Gut regulatory T cells mediate immunological tolerance in Salmonella-infected superspreader hosts by suppressing cytotoxic activity of T cells

Superspreader hosts carry out most pathogen transmission events and are often disease tolerant since they remain asymptomatic despite high pathogen burdens. Here we describe the superspreader immune state that allows for disease tolerance. In a model of Salmonella infection, superspreader mice develop colitis with robust CD4+ and CD8+ T-cell responses, however, they remain asymptomatic. We found that superspreaders have significantly more regulatory T cells (Tregs) in the distal gut compared to non-superspreader infected hosts. Surprisingly, the depletion of Tregs did not induce pathogen clearance but rather exacerbated weight loss, increased gut inflammation, and compromised epithelial intestinal barrier. This loss of tolerance correlated with dramatic increases in cytotoxic CD4+ and CD8+ T cells. Interestingly, CD4 neutralization in Tregs-depleted superspreaders was sufficient to rescue tolerance. Our results indicate that Tregs play a crucial role in maintaining immunologic tolerance in the guts of superspreader mice by suppressing cytotoxic CD4+ and CD8+ T-cell activities. AUTHOR SUMMARY Superspreader hosts are the main cause of disease transmission and a very important public health concern. Here, we evaluated the immunological tolerance of the Salmonella infected superspreaders in a mouse model. By manipulating Tregs, we demonstrated the immunological mechanism from the host to maintain health status and high pathogen burden. Tregs depletion in the superspreaders led to severe disease, with damage of the intestinal epithelia, and high morbidity without having any effect on shedding and systemic Salmonella burden. Furthermore, we demonstrated that the damage of the intestinal epithelia was related to cytotoxic activity of T cells. When Tregs were depleted, CD8+ T cells produced high levels of granzyme B and perforin. CD8+ T cells neutralization in Tregs depleted mice led to increased cytotoxic CD4+ T cells. Interestingly, neutralization of CD4+ T cells in the Tregs depleted mice led to a reduction in the CD8+ T cells producing granzyme B and it was sufficient to rescue host tolerance in this model. We demonstrate for the first time that cytotoxic CD4+ T cells damage the epithelial intestinal barrier and contribute to loss of tolerance in the context of a superspreader host. These findings open new perspectives to understand mechanisms of tolerance in the intestine of a superspreader host.


52
In contrast to resistant hosts, which control infection and minimize disease by reducing pathogen 53 load, immune tolerant hosts can sustain higher pathogen loads without losing health (1). 54 Importantly, immune tolerant hosts can become carriers and transmit disease. To exemplify the 55 impact of immune tolerant hosts, the Pareto 80/20 rule has been applied to demonstrate that 20% 56 of infected hosts can be responsible for 80% of diseases that may result in death (2). The 20% 57 responsible for transmitting disease are called superspreaders due to their capacity to spread 58 pathogens to other hosts and, for some diseases, remain asymptomatic. Thus, it is critical to 59 better understand superspreaders in order to address this important public health threat and 60 disease reservoir (3-5). 61 To understand the superspreader Salmonella host, previous work from our group demonstrated 62 that 129X1/SvJ mice can be chronically infected with Salmonella enterica serovar Typhimurium 63 (STm) (6) whereby 25-30% of the mice become superspreaders, shedding more than 10 8 bacteria 64 per gram of feces (7). Superspreader STm-infected mice are considered tolerant as they lack any 65 signs of disease (e.g., ruffled fur, weight loss, or decrease in temperature) and shed enough   88 To characterize the intestinal immune response to STm in the tolerant superspreader mice, we 89 infected 129X1/SvJ mice orally with 10 8 CFU of wild-type S. Typhimurium strain SL1344.

90
Approximately 25-30% of the mice became superspreaders around 21 days post-infection (p.i.), 91 shedding more than 10 8 CFU of pathogen per gram of feces ( Fig. 1a; Fig. S1a), similar to our 92 previously published reports (7, 9, 16). Superspreader mice are tolerant to the high load of STm 93 in the intestine as demonstrated by a lack of significant weight loss over the course of 28 days 94 p.i. (Fig. S1b). To characterize immune cell populations in the guts of infected mice, colonic 95 immune cells were extracted from non-superspreader mice (e.g., mice shedding 10 2 -10 7 CFU/g 96 feces) and superspreader mice ( Fig. 1b and Fig S1a) 28 days p.i. and cell populations were 97 analyzed by flow cytometry. As previously published by our group(7, 9), the superspreader mice was an increase in Th17 cells in superspreaders compared to non-superspreader mice (Fig. 1e). 105 We then measured the systemic levels of STm in the superspreaders compared to non-106 superspreaders (Fig. S1d), and there was no difference in the systemic levels of STm, which 107 reflects previous findings that Th17 cells impair bacterial translocation from the gut (17). Tregs 108 are well described to suppress the activity of T cells (Reviewed in (18)), to address if Tregs play 6 109 a role in the tolerance of superspreader STm infected hosts we measured the expression of the 110 transcription factor Foxp3 (expressed in Tregs) in the CD4 + T cells. Superspreader mice had a 111 significantly higher percentage of Tregs compared to non-superspreader mice (Fig. 1f). Indeed, 112 the levels of Tregs in non-superspreader mice were not significantly different from uninfected 113 control mice (Fig. 1f). Thus, Tregs levels positively correlated with CFUs in the feces of 114 superspreader mice (Fig. 1g), and the correlation was even bigger when correlating Tregs with 115 CFU from the mice that shed >10 7 CFU/g of feces (Fig. 1h), suggesting that Tregs are important 116 for maintaining tolerance in superspreader hosts.  As expected, (7, 16) two days after streptomycin administration, almost all mice shed >10 8 125 CFU/g of feces (Fig. 2a) without a significant difference in STm levels in the spleen (Fig. 2b).

126
The induced superspreader hosts did not exhibit significant weight loss or show signs of 127 morbidity after antibiotic treatment (Fig 2c). The antibiotic-induced superspreader hosts had 128 significant increases in the percentage and the total number of CD4 + T cells in the colon at day 129 21 p.i. (Fig. 2d), and by day 28 p.i. (Fig. 2h) the percentage of CD4 + T cells was similar to the 130 levels in non-induced superspreaders (Fig. 1f). The percentage of CD8 + T cells was not altered at 131 days 21 or 28 p.i. (Fig. 2e,i). Importantly, the mice had increased percentages of Th17 cells ( Fig.   7   132 2f,j) and Tregs (Fig. 2g,k) at days 21 and 28 p.i. when compared to the mice that did not received 133 the antibiotic treatment. These data indicate that these induced superspreader mice can be used as 134 a model to study tolerance mechanisms in the colon of superspreader hosts.

135
Tregs depletion in superspreader hosts leads to loss of tolerance 136 To test the role of Tregs in the tolerance phenotype of superspreader hosts, we depleted Tregs in    (Table S2 and Table S3). The transcriptional profiles in the colons of both STm- 196 infected and uninfected mice in F1-DEREG mice were distinct from the F1-DTR +/as calculated 197 by the log10 p-value and log2 fold change (Fig 5a). As expected, Foxp3 expression is 198 significantly higher in the F1-DTR +/infected or uninfected. However, there were no significant 199 changes in the expression of genes involved with wound healing response, such as the receptor 200 of Amphiregulin, EGF receptor (EGF-R) (28, 29) (Fig 5a, Fig S3f). In contrast, there was a  Methods. We confirmed that CD8 + T cells were depleted in the colons of infected mice by flow 225 cytometry (Fig 6a,b). Surprisingly, CD8 + T cell neutralization did not restore tolerance in F1-226 DEREG superspreader mice as weight loss was similar in the anti-CD8b cell antibody-and 227 isotype control antibody-treated mice (Fig. 6c). In addition, the absence of CD8 + T cells did not 228 alter the STm burden in the spleens and feces of superspreader F1-DEREG or F1-DTR +/mice 229 (Fig. 6d). Finally, the level of gut epithelial damage was assessed with the FITC-dextran 230 permeability assay in the CD8b-depleted mice and controls. The depletion of CD8 + T cells did 231 not reduce gut permeability in the superspreader F1-DEREG mice (Fig. 6e), suggesting that 232 other immune cell types contribute to loss of tolerance in these mice.

233
Therefore, we next investigated whether CD8 + T cell depletion would affect GZMB production 234 in Natural Killer cells (NK) and CD4 + T cells. The percentage of NK cells in the CD8b-depleted 235 mice did not increase (Fig. 6f), nor did the level of GZMB-producing NK cells increase when 236 compared to isotype-control treated (Fig. 6g). However, there was a very significant increase in 237 the percentage of CD4 + T cells in the colon of F1-DEREG mice treated with anti-CD8b when 238 compared to the isotype treated (Fig. 6h). Interestingly, these CD4 + T cells acquired a cytotoxic 239 phenotype as measured by the increased intracellular GZMB when CD8 + T cells were depleted 240 (Fig. 6i). Collectively, our results indicate that cytotoxic CD4+ T cells are playing a major role in 241 the loss of tolerance in Tregs-depleted superspreader hosts. 242 We next wanted to dissect the role of cytotoxic CD4+ T cells in loss of immune tolerance. Since 243 neither an antibody to neutralize GZMB in vivo nor GZMB-deficient 129X1/SvJ mice are 244 readily available, we neutralized both CD4 + and CD8 + T cells, or only CD4 + T cells in 245 superspreader F1-DEREG and F1-DTR +/mice using neutralizing antibodies. The depletion was 246 confirmed by flow cytometry (Fig 7e,g). Strikingly, F1-DEREG superspreader mice depleted of 12 247 CD4 + and CD8 + T cells did not lose significant weight or show signs of morbidity, although mice 248 that received the isotype control antibodies lost significant weight and became moribund (Fig.   249   7a). Surprisingly, STm-infected F1-DEREG mice depleted of only CD4 + T cells did not lose 250 weight or show signs of morbidity (Fig. 7a).

251
Neutralization of just CD4 + T cells, as well as both CD4+ and CD8+ T cells in STm-infected F1-252 DTR +/superspreader mice, did not result in significant weight loss (Fig. 7b). Additionally, there 253 was no change in the STm burden in the absence of CD4 + T cells or CD4 + and CD8 + T cells in 254 the feces (Fig. 7c) and only slight increases in the spleen (Fig.7d) of superspreader F1-DEREG 255 or F1-DTR +/mice. In addition, CD4 + T cell neutralization led to a decrease in the percentage of 256 CD8 + T cells as well as intracellular GZMB levels in the colon of STm-infected F1-DEREG 257 mice when compared to F1-DEREG mice (Fig. 7e,f). The neutralization of CD4 + T cell and 258 GZMB intracellular levels was confirmed by flow cytometry (Fig. 7g,h).

259
To gain more insight into how CD4 + T cells might contribute to the greater weight loss of Tregs- compared to isotype control mice, confirming that the neutralizing antibodies dampened the 268 cytotoxic response in the colon (Fig. S3a-f).

Importantly, the neutralization of both CD4 + and CD8 + T cells or only CD4 + T cells reduced gut
270 permeability in the superspreader F1-DEREG mice to levels that were similar to superspreader 271 F1-DTR +/mice (Fig. 7i), Taken together, our results indicate that Tregs-dependent suppression 272 of CD4 + T cells in the guts of STm superspreader mice is crucial for immunological tolerance 273 (Fig. 7j).     (Table S4). Methacarn and allowed to fix. After fixation, the linearized colon was rolled into a "Swiss roll".

431
Total RNA was extracted from colonic tissues using the RNAeasy Kit (Qiagen) following the 432 manufacturer's instructions and quantified by NanodropA total of 25 ng of RNA was used for 433 NanoString nCounter assay and the codeset for Mouse Fibrosis Panel (Nanostring) was utilized.

434
The hybridization, processing, and acquisition were performed at the Nanostring facility 435 (NanoString Technologies, Seattle, WA). The normalization and differential expression analysis 436 were conducted using NSolver 4.0 software (Nanostring).