Porcine intraepithelial lymphocytes undergo migration and produce an antiviral response following intestinal virus infection

The location of intraepithelial lymphocytes (IELs) between epithelial cells provide a first line of immune defense against enteric infection. It is assumed that IELs migrate only along the basement membrane or into the lateral intercellular space (LIS) between epithelial cells. Here, we identify a unique transepithelial migration of porcine IELs as they move to the free surface of the intestinal epithelia. The major causative agent of neonatal diarrhea in piglets, porcine epidemic diarrhea virus (PEDV), increases the number of IELs entering the LIS and free surface of the intestinal epithelia, driven by chemokine CCL2 secreted from virus-infected intestinal epithelial cells. Remarkably, only virus pre-activated IELs inhibits PEDV infection and their antiviral activity depends on the further activation by virus-infected cells. Although high levels of perforin is detected in the co-culture system, the antiviral function of activated IELs is mainly mediated by IFN-γ secretion inducing robust antiviral response in virus-infected cells. Our results uncover a unique migratory behavior of porcine IELs as well as their protective role in the defense against intestinal infection.

The authors present a manuscript documenting the migratory characteristics and anti-viral activity of intraepithelial lymphocytes (IEL) in pigs. IEL constitute a large proportion of the body's T cells but are comparatively understudied, despite displaying important anti-viral and anti-cancer potential. Moreover, IEL exhibit unconventional characteristics, such as innate-like reactivity and surveillance behaviour. While a growing body of work has documented the migratory behaviour and anti-viral activity of IEL in mice, how this may translate to humans is less clear. However, IEL have been implicated in coeliac disease and IBD, amongst other pathologies, suggesting that more information is needed regarding how these cells function in beneficial and pathological scenarios. Many IEL are γδ T cells, and pigs and ruminants show distinct γδ T cell populations from mice, warranting a closer look at IEL activity in these animals. This may shed some light into the basic questions about IEL that we still lack clear answers to, such as antigen specificity and molecular signals regulating their behaviour.
The authors begin by tracking localisation of IEL in intestinal samples from pigs as they mature. They record a previously unappreciated phenomenon, where apparently viable IEL could be found on the apical face of the epithelium i.e. in the intestinal lumen. They also see a steady increase in IEL numbers as pigs develop and are weaned, suggesting an influence from the microbiome. Moreover, following infection with porcine epidemic diarrhea virus (PEDV), an industrially-relevant pathogen, they see a dramatic and rapid increase in the number of T cells in the intraepithelial and luminal compartments. A ligated jejunal loop model is employed here to permit highly localised and acute analysis of this migratory response, which also occurs following bacterial inoculation. To further elucidate this transepithelial migration, the authors develop an in vitro system whereby IEL migration through an epithelial monolayer is observed when a spatially separated epithelial cell line is infected with PEDV. They then show that CCL2 but not CCL5 may induce transepithelial migration of IEL in this system, and that inhibition of CCL2 largely blocks their migration in response to PEDV infection.
The second half of the manuscript instead investigates the effector function of porcine IEL in in vitro PEDV infection. Contrary to mice, where anti-CD3 stimulation promoted anti-viral activity of IEL against norovirus, the authors saw no impact of unstimulated or anti-CD3 pre-treated porcine IEL upon PEDV replication in an in vitro co-culture assay. However, pre-treatment of IEL with whole inactivated (WI) PEDV induced a dramatic anti-viral activity in IEL upon subsequent coculture with a live PEDV-infected cell line. This anti-viral activity was contact-independent as shown by transwell assay. A similar effect was seen with IEL pre-activated by in vivo WI PEDV infection. Finally, the authors investigate the anti-viral activity in more detail. WI PEDV-preactivated IEL reduced overall PEDV infection of the cell line, but increased apoptosis of infected cells, alongside a large induction of perforin in the IEL. More mechanistically, they find a clear induction of IFN-γ in pre-activated IEL upon co-culture with infected cells, which correlated with the induction of interferon-stimulated genes in the infected cells and led to control of infection as demonstrated by inhibition of IFN-γ.
Overall, these findings are relevant to the IEL field as they are the first robust characterisation of porcine IEL behaviour at steady state and following viral infection. Given the similarities but also key differences between IEL in mouse and humans, better study of IEL in more closely related species will help address important questions relevant to human health and disease. The authors generally use the best techniques available, considering the obvious technical limitations of live imaging in pigs, and the necessity for in vitro assays in certain contexts. The observation of luminal T cells is intriguing, although probably needs to be better characterised in the future to ensure it is not an artefact of intestinal harvesting, which is prone to autolysis. Moreover, the strict requirement for acute pre-exposure to inactivated PEDV (either in vitro or in vivo) to activate IEL, which then only perform anti-viral activity in the presence of live PEDV-infected cells, suggests some interesting biology which may have implications for IEL responses in many systems. The authors should discuss these implications, namely that IEL appear to require microbe pre-exposure (over an extremely short timeframe) in order to induce a capacity to respond to what appears to be host-derived soluble signals following infection with a live virus. Moreover, there should be discussion of how the PEDV may activate the IEL, especially if anti-CD3 TCR stimulation did not have an equivalent effect.
Major concerns 1. My predominant concern with the manuscript is in the authors' conclusions, namely that the observed migratory behaviour of IEL is contributing to their anti-viral function. This is an extremely tenuous link, purely based upon the observation that a small percentage of IEL enter the lumen following PEDV infection, where it is possible that they directly sense PEDV, which may then link to the requirement for PEDV pre-exposure for the anti-viral assays. However, the authors repeat this often, and even in the title of the manuscript. While it is likely that migratory behaviour of IEL contributes to anti-viral surveillance (as reported by Hoytema van Konijnenburg et al Cell 2017, amongst others), this is not supported by the authors' experiments. The authors in effect have a manuscript describing IEL migration in viral infection, and separately, anti-viral function by IEL. They need to tone down these conclusions and amend the title.
2. My other concern regards the interpretation of the WI PEDV pre-activation. Apparently, this occurs over the course of 1hr culture of IEL with WI PEDV, prior to the IEL being transferred into contact with a live PEDV-infected cell line. However, IEL that were not first cultured with WI PEDV displayed no anti-viral effect upon co-culture with infected cells, despite the fact that PEDV was obviously present in these co-cultures (given plaque assays using culture S/N e.g. Fig 5B). I wonder if the authors can explain how 1hr contact with WI PEDV, but not 24hr contact with PEDV in the S/N of infected cells, resulted in IEL activation?
3. Throughout the paper, the IEL are treated as bulk, except in Figure 7A where γδ T cells are suddenly stained for no clear reason. It would be helpful if the authors could investigate, at least, whether αβ vs. γδ IEL show similar anti-viral behaviour, and requirement for WI PEDV preactivation. It seems that amongst porcine IEL, γδ T cells express the most NKG2D, which might be relevant to the perforin/cytolysis data in Figure 7 (Altmeyer et al 2017 Vet Immunol Immunopathol). In addition, it might add some useful data to ask whether the perforin/cytolysis results occur in the transwell system, where NKG2D-ligand interactions between IEL and infected cells would be prevented.
Minor concerns 4. The CCL2 experiments in Figure 4 are independently valid. However, the authors should reconcile the fact that WI virus does not induce CCL2 in their cell line with the observations of IEL migration changes following in vivo WI virus administration in Figure 5. Perhaps explaining this, Fig 4B suggests that WI PEDV introduction into the intestine causes CCL2 upregulation, but this is not confirmed by the figure legend. Was this in fact live or WI virus? 5. Perhaps the IEL do not migrate to CCL5 because they simply do not express CCR5? This could be easily investigated by qPCR if porcine antibodies are not available. 6. Figures 1 and 2 show summary data without error bars or statistics on the plots, despite pvalues being defined in the figure legends and in the text. These should be shown on the graphs ( Fig 1F, Fig 2B, Fig 2E). Moreover, the number of animals used in these experiments should be included in all figure legends. Figures 1  and 2 is helpful as absolute number, but frequency of total T cells in each area (%) should also be presented. This will provide evidence for the authors' claims that the frequency of luminal or intraepithelial-located T cells changes upon infection/ageing. Currently, the absolute counts merely suggest an overall influx of T cells, without a specific relocalisation. Additionally, basal T cells should be quantified in Figure 2 8. The multiple comparisons tests used, for example, in Figure 4F and Figure 5D, F G and H) are a little unclear. Are they truly comparing all groups, or are some groups combined before comparison (i.e. the brackets linking two columns). If they simply summarise equivalent p-values for the comparison between one group and two other groups, the individual comparisons should be displayed. Figure 7A is not convincing. The gating looks arbitrary. γδ TCR vs αβ TCR staining should be shown on the same plot in order to convincingly discriminate between the two populations. 10. Figure  The current study provides evidence of transepithelial migration of IELs in porcine and how this novel migratory behavior could protect against oral infection by PEDV. Finally, the study concludes that protection against PEDV is mediated by IFN gamma and the cytotoxic nature of activated IELs. Though these findings have some novelty, they are somewhat preliminary and are not backed by solid evidence.

The flow cytometric staining of γδ T cells in
Specific comments -1. Though authors found IELs on the luminal side, there is no direct evidence that these IELs leaked from the damage site of the epithelium layer. 6. In panel 5F, why are there no plaques seen in the mock of direct co-culture? 7. In Fig 7, it was concluded that granzymes and perforin mediated antiviral activity. However, this is just a correlation as Gzms and perforins were only measured by ELISA, and no direct evidence is provided that these molecules are mediating the death of epithelial cells. Moreover, apoptosis of epithelial cells can also be caused by IFN gamma. 8. In conclusion, the authors found a transepithelial behavior of porcine IELs, but its implications are not very clear from the current data presented in the manuscript. Therefore, the paper's title seems to be an overstatement in concluding that transepithelial migration leads to the antiviral state. The authors should discuss these implications, namely that IEL appear to require microbe pre-exposure (over an extremely short timeframe) in order to induce a capacity to respond to what appears to be host-derived soluble signals following infection with a live virus. Moreover, there should be discussion of how the PEDV may activate the IEL, especially if anti-CD3 TCR stimulation did not have an equivalent effect.
A: Thank you very much for your insightful comments and thorough analysis of our manuscript. We acknowledge and welcome your suggestions, which will be incorporated to improve the quality of our manuscript. We will respond to your specific questions one by one, perform the relevant experiments and correct a series of errors found in the manuscript. All the changes made by us are labeled with red font in the revised manuscript.
Q1. My predominant concern with the manuscript is in the authors' conclusions, namely that the observed migratory behaviour of IEL is contributing to their anti-viral function. This is an extremely tenuous link, purely based upon the observation that a small percentage of IEL enter the lumen following PEDV infection, where it is possible that they directly sense PEDV, which may then link to the requirement for PEDV pre-exposure for the anti-viral assays. However, the authors repeat this often, and even in the title of the manuscript. While it is likely that migratory behaviour of Q2. My other concern regards the interpretation of the WI PEDV pre-activation.
Apparently, this occurs over the course of 1hr culture of IEL with WI PEDV, prior to the IEL being transferred into contact with a live PEDV-infected cell line. However, IEL that were not first cultured with WI PEDV displayed no anti-viral effect upon coculture with infected cells, despite the fact that PEDV was obviously present in these co-cultures (given plaque assays using culture S/N e.g. Fig 5B). I wonder if the authors can explain how 1hr contact with WI PEDV, but not 24hr contact with PEDV in the S/N of infected cells, resulted in IEL activation? A2: Your careful review is much appreciated. We apologize for the missing description for IELs pre-activation. In fact, incubation with inactivated PEDV alone cannot induce IELs' antiviral potential (Fig. R1). It was only when we added a centrifugation step (IELs and WI PEDV were centrifuged at 1200×g for 15 min at 25°C) before incubation that the IELs became effectively pre-activated. With regards to why the centrifugal process is critical for the pre-activation of IELs by viral antigens, previous studies have shown that cultured lymphocytes in suspension have a low chance of contacting with virions during virus inoculation, while centrifugation could dramatically enhance the binding of the viral particles to lymphocytes (~40 fold) 1, 2, 3 . Based on this observation, we speculated that centrifugation may facilitate the pre-activation of IELs in the incubation phase by increasing the binding of WI PEDV to IELs. However, further studies are needed to confirm this speculation further. Thank you again for your careful review and thought-provoking questions. At present, we are unable to separate the αβ + or γδ + IEL from porcine intestinal IELs for the following reasons. 1) There are currently no commercial αβ + IEL antibodies available for flow cytometry staining; 2)Although previous study have considered γδ -T cells as αβ + IEL, the γδ + IELs still cannot be effectively sorted out since the grouping of γδ T positive cells was not obvious after anti-Pig γδ T lymphocytes antibody staining. (Fig. R2). Based on your kindly suggestions. The LDH release and perforin secretion in the two co-culture models (contact or noncontact) were further determined. As shown in Fig.R3, the pre-activated IELs from the two culture models both exhibited highly perforin secretion and similar cytotoxic responses. Regarding the mechanism underlying this observed phenomena, we can speculate that although the NKG2D-ligand interactions have been proven to be relevant for activating the cytolytic responses of γδ T cells, many other activating receptors (without the necessity of cell-to-cell contact) were also involved in triggering the cytotoxic activities of IELs 5, 6 . For example, gut-resident IELs also expressed high levels of cytotoxicity in activating receptor NKp46, which bound the ligands with various viral and bacterial proteins. In addition, the expression of OX40 in CD8 + T IEL was also associated with T-IEL-mediated cytotoxicity. As the ligand of OX40, OX40L was mostly expressed on the T and B lymphocytes, rather than on the intestinal epithelial cells 9 .   Table R1. Unexpectedly, the CCR5 gene was also expressed in the porcine intestinal IELs, although its transcription level was lower than the level of CCL2 in IELs (Fig. R4). However, there remains some uncertainty as to whether the CCR5 protein could really be expressed and distributed in the surface of porcine IELs due to a lack of appropriate commercial antibodies.
This issue still requires further exploration in future studies. Q6. Figures 1 and 2 show summary data without error bars or statistics on the plots, despite p-values being defined in the figure legends and in the text. These should be shown on the graphs (Fig 1F, Fig 2B, Fig 2E). Moreover, the number of animals used in these experiments should be included in all figure legends. A6: Thank you for pointing this out. We apologize for the missing information. We have added the error bars accordingly in the revised figures (including Fig 1F, Fig 2B, Fig   2E) and different letters to indicate significant differences. Moreover, all figure legends have also described the number of animals used. Thank you again for your careful review.

Q7. The quantification of T cell localisation to basal, intraepithelial and luminal areas
in Figures 1 and 2 is helpful as absolute number, but frequency of total T cells in each area (%) should also be presented. This will provide evidence for the authors' claims that the frequency of luminal or intraepithelial-located T cells changes upon infection/ageing. Currently, the absolute counts merely suggest an overall influx of T cells, without a specific relocalisation. Additionally, basal T cells should be quantified in Figure 2.
A7. Thank you for your kindly reminder. We have added the frequency of total T cells (%) in each area of Fig. 1F, Fig. 2B, and 2E. Moreover, the number of basal T cells in Fig. 2B and 2E was also be quantified.
Q8. The multiple comparisons tests used, for example, in Figure 4F and Figure Fig.3d, Fig.3e, Fig.4c, Fig.4f, Fig.5c,   Fig.5d, Fig.5f, Fig.5g, Fig.5h, Fig.6e, Fig.6f, Fig.7e Fig.7f, Fig.8d, Fig.8e, Fig.8f,   Fig.8g, Fig.8h, Fig.8i and Fig.S4). Thank you again for your careful review. Figure 7A is not convincing. The gating looks arbitrary. γδ TCR vs αβ TCR staining should be shown on the same plot in order to convincingly discriminate between the two populations. A9: We welcome and acknowledge your suggestion. The PE-labeled rat anti-Pig γδ T lymphocytes antibody (Cat. No. 561486; BD Biosciences, USA) is the only commercially available antibody for pig γδ T cells. When this antibody is used for staining, the population of γδ T positive cells cannot be clearly detected, although the positive cells were markedly shifted to the right. We have provided the FMO-isotype control to show our gating strategy in the revised Fig.3A (Fig. R5). it was not possible to determine the αβ TCR positive cells in porcine IELs in our present study.

Reviewer 2:
The current study provides evidence of transepithelial migration of IELs in porcine and how this novel migratory behavior could protect against oral infection by PEDV. Finally, the study concludes that protection against PEDV is mediated by IFN gamma and the cytotoxic nature of activated IELs. Though these findings have some novelty, they are somewhat preliminary and are not backed by solid evidence.
Thank you for your careful review and comments. We acknowledge and welcome your suggestions, which will be incorporated to improve the quality of our manuscript. We apologize for the imperfect study design and over-interpreting research results. At present, we have responded to your specific questions individually. As per your suggestions, we have further performed the relevant supplementary experiments and rewritten the relevant content in the "Discussion" and "Conclusion" sections. All the changes we made are labeled with red font in the revised manuscript. Q1. Though authors found IELs on the luminal side, there is no direct evidence that these IELs leaked from the damage site of the epithelium layer.
A1. Thank you for your careful review. I apologize for the misunderstanding caused by our inadequate interpretation. Indeed, our present study has found that the transepithelial migration of porcine IELs occurs even under physiological conditions, which exhibit age-and location-dependent characteristics. Furthermore, the integrity of the intestinal epithelial barrier (IELs resident) was evaluated by detecting tight junction protein Claudin-3. Immunohistochemistry staining showed that the intercellular movement of the IELs appeared to have no influence on the expression of the tight junctions in intestinal epithelium (Fig. R6). Therefore, we hypothesized that the porcine IELs primarily moved between the adjacent epithelial cells through the transient opening of the tight junctions, similar to what has been reported for transepithelial dendrites formation of intestinal dendritic cells. 13 . However, further studies are needed to elucidate the detailed mechanism involved. Thank you again for your kindly suggestions. A2: Thank you for your careful revision. I apologize for our incorrect description of the IPEC-J2-IEL co-culture system in the "Materials and methods" section. Actually, the IPEC-J2 were seeded onto the back side of the Transwell insert membrane. Thereafter, the isolated IELs were seeded on the basolateral membrane of the IPEC-J2 and cocultured for the indicated time. A schematic of the IPEC-J2-IEL co-culture system is shown in Fig. 3b and Fig. 4d, in which the upper face of IPEC-J2 was facing down and toward the lower compartment, and the IELs were co-cultured on the basolateral side of the IPEC-J2 cells. We have corrected our description of the culture system in the "Materials and methods" section of the revised manuscript.
Q3. Further, it was shown that upon PEDV infection, CCL2 was mainly present on the apical surface, which acts as a chemokine for the transepithelial migration of IELs.
Nevertheless, in the IPEC-J2-IEL system, IELs are already on the apical side and then moving towards the basal side. How do the authors corroborate these two observations?
A3. Sorry again for our incorrect description of the IPEC-J2-IEL co-culture system in the "Materials and methods" section. In our co-culture system, the IELs were cultured on the basolateral side of the polarized cultured IPEC-J2 cells and the moved towards the apical side, which was consistent with the secretion pattern of CCL2. The In the porcine small intestine, CD8α + γδ T cells and CD8α + αβ T cells comprised a substantial fraction of the IELs 14 . Among the CD8α + αβ T subsets, the cells expressed by CD8αβ + heterodimer (CD8αβ + αβ T cells) were considered to be induced IELs 15 .
These cells will likely include antigen-experienced effector or memory cells, which reside within the intestinal epithelial layer following prior infection, and recognized viral peptides presented by infected epithelial cells via their TCRs 16,17,18 . Unlike induced IELs, the natural IELs (including CD8αα + αβ T and CD8αα + γδ T cells) display unusual MHC characteristics, including the ability to directly recognize nonpeptide antigens in a manner similar to antibodies. 19,20 .
Considering that PEDV negative pigs (serum and fecal samples were negative for both PEDV antibody and nucleic acid) were used in our study, and the viral preactivation requirement of IELs, as well as the main function of CD8αα + αβ IELs are widely accepted as immune regulation 21 , we speculate that the main antiviral function of the IELs were mainly undertaken by the CD8α + γδ T cells. Comprising 50-60% of the IEL compartment in the porcine small intestine, the γδ T cells have several innate cell-like characters that allow for their early and rapid activation following recognition of cellular infection 22 . To accomplish these functions, the γδ T cells use both the T cell receptor (TCR) and additional activating receptors (notably NKG2D, and TLR) to respond to stress-induced ligands and infection. Among these, the TLRs play critical roles in the antiviral activities of γδ T cells 23,24 . The expression of TLRs in resting γδ T cells is usually weak or undetectable but can be quickly activated and upregulated by viral infection. Based on this observation, we further tested the expression of TLR in IELs, and the IEL processing method is the same as that presented in Fig.8A. The primers used for RT-qPCR are presented in Table R1.
Among the TLRs involved in viral infection, the expression of TLR2, TLR3 and TLR7 in pre-activated IELs were significantly increased after co-culturing with viral infected epithelial cells (Fig.R7, a and b). To further explore the specific TLR molecules involved in the virus recognition, the pre-activated IELs were treated with TLR inhibitors including C29 (TLR2) 25 , CU CPT 4a (TLR3) 26 and E6446 (TLR7) 27 , which were purchased from TargetMol (USA). As shown in Fig. R7. c to e, all these inhibitor treatments had significant suppressive effects on the antiviral activity of the activated IELs compared to the control. Moreover, the inhibitors used did not affect intestinal IEL viability at their working concentration (Fig.R7, f and g). Therefore, we speculate that TLR2, TLR3 and TLR7 appear to be the key players responsible for the virus recognition by the IELs, although these results were somewhat preliminary and require further validation.
Previous studies have shown that virus could directly activate γδ T cells by binding the viral receptors on the surface of the γδ T cells, which enables them to acquire a pre-activated phenotype that allows for the rapid induction of effector functions following the detection of cellular infections 28,29 . The expression of receptor molecules associated with PEDV binding 30,31 in porcine intestinal IELs has also been confirmed by us, including APN aminopeptidase N (APN), EGFR( epidermal growth factor receptor) and transferrin receptor 1 (TfR1) (data not shown). Prior to preactivation of WI PEDV, the porcine IELs were pre-treated with the inhibitors for these receptor molecules, including Bestin (Selleck, S1591) for APN 32 , Ferristatin II (Sigma-Aldrich, C1144) for TfR1 33 , AG1478 (NA) (Selleck, S2728) for EGFR 34 (Fig.   R8, a to c). At used concentrations, these inhibitors did intestinal IEL viability (Fig.   R8, d and e). Among them, the APN specific inhibitor Bestin significantly inhibited the antiviral function of porcine IELs in a dose-dependent manner (Fig. R8a).
Therefore, the PEDV binding receptor may play an important role in mediating IEL pre-activation and the molecular mechanisms involved merit further exploration.
Resembling to intestinal IELs, the antiviral activity of MLN T cells (isolated from PBS treated pig) is also dependent upon viral pre-activation. Previous studies have shown that the γδT cells were also present in the MLN, although they only account for a small proportion of the total T cell subsets 35 . Therefore, we speculate that the precise mechanisms of virus recognition by MLN T cells may be similar to that used by the IELs. However, the MLN T cells were actually used as a positive control in the IELs pre-activation experiment (in vivo), which represented the effector T cell activated by antigen presentation. Therefore, the mechanism underlying MLN T cells activation by PEDV will be further explored in our future work.    Q7. In Fig 7, it was concluded that granzymes and perforin mediated antiviral activity.
However, this is just a correlation as Gzms and perforins were only measured by ELISA, and no direct evidence is provided that these molecules are mediating the death of epithelial cells. Moreover, apoptosis of epithelial cells can also be caused by IFN gamma.
A7: Thank you for your insightful review. Although previous studies have shown that IELs exert their cytotoxic functions by secreting granzymes and perforin, the granzymes and perforin-mediated antiviral activity of IELs were unable to be proven by our current data. We do apologize for this over-interpreted conclusion.
To confirm the involvement of granzyme and perforin in the antiviral effects of IELs, we further conducted the following experiments. 1) The antiviral and cytotoxic activity of the IELs were determined after the granzyme and perforin inhibition treatment; 2) After treatment with IFN-γ blocking antibodies, the specific cytotoxicity of IELs against the epithelial cells was evaluated.
The antiviral activity of IEL (Fig. R10a) and the IEL-mediated epithelial cells apoptosis (Fig. R10b) were not influenced by treating with the perforin inhibitor concanamycin A (CMA) or granzyme inhibitor 3,4 Dichloroisocoumarin (DCI) 36,37 . It has also been confirmed that at the concentrations used, the two inhibitors will not affect the viability of the IELs (Fig. R10d). However, the IELs-induced apoptosis of epithelial cells was significantly inhibited by the IFN-γ antibody (25 ug) treatment (Fig. R10c).
The abovementioned results are consistent with the results in Fig.7H, where the antiviral effect of IELs could be almost entirely antagonized by IFN-γ antibody (Fig.7H).
Therefore, although high levels of perforin were detected in the co-culture system, we believed that the antiviral function of the activated IELs was mainly mediated by IFN-γ secretion that induced robust antiviral response in the virus-infected cells. We have added these supplementary results in the "Results" section of the revised manuscript. The relevant text in both the "Discussion" and "Conclusion" sections were also corrected. Thank you again for your careful review. Q8. In conclusion, the authors found a transepithelial behavior of porcine IELs, but its implications are not very clear from the current data presented in the manuscript.
Therefore, the paper's title seems to be an overstatement in concluding that transepithelial migration leads to the antiviral state.
A8. Thank you very much for your careful review and kind reminder. We apologize for over-interpreting our research conclusion. We acknowledge that the direct relationship between the observed migratory behavior of IELs as well as their antiviral activity remains speculative based on the data we currently have and requires further validation.
We have toned down these related conclusions and amended the title in the revised manuscript.