SLC15A4 controls endolysosomal TLR7-9 responses by recruiting the innate immune adaptor TASL

Nucleic acid sensing by endolysosomal Toll-like receptors (TLRs) plays a crucial role in innate immune responses to invading pathogens. In contrast, aberrant activation of these pathways is associated with several autoimmune diseases, such as systemic lupus erythematosus (SLE). The endolysosomal solute carrier family 15 member 4 (SLC15A4) is required for TLR7, TLR8 and TLR9-induced inflammatory responses and for disease development in different SLE models. SLC15A4 has been proposed to affect TLR7-9 activation through its transport activity, as well as by assembling in an IRF5-activating signalling complex with the innate immune adaptor TASL, but the relative contribution of these different functions remains unclear. Here we show that the essential role of SLC15A4 is to recruit TASL to the endolysosomal compartment, while its transport activity is dispensable. Targeting of TASL to the endolysosomal compartment is sufficient to rescue TLR7-9-induced IRF5 activation in SLC15A4-deficient cells. In line with this, lysosomal-localized TASL restored proinflammatory cytokines and type I interferon responses in absence of SLC15A4. Our study reveals that SLC15A4 acts as a signalling scaffold and that this transport-independent function is essential to control TLR7-9-mediated inflammatory responses. These findings further support targeting the SLC15A4-TASL complex as a potential therapeutic strategy for SLE and related diseases.


SUMMARY
Nucleic acid sensing by endolysosomal Toll-like receptors (TLRs) plays a crucial role in innate immune responses to invading pathogens. In contrast, aberrant activation of these pathways is associated with several autoimmune diseases, such as systemic lupus erythematosus (SLE). The endolysosomal solute carrier family 15 member 4 (SLC15A4) is required for TLR7, TLR8 and TLR9-induced inflammatory responses and for disease development in different SLE models. SLC15A4 has been proposed to affect TLR7-9 activation through its transport activity, as well as by assembling in an IRF5-activating signalling complex with the innate immune adaptor TASL, but the relative contribution of these different functions remains unclear. Here we show that the essential role of SLC15A4 is to recruit TASL to the endolysosomal compartment, while its transport activity is dispensable. Targeting of TASL to the endolysosomal compartment is sufficient to rescue TLR7-9-induced IRF5 activation in SLC15A4-deficient cells. In line with this, lysosomal-localized TASL restored proinflammatory cytokines and type I interferon responses in absence of SLC15A4. Our study reveals that SLC15A4 acts as a signalling scaffold and that this transport-independent function is essential to control TLR7-9-mediated inflammatory responses. These findings further support targeting the SLC15A4-TASL complex as a potential therapeutic strategy for SLE and related diseases.

INTRODUCTION
Detection of invading pathogens by the innate immune system is central to mount protective responses 1 . Microbial-derived nucleic acids are recognized by both cytosolic sensors as well as endolysosomal transmembrane Toll-like receptors (TLR) 3, 7, 8 and 9 [2][3][4][5] . These innate immune pathways play a critical role to control viral and bacterial infections by inducing antimicrobial genes, triggering the production of interferons and proinflammatory cytokines and priming tailored adaptive immune responses. Conversely, aberrant activation of nucleic acid-sensing pathways is involved in a broad spectrum of pathologies, ranging from interferonopathies to autoimmune conditions such as systemic lupus erythematosus (SLE) [6][7][8] .
A central pathogenic event in SLE and closely related autoimmune diseases is the engagement of endolysosomal TLRs, in particular TLR7, by endogenous, self-derived nucleic acids, resulting in the activation of immune cells, including primarily plasmacytoid dendritic cells (pDCs) and B cells [8][9][10][11] . These cells critically contribute to the development of the disease by producing type I interferons, proinflammatory cytokines and autoantibodies.
Over the past decade, the endolysosomal solute carrier family 15 member 4 (SLC15A4, also known as PHT1) has emerged as a critical component involved in TLR7-9-induced immune responses as well as in autoimmune diseases, a role strongly supported by both human genetics and animal studies. Indeed, evidences from genome-wide association studies (GWAS) implicated SLC15A4 in SLE [12][13][14][15][16][17] . The link between SLC15A4 and endosomal TLR7-9 responses was first revealed in an in vivo ENU mutagenesis screen assessing serum levels of type I IFNs upon injection of TLR7-9 agonists, which was impaired in Slc15a4-mutant feeble animals 18 . The requirement of this solute carrier for TLR7-9 function has been further established using conventional Slc15a4 -/mice and by investigating different infections and autoimmune disease models, including chemically-and genetically-induced SLE [19][20][21][22][23][24][25][26] . These studies demonstrated that SLC15A4 deficiency impairs TLR7-9-induced responses in multiple cells types, comprising pDC and B cells, and confers significant protection to autoimmune diseases in vivo. Interestingly, beside TLR signalling, SLC15A4 has been implicated in other innate immune pathway, including NOD1-2 responses and inflammasome activation, and Slc15a4 deficiency has been shown to be protective also in DSS-induced colitis models 24,[27][28][29][30] . Altogether, these studies support pharmacological inhibition of SLC15A4 as a potential therapeutic strategy for SLE and, possibly, other autoimmune and inflammatory conditions. Despite these findings, the mechanism(s) by which SLC15A4 affects TLR7-9 responses remains less clear, and multiple explanations have been proposed. The SLC15A family comprises five members (1-5) and the best characterized, SLC15A1 (PepT1) and SLC15A2 (PepT2), act as plasma membrane proton-coupled oligopeptide transporters 31 . Similarly, SLC15A4 has been described as an endolysosomal proton-coupled transporter mediating histidine/oligopeptide translocation from the lumen to the cytosol 21,30,[32][33][34] . Based on this function, SLC15A4 deficiency has been proposed to impair TLR7-9 function by altering endolysosomal homeostasis, pH and/or histidine concentration influencing thereby TLR maturation, TLR-ligand engagement, mTORC1 activity or cellular metabolic processes 18,21,24,33,35,36 . Furthermore, it was recently advanced that SLC15A4 deficiency compromises the trafficking of TLRs and their ligands to endolysosomes leading to defects in receptor engagement and in the generation of an endolysosomal organelle required for efficient signalling 23 .
Investigating this critical mechanistic aspect, we recently showed that SLC15A4 forms a signalling complex with a previously uncharacterized protein encoded on the chromosome X by the SLE-associated gene CXorf21, which we named TASL 37,38 . Loss of TASL or mutations impairing complex formation phenocopied SLC15A4 deficiency, resulting in impaired type I IFN and proinflammatory cytokine production upon TLR7-9 stimulation 37 . Importantly, both SLC15A4 or TASL knockout specifically impaired TLR7-9-induced IRF5 activation without affecting NF-kB and MAPK pathway, strongly suggesting that TLR-ligand engagement still occurs in SLC15A4-and TASL-deficient cells and that this complex specifically affects TLR signalling downstream of this initiating event. In line with this, we identified in the C-terminal region of TASL a pLxIS motif, which in the IRF3-adaptor proteins MAVS, STING and TRIF is required for IRF3 recruitment and activation 39 . Analogously, TASL pLxIS motif was essential for IRF5 binding, phosphorylation and downstream transcriptional responses 37 . Altogether, our study revealed that SLC15A4 controls IRF5 activation by mediating the recruitment to the endolysosomal compartment of TASL, which, through its pLxIS motif, acts as a novel IRF5activating immune adaptor 37 .
Collectively, these studies raise the question of the relative contribution for endolysosomal TLR7-9 responses of the two proposed functions of SLC15A4, i.e. transporter and TASLrecruiting signalling complex. Indeed, our data showed that the transport-inactivating mutations E465K/A, previously used to demonstrate the importance of the SLC15A4 transporter function 21 , also resulted in a complete impairment of TASL binding 37 . A detailed mechanistic understanding of the role of SLC15A4 is key to further evaluate its potential as drug target for SLE and inform efforts aiming at pharmacologically interfering with its function.
Here we show that the critical role of SLC15A4 in endolysosomal TLR7-9 responses is to mediate lysosomal recruitment of TASL, and that SLC15A4-mediated transport activity is dispensable for IRF5 activation and proinflammatory responses.

Fusion of TASL to transport-deficient SLC15A4(E465A) rescues TLR7/8 responses
In order to assess the relative contribution of SLC15A4 transport activity and endolysosomal TASL recruitment for TLR-induced responses, we devised a strategy to uncouple these two functions by fusing TASL coding sequence to the cytoplasmic C-terminus of SLC15A4, either wildtype or bearing the E465A substitution ( Figure 1A). Mutations of the key transmembrane residue E465 in SLC15A4 have been shown to impair both its transport function as well as TASL binding 21,37 . Accordingly, SLC15A4 E465 mutants failed to rescue TLR7-9-induced signalling when expressed in SLC15A4-deficient cells 37 . We first verified that fusion of TASL to SLC15A4 C-terminus did not alter its trafficking to the endolysosomal compartment.
SLC15A4-TASL and SLC15A4(E465A)-TASL fusion proteins showed the expected glycosylation when stably expressed in human monocytic THP1 cells ( Figure S1A). In line with this, both fusion proteins were detected on the LAMP2-positive lysosomes in these TLR7/8signalling competent cells ( Figure S1B).
Next, we investigated whether SLC15A4-TASL and, in particular, SLC15A4(E465A)-TASL fusions retained the full spectrum of SLC15A4 and TASL activities by assessing the effect on downstream inflammatory cytokines and chemokines production. Expression of SLC15A4-TASL and SLC15A4(E465A)-TASL rescued both TNF and CCL2 production in SLC15A4deficient THP1, while SLC15A4(E465A) had no effect as expected ( Figure 1D-E). In line with this (and correlating with STAT1 activation), IFNB induction upon R848 stimulation was restored in SLC15A4 knockouts expressing either SLC15A4-TASL or SLC15A4(E465A)-TASL, but not SLC15A4(E465A) ( Figure 1F). Interestingly, both SLC15A4-TASL fusions normalized cytokine/chemokine production and IFNB induction also in TASL knockout cells, suggesting that TASL does not need to be released from the endolysosomal compartment to fulfil its function ( Figure 1D-E and 1G).
Altogether, these results strongly suggest that SLC15A4 transport activity is not essential for TLR7/8-induced IRF5 activation and downstream signalling. Rather, they support the notion that the crucial role of this solute carrier in endolysosomal TLR pathway is to act as a signalling complex mediating the recruitment of TASL to the endolysosomal compartment.

Endolysosomal targeted TASL sustains TLR7/8-induced IRF5 activation independently of SLC15A4
These findings raised the question whether endolysosomal TASL localization is in itself sufficient to mediate TLR7/8-induced IRF5 activation independently of any other possible SLC15A4 function(s). To assess this, we targeted TASL to the lysosomal compartment independently of SLC15A4 by generating fusion constructs with LAMP1 or LAMTOR1, the lysosomal-anchoring component of the Ragulator complex ( Figure 2A) 40 . We selected these two lysosomal proteins because they allow to anchor TASL to this compartment using a different mechanism than the multitransmembrane SLC15A4. In the case of LAMP1, TASL sequence was inserted after the short, 12 amino acid-long cytoplasmic sequence which follows its single transmembrane domain. In contrast, LAMTOR1 does not contain any transmembrane domains and its lysosomal localization is mediated by N-terminal lipidation (myristoylation of Gly2 and palmitoylation of Cys3 and Cys4) 40 . LAMTOR1 N terminus has been previously shown to be sufficient for lysosomal localization, we therefore generated two different TASL fusion constructs, containing the first 1-39 or 1-81 amino acids of LAMTOR1 40 .
Of note, both constructs do not contain the LAMTOR1 region required for binding to the other subunits of the Ragulator complex (LAMTOR2-5), minimizing therefore any risk of interfering with its functions [40][41][42] . Upon stable expression in THP1 cells, LAMP1-TASL and LAMTOR1-TASL fusions were targeted to the lysosomal compartment ( Figure 2B-C, Figure S2A). We have previously shown that the first, N-terminal amino acids of TASL are required for SLC15A4 binding and that TASL N-terminal tagging impaired complex formation 37 . Therefore, we first investigated whether fusion of the lysosomal targeting proteins to TASL N-terminus would affect SLC15A4 binding. Indeed, TASL fusions failed to co-immunoprecipitate SLC15A4 when co-expressed in HEK293T cells ( Figure S2B). In line with this, immunoprecipitation of TASL fusion proteins stably expressed in THP1 cells did not recover endogenous SLC15A4 ( Figure   S2C). Altogether, these results indicate that LAMTOR1-and LAMP1-TASL localized to the lysosomal compartment independently of SLC15A4.
Next, we stably expressed these fusion constructs in SLC15A4-and TASL-deficient THP1 cells, along with SLC15A4-TASL and the respective wildtype controls ( Figure 2D).
Remarkably, the two LAMTOR1-TASL constructs as well as LAMP1-TASL fully normalized R848-induced TNF production in both knockout cell lines ( Figure 2E). CCL2 production was equally restored ( Figure 2F). Supporting these data, expression of lysosomal-localized TASL in SLC15A4-deficient cells efficiently rescued IRF5 activation, monitored by its phosphorylation and dimerization, as well as STAT1 phosphorylation ( Figure 2G, Figure S2D).
Consistently, LAMTOR1-and LAMP1-TASL efficiently complemented TLR7/8 signalling also in TASL knockout cells ( Figure 2H). Of note, compared to control sgRen cells, lysosomal TASL fusions, and especially LAMTOR1(1:39)-TASL, showed a mild increase in R848-induced IRF5 activation. This may be due to higher expression levels compared to endogenous TASL levels or, possibly, by its stable lysosomal association. To complement these data obtained upon stable expression of TASL fusion constructs, we generated SLC15A4-deficient cells inducibly expressing LAMTOR1(1:39)-TASL upon doxycycline treatment, allowing to transiently express this construct to levels comparable with endogenous TASL protein ( Figure S2E).

LAMTOR1
(1:39)-TASL restored IRF5 activation also in these conditions, indicating that lysosomal targeting of TASL in absence of SLC15A4 can achieve comparable activity as the endogenous protein in wildtype cells ( Figure S2E). Altogether, these results reveal that the key function of SLC15A4 essential for TLR7/8-induced responses is its ability to recruit TASL to lysosomes, as lysosomal localization of TASL is sufficient to fully restore IRF5 activation and downstream cytokine/chemokine production in SLC15A4-deficient cells. Therefore, SLC15A4-mediated transport and/or other direct or indirect metabolic effects appear to be dispensable for TLR7/8-induced IRF5-dependent responses.

Endolysosomal TASL restores TLR7-9 responses in SLC15A4-deficient pDC and B cells
To further confirm these findings, assess possible cell-type specific effects, and explore TLR9induced responses, we next investigated the human plasmacytoid dendritic cell line CAL-1 43 .
Plasmacytoid DCs are major producers of type I IFN production upon endolysosomal TLR stimulation by microbial or endogenous nucleic acids and play a central role in SLE pathogenesis 8,44 . Moreover, CAL-1 cells express TLR9 and we previously showed that stimulation with its ligand CpG triggers SLC15A4-and TASL-dependent IRF5 activation, allowing therefore to extend our investigation beyond TLR7/8 37 . Consistently with the data obtained in THP1 monocytes, stable expression of LAMTOR1-, LAMP1-and SLC15A4-TASL fusion constructs restored R848-induced IRF5 activation in both SLC15A4-and TASLdeficient CAL-1 cells ( Figure 3A-C, Figure S3A-B). Importantly, the capacity of lysosomaltargeted TASL to rescue IRF5 activation in absence of SLC15A4 was not specific to TLR7/8, but also observed upon TLR9 stimulation. Indeed, the impaired IRF5 activation observed in SLC15A4-or TASL-knockout CAL-1 upon stimulation with TLR9 agonist CpG was largely restored by expression of lysosomal TASL fusion proteins ( Figure 3D-E, Figure S3C-D).
Similar effects were observed when monitoring STAT1 activation, while, as expected, MAPK pathway proceeded independently of SLC15A4-TASL complex, confirming unaltered TLRligand engagement. Mirroring IRF5 activation, TNF and IL6 production upon R848 and CpG was efficiently rescued by lysosomal TASL in both CAL-1 knockout lines ( Figure 3F-G).
Next, we took advantage of the fact that lysosomal TASL fusions allow to uncouple localization and signalling activity to identify domains in TASL specifically required for IRF5 activation. For this we expressed a series of TASL deletion constructs bearing the lysosomal LAMTOR1(1:39)-targeting motif in SLC15A4-deficient cells ( Figure 3H-I, Figure S3E).
Interestingly, while lysosomal TASL deleted of amino acids 1-105 retained full activity, further removal of the following evolutionary conserved region (aa 106-144) resulted in a complete loss of IRF5 phosphorylation, suggesting a third functional motif in TASL in addition to the Nterminal lysosomal targeting region and C-terminal IRF5-recruiting pLxIS motif. Of note, stable expression of wildtype TASL in SLC15A4 knockout cells was unable to restore IRF5 activation, further confirming that lysosomal anchoring, and not increased TASL expression levels, is crucial for restoring signalling upon loss of this solute carrier ( Figure S3F).
Besides pDCs, B cells are key contributors to autoimmune disease pathogenesis 9,45 . In this context, accumulating evidences support a central role of cell-intrinsic endolysosomal TLR signalling in promoting the main processes by which B cells contribute to autoimmune diseases, including production of autoantibodies, antigen presentation to T cells and production of cytokines 9,11 . Slc15a4-deficiency in B cells affects their function and confers protection in murine SLE models 21,25,26 , but the role of this solute carrier in human B cells has not been yet investigated. Moreover, the function of TASL has not been explored in neither murine nor human B cells. We therefore assessed the involvement of the SLC15A4-TASL complex for endolysosomal TLR responses and IRF5 activation in EBV-immortalized human B cell lines. After verifying that R848 stimulation triggered IRF5 activation in two independent lines, we generated SLC15A4 and TASL knockout ( Figure S4A, S4C). In line with previous observations in THP1 and CAL-1 cells ( Figure 1B, 3A) 37 , deletion of SLC15A4 resulted in a concomitant reduction in TASL protein levels, suggesting functional complex formation also in these cells ( Figure S4A, S4C). Knockout of either SLC15A4 or TASL strongly impaired IRF5 phosphorylation and dimerization in both lines, with the reduction in IRF5 activation correlating with the knockout efficiency of the different sgRNAs observed in these cell populations ( Figure   4A-B, Figure S4B, S4D-E). These data demonstrate that the SLC15A4-TASL complex is essential for endolysosomal TLR-induced IRF5 activation also in human B cells, extending therefore its role beyond innate immune monocytes and pDCs, and demonstrating its central

DISCUSSION
In this study, we show that the crucial role of SLC15A4 in controlling TLR7-9-induced IRF5 activation and cytokine production is to recruit TASL to the endolysosomal compartment, while its substrate transport activity is not required per se.
How SLC15A4 controls endolysosomal TLR7-9 responses remained unclear and different mechanistic explanations, not all necessarily mutually exclusive, have been proposed. Early studies support a model in which loss of SLC15A4 proton-coupled histidine/oligopeptide transport activity results in the accumulation of its substrates in the endolysosomal lumen, altering thereby pH and/or histidine levels 18,21,24,36 . This in turn would impact on TLR maturation and/or ligand-receptor engagement. Altered intralysosomal environment and consequent impaired mTORC1 activation have been also proposed to influence TLR responses by affecting a feedforward loop mediating IRF7 upregulation 21 . Finally, other studies linked SLC15A4-deficiency to perturbed trafficking and colocalization of TLRs with their ligand, resulting in impaired receptor-ligand engagement 23 , or to metabolic perturbations affecting the TCA cycle, autophagy and/or mitochondrial integrity 33,35 . The finding that lysosomal localization of TASL is sufficient to rescue TLR7-9 responses in SLC15A4-deficient cells demonstrates that translocation or altered concentrations of potential substrates of SLC15A4 is not critically involved in TLR pathway regulation, at least in terms of IRF5 activation and production of proinflammatory cytokines and Type I IFNs. Importantly, this does not exclude the possibility that SLC15A4 transport activity may indirectly affect the TLR7-9 pathway by controlling TASL recruitment. Indeed, it is conceivable that transport-dependent conformational changes in SLC15A4 could impact its ability to bind TASL, and therefore indirectly regulate IRF5 activation and downstream signalling. Whether the ability of SLC15A4 to recruit TASL is dependent on its conformation and/or its localization along the endolysosomal system is an intriguing question to be addressed in future studies.
Concerning the function of SLC15A4 as transporter, it should be noted that this solute carrier is expressed broadly, including in cell types that do not express TLR7-9, TASL nor IRF5 37 .
This strongly suggests that SLC15A4 is involved in other cellular processes beside its specific TASL-recruiting function in endolysosomal TLR signalling. This is consistent with its reported function in regulating cellular metabolism, mTORC1 activation, NOD-ligand transport and mast cell responses 24,[27][28][29][30]35,46 .
The involvement of endosomal TLR responses in SLE and related diseases is strongly supported by both animal studies and human genetics. Notably, the three components of the signalling axis we described, SLC15A4, CXorf21/TASL and IRF5, have all been identified in GWAS on human SLE, with IRF5 being one of the best characterized and strongly associated factors 12,13,47,48 . In line with this, IRF5-deficient mice show strong protection in a broad range of SLE disease models [49][50][51][52][53][54] . These evidences and the fact that solute carriers are an eminently druggable class of proteins, have put forward SLC15A4 as an attractive drug target for SLE and related diseases [55][56][57] . Further supporting this notion, here we show that the SLC15A4-TASL complex is essential for IRF5 activation not only in monocytes and pDCs, but also in human B cells, demonstrating therefore its general requirement in all the endolysosomal TLRresponding cells shown to be critically involved in SLE pathogenesis. The data presented strongly suggest that future efforts to pharmacologically target SLC15A4 should aim at interfering with SLC15A4-TASL complex function, either by direct inhibition of its assembly or, possibly, indirectly by interfering with the trafficking of the complex to the endolysosomal compartment. In contrast, our data imply that inhibition of SLC15A4 transport activity may not be in itself sufficient to inhibit endolysosomal TLR responses if this does not concomitantly result in interfering with complex formation or localization.          sgRNA sequences targeting SLC15A4 (sgSLC15A4-1 and sgSLC15A4-2), TASL (sgTASL-1 and sgTASL-2) or a non-targeting, control sgRNA sequence designed against Renilla (sgRen) previously described (Table S1) 37 .

Generation of stable knockout and overexpressing cell lines by lentiviral transduction
Lentiviral transduction was performed as previously described 37 . Briefly, HEK293T cells were transfected with sgRNA-or cDNA-encoding lentiviral vectors and packaging plasmids psPAX2 and pMD2.G (plasmid no. 12260 and plasmid no. 12259 from Addgene) using PEI (Sigma).
The medium was changed with fresh RPMI, supplemented with 10% (v/v) fetal bovine serum (FBS) and antibiotics (100 U/ml penicillin, 100 μg/ml streptomycin) 6h post transfection. After Prior to transfer, Phos-tag SDS-PAGE gels were incubated (2 times 10 min) with transfer buffer supplemented 10 mM EDTA and then washed 10 min in transfer buffer without EDTA.
For IRF5 Native-PAGE, 10ug of lysate was separated by 7.5% polyacrylamide gel without SDS. Before transfer, gels were soaked in running buffer with 0.01% SDS for 30 minutes at room temperature. After transfer, membranes were blocked by 5% non-fat dry milk in TBST and probed with indicated antibodies. In experiments in which multiple antibodies were used, equal amounts of samples were loaded on multiple SDS-PAGE gels and western blots sequentially probed with a maximum of two antibodies.

Co-immunoprecipitation
For immunoprecipitation assays with tagged proteins, cells were lysed in E1A buffer. The appropriate amount of whole-cell lysate was used as input, and the remaining was subjected to immunoprecipitation using equilibrated CaptureSelect C-tagXL Affinity matrix beads (Thermo Fisher Scientific) overnight at 4 °C on a rotating wheel. Beads were washed three times with E1A buffer and eluted with SDS loading buffer. After quantification of the whole-cell lysate and immunoprecipitated proteins were analysed by SDS-PAGE and immunoblotting.

PNGase F treatment
Statistical analyses and graphs were made using GraphPad Prism 9 software (GraphPad).
The number of experiments or biological replicates (n) used for the statistical evaluation of each experiment is indicated in the corresponding figure legends. The data are plotted as a mean ± SD as indicated.