View all hidden authors and organizations

Edited by Jorge Galán, Department of Microbial Pathogenesis, Yale University, New Haven, Connecticut; received July 28, 2021; accepted November 11, 2021

Some infections are extremely difficult to cure, even with adequate antibacterial chemotherapy. How pathogens survive exposure to antimicrobial agents in tissues is still poorly understood. Using three-dimensional whole-organ tomography, we showed that Salmonella colonization in mouse spleens was uneven. The low Salmonella density in the white pulp only triggers limited local infiltration of inflammatory cells, which is essential to support antibacterial Salmonella clearance. Inflammatory cell density decreased further during treatment to cope with the decrease in Salmonella load, leading to insufficient support for removal and eradication failure. However, continuous inflammation during antibacterial chemotherapy can be effectively eliminated. Our results identified heterogeneous host-pathogen interactions in separated tissues as the main mechanism for the persistence of Salmonella antibiotics.

Even without antimicrobial resistance, antimicrobial chemotherapy cannot eradicate pathogens. Persistent pathogens then cause disease recurrence. In vitro studies have shown various mechanisms of antibiotic persistence, but due to the difficulty of studying the survivors of rare pathogens in complex host tissues, their in vivo relevance remains unclear. Here, we used high-resolution whole-organ tomography to locate and characterize the rare surviving Salmonella in the mouse spleen. Chemotherapy eliminated >99.5% of Salmonella, but it was not effective against the small Salmonella subgroup in the white pulp. Previous models cannot explain these findings: drug exposure is sufficient, Salmonella continues to replicate, and host pressure induces only limited Salmonella drug tolerance. In contrast, antibacterial clearance requires the support of neutrophils and monocytes that kill Salmonella, and the density of such cells in the white pulp is lower than other spleen compartments with higher Salmonella burden. Neutrophil density decreased further during treatment with the reduction of Salmonella load, resulting in insufficient support for the removal of Salmonella from the white pulp and the failure of eradication. However, adjuvant therapies that maintain inflammatory support can be effectively eliminated. These results identified uneven Salmonella tissue colonization and spatiotemporal inflammatory dynamics as the main reasons for the persistence of Salmonella, and established a powerful method to study rare but influential pathogen subgroups in complex host environments.

In difficult-to-treat bacterial infections, including tuberculosis, deep Staphylococcus aureus infection, and invasive salmonellosis, adequate antibacterial chemotherapy can initially eliminate most pathogenic cells and resolve clinical symptoms. However, even in the absence of relevant antimicrobial resistance, even prolonged treatment often fails to eradicate the pathogen. The persistence of a small number of pathogenic cells will cause disease recurrence and accelerate the emergence of antimicrobial resistance (1⇓ ⇓ –4). The eradication of such difficult-to-treat infections represents an urgent medical need.

The development of effective treatments requires a detailed understanding of its underlying mechanisms. Various mechanisms have been proposed to explain the persistence of pathogens during antimicrobial chemotherapy. First, the antimicrobial agent may not reach all bacterial cells in sufficient amounts, because the anatomical permeability barrier restricts the drug from entering certain tissue areas of the bacteria (5, 6). Second, bacteria may adopt physiological states in host tissues, making them resistant to antibiotic exposure. This state of tolerance may be triggered by the host immune system's pressure on the pathogen (7⇓ ⇓ ⇓ ⇓ ⇓ –13). In addition, in some cases, due to low ATP levels, limited nutrient supply and stress conditions can slow down bacterial proliferation, thereby increasing resistance to most antibiotics (14⇓ ⇓ ⇓ –18) (19). Third, pathogen heterogeneity is considered to be the main reason for treatment failure (1, 2, 20⇓ –22). Some bacteria may stop replicating due to random internal processes or in response to external triggers. This non-replicative "persistence" can survive exposure to other lethal antibiotic concentrations. Other forms of heterogeneity may also lead to treatment failure, including asymmetric cell division (23⇓ –25), uneven distribution of efflux pumps between daughter cells (26), and heterogeneous expression of prodrug activating enzymes (27) , Transient gene amplification (28)), and heterogeneous induction of specific stress responses (29). The relevance of the various mechanisms by which these pathogens persist in host tissues is still unclear, because the supporting data are almost entirely obtained using in vitro models (1⇓ –3, 20⇓ –22, 30). In vivo data is very important, because bacterial susceptibility depends on environmental factors, and the complex and diverse microenvironment in infected tissues is difficult to simulate in vitro (31).

A small animal infection model suitable for in vivo studies is systemic salmonellosis in mice, which serves as a model for invasive salmonella infection in humans. Such infections, including typhoid and paratyphoid (enteric fever) and non-typhoid Salmonella (NTS) bacteremia, are major health problems worldwide (32, 33). Antibiotic chemotherapy often fails to eradicate Salmonella and causes disease recurrence, even if the bacterial strain is sensitive to the drug (34⇓ ⇓ –37). The mouse model of invasive salmonellosis reproduced these eradication challenges (38⇓ ⇓ –41). We have previously shown that clinically relevant doses of fluoroquinolone antibiotics can only slowly eliminate Salmonella in mouse tissues, because Salmonella replicates slowly in the body, with a generation time of about 6.5 hours (42). However, the removal of colony forming units (CFU) continued to exponentially decay in a single phase for at least 5 days. Here, we continued the treatment for a longer time interval and observed a decrease in clearance rate and eradication failure at a later point in time. We aim to unravel the underlying mechanism of the persistence of Salmonella, an antibiotic.

Because it is difficult to locate and characterize a few, sparsely distributed micron-sized pathogen cells in the entire centimeter-sized host organ, the in vivo study of pathogen cells that survived chemotherapy has been hindered. We addressed these limitations by using serial two-photon tomography (STP) (43) to detect individual Salmonella cells in the entire spleen of infected mice. We solved the problem of interfering with tissue autofluorescence, developed an automated pipeline for identifying Salmonella cells with TB-level imaging data, and verified the accurate detection of a single Salmonella cell with a density as low as 100 mm3 in tissues. We used STP to locate Salmonella survival treatments in two clinically relevant antibiotic classes. We combine STP with laser capture microdissection, flow cytometry, Salmonella reporter gene constructs, and adjuvant therapies to determine non-replicating Salmonella persistence, stress-triggered drug tolerance, uneven drug delivery, and neutrality Correlation of tissue infiltration of granulocytes and inflammatory monocytes. We found that Salmonella colonization in the spleens of untreated mice was uneven. A small number of Salmonella subgroups colonize the white pulp (WP) and only induce local infiltration of inflammatory monocytes and neutrophils. These Salmonella-killing host cells support the elimination of Salmonella during chemotherapy, but their density decreases during chemotherapy as the local Salmonella load decreases, resulting in a higher survival rate of Salmonella in WP than in other initially colonized and inflamed splenic compartments. Times. In contrast, there is no correlation between Salmonella dormancy, stress-induced antimicrobial resistance, or inefficient antibiotic delivery. Therefore, the host tissue structure and the topological structure and dynamics of host-Salmonella interaction lead to locally different antibacterial activities, and ultimately lead to eradication failure.

We infected genetically susceptible BALB/c mice by oral administration of Salmonella typhimurium serotype. Once clinical symptoms appeared, we started chemotherapy with the recommended dose of the fluoroquinolone antibiotic enrofloxacin. This treatment prevented the death of the mice, resolved the clinical symptoms, and reduced the Salmonella load in the spleen of the main target organ by approximately 200 times, and exhibited a slow single-phase exponential kinetics during 6 days (42) (Figure 1A and B). However, the clearance rate slowed down thereafter, even if the Salmonella viable cells remained susceptible in vitro (minimum inhibitory concentration, MIC, 0.03 to 0.06 mg/L). After 10 days of treatment, Salmonella was found in the spleen (~2,000 CFU), liver (~2,000), mesenteric lymph nodes (~1,000), and Pyle's nodules (~200 of the last small intestinal plaque) (SI appendix, Figure S1A) . Surviving Salmonella pools in these and other organs have previously been observed in intravenously infected mice (38) and streptomycin-pretreated enteritis models (44).

Whole spleen monitoring of Salmonella survival during antibiotic chemotherapy. (A) Disease score of mice infected by mouth. From day 5 to day 14, mice received 5 mg/kg of enrofloxacin per day. The arithmetic mean and SD of 3 to 26 mice are shown. (B) Salmonella load in the spleen of oral infected mice as determined by electroplating (triangle), flow cytometry (diamond) or continuous two-photon tomography (circle). Vaccination and flow cytometry results are shown as geometric mean and standard deviation for groups of 3 to 20 mice (yellow labels indicate previously reported values) (42). Each tomographic data point represents a single mouse. The dashed line from day 2 to day 5 shows the exponential fit; the line from day 5 to day 15 shows the exponential decay of the y offset; the line from day 15 to day 18 connects the geometric mean. (C) The principle of STP tomography. The vibrating knife removes the surface of the tissue block. Before the vibrating microtome removes the next part, the two-photon microscope scans the block surface. (D) Side view and top view of collagen fibers (trabeculae) detected by STP tomography of a 5 mm spleen section. (E) Detection of GFP-Salmonella by STP tomography in an optical section (left). Neutrophils are stained by injecting Ly-6G antibody labeled with phycoerythrin in the body. The corresponding vibrating section was stained with Salmonella LPS antibody and imaged with a confocal microscope (right image) to confirm the green fluorescent particles. (Scale bar, 3 μm.) (F) Analyze the GFP fluorescence intensity distribution of a single Salmonella cell in the spleen on the 5th day after oral infection by flow cytometry (n = 7,685) or STP tomography (n = 681). The intensities are normalized to their respective median values. (G) An example of a single Salmonella cell and microcolonies in the spleen detected by STP tomography. The number on the right represents the estimated number of Salmonella cells based on total GFP fluorescence. (Scale bar, 5 μm.) (H) Size distribution of Salmonella objects in oral infected mice during (day 5) and antibacterial chemotherapy. The data represents aggregated data from three mice per group and 39 to 362 Salmonella subjects per mouse. The inset shows the proportion of a single Salmonella among all Salmonella subjects, and each circle represents a mouse.

Salmonella survivors caused disease recurrence after stopping treatment in 7 of the 8 tested mice (Figures 1A and B and SI appendix, Figure S1B). The second regimen of enrofloxacin once again reduced clinical symptoms to baseline (SI Appendix, Figure S1B), indicating that Salmonella is still sensitive to treatment in the body. At the end of the second protocol, we still found surviving Salmonella in the spleen, liver, and mesenteric lymph nodes, but not in the last Pyle's nodule (SI appendix, Figure S1A). Using enrofloxacin at a dose 60 to 120 times higher than the recommended dose (to avoid adverse reactions), it will still take 5 weeks to eradicate Salmonella (39, 45, 46).

Similarly, after 10 days of treatment with β-lactam ceftriaxone, 480 ± 190 Salmonella cells survived in the spleen (SI Appendix, Figure S1C). Recurrence of the disease after treatment interruption occurred in all four test mice, and the second regimen of ceftriaxone can be used for treatment (SI Appendix, Figure S1B). Therefore, the two major classes of antibiotics widely used to treat invasive salmonellosis in humans have failed to eradicate Salmonella that infect mice orally.

Our goal is to determine the survival mechanism and host microenvironment of persistent Salmonella. The currently available method (47⇓ ⇓ ⇓ -51) is impractical for characterizing the micron-sized Salmonella cells scattered in the centimeter-sized host organ. However, STP tomography provides a robust method for similar tasks—detecting fine neuron axons with submicron resolution in the entire mouse brain (43) (Figure 1C). Therefore, we are interested in applying STP to infectious diseases. Our initial attempts showed that the standard STP protocol is hampered by the autofluorescence of a large number of tissues in the spleen, which becomes particularly strong during infection (SI appendix, Figure S2A) and masks the fluorescent signal of Salmonella expressing GFP. We were unable to increase the GFP level because it would reduce the in vivo adaptability of Salmonella (52), but we found that storing the perfused fixed spleen in a cryoprotectant at -20 °C can effectively inhibit the autofluorescence of interfering tissues (SI Appendix , Figure S2A)). As another challenge, the illumination was uneven and varied between samples (SI Appendix, Figure S2B), and it was not possible to subtract a single reference background image as was done for the brain data set. The application of "Corrected Intensity Distribution Using Regularized Energy Minimization" (CIDRE) (53) solves this problem (SI appendix, Figure S2B).

Imaging of a 5 mm spleen section with a horizontal resolution of 0.435 μm and a vertical resolution of 10 μm (Figure 1D) shows that Salmonella-like particles have the spectral characteristics of GFP (Figure 1E). A single three-dimensional (3D) image stack of 5 mm spleen slices contains approximately 0.5 TB of data, and manual analysis is prohibited. Simple threshold-based Salmonella detection is hindered by residual autofluorescence and imaging artifacts. To solve this problem, we used machine learning, including support vector machines and deep convolutional neural networks, which distinguish GFP-Salmonella from the background with >99% sensitivity and specificity. We verified the identification of GFP-Salmonella by antibody staining of Salmonella lipopolysaccharide in spleen slices extracted from the tomography scanner (Figure 1E), and it was not found in the spleen of mice infected with yellow fluorescent protein (YFP)-Salmonella. The detection of a signal classified as GFP-Salmonella indicates specific detection. The second harmonic signal of collagen (54) and in vivo staining of host cell surface antigens with fluorescently labeled antibodies revealed the tissue environment of Salmonella cells (Figure 1D and E). In this study, we focused on the infected spleen, but STP also found GFP-Salmonella in the liver, mesenteric lymph nodes, and small intestinal Peyer masses (SI appendix, Figure S3 and movie S1-S3).

A single Salmonella cell has a narrow GFP fluorescence intensity distribution, which is consistent with the flow cytometry data of the same GFP-Salmonella strain in the spleen (55) (Figure 1F), indicating that the fluorescence of STP is quantitatively reliable. For unresolved Salmonella microcolonies, we divided the total fluorescence by the median single-cell GFP value to estimate the number of Salmonella cells per microcolony (Figure 1G). The results showed that about 40% of Salmonella exist alone, and there are no homologous bacteria in their direct (<5 μm) neighborhood, which is consistent with previous findings (55, 56) (Figure 1H). Based on the number and size of microcolonies and the volume of tissues imaged, we estimated the total number of Salmonella cells in each spleen. These data are consistent with the plating results and flow cytometry of independently infected mice (Figure 1B). Therefore, STP tomography reliably locates and quantifies a single Salmonella cell with a population density as low as one cell in 100 mm3 tissue.

Comparison of Salmonella subjects before and after chemotherapy with enrofloxacin or ceftriaxone for 10 days showed no change in microcolony size distribution. During the 99% antibacterial clearance of all Salmonella, we did not detect the preferential survival of a single isolated Salmonella cell, which is inconsistent with the proposed unique survival of a single non-replicating Salmonella persistent cell (57).

STP tomography allows us to locate rare surviving Salmonella during antibacterial chemotherapy at the whole organ level. The spleen has three main compartments (58) with different physiological functions: red pulp (RP), which serves as a blood reservoir and removes aging red blood cells; a wettable powder containing high-density B and T lymphocytes, which can induce blood loss Adaptive immunity to pathogens; and the intermittent marginal zone (MZ) that captures blood-borne particles and pathogens. We identified these three regions by in vivo staining of CD169 (also known as Siglec-1 or Sialoadhesin), which is a marker of macrophage subtypes in MZ (59) (Figure 2A and B). We noticed that the Salmonella clearance kinetics in RP is similar to that in MZ. Therefore, we merged the data of these two compartments (RP/MZ) and compared them with the data of WP.

Elimination of isolated Salmonella during antibacterial chemotherapy. (A) A tomographic image of a 5 mm spleen section stained in vivo with anti-CD169 antibody. (B) The localization of Salmonella cells (cyan, arrow) in the spleen compartment (RP; MZ, marginal zone containing CD169-positive macrophages; WP) of oral infection mice. (C) Salmonella load in the spleen compartment of mice that were orally infected during enrofloxacin chemotherapy. Each circle represents the STP data of a single mouse. These lines are a single exponential decay fit with a y offset. (D) The ratio of Salmonella load in WP to RP/MZ during chemotherapy with enrofloxacin (Enro.; the P value whose slope deviates from zero after logarithmic transformation data linear regression is shown as a dashed line). Data for 10 days of treatment with ceftriaxone (Ceftr.) are also shown. Each symbol represents the STP data of a single oral infection mouse, which is based on the localization of C. (E) Salmonella expressing GFP in the CD68-positive macrophages outside the B220-positive B cell area of ​​the spleen WP in the oral cavity. Mice after 10 days of star treatment. (Scale bar, 5 microns.)

Before treatment, approximately 95% of approximately 200,000 Salmonella cells were present in RP/MZ (60, 61) (Figure 2C). Enrofloxacin removes Salmonella in these compartments with single-phase exponential kinetics at a rate of 0.44 ± 0.23 log per day for about 7 days, after which the removal slows down significantly. After 10 days, about 300 Salmonella cells or about 0.15% of the initial load survived. In contrast, at the beginning of treatment, only ~10,000 (~5%) Salmonella cells were present in WP. The initial clearance rate of this smaller Salmonella subgroup is 2 times slower than RP/MZ (0.21 ± 0.12 log per day), and has been significantly slowed down after 4 days of treatment (9th day after infection, 9th day after infection). The Salmonella load decreased by less than 3 times in the next 6 days. After a total of 10 days of treatment, about 1,000 Salmonella cells or about 10% of the initial load of WP survived, which is 70 times larger than RP/MZ (Figure 2C and D).

The difference between ceftriaxone and enrofloxacin is that it targets peptidoglycan transpeptidase instead of DNA topoisomerase, and lacks activity against non-replicating bacteria (62⇓ ⇓ –65), which is similar to RP/MZ In contrast, the similar heterogeneous activity against Salmonella in WP (Figure 2D) indicates that the survival rate of different Salmonella does not depend on the specific antibacterial action mode or activity against non-replicating bacteria.

Immunohistochemistry showed that the Salmonella surviving in WP is located in CD68+ Ly-6C− F4/80− CD21/35− FDC− ER-TR7− gp38− CD3− CD19− WP macrophage (59 ) Inside the cell (Figure 2E). We did not find Salmonella cells in B or T cells, follicular dendritic cells, dendritic cells, inflammatory monocytes, lymphoepithelial cells or reticulofibroblasts. Our data is different from previous studies that attempted to identify the host cells of viable Salmonella cells by analyzing tissue homogenates by flow cytometry. Homogenization destroys most Salmonella-infected cells, even if done gently (55). The released bacteria can bind to other host cells in the tissue homogenate, such as B cells (66), while intact tissues do not contain Salmonella (61).

In summary, these data indicate that antimicrobial chemotherapy is particularly ineffective against a small number of Salmonella populations in WP.

The low antibacterial clearance rate of Salmonella in WP may reflect low drug delivery efficiency (67). Enrofloxacin effectively penetrates tissues (64); however, as opposed to directly supplying blood to the RP/MZ, the spleen WP only receives blood solutes (69) through diffusion through the catheter system (68). In addition, oral Salmonella infection caused extensive thrombosis in the splenic vein (70) (Figure 3A), and ultrasound imaging showed decreased blood circulation in the spleen (Figure 3B and C).

The role of drug delivery and non-replication persistence. (A) Autofluorescence photomicrograph of spleen section, with highly autofluorescent thrombus in the large vein. Similar images were obtained for all orally infected mice. (B) Ultrasonic 3D power Doppler mode images of blood flow in the spleen of uninfected and orally infected mice. (Bottom) Corresponding blood flow velocity in the splenic artery determined by pulsed wave Doppler ultrasound. (C) Average velocity of splenic arteries and veins in uninfected and oral infected mice. Each circle represents a mouse (two-way analysis of variance comparing naive and infected mice with repeated measures). (D) Enrofloxacin concentration in different spleen compartments of mice infected by mouth. Each circle represents a mouse (two-way ANOVA with repeated measures comparing RP and WP). For comparison, the MIC of enrofloxacin against Salmonella strains measured in Mueller-Hinton broth is shown. (E) Distribution of Salmonella replication rates in different spleen compartments. The graph represents the aggregated data of three orally infected mice at each time point. The few Salmonella cells that did not divide included Salmonella corpses killed by enrofloxacin that retained the TIMERbac protein. Previously reported in vivo data for ssrB mutants showed replication arrest in the mouse spleen (42) for comparison (grey). (F) The average Salmonella division rate, each circle represents a single oral infection mouse (P value: repeated measurement comparing red and white dental pulp by two-way analysis of variance).

To evaluate drug delivery, we used laser capture microdissection (SI appendix, Figure S4A) to separate WP and RP/MZ, and used enrofloxacin-specific ELISA to determine drug concentration. The enrofloxacin level in RP/MZ was slightly higher than WP, but on day 9 after infection, the enrofloxacin level in the two compartments was >1 mg mL-1 (Figure 3D), so it exceeded the effective killing Pharmacokinetics of bacteria-more than three times the pharmacodynamic target (peak concentration): MIC ratio>12) (71⇓ ⇓ –74). The killing test of Salmonella cultures grown in chemostat cultures under simulated in vivo conditions, the generation time was about 6 hours (42), and it was confirmed that enrofloxacin concentrations higher than 1 mg mL-1 would lead to saturation killing ( SI appendix, Figure S4B) (73). The accumulation of fluoroquinolones in phagocytes (75) indicates that Salmonella resident in macrophages has a higher exposure to enrofloxacin compared to the overall average concentration in WP (Figure 2E). These data indicate that there are enough medicines available locally for effective killing.

In order to track the killing of Salmonella by enrofloxacin in the body, we treated mice with enrofloxacin on the 5th day after infection. At different times before and after the administration, we killed the mice and prepared a spleen homogenate in a detergent-containing buffer that released 90% of the host cells and tissue debris as a suspension of a single bacteria Salmonella above (76). We use flow cytometry to classify Salmonella and determine their survival rate by comparing CFU (live Salmonella) and flow cytometry (live and drug-killed Salmonella) on the plate (42). To ensure that there is a sufficient number of Salmonella for rapid sorting, we use intravenously infected mice (first dose of enrofloxacin) on the 5th day after infection. Salmonella has an indistinguishable replication rate and sensitivity to enrofloxacin after oral or intravenous infection (42). The results showed that enrofloxacin killed 90% to 95% of Salmonella within the first hour (42), consistent with achieving the pharmacokinetic-pharmacodynamic goal (see above). However, almost no further killing occurred in the next 3 hours (SI appendix, Figure S4C), which was consistent with the results of chemostat cultures grown under simulated in vivo conditions (SI appendix, Figure S4D), and reported The in vitro data of enrofloxacin and other fluoroquinolones were reported (74, 77). This biphasic killing may be the result of the SOS response ("drug-induced tolerance") of the bacteria to fluoroquinolone exposure (77).

In general, compared with the bacteria that replicate rapidly in thick broth, the killing of Salmonella slowly replicating in mice and chemostats is quite limited, with similar fluoroquinolone concentrations killing a few within 1 hour of exposure. A logarithmic number of bacteria (42, 74, 77). Antibiotics are also moderately active in killing invasive Salmonella in human patients (78, 79). Salmonella with a slightly reduced sensitivity to fluoroquinolones is associated with a longer antipyretic time and an increased risk of failure (37, 80). Other Enterobacter species have similar and still effective treatment for reduced sensitivity (Clinical and Laboratory Standards Institute, CLSI, ciprofloxacin sensitivity breakpoint: Salmonella, ≤0.06 μg/mL; other Enterobacter species, ≤0.25 μg/mL; http://em100.edaptivedocs.net/Login. aspx).

From 1 hour to 4 hours after administration, the levels of enrofloxacin in RP/MZ and WP decreased by 5 to 10 times (Figure 3D), which is the same as that of mice that eventually eliminated enrofloxacin within 1 hour The half-life is the same (74, 81, 82). Gradually reducing the level of enrofloxacin will eventually stop the enrofloxacin-mediated killing, and even when the level of enrofloxacin is lower than the MIC, it will even promote the re-growth of some Salmonella (see below). It is known that sub-MIC trough levels appear at the recommended doses of mice (74) (also in human patients) (83), but it may not be a problem for the treatment of faster-growing pathogens, which are affected by Enno Peak levels of floxacin kill more effectively after each administration. In fact, the Cmax/MIC ratio is an indicator of the effective treatment of fluoroquinolones against various other pathogens, while the portion of time above the MIC is not (84). Higher or more frequent doses than recommended may reduce the duration of sub-MIC enrofloxacin levels, but also increase adverse reactions (85).

Taken together, these results show the pharmacokinetics-pharmacodynamic goals in the two splenic compartments. Therefore, insufficient drug delivery does not explain the slow clearance of Salmonella in WP from the 9th day after infection. However, due to the low replication rate of Salmonella, killing is mainly limited to the first hour after administration, and is generally quite limited.

It has been suggested that only non-replicating Salmonella survive extensive antibiotic treatment in the body (57). The study used continuous administration of enrofloxacin at a dose 30 times higher than recommended for rodents. The viability of Salmonella after treatment was evaluated indirectly based on the fluorescence changes during tissue lysis with detergent and incubation in rich medium. The observed ~65% decrease in fluorescence can reflect that the pre-loaded GFP was diluted by one or two Salmonella during the 12-hour incubation period. This is considered as evidence of viability (57), but the data does not exclude some GFP leakage from the recipient. The damaged Salmonella corpse serves as another explanation. The other two studies did not detect any colony-forming Salmonella under the same experimental conditions (42, 45), which indicates that non-replicating Salmonella particles are basically unable to survive after treatment. Other studies have shown that slowly replicating Salmonella can survive in mice treated with more clinically relevant doses of fluoroquinolones (44, 86).

To determine the replication status of surviving Salmonella in our model, we used STP tomography to quantify the fluorescence of Salmonella expressing TIMERbac, a reporter protein with different maturation kinetics between green and orange fluorophores. The ratio of green to orange fluorescence depends on the overall protein dilution associated with bacterial growth and division, and therefore indicates the rate of single cell replication (42). The TIMERbac data of Salmonella in infected mice are consistent with independent measurements of in vivo replication rates based on dilution of non-replicating plasmids. The TIMERbac data are also consistent with the net growth of Salmonella wild-type (WT) and attenuated mutants in mouse strains with different resistance to Salmonella infection (42, 87). STP tomography of TIMERbac-Salmonella showed that the heterogeneous replication rate in RP/MZ (42) was higher than that in WP. Due to the preferential killing of rapidly dividing subpopulations (42) (Figure 3 E and F), the median rate of 24 hours after the last enrofloxacin dose decreased in both compartments during the treatment period. However, most Salmonella cells in the two compartments continue to divide, consistent with the sub-MIC trough level (see above), and the TIMERbac pattern is still different from that of the ssrB mutant, which shows replication after several initial divisions in the body Stagnation (42, 88). This was also true on the 9th day after vaccination, when the clearance in WP basically ceased (Figure 2C). The few "non-dividing" Salmonella cells include Salmonella corpses killed by enrofloxacin with fatal DNA damage, but the intact envelope retains the TIMERbac protein (42). Direct observation of dividing cells 10 days after tissue section processing also supports ongoing Salmonella replication (day 15; Figure 2E and SI appendix, Figure S5). In addition, during the eradication of 99.5% of all bacteria, we did not detect any preferential survival of isolated single Salmonella cells (Figure 1H). Taken together, these data indicate that replication arrest is not a critical factor for Salmonella survival during treatment.

Host pressure, such as reactive oxygen species (ROS) and nitric oxide (NO), can promote bacterial resistance to antibiotics (7⇓ ⇓ ⇓ ⇓ ⇓ –13). Salmonella is exposed to various stresses in the infected spleen, and these stresses are highly heterogeneous at the single-cell level (89). This heterogeneity provides a unique opportunity to directly compare the antimicrobial susceptibility of Salmonella subgroups with high or low stress in the same host tissue. To test this idea, we infected mice with various Salmonella stress reporter strains. On the 5th day after infection, we administered enrofloxacin, and 1 hour later, we prepared detergent-treated spleen homogenate to release >90% of all Salmonella from host cells and tissue debris as a single bacterial suspension (76). We use flow cytometry to classify the subgroups of stressed and non-stressed Salmonella and compare the CFU (live Salmonella) and flow cytometry (live and drug-killed Salmonella) on the plate (42) (Figure 4A) ) To determine their survival rate. To ensure that there were enough Salmonella bacteria for rapid sorting, we used intravenously infected mice on day 5 (the first enrofloxacin dose). Salmonella has an indistinguishable replication rate and sensitivity to enrofloxacin after oral or intravenous infection (42).

The role of stress tolerance. (A) Strategies to determine the survival rate of stressed Salmonella subgroups during chemotherapy in intravenously infected mice. Salmonella carrying the GFP reporter plasmid for oxidative stress showed heterogeneous fluorescence in the spleen at 1 hour (left) (61). This signal is not affected by enrofloxacin exposure or various mutations in the body for 1 hour (SI appendix, Figure S6). Sort the GFPlow and GFPhigh subsets and determine the survival rate by comparing flow cytometry (live and dead Salmonella) with plate counts (live Salmonella). (B), the relative survival rate of Salmonella subgroups of different reporter gene constructs and Salmonella genotypes. Each symbol represents a single iv-infected mouse (*P <0.05; **P <0.01; ***P <0.001; ****P <0.0001; one-sample t-test of logarithmic transformation data and Holm- Šídák correction for multiple comparisons).

The results showed that about 10% of Salmonella subgroups with high PkatG-gfp activity showed exposure to host-derived ROS (61), which was about twice as good as enrofloxacin that survived GFPlow Salmonella with low oxidative stress (Figure 4A) And B and SI appendix, Figure S6A). Salmonella requires acrB instead of katG to achieve this protective effect, indicating that host ROS antagonizes the activity of enrofloxacin by stimulating AcrB-mediated drug efflux (90) rather than by changing redox physiology (61, 91).

Salmonella cells with high PhmpA activity showed that they were exposed to host-derived NO (61), and their survival rate was about 1.5 times lower than that of unexposed Salmonella (Figure 4B and SI appendix, Figure S6B). Salmonella mutants with detoxification defects ("Δ4", ΔhmpA Δhcp-hcr ΔytfE) increase the NO concentration in exposed bacteria (61) and have similar survival patterns. Therefore, physiological NO by itself may not increase lethality, but represents a microenvironmental marker that enhances the activity of enrofloxacin.

The survival rate of Salmonella cells with high PphoP-1 (92) activity is about 2.5 times lower than that of cells with low PphoP-1 activity (Figure 4B and SI appendix, Figure S6B). PphoP-1 is associated with Salmonella replication rate (SI appendix, Figure S4C), which is associated with increased sensitivity to enrofloxacin in the body (42) (Figure 4B). However, simultaneous monitoring of replication rate and PphoP-1 activity revealed an additional ~2-fold replication-independent effect of high PphoP-1 activity (Figure 4B and SI appendix, Figure S6D). Therefore, the microenvironment that stimulates PphoP-1 enhances the activity of enrofloxacin through factors that promote replication and are independent of replication. It is known that the low magnesium or low pH stimulation (93) that can induce PphoP-1 may be related to the enhanced enrofloxacin activity, because the magnesium-unresponsive Salmonella phoQ H157R mutant (94) shows a similar killing mode, while Low pH will reduce enrofloxacin activity (MIC value: pH 7.4, 0.03 mg per liter; pH 5.5, 0.06 mg per liter; pH 5.0, 0.24 mg per liter). Antimicrobial peptides, which also induce PphoP-1 (93), can promote antibiotic action (95), but it is difficult to test their effects in mice because the only described antimicrobial peptide non-responsive Salmonella mutants have additional signaling defects (96 ).

Taken together, these data show the complex regulation of Salmonella susceptibility by pressure exerted by heterogeneous hosts. However, the size of the impact is moderate, and some of the pressure is actually related to enhanced antibiotic activity. Opposing host pressure is a key factor in supporting the survival of Salmonella during antibacterial chemotherapy.

Salmonella infection triggers an inflammatory host response, with infiltration of neutrophils and inflammatory monocytes. These cells kill Salmonella through oxidative damage (61, 97, 98) and envelope destruction (99), and partly control disease progression (100, 101). Host inflammation supports the antibacterial clearance of bacteria (44, 66, 102⇓ ⇓ ⇓ ⇓ –107), so it may be a related factor affecting the survival of Salmonella.

Inflammation is not constant during treatment. On the contrary, various parameters showed a decrease in inflammation: clinical symptoms disappeared during chemotherapy (Figure 1A) and the serum concentration of the inflammatory cytokine IFNγ decreased (SI appendix, Figure S7A). In the spleen, the infiltration of neutrophils and inflammatory monocytes (Figure 5A and B), and the abundance of inflammation-related transcripts Ifng, Tnfa, and Mpo decreased (Figure 5C). A similar decrease in inflammation was previously observed during high-dose ciprofloxacin treatment in a mouse model of Salmonella enteritis (44).

Uneven and reduced inflammation can impair the removal of Salmonella. (A) Immunohistochemistry of mouse spleen sections infected with WT Salmonella or a mixture of wild-type and Enrofloxacin-resistant Salmonella (ER/WT) orally. Enrofloxacin treatment started on the 5th day. Ly-6G stains neutrophils and Ly-6C stains inflammatory monocytes. The fluorescence micrograph was inverted to improve visibility. (Scale bar, 1 mm.) (B) Neutrophil density at different time points after oral infection. Each symbol represents a single oral infection mouse. The bars represent the mean (**P <0.01; two-way analysis of variance for single and mixed infections on day 9). (C) Transcription abundance in different spleen compartments (red, RP/MZ; blue, WP) of mice infected with WT Salmonella or a mixture of WT and Enrofloxacin-resistant Salmonella (ER/IR). Each symbol represents an orally infected mouse (two-way analysis of variance used to compare log-transformed data of single infection and mixed infection on day 9). (D) Salmonella load in the spleen at different time points after oral infection with a mixture of WT and ER Salmonella. The red dotted line represents the fit of mice infected only with WT Salmonella shown in Figure 1B. Enrofloxacin treatment started on the 5th day. Each symbol represents a single mouse (circle, data from STP tomography; triangle, plating result; *P <0.05; ****P <0.0001; one-way analysis of variance using Holm logarithmic transformation data – Šídák correction for multiple comparisons). (E) During enrofloxacin chemotherapy, the clinical score of mice infected with only WT Salmonella (same data as shown in Figure 1A) or a mixture of wild-type and enrofloxacin-resistant Salmonella (ER/WT) ( Corrected with Geisser-greenhouse). (F) Continuous two-photon micrograph of mouse spleen on day 5 after oral infection with wild-type (WT, GFP) and enrofloxacin-resistant (ER, YFP) Salmonella mixture. (G) On the 5th day after oral infection with WT Salmonella or a mixture of WT and Enrofloxacin-resistant Salmonella (ER/IR), the distance distribution to the nearest neighbor of the same or another Salmonella strain. Data was pooled from three mice in each group. (H) The median nearest neighbor distance displayed in G for comparison. Each symbol represents an individual mouse that was orally infected (** P <0.01 used to compare WT-ER with all other groups; one-way analysis of variance with Holm-Šídák correction for multiple comparisons). (I) After 6 days of enrofloxacin treatment, use TGFβ neutralizing antibody or isotype antibody (Ctrl.) as adjuvant therapy on the impact of Salmonella tissue load, and use enrofloxacin and adjuvant anti-TGFβ therapy for the last 5 days. Each symbol represents CFU data from oral-infected mice (t-test of log-transformed data). (J) The mouse clinical score shown in I. The magenta line indicates anti-TGFβ therapy (enrofloxacin starts on day 5).

In addition, compared with WP, the neutrophil density and transcription abundance in RP/MZ were significantly higher than WP (Figure 5A-C). In particular, from day 5 to day 9 after inoculation, the abundance of Mpo transcripts, which are markers of neutrophils and inflammatory monocytes, decreased by> 1,000 times in WP, which is in line with what was observed by immunohistochemistry. The local consumption of these cell types is consistent. These data show that the inflammation is uneven and decreasing during antibacterial chemotherapy, which may be a response to local Salmonella load (100, 108⇓ –110). In particular, the inflammation at the beginning of the WP treatment was very weak, there was almost no Salmonella colonization, and as the salmonella load decreased, the inflammation in both compartments was decreasing.

The comparison of spatiotemporal inflammatory kinetics and Salmonella clearance rate shows that the clearance efficiency is particularly low when the density of neutrophils and monocytes is low. The important role of neutrophils and monocytes is confirmed by antibody-mediated depletion of these cells, which further impairs Salmonella clearance (66) (SI appendix, Figure S7B). These data indicate that weakened support of neutrophils and other inflammatory processes during treatment may lead to decreased clearance and failure to eradicate.

To test this idea, we aim to delay the decline of inflammation. We co-infected mice with the GFP-Salmonella WT strain (MIC, 0.03 μg/mL) and the reduced susceptibility YFP-Salmonella (ER) strain (MIC, 0.25 μg/mL, still lower than CLSI enrofloxacin sensitivity) ≤0.5 μg/mL breakpoint, http://clsivet.org/Login.aspx). The ER strain should be cleared more slowly during chemotherapy because its sensitivity is reduced to maintain stimulation of the inflammatory response, while the WT strain will be able to monitor the clearance of susceptible Salmonella under these conditions. At the beginning of treatment (day 5 after infection), mice infected with a mixture of strains had similar Salmonella burden and clinical scores compared to mice infected with WT alone (Figure 5D and E). As predicted, the efficiency of enrofloxacin treatment in clearing YFP-Salmonella ER in a single infection is lower than that of WT, which is related to the slower decline of disease symptoms (Figure 5D and E) and serum IFNγ (SI appendix, Figure S7A) Related, neutrophil and monocyte infiltration, and inflammation-related transcripts (Figure 5A-C).

In the presence of such persistent inflammation, antimicrobial chemotherapy successfully eradicated the co-infected susceptible WT strain, in sharp contrast to GFP-WT single-strain infection, in which Salmonella is impossible to eradicate (Figure 5D). This supports the idea that the reduced contribution of neutrophils and monocytes to Salmonella clearance leads to treatment failure. Salmonella ER maintained a high load under the same conditions, indicating that their sensitivity to enrofloxacin was reduced by about eight times, resulting in poor clearance even with sustained inflammation support. Human patients infected with Salmonella typhi serotype and moderately reduced sensitivity to fluoroquinolones also experience treatment failure (37, 111).

The direct competition between ER and WT Salmonella strains does not explain the accelerated clearance of WT in mixed infections, because these two strains do not co-infect the same host cell, but are separated by several uninfected cells (Figure 5 FH ) Consistent with clonal spread, there are no mixed foci of infection (56). Limited Salmonella competition in the spleen resulted in a constant Salmonella single-cell replication rate, a hundred-fold change in Salmonella tissue load during acute infection (42) and comparable Salmonella load in mixed and single-strain infections on day 5 (Fig. 5D) .

Our data are broadly consistent with previous observations that stimulating inflammation can reduce the antibiotic persistence of Salmonella in mouse enteritis models (44). However, injection of inactivated Salmonella or purified lipopolysaccharide (LPS) in our model did not accelerate the clearance of susceptible strains by Salmonella. The injected LPS is captured in the MZ (and other organs), but has limited spread to the spleen RP and WP (112, 113), which may limit its impact on locally persistent Salmonella.

In supplementary experiments, we determined the Salmonella clearance rate at low Salmonella, neutrophil and monocyte density. For this reason, we have started chemotherapy on the first day after vaccination. At that time, the salmonella load was still very low, neutrophils and inflammatory monocytes were still very few (SI appendix, Figure S7 C and D), and there were no clinical symptoms . We use intravenous infection in these experiments to reduce experimental variation. Early-onset enrofloxacin treatment prevented disease progression, but failed to reduce Salmonella load at all (SI appendix, Figure S7E). Compared with the beginning of treatment, Salmonella surviving after 3 days of treatment has an indistinguishable replication rate (SI Appendix, Figure S7F), again opposing the related effects of non-replicating cells.

In vitro cell culture infections, Salmonella survival from antimicrobial chemotherapy requires type III secretion system 2 (T3SS-2) and T3SS-2 effector protein SteE/SarA to induce anti-inflammatory, M2-like polarization of infected macrophages (114 ). In contrast, early treatment of infected mice is also ineffective against WT Salmonella, Salmonella ΔssrB on T3SS-2 deficiency and Salmonella ΔsteE (SI Appendix, Figure S7G), indicating that T3SS-2 activity and SteE/SarA are not required for antibiotics Persistence in the body. Salmonella cell culture infections are fundamentally different from in vivo conditions in all aspects (115), including the advantage of non-replicating persistence (in vitro) and almost nonexistence (in vivo) (42). The mutant phenotype also shows that there is a mechanistic difference between the survival of Salmonella during chemotherapy and the persistence of Salmonella in chronic infection (this requires SteE/SarA) (116), which is consistent with other differences in host cell types (Figure 2E) ( 116, 117) and Salmonella control mechanisms (antibiotics and CD4 T lymphocytes and TNFα) (102, 116, 118).

In summary, these data indicate that the continuous contribution of neutrophils and other inflammatory mediators can effectively eliminate Salmonella, but it is not without these contributions. Therefore, the always weak inflammation in WP and the further decrease in inflammation during antibacterial chemotherapy are the main reasons for the decreased clearance rate and the persistence of Salmonella. Uneven drug delivery, non-replicating persistent or slow-growing Salmonella subpopulations, stress-induced tolerance, and Salmonella’s immunomodulatory capacity must all have limited impact, because these factors are all factors when sustained inflammation supports chemotherapy. Can not prevent the removal of Salmonella. This is consistent with our direct analysis of these factors (see above).

Using continuous bacterial stimulation to maintain inflammation during chemotherapy is effective in eradicating susceptible Salmonella. However, this level of stimulation is also associated with prolonged clinical symptoms, which is unacceptable for improving patient treatment. A previous study found that in mice treated with lipopolysaccharides or CpG nucleotides, the durability of Salmonella antibiotics was reduced (44), but these reagents also have unacceptable safety risks. More targeted interventions may divorce the supporting role of certain inflammatory responses from aggravating disease symptoms. One possibility is to target the main cytokine TNFα, which plays a key role in Salmonella control during acute infections. However, TNFα apparently has no effect on the antibacterial clearance of Salmonella (102).

In order to determine a more effective and safe adjuvant therapy, we believe that infiltrating neutrophils and inflammatory monocytes capture and destroy Salmonella, and provide the main contribution to the antibacterial clearance of Salmonella (Figure 5A-C). Our goal is to use safe strategies to stimulate neutrophil function. These strategies are currently in clinical development as adjuvant cancer therapies. Specifically, we used neutralizing antibodies to inhibit the inhibitory effect of TGFβ on neutrophil function (119). TGFβ inhibition is currently in clinical development (120). We also tested IL-15 superagonist, an IL-15 complex with improved pharmacokinetics to stimulate the natural killer cell/neutrophil axis, resulting in neutrophils with excellent antibacterial properties (121). Various preparations of IL-15 are in active clinical development (122). We observed that anti-TGFβ and IL-15 superagonists both increased the antibacterial clearance rate of Salmonella by 85% to 95% (Figure 5I and SI appendix, Figure S8A) without aggravating or prolonging clinical symptoms (Figure 5J and SI appendix, Figure 5A) .S8B). Therefore, safe adjuvant immunotherapy is helpful for chemotherapy of difficult-to-treat salmonella infections.

In this study, we developed STP tomography to locate and characterize rare Salmonella cells in the entire host organ. Using this method, we found that the survival rate of a small Salmonella subgroup in the spleen WP in antimicrobial chemotherapy is several orders of magnitude better than that of the Salmonella group. Common assumptions cannot explain this observation: antibiotics reach WP in sufficient amounts, Salmonella continues to replicate, and host stress-induced Salmonella resistance and Salmonella's immunomodulatory ability have only a slight impact. These findings suggest that concepts derived from laboratory models of bacterial persistence may have limited relevance to the in vivo conditions of complex host tissues.

In contrast, antibacterial clearance of Salmonella requires the support of inflammatory cells such as neutrophils and monocytes, partly because antibiotics are less active in vivo against Salmonella slowly replicating in mice (42). The infiltration of neutrophils and monocytes follows the local Salmonella density, which is high in RP and MZ but low in WP (this Salmonella tropism may reflect the low abundance of target phagocytes) (61). Although a large number of neutrophils and monocytes support continuous antibacterial clearance from RP and MZ until the end of treatment, the number of such cells in WP is low and declining, and only supports slow clearance within a few days. In the end, the number of local Salmonella was below the threshold of pathogen-related molecular patterns to recruit inflammatory cells. Therefore, the bacteria largely escaped the killing, but survived and slowly replicated. The key role of inflammatory cell infiltration was confirmed by adjuvant therapy to maintain inflammation, which led to successful eradication. Therefore, the complex tissue structure, Salmonella tropism and related spatiotemporal host-pathogen dynamics determine the success or failure of antibacterial chemotherapy in vivo. These mechanisms require high-resolution, large-scale tissue imaging to discover, and are difficult to reproduce under laboratory conditions.

Previous studies of mouse strains genetically resistant to Salmonella infection also revealed the key role of host immunity on the outcome of antibacterial chemotherapy. Although treatment with clinically relevant doses of ciprofloxacin for 10 or 12 days failed to eradicate Salmonella from the spleen and liver (for susceptible mice), resistant mice can control the surviving Salmonella without recurrence (39, 123 ). In areas where typhoid fever is endemic, as many as 8% of people carry Salmonella in the liver, gallbladder, and/or gallstones, but do not show symptoms of typhoid disease (124⇓ –126), indicating that the infection is persistent but controlled. Therefore, although the failure of microbial eradication is similar to that observed in genetically resistant mice, some human typhoid fever patients may still show clinical cure. Failure to eradicate that leads to the recurrence of clinically detected disease may occur more frequently in patients with weakened immunity. This can be tested in future clinical studies.

Our results indicate that STP tomography is a powerful method to track and characterize rare micron-sized pathogen cells throughout the organ host environment. We focused our analysis on the spleen, but survivors of Salmonella were also observed in Pyle's mass, mesenteric lymph nodes, and liver. All these reservoirs can cause recurrence. The immediate increase of Salmonella in the spleen after stopping treatment in our model indicates local regeneration, but we cannot rule out migration between different parts. STP can be used to analyze all relevant organs, providing an opportunity to identify and compare the host microenvironment of surviving Salmonella and the fate of Salmonella during recurrence in tissues with different anatomy and functions.

However, STP also has important limitations. The optical resolution of the two-photon microscope prevents the detection of individual microbial cells within the microcolonies. The currently available optical sheet systems still have poor optical resolution for large tissues (127), but future improvements may improve this problem. In vivo antibody staining can reveal surface-exposed epitopes in RP and MZ, but cannot reveal intracellular structures or any epitopes in WP, excluding circulating antibodies in intact animals (68). Tissue removal methods can achieve antibody staining and visualization in permeabilized organs (127) and may be compatible with STP.

Our findings indicate an opportunity to improve the treatment of recurrent human invasive salmonellosis. In particular, adjuvant immunotherapy, which has been in clinical development for other medical indications, has promoted the elimination of Salmonella. Future research may unravel the optimal dose of this type of therapy. Further understanding of the inflammatory mechanisms that contribute to antibacterial clearance of Salmonella can improve specific targeting while minimizing adverse reactions. Our strategies and methods may be suitable for elucidating the mechanism of the persistence of bacteria in other difficult-to-treat infections (such as tuberculosis). This is also related to the colonization of heterogeneous bacteria. Even after long-term chemotherapy, the replication of mycobacteria in the body is slow but continuous, and antibacterial. Poor agent activity is related to dynamic changes in inflammation (128).

In this study, we developed an STP tomography method to study Salmonella cells that persist in infected animals. Randomize mice for comparative treatment experiments. The researcher was unaware of the allocation during the experiment and evaluation of the results. However, tomography and flow cytometry are performed automatically. We estimate the sample size through a sequential statistical design. We first infected 2 to 3 mice based on the effect size and variation observed in the previous study (129), and used the results to estimate the group size to obtain statistical significance with sufficient power. Where applicable, the number of repetitions is indicated.

The Salmonella strain used in this study is based on the Salmonella typhimurium serotype SL1344 hisG xyl (130, 131). Strains expressing GFP or YFP carry a fluorescent protein coding gene in the sifB locus, which has homogenous high activity in vivo (55). TIMERbac is expressed by the PybaJ promoter (132) on the low-copy free pSC101 derivative (42). The reporter strain carries a pSC101 derivative, which constitutively expresses mCherry or mtagbfp2 from the PybaJ promoter, and a transcriptional fusion between the candidate promoter and gfp-ova, encoding a degradable fluorescent protein variant (with a half-life in the range of 30 minutes) (133) . Salmonella mutants with gene deletions were obtained by two consecutive single crossovers, with a positive selection for kanamycin resistance and a negative selection for levansucrase-mediated sucrose sensitivity (134).

In order to simulate in vivo conditions, we used 100 mM 2-(N-morpholino)ethanesulfonic acid (MES), 5 mM KCl, 15 mM NH4Cl, 0.5 mM K2SO4, 1 mM KH2PO4, 50 μM MgSO4, 0.02% casamino acid Culture Salmonella 0.02% glycerol, 0.0042% N-acetylglucosamine, 0.003% glucose, 0.0018% glucosamine, 0.005% histidine, 25 mM NaHCO3, pH 5.5, and use 10% O2/5% in a micro thermostat The generation time of CO2 constant flow is 6 h. Before adding antibiotics, the growth conditions were maintained for 72 hours.

We used female BALB/c mice (Charles River Laboratories) between 10 and 16 weeks of age. Mice were housed in groups in individually ventilated cages with a 12-hour light/dark cycle. Mice have free access to food and acidified drinking water. All animal experiments were approved (permit 2239, Kantonales Veterinäramt Basel) and were carried out in accordance with local guidelines (Tierschutz-Verordnung, Basel) and Swiss Animal Protection Law (Tierschutz-Gesetz).

Age-matched female BALB/c mice (Charles River Laboratory) from 10 to 16 weeks of age were deprived of food and water for 4 hours, and then orally infected with approximately 5 × 107 CFU. On the 5th day after infection, 50% to 80% of the mice showed obvious disease symptoms (scoring 4 or 5, see below). In order to minimize the effect of variable incubation time after oral infection, we only analyzed those mice with disease. For some experiments, mice were infected by injecting approximately 1,000 CFU through the tail vein. Intravenously infected mice show smaller changes in Salmonella tissue load between individual mice, so fewer experimental animals are required to detect differences. Animals without iv infection were excluded from the analysis. Salmonella has an indistinguishable replication rate, net growth, and enrofloxacin sensitivity in the spleen after oral or intravenous infection (42). The size of the inoculum is determined by each infection plate.

The SI appendix, SI Materials and Methods describes detailed information on antibody injection, disease sign scores, and organ preparation.

The spleen is homogenized in ice-cold phosphate buffered saline containing 0.2% Triton X-100. All samples are kept on ice until analysis. Remove large host cell debris by centrifugation at 500 × g for 5 minutes. The supernatant contains more than 90% Salmonella as discrete single cells (76). The relevant spectral parameters were recorded in a FACS Fortessa II equipped with 405-nm, 488-nm and 561-nm lasers (Becton Dickinson), using side scatter (SSC) thresholds to exclude electronic noise. The details of the experiment are described in the SI appendix, SI Materials and Methods.

We embedded the fixed tissue in 5% oxidized agarose, which was freshly prepared by incubating type I agarose (Sigma-Aldrich, A0169) in 10 mM NaIO4 (Sigma S1878) for 2 hours. Covalently incubate the tissue with agarose by incubating for 2.5 hours in an excess of borohydride buffer (10 mM NaBH4, Sigma-Aldrich 452882, in 50 mM borax, Sigma 221732/50 mM boric acid, Sigma B6768, pH 90.5) Cross-linked. , Stir overnight at 4 °C to dissolve). The spleen infected with Salmonella expressing TIMERbac was embedded in unoxidized 5% type I agarose (Sigma-Aldrich, A0169) without cross-linking to prevent changes in the emission spectrum of TIMERbac. We use a two-photon whole organ tomography scanner (TissueCyte 1000, TissueVision) to image the embedded tissue. We cut physical sections every 50 microns (spleen), 20 microns (liver), or 30 microns (Peyer's spot and mesenteric lymph nodes) at a rate of 0.3 mm/sec (0.05 mm/sec for liver and non-crosslinked spleen). After each slice, we used three channels (blue channel: 450 to 500 nm; green channel: Chroma Technology ET510/20m, 500 to 520 nm for GFP and green TIMERbac emission, built-in 500 to 560 nm for GFP/ YFP; red channel: 560 to 650 nm) uses a 940 nm excitation laser (Mai Tai eHP, Spectraphysics). We use a 20x objective lens with a numerical aperture (NA) 1.0 (Plan Apo, Zeiss 421452-9880-000) to image in the x and y directions with a resolution of 0.435 μm.

We use the CIDRE method to retrospectively correct the illumination for each channel and optical plane (53), and further correct the imaging depth. Each imaging part is formed by 800-×800-μm "tiles" with an x/y overlap of 10%. z There is no overlap. Use the MATLAB-based software package StitchIt (https://github.com) to stitch the light correction from the entire spleen scan based on their expected position in the x/y tiled grid (not registered in the tiled overlapping area) Tile/SainsburyWellcomeCentre/StitchIt) (136). We train the MATLAB support vector machine classifier through GFP-based blue and green channels and all three-color channels of YFP, and use supervised machine learning to segment Salmonella cells expressing GFP and YFP. In order to eliminate the autofluorescence between particles classified as "YFP", we trained an eight-layer deep convolutional neural network built with MATLAB layer functions. All analyses were performed using MATLAB 2017b and 2018b. The custom code is available on GitHub (https://github.com/BumannLab/Li_BumannLab_2020).

The 50-micron-thick spleen section taken from TissueCyte was stored in cryoprotectant at -20 °C (300 g sucrose, 10 g polyvinylpyrrolidone 40, 400 mL ethylene glycol, 400 mL PB buffer (10 mM NaH2PO4, 40 mM Na2HPO4), after dissolution, add 1 liter of PB buffer) (137). The sections were washed three times with Tris buffered saline pH 7.4 (TBS) for 5 minutes each time. For antigen retrieval, sections are incubated in 5 mL of pre-warmed 10 mM sodium citrate (pH 8.5) at 37 °C for 45 minutes. The sections were washed three times with TBST (0.1% Triton X-100 in TBS), blocked with 1% BSA fraction V and 2% mouse serum in TBST for 1 hour, and then washed three times with TBST. Sections were incubated with 10 µg/mL primary antibody in TBST overnight at 4°C, washed three times with TBST, and stained with 10 µg/mL secondary antibody for 1 hour at room temperature (antibodies are listed in the SI appendix, SI materials and methods ) ). The sections were washed three times with TBST, mounted in 70 µL mounting medium (Dako), and inspected with a Leica SP8 confocal microscope.

The freshly prepared unfixed spleen was wrapped in aluminum foil and placed in a 50 mL tube pre-cooled in liquid nitrogen. Frozen spleen is stored at -80°C. Use a cryostat (Leica, CM1950) to cut frozen 50 µm thick sections at -14 °C, place on a PET FrameSlide (Zeiss, 415190-9101-000), and dry in a 50-mL tube containing 10 mL Silica gel orange (Sigma, 13767) overnight. The dry section is cut with a laser capture microdissection microscope (Zeiss PALM-Microdissection), with a 5x objective lens, 100% laser power, 68% focus and three cutting cycles. Record and summarize all cut areas. Use fine tweezers to collect the slices under a macroscope (Olympus SZ60). Use MaxSignal Enrofloxacin ELISA Kit (Bioo Scientific 1017-01) to quantify enrofloxacin, and convert it into a concentration value based on the corresponding microdissection area and 50 µm slice thickness.

For RT-PCR, use a cryostat (Leica, CM1950) to cut frozen 50 µm thick sections at -14 °C, place them on a PET FrameSlide (Zeiss, 415190-9101-000), and store in a 50 mL test tube Use 45 mL of pure ethanol (Sigma, 1117271000). Before laser dissection, remove the slides and place them in a 1 L bottle containing 500 mL of silica gel orange (Sigma, 13767) for 30 minutes. The dry section is cut with a laser capture microdissection microscope (Zeiss PALM-Microdissection), with a 5x objective lens, 100% laser power, 68% focus and three cutting cycles. Collect the microdissected tissue in TRIzol reagent (Fisher Scientific, 15596026). Use Quick-RNA Microprep kit (Zymo, R1050) to extract RNA. RNA was reverse transcribed with RT-PCT kit (Promega, A6010) and random oligonucleotide hexamer. Using glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as an internal control, cDNA was quantified using a real-time PCR system (Applied Biosystems StepOnePlus). The following oligonucleotides were used (GAPDH, TGTGTCCGTCGTGGATCTGA, CCTGCTTCACCACCTTCTTGAT; Ifng, AACGCTAACACTGCATCTTGG, GACTTCAAAGAGTCTGAGG; Tnfa, GGATACCAACTATTGCTTCAGCTCC, AGGCTCCAAAATAAATAGGGGCAGGGTC; and GCAGTCAAGTC).

A Vevo3100 (Visualsonics) equipped with a 32 MHz MX550D transducer was used to measure blood flow in mice under isoflurane anesthesia.

As shown in the figure legend, use GraphPad Prism 8.0.2 for statistical testing. Unless otherwise stated, error bars represent the mean ± SD. Statistical tests and related P-values ​​are reported in the results or legend. All statistical tests are bidirectional.

The MATLAB code used to analyze SPT tomography data has been stored on GitHub (https://github.com/BumannLab/Li_BumannLab_2020). All other research data is included in the article and supporting information.

We thank Sandra Meyer (Visualsonics) for help with ultrasound measurement, Wolf Heusermann (Biozentrum, Imaging Core Facility) for technical support at STP, and Frédéric Goormaghtigh (Biozentrum) and Guy Riddihough (Life Science Editor) for their helpful comments. This work was supported by the Swiss National Science Foundation 310030_156818 and 310030_182315 (for DB).

↵1J.L., BC and JF made equal contributions to this work.

Author contributions: JL, BC, JF, and DB design research; JL, BC, JF, and FRC research; NC, FRC, and RAAC contributed new reagents/analysis tools; JL, BC, JF, NC, FRC, and DB Analyze the data; JL and DB wrote this paper.

The author declares no competing interests.

This article is directly contributed by PNAS.

This article contains online support information https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2113951118/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.