Anti-rotavirus Properties and Mechanisms of Selected Gram-Positive and Gram-Negative Probiotics on Polarized Human Colonic (HT-29) Cells

Probiotics have been investigated to improve the universal rotavirus (RV) vaccination as well as to ameliorate the RV infection. However, underlying mechanisms how probiotics mediate beneficial effects needs more investigation. Thus, in the present study, we used polarized HT-29 cells to assess the anti-RV properties of Gram-positive, (Lactobacillus acidophilus, Lacticaseibacillus rhamnosus GG, and Bifidobacterium subsp. Lactis Bb12) and Gram negative, (Escherichia coli Nissle 1917) probiotics and study their underlying mechanisms. Our results showed that pre-treatment of HT-29 cells for 4 h with probiotics, significantly reduced (p < 0.05) human RV replication and this effect was most pronounced for E. coli Nissle followed by L. acidophilus and L. rhamnosus GG. Strikingly, only pre-treatment with live bacteria or their supernatants demonstrated anti-RV properties. Except Gram negative E. coli Nissle, the Gram-positive probiotics tested did not bind to RV. Ingenuity pathway analysis of tight junction (TJ)- and innate immune-associated genes indicated that E. coli Nissle or E. coli Nissle + RV treatments improved cell–cell adhesion and cell contact, while L. acidophilus or L. acidophilus + RV treatments also activated cell–cell contact but inhibited cell movement functions. RV alone inhibited migration of cells event. Additionally, E. coli Nissle activated pathways such as the innate immune and inflammatory responses via production of TNF, while RV infection activated NK cells and inflammatory responses. In conclusion, E. coli Nissle’s ability to bind RV, modulate expression of TJ events, innate immune and inflammatory responses, via specific upstream regulators may explain superior anti-RV properties of E. coli Nissle. Therefore, prophylactic use of E. coli Nissle might help to reduce the RV disease burden in infants in endemic areas.


Introduction
Infectious gastroenteritis is a leading cause of morbidity and mortality in infants worldwide. Rotavirus (RV)-induced gastroenteritis is one such vaccine preventable disease associated with 215,000 deaths annually worldwide in 2013 [1]. Several factors, including malnutrition, micronutrient deficiencies, breastfeeding, maternal immunity, histo-blood group antigen type, composition of gut microbiota, and medication use have been suggested to reduce efficacy of enteric vaccines in developing countries [2][3][4][5][6][7][8][9]. Typical symptoms include vomiting, watery diarrhea, and fever. The fecal-oral route is the established mode of transmission. Upon ingestion, RV infects mature small intestinal enterocytes and leads to diarrhea via (i) destruction of enterocytes, (ii) downregulation of the absorptive enzymes, (iii) inflammation of the gut, and (iv) compromised gut barrier [10,11]. Thus, an efficacious and universal treatment to prevent or alleviate RV diarrhea in infants must be capable of (i) regulating immune responses (inflammation), (ii) restoring barrier functions, (iii) stimulating enterocyte proliferation and repair, and (iv) inhibiting RV replication in the gut.
Probiotics represent a potential universal anti-RV treatment, and thus, their effects are being studied widely in combination with vaccines, antibiotics, and/or oral rehydration therapies in clinical and animal studies [12][13][14][15][16][17][18]. Most of the probiotics are Gram-positive (G+) bacteria; therefore, a wealth of literature regarding probiotic effects on amelioration of RV disease is derived from bacterial species that belong to Lactobacillus and Bifidobacterium genera [16]. The clinical efficacy of probiotic Lactobacillus acidophilus against RV was demonstrated in infants where L. acidophilus treatment resulted in decreased severity of RV disease, characterized by improved stool consistency and reduced duration of diarrhea [19][20][21]. Similarly, in randomized clinical trials, Lacticaseibacillus rhamnosus GG administration was shown to shorten the duration of RV diarrhea in children [21][22][23]. Further, prophylactic supplementation of L. rhamnosus LGG reduced the risk of nosocomial diarrhea and rotavirus gastroenteritis in infants [24]. In a clinical study that compared probiotic treatment to oral rehydration therapy, children consuming probiotic Bifidobacterium subsp. lactis exhibited significantly reduced duration of RV diarrhea [25]. Our group and others have demonstrated that L. acidophilus, L. rhamnosus GG, and B. lactis Bb12 were efficacious in reducing the severity of RV diarrhea in the gnotobiotic piglet model [13,16,18,26]. However, the Gram-negative probiotic Escherichia coli Nissle-1917 has not been tested for anti-RV properties in infants [27], although several animal studies from our group highlighted its superior characteristics compared to other Gram-positive probiotics in ameliorating RV disease [13,14,18]. A randomized, double-blind clinical trial indicated that administration of E.coli Nissle successfully alleviated the idiopathic chronic constipation without any major side effects [28].
Despite cumulative evidence of extensively evaluated Gram-positive and Gram-negative probiotics in clinical and animal studies [13], their anti-RV properties and mechanistic insights are poorly investigated [29]. Therefore, the objectives of this study were to (i) investigate and compare anti-RV properties of Gram-positive probiotics (L. rhamnosus GG, L. acidophilus, B. animalis subsp. lactis Bb12) and Gram-negative probiotic (E. coli Nissle1917), and (ii) investigate the mechanisms by which E. coli Nissle modulates rotavirus infections in vitro. Numerous cell lines have been used to investigate the in vitro mechanisms of probiotics. The distinct features of human colonic adenocarcinoma (HT-29) cells closely mimicking the in vivo functional intestinal epithelium [30][31][32][33][34] make the HT-29 cells an ideal model. These differentiated cells possess apical brush border proteins, Cl-channels, Cl-secretion, mucus production, disaccharidases and peptidases, domes on impermeable substrates, trans-epithelial resistance (TER), and intracellular tight junction proteins similar to intestinal epithelium. In this study, we have established a polarized HT-29 cell monolayer model to investigate anti-RV properties of the abovementioned probiotics. We tested three different probiotic treatment regimens to determine their effects on RV infection. (1) Pre-inoculation: This approach mimics the in vivo effects of probiotics administered prior to RV infection as a prophylactic measure in humans. (2) Co-inoculation: This regimen was used to model probiotic administration during ongoing RV infections in infants. (3) Pre-incubation and coinoculation: Our rationale was that incubating RV with probiotics would provide sufficient time to induce structural and functional alterations that would not be possible in the direct co-inoculation experiment. Of the four probiotics tested in this study, E. coli Nissle 1917 (Dr. Ulrich Sonnenborn) pretreatment exhibited the most prominent anti-RV properties, while L. acidophilus (ATCC 700396) and L. rhamnosus GG (ATCC-53103) induced intermediate and the least effects, respectively. E. coli Nissle's superior anti-RV properties were attributed to its ability to bind RV, modulate expression of TJ, innate immune response and PRR signaling genes, via specific upstream regulators.  [35,36]. The virulent human RV Wa G1P [8] strain at pig passages 25-26 was used in this study [37].

Culturing of Polarized HT-29 Cells
A unique feature of HT-29 cells observed in the absence of glucose and presence of galactose is that the cells closely mimic the in vivo enterocyte architecture. Therefore, in the present study, we adapted previously established protocols to induce HT-29 polarization [31][32][33]. Briefly, HT-29 cells (ATCC HTB-38 ™ ) were cultured in Dulbecco's modified Eagle's medium (Invitrogen, Grand Island, NY, USA) containing 4.5 g/L D-glucose (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco, Amarillo, TX, USA), 2 mM glutamine, 1% nonessential amino acids (Gibco, Amarillo, TX), and streptomycin-penicillin antibiotic (Sigma-Aldrich, St. Louis, MO) mix for 2 days. Subsequently, cells were cultured in DMEM with gradually decreasing concentrations of D-glucose (4 mM, 3 mM, 2 mM) with other cell culture ingredients added daily as mentioned above. After that, the cells were cultured in the presence of 1 mM glucose and 1 mM galactose (Sigma-Aldrich, St. Louis, MO, USA) for 24 h before replacing glucose source from medium. In subsequent days, galactose source was gradually increased in 1 mM increments to reach the final concentration of 5 mM with other ingredients. Finally, the cells were sub-cultured to from monolayer and freezer stocks of polarized HT-29 cells were prepared using routine cell culture techniques.Transmission electron microscopy (TEM) was carried as described previously [38] at the Molecular and Cellular Imaging Center (http:// www.oardc.ohio-state.edu/mcic) to confirm the polarization of the HT-29 cells. Here and onward, all in vitro experiments (unless specified otherwise) were performed using the polarized HT-29 monolayers in 96-or 48-wells plates with passage number ranging from P1 to P13. Each treatment including controls were performed in triplicate wells in 2 to 3 independent experiments. Average mean and standard deviation were used to express the results.

Probiotics-HT-29 Cell Adhesion Assay
E. coli Nissle, L. rhamnosus GG, L. acidophilus, and B. lactis Bb12 were tested to assess their cell adhesion properties after 30 and 60 min of incubation using multiplicity of infections (MOI) 0.01 as described previously [39]. HT-29 cells were washed with Dulbecco's phosphate-buffered saline (DPBS) without CaCl 2 and MgCl 2 (Invitrogen, Grand Island, NY, USA) 2-times and maintained in the cell culture medium without FBS for 2 h. Bacterial cultures were pelleted at 10,000 × g for 10 min and washed with antibiotic-free cell culture medium. Desired OD600 nm for different probiotic bacteria were adjusted in antibiotic-free cell culture medium according to the established standard curve (OD600 nm E. coli Nissle ~ 0.10, L. rhamnosus GG ~ 0.16, L. acidophilus ~ 0.23, and B. lactis Bb12 ~ 0.78) that corresponds to MOI of 0.01. HT-29 cells were inoculated with bacterial cells resuspended in antibiotic free cell culture media in triplicate wells. E. coli K-12 strain (OD600 nm ~ 0.10) and the cell culture medium were included as controls. At the end of the assay, media were carefully removed, and the cell monolayers were washed 2 times using the cell culture medium with antibiotics. The HT-29 cells were then lysed with 0.1% Triton-100X, and suspension was used for colony forming unit (CFU) enumeration.

RV Infection of Polarized HT-29 Cells
Earlier studies have demonstrated that RV can effectively infect HT-29 cells [40]. However, the efficacy of RV replication could be affected by HT-29 polarization status and RV strains used, which prompted us to determine the optimal conditions for RV infection in HT-29 cells. The infectivity of RV Wa strain to polarized HT-29 cells was evaluated using tenfold dilutions [1 × 10 8 to 1 × 10 3 focus forming units (FFU/mL)] of RV Wa. RV infection and quantification were measured as described previously [41].

Probiotic Treatment Regimens
We tested three different probiotic treatment regimens (pre-inoculation, co-inoculation, and pre-incubation and co-inoculation) to determine their effects on RV infection: 1. Pre-inoculation: This approach mimics the in vivo effects of probiotics administered prior to RV infection as a prophylactic measure in humans. Our rationale was since probiotic treatment induces beneficial changes in the host cells by up-regulating innate immune and tight junction genes that would either prevent or inhibit RV replication. 2. The co-inoculation: This regimen was used to model probiotic administration during ongoing RV infections in infants. Our hypothesis was that probiotic and RV may share similar binding sites on HT-29 cells, and thus probiotics could interfere with RV binding when probiotics were mixed with the RV particles and then allowed to infect HT-29 cells simultaneously. 3. Pre-incubation and co-inoculation: Our rationale was that incubating RV with probiotics would provide sufficient time to induce structural and functional alterations that would not be possible in direct co-inoculation experiment. We expected that direct interactions between probiotics and RV may alter ligand/receptor expression needed for productive infection of HT-29 cells. (a) Pre-inoculation: probiotic bacteria to HT-29 cell ratio 100:1 was used to pretreat the HT-29 cell monolayers for 1, 2, and 4 h prior to RV Wa infection. (b) Co-inoculation: RV and probiotic bacteria were mixed at the ratio of 1:100 and were used to inoculate the HT-29 cells.
(c) Pre-incubation/co-inoculation: a 1:100 mixture of RV and probiotic bacteria was prepared and incubated at room temperature for 1 and 2 h. At respective time interval, suspension was used to infect HT-29 cells.

Optimizing Probiotic Cell Killing
The cyclic freezing-thawing was used to kill the probiotic cells to better preserve the outer membrane structure compared to dry heat killing. For this, 5 mL of OD adjusted bacterial cultures were frozen (− 80 °C) for 24-, 48-, 72-h, and 1-week time interval and followed by thawing at 37 °C and an aliquot of the culture was used to enumerate the number of live bacterial cells on the respective solid agar. Increasing freeze-thaw cycles (7 freeze cycles for 1-week incubation) decreased the number of live cells but it did not result in 100% loss of viability. Thus, the dry heat killing method was used to conduct the experiment. The dry heat-killing protocol was optimized for these probiotics by incubating at different temperatures and for variable durations. The lowest temperature and the shortest duration that resulted in 100% loss of viability were selected to ensure that the bacterial outer membrane surface architecture was preserved as much as possible. The optimal condition was achieved for E. coli Nissle using 65 °C for 5 min [42] and for L. acidophilus, L. rhamnosus GG, and B. lactis Bb12 using 65 °C for 10 min. These conditions were used throughout the study. Killed bacteria were pre-incubated with the HT-29 cells for 4 h to determine the anti-RV effects.

Probiotic-RV Binding Using Flow Cytometry
Binding of probiotics to RV was determined by flow cytometry as described previously [13]. Briefly, probiotic bacteria stained with 5 µM SYTO 9 (Life Technologies, Carlsbad, CA, USA), followed by incubation with semipurified RV or Alexa Fluor 647 (Life Technologies, Carlsbad, CA)-conjugated rotavirus like particles (VLP) at 37 °C for 1.5 h. The unbound virus particle on the probiotic bacteria were removed by washing 3 times with sterile PBS and bacteria were incubated with Alexa Fluor 647-conjugated anti-RV mAb (clone RG23B9C5H11) or isotype control at 4 °C for 45 min. The samples were washed and bacterial-RV complexes were acquired using BD Accuri C6 flow cytometer (Ann Arbor, MI, USA).

Selection of Probiotics and In vitro Strategy to Investigate Anti-RV effects
Unlike co-incubation strategies, pre-incubation of probiotic has a number of advantages to demonstrate the probiotics anti-RV properties like the following: (i) probiotics can induce host innate and adaptive responses like defensins, anti-microbial peptides, etc. that might prevent or inhibit RV infection; (ii) probiotics can up-regulate tight junction proteins like ZO1 and occludin, thereby improve gut barrier function resulting in prevention/inhibition of RV infection; and (iii) probiotics that are strongly adhered to host cells might even compete or interfere with RV binding activity. Because of these advantages and as expected, outcome of different probiotic treatment strategies led us to investigate the mechanisms regulating probiotic effects using the preincubation regimen. We have focused on evaluating two probiotics (E. coli Nissle and L. acidophilus) that showed significant anti-RV properties compared to other two probiotics when incubated for 4 h.

Effect of E. coli Nissle and L. acidophilus Pretreatment on HT-29 Cells
Prior to conducting the experiments, we assessed the cytotoxicity of E. coli Nissle and L. acidophilus pre-treatment. The total number of cells and % dead cells were compared to no probiotic treatment at 0 and 4 h using conventional trypan blue staining technique.

Effects of E. coli Nissle and L. acidophilus Supernatant on RV Infection
Bacterial culture supernatants were harvested to determine their contribution to the observed anti-RV properties of E. coli Nissle and L. acidophilus. Briefly, overnight grown probiotic culture was centrifuged for 10,000 × g for 10 min to collect the supernatants. Collected supernatants were filtered through 0.22-micron filter to generate cell-free extracts. Further, portions of the original filtrates were used to prepare 10 × filtrates using vacuum concentrator (Thermo Fischer Scientific, Waltham, MA, USA). Instrument was run on low vacuum mode to preserve any labile molecule until one tenth of volume remained. In vitro pre-incubation experiment was performed at 4 h of treatment with 1 × and 10 × cell-free extract as described above.

RT 2 Profiler™ PCR Arrays Analysis
The expressions of sets of 84 genes involved in each PCR arrays, i.e., Human Tight Junction (TJ), Innate Immune Response were determined using RT 2 Profiler ™ PCR Array (Qiagen, Array # PAHS-143Z and PAHS-148Z, Germantown, MD, USA). The list of genes, 96 well format, protocol used to perform these arrays can be found on Qiagen RT 2 Profiler PCR array [42,43]. The genes list for TJ array included critical genes encoding proteins that form impermeable barriers between epithelia cells to regulate polarity, proliferation and differentiation (https://www.qiagen.com/us/shop/pcr/primer-sets/ rt2-profiler-pcr-arrays?catno=PAHS-143Z#geneglobe) while for, innate immune response and PRR signaling included cellular growth and development, proliferation and maintenance, and anti-inflammatory and proinflammatory responses and antimicrobial responses and cell apoptosis-associated genes. HT-29 cells in 48 well tissue culture plates were treated with E. coli Nissle and L. acidophilus for 4 h and then infected with RV as described previously [41]. Controls were included in the experiment: E. coli Nissle, E. coli Nissle+RV, L. acidophilus, L. acidophilus+RV, and RV. Treated HT-29 cells were washed and suspended in the TRIzol reagent (Life technologies, Carlsbad, CA USA). Total RNA was extracted from the wells using the miRNeasy Mini Kit (Qiagen, Germantown, MD),and traces of DNA were removed using the Qiagen RT2 First Strand Kit (Qiagen, Germantown, MD, USA). The cDNA was synthesized using the Qiagen RT2 First Strand Kit and analyzed using RT² Profiler ™ PCR Arrays. The Ct-values for each gene were normalized using house-keeping genes that were included in the TJ and innate response arrays. Subsequently, fold-changes in genes expression were determined using the ΔΔCt method. To examine the potential functions and cellular pathways that were modulated in HT-29 cells in response to different treatments, ingenuity pathway analysis (IPA; Qiagen, Redwood City, CA, USA) was performed as described previously [44]. Each treatment was performed in four replicate wells and was repeated in three independent experiments.

Statistical Analysis
One-way analysis of variance (ANOVA) followed by Sidak's multiple comparisons test was used for the probiotic adhesion assay, while unpaired t-test was used to analyze the results of the RV infection of HT-29 cells.
The data generated in all other experiments were analyzed using two-way ANOVA followed by Tukey's multiple comparisons. ANOVA was also used to analyze the qPCR data. A fold change of ± 1.5 ⩾ or ⩽ 1.5 and a p value ≤ 0.05 was used to determine significant differences in the expression of the genes. All the statistical analysis were performed using a GraphPad Prism software (San Diego, CA). Significance level of the data analyzed in IPA was determined by default Fisher's Exact test. In all statistical analysis a p value of ≤ 0.05 was used to determine level of significance.

TEM Confirmed Polarization of HT-29 Cells
HT-29 cells possessed both differentiated and undifferentiated epithelial cells of different types including goblet cells, M cells, and enterocytes, more closely mimicking in vivo conditions when grown in the presence of galactose [34,45]. Thus, we adapted the above protocol to induce HT-29 cells and confirmed the polarized state of HT-29 cells. Transmission electron microscopy (TEM) imaging indicated that polarized HT-29 cells were characterized by the presence of dense granules, large vacuoles, tight junctions, and microvilli (Fig. 1A, B), closely resembling the in vivo structure of intestinal epithelial cells.

Probiotic Bacteria Adhere to HT-29 Cells
Probiotic bacteria like B. lactis Bb12 and E. coli Nissle were shown to adhere to epithelial cell lines like Caco2; however, many factors are also known to influence the host cell adhesion ability [46]. Thus, in the present study, we confirmed the probiotic ability to adhere to the HT-29 cells before investigating their anti-RV properties. Compared to 30 min incubation (data not shown), higher numbers of bacterial cells adhered to HT-29 cells at 60 min. Gram-negative bacteria E. coli Nissle and E. coli K-12 adhered more efficiently to HT-29 cells compared to Gram-positive probiotic bacteria ( Fig. 2A).
There were no significant differences in the binding ability of the Gram-positive probiotic bacteria where 10% of bacterial cells were bound to HT-29 cells.

RV Infects Polarized HT-29 Cells
Rhesus monkey kidney (MA104) cells are more commonly used for the growth and characterization of both animal and human culture-adapted RVs. In addition to MA104 cells and depending upon the strain, RVs were shown to infect other types of continuous cell lines including HT-29 cells [40,47]. Our immediate question to answer was whether polarized HT-29 cells are equally efficiently infected by RV Wa strain used in our present study. Concurrent with previous findings, RV Wa strain infected polarized HT-29 cells, but to some degree less efficiently compared to MA104 cells. MOI of RV Wa strain for MA104 cells varied from 0.1 to 0.5 while HT-29 cells required 1 to 0.1 MOI [47]. Infection with either 1.0E+07 and 1.0E+06 FFU of RV resulted in comparable number of FFU (Fig. 2B); therefore, we used 1.0E+07 FFU for our subsequent infections. Interestingly, no RV was detected when infected with lower than 1.0E+06 FFU.

Pre-and Co-inoculation with E. coli Nissle, L. rhamnosus GG, and L. acidophilus Inhibited RV Replication in HT-29 Cells
Except B. lactis Bb-12, other probiotic pretreatments significantly (p < 0.05) inhibited RV replication in HT-29 cells (Fig. 3A). Irrespective of E. coli Nissle pre-treatment duration, RV Wa replication was inhibited, although most inhibition was observed at 4 h of pretreatment. While comparable trends were observed for L. acidophilus and L. rhamnosus GG, the magnitude of the inhibition was lower, and reached significance only at 4 h of pre-treatment (Fig. 3A). In the co-inoculation treatment, an overall similar trend was observed with pre-treatment, wherein co-incubation with E. coli Nissle, L. acidophilus and L. rhamnosus GG for 4 h resulted in significant RV inhibition in HT-29 cells (Fig. 3B). As observed with pre-treatment above, B. lactis Bb12 coinoculation did not affect RV replication (Fig. 3B). In the pre-incubation and co-inoculation regimen, like preand co-inoculation experiments, E. coli Nissle induced the most significant inhibition of RV infection, followed by L. acidophilus and L. rhamnosus GG that resulted in a slight numeric, but not significant inhibition. In contrast, B. lactis Bb12 incubation and co-inoculation resulted in significantly increased RV replication compared to RV alone and other probiotic treatments (Fig. 3C). Interestingly, RV alone infectivity in HT-29 cells decreased progressively during incubation at room temperature, which precluded us from performing the 4 h incubation and co-inoculation experiment.

Observed Anti-RV Effects of Probiotics Were Induced by Live Bacteria
To investigate the mechanisms of the probiotic effects, we first sought to understand whether live probiotics are essential to induce the observed anti-RV effects. Except for B. lactis Bb-12, pre-treatment with other live probiotics resulted in inhibition of RV as observed previously. However, this effect was not observed when heat-killed bacteria were used. Neither live nor dead B. lactis Bb-12 pre-treatment resulted in RV inhibition (Fig. 4A).

Only E. coli Nissle Binds to RV But Not L. acidophilus, L. rhamnosus GG, and B. lactis Bb12
The ability of probiotics to directly bind RVs is one of the mechanisms that contributes to the anti-RV properties as we reported previously [13]. Therefore, we investigated the probiotics-RV binding properties. Less than 2% of L. acidophilus and B. lactis Bb12 bound to RV (Fig. 4B), while significantly higher percentages (15%) of E. coli Nissle bind to RV and no RV binding by L. rhamnosus GG was evident.

E. coli Nissle and L. acidophilus Pre-treatments Did Not Result in Cytotoxic Effects in HT-29 Cells
Increasing duration of probiotic treatment is generally not recommended due to rapid multiplication of bacteria that ultimately damage the host cell monolayer. In our present study, except for E. coli Nissle, probiotic treatments did not induce any monolayer alterations. After, 4 h of treatment, E. coli Nissle induced up to 50% HT-29 cells monolayer clumping, which raised a question whether the observed superior anti-RV effects of E. coli Nissle were due to the loss of host cells viability resulting in reduced numbers of HT-29 cells available for RV infection. To address this concern, we determined the total number of cells and % dead cells. There was no difference in the total numbers and viability of HT-29 cells compared to no probiotic treatment contros. This suggests that the observed anti-RV effects were due to probiotic induced alterations of the cell physiology and/or structure (Fig. 5A).

Both E. coli Nissle and L. acidophilus Culture Supernatants Possess Anti-RV Properties
To establish if probiotic secreted factors present in culture supernatant induce beneficial host responses to prevent or interfere with RV replication in HT-29 cells, we assessed the effects of pretreatment with 1 × and 10 × cell free culture supernatants prepared from probiotics grown in their respective media [35,36]. Both 1× and 10× E. coli Nissle and L. acidophilus supernatants possessed RV inhibitory properties; however, therewere no differences between pretreatment with 1× and 10× E. coli Nissle and L. acidophilus supernatants. This suggests that E. coli Nissle and L. acidophilus secrete biologically active molecules essential for the observed anti-RV effects (Fig. 5B).

IPA
To further study the mechanisms of the E. coli Nissle and L. acidophilus anti-RV properties, we performed targeted PCR arrays and compared the fold-change expression of genes in HT-29 cells in probiotic (E. coli Nissle, L. acidophilus, E. coli Nissle + RV, and L. acidophilus + RV) and RV infection alone treated groups (Tables S1 and S2 and Figs. S1 and S4). A comprehensive summary of gene regulations was depicted as radar diagrams (Figs. 6 and 11). The IPA tool was used to predict the biological functions by analyzing networks and canonical pathways. The top two gene regulatory networks associated with E. coli Nissle, L. acidophilus, RV, L. acidophilus + RV, and E. coli Nissle + RV based on focus molecules and scores are shown in Table 1.

RV Has Limited Influence on TJ Gene Functions and Pathways Compared to Either of the Probiotic Treatments
RV influenced a limited number of pathways compared to probiotic and probiotic + RV treatments. For example, RV did not influence the electric resistance, transmigration of granulocytes (lymphocyte and neutrophil) functions (Fig. 7A). Correspondingly, only leukocyte excavation and PTEN signaling were affected by RV treatment (Fig. 7B). Paxillin signaling, which is involved in cell adhesion, was activated only by E. coli Nissle or E. coli Nissle + RV but not in other treatments. E. coli Nissle or E. coli Nissle + RV affected the most functions and pathways compared to others. Interestingly, L. acidophilus + RV treatment activated functions and pathways that were not affected by RV treatment alone. Though RV treatment influenced the function and pathways the least, it affected the greatest number of upstream regulators compared with that of probiotic or probiotic + RV treatments. The one upstream regulator CDX2, was only influenced by probiotic + RV treatment (Fig. 7C). CDX2 is a nuclear transcriptional factor involved in intestinal cell proliferation, differentiation, adhesion, and apoptosis. Overall, E. coli Nissle or E. coli Nissle + RV treatment affected several biological functions and cellular pathways while RV alone influenced the least. The important TJ gene clusters that were differentially regulated on RV and probiotics alone (E. coli Nissle and L. acidophilus) or probiotic + RV (E. coli Nissle + RV, and L. acidophilus + RV) treatments are indicated in Table 2.

E. coli Nissle Treatment Improves Cell-Cell Adhesion and Cell Contact
The five major predicted TJ cellular functions affected by the E. coli Nissle treatment alone included: cell polarity formation, cell-cell contact, formation of adherens junctions, cell movement, and cell-cell adhesion, wherein more specifically, cell-cell adhesion and cell contact biological functions were highly activated while, cell polarity was least activated (Fig. S2). The genes like ACTN1&4, AFDN, alpha actinin, CTNNA3, ERK1/2, JAM3, and MAGI1 lead to activation of cell-cell contact, while AFDN, alpha catenin, CTNNA3, and MAGI1 genes are responsible for the activated biological function of cell-cell adhesion. A complete list of genes involved in the respective biological functions is highlighted in Fig. S2. Similar cellular pathways (cell polarity formation, cell-cell contact, cell-cell adhesion, polarity of cells, formation of adherens junctions, and cell movement) were affected in E. coli Nissle + RV-treated cells; however, one additional pathway was also influenced, i.e., polarity of cells. Though all cellular pathways were activated, cell-cell contact, and polarity of cells were highly triggered (Fig. 8). The polarity of cells was activated due to ACTN4, F-actin, JAM3, and PRKCZ genes while the cell-cell contact was activated via ACTN1, ACTN4, AFDN, alpha catenin, ERK1/2, MAGI1, and Par6 genes.

Cell-Cell Contact Activated While Cell Movement Inhibited upon L. acidophilus Treatment
Affected cellular functions by the LA treatment alone included: cell polarity formation, cell-cell contact, organization of cytoskeleton, formation of cellular protrusions, formation of adherens junctions, and cell-cell adhesion. All the cellular functions were activated of which, cell-cell contact, and cell-cell adhesion were highly stimulated by the LA treatment (Fig. S3). Genes involved in the respective cellular functions are illustrated in Fig. S3. The genes ACTN1, AFDN, alpha catenin, CTNNA3, and MAGI1 simultaneously activated the functions of cell-cell contact and cell-cell adhesion (Fig. S3). While after LA + RV treatment, formation of tight junctions, cell polarity formation, cell-cell contact, formation of adherens junctions, cell movement, formation of intercellular junctions, and polarity of cells were affected (Fig. 9). (2) Cell-to-cell signaling and interaction, cellular assembly and organization, cellular function, and maintenance 37 16 (2) Antimicrobial response, cell death and survival, inflammatory response 21 10 L. acidophilus NA (1) Cell morphology, cellular assembly and organization, cellular development 55 22 (2) Cell-to-cell signaling and interaction, cellular assembly and organization, cellular function, and maintenance 40 17 L. acidophilus + RV NA (1) Cell-to-cell signaling and interaction, cellular assembly and organization, cellular function, and maintenance 75 27 (2) Cardiovascular system development and function, cell morphology, cell-to-cell signaling, and interaction 18 9 Compared to other cellular functions, cell-cell contact, and cell movements were highly activated or inhibited, respectively. The activated function of cell-cell contact was due to AFDN, ERK1/2, JAM3, and MPDZ genes while inhibited cell movement function was due to AFDN, CLDN19, CLDN5, LLGL1, MAGI2, PARD3, and PATJ genes (Fig. 9).  Table 2 The fold change expressions of selected tight junction (TJ) genes in signaling pathways The fold change expressions were calculated using HT-29 cells basal level genes expressions, where a cutoff value of ± 1.5-fold change was considered to tabulate

The Migration of Cells Is Inhibited upon RV Infection
Six TJ biological functions like formation of tight junctions, cell-cell contact, cell-cell adhesion, formation of intercellular junctions, morphology of tight junctions, and migration of cells were modulated on RV infection of HT-29 cells, wherein most of the functions were activated except migration of cells (Fig. 10). The inhibited function of migration of cells was associated with AFDN, AMOTL1, CLDN19, Cr3, LLGL1, and MAGI1 genes (Fig. 10).

Fig. 8
Ingenuity pathway analysis (IPA) predicted TJ cellular functions that are consistently activated and inhibited following E. coli Nissle + RV treatment. The networks of differentially expressed genes were algorithmically generated based on their connectivity such that the highly interconnected networks likely represent significant biological function. The Fischer's exact test was used to calculate a p value for each biological function assigned to a particular network. The name and number of molecules involved are described in the table

Innate Functions and Pathways Not Affected by RV Were Influenced in the Presence of E. coli Nissle Treatment
Since only E. coli Nissle showed significantly higher RV binding characteristics, innate response PCR array was analyzed for E. coli Nissle and control groups but not for LA group (Figs. 11 and S4, and Table S2). One function of innate genes not affected by RV, i.e., differentiation of blood cells, was influenced on E. coli Nissle or E. coli Nissle + RV treatments (Fig. 12A). Similarly, LPS stimulated MAPK, B cell receptor, PI3K in B lymphocytes, PI3K/AKT, role of NFAT, Nf-kB, LXR/RXR, Ga12/13 signaling were only influenced in the presence of E. coli Nissle treatments suggesting E. coli Nissle activates robust innate immune responses (Fig. 12B). For upstream regulators, RV influenced all the regulators but with lesser intensity compared with E. coli Nissle or E. coli Nissle + RV treatments (Fig. 12C). The important innate gene clusters that were differentially regulated on RV and E. coli Nissle alone or E. coli Nissle + RV treatments are indicated in Table 3.

Fig. 9
Ingenuity pathway analysis (IPA) predicted TJ cellular functions that are consistently activated and inhibited following LA + RV treatment. The networks of differentially expressed genes were algorithmically generated based on their connectivity such that the highly interconnected networks likely represent significant biological function. The Fischer's exact test was used to calculate a p value for each biological function assigned to a particular network. The name and number of molecules involved are described in the table 1 3

E. coli Nissle Activates Innate Immunity and Inflammatory Response via TNF Production
The top 5 biological functions that were affected on E. coli Nissle treatment alone included: production of cytokines, production of proteins, innate immune responses, function of dendritic cells, and function of phagocytes (Fig. S5). All the functions were activated, and the genes involved in activating these functions in turn activated the robust production of TNF. On E. coli Nissle + RV treatment, six biological functions that were affected included, inflammatory response, cytokine and chemokine mediated signaling pathway, cell movement of granulocytes, immune response of cells, activation of phagocytes, and activation of macrophages (Fig. 13). Interestingly, activation of phagocytes and inflammatory responses were highly activated. The genes involved in the activation of inflammatory response include CXCL8, Cpla2, Eotaxin, Ifn, Ifn gamma, LBP, pro-inflammatory, and Tnf (family) genes. The detailed list of genes involved and their status in respective biological pathways can be found in Fig. 13.

NK Cells and Inflammatory Response Are Strongly Activated on RV Infection
Eight biological functions were affected on RV infection alone, i.e., activation of cells, inflammatory response, innate immune responses, production of cytokines, function of antigen The networks of differentially expressed genes were algorithmically generated based on their connectivity such that the highly intercon-nected networks likely represent significant biological function. The Fischer's exact test was used to calculate a p value for each biological function assigned to a particular network. The name and number of molecules involved are described in the table presenting cells, activation of natural killer (NK) cells, replication of virus, and antiviral response (Fig. 14). Of these, activation of NK cells and inflammatory responses were strongly activated, and the genes involved in both these functions are shown in Fig. 14.

Discussion
Understanding interactions among pathogens, probiotics and host epithelial cells are of utmost importance to maintain enteric health. In this study we utilized polarized HT-29 cells as an in vitro model to investigate the anti-RV properties of the selected Gram-positive and Gramnegative probiotics and their underlying mechanisms. The Gram-positive and Gram-negative probiotics were tested for their ability to inhibit RV replication using established three-way treatment strategies mainly pre-inoculation, coinoculation, and pre-incubation/co-inoculation. The E. coli Nissle and L. acidophilus in a pre-inoculation strategy showed significantly higher ability to prevent RV replication in HT-29 cells. In agreement with our findings, recent studies also highlighted the RV inhibitory effects of probiotics like Lactobacillus and Bifidobacterium species using in vitro cell culture system [20,48,49], wherein up to 50% reduction in plaque forming units (PFUs) was observed. Yet, in another study, six different Bifidobacterium Fig. 12 An overall summary of innate response genes' IPA predicted A cellular functions, B their corresponding pathways that are in turn regulated by C upstream regulators. The heatmap shows the predicted activation (blue color) or inhibition (brown) of several selected cellular functions for each of the treatment groups. The intensity of color corresponds to expression status. The functions, pathways and upstream regulators are organized in descending order of -log(p) values that represents the extent to which the gene set for a particular group overlaps with the given cellular function/pathways. The red color box highlights the unique signatures at each level Table 3 The fold change expressions of selected innate genes in signaling pathways The fold change expressions were calculated using HT-29 cells basal level genes expressions, where a cutoff value of ± 1.5-fold change was considered to tabulate 3.14 strains were tested for anti-RV activity using both HT-29 and MA-104 cell lines and showed up to 50% reduction in PFU, and further higher RV reduction was evident in HT-29 cells compared to MA-104 cells, more specifically in a pre-inoculation strategy [50]. Like previous studies, our findings confirmed the ability of the Lactobacillus probiotics (L. acidophilus and L. rhamnosus GG) to reduce RV replication and showed that the pre-inoculation strategy was more efficient in reducing RV load in the HT-29 cells.
Although the exact reasons for the lack of effects of Bb12 treatment is unknown, factors like, probiotic strain and/or species specific effects, RV strain used, dose and duration of probiotic treatment, etc., can undoubtedly influence the experimental outcome. Pre-inoculation of HT-29 cells with probiotics was more efficient in inhibiting RV in our present study and from others' studies [48,49,[51][52][53][54]. This strategy has numerous advantages including the following: (i) limiting the cell adsorption and internalization of RV due to the direct trapping of the virus by the probiotic bacteria [13,53,56], (ii) "cross-talk" with the host epithelial cells in establishing antiviral protection [44,52,53,55], (iii) enhancing gut mucosal barrier thereby reducing permeability [14,54], (iv) production of metabolites with direct antiviral properties [52,55], and (v) interfering with intracellular RV replication by regulating NSP4 and Ca2+ [49,51]. Interestingly, anti-RV effects were not observed when dead probiotic cells were used (Fig. 4A). However, RV replication inhibition was still evident when cell culture supernatants were used (Fig.  5B) in the pre-inoculation regimen, suggesting that live probiotics cells and their secreted products are necessary to provide the RV inhibition effects. In this regards, recent studies have demonstrated that E. coli Nissle supernatant contain both soluble proteins and outer membrane vesicles that positively modulate the epithelial barrier through upregulation and redistribution of TJ proteins mainly, ZO-1, ZO-2, and CLDN14 [56,57]. Similarly, in our study, both E. coli Nissle and L. acidophilus pretreatment upregulated the expression of CLDN 14 ( Table 2). The probiotic's ability to bind RV particles was also shown to contribute to the anti-RV properties. In the present and in our previous study [13], none of the tested probiotics other than E. coli Nissle Fig. 13 Ingenuity pathway analysis (IPA) predicted innate response cellular functions that are consistently activated and inhibited following E. coli Nissle + RV treatment. The networks of differentially expressed genes were algorithmically generated based on their connectivity such that the highly interconnected networks likely rep-resent significant biological function. The Fischer's exact test was used to calculate a p value for each biological function assigned to a particular network. The name and number of molecules involved are described in the table possessed any significant RV binding properties (Fig. 4B). This further suggests that the direct binding of RV particles by E. coli Nissle is an important mechanism by which the reduction in RV replication is achieved following E. coli Nissle treatment. Additionally, E. coli Nissle decreased cell movements and increased cell-to-cell contact and TJ formation (Figs. 8 and S2), the critical cellular functions that are important for RV proliferation in the intestinal cells.
Previous study has shown that in polarized MDCK cells RV infection activates RhoA/ROCK/MLC signaling, which alters TJ protein distribution and disrupts TJ integrity thereby facilitating RV access to coreceptors and entry into the cells [58]. Though no specific activation of MLC signaling was observed, we found that RVinduced inhibition of morphology of TJ and formation of TJ/intercellular junctions (Fig. 9). Inhibition of these TJ formation events help RV to 14 Ingenuity pathway analysis (IPA) predicted innate response cellular functions that are consistently activated and inhibited following RV infection. The networks of differentially expressed genes were algorithmically generated based on their connectivity such that the highly interconnected networks likely represent significant biological function. The Fischer's exact test was used to calculate a p value for each biological function assigned to a particular network. The name and number of molecules involved are described in the table readily access the TJ and adherence junction's proteins that acts as receptors or coreceptors for RV infection [59,60]. Specifically, JAM2, occludin, and ZO-1 of TJ proteins play important receptor or coreceptor roles during RV entry into the cells [59]. Our study reports that upregulation of JAM2 gene by RV infection, which to a certain extent was repressed on probiotic treatments (Table 2). Another gene CLDN14 critical for TJ formation is highly expressed after E. coli Nissle and LA treatments (Table 2) suggesting roles for these probiotics in improving TJ formation [61]. Previously, E. coli Nissle was shown to induce protein kinase C-ζ (PKCζ) and extracellular-signal-regulated kinase 1/2 (ERK1/2) phosphorylation mediated events to upregulate the TJ protein claudin-14 [61]. Further, we found that transcription factor 2 (CDX2), which is critical in early intestinal differentiation and has been implicated as a master regulator of intestinal homeostasis and permeability is only regulated and inhibited by probiotic+RV infection (Fig. 7) suggesting the tested probiotics likely negate the RV-induced intestinal inflammation and perturbed homeostasis [62].
As predicted, E. coli Nissle modulated the expression of key genes involved in the innate immune and inflammatory responses. TLR genes like TLR1, TLR5, and NOD1 were upregulated on E. coli Nissle treatment ( Fig. 9 and Table 3). It is likely that E. coli Nissle may antagonize the RV induced inhibition of NF-κB pathway, as it was previously reported that RV employs several strategies to inhibit immune responses in cells, specifically the prevention of the nuclear accumulation of NF-κB [63]. Another protein, encoded by the RELA gene, is also uniquely upregulated by E. coli Nissle treatment (Table 3). This protein forms a dimer with NF-κB in the nucleus and acts as the transcription factor for many of the genes that are regulated by NF-κB [64]. The upregulation of RELA by E. coli Nissle treatment supports the idea that E. coli Nissle can boost the transcription of genes regulated by NF-κB, which are inhibited by RV. Interestingly, TNF was highly upregulated by E. coli Nissle treatment. Besides the implications of higher levels of TNF for the inflammatory response, TNF is known to recruit proteins that interact with RIPK1, causing it to promote cell survival via the activation of the NF-κB pathway [65]. Other mechanisms that E. coli Nissle exploits to antagonize RV infection include, the enhancement of antiviral sensing in HT-29. The protein encoded by the CARD6 gene (also uniquely upregulated by E. coli Nissle; Table 3 and Fig. 9) has been shown to play a key role in recognition of intracellular viral dsRNA [66]. Since RV is a dsRNA virus, and if E. coli Nissle treated cells have an enhanced ability to detect the presence of RV, then host cells may have an enhanced ability to mount a virus specific attack and/or signal other cells about the presence of RV.

Conclusion
Taken together, our in vitro study, which represents a simplified model for probiotic-RV-epithelial cell interactions, demonstrated that the probiotic E. coli Nissle acts via multiple mechanisms: (i) RV binding, (ii) up-regulation of critical TJ genes via specific upstream regulator (e.g., CDX2) to maintain gut barrier homeostasis, and (iii) activation of innate immune and inflammatory responses to neutralize RV infection. Our in vitro findings further suggest that E. coli Nissle supplementation, especially prior to enteric infections in infants, can confer more benefits in reducing enteric infectious diseases.