Events associated with susceptibility to invasive Salmonella enterica serovar Typhi in BALB/c mice previously infected with Plasmodium berghei ANKA

Numerous mechanisms have been proposed to explain why patients with malaria are more susceptible to bloodstream invasions by Salmonella spp., however there are still several unknown critical factors regarding the pathogenesis of coinfection. From a coinfection model, in which an S. enterica serovar Typhi (S_Typhi) was chosen to challenge mice that had been infected 24 h earlier with Plasmodium berghei ANKA (P.b_ANKA), we evaluated the influence of malaria on cytokine levels, the functional activity of femoral bone marrow-derived macrophages and neutrophils, and intestinal permeability. The cytokine profile over eight days of coinfection showed exacerbation in the cytokines MCP-1, IFNγ and TNFα in relation to the increase seen in animals with malaria. The cytokine profile was associated with a considerably reduced neutrophil and macrophage count and a prominent dysfunction, especially in ex vivo neutrophils in coinfected mice, though without bacterial modulation that could influence the invasion capacity of ex vivo S_Typhi obtained from liver macerate in non-phagocyte cells. Finally, irregularities in the integrity of intestinal tissue evidenced ruptures in the enterocyte layer, a presence of mononuclear leukocytes in the enterocyte layer, an increase of goblet cells in the enterocyte layer and a high volume of leukocyte infiltrate in the sub-mucosa were greatly increased in coinfected animals. Increases of mononuclear leukocytes in the enterocyte layer and volume of leukocyte infiltrate in the sub-mucosa were also seen in monoinfected animals with P. berghei ANKA. Our findings suggest malaria causes a disarrangement of intestinal homeostasis, exacerbation of proinflammatory cytokines and dysfunction in neutrophils that render the host susceptible to bacteremia by Salmonella spp.

Numerous mechanisms have been proposed to explain why patients with malaria are more susceptible to bloodstream invasions by Salmonella spp., however there are still several unknown critical factors regarding the pathogenesis of coinfection. From a coinfection model, in which an S. enterica serovar Typhi (S_Typhi) was chosen to challenge mice that had been infected 24 h earlier with Plasmodium berghei ANKA (P.b_ANKA), we evaluated the influence of malaria on cytokine levels, the functional activity of femoral bone marrow-derived macrophages and neutrophils, and intestinal permeability. The cytokine profile over eight days of coinfection showed exacerbation in the cytokines MCP-1, IFNγ and TNFα in relation to the increase seen in animals with malaria. The cytokine profile was associated with a considerably reduced neutrophil and macrophage count and a prominent dysfunction, especially in ex vivo neutrophils in coinfected mice, though without bacterial modulation that could influence the invasion capacity of ex vivo S_Typhi obtained from liver macerate in non-phagocyte cells. Finally, irregularities in the integrity of intestinal tissue evidenced ruptures in the enterocyte layer, a presence of mononuclear leukocytes in the enterocyte layer, an increase of goblet cells in the enterocyte layer and a high volume of leukocyte infiltrate in the sub-mucosa were greatly increased in coinfected animals. Increases of mononuclear leukocytes in the enterocyte layer and volume of leukocyte infiltrate in the sub-mucosa were also seen in monoinfected animals with P. berghei ANKA. Our findings suggest malaria causes a disarrangement of intestinal homeostasis, exacerbation of proinflammatory cytokines and dysfunction in neutrophils that render the host susceptible to bacteremia by Salmonella spp.
Bacteremia caused by Salmonella enterica (typhoid or non-typhoid) in malaria patients is a major public health concern in many Africa countries [1][2][3][4][5][6][7][8] . To date, numerous mechanisms have been proposed to explain susceptibility to malaria-induced Salmonella 6,[9][10][11][12][13][14] , and hemolysis caused by Plasmodium sp. causes the impairment of a variety of host defense mechanisms. Pioneering studies have already associated macrophage and neutrophil dysfunctions as having an important role in increasing susceptibility to Salmonella bacteremia, since these cells are one

Results
Susceptibility of mice with malaria to Salmonella enterica sv Typhi infection. Most Salmonella enterica serovares (sv) can cause intestinal infections. For the selection of the bacterial strain to be used in the coinfection model, the three following strains were used: Salmonella enterica serovar Typhi (S_Typhi), Salmonella enterica serovar Chloreaesius (S_Chloreaesius) and Salmonella enterica serovar Salamae (S_Salamae) ( Figure S1). Only the S. enterica serovar Typhi managed to evolve in the invasion, and increased the number of colonies established in 48 h. Based on the hypothesis that malaria infection would act as a susceptibility factor to Salmonella invasion, the experimental design of coinfection was tested in order to define the minimum bacterial concentration in which bacteremia occurred only in mice previously infected with malaria, and not in healthy mice. The Plasmodium-Salmonella coinfection model was established by comparing three different doses of S_Typhi (1 × 10 5 , 1 × 10 4 and 1 × 10 3 CFU). For comparison, another three groups of S_Typhi monoinfected mice received only 1 × 10 5 , 1 × 10 4 1 × 10 3 CFU, respectively. The bacterial culture was positive in all coinfection plates, on all sampled days (D2, D4 and D8), while the group that was monoinfected with S. enterica sv Typhi infected with 1 × 10 3 CFU did not show bacterial growth in any of the macerated organs on any of the analyzed days. The presence of colony-forming units in the macerate of the liver, spleen and intestine was only found in the respective Plasmodium-Salmonella-coinfection group ( Fig. 1A-C). The results allowed us to determine that the inoculum of 10 3 CFU of S. enterica sv Typhi was only able to develop bacteremia in animals previously infected with malaria. The parasitemia of P. berghei ANKA in Plasmodium-Salmonella-coinfected mice on days D2, D4 and D8 did not differ from those of malaria-monoinfected mice ( Figure S2).

Figure 1.
Plasmodium berghei ANKA and Salmonella enterica serovar Typhi coinfection model. In the malaria infection model developed, three groups of animals received an intraperitoneal dose of 10 6 erythrocytes infected with P. berghei ANKA on D0, and challenged with S_Typhi on D1. Group 1 received 10 5 S_Typhi CFUs on D1 (Pb + S_Typhi 10 5 ). Group 2 received 10 4 S_Typhi CFUs on D1 (Pb + S_Typhi 10 4 ), and Group 3 received 10 3 S_Typhi CFUs (Pb + S_ Typhi 10 3 ). Three other groups were challenged with only S_Typhi on D1, Group 4 (S_Typhi 10 5 ), Group 5 (S_Typhi 10 4 ) and Group 6 (S_Typhi 10 3 ). The animals were euthanized for the quantification of CFUs of the liver, spleen and intestine in SS-Agar at 37 °C. This experiment was replicated once. www.nature.com/scientificreports/ Plasma cytokine profile during malaria and Salmonella enterica sv Typhi coinfection in relation to malaria-and Salmonella-monoinfections. Since production of inflammatory cytokines and chemokines are hallmarks of pathology in bacteremia for Salmonella or other invasive bacteria in malaria infection, the inflammation response was evaluated using a cytometric bead array (CBA Mouse Inflammation Kit; BD-Biosciences/USA). MCP-1, TNFα and IFNγ were the most secreted followed by IL-6, IL-10 and IL-12 ( Fig. 2). Both Pb_ANKA-infected BALB/c mice and coinfected mice showed exacerbation in the cytokines MCP-1, TNFα and IFNγ, which may contribute to the susceptibility of the organism to Salmonella invasion. On the other hand, no differences were found with IL-6 and IL-10 between groups, while a slight increase in IL-12 was observed with animals with S_Typhi monoinfection. Nonetheless, these data presented inconclusive results.
Evaluation of phagocytic activity in medullary subpopulations. The study of susceptibility to bacterial coinfected in malaria patients has focused on dysfunction of phagocytic cells macrophages and neutrophils. Our study evaluated the phagocytic activity of ex vivo, PMA-activated, adherent, medullary subpopulation cells against a suspension of 10 3 CFU of formalin-inactivated S_Typhi (see the self-explanatory diagram of the ex vivo phagocytic assay in Figure S3). The total number of adherent cells showed no differences in all groups Fig. 3). Coinfected mice showed a very low percentage of (neutrophil and macrophage) phagocytes cells based on their characteristics under Romanowsky's staining method when compared to other groups. The percentage of neutrophils with at least one phagocyted bacteria were more greatly affected in coinfected mice in relation to other groups, including in Pb_ANKA-infected BALB/c mice which was reduced in relation to the control mice.
The percentage of monocytes with at least one phagocyted bacteria was also affected in coinfected mice, but it was possible though to note that ex vivo neutrophils showed a prominent phagocytic dysfunction.
Evaluation of the presence of bacterial modulation. In malaria monoinfection, an intraperitoneal dose with 10 6 erythrocytes infected with P. berghei ANKA on D0 is able to cause BALB/c mice to die around www.nature.com/scientificreports/ day 20 due to high levels of parasitemia (> 60-70%). Coinfection did not lead to a worse outcome because parasitemia was similar to that of P. berghei ANKA-monoinfected mice, when the animals were euthanized around twenty days (data not shown). It was also investigated whether the strain of S_Typhi recovered from the livers of coinfected mice underwent some modulation that could influence the invasion capacity in non-phagocyte cells, whereas bacterial pathogens tightly regulate the expression of virulence genes in response to very specific conditions. The HeLa cell invasion assay was used to compare the S_Typhi strain used in pre-inoculum conditions with the ex vivo strain obtained from liver macerate (see the self-explanatory diagram of invasion assay in Figure S4). Both behaved similarly in relation to the amount of invaded cells (Fig. 4A), however, the pre-inoculum invaded more HeLa cells after six hours, which indicates that the post-invasion bacterium did not undergo any modulation that made it more invasive (Fig. 4B).
Histopathological analysis by photomicroscopy. The above results showed that the invasion by S_ Typhi was mainly due to the presence of malaria rather than the invasive capacity of the bacterium. To assess whether malaria infection affects intestinal tissue integrity or irregularities on mucosa sublayers are a result of coinfection, histopathological sections were submitted to morphometric analysis. The volume and intestinal wall did not differ between the groups following the quantification by Cavalieri technique, nor did the submucosal, muscular and serous layers show differences among the groups (data not shown). Morphometric measurements of intestinal volume in the cross sections of histopathological sections increased by magnification 100. The intestinal tissue integrity and irregularities on mucosa sublayers were evaluated by images of sections using a stereomicroscope. Blind analyses were evaluated in longitudinal extensions at magnification 400, where counting systems containing points were superimposed on the images in the program Imod version 4.7/stereology module 22 ( Figure S5). Four types of changes were evident ( Pb_ANKA monoinfected animals showed a slight increase in ruptures, while the S_Typhi-monoinfected mice (ST) showed very few ruptures, which indicates that these ruptures may be the gateway for the bacterium (Fig. 6A).
The intestinal mucosa showed the presence of mononuclear leukocytes among the enterocytes (column II, Fig. 5). In this case, the three groups presented more than the control group on D4, the mononuclear cells infiltrate was much higher on D8 in the coinfected ones in relation to S_Typhi and control groups, however it did no differ of P. berghei ANKA infected mice (Fig. 6B). Regarding the presence of mucoid or goblet cells in the epithelial layer (column III, Fig. 5), only the co-infected animals showed an increase of these cells (Fig. 6C). The volume of leukocyte infiltrate in the mucosa showed the main differences (column IV, Fig. 5) and the three groups differed from the control on the 2 days, while the volume in coinfected animals was higher than both groups which were monoinfected with P. berghei ANKA and S_Typhi (Fig. 6D). Interestingly, the volume of leukocyte infiltrate in S_Typhi-infected mice recovered normal volume on D8 that did not differ of control group. Evaluation of phagocytic activity of ex vivo medullar neutrophils and macrophages. After 24 h of ex vivo incubation on glass slides of cells recovered from the femur of anesthetized mice from coinfected, monoinfected groups with P. berghei ANKA or S_Typhi and control group. After washing, the adherent cells were incubated with Phorbol 12-myristate 13-acetate (PMA) for activation in the phagocytosis assay. A suspension of S_Typhi killed by formalization was incubated with the adherent cells. Finally, the slides were washed, fixed in picric acid and stained using the Romanowsky method. Neutrophils and monocytes were visualized through staining characteristics and the total number of phagocytic bacteria was calculated. The phagocytic activity in cells between groups was compared using the percentage of phagocytes between adherent cells; % of neutrophils and % of macrophage containing phagocytic bacteria, respectively. The phagocytic activity in cells was compared by multiple comparison by Tukey's method. The data are shown in a scattered plot with mean and standard deviations. Asterisks: level of significance of *p < 0.05, **p < 0.005 and ***p < 0.0005. www.nature.com/scientificreports/ In relation to enterocyte basal layer, the volume was inconclusive (Fig. 6E). With this, it is possible to suggest that malaria causes a disruption of intestinal homeostasis and makes it susceptible to Salmonella invasion.

Discussion
Due to the overlap in terms of geographical distribution, children with recent or acute malaria have a higher risk of bacterial infection and death. This knowledge comes from clinical studies according to which the effects of malaria on the immune response against Salmonella bacteremia 1,2,7 . The exact nature of this association is still unknown, due to heterogeneity of data and lack of adequate controls corresponding to severity (reviewed by 6 ). Thus, little is known about whether malaria affects host resistance to intestinal colonization with typhoidal or non-typhoidal Salmonella enterica after bacterial ingestion. Herein, we managed to develop a model of coinfection by Salmonella enterica serovar Typhi and P. berghei ANKA in BALB/c mice, in which animals were already susceptible to Salmonella bacteremia after a day of malarial infection. Our model showed that a level of 1% parasitemia and 24 h of intraperitoneal infection was able to make the animals susceptible to invasive salmonellosis. These parasitemias are commonly observed in human malaria, especially in African children. Therefore, the leukocyte hemozoin pigment is a stronger predictor of malaria disease than parasitemia itself, which shows that concomitant infection can mask the diagnosis of salmonellosis infection when antimalarial treatment has still not improved the child's symptoms 9,23 . Although malaria is an acute systemic infection, it is known that gastrointestinal symptoms, such as nausea, vomiting, abdominal pain and diarrhea, are common during the acute phase of malaria. Nutrient malabsorption has been reported in severe falciparum malaria, suggesting that the environment in the intestinal lumen could be altered 18,24 . In a study on malaria falciparum, damage in the intestinal epithelium was evidenced due to permeation of compounds, such as monosaccharides and disaccharides, that are normally relatively impermeable in a healthy intestinal mucosa but in this case they were observed in the urine of these patients 18 . In this context, we performed a thorough analysis of intestinal tissue in the histopathological sections in malaria infections on D8 in order to evaluate changes in intestinal volume and tissue. The volume of the lumen and intestinal wall were not altered between the groups probably because on the eighth day the changes are probably still focal. The intestinal epithelium provides a protective barrier against enteric pathogens and upon disruption increases susceptibility to infections or impairs immunity 25 . One of the main findings observed here was that monoinfection by P. berghei ANKA malaria was able to alter the integrity of the epithelial layer of the intestine. These findings are in agreement with altered intestinal epithelium permeability in both Plasmodium falciparum infected patients and murine malaria model 10,18,25 . The appearance of ruptures in the enterocyte layer seems a detachment of the intestinal epithelia demonstrated by Taniguchi and cols 10 . Here, ruptures in the enterocyte layer were exacerbated in coinfected animals suggesting that ruptures in the epithelial layer caused by malaria predispose the host to bacteremia 23,26 .
Changes in the architecture of the intestinal tissue included the increase in the number of goblet cells and in the cell density, which in the latter is characterized by volume of the leukocyte infiltrate in the sub-mucosa. Pre-inoculum aliquot was obtained after growth of nine hours in LB broth and stored at − 80 °C for more than 15 days until the moment of the assay. The ex vivo aliquot was obtained from a macerated mouse liver coinfected on D4 and frozen at − 80 °C. The two ex vivo and pre-inoculum bacteria were thawed and set for invasion assay in HeLa cells. Pre-inoculum bacteria were diluted in a series of 10-10 and sown in SS-Agar medium to define the inoculum equivalence of ex vivo bacteria. The invasion time was 1, 3 and 6 h. After these periods, medium supplemented with gentamicin was added to the culture medium for 1 h in order to eliminate external bacteria. Slides were then washed, stained using the Romanowsky method and bacteria visualized. The data are shown in a box plot with mean and standard deviations. Asterisks: level of significance of *p < 0.05, **p < 0.005 and ***p < 0.0005. www.nature.com/scientificreports/ The goblet cells are responsible for the layer of mucus that covers the epithelium to facilitate the removal of food content and are also part of the first line of defense against damage caused by pathogenic bacteria and bacterial products 27,28 . Here, coinfected mice showed several of these mucoid cells in the enterocyte layer, which indicates that the innate response to the infection bacteria was active in these animals. However, the disarrangement of intestinal homeostasis caused by malaria may compromise the fight against the Salmonella infection, despite abundance of the goblet cells. In relation to cell density, leukocyte infiltrate was greatly increased in coinfected mice. Several coinfection models showed that these infiltrates are composed of inflammatory monocytes and neutrophils 6,9,21,25 . Despite this increase, the phagocytic activity has been one of the most widely-accepted explanations for susceptibility to bacteremia in malaria 2,4,14,15 . Herein, phagocytic dysfunctions of macrophage and neutrophil progenitors in co-infected mice were greatly reduced, which indicates that these effector cells would have already been compromised in the primary lymphoid organs. In addition, hematopoiesis impairment may be associated with factors associated with bacteria, since MCP-1, TNFα and IFNγ levels and parasitemia in coinfected mice did not differ from those monoinfected by P. berghei ANKA.

Scientific
In relation to the bacteria, the survival and growth of salmonellae within host cells are important for bacterial virulence, and many genes that are required for invasion are not expressed when bacteria are grown on rich media 29 . Herein, we demonstrated that phenotypically the ex vivo S_Typhi recovered from the liver of coinfected animals did not become more invasive than pre-inoculum bacteria. One study showed that animals previously infected with P. yoelii exhibited intensified colonization of a noninvasive, commensal, Escherichia coli strain, which suggests that the disturbance of the microbial community by malaria opens up an ecological niche that can be harnessed by pathogenic bacteria 21 . However, we are cautious as to whether our data show that malaria infection makes mice susceptible to invasion of pathogenic bacteria, despite some modulation in vivo, or if, phenotypically, the S_Typhi grown in rich media would already be able to invade the host cells 9,14 .
One limitation of the study was to use a serovar Typhi from the Salmonella enterica line, since pediatric bacteremia caused by P. falciparum infection mainly involves non-typhoid invasive Salmonella. In addition, the models of coinfection or monoinfection in mice use S. enterica serovar Typhimurium because it spreads systemically in this host 9,12,14,21,30 . From three serovars of Salmonella enterica, we defined a dose with S_Typhi that managed to evolve in the invasion of BALB/c mice only in those previously infected with malaria. Other limitation was that in our model, the host was challenged with the bacterium the day after malaria infection, while children who develop malaria-Salmonella coinfection are probably already colonized by these pathogenic bacteria 15 . Although, that is assumption that Salmonella infection is subsequent to malaria, i.e. it is reasoned that in endemic areas individuals would be most likely to become infected with typhoidal or non-typhoidal Salmonella after contracting malaria 9 . The number of mice used here seemed another limitation, however they were sufficient to demonstrate the major differences. Despite these limitations, our model provided us with a new view on intestinal mucosal alteration that predisposed BALB/c mice to bacteremia due to malaria.
In conclusion, our findings provide descriptive insights into how mucosal responses to a bacterial pathogen are altered in a simultaneous malaria infection. Although the knowledge of malaria as a susceptibility factor to Salmonella bacteremia is evident, the pathophysiological mechanisms are still not well understood due to their complexity and lack of well-established experimental models. This study aimed to establish an experimental model and evaluate the pathophysiology with better coverage of intestinal histopathology, systemic immune factors and functional assays of phagocytic dysfunction. The inflammatory exacerbation evidenced by the increase of cytokines and chemokines with the dysfunction especially of neutrophils is also well known and was well characterized as an effect of an exacerbated inflammatory response. Finally, our main findings were the alteration in the epithelial integrity of the intestine caused by monoinfection with P. berghei, which was identified as a possible cause of susceptibility to Salmonella invasion. Knowledge on the influence on intestinal tissue and the interaction of effector immune mechanisms in Salmonella-malaria coinfection is absolutely essential for creating strategies to combat the bacterial infections in affected populations.

Material and methods
Strains. Currently, the Salmonella nomenclature system defines the genus as being composed of three species: S. subterranea, S. bongori and S. enterica, the latter having more than two thousand serotypes and, as such, it is classified as a subspecies 31 . The Salmonella enterica serovar Typhi (S_Typhi), Salmonella enterica serovar Chloreaesius (S_Chloreaesius) and Salmonella enterica serovar Salamae (S_Salamae) strains were provided by Animals. Female BALB/c mice aged between 6 and 8 weeks were used. In all experiments, the welfare of the animals was taken into consideration. The mice were housed in a standard polycarbonate cage with wood shavings bedding, with a maximum of 5 mice per cage. We also attempted to reduce the stress of individual housing (when necessary) by environmental enrichment with nestlets and small play tunnels. The animal room was kept at a controlled temperature and humidity, with a light and dark cycle of 12 h. Mice had ad libitum access to food and water.
Anesthesia and euthanasia. All efforts were made to prevent undue stress or pain to the mice. The mice were humanely euthanized once they showed the following clinical signs: lethargy; hypothermia and/or difficulty of breathing. The mice were euthanized with ketamine (300 mg/kg) (Vetbrands, Brazil) and xylazine Infection. Initially, one Salmonella enterica serovar was selected to be used in the coinfection model with the Plasmodium berghei ANKA. The strategy was to select the most invasive of the three bacteria (S_Typhi, S_Chloreaesius and S_Salamae) i.e., the one that presented the most colony forming units (CFU) in the liver, spleen and intestine at 24 h and 48 h post infection. The animals were divided into three groups containing four animals to be individually challenged with the three different serovares. The time necessary to reach maximum exponential growth that was ideal for infection was defined as nine hours of growth in LB broth, and the oral inoculation was performed by gavage technique with a concentration of 1 × 10 9 CFU in 100 µl in each individual. After euthanasia and removal of organs, they were homogenized, diluted and sown in serial dilution in Salmonella-Shigella agar culture medium (SS-Agar) and incubated at 37 °C. After 24 h, the colony forming units (CFU) were quantified.  www.nature.com/scientificreports/ received an intraperitoneal dose of cryopreserved aliquots containing 5 × 10 6 erythrocytes infected with P. berghei ANKA on day 0 (D0). One group received a dose of 1 × 10 5 CFU of S_Typhi, the second received 1 × 10 4 CFU and the third 1 × 10 3 CFU on D1, by gavage. For the comparison, three other groups containing four mice with monoinfection by S_Typhi, received, a dose of 1 × 10 5 CFU by gavage on D1. The second received 1 × 10 4 CFU and the third 1 × 10 3 CFU, by the same means. The experimental design of malaria-Salmonella coinfection was evaluated by the presence and/or absence of colony-forming units in the macerate of the liver, spleen and intestine obtained on D2, D4 and D8. After euthanasia, the organs were removed whole, homogenized, diluted and sown in serial dilution in Salmonella-Shigella agar culture medium (SS-Agar) and incubated at 37 °C. The group in which S_Typhi monoinfected mice did not show bacterial growth in any of the macerated organs on any of the analyzed days was chosen. Parasitemia of P. berghei ANKA groups was monitored by flow cytometry.

Establishment of
Plasmodium berghei ANKA and Salmonella enterica serovar Typhi coinfection model. Three groups containing four mice were classified as S_Typhi monoinfection, P. berghei ANKA monoinfection and Plasmodium-Salmonella-coinfection. The P. berghei ANKA monoinfection and Plasmodium-Salmonella-coinfection groups received an intraperitoneal dose of cryopreserved aliquots containing 5 × 10 6 erythrocytes infected with P. berghei ANKA on day 0 (D0). The controls and S_Typhi monoinfected mice received an equal volume of isotonic and neutral phosphate buffer. The determination of parasitemia was performed by flow cytometry. The controls and S_Typhi monoinfected mice received an equal volume of isotonic and neutral phosphate buffer. The animals were submitted to the Plasmodium-Salmonella coinfection and S_Typhi monoinfection mice received 100 µL S_Typhi 10 3 in 24 h LB broth via gavage one day after the malarial infection (D1). Mice of control and P. berghei ANKA monoinfection groups received an equal volume of isotonic and neutral phosphate buffer. Parasitemia of P. berghei ANKA monoinfection and Plasmodium-Salmonella-coinfection groups was performed by flow cytometry. Extraction from the intestine occurred on days four and eight (D4 and D8) after malarial infection. The animals were euthanized for the quantification of CFUs in the liver, spleen and intestine, and the homogenized ones were sown in SS-Agar at 37 °C for quantification of CFU.
Evaluation of functional activity. The phagocytic activity of neutrophils and macrophages was evaluated using cell culture after incubation, which was based on the literature and optimized for the methodology (See the diagram in Figure S3). Briefly, after induction of anesthesia and analgesia, the animals were sacrificed for the removal of medullary subpopulations from the femur on D8. The recovered cells were plated on glass slides (Knittel, Brazil) in DMEM 24-well plates at 37 °C, 5% CO 2 for panning of adherent cells. After 24 h of incubation, nonadherent cells were removed after three cycles of washing in DMEM medium. The adherent cells were incubated with Phorbol 12-myristate 13-acetate (PMA) for 1 h for activation, and then washed three times in DMEM medium. A suspension of 10 3 CFU of formalin inactivated S_Typhi that had been treated previously with 37% formalin for 1-h and washed 4 times with neutral phosphate buffer was applied for 3 h at 37 °C, 5% CO 2 . Finally, the slides were washed, fixed in picric acid and stained using the Romanowsky method. Neutrophils were differentiated by the "busy" aspect of nucleus showing several lobes, while cells characterized by horseshoe-shaped nucleus with dishwater-gray cytoplasm and a few tiny granules were characterized as monocyte-macrophage lines. The quantity of phagocytic activity in cells was compared by multiple comparison by Tukey's method.

Bacterial invasion assay in HeLa cells.
The possibility of ex vivo Salmonella strains having been modulated after infection and influencing phagocytic activity was analyzed. For this, the livers of mice that were coinfected on D4 were macerated and frozen at − 80 °C. For comparison, an aliquot of S_Typhi was used under the same conditions as the pre-inoculum (in growth phase of nine hours in LB broth) and stored at − 80 °C for 15 days until the moment of the assay ( Figure S4). HeLa cells were grown in 24-well plates containing glass slides (Knittel, Brazil) in DMEM medium with 10% fetal bovine (Sigma, Brazil) at 5% CO 2 at 37 °C until total confluence. The two ex vivo and pre-inoculum bacteria were thawed and set for the invasion assay in HeLa cells. Pre-inoculum bacteria were diluted in a series of 10-10 and sown in SS-Agar medium to define the inoculum equivalence of ex vivo bacteria. The invasion time was 1, 3 and 6 h. After these periods, medium supplemented with gentamicin was added to the culture medium for 1 h in order to eliminate external bacteria. Slides were then washed, stained using Romanowsky staining and bacteria were visualized.
Analysis of serum cytokines. Cytokines were measured on D2, D4 and D8 for each of the infection groups. The plasma from four mice was collected for the determination of cytokines using cytometric bead array for mouse inflammation factors (CBA Mouse Inflammation Kit; BD-Biosciences/USA).
Histopathological evaluation of the intestine. After euthanasia, the intestines of the animals were removed and kept in buffered formalin for 48 h at room temperature. The samples were processed in the Quantitative Morphology Laboratory (LaMiq/UFAM). For this, the intestines were dehydrated in increasing concentrations of 70 and 96% ethanol (2 h in each concentration), pre-infiltrated in 96% ethanol + hydroxyethyl methacrylate plastic historesin solution (Technovit 7100, Külzer-Heraues, Germany) overnight and infiltrated in 100% resin. The samples were arranged in individual Histobloc teflon molds (Külzer-Heraues, Germany) set in plastic resin + polymerizing solution. The molds were heated in an oven at 37 °C until complete polymerization. The analysis of epithelial integrity of the intestine and morphometric measurements were made using stereology of the Cavalieri principle of volume and the Delesse principle, both operated using the Imod program (see Supplemental Material 1). www.nature.com/scientificreports/ Statistical analysis. The prism 5.0 statistical program (GraphPad Software, Inc. USA) was used for the statistical and graphical analysis of this study. The independent variables (treatments) were tested for their normality by the Kolmogorov-Smirnov test. An ANOVA test using a multiple Tukey comparison was used for the following assays: (1) cytokines on days D4 and D8, (2) phagocytosis assays on D8, and (3) S_Typhi invasion of pre-inoculum and ex vivo in HeLa cells. In relation to histopathological analyses, on average 20 serial sections of the intestine were calculated by appropriate equations developed for stereological analyses 22,32,33 . An error coefficient of 5% and deviation of 15% were considered acceptable. The estimate of the volume was determined according to Cavalieri's principle, and the results were analyzed using one-way ANOVA (see Supplementary material).