In vitro digestion of SARS-CoV-2 contaminated berries reveals high inactivation of infectious virus during gastrointestinal passage

ABSTRACT

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infectivity was shown to be stable on frozen berries for at least a month; however, whether the virus on berries remains infectious as it passes through the gastrointestinal tract is unknown. Here, the stability of SARS-CoV-2 infectivity on berries under in vitro digestion was investigated. The INFOGEST (an international network of excellence on the fate of food in the gastrointestinal tract) and the biorelevant fast- and fed-state models were used. SARS-CoV-2 (~7 log 50% tissue culture infective dose [TCID₅₀]/mL) infectivity was assessed in blueberry and strawberry homogenates at 4°C and 37°C, and following consecutive oral (2 min), gastric (10 and 60 min), and intestinal (15 and 120 min) digestion at 37°C. SARS-CoV-2 infectivity was quantified on Vero-E6 cell line using the TCID₅₀ assay. SARS-CoV-2 infectivity in berry homogenates was stable for 120 min at 4°C; however, the virus was completely inactivated at 37°C within 60 min. The biorelevant fed-state model closely mimicked the INFOGEST model. In the oral phase, SARS-CoV-2 infectivity decreased by ~1 log, which did not change during the 10-min gastric digestion. By 60 min of gastric digestion, strawberries had a protective effect on SARS-CoV-2, showing less inactivation (~2 log) in comparison to blueberries (~3–3.4 log). The viruses entering the intestinal phase showed a 2.6–3.3 log inactivation within 15 min. In contrast, SARS-CoV-2 in the fast-state model showed high inactivation at the 10- and 15-min gastric and intestinal time points (2.5 and 4 log, respectively). Therefore, under high levels of SARS-CoV-2 contamination of berries, it is possible for the virus to infect the oral cavity but less likely to infect the intestine.

IMPORTANCE

During the pandemic, news outlets occasionally reported on the detection of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) RNA on various foods, raising concerns over contaminated foods initiating infections. Coronavirus disease 2019 (COVID-19) patients often experience gastrointestinal symptoms and shed SARS-CoV-2 RNA in their feces. In addition, active virus replication in the gastrointestinal tract was shown; however, infectious viruses were rarely detected in feces. We previously showed that SARS-CoV-2 remained infectious on frozen berries for at least a month. Here, in vitro digestion models showed that SARS-CoV-2 on berries exhibits minimal inactivation at the oral phase and the virus may escape gastric inactivation early during feeding. However, high intestinal inactivation of the virus on berries suggested that SARS-CoV-2 was less likely to initiate infection in the small intestine. In contrast, the oral cavity is a potential site where infection might be initiated, providing more input for the gastrointestinal tract. High intestinal inactivation might explain the difficulty of detecting infectious SARS-CoV-2 in feces but not of virus RNA.

INTRODUCTION

In December 2019, a novel virus emerged in Wuhan, China, causing severe respiratory illness. The novel virus belonged to the same species as Severe acute respiratory syndrome coronavirus (SARS-CoV) and was named SARS-CoV-2 (1). The disease caused by SARS-CoV-2, coronavirus disease 2019 (COVID-19) , can be mild or severe, with symptoms ranging from fever, cough, fatigue, muscle aches, loss of smell or taste, nausea, vomiting, and diarrhea to severe pneumonia (2). The most affected are the elderly, the immunocompromised, and individuals with comorbidities. Since the beginning of the COVID-19 pandemic, it was reported that some COVID-19 patients experience gastrointestinal illness with or without respiratory symptoms (3). In addition, the viral RNA was detected in stools of 38%–85% COVID-19 patients (4–7). The viral titer in stool can reach as high as 7.5 log₁₀ copies per gram on the second week post symptoms onset (7), and the duration of viral shedding can last for weeks (5, 8, 9). Interestingly, children seem to experience more gastrointestinal (GI) illness as diarrhea is a frequent feature of pediatric COVID-19 patients (10) and they continue to shed the virus in feces 2 weeks longer than respiratory shedding (11). The virus receptor angiotensen coverting enzyme 2 (ACE-2) is highly expressed in the oral cavity and along the GI tract (12, 13) and active replication of the virus was shown to occur in the oral cavity (14, 15) and GI epithelium of COVID-19 patients (5, 7). Although, isolation of infectious SARS-CoV-2 from stool of COVID-19 patients was reported to be difficult, a number of studies were successful (9, 16–18). Thus, SARS-CoV-2 is a respiratory virus with enteric tropism (3, 10), raising concern for possible oral or GI infection through ingestion of the virus in contaminated food.

SARS-CoV-2 RNA detection on food has been demonstrated in several cases, especially from frozen food (19, 20). This is plausible because SARS-CoV-2-infected individuals carry high viral load in their oral fluids and can potentially contaminate food through large droplets deposited by speaking or coughing (21). For example, uninterrupted speaking for 3 min generated ~1,095 droplets covering an area of 0.75 × 1.25 m which can theoretically contain at least 3.5 × 10³ infectious viruses (22). A single cough event is expected to generate ~10⁵ SARS-CoV-2 particles (23). Regardless of the source or route of contamination of food with SARS-CoV-2, several laboratory studies showed that once SARS-CoV-2 contaminates food, it can remain infectious for days as measured in cell culture infectivity assays. For example, on meat such as beef steak, ground beef, pork chops, ground pork, and turkey stored at 4°C, the virus added at ~4 log per gram remained infectious for 2–3 weeks; with a loss of 1–2 log within 7 days (24). SARS-CoV-2 in acidic fermented milk (pH 3.45) stored at 4°C and in ice cream (fat content 10.2%) stored at −20°C or −80°C remained infectious, showing ~3 and 1.8 log reduction in infectivity after 7 and 60 days of storage, respectively (25). However, thermal processing such as cooking meat at 71°C for 2 min and pasteurization of milk at 63°C for 30 min or 80°C for 15 seconds were shown to inactivate SARS-CoV-2 (24, 25). Therefore, the concern was mainly associated with SARS-CoV-2 contamination of food that are consumed raw without any processing of such fruits and vegetables. A number of studies showed that infectious SARS-CoV-2 can survive on various produce for days, depending on the contamination level and storage temperature. For example, infectious virus was retrieved from grapes and tomatoes contaminated with 4 log of SARS-CoV-2 for 14 days, showing only ~2.2–2.5 log reduction (24). Our previous research showed that it takes 3 days for SARS-CoV-2 contamination on blueberries and strawberries stored at 4°C to show ∼1–2 log 50% tissue culture infective dose (TCID₅₀) per gram reduction (26). However, when contaminated berries were frozen, no significant reduction in SARS-CoV-2 infectivity was observed for at least a month (26). Contamination of frozen produce with SARS-CoV-2 is concerning because these foods can be intended for consumption without any processing, have long shelf life, and can be shipped between countries.

While food may act as vehicle for a SARS-CoV-2 transmission, no reported infections have been demonstrated as a result of consuming SARS-CoV-2 contaminated food (27). Indeed, a number of epidemiological studies linked dining at restaurants or sharing meals with increased risk of SARS-CoV-2 infections; however, aerosol-mediated infection was not ruled out in these settings [reviewed in reference (22)]. Because SARS-CoV-2 is a biosafety level-s (BSL3) pathogen, limited studies addressed the stability of SARS-CoV-2 infectivity during gastrointestinal passage. Two previous studies used commercially available biorelevant GI fluids on either SARS-CoV-2 or a recombinant SARS-CoV-2 mNeonGreen but failed to supplement the gastric or intestinal fluids with digestive enzymes or to perform consecutive digestions from oral to gastric to intestinal (28, 29). These biorelevant fluids simulate the GI tract pH and the chemical composition of the gastric and intestinal contents under both fasting and feeding states and are extensively used in oral drugs development (30, 31). We previously showed that SARS-CoV-2 incubated with these GI biorelevant fluids supplemented with digestive enzymes was highly inactivated within 10 min in gastric followed by intestinal fluids (>4 log inactivation) simulating fasting (32). However, when SARS-CoV-2 was incubated with fluids simulating fed state, the virus was less inactivated (~1 log) within 10 min of gastric followed by intestinal digestion (32). Furthermore, we showed that the latter is likely mediated by higher gastric pH (≥3.5–6) and higher intestinal bile concentrations (~15 mM) that occur during feeding (32).

There are no previous studies investigating the effect of oral or gastrointestinal fluids on SARS-CoV-2 ingested in food. Given that we showed extended survival of infectious SARS-CoV-2 on berries, the objective of this study was to evaluate the stability of SARS-CoV-2 infectivity on berries during in vitro simulated digestion (IVD). Berries are known as a high-risk food commodity for the spread of foodborne enteric viruses such as human norovirus and hepatitis A virus. Also, berries contaminated in one country can transmit enteric viruses to other countries [reviewed in reference (33)]. This is important in light of the World Health Organization report on the origins of the SARS-CoV-2, suggesting that the re-introduction of SARS-CoV-2 via cold/food chain is possible (34). Here, we first determined the effect of berry homogenates (mimicking berry smoothies) on SARS-CoV-2 infectivity. Then, we used two IVD models to evaluate the effect of digestion on infectious SARS-CoV-2 mixed in homogenized berries. The first model was based on INFOGEST (an international network of excellence on the fate of food in the gastrointestinal tract) standardized model (35) and the second was based on commercially available biorelevant fluids simulating both fasting and feeding states. All fluids were supplemented with digestive enzymes as described previously (32).

RESULTS

Effect of time and temperature on infectious SARS-CoV-2 in berry homogenates

At 4°C and in water alone, SARS-CoV-2 showed no significant changes in its infectivity titers over the 120-min incubation period (Fig. 1A). Similarly, both berry homogenates did not significantly affect the virus infectivity titers over the 120-min period (Fig. 1A). However, as observed at the 2-min time point, the virus was recovered at significantly lower titer from strawberry (3.3 log₁₀ TCID₅₀/mL) as compared to blueberry homogenates (4.4 log) or to water (5.2 log) (Fig. 1A). The latter suggests that virus recovery varied with the type of berry either due to instantaneous inactivation at 2 min or differential loss of virus during virus recovery from berry homogenates. Regardless, SARS-CoV-2 in berry smoothies prepared in water remained infectious under refrigeration for at least 2 h.

Fig 1.

Time and temperature effects on SARS-CoV-2 incubated in water (control) or blueberry or strawberry homogenates. Berries were homogenized in water at a ratio of 1:1 (vol/vol) using a commercial smoothie blender. The virus was mixed with berry homogenates at a ratio of 1:10 (vol/vol) and incubated at (A) 4°C or (B) 37°C. In addition, (C) the virus was mixed with berries that were pre-neutralized to pH 7 and incubated at 37°C. Virus-berry homogenates were incubated for a period of 2, 60, and 120 min, before being neutralized, and centrifuged and the supernatants were tested using TCID₅₀ assay for virus quantification. Comparing treatments within each panel, means with different letters differ significantly (P < 0.05). Significant differences between temperatures among corresponding treatments were designated with asterisks sign (*) for comparing panel A to B, with the number sign (#) for comparing panel B and C, and with dot sign (•) for comparing panel C to A. Dotted line represents this assay’s detection limit at 1.8 log₁₀ TCID₅₀/mL.

Next, the experiments above were repeated using incubation at 37°C, a temperature relevant to gastrointestinal digestion. When SARS-CoV-2 was incubated in water, there was a significant reduction (~0.5 log) in virus titer by the end of the 120-min incubation period (Fig. 1B). Again, as observed with the 4°C experiments, at 37°C and 2 min, the virus was recovered at a significantly lower infectivity titer from strawberry (3.3 log) as compared to blueberry homogenates (4.3 log) or to water (5.3 log) (Fig. 1B). At 60 and 120 min, there was a complete loss of infectious virus in both berry homogenates (at or below detection limit of 1.8 log) (Fig. 1B). In comparison to the 4°C experiments, these data suggest that higher temperature (37°C) exerted a significant inactivation effect on the virus when mixed in berry homogenates (* in Fig. 1A as compared to Fig. 1B). The latter indicates that certain factors in berry homogenates are activated at higher temperature (37 vs 4°C) which leads to complete loss of SARS-CoV-2 infectivity.

Effect of pH of berry homogenates on infectious SARS-CoV-2

The pH for strawberry homogenates ranged between experiments from 3.1 to 3.6 while that for blueberries between 2.8 and 3.4. Therefore, to understand whether the berry pH affected the virus infectivity, the pH of berries was pre-neutralized to ~7 prior to incubation with SARS-CoV-2 at 37°C. Surprisingly, neutralizing the pH of the strawberry homogenates significantly enhanced the recovery of SARS-CoV-2 at the 2-min time point by 1.4 log (# in Fig. 1B as compared to C). Also, the latter rendered the virus recovered at 2 min not to be significantly different between berries (Fig. 1C). Additionally, at 60 min, the infectivity of the virus in blueberry homogenates was partially restored at 37°C but still significantly lower than that observed at 4°C (3.1 log versus 4.03 log, respectively) (• in Fig. 1C as compared to A). As for the virus in strawberry homogenates at 60 min, the infectivity titer was restored to a significantly higher titer than that recovered under 4°C incubation (3.9 log versus 3.2 log). A similar effect for neutralizing the pH was observed at 120 min, with the exception that the virus titers in strawberry homogenates were not significantly different than that at 4°C (3.4 log versus 3.1 log, respectively) (Fig. 1C and A). Comparing between the two incubation periods at 60 and 120 min, the infectivity titers did not significantly differ for both pre-neutralized berry homogenates (Fig. 1C).

Taken together, both higher temperature and lower berry pH contributed to the loss of SARS-CoV-2 infectivity in berry homogenates at 60 to 120 min (Fig. 1A to C). Blueberries and strawberries may have different intrinsic factors affecting the virus infectivity as observed by the difference in the virus recovery at 2 min and its enhancement with pH neutralization in strawberry homogenates more than in blueberry. Time only had significant effects between the 2 and 60 or 120 min but not between 60 min and 120 min (Fig. 1A to C).

Effect of simulated in vitro digestion on the infectivity of SARS-CoV-2 in berry homogenates

The stability of SARS-CoV-2 infectivity in berry homogenates as it passes through the oral and GI tract was evaluated using two models of digestion simulating the oral, gastric, and intestinal phases of digestion. In the first model, the fluids’ composition and digestive enzymes were based on the INFOGEST standardized model. In the second model, the commercially available biorelevant fluids which simulate western diets during either the fasting or feeding states were used. Both IVD fluids were supplemented with similar concentrations of digestive enzymes. The log reduction in infectivity of SARS-CoV-2 was determined after consecutive incubations in the oral (2 min), gastric (additional 8 and 58 min for a total of 10 and 60 min), and intestinal (additional 5 and 60 min for a total 15 and 120 min) phases of digestion at 37°C.

In vitro digestion based on INFOGEST model

During the 2-min oral phase, SARS-CoV-2 infectivity showed higher significant reduction when incubated in strawberry homogenates as compared to incubation in fluids alone (1.2 log vs 0.6, respectively) (Fig. 2A). The virus mixed in blueberry homogenates showed a ~1 log reduction in infectivity, but this reduction was not significantly different from virus incubated in salivary fluids or strawberry homogenates (Fig. 2A).

Fig 2.

Reduction in SARS-CoV-2 infectivity titer (log TCID₅₀ per milliliter) in berry homogenates subjected to in vitro digestion simulating consecutive (A) oral, (B) gastric, and (C) intestinal phases. Fluids were prepared based on INFOGEST standard model. Samples were incubated for a period of 2 (oral), 10 and 60 (gastric), and 15 and 120 min (intestinal) at 37°C. Samples were neutralized after the gastric phase prior to addition of intestinal fluids. At end of incubation periods, samples were centrifuged, and the supernatants were tested using TCID₅₀ assay for virus quantification. Comparing treatments within a digestion phase, means with different letters differ significantly (P < 0.05). Significant differences among oral, gastric, and intestinal phases within corresponding treatments were designated with asterisk sign (*) for comparing panel A to B, with the number sign (#) for comparing panel B and C, and with dot sign (•) for comparing panel C to A.

Early in the gastric phase, at 10 min, the overall virus log reduction (≤1 log reduction) was not significantly different between berry homogenates or as compared to virus in fluids (Fig. 2B). This reduction in virus infectivity was not significantly different from that obtained during the oral phase (Fig. 2A and B). However, by 60 min, virus in blueberry homogenates showed the highest significant reduction as compared to virus in strawberry homogenates or fluids (3 log versus 2 and 1.5 log, respectively) (Fig. 2B). This reduction in virus infectivity at 60 min was significantly higher for virus in both berry homogenates as compared to that at 10 min (Fig. 2B).

During 15 min of intestinal digestion, the virus log reduction again was not significantly different between berry homogenates or fluids (~2.8 to 3.3 log) (Fig. 2C). This reduction, however, was significantly higher than that observed during the 10-min gastric phase for all treatments (# in Fig. 2B as compared to C). By the end of 120 min of consecutive digestion, there was a ≥3.9 log reduction in virus infectivity in fluids or berry homogenates (Fig. 2C). This increase at 120 min was only significantly higher for virus in strawberry homogenates as compared to that at 15 min (Fig. 2C). In addition, comparing virus reduction during the time points 60 min gastric to 120 min intestinal phases, a significant increase in virus reduction was observed for virus in fluids and strawberry homogenates, but not in blueberry homogenates (# Fig. 2B as compared to C).

In vitro digestion based on biorelevant fluids simulating fasting state

During the 2-min oral phase, virus in blueberry homogenates showed significantly higher reduction as compared to virus in fluids but not to strawberry homogenates (1.6 vs 1 and 1.2 log, respectively) (Fig. 3A).

Fig 3.

Reduction in SARS-CoV-2 infectivity titer (log TCID₅₀/mL) in berry homogenates subjected to in vitro digestion using biorelevant fluids simulating the western diet (A) oral phase (2 min), (B) followed by gastric phase (fasting state simulating gastric fluid) and then followed by (C) intestinal phase (fasting state simulating intestinal fluid) under fasting state. Sample processing and statistical analyses are consistent with the previous figure.

Within 10 min under gastric digestion simulating fasting state, there was no significant difference in SARS-CoV-2 infectivity between fluids or blueberry and strawberry homogenates (~2.3, 2.3, 1.8 log, respectively) (Fig. 3B). This reduction in virus infectivity was significantly higher than that obtained during the oral phase (* in Fig. 3B as compared to A). Within 60 min of gastric digestion under fasting state, the virus infectivity reduction in blueberry was significantly higher than that of virus in fluids, but not different from that of virus in strawberry homogenates (4.4 vs 3.3 and 3.9, respectively) (Fig. 3B). All virus infectivity reductions at 60 min were significantly higher than those at 10 min (Fig. 3B).

Within 15 min of the intestinal digestion simulating fasting, reduction in SARS-CoV-2 infectivity was significantly higher in strawberry homogenates as compared to both virus in blueberry homogenates or virus in fluids (4.4 vs 3.5 and 4 log, respectively) (Fig. 3C). By the end of 120 min, there was a ≥4.4 log reduction in virus infectivity in fluids or berry homogenates (Fig. 3C). This increase at 120 min was significantly different for virus in blueberry homogenates as compared to that at 15 min (Fig. 3C).

In vitro digestion based on biorelevant fluids simulating feeding state

The oral phase of digestion experiment was common between both the fasting and feeding IVD simulation. Early during gastric digestion simulating feeding state, the virus reduction in fluids was relatively low (<1 log) and not significantly different between berry homogenates (Fig. 4A). Within 60 min of gastric digestion, there was a significantly higher virus reduction in blueberries as compared to virus in both fluids or strawberry homogenates (3.4 vs 1.5 and 2 log) (Fig. 4A). This increase in virus inactivation was significantly higher as compared to the one observed at 15 min especially for berries (asterisks in Fig. 4A). With the subsequent addition of the intestinal fluid and by the end of 15 min, virus reduction reached ~2.6 log for all treatments (Fig. 4B). By the end of 120 min, there was a ≥3.4 log reduction in virus infectivity in fluids or berries (Fig. 3C). This increase at 120 min was significantly different for virus in strawberry homogenates as compared to that at 15 min (Fig. 4B).

Fig 4.

Reduction in SARS-CoV-2 infectivity titer (log TCID₅₀ per milliliter) in berry homogenates subjected to in vitro digestion using biorelevant fluids simulating the western diet (A) gastric (FEDGAS: Fed state gastric) then (B) followed by intestinal (FeSSIF: Fed State Simulated Intestinal Fluid) phase under feeding state. Sample processing and statistical analyses are consistent with the previous figure.

Overall, both models showed that during oral phase ~1 log inactivation in SARS-CoV-2 infectivity occurred in all groups. A consistent trend was observed under all models, revealing that the virus in fluids showed less inactivation at the 60-min incubation under gastric digestion and the virus in strawberry showed less inactivation than virus in blueberry homogenates. The biorelevant feeding model of digestion closely mimicked the INFOGEST model revealing that SARS-CoV-2 is minimally inactivated (~1 log) early during the gastric phase regardless of the matrix (fluids or berries). However, surviving infectious SARS-CoV-2 entering the small intestine are inactivated in all groups by ~2.6–3 log within an additional 5 min of intestinal digestion. Those viruses surviving during the 60-min gastric digestion were further inactivated in all groups during the additional 60-min intestinal digestion reaching a 3.4 log reduction. The latter is the maximum log inactivation that can be detected with our fed-state assay. In contrast, the biorelevant fasting model of digestion revealed that SARS-CoV-2 is inactivated early on by ~2 log at the gastric phase regardless of the matrix (fluids or berries) and more inactivation occurred at the intestinal phase, within an additional 5 min, in all groups (≥ 3.5 log). As gastric digestion proceeded, more inactivation occurred (≥3.3 log) in all treatments. Those viruses surviving during the 60-min gastric digestion were highly inactivated in all groups, reaching the maximum log inactivation that can be detected with our fast-state assay during the additional 60 min of intestinal digestion (~4.4 log).

DISCUSSION

Previous studies examined the effect of individual simulated digestive fluids on either SARS-CoV-2 or on other human coronaviruses, often without the addition of digestive enzymes, testing the consecutive effect of these fluids or including food (28, 29, 36, 37). The consensus among these studies is that SARS-CoV-2, MERS-CoV (Middle East Respiratory Syndrome coronavirus), and human coronavirus-229 and -OC43 (HCoV-229E and HCoV-OC43) are highly inactivated in fast-state gastric fluids (pH 1.5), showing >3.5–5 log reduction, and in fed-state intestinal fluids (pH 5), showing >2.5 to 4 log reduction in 60 min. However, the discrepancy between these studies is regarding the effect of intestinal fluids during fasting state (pH 6.5). Two studies using 229E, SARS-CoV-2, and recombinant SARS-CoV-2 tagged to a fluorescent protein found minimal inactivation (0.1–0.6 log) after 60 min of incubation in these fast-state intestinal fluids (28, 29), while our previous research using SARS-CoV-2 and another study using OC43 found ~4.5 and 2.5 log inactivation in 60 min, respectively, in the same biorelevant fast-state intestinal fluids (32, 38). With the exception of our previous study, none of these studies supplemented the intestinal fluids with pancreatin. We previously showed that pancreatin in intestinal fluids contributes to a significant dose-dependent inactivation of the virus (~2 log inactivation) (32). Additionally, these previous studies did not examine the effect of fed-state gastric fluids at pH 3 on SARS-CoV-2. The gastric fed-state fluids were examined for their effect on either MERS-CoV, OC43, 229E, or on SARS-CoV-2 but using fed-state gastric fluids of pH 5. Specifically, in one study, 229E and MERS-CoV were found to be completely resistant to gastric fed-state fluids (pH unknown) after 120-min incubation (37). In contrast, in gastric fed-state fluids (pH 3), OC43 and 229E were found to be inactivated by 3.5 and 2.4 log in 60 and 120 min, respectively (36, 38). In our previous study and this current study, we found that SARS-CoV-2 in fed-state gastric fluid at pH 3 was reduced by 3.5 log (32) and 1.5 log at 60 min. The difference between our two studies is that in this current study, the gastric fed-state fluid contained salivary fluids (oral phase) which may have protected the virus during the gastric phase (discussed below). The one study examining fed-state gastric fluids effect on SARS-CoV-2 and also 229E virus used the fluids at pH 5 to show no effect on the infectivity of both viruses after 120-min incubation (28). The discrepancy between all of the studies is likely due to the exact pH of the fed-state gastric fluid. We have shown that a slight difference in pH especially between pH 3 and 3.5 has a significant difference on virus inactivation (1.2 vs 3.2 log reduction), while pH 5 has no significant inactivation effect (32). Taken together, and despite the aforementioned limitations in these previous studies, it can be concluded that if SARS-CoV-2 is ingested in fed state, the virus can escape gastric inactivation depending on emptying time and the pH of the stomach.

The primary function of the oral cavity is to make the food ready to be swallowed. The texture of the food is changed after being wetted and lubricated by salivary fluids (35). Under both models, our results showed consistently that during the oral phase, SARS-CoV-2 is inactivated by ~1 log in salivary fluids or in berry homogenates. It is known that the oral mucosa is a site for SARS-CoV-2 infection and replication (15). In fact, the severity of SARS-CoV-2 infection has been correlated to oral viral load, and thus, detection of the virus in saliva may represent early stages of infection (14). Once SARS-CoV-2 is in the mouth, it can replicate in the oral glands and mucosa and provide input to the gastrointestinal tract as well as be source for viral transmission to other hosts (15). Additionally, our results showed that upon consecutive addition of gastric fluids, SARS-CoV-2 in saliva was less inactivated than viruses in berry homogenates during the gastric phase for as long as 60 min. Saliva may have protective effect on SARS-CoV-2 as has been shown for swallowed influenza viruses (39). Thus, our results raise concern that SARS-CoV-2 on contaminated food may initiate infection of the oral mucosa.

The main purpose of the stomach is to regulate the amount of digesta emptied to the small intestine in order to optimize digestion (35). Our study is the first to show that under consecutive phases of digestion and when mixed in berries or salivary fluids, SARS-CoV-2 can escape stomach gastric barrier early during digestion (≤1 log reduction in 10 min) under fed state but not under fast state (≥2 log in 10 min). Different food may have different intrinsic factors that may enhance or protect SARS-CoV-2 during GI passage. Interestingly, in the gastric phase and under both IVD models and both feeding and fasting states, we observed a trend whereby a lower inactivation of SARS-CoV-2 was shown to occur at 60 min in strawberry homogenates as compared to virus in blueberry homogenates. One explanation could be the lower pH of blueberry compared to strawberry, and as shown in the first experiments, pH contributed differential inactivation of SARS-CoV-2 in berries (Fig. 1C). A previous study using human coronavirus OC43 demonstrated that the virus on cucumber and under biorelevant fluids simulating gastric fed state (pH 3) showed 1.2 log less inactivation as compared to virus in fluids alone (38), whereas in biorelevant fluids simulating fast state, cucumber had no significant difference on virus inactivation for either gastric or intestinal digestions (38).

From a microbial viewpoint, the stomach is deemed as a gastric barrier against invading pathogens due to its high acidity. However, enteric non-enveloped viruses such as rotaviruses and enteroviruses infectivity titers were shown to be relatively stable under both fast and fed gastric biorelevant fluids for 120 min (28, 29), and thus are much better at bypassing the gastric acidity barrier (40). On the other hand, the notion that stomach acidity is enough to completely inactivate enveloped viruses may not always hold. This is because the gastric pH varies between 1.23 and 6.7 depending on whether the individual is in a fasting or feeding state, respectively (41). During eating, the pH of the stomach rises to ~6, and the meal’s effect on gastric pH may still be apparent for over 3 h, after which the pH decreases to ~2.5 (41). It is worth mentioning that the fed-state gastric pH 3 used in our study represents when the stomach is 75% empty within 3–6 h of consuming a meal. During the first hour of digesting a meal, the gastric pH is closer to 6 and therefore more SARS-CoV-2 is expected to escape inactivation as shown by Lee et al. where SARS-CoV-2 infectivity was stable under fed-state gastric pH 5 for 120 min (28). In fact, MERS-CoV injected intra-gastrically in mice survived the stomach acidity and was shown to replicate in the small intestine, causing enteric infection with a sequential respiratory infection, and the outcomes were worse when the mice were pre-treated with anti-acids drugs (37). Similarly, SARS-CoV-2 intra-gastrically injected in rhesus monkeys resulted in productive infection of digestive tissues, impairment of GI barrier, and severe infections (42). Taken together, our results provide evidence that due to the dynamic nature of gastric emptying over time and the higher gastric pH early during eating, SARS-Cov-2 can escape stomach inactivation. Furthermore, strawberries had a relatively more protective effect compared to blueberries during gastric digestion.

The small intestine provides a large surface area ~60 m² for absorption of nutrients. SARS-CoV-2 ACE2 receptor is found at the highest expression level across the full length of the small intestine, in low expression on the colon, and very low in the stomach and the rectum (12). Thus, there are ample viral receptors on the intestine to allow SARS-CoV-2 initiation of infection. Our results showed that SARS-CoV-2 may retain its infectivity as it passes through the stomach. However, early on during intestinal digestion (15 min), there was high inactivation of SARS-CoV-2 (2.3–3.3 log INFOGEST to 2.8 in biorelevant feeding IVD models). The latter suggested that only in cases of high SARS-CoV-2 contamination of berries (>3.3 log TCID₅₀ per milliliter), and when these are consumed as part of a meal, would escaping infectious SARS-CoV-2 have a chance to interact with intestinal ACE2 receptors. Further inactivation of SARS-CoV-2 infectivity occurred in the intestine as digestion proceeded (~3.5–4 log in 120 min), thus initial replicating viruses in the intestine would be expected to be inactivated as they are released to the intestinal lumen. The latter may explain why SARS-CoV-2 RNA is frequently detected in feces whereas infectious viruses are rarely detected. However, this requires further investigation in animal models fed SARS-CoV-2 with food. Previous animal studies showed that ferrets fed with low dose of SARS-CoV-2 (~2 log₁₀ RNA copies per milliliter in a fecal specimen) experienced mild symptoms such as slight increase in body temperature and rhinorrhea (43), while hamsters ingesting 10⁵ infectious SARS-CoV-2 showed asymptomatic mild infection of the respiratory and GI tract (28). However, using a higher dose of 10⁷ plaque forming unit (PFU) SARS-CoV-2 injected intra-gastrically to rhesus monkeys caused impairment of the GI barrier and severe infections in both the lung and the intestine (42). Therefore, while our data using IVD models show cumulative inactivation of the virus as digestion proceeded from oral to gastric to intestinal, surviving viruses may initiate infection along the GI tract especially in the oral and intestinal sites, depending on the original SARS-CoV-2 viral load on foods.

Conclusions

Our data showed for the first time that infectious SARS-CoV-2, when ingested on berries, can escape stomach barrier. Strawberries are more protective than blueberries during gastric digestion but as the virus enters the intestinal digestion phase, a high level of inactivation occurred. Our laboratory findings suggest that infection from SARS-CoV-2 on berries is only plausible under very high levels of virus contamination. Further testing in animal models is needed to confirm these conclusions.

MATERIALS AND METHODS

Virus propagation and cell culture

The USA reference strain SARS-CoV-2 USA-WA 1/2020 (BEI resources NR-52281) was propagated in African Green Monkey kidney cells (Vero-E6 ATCC CRL-1586) as described previously (26). Handling of SARS-CoV-2 was done under strict BSL3 biosafety protocols at the Center for Food Safety BSL3 laboratory. Briefly, 1- or 2-day-old 90% confluent cells were used to prepare virus stocks using a multiplicity of infection of 0.01. Infection media consisted of Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 2% heat-inactivated fetal bovine serum and 1% antibiotic-antimycotic cocktail. Harvesting the virus was done at 72 h post-infection. Infected cells were collected from the flasks and centrifuged at 450 × g for 5 min at 4°C to pellet the cell debris, while supernatants containing the virus were ultra-filtered through Amicon 100 KDa Ultra-15 centrifugal devices (Millipore) immediately after harvest to concentrate the viruses 10 times and to partially remove the cell culture media used in the infection. An aliquot of the virus was immediately titrated as described below. The original viral titer generated was ~7 log₁₀ TCID₅₀ per milliliter, while the ultrafiltered virus titer was ~8 log₁₀ TCID₅₀ per milliliter.

SARS-CoV-2 in berry homogenates

Strawberries and blueberries were bought from local grocery stores, washed, and homogenized in sterile water at 1:1 (wt/vol) using Hamilton Beach 2-speed Hand Blender (Walmart, Griffin, Georgia). Berry homogenates were aliquoted in Eppendorf tubes and transferred to the BLS3 lab. SARS-CoV-2 at a titer of ~7 log TCID₅₀ per milliliter was added to freshly made berry homogenates at 1:10 (vol/vol) and immediately incubated at 4°C or 37°C. Control positive was virus added to sterile water, while control negative was berry homogenates without any virus added. The incubation period was for 2, 60, and 120 min. The pH of the blueberry and strawberry homogenates was pre-determined on a subset of aliquoted samples (pH 2.8–3.4 and 3.1–3.6, respectively) and the volume of 2.5 M NaOH required to neutralize the pH was determined and back-confirmed on another set of berry homogenates aliquots. At the end of the incubation period, the berry homogenate samples were neutralized. Then, all samples were centrifuged at 10,000 rpm for 5 min at 4°C. The supernatants were transferred to new Eppendorf tubes containing 1% antibiotic-antimycotic cocktail and immediately used to quantify virus titers using the TCID₅₀ assay described below.

Effect of in vitro simulated digestion on the infectivity of SARS-CoV-2 in berry homogenates

The stability of SARS-CoV-2 infectivity in berries as it passes through the GI tract was evaluated under three phases of simulated digestion using two IVD models: INFOGEST (35) and biorelevant which includes both fasting and feeding states (Biorelevant, UK). First, the SARS-CoV-2 infectivity was evaluated under the oral phase alone (2 min), then under consecutive gastric (additional 8 and 58 min for a total of 10 and 60 min), and then intestinal (additional 5 and 60 min for a total 15 and 120 min) phases of digestion at 37°C.

INFOGEST model fluids

Salivary, gastric, and intestinal fluids were prepared as described in the INFOGEST protocol (35). Briefly, stock solutions of 0.5 M KCl, 0.5 M KH₂PO₄, 1 M NaHCO₃, 2 M NaCl, 0.15 M MgCl₂(H₂O)₆, 0.5 M (NH₄)₂CO₃, and 0.3 M CaCl₂ (H₂O)₂ were prepared and sterilized by autoclaving. To make the simulated salivary, gastric, and intestinal fluids, these chemicals were mixed at the volumes indicated for each fluid as described in the published protocol (35). The pH for salivary, gastric, and intestinal fluids was adjusted to 7, 3, and 7 respectively. The digestive enzymes were prepared as described previously (32, 35). Briefly, α-amylase from human saliva was prepared as 1,000 units per milliliter in salivary fluids (pH 7) and used at a final concentration of 75 U/mL. Porcine pepsin was prepared in gastric fluid (pH 3) as 100 mg/mL and used at a final concentration of 8 mg/mL. Porcine pancreatin was prepared in intestinal fluids (pH 7) as 50 mg/mL and used at a final concentration of 5 mg/mL. Porcine bile was prepared in intestinal fluids (pH 7) as 20 mg/mL and used at a final concentration of 2 mg/mL. All chemicals and enzymes were bought from Sigma-Aldrich. Enzymes, bile, and CaCl₂ (H₂O)₂ were added freshly to the IVD fluids just prior to their use in the experiments.

Biorelevant model fluids

The biorelevant fluids used were re-constituted as per manufacturer’s instructions (Biorelevant, UK). The chemical composition of these fluids are as follows: the fasting state simulating gastric fluids (FaSSGF, pH 1.6) contained 0.08 mM taurocholate, 0.02 mM phospholipids, 34 mM sodium, and 59 mM chloride. The fasting state simulating intestinal fluids (FaSSIF, pH 6.5) contained 3 mM taurocholate, 0.75 mM phospholipids, 148 mM sodium, 106 mM chloride, and 29 mM phosphate. For fed state, the gastric fluid fed state Gastric [(FEDGAS), pH 3] representing a high-fat Food and Drug Adminstration (FDA) meal and containing fats, carbohydrates, dietary fibers, and bile salts at 62.5, 62.5, 3.1, and 0.3 g/900 mL, respectively was used. FEDGAS pH 3 represents a high-fat FDA meal when the stomach is 75% empty within 3–6 h after a meal. The intestinal fluid simulating feeding was Fed State Simulated Intestinal Fluid [(FeSSIF), pH 5] containing 15 mM taurocholate, 3.75 mM phospholipids, 319 mM sodium, 203 mM chloride, and 144 mM acetic acid. The gastric fluids FaSSGF and FEDGAS pH 3 were supplemented with pepsin (8 mg/mL) while the intestinal fluids FaSSIF and FeSSIF were supplemented with pancreatin (5 mg/mL).

For both IVD models: berry homogenates were prepared as described above; however, instead of water, they were homogenized directly in simulated salivary fluids. There are no commercial salivary fluids for the biorelevant model; therefore, human saliva (Lee Biosolutions, Maryland Heights, MO, USA) was used to prepare berry homogenates. The human saliva was mixed at 1:1 (vol/vol) ratio with berries homogenized in simulated salivary fluids. SARS-CoV-2 was added to fluids or berry homogenates at 1:10 (vol/vol). Control positive was virus inoculated in water only (without any subsequent treatments) and incubated at 37°C for same incubation periods. For gastric samples proceeding to intestinal digestion, the samples were neutralized. The latter will also inactivate pepsin, as neutral pH denatures this enzyme (44). All samples were neutralized, adjusted to the same final volume using sterile water before being centrifuged at 10,000 rpm for 5 min at 4°C. The supernatants were transferred to new Eppendorf tubes containing 1% antibiotic-antimycotic cocktail and placed on ice to stop the action of intestinal enzymes. For biorelevant simulating feed-state intestinal fluids, there was a thin layer of fat forming on the surface of the liquid following centrifugation. The liquid in between was carefully transferred into new tubes, supplemented with 1% antibiotic-antimycotic, and used immediately to test for SARS-CoV-2 infectivity. Cytotoxicity was always evaluated by incubating a serial dilution of the treatment alone on Vero-E6 cells. The detection limit for all fluids was 1.72 log TCID₅₀ per milliliter with the exception of feeding state experiments, the detection limit was 2.72 log TCID₅₀ per milliliter.

Determination of virus infectivity

The TCID₅₀ assay was performed as described in our previous research (45). Briefly, 1 - to 2-day-old 90% confluent cell monolayers in 96-well plates were infected in quadruplet with serially diluted samples (1:10) in cell culture media supplemented with 2% fetal bovine serum (FBS) and 1% anti-anti and incubated at 37°C. The plates were inspected for cytopathic effects between day 4 and 5 post-infection. Viral titers were estimated following the Reed-Muench equation for the calculation of TCID₅₀ (46). In addition to experimental controls, control positive (virus with known titer) and control negative (cell culture media) were included in each experiment.

Statistics

Experiments were independently repeated three times and three technical replicates were tested for each treatment. GraphPad Prism version 5 (GraphPad Software, USA) was used for all statistical analyses. The entire data set was transformed to log₁₀. The log reductions in infectivity were calculated based on SARS-CoV-2 infectivity in control samples incubated under the same conditions. One-way or two-way analysis of variance followed by Tukey or Bonferroni post-tests, respectively, were used to determine significant differences in mean infectivity titers. The factors analyzed included time, temperature, and treatment. Differences in means were considered significant when the P-value was less than 0.05 and are denoted in the figures by different alphabets, asterisks (*), number sign (#), and dot (•). Data are expressed as the mean ± standard error.