SifA SUMOylation governs Salmonella Typhimurium intracellular survival via modulation of lysosomal function

One of the mechanisms shaping the pathophysiology during the infection of enteric pathogen Salmonella Typhimurium is host PTM machinery utilization by the pathogen encoded effectors. Salmonella Typhimurium (S. Tm) during infection in host cells thrives in a vacuolated compartment, Salmonella containing vacuole (SCV), which sequentially acquires host endosomal and lysosomal markers. Long tubular structures, called as Salmonella induced filaments (SIFs), are further generated by S. Tm, which are known to be required for SCV’s nutrient acquisition, membrane maintenance and stability. A tightly coordinated interaction involving prominent effector SifA and various host adapters PLEKHM1, PLEKHM2 and Rab GTPases govern SCV integrity and SIF formation. Here, we report for the first time that the functional regulation of SifA is modulated by PTM SUMOylation at its 11th lysine. S. Tm expressing SUMOylation deficient lysine 11 mutants of SifA (SifAK11R) is defective in intracellular proliferation due to compromised SIF formation and enhanced lysosomal acidification. Furthermore, murine competitive index experiments reveal defective in vivo proliferation and weakened virulence of SifAK11R mutant. Concisely, our data reveal that SifAK11R mutant nearly behaves like a SifA knockout strain which impacts Rab9-MPR mediated lysosomal acidification pathway, the outcome of which culminates in reduced bacterial load in in vitro and in vivo infection model systems. Our results bring forth a novel pathogen-host crosstalk mechanism where the SUMOylation of effector SifA regulated S. Tm intracellular survival.


Introduction
The Gram-negative bacterium Salmonella Typhimurium (hereafter to be referred as S. Tm) causes gastroenteritis in humans.Although, the infection clears out in a week's span in healthy individuals, there is impending danger of bacteremia in immunocompromised individuals and infants.Infections caused by S. Tm are frequent, leading to a significant health burden in both developing and developed world [1].The advent of multi drug resistant invasive strains of S. Tm also possesses additional threats notably in Sub-Saharan African countries and among children [2,3].Information regarding detailed molecular mechanisms are required to gain better insights behind the pathogenesis of this bacterium.
S. Tm pathogenicity is attributed to the presence of unique cluster of genes in its genomes making up the Salmonella pathogenicity islands.Among many SPIs, Salmonella pathogenicity island 1 (SPI1) and Salmonella pathogenicity island 2 (SPI2) are well explored [4].A coordinated action of SPI-1 effector proteins aid in S. Tm entry into host along with induction of proinflammatory signaling mechanisms [5,6].
Post uptake, S. Tm resides in a secluded membranous compartment called Salmonella-Containing Vacuole (SCV) which sequentially acquires markers of endocytic pathway.Early endosome antigen 1 (EEA1), GTPases Rab5 and Rab11 are initially recruited to SCV which are later interchanged with late endocytic markers Rab7, Lysosomal associated membrane proteins 1,2.3 (LAMPs) and vacuolar ATPases [7].Sequestration of S. Tm within SCV is important since it enables intracellular replication and evasion of host defense mechanisms [8][9][10].Members of the Rab family are crucial regulators of the intracellular membrane trafficking, thereby, they also modulate the fate of SCV [11].Aided by effector functions, S. Tm are known to target the host GTPase Rab7 to modify SCV maturation [12].Recently our group also showed that S. Tm manipulates Rab7 function by interfering with Rab7-SUMOylation, a posttranslational modification mechanism [13].Several crucial aspects of SCV biogenesis and its stability are majorly fine-tuned by an SPI-2 effector Salmonella induced filament A (SifA) protein [14].SifA directs PLEKHM1 (Pleckstrin homology and RUN domain containing M1), Rab7 and HOPS (Homotypic fusion and protein sorting) complex to mobilize late endosomal membrane compartments for SCV generation [15].PLEKHM1 directly interacts with both Rab7 and SifA, through its pleckstrin homology domain (PH domain) [16].A ΔSifA mutant of

PLOS PATHOGENS
Salmonella escapes SCV and fails to replicate in macrophages and thereby ends with compromised pathogenicity in vivo [17].
Tube like membranous extensions generated from SCV membrane, collectively called as Salmonella-induced filaments (SIFs), are generated during later stages of infection [18,19].SIFs are required for SCV nutrition and stability [20][21][22].SifA function alters microtubule dynamics which is required for SIF formation [23] as well as recruitment of lysosome associated membrane protein 1 (LAMP1) on to the tubules [24,25].Furthermore, SifA forms a complex with SKIP (SifA and Kinesin Interacting Protein) also known as PLEKHM2 (Pleckstrin homology and RUN domain containing M2 protein) which enable regulation of plus-enddirected motor kinesin-1 on SIFs [26][27][28].Though, SCV later fuses with lysosomes, they are devoid of recycling mannose-6-phosphate receptors (MPRs) responsible for trafficking lysosomal hydrolases into lysosomes [29,30].The GTPase Rab9 through its interaction with SKIP associates with MPRs for routing these hydrolases into lysosomes [31,32].During infection S. Tm interferes with Rab9-SKIP interaction and MPR pathway, together these mechanisms leads to blockade of lysosomal acidification [32,33].Thus, several mechanisms and functions crucial to Salmonellosis are attributed to SifA via its interaction with numerous host adapters.Being able to carry out several regulatory roles by SifA is a paradigm, the mechanistic details of which are still not fully understood.To accomplish multiple functions, effectors often utilize host PTMs.
Our group demonstrated that S. Tm control several aspects of host cell physiology by altering the SUMOylation machinery [34].SUMOylation, a small ubiquitin like modifier system is known to post-translationally modify proteins and thereby affecting their function.SUMOylation being a powerful machinery can impact stability, folding, interaction and localization of their substrates [35].SUMO protein exists as multiple isoforms in mammalian cells.SUMO-1 is 48% identical to SUMO-2 and shares 46% identity with SUMO-3, whereas SUMO-2/3 are 95% identical to each other.SUMO2/3 form poly chains on target proteins by conjugation, while SUMO-1 chains are only observed under in vitro condition [36,37].
The global SUMOylome in hosts undergoes a major alteration during S. Tm infection, several of which belong to vesicular transport pathway [13].Also, bacterial effectors from certain pathogens Anaplasma [38] and E. chaffenis [39] undergoing SUMOylation has been shown to be beneficial for pathogenesis.It is unclear if SifA or any of S. Tm effectors undergo PTM modification by SUMO.This study aims to identify whether SifA undergoes such processes and if so, how this modification could add complexity to S. Tm pathogenicity.

Lysine 11 of SifA gets SUMOylated during infection
Since, SifA and Rab7 reside in proximity in the SCV and Rab7 SUMOylation is prevented by S. Tm during infection [13], we investigated possible interaction of SifA with components of SUMOylation machinery.In our preliminary analysis, a yeast two hybrid assay was conducted to probe the interaction between SifA with Smt3, a yeast SUMO1 ortholog or Ubc9, the human E2 enzyme of SUMOylation pathway.The genes encoding SifA, Smt3, Smt3ΔGG and human Ubc9 were cloned in both yeast bait expression vector pGBKT7 expressing GAL4 DNA binding domain (BD) and pGADT7 vector designed to express GAL4 activation domain (AD) separately.Yeast cells (Y187, Clonetech's matchmaker Y2H system) expressing both bait (pGBKT7) and prey vector (pGADT7) were confirmed by their growth on selection plates (Fig 1A).We observed that both Smt3 and Ubc9 showed interaction with SifA as seen by the growth of yeast on plates devoid of histidine and adenine.Notably, the mutant protein Smt3ΔGG, where C-terminal diglycine motif was deleted showed no growth.The diglycine motif at the c-terminal tail of SUMO paralogs participate in the iso-peptide bond formation with the lysine residue of the target protein, therefore the absence of this hinders SUMOylation [40,41].Screening was further carried on selection plates containing 2-Amino-1H-1,2,4-triazole (3-AT) at a concentration of 2.5mM adding stringency to the histidine-based growth selection.Similarly, selection comprising either one empty parent vector or between parent vectors led to complete inhibition of growth (Fig 1A).These data led us to conclude that SifA interacts with Smt3 and Ubc9.It was therefore important to examine if this interaction was a result of SifA undergoing SUMOylation.

PLOS PATHOGENS
The primary sequence of SifA was analyzed through various online 'SUMO' prediction software, including GPS-SUMO, JASSA and SUMO plot [42,43], for identifying the presence of probable SUMO consensus sites.Multiple lysine harbored by SifA were picked by this software (Fig 1B).However, the search from all these above-mentioned programs collectively pointed at lysine11 (K11) lying within a SUMO consensus motif CKXE/D [44] displaying higher score in comparison to the cut-off (Table 1).Apart from the indicated SUMO consensus sites, primary sequence of SifA accommodate two probable SUMO interaction motif (SIM) sites spanning from amino acids 97-101 and 173-177 although this SIM site is not picked by the prediction software used in the study.SIM sites are short stretch of hydrophobic amino acids present in a protein through which it can interact with other proteins that are SUMOylated [45].
To assess the possibility of SUMOylation at lysine 11, a mutant was generated with an arginine substitution (K11R) via site directed mutagenesis.The recombinant proteins wild type SifA (GST-SifA WT ) and lysine mutant (GST-SifA K11R ) were expressed in E. coli Rosetta strain and purified.The individual protein fractions obtained after thrombin cleavage from chromatographic column were run on SDS-PAGE where a band size corresponding to SifA (37 kDa) and GST (25 kDa) were visible (Fig 1C).In vitro SUMOylation assay was conducted with these purified fractions of SifA, using components from ENZO SUMOylation kit.Post reaction the mixture was run on SDS-PAGE and probed with anti-SUMO-1 and anti-SUMO2/3 antibody separately.As indicated in the immunoblot in Fig 1D, higher molecular weight bands were observed in SifA WT when probed with SUMO1 antibody, whereas the intensity of bands was significantly less in SifA K11R , also quantified in the graph (Fig 1D).Presence of thrombin cleavage through chromatography by using superdox 200 column run on an SDS PAGE (12%) and Coomassie stained for expression analysis.Bands at 25 kDa correspond to GST carryover.(D, E) Resultant immunoblot of isolated proteins subjected to in vitro SUMOylation assay separately probed for both SUMO1 and SUMO2/3 isoforms.The results indicated the presence of higher molecular bands in wild type SifA incubated with SUMO-1 isoform (indicated in the densitometry plot, right panel), whereas the intensity of bands was remarkably less in K11R mutant.(F) In-bacto SUMO conjugation assay for GST tagged SifA expressed in E. coli BL21 along with pE1E2S1 plasmid encoding SUMO enzyme machinery.The blots were probed by anti-GST antibodies.Rab7 was used as a positive control [13] (G) Coimmunoprecipitation of eGFP-SifA expressed in HCT-8 cells pulled down by SUMO-1 antibody, succeeding immunoblot probed by anti-GFP antibody showed interaction only in wild-type SifA expressing lysates.(H) Spectra displaying fragmentation of SifA peptide bearing diglycine at K11. Inset represents schematic of SifA peptide conjugated to di-glycine of SUMO-1.The blots shown here are representative of at least three biological replicates.and 1E).Purified RanGAP1 protein was used as positive control since it is known to undergo SUMOylation by both isoforms of SUMO.SUMOylation of RanGAP1 yields multiple bands observed above 40 kDa [46].The above observations were further reinforced by another assay referred to as 'in bacto-SUMOylation assay.'This assay majorly relies on the expression of plasmid pE1E2S1 encoding SUMO machinery enzyme components (E1, E2) and SUMO1 (S1) besides the expression of target protein, SifA in E. coli [47].The pE1E2S1 plasmid along with constructs encoding SifA WT or SifA K11R were co-transformed in E. coli BL21 cells.The lysates from these cells were run on SDS-PAGE and immunoblotted.As depicted in Fig 1F, a prominent band around 120 kDa was observed, indicating possible SUMO-modification of SifA, only in sample comprising co-transformed wild type SifA and pEIE2S1 construct, but not with those with pE1E2S1 and K11R mutant.As seen from our previous assays, presence of multiple bands in the blot may be due to SUMO-chain formation.The mammalian protein Rab7 and its corresponding SUMO mutant, Rab K175R were also included here as a positive control for the assay respectively [13] (Fig 1F).As anticipated, higher mobility bands in case of Rab7 but not for Rab K175R was observed.
In order to validate these observations in vivo in mammalian cells, a co-immunoprecipitation (co-Ip) assay was conducted using human intestinal epithelial cells, HCT-8.HCT-8 cells were transfected by plasmids bearing both SifA WT and SifA K11R genes cloned in mammalian peGFP-C1 vector separately.After 24 hours of transfection, the transfected cells were subjected to infection by ΔSifA mutant for 7 hours.Immunoprecipitation was carried by anti-SUMO1 antibody followed by immunoblotting with anti-GFP antibody.A band corresponding to GFP-SifA WT around 62 kDa and prominent upper band, representing SUMOylated version of SifA, was observed only in wild type samples.Notably, in line with our above results, we did not observe any bands in SifA K11R transfected samples making it indistinguishable from isotype control (Fig 1G).The expression of SifA K11R protein was observable in the input sample (Fig 1G).To investigate whether SifA undergoes SUMOylation during infection, HCT-8 cells were infected by SifA WT and SifA K11R complemented strains in ΔSifA (pSifA WT & pSifA K11R respectively) using prokaryotic expression vector pACYC-184.The infection was allowed to proceed till 7 hours at a MOI (Multiplicity of Infection) of 1:40.A faint band corresponding to SUMOylated SifA was identified only in SifA WT  Finally, an in-situ detection of SUMO modified SifA in cells were accomplished through a proximity ligation assay (PLA).This assay was conducted in HeLa cells transfected by peGFP-C1 plasmids containing SifA WT or the non-SUMOylatable K11R mutant SifA (hereafter to be also referred as SUMOylation deficient SifA) for 24 hours.The transfected cells were probed by anti-GFP antibody (rabbit) and anti-SUMO-1 antibody (mouse).PLA mainly relies on the attachments of oligonucleotide labeled secondary antibodies against the primary that will only be joined and ligated if the interacting proteins are in proximity.The interaction is expected to result in the formation of a closed circular DNA template which is eventually amplified and detected by detection oligos which can be visualized as discrete puncta (S1B Fig) .In the control sample, one primary antibody (anti-GFP) was omitted.As shown in the S1C Fig, the puncta from individual cells quantified suggested interaction between SifA and SUMO-1.The number of punctae was observed highest in SifA WT transfected cells in comparison to those with SifA K11R (S1C Fig), while, there was no signal in the negative control (left panel).Based on the results from all these above experiments, we concluded that SifA undergoes SUMO-modification by SUMO-1 isoform at lysine 11.

Ectopically expressed SifA K11R displays compromised stability
As it is well established that SUMOylation can potentially impact substrate's stability, folding, interaction or/and compartmentalization [35], we therefore set out to investigate the biological consequences of SifA SUMOylation.
HCT-8 cells transfected by peGFP-C1 plasmids containing SifA WT or SifA K11R were monitored for indicated time periods as shown in supplementary S2A Fig.There were no notable differences in expression at 24-or 36-hours post transfection between SifA WT and SifA K11R (S2B Fig).However, at 48 hours post transfection, compared to SifA WT a significant decrease in pools of SifA K11R was observed as shown in S2B Fig. Next, cycloheximide chase assay was executed to determine the steady state stability and half-life of both proteins [48].To accomplish this, cycloheximide at a concentration of 100 μg/ml was incorporated into transfected HCT-8 cells succeeding 24 hours of transfection (Fig 2A).The decay curve represents amount of remaining percent protein acquired from three independent chase (individual replicates are plotted) at each time point in comparison to start of the experiment (time 0).As it is evident from the curve (Fig 2B), the stability of SifA K11R was considerably less compared to the wild type protein beginning at 2 hours.At 4 hours the pools of SifA K11R were at 25% of the initial levels (two replicates), whereas the levels of SifA WT were around 50% of initial (two replicates).Now to ascertain the pathway leading to the degradation of SifA K11R , transfected cells were treated with MG132 (inhibitor of ubiquitin proteasomal pathway) or bafilomycin A1 (blocker of lysosomal degradation).It was evident that the levels of both the proteins did not alter significantly upon MG132 treatment (Fig 2C and 2D), only a mild increase in both proteins were seen.However, treatment of bafilomycin A1 (BA1.)restored the levels of SifA K11R in addition to significant increase in the levels of both proteins compared to untreated samples (Fig 2C and 2D).Thus, the relative pools of wild type and SifA K11R were identical in presence of this drug indicating a role of lysosomal activity in regulation SifA K11R expression.Additional proteins cyclin B and c-Jun were included as controls.Cyclin B is known to undergo proteasomal lysis [49] Over expression of SifA results in increased lysosomal activity compared to vector control, while lysotracker staining intensity further escalated in presence of GFP SifA K11R hinting the role of SifA in modulating lysosomal function through its SUMOylated lysine 11.Hence, the data from above experiments implied that stability of ectopically expressed SifA K11R is lesser due to active lysosomal activity compared to its wild type counterpart.

Infection with Salmonella bearing SUMO deficient SifA results in altered expression of host proteins, SKIP and MPR
Taking into consideration, the role of lysosomes in regulating stability of SifA from our previous findings and the fact that one of major functions of this effector is protection of the bacterium from lysosomal degradation, we decided to decode how each downstream adapter involved in this process respond to SifA SUMOylation.As mentioned in the introduction, SifA is involved in subverting retrograde trafficking of mannose-6-phosphate receptors (MPRs) hence inhibiting lysosomal acidification.It is known that SifA-SKIP complex is formed during Salmonella infection enables sequestration of Rab9 protein leading to misrouting of MPRs [32].Also, the host interactor of SifA, PLEKHM2 (SKIP) is regarded as a negative regulator of lysosomal activity [32] and recently its expression was known to be modulated by certain strain of Mycobacteria [51].We began by examining whether SKIP expression gets modulated during Salmonella infection.Accordingly, HCT-8 cells were infected by the following strains of Salmonella; SL1344 (wild type), ΔSifA (SifA gene knock-out in SL1344 background), ΔSifA complemented with SifA WT (referred as SifA WT ) or SifA K11R (referred as SifA K11R ) (cloned in pBH-HA vector) individually at a MOI of 1:40 for 4 hours.Samples infected by SL1344 showed some increase (although not statistically high) in SKIP levels compared to lysates generated from uninfected control, but the levels did not rise in ΔSifA infected samples (Fig 3A).Notably, there was a decrease in SKIP expression in SifA K11R infected samples (p value = 0.056) (Fig 3A).Similarly, the same samples were probed for the presence of Cation Independent mannose 6-phosphate receptor (CI-M6PR).The blot in Fig 3B showed the levels of CI-M6PR are high in ΔSifA and SifA K11R infected samples in comparison to their respective wild-type proteins.Lysates obtained from ectopically expressed SifA WT and SifA K11R in HCT-8 cells after 24 hours of transfection exhibited a drastic variation in the expression of protein CI-M6PR (Fig 3C).The levels of CI-M6PR protein were significantly enhanced in case of SifA K11R expressing lysates (Fig 3C).We surmised that, unlike its wild type counterpart, SifA K11R is deficient in the known function of preventing lysosomal acidification.Further, we also checked the ability of SifA K11R to interact with its adapters SKIP and Rab9 via independent co-IP experiments from transfected HCT-8 cells.As depicted in Fig 3D, band of SKIP protein from IP fraction corresponding to SifA K11R pull down was fainter compared to SifA WT .Likewise, the interaction between SifA K11R and Rab9 necessary for trafficking of lysosomal hydrolases was also lesser compared to wild type (Fig 3E), hinting inability of SUMO mutant of SifA in sequestering Rab9 through its interaction with SKIP.Progressing, Rab9 CI-M6PR co-localization was compared during an infection with complemented strains SifA WT and SifA K11R .Co-localization assays were conducted in HeLa cells, as this cell line is widely employed for imaging studies related to Salmonella infection [52][53][54].HeLa cells were infected by SifA WT or SifA K11R separately at a MOI of 1:40 for a duration of 7 hours, after which these infected cells were fixed, stained for CI-M6PR (green channel), Rab9 (red channel) cell nuclei and indicated strains were stained by DAPI (blue channel) (S3A Fig) .The combined analysis from individual cells portrayed a statistically significant increase in colocalization between CI-M6PR and Rab9 in case of infected cells by SifA K11R compared to wildtype.Since the co-localization between Rab9 and CI-M6PR were affected, the distribution of both these proteins over SCVs were also assessed.CI-M6PR receptor was found to be better localized to SCV marker LAMP1 on SCVs from SifAK 11R infected cells (5.Thus, the expression and colocalization of CI-M6PR were differently affected during SifA WT  and SifA K11R infection.The inference from above observations concur SUMOylation of SifA at 11 th lysine promotes misrouting of Rab9-MPRs trafficking.Thus, it is hypothesized that SifA K11R is less potent of providing protection to SCV from digestion by lysosomal hydrolases.

Sumoylation of Salmonella effector
The predicted outcome in the current scenario stipulates escalated lysosomal activity in SifA K11R infected cells compared to SifA WT .

SUMOylation of SifA is required for preventing lysosomal acidification and eventual lysosomal clearance of Salmonella
HeLa cells were treated with a ratiometric probe called lysosensor yellow/blue DND to monitor the difference in lysosomal acidity across the samples.The dye fluorescence shifts from blue to bright yellow as the pH shifts from neutral to acidic environment [55,56].As shown, in Fig 4A , from the fields corresponding to infection with SifA WT and SifA K11R , the intensity of yellow panel was higher in latter compared to the former, thus implying that SifA K11R is incapable of lowering the lysosomal hydrolase activity.Fluorescence from uninfected cells was also shown for reference (Fig 4A and 4B).Further, we quantitated the amount of a lysosomal hydrolase cathepsin D over SCV.HeLa cells were infected by ΔSifA, SifA WT or SifA K11R at the regular MOI and SCVs from infected cells were co-stained for cathepsin D and LAMP1 (Fig 4C).The presence of cathepsin D (red) was indicated graphically as signals overlapping on SCV marker LAMP1 (green) and bacteria (blue channel) on selected SCVs as ROIs (Fig 4C and 4D).From these images it was evident that cathepsin D co-localized more on SCVs from ΔSifA and SifA K11R infected cells, whereas as they were barely present on SCVs from SifA WT infected cells (Fig 4E).From here on, we concluded that SUMOylation of SifA is necessary for S. Tm to overcome degradation by lysosomal hydrolases.

Host cells infected by S. Tm with SUMO deficient SifA display fewer SIFs
As mentioned earlier, S. Tm pathogenicity relies greatly on the ability to generate SCV and SIFs in the endocytosed host cells, both of which are majorly attributed to effector SifA [14,23].These phenotypes were surveyed in cells infected with SifA WT and SifA K11R to understand the role of SUMOylation in SifA function.The protein galectin-8 is regarded as an indicator of SCV integrity.Galectin-8 binds to exposed host glycans on damaged SCVs activating xenophagy in host [10].Galectin-8 (Gal-8) GFP plasmid transfected HeLa cells were infected by complemented strains possessing SifA WT or SifA K11R or SL1344 or ΔSifA at an MOI of 1:40.Live cells were imaged 1 hour pi to determine the acquisition of galectin-8 protein on SCV.From Fig 5A it was evident that gal-8 (green channel) localized on SCV (red channel) distinctly during infection by shown strains which is also represented graphically in Fig 5B .ΔSifA had high amount of galectin-8 over SCVs compared to wild type, indicating deformed vacuole formation, followed by SifA K11R , although there was no drastic difference in their accumulation compared to SifA WT (Fig 5B).Proceeding further, infected HeLa cells were enumerated for visible SIFs to understand the effect of SifA SUMOylation on SIF generation.In general SIFs are formed at later hours of infection by fusion of SCV with late endocytic compartments carrying membranes enriched in glycoprotein LAMP1 [57].Thus, to enable visualization of SIFs, HeLa cells after 10 hours pi, were fixed and stained for the glycoprotein marker LAMP1.Image fields displaying LAMP1 (green channel) captured from more than forty fields infected by indicated strains are shown in Fig 5C .The white arrows specify elongated SIFs which appear like thin strands of thread.The number of SIFs counted from acquired frames corresponding to each strain were calculated in percentage and are presented graphically in Fig 5D .Overall, from the data it was clear that presence of SIFs was maximally seen in cells infected by bacteria expressing SifA WT .Besides no SIFs were produced in cells in response to ΔSifA infection.Intriguingly, cells infected by bacteria expressing SifA K11R , akin to ΔSifA, showed substantially fewer SIFs compared to the wild type strains.In view of the above data, we concluded that SUMOylation of SifA functionally regulates critical aspects related to S. Tm pathogenesis.

SifA SUMOylation is necessary for Salmonella replication and virulence
Virulence of bacterial pathogens is mainly dependent on their ability to replicate and disseminate inside their host.We therefore set forth to investigate the survival and virulence potential of wild type SifA and SUMO mutant SifA expressing Salmonella strains.Both the wildtype SifA complemented ΔSifA strain (SifA WT ), and SUMOylation deficient SifA complemented ΔSifA strain (SifA K11R ) showed similar SifA expression (S5A Fig) .Furthermore, SL1344, ΔSifA, SifA WT and SifAK 11R followed similar growth kinetics (S5B Fig) .Gentamycin protection assay (GPA) was performed in HCT-8 cells infected with above mentioned four strains to investigate intracellular bacterial load.Cells at 2 hours post-infection (pi) revealed no change

PLOS PATHOGENS
in colony forming units (CFU) (Fig 6A).However, at 7 hours pi, CFU from SifA K11R infected cells was considerably lesser compared to other strains (Fig 6A).The replication of ΔSifA at this point was very high, as expected due to hyper replication of this strain owing to loss of vacuole [58].After 16 hours pi, the difference in intracellular bacterial load from cells infected by both ΔSifA and SifA K11R was significantly lower in comparison to their respective wild type controls (Fig 6A).Aside from epithelial cells, we also carried out GPA in isolated primary bone marrow derived macrophages (BMDMs) from mice.ex vivo infection of BMDMs by the above-mentioned strains for duration of 16 hours, yielded similar pattern where strain SifA K11R replicated at a slower rate in comparison to SifA WT  The pattern observed was analogous to the data from HCT-8 and macrophages.Thus, it was evident that SifA K11R exhibited a severely compromised replication in vitro in a range of different cell types.Since most of our experiments focused on K11 mutant, we were intrigued by the fact whether the same lysine is targeted by ubiquitin machinery.Multiple bands that can be seen in the blot across all the lanes may be representing the poly-ubiquitination in other sites (S5F Fig) .Notably, a unique and prominent band was seen exclusively in GFP-SifA WT suggesting lysine 11 is also targeted by ubiquitin machinery (S5F Fig).Now to answer the obvious question, whether the phenotypes associated with SifA K11R mutant is due to SUMOylation or ubiquitination, we performed a bacterial burden assay after knocking down Ubc-9, the sole E-2 enzyme of SUMOylation pathway (S5G Fig) .When Ubc-9 was knocked down, SifA WT replicated at similar rates compared to pSifA K11R suggestive of the fact that SUMOylation is required for SifA function (S5H Fig).
The attenuated replication of S. Tm bearing SifA K11R detected in vitro prompted us to examine the behavior of this strain in mice models of Salmonella infection.Streptomycin pretreated C57BL/6 female mice model was adapted to carry out all the infections.A competitive index (CI) assay was conducted [59] to assess the fitness of SUMOylation deficient SifA K11R strain in comparison to SifA WT .To actualize this assay, ΔSifA complemented with SifA WT cloned in pET28a vector (vector 1, kanamycin selection) was mixed with SifA K11R complemented in ΔSifA via pET21a vector (vector 2, ampicillin selection) in equal ratio to achieve a MOI of 5x10 7 CFU/ml.The parental Salmonella strain SL1344 used here is streptomycin resistant.The input inoculum contained equal volume of both cultures since they multiplied at same rate in LB media.Streptomycin treated mice post 24 hours, were fed with the above stated mixed cultures by oral gavage, for a total duration of 48 hours after which animals were euthanized and tissue lysates plated for CFU enumeration.The panel (Fig 6C ) indicates the categorized mice groups (N = 3) fed with the corresponding strains as shown.Colon harvested from these mice groups were assessed for their morphology, which exhibited maximum inflammatory features in group A (bearing pSifA WT + pSifA K11R ) whereas these features were minimally seen in group D (bearing Vector1+Vector2) (Fig 6D).Furthermore, the tissue homogenate from caeca, spleen, mesentric lymph nodes (MLNs), and fecal pellets were used to calculate CFUs on McConkey agar with appropriate selection markers (kanamycin for SifA WT and ampicilin for SifA K11R ).The following formula was used to calculate the Competitive Index: index of CI values from three mice is also shown in the graph.As evident from these graphs, both SifA WT and SifA K11R strains were able to colonize and multiply similarly in mice gut like other groups.However, the index values for group A varied from group D significantly in other organs checked.Furthermore, the index values were comparable between groups A and B. However, the values for groups C and D were distinct from groups A and B meaning wild type strain outcompeted K11R which replicated and disseminated identical to ΔSifA strain.To support the above data, CFU involving single infections (5x10 7 CFU of individual strain per mice) performed in same organs under similar conditions, were also added beside their corresponding CI plots (Fig 6E -6H).Differences in bacterial load were largely observed in MLNs and faecal pellet.Histopathological analysis involving Hematoxylin and Eosin (H & E) studies done in proximal colon tissue sections from mice infected by indicated strains showed severity of inflammation varied between wild type and SUMO deficient SifA (Fig 6I).The histopathology score tested was higher for SifA WT compared to SifA K11R .Finally, a survival assay was administered comprising SL1344 and ΔSifA in addition to complemented strains for evaluating their virulence in mice.Mice groups orally infected by depicted four strains (N = 4) separately at the same time were left monitored to decide the progression of infection leading to bacteremia and death in these animals.The curve in Fig 6J indicated the mice infected with complemented strain SifA K11R survived for more days (till day 10), like ΔSifA strain.However, mice groups infected by the wild type strains succumbed to infection much earlier.These data sets identified that mutant K11R of SifA bearing strain is comparably attenuated in virulence and potential to spread Salmonellosis in animals.
Hereby, we concluded that SUMOylation of SifA during infection is essential for Salmonella to thrive in the host, overcoming lysosomal attack and dissemination.A schematic of events associated during pathogenesis of Salmonella with and without SUMOylation of SifA is represented in Fig 7.

Discussion
The present study emphasizes a novel regulatory mechanism, where the effector SifA is seen to undergo SUMOylation, a host PTM mechanism, for the progression of Salmonellosis.SPI-2 effector SifA is required for dissemination of Salmonella and establishing systemic infection in mice [17,60].At the cellular level, SifA is responsible for SCV maintenance and SIF generation, two phenotypic hallmarks associated with the virulence of Salmonella [14,23].Mechanistically, the presence of SifA subverts trafficking of Rab9-M6PR mediated lysosomal hydrolases targeting SCVs, thereby protecting it from lysosomal degradation [32].To perform these numerous functions, SifA interacts directly or indirectly with multiple host adaptors including SKIP, PLEKHM1, Rab7, Rab9, LAMP1 simultaneously associating with few other SPI-2 effectors [15,26,61,62].In this given scenario, the imperative question that needs to be addressed is the factors regulating such strategic placements of these proteins within the complex network.PTM mechanisms are tools regulating almost entire cellular events, which have been shown to be targeted by pathogenic effectors to steer control of host cellular physiology.As reported earlier, SifA undergoes PTMs prenylation and S-acylation for membrane anchorage and achieve with significant variation in group A (pSifA WT + pSifA K11R ) and D (Vector control).(E-H) Relative virulence ratio calculated by the equation described in the text.CI with mean index is displayed for bacterial counts obtained from caeca, spleen, MLNs, and fecal pellet.The bar graph at the right compares the bacterial load plotted from mice individually infected by the indicated strains.(I) H and E studies showing increased intensity of inflammation in pSifA WT infected mice proximal colon sections.The histopathology of the colon was quantified by considering parameters like epithelial damage, goblet cell loss and immune cell infiltration which were scored blindly by a trained pathologist.(J) The survival curve representing no. of days lived by every infected mouse from all groups.Each drop signifies death of an animal in the group.Mice infected by ΔSifA and pSifA K11R strains showed prolonged survival owing to attenuated virulence.https://doi.org/10.1371/journal.ppat.1011686.g006destined localization in infected cells [63].Our group reported PTM SUMOylation to be drastically altered upon Salmonella infection [34].Besides, it was demonstrated that the intracellular fate of Salmonella is significantly impacted by the SUMOylation of GTPase Rab7 [13].
Further, other relating studies have shown effectors of various pathogens e.g., AmpA from Anaplasma [38] and TRP120 from E. chaffeensis [39] getting SUMOylated during infection for aiding the pathogen survival inside hosts.Hence, the strong background from all these studies and a need for convincing hypothesis to provide insights regarding functional regulation of SifA paved the foundation of this study.
The current study identified SifA undergoing SUMOylation at its 11 th lysine, and further proved how Salmonella exploited this modification for its benefit.It is well known that through its interaction with SKIP, SifA contribute to the regulation kinesin-1 mediated anterograde movement of late endosomal compartments [64].The functions of SUMOylation are well explored in processes like transcriptional regulation, DNA repair, sub cellular localization, substrate folding etc., [65][66][67][68] nonetheless their role in regulation of cytoskeletal elements is currently surfacing.Recent studies have revealed that actin gets SUMOylated for nuclear

PLOS PATHOGENS
transport [69], also alpha tubulin undergoes SUMO modification to exert control over microtubule assembly [70].One important aspect associated with control of cytoskeletal proteins is cellular trafficking.SUMOylation enable their substrate cargoes to be transported in both anterograde and retrograde directions reversibly.A study from literature concerning RNA binding La chaperone protein demonstrated that directionality of transport is dependent on SUMOylation at lysine 41 in neuronal axons [71].
Hence, we presume that SifA SUMOylation provides a scaffold to carry out various cellular trafficking that decide the fate of bacterial existence in host cells.Misrouting Rab9 conjugated MPR vesicles preventing SCV acidification, [32],utilizing PLEKHM1 Rab7 complex for mobilizing phagolysosomal membranes for SCV generation, [15,62] binding to SKIP Kinesin-1 complex for SIF extension [61] are few of the cellular transport orchestrated by SifA.SifA SUMOylation supposedly imparts control to these complex trafficking events during pathogenesis.The host SUMO ligases and deSUMOylases involved in maintenance of SifA SUMOylation and the spatiotemporal regulation of above stated events directed by SUMO are promising fields for future investigations.Nevertheless, our work indicates SUMO is integral component of functional SifA and hampering it collapses the prominent phagolysosomal alterations brought by SifA during pathogenesis.Moreover, the present work cites the first evidence of a Salmonella effector undergoing SUMOylation for sustenance inside hosts and discloses additional layers of events in host pathogen cross-talk.

Ethics statement
All experiments included in this manuscript were performed after obtaining due approvals from Institutional Bio Safety Committee (IBSC).Animal experiments were also cleared through Institutional Animal Ethics Committee (IAEC), under the approval numbers-RCB/ IAEC/2017/019 & RCB/IAEC/2021/087.

Cell culture and treatments
Human adenocarcinoma Cell lines HCT-8, were cultured in Rosewell Park Memorial Institute (RPMI) media supplemented with 14 mMNaHCO3 (Sigma), 15 mM HEPES buffer (pH 7.4), 1 mM sodium pyruvate (GIBCO), 40 mg/l penicillin (GIBCO), 90 mg/l streptomycin (GIBCO), and 10% fetal bovine serum (FBS) (GIBCO).Human colorectal adenocarcinoma cell line CaCo2 and human cervical carcinoma cell line HeLa, were grown in Dulbecco's modified Eagle's medium (DMEM) comprising14 mM NaHCO3, 15 mM HEPES buffer (pH 7.5), 40 mg/l penicillin, 90 mg/ l streptomycin, and 10% FBS.Bone Marrow Derived Macrophages (BMDMs) were prepared from the femur of C57BL/6 mice by flushing them aseptically in DMEM.Cells thus obtained were first treated with RBC lysis buffer (Sigma) and then cultured and differentiated in DMEM supplemented with10 mM HEPES, non-essential amino acids (GIBCO), 10% FBS and 20% L929 conditioned media.The pharmacological inhibitors included in this study include cycloheximide used at a concentration of 100 ug/ml till the indicated time, proteasomal inhibitor MG132 at 20 μM working concentration for a total duration of 4 hours.Vacuolar ATPase inhibitor bafilomycin A1 treatment was given at a concentration of 100 nM for total duration of 4 hours.

Bacterial Strains, Plasmids, and In vitro infection
Salmonella Typhimurium strain SL1344 was used throughout the studies.The SifA deletion knockout strain (ΔSifA) in SL1344 background together with the parent strain was kindly gifted by Beth McCormick, University of Massachusetts Medical School, MA.The coding sequence of SifA amplified from SL1344 genome were cloned in a series of prokaryotic expression vectors including pBH-HA from Roche, pET28a, pET21a, pACYC-HA and pGEX-6p1 vectors for various experiments.SUMO intolerant SifA (SifA K11R) was generated via site directed mutagenesis (SDM) in all these above vectors.Plasmids bearing wild type and mutant clones were transformed into the ΔSifA strain via electroporation for the generation of complemented strains ΔSifA/pSifA WT and ΔSifA/pSifA K11R respectively.All the mentioned bacterial strains were cultured in Luria Bertani (LB) broth at 37˚C aerobically for 8 hours serving as the primary culture.A secondary culture grown by inoculating the primary culture in 12 ml LB broth in standing white cap tubes (round bottom polypropylene dual position snap cap tubes) under stationary hypoxic conditions overnight at 37˚C were used to infect epithelial cells HCT-8, HeLa and Caco2 at a multiplicity of infection (MOI) of 1:40.BMDMs were infected similarly at a MOI of 1:10.Mammalian vector bearing SifA cloned in peGFP-C1 vector was used for mammalian over expression studies.

Cell transfection
Transfection studies were conducted in HCT-8 and HeLa cells by the standard transfection protocol employing reagents lipofectamine 2000 and lipofectamine LTX (Invitrogen, USA) respectively.Day prior to transfection, 2.5×10 5 cells were plated in 24-well plates to obtain 80% confluency.Briefly, 1μg of required plasmid and the transfectant reagent taken at a ratio of 1μg plasmid: 2 μl of reagent were separately diluted in minimum serum media Opti-MEM (Invitrogen, USA) and incubated for 5 minutes.The two mixtures were then mixed and incubated for another 15 minutes at room temperature and subsequently supplied to cells growing in Opti-MEM followed by an incubation of 24 hours, after which cells were replenished with complete media.

Yeast two hybrid assay
The bait fusion construct in vector pGBKT7 expressing GAL4 DNA binding domain (BD) and prey fusion construct in vector pGADT7 expressing GAL4 activation domain (AD) were cotransformed in yeast strain (Y187, Clonetech's matchmaker Gold Y2H system).Transformants grown on SD media plates devoid of auxotrophic markers leucine and tryptophan at 30˚C were replated on selection plate further lacking histidine and adenine to monitor bait prey interaction.

Protein Purification and In vitro SUMOylation assay
E. coli expression vector pGEX6P1 encoding SifA WT or SifA K11R were transformed in strain BL21 followed by IPTG induction at a concentration 0.05 mM at 37˚C.The fusion construct GST SifA WT or GST SifA K11R were purified by using glutathione sepharose beads (GE Healthcare) and further cleaved with thrombin for tag separation.Proteins further purified by chromatography using superdox 200 columns were subjected to SUMO conjugation reactions with the help of commercially available in vitro SUMOylation assay kit (Enzo Life Sciences).Kit components comprising SUMO machinery enzymes and ATP were added to 250 nM of wild type SifA and K11R SifA proteins separately in a 20 μl reaction mix along with provided controls at 37˚C for 1 hour as per manufacturer's instructions.The reactions were stopped and processed by adding Laemmle buffer after which it was loaded onto SDS-PAGE and immunoblotted to be probed by SUMO-1 and SUMO2 antibodies.

In-bacto SUMO Conjugation assay
This assay again included co-expressing SifA WT or SifA K11R from pGEX6P1 vector in E. coli BL21 along with plasmids encoding SUMO machinery enzymes pE1E2S1 (SUMO-1) or pE1E2S2 (SUMO-2) for the overproduction of SUMO conjugated substrates in bacteria [47].Rab7 was chosen as positive control for this assay [13].Double transformants growing on plates supplied with ampicillin and chloramphenicol were cultured in LB in presence of 0.05 mM IPTG for induction at 37˚C for 3 hours which is followed by lysates preparation and analysis by western blot.

Immunoprecipitation
For immunoprecipitation HCT-8 cells were lysed in non-denaturation immunoprecipitation lysis buffer (Pierce, Thermo Scientific) supplemented with 20 mM N-Ethyl Maleimide (NEM), (Sigma) and 2x protease inhibitor cocktail (G biosciences) for 20minutes on incubation on ice.Lysates were prepared by centrifugation at 15700 g for 10 minutes at 4˚C and collecting the supernatant.The lysates obtained were subjected to preclearing by incubation with protein G sepharose beads (GE) for 30 minutes at 4˚C on an end-to-end rotor.Precleared lysates after centrifugation were then incubated with anti-SUMO1 antibody at 4˚C on an end-to-end rotor for overnight.Next day, the incubated lysates were precipitated by addition of protein G sepharose beads for 3hours at 4˚C.Precipitations were also carried by IgG raised in same host (Merck Millipore) as the isotype control.The antigen-antibody complex obtained after washing with lysis buffer was processed in Laemmle buffer for immunoblotting.

Immunofluorescence
HeLa and HCT-8 cells were grown on cover slips embedded in 24-well plates.In assays monitoring lysosomal activity ratiometric probes like DQ-BSA, lysotracker and lysosensor were incorporated in cells prior to fixation as per manufacturer's instructions.Post infection, cellswere washed thrice in 1× phosphate buffered saline (PBS) and fixed in 4% methanol free paraformaldehyde (CST) for 15 min at room temperature (RT).Following washing, cells were permeabilized in 0.5% saponin in PBS for 10 minutes at RT after which they were blocked in blocking solution (1%BSA in PBS containing 0.1% Tween 20) for 1 hour at RT. Cells were incubated in primary antibodies against LAMP1 (Sigma), Rab9 (Invitrogen), GFP (1:200), Galectin-8 (Abcam), CI-M6PR (Abcam), SUMO-1(Invitrogen), Salmonella LPS (Abcam) prepared in blocking solution for overnight at 4˚C.Next day, cells were washed four times in PBS and proceeded for incubation in fluorophore conjugated secondary antibodies for 2 hours at RT in PBS containing 0.5% BSA and 0.1% Tween-20.Cells were finally washed three times and incubated in 4,6-diamidino-2-phenylindole (DAPI) (1 μg/ml; Sigma-Aldrich) for 5 min in dark.Mounting was done using prolong diamond antifade (Invitrogen).Images were acquired in Leica SP8 confocal microscopy under 63 × oil immersion objectives.Co-localization analysis and intensity calculations were obtained from acquired images using SP-8 LasX software and Image J.

Mass spectrometry
In vitro SUMOylated purified SifA samples were run on normal SDS PAGE gel (5% stacking, 12% resolving), stained in CBB staining solution (0.1% Coomassie Brilliant Blue R-250, 50% methanol and 10% glacial acetic acid) followed by destaining in solution (40% methanol and 10% glacial acetic acid).Samples were excised with a clean scalpel into finer slices and taken for further processing.The gel slices were further destained with treatment of 50% ammonium bicarbonate and 50mM acetonitrile solution mix.The destained gel slices are subjected to trypsin and chymotrypsin digestion in the ratio of 1:20 (1μg trypsin for 20 μg protein) and left for 16hours at 37-degree Celsius.Next day the peptides are extracted in 50% ACN and proteome grade water containing 0.1% formic acid.The peptides were concentrated in a vacuum drier, treated with ZipTip (Thermo) for salts and SDS removal and handed over for injection in mass spectrometer (ab sciex 6500).The peptides obtained were searched against uniprot ID of SifA (E8XFF1-SALT4) for specific modifications.

Proximity ligation assay
Proximity Ligation Assay (PLA) was conducted using Duo-link PLA kit (Sigma) as per manufacturer's protocol.Cells after permeabilization as described before were blocked in PLA blocking buffer followed by incubation in primary antibody against GFP (rabbit) and SUMO-1(mouse) overnight.Next day, the coverslips were washed in buffer A provided in the kit.Further, the samples were incubated with plus and minus end PLA probes supplied by the manufacturer.Hereafter the attached probes were ligated and subsequently PCR amplified for 100 minutes using the provided reagents.All the reactions were performed at 37˚C in a humidity chamber.Finally, the samples were washed in buffer B and mounted onto a slide with the supplied mountant containing DAPI for visualization.The PLA signal punctae were quantified from indicated no. of cells across each sample and plotted using the graph pad software.

Gentamycin protection assay
Infection in cells by various strains at an MOI of 1:40 in medium lacking pen strep were allowed to proceed for 1 hour at 37˚C in the incubator, followed by treatment with antibiotic Gentamycin (Gibco) at a higher concentration of 100 μg/ml in media for 1 hour to eradicate extracellular bacteria.After this step, cells were washed in sterile PBS followed by further incubation in the media containing gentamycin at a concentration of 20 μg/ml for the rest of the infection.At the end, infected cells were lysed using PBS containing 0.1% triton X-100 detergent followed by dilution plating on LB plates added with streptomycin (50 μg/ml).The plates were incubated in a 37˚C incubator and individual CFU were counted and graphically represented.

Mice infections
All in vivo studies were conducted in C57BL/6 female mice (6-8 weeks).All the animal experiments were carried out in the Small Animal Facility of Regional Centre for Biotechnology (RCB).The experiments were performed after approval by the RCB Institutional Animal Ethics Committee (approval no.RCB/IAEC/2017/019 & RCB/IAEC/2021/087).For the induction of colitis streptomycin pre-treated colitis model was employed [72].Food and water were removed 4 hours prior to infection followed by a treatment with 20 mg/Kg of streptomycin by oral gavage for the clearance of microbiota.After 24 hours, Salmonella SL1344 strain at a MOI of 5x10 7 CFU per mice was fed using oral gavage after 4 hours of withdrawal of food and water.Post 48 hours of infection, mice were euthanized and various organs including colon, spleen, MLNs, and faecal pellets were harvested.The tissues from organs were homogenized in PBS+ 0.1% Triton-X solution in Precelly's bead beater followed by dilution plating on McConkey agar plates containing streptomycin (100 μg/ml) for bacterial enumeration.For the survival assay, mice infected by the mentioned strains as described above were left monitored for death succumbing to progression of infection.A survival curve indicating the no. of days each mouse survived across all groups were plotted using graph pad prism software.

Competitive index assay
This assay involves infecting the same mice with mixed inocula comprising both wild type and mutant protein expressing strains in equal ratio [59,73].The strains ΔSifA/pSifA WT (strain1) ΔSifA/pSifA K11R (strain2) were cloned in pET28a vector (Kan resistance) and pET21a vector (Amp resistance) respectively, thus enabling a plasmid-based antibiotic marker differentiation.The input bacterial inoculum containing equal concentrations of both strain 1 and strain 2 totaling to a MOI of 5x10 7 CFU standardized by absorbance based CFU determination were fed to the C57BL/6 mice as described earlier.Homogenized tissue lysates from organs collected post 48 hours of infection were simultaneously plated on separate McConkey agar strep plates supplied with kanamycin (50 μg/ml) or ampicillin (100 μg/ml).CI was calculated as the CFU of mutant-to-wild-type ratio obtained from target sample, divided by the corresponding CFU ratio in the inoculum (input sample).

Haematoxylin and Eosin staining
Proximal colon sections were fixed in 10% formalin buffer for two days at room temperature and frozen in holders with a thick layer of cryomatrix (Thermo Scientific) and stored at -80˚C till use.Five micrometre thick sections were cut onto glass slides in a microtome and processed for haematoxylin (Sigma) and eosin (Sigma) staining.The slides and mounted using DPX mountant (Sigma) and bright field images were taken using a Nikon (NY, USA) inverted fluorescence microscope.

Statistics
The results are expressed as the mean standard error from an individual experiment done in triplicate.Data were analyzed either with one way ANOVA or by standard unpaired two-tailed Student's t test, with p values of 0.05-0.001considered statistically significant.All graphs and stats were plotted using the graph pad prism software.

Fig 1 .
Fig 1. Lysine 11 of SifA gets SUMOylated during infection.(A) Yeast two hybrid (Y2H) assay indicating interaction between SifA and SUMO1, Smt3 (yeast SUMO1 ortholog), Smt3ΔGG (Smt3 with C-terminal di glycine deleted) and human Ubc9 (E2 enzyme of SUMO pathway).(B) Schematic representation of SifA primary sequence highlighting SUMO consensus motif and SIM sites.The lysine at 11 th position within a SUMO consensus site was predicted as a potential SUMOylation site with high probability by multiple in silico tools.(C) Fractions of purified SifA WT and SifA K11R corresponding to 37 kDa obtained after infected sample as shown in S1A Fig.Further the validity of SifA SUMOylation at site 11 was validated by mass-spectrometry (MS peptide summary provided in S1 Data file), where the reaction consisting in vitro SUMOylation of SifA was subjected to mass-spectrometry.The spectra corresponding to the peptide bearing SUMOylated SifA at lysine 11 is shown in Fig 1H, also the fragmentation for same peptide is shown in inset.
, thus showing maximum recovery in MG132 treated samples (Fig 2E), and similarly c-Jun is known to undergo degradation via both the pathways [50] (Fig 2F).Next, live cell imaging was carried out in SifA WT or SifA K11R transfected HCT-8 cells, to assess the co-localization between SifA and lysosomes.Lysosomes were stained by the cell permeable fluorescent dye lysotracker red (S2C Fig).Live cell snaps acquired via confocal imaging displayed diminished GFP fluorescence (green channel) in case of GFP SifA K11R (S2C and S2D Fig) accounting to their diminished stability.Interestingly, we have also seen increased lysotracker intensity in case of GFP SifA K11R in comparison to GFP SifA WT (S2E Fig), although both proteins co-localized with lysosomes in similar fashion (S2F Fig).

Fig 2 .
Fig 2. Ectopically expressed SifA K11R protein is unstable in comparison to wild-type.(A) Cycloheximide (100μg/ml) was incorporated into SifA WT and SifA K11R transfected HCT-8 cells after 24 hours of transfection.Expression levels of both proteins are indicated till 8 hours of treatment duration (cyclo chase).(B) Decay curve indicating the stability of both proteins.The level of each protein at time 0 was set as 100% and the percentage of protein remaining at each time point from three independent experiments was calculated and plotted.(C) Blots representing the expression levels of wild-type and SUMO mutant SifA along with control proteins cyclin B and c-Jun and its (D) relative quantification on proteasomal inhibitor MG132 (20μM) treatment and on lysosomal acidification inhibitor bafilomycin A1 (100nM) treatment.Normalized expression levels of (E) cyclin B and (F) c-Jun in response to drug treatments.All statistics had been performed by students unpaired t-test.The blots shown here are representative of at least three biological replicates.https://doi.org/10.1371/journal.ppat.1011686.g002 5 hours pi) (S3B Fig).The distribution of Rab9 on SCVs from all four strains (S4A and S4B Fig) showed no change.

Fig 3 .
Fig 3. SifA-K11R infection leads to alteration in expression levels of host proteins involved in lysosomal acidification.(A) PLEKHM2 (SKIP) expression levels and their quantification in response to infection with depicted strains of Salmonella.(B) CI-M6PR expression levels in response to infection by indicated strains of S. Tm. (C) Immunoblot and its densitometric analysis representing CI-M6PR expression in transfected samples.Control refers to lysates prepared from non-transfected samples.(D) Co-IP assays showing interaction of SifA WT and SifA K11R with SKIP protein and (E) small GTPase Rab9 from transfected HCT-8 cells.The blots shown here are representative of at least three biological replicates.https://doi.org/10.1371/journal.ppat.1011686.g003

Fig 4 .
Fig 4. SUMOylation of SifA is required for preventing lysosomal acidification and eventual lysosomal clearance of Salmonella.(A) Images of HeLa cell fields infected by shown strains acquired for lysosomal activity measurement using lysosensor yellow/blue which emits bright yellow fluorescence in response to acidified environments (scale 25 micron), the images are shown in both colored (top panel) and grey scale (bottom panel).(B) Quantification and comparison of yellow fluorescence (acidic environment) emitted from lysosensor stained cells across indicated samples.(C) Images of cathepsin D localization over SCVs obtained after infection by indicated strains in HeLa cells for 4 hours.(D, E) Quantification of cathepsin D (red) signal overlaps with LAMP1 (green) and DAPI and its co-localization over SCVs.https://doi.org/10.1371/journal.ppat.1011686.g004

Fig 5 .
Fig 5. Cells infected by SifA K11R displays fewer SIFs.(A) Images from HeLa cells infected by indicated strains transformed by mCherry plasmid compared for their ability to form SCVs 1hour post infection.The integrity of SCVs was assessed by detecting the presence of the marker galectin-8 via live cell imaging (scale 8 micron).(B) Gal-8 positive rods were enumerated out of total rods in a cell and expressed as percent.(C) Image fields with cells infected with indicated strains stained by SIFs marker LAMP1 probed after 10 hours post infection (scale 50 micron).(D) Quantitative representation of percent of cells containing SIFs from samples infected by varied strains of Salmonella.https://doi.org/10.1371/journal.ppat.1011686.g005 (Fig 6B).Finally, the virulence potential of SifAWT and SifA K11R were further tested in other cell lines HeLa (S5C Fig), CaCo2 (S5D Fig) and RAW 264.7 (S5E Fig) post 16 hours pi.

Fig 6 .
Fig 6.SifA SUMOylation is necessary for implementing virulence.(A) Gentamycin protection assay results indicating bacterial burden in HCT-8 cells infected by the shown strains at 2,7-and 16-hours post infection (* denotes p value of 0.0215 and ** equals p value of 0.0018).(B) Intracellular bacterial load from isolated primary BMDMs infected by shown strains at 16 hours post infection.(C) The schematic describing the mice groups infected with the indicated strains.CI assay involves infecting the animal with mixed inoculum consisting strain 1 (WT) and strain 2 (K11R) in equal ratio.(D) The colon morphology comparison from infected animals across the groups.Graphical representation of colon length from shown groups

Table 1 . Displaying potential SUMOylatable lysine(s) within SUMO consensus sites from primary sequence of SifA as predicted by indicated software. *
Score above 0.75 are considered HIGH and where therefore selected.