SNRPD2 Is a Novel Substrate for the Ubiquitin Ligase Activity of the Salmonella Type III Secretion Effector SlrP

Simple Summary Salmonella is a genus of bacterial pathogens that can cause several diseases in humans and other animals. These bacteria can inject proteins known as effectors into animal cells through a secretion system. One of these effectors, SlrP, promotes the covalent addition of ubiquitin, a small eukaryotic protein, to specific host proteins, leading to an alteration of their stability or function. Here, we have performed a genetic screen to find new human targets of SlrP. In this way, we have identified SNRPD2, a core component of the spliceosome, the ribonucleoprotein complex that removes introns from eukaryotic pre-mRNA. SNRPD2 physically interacts with SlrP and is also a substrate of its ubiquitination activity. Lysines at positions 85 and 92 in SNRPD2 are among the residues that were ubiquitinated in the presence of SlrP. The identification of new host targets of Salmonella effectors contributes to a better understanding of the biological processes that are highjacked by these pathogens during infection, and can help in the design of future therapeutic strategies. Abstract SlrP is a protein with E3 ubiquitin ligase activity that is translocated by Salmonella enterica serovar Typhimurium into eukaryotic host cells through a type III secretion system. A yeast two-hybrid screen was performed to find new human partners for this protein. Among the interacting proteins identified by this screen was SNRPD2, a core component of the spliceosome. In vitro ubiquitination assays demonstrated that SNRPD2 is a substrate for the catalytic activity of SlrP, but not for other members of the NEL family of E3 ubiquitin ligases, SspH1 and SspH2. The lysine residues modified by this activity were identified by mass spectrometry. The identification of a new ubiquitination target for SlrP is a relevant contribution to the understanding of the role of this Salmonella effector.


Introduction
Salmonella are Gram-negative bacteria that belong to the Enterobacteriaceae family. This genus includes thousands of serovars that can infect a wide variety of animals, causing different diseases, including gastroenteritis and typhoid fever, depending on the combination of serovar/host [1]. Salmonella enterica has two type III secretion systems (T3SS) encoded by Salmonella pathogenicity island 1 (SPI1) and SPI2, respectively, that are essential for the interaction with eukaryotic host cells [2][3][4]. These systems are present in many gram-negative symbionts and pathogens of animals and plants and are used to translocate effector proteins into the host cell cytoplasm [5]. The SPI1-encoded T3SS is expressed at the beginning of infection and is involved in a cell invasion mechanism that operates through the localized reorganization of actin filaments and the formation of membrane ruffles on the surface of host cells [6]. SPI2-encoded T3SS is expressed when Salmonella is Table 1. Microbial strains and plasmids used in this study.

DNA Amplification with the Polymerase Chain Reaction and Sequencing
The amplification reactions were carried out on a T100 Thermal Cycler (Bio-Rad, Hercules, CA, USA) using Velocity DNA polymerase or MyTaq Red DNA polymerase (Bioline, London, UK) according to the supplier's instructions. Oligonucleotides are described in Table 2. Constructs were sequenced with an automated DNA sequencer (Stab Vida, Oeiras, Portugal).

Plasmid Construction
Bacterial genes were amplified using the wild-type strain of S. enterica serovar Typhimurium 14028 as template, while eukaryotic genes were amplified from a human Jurkat cDNA library. To generate the plasmids pIZ3597 and pIZ3598, sspH1 and sspH2 inserts were PCR amplified using the primers P1-pQE80-sspH1-fw and P2-pQE80-sspH1-rv, or P1-pQE80-sspH2-fw and P2-pQE80-sspH2-rv, respectively. The pQE80L plasmid was amplified using the pQE80fw and pQE80rv primers. Then, both PCR products were ligated by Gibson assembly [28]. For the rest of the constructions, a classical cloning strategy based on enzymatic digestion was used.

Yeast Two-Hybrid Methods
A human Jurkat cDNA library constructed in fusion with the activation domain of Gal4 in pGAD1318 was screened. S. cerevisiae strain L40 was sequentially transformed with pIZ1628 (pLEX10-SlrP) and the library by the lithium acetate procedure [29]. The transformants were seeded in yeast drop-out medium lacking leucine, tryptophan, and histidine. The plates were incubated at 30 • C for 3 to 8 days and then colonies were patched on the same medium and replica-plated on Whatman 40 filters placed on drop-out medium lacking leucine and tryptophan to test the β-galactosidase activity [30]. Positive clones were rescued, tested for specificity using empty pLEX10, and sequenced with primer Gal4AD.

In Vitro Ubiquitination Assays
Ubiquitination reactions were carried out in a 20-µL mixture containing buffer A (25 mM Tris-HCl, pH 7.5, 50 mM NaCl, 5 mM ATP, 10 mM MgCl 2 , 0.1 mM DTT), 2 µg of HA-tagged ubiquitin, 0.25 µg of E1 (Boston Biochem, Cambridge, MA, USA) and 1 µg of E2 (human recombinant UbcH5b from Boston Biochem) in the presence or absence of 1 µg of 6His-SlrP, GST, or GST-SNRPD2. Reactions were incubated at 37 • C for 1 h with shaking and stopped by adding an equal volume of Laemmli sample buffer containing 100 mM DTT and boiling. Some reactions were carried out with GST fusion proteins bound to glutathione-agarose beads and the beads were washed five times with NP40 buffer before boiling in Laemmli sample buffer with 100 mM DTT.

Analysis of SNRPD2 Ubiquitination Sites by MALDI-MS(/MS)
Ubiquitinated and non-ubiquitinated purified proteins were loaded on a 10% acrylamide gel and Coomassie stained with Instant Blue (Abcam, Cambridge, UK). Bands of the ubiquitinated protein with a higher molecular weight compared with those of the nonubiquitinated form were excised and analyzed in the BIO-MS mass spectrometry facility of the Universidad Pablo de Olavide.
Acrylamide plugs were destained with NH 4 HCO 3 and acetonitrile. Cysteine residues were reduced with DTT and alkylated with iodoacetamide. The protein was digested with trypsin. After digestion, the peptides were extracted with acetonitrile, acidified, and desalted in a C18 column. Mass spectra were obtained with a MALDI-TOF Ultraflextreme (Bruker, Billerica, MA, USA) in the TOF-TOF mode. The obtained fingerprint spectrum was compared against the simulated trypsin digestion of SNRPD2 sequence considering cysteine carbamidomethylation as a fixed modification and methionine oxidation and lysine ubiquitination as possible modifications. Peptides with a predicted ubiquitinated lysine were fragmented to confirm the presence of the modification.

Quantification of Protein Bands and Statistics
Fluorescent or luminescent signals from immunoblots detected with the Odyssey Fc imaging system (LI-COR) were quantified using Image Studio Lite software (LI-COR). SNRPD2 signals were corrected using β-actin as internal loading control. Means and standard deviations were calculated and one-way ANOVA with Dunnett's post-hoc test or a Student's t test were used to evaluate if the differences between conditions were significant.

Identification of Mammalian Binding Partners for Salmonella SlrP through a Yeast Two-Hybrid Screen
To find new interacting partners for the Salmonella T3SS effector SlrP, we carried out a yeast two-hybrid screen using pLEX10 as bait vector and a human cDNA library that was prepared using the vector pGAD1318. The screening was carried out in strain L40 of S. cerevisiae, which carries two reporter genes to detect the interactions: HIS3, which complements an auxotrophy, and lacZ, which codes for the enzyme β-galactosidase. A total of 3 × 10 6 clones were screened and 1400 colonies were able to grow in synthetic medium lacking histidine that was used to select for the interactions. Furthermore, 588 of these clones also showed β-galactosidase activity. The plasmids recovered from these clones were subjected to PCR using primers specific for thioredoxin cDNA, since thioredoxin is an SlrP partner identified in a previous screen. Indeed, 220 candidates were identified as cDNA encoding thioredoxin. The sequencing of some of the remaining candidates revealed that most of them corresponded to cDNA for SNRPD2 (small nuclear ribonucleoprotein D2), which plays a role in pre-mRNA processing. Finally, 30 new potential partners for SlrP were identified by a combination of DNA sequencing and PCR amplification. A clone of each candidate was reintroduced in yeasts containing pLEX10-SlrP or the empty vector to test the specificity of the interactions. A specific interaction in the two-hybrid system was observed for 14 candidates in addition to thioredoxin ( Figure 1 and Table 3).
total of 3 × 10 6 clones were screened and 1400 colonies were able to grow in synthetic medium lacking histidine that was used to select for the interactions. Furthermore, 588 of these clones also showed β-galactosidase activity. The plasmids recovered from these clones were subjected to PCR using primers specific for thioredoxin cDNA, since thioredoxin is an SlrP partner identified in a previous screen. Indeed, 220 candidates were identified as cDNA encoding thioredoxin. The sequencing of some of the remaining candidates revealed that most of them corresponded to cDNA for SNRPD2 (small nuclear ribonucleoprotein D2), which plays a role in pre-mRNA processing. Finally, 30 new potential partners for SlrP were identified by a combination of DNA sequencing and PCR amplification. A clone of each candidate was reintroduced in yeasts containing pLEX10-SlrP or the empty vector to test the specificity of the interactions. A specific interaction in the two-hybrid system was observed for 14 candidates in addition to thioredoxin ( Figure 1 and Table 3). Figure 1. Analysis of interactions of SlrP with human proteins in the yeast two-hybrid system. Derivatives of plasmid pGAD1318 expressing the indicated proteins (or C-terminal fragments of these proteins) were introduced in yeast strain L40 together with pLEX10 or pLEX10-SlrP. The interaction between SlrP and human proteins is shown by the detection of blue color in the presence of X-Gal after a β-galactosidase filter assay. Empty vectors were used as negative controls. Table 3. Candidate host partners of SlrP identified in a yeast two-hybrid screen.

mber of lones
Description of the Product Amino Acids Encoded in Different Clones 1 5 Serine hydrolase with lysine deacetylase activity 1-210 2 Acidic leucine-rich nuclear phosphoprotein 25-249 5 Centrosomal protein 120-865 Figure 1. Analysis of interactions of SlrP with human proteins in the yeast two-hybrid system. Derivatives of plasmid pGAD1318 expressing the indicated proteins (or C-terminal fragments of these proteins) were introduced in yeast strain L40 together with pLEX10 or pLEX10-SlrP. The interaction between SlrP and human proteins is shown by the detection of blue color in the presence of X-Gal after a β-galactosidase filter assay. Empty vectors were used as negative controls.

Confirmation of the Interaction of SlrP with SNRPD2
Most of the clones detected in the two-hybrid screen described in the previous section expressed the host protein SNRPD2. In addition, several independent clones encoding this protein were isolated (clones of different sizes as indicated in Table 3). Therefore, we decided to focus on studying the interaction of this protein with SlrP. Two independent approaches were used to confirm the interaction. For the first approach, purified 6His-SlrP was incubated with GST or GST-SNRPD2 bound to glutathione-agarose beads for an hour. After extensive washing with NP40 buffer, the copurification of SlrP with SNRPD2 was demonstrated by Western blotting with anti-His antibodies. Copurification was not observed with GST alone, which was used as a control (Figure 2A). For a second approach, epithelial human HeLa cells were transfected with a plasmid expressing 3HA-SNRPD2, and protein extracts from these cells were incubated with 6His-SlrP bound to Ni-NTA agarose beads. The copurification of 6His-SlrP and 3HA-SNRPD2 was shown by Western blot using anti-HA antibodies. SNRPD2 was not copurified with 6His-SseK1, an unrelated Salmonella effector that was used as a control ( Figure 2B). The copurification of 6His-SlrP with endogenous SNRPD2 was also observed ( Figure 2C). In this case, a weaker band is detected, probably due to the lower sensitivity of the anti-SNRPD2 antibody.

SNRPD2 Is a Target of the E3 Ubiquitin Ligase Activity of SlrP
The main objective of this study was the detection of new targets for the E3 ubiquitin ligase activity of SlrP. As an interacting partner of SlrP, SNRPD2 may also be a substrate of its catalytic activity. To test this hypothesis, in vitro reactions were performed mixing HA-ubiquitin, E1, E2, 6His-SlrP in the presence or absence of GST-SNRPD2. As seen in Figure 3A, intense signals that may correspond to the ubiquitinated forms of GST-SNRPD2 were observed. Less intense bands were also detected in the absence of

SNRPD2 Is a Target of the E3 Ubiquitin Ligase Activity of SlrP
The main objective of this study was the detection of new targets for the E3 ubiquitin ligase activity of SlrP. As an interacting partner of SlrP, SNRPD2 may also be a substrate of its catalytic activity. To test this hypothesis, in vitro reactions were performed mixing HA-ubiquitin, E1, E2, 6His-SlrP in the presence or absence of GST-SNRPD2. As seen in Figure 3A, intense signals that may correspond to the ubiquitinated forms of GST-SNRPD2 were observed. Less intense bands were also detected in the absence of SNRPD2. These are probably polyubiquitinated forms of ubiquitin induced by SlrP [17]. To confirm these results, new ubiquitination reactions were carried out with GST-SNRPD2 or GST bound to glutathione-agarose beads. The beads were then extensively washed before ubiquitination and analyzed by immunoblotting with anti-HA antibodies. As seen in Figure 3B, with this protocol, ubiquitinated bands were only observed in the reactions with GST-SNRPD2 but not with GST alone. The ubiquitination ladder obtained suggests that several ubiquitin adducts can be attached to SNRPD2 to generate polyubiquitinated forms of this substrate.

PEER REVIEW 10 of 19
ladder obtained suggests that several ubiquitin adducts can be attached to SNRPD2 to generate polyubiquitinated forms of this substrate.

Specificity of the Interaction and Ubiquitination of SNRPD2
SspH1 and SspH2 are two effectors of Salmonella that, together with SlrP, belong to the same family of NEL E3 ubiquitin ligases. Therefore, we decided to test whether these effectors were also able to interact and/or ubiquitinate SNRPD2. Interactions were studied using the yeast two-hybrid system and pull-down experiments. As seen in Figure 4A, only SlrP, but not SspH1 or SspH2, was able to interact with SNRPD2 in vivo in the two-hybrid system, although interaction was also observed between 6His-SspH1 and 3HA-SNRPD2 in an in vitro pull-down assay ( Figure 4B).

Specificity of the Interaction and Ubiquitination of SNRPD2
SspH1 and SspH2 are two effectors of Salmonella that, together with SlrP, belong to the same family of NEL E3 ubiquitin ligases. Therefore, we decided to test whether these effectors were also able to interact and/or ubiquitinate SNRPD2. Interactions were studied using the yeast two-hybrid system and pull-down experiments. As seen in Figure 4A, only SlrP, but not SspH1 or SspH2, was able to interact with SNRPD2 in vivo in the two-hybrid system, although interaction was also observed between 6His-SspH1 and 3HA-SNRPD2 in an in vitro pull-down assay ( Figure 4B). To study the ability of these effectors to specifically ubiquitinate SNRPD2, first we checked that the 6His-tagged forms of the three effectors were active as E3 ubiquitin ligases, as in the presence of E1 and E2 they were able to induce the polyubiquitination of ubiquitin ( Figure 5A). Importantly, the in vitro ubiquitination experiment shown in Figure 5B indicates that SNRPD2 is not a substrate for the activity of SspH1 or SspH2. This experiment also shows that the ubiquitination activity of SlrP on SNRPD2 is lost when the catalytically important residue Cys 546 is changed into Ala [17]. To study the ability of these effectors to specifically ubiquitinate SNRPD2, first we checked that the 6His-tagged forms of the three effectors were active as E3 ubiquitin ligases, as in the presence of E1 and E2 they were able to induce the polyubiquitination of ubiquitin ( Figure 5A). Importantly, the in vitro ubiquitination experiment shown in Figure 5B indicates that SNRPD2 is not a substrate for the activity of SspH1 or SspH2. This experiment also shows that the ubiquitination activity of SlrP on SNRPD2 is lost when the catalytically important residue Cys 546 is changed into Ala [17]. 022, 11, x FOR PEER REVIEW 12 of 19

Analysis of SNRPD2 Ubiquitination by Mass Spectrometry and Mutagenesis
Next, we made efforts to identify residues in SNRPD2 whose ubiquitination is catalyzed by SlrP. The untreated GST-SNRPD2 protein was compared with the same protein after the in vitro ubiquitination assay by SDS-PAGE and Coomassie staining ( Figure 6A). Some putative ubiquitinated bands were cut from gels and analyzed by mass spectrometry. As a result of this analysis, lysines at positions 85 and 92 in the SNRPD2 sequence were identified as ubiquitinated residues. These two lysines are contained in the tryptic peptide 83-92: GKKKSKPVNK. To confirm the relevance of these residues in the ubiquitination process under study, we generated point mutants where these lysines were mutated to alanines (mutants K85A and K92A) and a double mutant (K85A K92A). Furthermore, since the peptide that contains these lysines contains five lysines, we also

Analysis of SNRPD2 Ubiquitination by Mass Spectrometry and Mutagenesis
Next, we made efforts to identify residues in SNRPD2 whose ubiquitination is catalyzed by SlrP. The untreated GST-SNRPD2 protein was compared with the same protein after the in vitro ubiquitination assay by SDS-PAGE and Coomassie staining ( Figure 6A). Some putative ubiquitinated bands were cut from gels and analyzed by mass spectrometry. As a result of this analysis, lysines at positions 85 and 92 in the SNRPD2 sequence were identified as ubiquitinated residues. These two lysines are contained in the tryptic peptide 83-92: GKKKSKPVNK. To confirm the relevance of these residues in the ubiquitination process under study, we generated point mutants where these lysines were mutated to alanines (mutants K85A and K92A) and a double mutant (K85A K92A). Furthermore, since the peptide that contains these lysines contains five lysines, we also generated a deletion mutant that lacks all of these residues (SNRPD2 mutant ∆84-92). These mutant versions of SNRPD2, together with the wild-type protein, were tested as SlrP substrates in an in vitro ubiquitination assay. As seen in Figure 6B,C, a partial but significant reduction in ubiquitination was detected for the K92A mutant and the double mutant K85A K92A. The reduction was more dramatic in the deletion mutant that lacks five lysines. 022, 11, x FOR PEER REVIEW 13 of 19 generated a deletion mutant that lacks all of these residues (SNRPD2 mutant ∆84-92). These mutant versions of SNRPD2, together with the wild-type protein, were tested as SlrP substrates in an in vitro ubiquitination assay. As seen in Figure 6B,C, a partial but significant reduction in ubiquitination was detected for the K92A mutant and the double mutant K85A K92A. The reduction was more dramatic in the deletion mutant that lacks five lysines.

Lack of Effect of SlrP on SNRPD2 Levels
In many cases, ubiquitination is a signal for proteasome-dependent degradation. Therefore, we decided to investigate whether the levels of SNRPD2 were altered by the presence of SlrP. Subsequently, human HEK293T cells were transfected with different amounts of plasmid pIZ1720, a derivative of plasmid pCS2 encoding SlrP-3xFLAG, or an empty vector. As seen in Figure 7, the presence of SlrP did not change the amount of SNRPD2 detected with anti-SNRPD2 antibodies. The quantification of bands from three

Lack of Effect of SlrP on SNRPD2 Levels
In many cases, ubiquitination is a signal for proteasome-dependent degradation. Therefore, we decided to investigate whether the levels of SNRPD2 were altered by the presence of SlrP. Subsequently, human HEK293T cells were transfected with different amounts of plasmid pIZ1720, a derivative of plasmid pCS2 encoding SlrP-3xFLAG, or an empty vector. As seen in Figure 7, the presence of SlrP did not change the amount of SNRPD2 detected with anti-SNRPD2 antibodies. The quantification of bands from three independent experiments indicated that the ratio between the amount of SNRPD2 in SlrP transfected cells and vector transfected cells is not significantly different from 1 (mean 1.33, standard deviation 0.48, p = 0.15). independent experiments indicated that the ratio between the amount of SNRPD2 in SlrP transfected cells and vector transfected cells is not significantly different from 1 (mean 1.33, standard deviation 0.48, p = 0.15).

Discussion
A previous yeast two-hybrid screen carried out in our laboratory identified two binding partners for the T3SS effector SlrP that was expressed from the plasmid pGBT10 [17,18]. In order to find new interacting partners, we carried out a screen using pLEX10 as vector. Unlike pGBT10, which contains a truncated ADH1 promoter, this vector contains the full-length promoter, allowing higher expression of the bait fusion protein and improving screen sensitivity [31]. In fact, using this vector, we were able to identify 14 additional candidate partners, suggesting that many more interacting partners may be revealed by different experimental approaches.
The vast majority of clones isolated in this two-hybrid screen expressed either thioredoxin or SNRPD2. Our group had previously studied the interaction of SlrP with thioredoxin [17,27]. Therefore, we decided to focus this work on the SlrP-SNRPD2 interaction. This interaction was confirmed using two additional independent methods. SNRPD2, small nuclear ribonucleoprotein (snRNP) SmD2, is a protein of 118 amino acids that is a component of the spliceosome, a complex consisting of five snRNPs and numerous associated proteins known as splicing factors [32]. SNRPD2 is one of the 141 proteins considered core components of the human spliceosome based on their high abundance or function [33,34]. It is found in the precatalytic spliceosome B complex, the activated spliceosome C complexes, and the minor U12 spliceosome [35][36][37]. This protein plays a role in the appropriate cohesion of sister chromatids and cell proliferation [38] and in the nuclear retention of lncRNAs [39]. Its expression has also been proposed as a marker of prognosis in hepatocellular carcinoma [40,41] and as a factor that bridges mild cognitive impairment and Alzheimer's disease [42].

Discussion
A previous yeast two-hybrid screen carried out in our laboratory identified two binding partners for the T3SS effector SlrP that was expressed from the plasmid pGBT10 [17,18]. In order to find new interacting partners, we carried out a screen using pLEX10 as vector. Unlike pGBT10, which contains a truncated ADH1 promoter, this vector contains the full-length promoter, allowing higher expression of the bait fusion protein and improving screen sensitivity [31]. In fact, using this vector, we were able to identify 14 additional candidate partners, suggesting that many more interacting partners may be revealed by different experimental approaches.
The vast majority of clones isolated in this two-hybrid screen expressed either thioredoxin or SNRPD2. Our group had previously studied the interaction of SlrP with thioredoxin [17,27]. Therefore, we decided to focus this work on the SlrP-SNRPD2 interaction. This interaction was confirmed using two additional independent methods. SNRPD2, small nuclear ribonucleoprotein (snRNP) SmD2, is a protein of 118 amino acids that is a component of the spliceosome, a complex consisting of five snRNPs and numerous associated proteins known as splicing factors [32]. SNRPD2 is one of the 141 proteins considered core components of the human spliceosome based on their high abundance or function [33,34]. It is found in the precatalytic spliceosome B complex, the activated spliceosome C complexes, and the minor U12 spliceosome [35][36][37]. This protein plays a role in the appropriate cohesion of sister chromatids and cell proliferation [38] and in the nuclear retention of lncRNAs [39]. Its expression has also been proposed as a marker of prognosis in hepatocellular carcinoma [40,41] and as a factor that bridges mild cognitive impairment and Alzheimer's disease [42].
Interestingly, the two-hybrid screen also revealed the interaction of SlrP with LSM2, another core component of the U4/U6-U5 tri-snRNP complex, which is involved in the assembly of spliceosomes and is a component of the precatalytic spliceosome B complex [43]. The fact that SlrP can interact and even ubiquitinate some components of the spliceosome suggests that this is a new host process that may be targeted by S. enterica serovar Typhimurium. Interestingly, there are some precedents in this regard. In fact, infections of human primary macrophages with Salmonella or Listeria induce differential isoform usage for many genes [44], notably genes involved in immune responses. These alternative splicing changes have been suggested to be critical for regulating innate immune gene expression and controlling infection outcomes in macrophages [45]. In a previous work, infection with Salmonella or Yersinia was also shown to amplify the alternative splicing of pre-mRNA for the HLA-B27 class I major histocompatibility complex that leads to the generation of a cell-free soluble protein isoform [46]. The signal inducing this event was not identified, but appeared to be specific, since incubation with IFNγ or lipopolysaccharide did not produce the same effect. Interestingly, another member of the NEL family of E3 ubiquitin ligases, IpaH9.8 from Shigella flexneri, targets the splicing factor U2AF35 (also known as U2AF1) [47]. Another recent study that aimed to reveal the interactome of Salmonella T3SS effectors showed that SlrP may interact with various host partners involved in RNA processing [48]. The results obtained in the present work suggest that the effector SlrP may be one factor that allows Salmonella to manipulate the host splicing system.
The ubiquitination of SNRPD2 catalyzed by SlrP was characterized at the molecular level by mass spectrometry. Lysines 85 and 92 were identified as ubiquitinated and mutagenesis analysis confirmed that residue 92 was indeed a preferential target for the catalytic activity of SlrP. However, the mutation of both residues did not completely abolish ubiquitination. This is not surprising since there are examples of other E3 ligases with low lysine specificity that target a large ubiquitination zone, where several lysines can be ubiquitinated [49]. The mutation of preferred ubiquitination sites does not prevent these E3s from ubiquitinating another site on the same substrate [50]. Lysines represent 5.7% of the total human proteome [51] but SNRPD2 has higher contents with 14.4% of this amino acid. Furthermore, the two residues identified as ubiquitinated are in a 20 residue stretch with 50% lysines that may represent the preferred ubiquitination zone targeted by SlrP in this protein.
The expression of SlrP in HEK293T cells did not appear to trigger SNRPD2 degradation. Although there is the possibility that the conditions used in our experiments were not sensitive enough to detect a low level of degradation, this result suggests that ubiquitination in this case may have a non-proteolytic function. In fact, there are many examples of the involvement of mono-and polyubiquitination in functions such as protein kinase activation, DNA repair, vesicle trafficking or transcription factor activity regulation, without affecting the stability of the target proteins [52]. A plausible consequence of the ubiquitination of SNRPD2 would be the interference with a proper spliceosome assembly. SNRPD2 (also known as SmD2) is one of seven core Sm proteins (B/B', D1, D2, D3, E, F, and G). The proposed assembly of these proteins suggests a heptamer model that form a ring that interacts with the Sm site in U1, U2, U4 and U5 small nuclear RNAs (snRNAs) to form the corresponding core snRNPs. In this structure, strand β4 of the SmD2 protein pairs with strand β5 of the SmD1 protein [53]. One of the residues ubiquitinated by SlrP, Lys92, is included in the β4 strand. Another ubiquitinated residue, Lys85, is in loop L4 (residues 76-90). Interestingly, this is a not ordered region bearing several positively charged side chains that have been suggested to interact with a secondary structural element that is present in the majority of U snRNAs [54]. Together, these data suggest that the ubiquitination of SNRPD2 by SlrP at residues Lys85 and Lys92 may hinder the stability of spliceosomal snRNP particles. Additional experiments will be needed to test this hypothesis.
The genes encoding the NEL E3 ligases found in S. enterica have a complex phylogenetic distribution [55]. While slrP and sspH2 are common to most salmonellae, sspH1 has a more restricted dissemination [10] and is absent in some laboratory Typhimurium strains as well as in the better characterized strains of typhoidal Salmonella. However, the simultaneous presence of two or more effectors of this family in the same bacteria raises questions about the specificity and redundancy of the activity of E3 ligases toward particular substrates. For example, strain 14,028 of S. enterica serovar Typhimurium, used in our experiments, expresses SlrP, SspH1, and SspH2. These three proteins have similar architectures, with an N-terminus involved in T3SS-dependent translocation, a central LPX domain, and the catalytic C-terminal NEL domain [15]. The LPX domains are supposed to be necessary for physical interactions with substrates, and previous studies indicate that they are involved in the regulation of the activity of the adjacent NEL domain [56]. In addition, our previous structural studies of SlrP demonstrated that the linker region between the LPX and NEL domains plays an essential role in substrate binding [27]. The LPX domains of the three effectors contain a different number of LRR motifs: 10 for SlrP, 8 for SspH1, and 12 for SspH2 [15], and BLASTP comparisons indicate that while SspH1 and SspH2 are 62% identical, the overall sequence identity of SlrP with the other two members of the family is about 40%. The results presented here indicate that SlrP was the only one able to interact with SNRPD2 using the two-hybrid system, although some interaction was also observed for SspH1 in a pull-down experiment, suggesting a weaker in vitro interaction of SNRD2 with this effector. More importantly, SNRPD2 was specifically ubiquitinated in the presence of SlrP, but not in the presence of SspH1 or SspH2. These results support the idea that these proteins, in spite of their similar domain composition and catalytic activity, have specific host targets and might manipulate different host processes during infection.

Conclusions
In this work, we carried out a genetic screen to identify new targets for the Salmonella effector SlrP. A total of fourteen new interacting partners were identified and, among them, the splicing factor SNRPD2 was shown to be a specific substrate for the E3 ubiquitin ligase activity of SlrP.
The identification of host targets for Salmonella T3SS effectors is the first step in understanding the cellular processes that are manipulated by these virulence molecules. In fact, very few catalytic substrates have been identified for the three Salmonella members of the NEL family: PKN1 for SspH1 [57], NOD1 for SspH2 [58], and thioredoxin for SlrP [17]. The identification of SNRPD2 as a new target for SlrP ubiquitination is a relevant contribution to understanding the role of this Salmonella effector and opens new perspectives to investigate how this fascinating intracellular pathogen hijacks essential host functions such as mRNA metabolism.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.