Salmonella comprises two species and more than 2500 serovars with marked differences in host specificity; and is responsible for a wide spectrum of diseases, ranging from localized gastroenteritis to severe life-threatening invasive disease (Majowicz et al., 2010). It is generally transmitted to humans through consumption of contaminated food, or through fecal-oral route (Betancor et al., 2010; Gopinath et al., 2012; Bonardi et al., 2017). Even when salmonellosis cases are mild, it can be lethal, depending mainly on the particular infecting serotype and immunological status of the host. Salmonella has evolved multiple mechanisms, comprising a sophisticated repertoire of virulence factors, to compete with the gut microbiota for access to the host cells and successfully infect them, thereby avoiding or subverting host immune responses (Shea et al., 1996). A distinctive feature of Salmonella pathogenesis is the capacity to invade and live inside professional phagocytic cells inside Salmonella containing vacuoles (SCV). Particularly, macrophages represent an ideal niche for this purpose, whereas other innate immune cells such as neutrophils kill bacteria through oxidative burst, among other mechanisms (Burton et al., 2014). The initiation of an inflammatory response, triggered by many Pathogen-Associated Molecular Patterns (PAMPs) that Salmonella possesses, recruits neutrophils and inflammatory monocytes, in order to restrain the infection (Pham and McSorley, 2015). Besides the classical phagocytic role, these recruited cells have an additional mechanism for controlling bacteria that relies on the release of extracellular traps (ETs), a newly described pathway of programmed cell death known as ETosis. This released net-like material is composed of cellular DNA decorated with histones and cellular proteins (Brinkmann et al., 2004). These trapping structures have shown ability to neutralize and kill all kinds of microorganisms, ranging from viruses and bacteria to fungi and parasites (Brinkmann et al., 2004; Urban et al., 2006; Saitoh et al., 2012; de Buhr et al., 2018). Their relevance during pathogen elimination becomes apparent by the clinical finding that hereditary NET deficiency leads to invasive aspergillosis, that is later controlled by restoration of NET formation ability with gene therapy (Bianchi et al., 2009). Regarding Salmonella, it was the first microorganism that was reported to be killed by NETs, and since, several other studies have confirmed it and deepened into its variants (Brinkmann et al., 2004; Pieper et al., 2017). ETosis has been also attributed more recently to macrophages [reviewed in (Doster et al., 2018)] Salmonella-induced MET release, however, has never been reported so far, although other enterobacteriaceae family members as E. coli are known inducers (Webster et al., 2010). In this work, we observed the production of METs induced by Salmonella enterica serovar Typhimurium and recorded their effect on bacteria. Our results show for the first time that macrophages release extracellular traps upon Salmonella infection, where bacteria are retained and killed. This suggests a new defense mechanism against these highly prevalent bacteria. Although these results are preliminary they provide new insights in macrophages' role against Salmonella infection.

We first set up the conditions to evaluate MET release by J774A.1 macrophages, using uninfected and PMA-treated cells as negative and positive controls, respectively. MET release was not observed in untreated macrophages. Instead, PMA addition induced release of MET, as identified by fluorescent staining with the DNA-binding dye Hoechst, and laser scanning confocal microscopy ( Figure 1 ).

We then evaluated whether Salmonella induces MET formation in J774A.1 macrophages. Infection was carried out at a high MOI as a proof of concept, 40:1 (bacteria:cells ratio), and followed up for 1 hour (t = 10, 30 and 60 minutes) after infection. We found MET release after Salmonella infection; average size of the fibers 16.7 ± 10 μm, ranging between 6.6 and 50.3 μm ( Figure 2 ). MET release was observed as fast as 10 minutes after infection ( Figure 2 , central row, left panel), and by 30 minutes we still observed METs and macrophages in normal shapes ( Figure 2 , central row, central panel). After 60 minutes of infection, most macrophages were no longer viable as inferred from the lack of defined membrane and nucleus, as well as the presence of cellular debris ( Figure 2 , central row, right panel). MET structures were observed beyond cellular boundaries, close to the macrophages ( Figure 2 , central row, left panel). They presented uniform organization compared to a normal nucleus, revealed by the intensity of the staining. As the Hoechst 33342 dye binds to the minor groove of the DNA at AT-rich sequences differentiates distinct degrees of chromatin condensation state. The nuclear DNA with different zones of compaction is reflected by different degrees of Hoechst intensity as more intense is the fluorescence more condensed is the chromatin, while the METs that mainly consist of DNA fibers and enzymes are observed as an even blue color.

Since METs are mainly composed of DNA, we confirmed the nature of this process by inducing them in the presence of DNAse I ( Figure 2 , bottom row). METs were not observed when macrophages were pre-treated with DNAse I, confirming that DNA is the major structural component of these fibers. Moreover, cell damage was still observed, even in the presence of DNAse I, suggesting that the process of ETosis was taking place but METs were no longer visualized because DNAse I had already digested them. Another clue that this phenomenon is related to ETosis is the presence of several enucleated cells observed 30 minutes after infection ( Figure 2 , bottom row, central panel).

Extracellular DNA was quantified 30 minutes after MET release induction ( Figure 3 ), showing that. Salmonella infection significantly increased extracellular DNA. DNAse I pre-treatment resulted in a significant decrease of quantified DNA, confirming that a significant amount of DNA was exposed to DNAse I activity, hence outside cell boundaries.

Live imaging proved that Salmonella became trapped in METs ( Movie 1 ). Once trapped, bacteria stopped moving after 30 seconds of efforts to escape ( Movie 1 , i.e. minute 01:30.213 to 02:00.463). The trajectories of Salmonella tracked from Movie 1 are shown in Figure 4 . The analysis revealed that 11 bacteria were trapped and immobilized within the MET from the beginning of the movie (grey trajectories); 4 bacteria were detected swimming in the media and trapped within the MET, where they stayed until the end of the movie (red trajectories); 4 bacteria were trapped temporarily within the MET and then escaped the trap (blue trajectories), and a total of 72 swimming Salmonella entered and left the movie frame transiently without getting in contact with the MET (trajectories not shown). Tracking analysis also revealed that some bacteria were constrained to one particular area of the image, which corresponds to a MET (Still image 2, Movie 2 ). These trapped bacteria showed confined displacements of only a few microns ( Movie 2 ); and compared to the free bacteria, they have less movement.

Once bacteria were trapped, the GFP started to progressively diminish until the end of the movie acquisition. A GFP intensity decrease was observed by a time-lapse total fluorescence plot (data not shown). Swimming Salmonella did not show an intensity decrease after being exposed to the highest laser power for 3 minutes (data not shown). In another approach to visualize this phenomenon, we processed Movie 1 so as the emission color of the channel was changed to a red/green lookup table (LUT) ( Movie SM1 ) where high fluorescence intensity values are observed in green and low intensity values in red. Besides, the same LUT was applied for both green and blue channel (for GFP and DNA, respectively), as bacteria were stained with both fluorophores. This allows observing that most trapped bacteria undergo a switch from green to red, indicating a decrease in fluorescence; whereas DNA signal from the nucleus remains constant (bluing one in Movie 1 ). This last observation rules out that GFP dimming is due to photobleaching.

We then assessed through time the percentage of MET releasing cells at different MOIs, ranging from low (1) to high (40), in order to evaluate whether the process was influenced by bacterial load. As shown in Figure 5 , Salmonella infection rapidly increases the number of cells undergoing MET release, irrespective of the multiplicity of infection. Noteworthy, by 10 and 30 minutes the percentage of MET releasing cells is similar, while by 60 minutes uninfected cells exhibit comparable MET induction to Salmonella infected cells. This could be explained by the fact that minimum amount of extracellular material passively released by uninfected cells (e.g. in case of cell death) is sufficient to act as a DAMP (damage-associated molecular pattern) that triggers a proinflammatory cascade that includes cytokines as IFNγ or TNFα, both which are known MET inducers (Mohanan et al., 2013; Wong and Jacobs, 2013). In this way, sterile MET production could still be induced, amplifying the effect even when no infection is occuring. Indeed, this hypothesis was confirmed by the observation of MET release by fresh cells after the addition of supernatant from another culture, whereas this event was not observed after substituting supernatant with fresh medium ( Supplementary Figure S1 ).

Salmonella quantification after MET release revealed that approximately 10% of the initial inoculum died after 60 minutes of infection (p = 0.0034, Student T-test) ( Figure 6 ). No bacterial death was observed when DNAse I pretreatment was applied, suggesting that MET dismantling impairs bacterial death. Hence, the above mentioned GPF dimming could be attributed to a decreased GFP active expression by dying bacteria, most likely due to the action of enzymes present within the MET. Altogether, these observations reinforce our hypothesis that MET could trap and kill Salmonella.

Finally, this phenomenon was confirmed in primary macrophages. MET release was also observed in bone marrow-derived macrophages infected with Salmonella as for J774A.1 macrophages (30 minutes, MOI 40:1) ( Figure 7 ).

Salmonella is a major pathogen that still lacks efficient methods of control. Understanding the complexity of the immune response to Salmonella infection shall facilitate the development of improved vaccines and other tools to control the burden of Salmonella induced diseases. Here we demonstrated that MET release is part of this picture and contributes to killing of extracelullar bacteria. To date, MET release has been observed in response to many pathogens, including certain gram negative bacteria such as E. coli, H. influenzae and K. pneumoniae, amongst others. Hereby, we reported and characterized for the first time that Salmonella Typhimurium can induce METs. This event was observed as the extrusion of extracellular DNA fibers from macrophages, which are no longer visualized in the presence of DNAse I. The size of these fibers ranges from 10 to 50 μm approximately, which is in accordance with the reports of Brinkmann and Doster groups (Brinkmann et al., 2004; Doster et al., 2018).

The addition of live Salmonella to J774A.1 macrophage cultures induced METs release within 10 minutes. Our results are in line with those from Hellenbrand and Muñoz-Caro groups, which reported MET release triggered by other pathogens in less than 30 minutes (Hellenbrand et al., 2013; Munoz-Caro et al., 2014). The original description of NETs showed a rapid release of extracellular DNA (10 minutes after stimulation with PMA), but latter reports of the same group have suggested ETosis could take several hours (Brinkmann et al., 2004; Fuchs et al., 2007). In this case, one hour after infection, most macrophages were no longer viable, indicating that MET formation involves cell death. Nevertheless, it has been described that in certain circumstances vital ETosis could also occur (Byrd et al., 2013; Pilsczek et al., 2010; Rochael et al., 2015). The potential benefit of macrophage suicidal METosis could be related to Salmonella ability to survive inside them in SCV. In this way, two complementary functions against Salmonella could be taking place: to retain and kill extracellular bacteria within METs and also to avoid invasion and survival of bacteria inside macrophages, through self-killing. Indeed when infection is carried out at higher bacterial load (i.e. MOI 100:1), MET formation is also detected and macrophages are severely damaged ( Figure S2 ). However, surviving macrophages contain bacteria trapped inside SCVs ( Movie SM2 ).

Further analysis reveals that Salmonella trapped within these METs display restricted movement when compared to free swimming bacteria. In addition, once movement is restricted due to entrapment within MET bacteria die, as evidenced by GFP fluorescence dimming. The antimicrobial activity of ETs has been mostly attributed to enzymes such as elastase or myeloperoxidase; nevertheless, the antimicrobial effect of histones themselves has been reported decades ago (Hirsch, 1958), and DNA has been particularly linked to NET bactericidal activity more recently (Halverson et al., 2015). Since we did not define the MET composition besides DNA, we are not able to provide more information about the precise antimicrobial mechanism against these particular bacteria. Further studies are needed in order to define the nature of this new anti-Salmonella phenomenon. Zhao et al. (2018) have reported that BCA-enhancement of MET release is an efficient way to eliminate extracellular Salmonella, either in vitro or in vivo, therefore reinforcing our results that pose this as a bactericidal mechanism. We noticed a discrete bacterial killing by MET that could not be explained by phagocytosis, suggesting that Salmonella-induced METs have effective but limited antimicrobial capacity. This is in accordance with a previous report for E. coli, a model enterical pathogen, which proposed a role for METs on limiting dissemination rather than directly killing extracellular bacteria (Liu et al., 2014). The potential physiological role of ET release as an important strategy against pathogens has been recently reviewed (Ravindran et al., 2019). Here we show the following sequence of events: 1) Salmonella triggers rapid MET release, 2) bacteria are trapped within these METs and die, and 3) macrophages die.

Different stimuli have been reported to trigger MET release [reviewed in (Doster et al., 2018)] while the precise mechanisms are poorly understood. The particular signals that elicit METs formation upon Salmonella infection are still unknown. It has been described that amyloid fibers trigger NET release (Azevedo et al., 2012). Salmonella, as other enterobacteria, produces curli fibers, which are bacterial amyloids. These fibers are potent inductors of an inflammatory response and are involved in early stages of cellular aggregates and biofilm formation (Barnhart and Chapman, 2006). Since LVR01 is a biofilm producing strain, it is then possible that Salmonella fimbria curli could be inducing MET release.

To sum up, we observed for the first time that Salmonella Typhimurium LVR01 triggers MET release under experimental conditions, which subsequently leads to bacterial clearance and macrophage death. Further studies are needed to fully address the relevance of this mechanism in immunity against Salmonella as well as the precise molecular mechanisms involved.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Comisión Honoraria de Experimentación Animal (CHEA), Facultad de Medicina, Uruguay.

AM, MM, and PS contributed to the conception and design of the study. AM and MM performed the experiments. PS, NC, JJ-W, and SH obtained, processed and analyzed confocal images. JAC contributed to interpretation of the data. AM wrote the first draft of the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by the Chilean Millennium Science Initiative P09-015-F; FONDECYT 1181823; FONDECYT 1211988; DAAD 519605, CORFO 16CTTS-66390 and ACM 170003 to SH, PS, JJ-W, NC-H; AUCI Program for South-South Collaboration Uruguay-Chile.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.