-CoV-2 . In gut, ACE2 acts as a key regulator of dietary amino acid homeostasis, innate immunity, gut microbial ecology, and susceptibility to develop inflammation in a RAS independent manner (Hashimoto et al., 2012) . In the current study on D5, during the acute infection, we observed by immunohistology a significantly higher expression of ACE2 in the duodenums of OL-1 and OL-2 group ferrets compared to placebo group animals (Figure 3 ), but by qPCR lower expression of ACE2 in all groups compared to D0 in the main study ( Figure S2B ). As it has been shown that SARS-CoV-2 infection decreases ACE2 on the cell membrane (Verdecchia et al., 2020) , the results suggest that OL-1 and OL-2 groups may have had lower duodenal viral infection and replication compared to placebo. This hypothesis is supported by increased IFNA expression by OL-1 and OL-2 compared to placebo at D5 ( Figure 2 ) and suggests activation of the type I IFN pathway in the gut. On the other hand, ACE2 also acts as the SARS-CoV-2 receptor, and further, higher ACE2 expression has cytoprotective effects (Gheblawi et al., 2020) , and thus it is not clear what is an optimal level of ACE2 expression (Chaudhry et al., 2020) . Despite increase in ACE2, we did not observe differences in the duodenal tissue inflammation (Figure 2 ) or in inflammatory gene expression response between the treatments at D5 (Figure 3) . Interestingly, the Mf and DC treatment with OL-1 and OL-2 decreased expression of ACE ( Figure 7 ) that has opposite function to ACE2. The results indicate that the effect of the probiotic consortia on the gut ACE-ACE2 balance and RAS system could be cytoprotective. The transcriptomics data also showed a decreased expression of NRP1 that acts as a cofactor for ACE2-mediated viral cell entry (Figure 7 ) (Cantuti-Castelvetri et al., 2020) , which adds to the evidence on the influence of the

At D10 in the main study, duodenal ACE2 expression decreased in OL-1 and OL-2 groups compared to placebo, and also compared to D5 OL-1 and OL-2 levels (Figure 3) , perhaps suggesting faster resolution to baseline ACE2 expression in probiotic groups, however, with lack of non-infectious samples for baseline this hypothesis remains to be tested.

The inflammatory gene expression at D10 was comparable to D5 in all groups, however, the histological inflammatory scoring showed a decrease from D5 to D10 ( Figure 3I and 3J) , in line with resolution of intestinal infection. The effects of OL-1 and OL-2 compared to placebo were relatively pronounced and consistent with duodenal gene expression. Both consortia increased the expression of type III IFN IFNL1, monocyte (CCL2) and neutrophil (IL8) targeting chemokines, and TLR8 that recognizes SARS-CoV-2, suggesting stimulation of anti-SARS-CoV-2 immunity. OL-2 also induced cytosolic SARS-CoV-2 pattern recognition receptor IFIH1 (MDA5), IL6, and TNFA, and OL-1 decreased IFNG, that points to consortia specific differences on immune stimulation. The differences between OL-1 and OL-2 were also observed in duodenal gene expression in the pilot study, where OL-2 decreased IFN responses at D5, and OL-1 decreased pro-inflammatory and IFN gene expression at D21 compared to placebo. In addition, OL-1 and OL-2 showed significant reduction in ACE2 transcripts when compared to placebo at D21.

So far little has been done to evaluate therapies or intervention to target the gastrointestinal tract pool of SARS-CoV-2. Based on the differential expression of ACE2 in the duodenum we propose that OL-1 and OL-2 intervention could potentially be effective in managing the SARS-CoV-2 load in the duodenum, potentially also in humans as supported by decreased expression of NRP1 and ACE in human DC model.

In COVID-19, human lungs are the key site for pathology and viral replication and thus ferret lung immune response modulation could be of relevance, although SARS-CoV-2 is mostly upper respiratory infection in ferrets (Munoz-Fontela et al., 2020) . The gene expression results at D5 and D10 of the main study compared to non-infected ferrets from the pilot study, show that SARS-CoV-2 induces type I (IFNA) and III (IFNL1) IFN responses, but suppresses type II response (IFNG), and stimulates pro-inflammatory (IL6 and IL8) and anti-inflammatory (IL10) gene expression ( Figure S2B) .

Overall, the cytokine profiles indicate balanced response of the ferrets to the virus where the viral replication is controlled without cytokine storm. We also observed at D5 and D10 in all groups increased expression of ACE2 that has anti-inflammatory and cytoprotective properties in addition to its role as a receptor for SARS-CoV-2 (Gheblawi et al., 2020) . The results above are in line with the lack of symptoms in the ferrets and low viral load by D10 (Figure 1 ). In the pilot study minimal changes in gene expression was observed that likely reflects the low infectivity ( Figure S1 and S2) .

Similar immune responses in ferrets to SARS-CoV-2 as in the main study have been observed previously (Blanco-Melo et al., 2020) that supports our use of non-infected placebo treated animals J o u r n a l P r e -p r o o f from the pilot study to normalize the response of the infection in the main study ( Figure S2B) , however, the results need to be interpreted cautiously.

Probiotic effect compared to placebo on the lung gene expression was low. In the main study at D5, OL-2 increased expression of chemokines (CXCL10 and IL8) targeting NK cells and neutrophils, and TLR8 that recognizes viral RNA (Figure 2) . This effect was not observed at D10 and could be due to resolution of the infection. In the pilot study OL-1 at D0 decreased CXCL10 and TLR8, suggesting effects on the lung immune function pre-infection, however, no changes were observed in duodenum. At D5 OL-1 increased IFNG and at D21 post infection the differences were most pronounced for OL-1 that increased IFNA, IFNL1, IL10, IL6, and TNFA, and decreased CCL2, also suggesting effects post-infection. In summary, probiotic consortia seem to induce lung immune system, but the effects seem mild and somewhat inconsistent in this data set. On the other hand, OL-1 and OL-2 reduced the viral load that is suggestive of influence on nasal immune system as well.

The effect of the probiotic consortia on gut-lung or gut-respiratory axis warrants further investigation.

Innate immune systems between ferrets and humans, and mammals in general are homologous, however, to gain further understanding on potential effect of the consortia in humans we exposed human blood-derived Mfs and DCs to OL-1 and OL-2 and viral RNA analog cocktail pIC+R848 and analyzed their cytokine and transcriptome responses.

The studies on SARS-CoV-2 infection and pathogenesis indicate that inadequate innate immune and IFN response increase susceptibility to more severe COVID-19 (Schultze and Aschenbrenner, 2021) that is supported by increased risk of more severe disease by aging. Our results show that exposure of Mfs to OL-1 and OL-2 resulted in the secretion of cytokines important for Mf activation and induction of type I immunity (IL-12 and IFN-γ), but also drove pro-and anti-inflammatory cytokine production (Figure 4) . These results were supported by transcriptomics analysis showing activation of IL-1 and IL-12 cytokine family genes, upregulation of co-stimulatory molecules, MHC-I/II and chemokines targeting monocytes, NK cells, and neutrophils critical for early defense in viral infections ( Figure 6) . DC response to OL-1 and OL-2 further show Th1 polarization as determined by increased IL-12p70 production (Figure 4) . The transcriptomics analysis further showed an increase in STAT1 (Figure 7 ) that drives in DCs Th1 polarization (Johnson and Scott, 2007) , induction of costimulatory and MHC-I molecules, and upregulation of genes related to T cell growth and attraction. Lu et al. 2021 showed that SARS CoV-2 (24h infection with clinical isolates) induces IL1A, IL1B, IL8, CXCL10, CCL2, and CCL3 mRNA in human PBMC-derived myeloid cells (Lu et al., 2021) . We see quite similar gene induction pattern in our monocyte-derived Mfs (Figure 7) , where both consortia and pIC+R848 induce some of the same genes as SARS-CoV-2, suggesting induction of anti-viral pathways. In summary, the results show that probiotic consortia could prime or train the innate immune system also in humans prior to SARS-CoV-2 infection to enhance resilience of the host.

Priming of the innate system by OL-1 or OL-2 is a potential mechanism to explain the reduced viral load that we observed in the ferrets after 7 days of supplementation (Figure 1 ). Although we have limited data pre-infection, the post infection gene expression results suggest enhanced response in duodenum, but also potentially in lungs (Figure 2) . Netea et al. have suggested that the training of the human innate immune system, for example by vaccines, prior to SARS-CoV-2 infection could reduce the susceptibility to and the risk of severe COVID-19 (Netea et al., 2020) . Similar evidence on priming of the immune system against viral infections by probiotics have been noted more broadly in vitro and in human clinicals (Lehtoranta et al., 2020; Turner et al., 2017) .

Studies on the early pathogenesis of SARS-CoV-2 show that it infects the cells via ACE2 that is internalized. The host cells recognize the virus by endosomal (TLR3, TLR7, and TLR8) and cytosolic (RIG-I and MDA5) receptors that recognize ssRNA and dsRNA. This leads to an MyD88 and NF-κBdependent activation of pro-inflammatory response and IRF3/7 -dependent type I IFN response. IFNs drive STAT1/2 and IRF9 mediated ISG response. SARS-CoV-2 produces proteins that suppress the activation of this cascade (Schultze and Aschenbrenner, 2021) . The results of our transcriptomics study show that OL-1 can significantly modulate the KEGG COVID-19 pathway in Mfs, and that OL-2 shows similar gene expression pattern in Mfs but does not result in a significant change on a pathway level (Figure 7) . In DCs, the result on a pathway level is significant only for pIC+R848, however, OL-1 and OL-2 show similar pattern of significant gene expression (Figure 7) . Several signaling molecules on the pathways in Mfs and DCs show upregulation by the consortia including IRF3, IRF9, NFKB1, STAT1, STAT2, and ISGs ISG15, MX1, MX2, OAS1, OAS2, OAS3, and CGAS. In addition, viral RNA sensors IFIH1 (MDA5) and DDX58 (RIG-I) show an increased expression, whereas TLR7 and TLR8 expression have mixed results. Interestingly type I IFN family gene expression seems to be suppressed or unimpacted by OL-1 and OL-2, but mainly also by pIC+R848 (Figure 7) . Overall, the results suggest that OL-1 especially, but also OL-2, could potentially counteract suppression of the innate immune system activation by SARS-CoV-2. Delayed or poor IFN response has been J o u r n a l P r e -p r o o f associated with more severe COVID-19 disease in human clinical samples (Hadjadj et al., 2020) and lower levels of pro-inflammatory cytokines and chemokines noted with SARS-CoV-2 infection (Blanco-Melo et al., 2020; Chu et al., 2020) , suggesting that targeting innate immunity with probiotic consortia could provide benefits in managing SARS-CoV-2 infection in humans, however, making conclusions on efficacy of probiotics in humans requires clinical data.

In this study we have shown that probiotic consortia OL-1 and OL-2 stimulate ferret immune function and reduce viral load during SARS-CoV-2 infection (Figure 8) . Results of the human immune cell stimulation indicate that pathways and cytokine secretion associated with SARS-CoV-2 immune defense are activated and thus these probiotic consortia could potentially provide benefits to support immune function against SARS-CoV-2, perhaps by priming or training the innate immune system (Lehtoranta et al., 2020; Netea et al., 2020) (Figure 8) . We have also shown influence of the probiotic consortia on duodenal immune function and ACE2 expression. As many COVID-19 patients suffer from gastrointestinal symptoms, these probiotic consortia could potentially support intestinal health and immune function. Probiotics in general have shown efficacy in meta-analyses against respiratory tract infections (Hao et al., 2015; King et al., 2014) , however, the results between the strains or their combinations vary. Thus, it is warranted to use specific strains or consortia of probiotics for immune stimulation (Hill et al., 2014) . Even small differences in genome or within strains may result in vast differences in phenotypes and metabolism (Morovic et al., 2018; Zabel et al., 2020) . Human clinical studies should be conducted to better understand the effect of OL-1 and OL-2, and other microbial therapies in managing SARS-CoV-2 infections.

We have discussed the limitations of the study broadly in the discussion, however, it is noteworthy to highlight here that in the ferret study the results between the main and the pilot studies were somewhat conflicting, justifying further investigations into the effect of probiotics on immune function and anti-SARS-CoV-2 immunity. In addition, we did not specifically study the colonization of the probiotics or their effect on the microbiota in this study. The in vivo and human macrophage and dendritic cell studies would have benefited on inclusion of non-probiotic bacteria or other wellstudied probiotic strains to better evaluate the effect size and specificity of the immune response to OL-1 and OL-2 consortia. suggesting that these probiotic consortia could also be beneficial in humans.

Provided as separate excel-file.

Provided as separate excel-file.

Any further information needed on the resources or reagents used should be directed to the lead contact Charles Budinoff (Charles.R.Budinoff@iff.com).

This study did not generate any unique materials or reagents.

• Data regarding the entire project is available from Gene Expression Omnibus (GEO) under the SuperSeries number GSE180391. Individual datasets are also available from GEO for the ferret qPCR (GSE180390) and human RNA-seq (GSE180389) studies.

• This paper does not report original code involved in the analysis of the data.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

The probiotic strains used in the experiments were for OL-1: Bifidobacterium longum subsp. infantis 

For ferret studies, potato maltodextrin was used as a placebo and probiotics were delivered as freezedried powder in a capsule that was opened, resuspended to saline, and gavaged. All the strains are owned by and were manufactured along with placebo by IFF (Madison, Wisconsin, USA).

For in vitro cell stimulation assay, Bifidobacterium strains were cultured anaerobically at 37°C in Bifidobacterium medium 58 (Deutsche Sammlung von Mikroorganismen und Zellkulturen, DSMZ) (Salli et al., 2021) . Bacteria belonging to Lacticaseibacillus, Ligilactobacillus, Limosilactobacillus, Lactobacillus, Lactococcus, and Streptococcus genera were cultured anaerobically at 37°C in de Man, Rogosa and Sharpe medium (LAB M Ltd, Lancashire, United Kingdom), except for Limosilactobacillus fermentum SBS-1, which was grown aerobically. Strains were grown to logarithmic growth phase, collected by centrifugation, washed once with PBS and suspended to cell culture medium. The optical density (OD)600 was adjusted to correspond to bacteria: host cell ratio of 10:1. The bacteria in the consortia were applied on cells in equal proportions 2+2+2+2+2. 

Pilot Study: From D-21 through study D21, ferrets in group 1 received a daily oral gavage with the OL-1 consortia (20B CFU/ strain), and ferrets in group 3 received a daily oral gavage with the Placebo.

Ferrets in group 2 received a daily oral gavage with the OL-2 consortia (20B CFU/ strain) from study D1 through study D21. The virus was administered intranasally (i.n.) as droplets on study D0. On study D1, D3, D5, D7, and D9 nasal washes were collected. Five ferrets from each OL-1 and placebo groups were necropsied on D0 and five ferrets from each group were necropsied on D5 and D21 in the pilot study.

Main Study: From study D-7 through study D10, ferrets in group 1 received a daily oral gavage with the OL-1 consortia (20B CFU/ strain), ferrets in group 2 received a daily oral gavage with the OL-2 consortia (20B CFU/ strain), and ferrets in group 3 received a daily oral gavage with the Placebo. Each probiotic capsule was opened and content diluted in 3 mL sterile PBS for each daily oral gavage. 2019

Novel Coronavirus, isolate USA-WA1/2020 (SARS-CoV-2) was used to challenge the ferrets. The viruses were produced by infecting VE6 cells at IITRI with the SARS-CoV-2 and tissue culture infection dose (TCID50) was calculated. The virus in DMEM was administered i.n. as droplets on study day 8. Ferrets were anesthetized with Ketamine/Xylazine for delivery of the virus. A total of 0.5 mL of virus (target of 10 5 TCID50) was delivered to each study ferret. The 0.5 mL dose was split between each nostril (0.25 mL/nare). To confirm the inoculation titer of the virus, aliquots of the prepared virus solution were collected, the aliquots were stored at ≤ -65 °C, and a viral titer assay (TCID50) was performed.

On study days D1, D3, D5, D7, D9, and D10 nasal washes were collected. Ferrets were anesthetized with Ketamine/Xylazine, and 0.5mL of sterile PBS containing penicillin (100 U/mL), streptomycin (100 µg/mL) and gentamicin (50 µg/mL) were injected into each nostril and collected in a specimen cup when expelled by the ferret. The recovered nasal wash was collected, volume recorded, and aliquoted.

One aliquot (~100 µL) was treated with DNA/RNA Shield (Zymo Research, Irvine, CA) and then stored at room temperature for determination of viral load by RT-qPCR. Five ferrets from each group were necropsied on D5, and remaining ferrets from each group were necropsied at D10 in the main study.

Lungs were collected and 0.5-1.0 cm 3 piece of lung was flash frozen, and then stored at ≤ -65 °C. The duodenum was harvested, cut longitudinally to expose the inner mucosal layer, rinsed with sterile PBS, and divided into five roughly equal portions. Samples were flash frozen and stored at ≤ -65 °C for RNA isolation and samples were fixed in 10% formaline for IHC.

The concentration of virus in nasal washes from study days D1, D3, D5, D7, D9, and D10 was determined by RT-qPCR assay. Briefly, RNA was extracted from samples stored in RNA/DNA Shield using the Quick-RNA Viral Kit (Zymo Research) according to manufacturer's protocol. RNA was eluted with 100 µL nuclease-free water. A standard curve was prepared by using blank ferret nasal wash collected from ten naïve ferrets and spiked with known concentrations of viral RNA. Each RT-qPCR plate included 9 RNA standards (5 x 10 7 , 5 x 10 6 , 5 x 10 5 , 5 x 10 4 , 5 x 10 3 , 5 x 10 2 , 50, 20, 5 copies per RT-qPCR well) in duplicate, a NTC (no template control) and a positive control in triplicate well. Each test sample was analyzed in duplicate wells.

RT-qPCR was performed using the iTaq universal probes onestep kit (Bio-Rad). 5 µl viral RNA was used for RT-qPCR. Total reaction volume was 15 µL (5 µl RNA + 10 µL master mix). The following RT-PCR cycling conditions will be used: If the interaction effect Treatment×Day was significant, ANOVA was carried out to compare dCq across three treatment groups within the same tissue and day, followed by post-hoc analysis through obtaining estimated marginal means (EMMs) (Lenth et al., 2021) and pairwise comparison thereof.

The multiple comparison p-values were adjusted by Benjamini-Hochberg (BH) method. Log2 fold change was calculated using the log2 of the ratio of EMM dCq test/EMM dCq control. For EMM dCq control we used the placebo for each gene at the same time point or the pre-infection D0 placebo

time point for each gene tested.

Tissue samples were collected from PBS control and SARS CoV-2 -infected ferrets and incubated in 10% neutral-buffered formalin for fixation before they were embedded in paraffin based to standard procedures. The embedded tissues were sectioned and dried for 3 days at room temperature. To detect the ACE2 by immunohistochemistry, Goat Anti-ACE2 antibody (R&D, Cat#: AF933, Lot#:

HOK0620051) was used as the primary antibody. Antigen was visualized using citrate buffer (pH6.0), microwaved for 3 min and steam for 15 min.

Slides were viewed and digitized pictures were taken by Olympus VS120 scanner and analyzed by a senior Pathologist using Olyvia 3.2 software (Olympus Corporation, Tokyo, Japan). (Patro et al., 2017) . Two samples did not pass the 5 million read minimum, which were both from pIC+R848 treated DC and were not included in the analysis. 38 total samples were processed, sample quality metrics for RNA quality and alignment to the reference can be found in Supplemental   Table S4 .

The area under the curve (AUC) over the 10 days post infection was calculated from the viral titer trend lines for each ferret using the R statistical programming language. Kruskal-Wallis nonparametric test was performed to evaluate the differences in the AUC among the three treatment groups (sig. level = 0.05), and post-hoc pairwise Wilcoxon tests were deployed for multiple comparisons using

Benjamini-Hochberg method for p-value adjustment.

The qPCR dCq values were calculated based on the GAPDH housekeeping gene. Mixed-effect linear models were fit to test for differences between the treatments and control, fitting the model of dCq value = Treatment + Day + (1|Animal ID), with the animal ID being the random effect. The package emmeans (v 1.6.1) (Lenth et al., 2021) was used to conduct post-hoc tests and pairwise comparisons of the treatments, with p-values corrected using the Benjamini-Hochberg method.

Histopathological examination of duodenum ACE2 expression was scored by using H-score. The H-score is based on a predominant staining intensity. The Intensity scores are classified as negative (0), low intensity (2), medium intensity (2) and high intensity (3). The percentage of cells at each staining intensity level is counted (estimated), and finally, an H-score is assigned using the following formula:

Total H Score= 1 × (% cells 1+) + 2 × (% cells 2+) + 3 × (% cells 3+)

Data was presented as mean ± SEM. Statistical analysis was performed using one-way, ANOVA followed by Dunnett's multiple comparisons test. ns, not significant, **, p < 0.01; ***, p < 0.001.

Graphs were generated with GraphPad Prism (GraphPad, San Diego,USA).

The data from immune cell ELISA analyses (Figure 4 ) were log2-transformed prior to statistical analysis. The data were analyzed using a linear model that included the main effects of treatment and donor (2-way ANOVA). Pairwise-comparisons between the treatments were performed using contrasts of estimated marginal means and corrected for false positive rate using Sidak's method.

The fit of the models was checked by inspecting the normality of the model residuals. q q q q q q q q q q q q q q q q q q q q q q q q q q q q IFNa  IFNg  IFNL1  IL10  IL1b  IL6  IL8  RPS18  TLR8  TNFa  ACE2  CCL2  CCL5  CXCL10  IFNa  IFNg  IFNL1  IL10  IL1b  IL6  IL8  RPS18  TLR8  TNFa  ACE2  CCL2  CCL5  CXCL10  IFNa  IFNg  IFNL1  IL10  IL1b  IL6  IL8  RPS18 CCL5  CXCL10  IFIH1  IFNa  IFNg  IFNL1  IL10  IL1b  IL6  IL8  NFkB  RPS18  TLR8  TNFa  ACE2  CCL2  CCL5  CXCL10  IFIH1  IFNa  IFNg  IFNL1  IL10  IL1b  IL6  IL8  NFkB  RPS18  TLR8  TNFa   OL IFNGR1  IFNGR2  IL1A  IL1B  IL2  IL7  IL10  IL12B  IL15  IL16  IL17C  IL18  IL22  IL23A  IL26  IL27  IL32 Cytokine Genes q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q Chemokine Genes q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q 

Human Nasal and Lung Tissues Infected Ex Vivo with SARS-CoV

Provide Insights into Differential Tissue-Specific and Virus-Specific Innate Immune Responses in the Upper and Lower Respiratory Tract

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

Neuropilin-1 facilitates SARS-CoV-2 cell entry and infectivity

Manipulation of ACE2 expression in COVID-19

Comparative Replication and Immune Activation Profiles of SARS-CoV-2 and SARS-CoV in Human Lungs: An Ex Vivo Study With Implications for the Pathogenesis of COVID-19

CD31 is a key coinhibitory receptor in the development of immunogenic dendritic cells

Putative modulation of the gut microbiome by probiotics enhances preference for novelty in a preliminary double-blind placebocontrolled study in ferrets

Angiotensin-Converting Enzyme 2: SARS-CoV-2 Receptor and Regulator of the Renin-Angiotensin System: Celebrating the 20th Anniversary of the Discovery of ACE2

Potential intestinal infection and faecal-oral transmission of SARS-CoV-2

Impaired type I interferon activity and inflammatory responses in severe COVID-19 patients

Probiotics for preventing acute upper respiratory tract infections

Viral Infections, the Microbiome, and Probiotics. Frontiers in Cellular and Infection Microbiology 10

ACE2 links amino acid malnutrition to microbial ecology and intestinal inflammation

Expert consensus document. The International Scientific Association for Probiotics and Prebiotics consensus statement on the scope and appropriate use of the term probiotic

Leveraging the antiviral type I interferon system as a first line of defense against SARS-CoV-2 pathogenicity

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

STAT1 expression in dendritic cells, but not T cells, is required for immunity to Leishmania major

Toward understanding the origin and evolution of cellular organisms

KEGG: integrating viruses and cellular organisms

KEGG: kyoto encyclopedia of genes and genomes

Infection and Rapid Transmission of SARS-CoV-2 in Ferrets

Effectiveness of probiotics on the duration of illness in healthy children and adults who develop common acute respiratory infectious conditions: a systematic review and meta-analysis

Human CD36(hi) monocytes induce Foxp3(+) CD25(+) T cells with regulatory functions from CD4 and CD8 subsets

Coronaviruses: Innate Immunity, Inflammasome Activation, Inflammatory Cell Death, and Cytokines

Role of Probiotics in Stimulating the Immune System in Viral Respiratory Tract Infections: A Narrative Review

emmeans: Estimated Marginal Means, aka Least-Squares Means

Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

SARS-CoV-2 exacerbates proinflammatory responses in myeloid cells through C-type lectin receptors and Tweety family member 2

Nonpathogenic Lactobacillus rhamnosus activates the inflammasome and antiviral responses in human macrophages

Transcriptional and Functional Analysis of Bifidobacterium animalis subsp. lactis Exposure to Tetracycline

Animal models for COVID-19

An overview of preclinical animal models for SARS-CoV-2 pathogenicity

Trained Immunity: a Tool for Reducing Susceptibility to and the Severity of SARS-CoV-2 Infection

Salmon provides fast and bias-aware quantification of transcript expression

The furin cleavage site in the SARS-CoV-2 spike protein is required for transmission in ferrets

Dose-dependent response to infection with SARS-CoV-2 in the ferret model and evidence of protective immunity

Selective Utilization of the Human Milk Oligosaccharides 2'-Fucosyllactose, 3-Fucosyllactose, and Difucosyllactose by Various Probiotic and Pathogenic Bacteria

COVID-19 and the human innate immune system

Modulation of gut microbiota protects against viral respiratory tract infections: a systematic review of animal and clinical studies

Critical Role of Type III Interferon in Controlling SARS-CoV-2 Infection in Human Intestinal Epithelial Cells

Effect of probiotic on innate inflammatory response and viral shedding in experimental rhinovirus infection -a randomised controlled trial

The pivotal link between ACE2 deficiency and SARS-CoV-2 infection

Could Perturbation of Gut Microbiota Possibly Exacerbate the Severity of COVID-19 via Cytokine Storm? Frontiers in Immunology 11

ROntoTools: R Onto-Tools suite

Lactobacillus acidophilus induces virus immune defence genes in murine dendritic cells by a Toll-like receptor-2-dependent mechanism

Probiotic supplementation for respiratory and gastrointestinal illness symptoms in healthy physically active individuals

ggplot2: Elegant Graphics for Data Analysis

Attenuated Interferon and Proinflammatory Response in SARS-CoV-2-Infected Human Dendritic Cells Is Associated With Viral Antagonism of STAT1 Phosphorylation

Strainspecific strategies of 2′-fucosyllactose, 3-fucosyllactose, and difucosyllactose assimilation by Bifidobacterium longum subsp. infantis Bi-26 and ATCC 15697

Acute SARS-CoV-2 Infection Impairs Dendritic Cell and T Cell Responses

Heightened Innate Immune Responses in the Respiratory Tract of COVID-19 Patients

• Probiotic consortia decrease SARS-CoV-2 viral load in ferret nasal washes • Ferret duodenal ACE2 but not inflammation was modulated by the consortia • Immune response genes in duodenum and lungs were induced by probiotics• Probiotic consortia induce anti-viral response genes in human immune cells ex vivo J o u r n a l P r e -p r o o f