SLC46 Family Transporters Facilitate Cytosolic Innate Immune Recognition of Monomeric Peptidoglycans

Fragments of peptidoglycan (PGN) are strong activators of innate immune responses in both mammals and insects. In mammals, γ-D-Glu-meso-DAP, a dipeptide derivative from DAP-type PGN, is the minimal activator of NOD1. In contrast, muramyl dipeptide (MDP), a monosaccharide dipeptide derived from either Lysine-type or DAP-type PGNs, activates NOD2. Both these nucleotide-binding oligomerization domain (NOD) receptors trigger NF-κB and MAPK signaling pathways, driving production of proinflammatory cytokines and chemokines (1). In addition, NLRP1 and NLRP3 form inflammasome complexes in response to MDP (2, 3). Tracheal cytotoxin (TCT) is a monomer of DAP-type PGN released by Bordetella pertussis and Neisseria gonorrhoeae, and is known to cause massive cell death in ciliated epithelia in mammals as well as to activate Drosophila innate immune responses (4, 5).

In Drosophila, the PGN recognition proteins directly recognize PGN and trigger either of the two major Drosophila immune response pathways, Toll and Imd, culminating in robust induction of antimicrobial peptides (6–8). Among the 13 PGN recognition proteins in Drosophila, PGRP-LC and PGRP-LE bind specifically to DAP-type PGN and trigger the Imd pathway. PGRP-LC is a transmembrane protein that recognizes both polymeric DAP-type PGN and monomeric TCT in the extracellular milieu, whereas PGRP-LE senses TCT in the cytosol (9). However, the molecular mechanisms by which TCT accesses the cytosol to be sensed by PGRP-LE have not been explored.

In this study, we examine the delivery of TCT to the cytosolic PGN receptors PGRP-LE in Drosophila and NOD1 in mammals. Drosophila SLC15 homologs, which have been previously associated with the transport of Tri-DAP (a NOD1 agonist) and MDP (a NOD2 agonist), did not facilitate TCT recognition in Drosophila. In contrast, a targeted RNA interference (RNAi) screen identified CG8046, a Drosophila SLC46, as a novel transporter facilitating cytosolic recognition of TCT. We further show that mammalian SLC46As also facilitate TCT- and MDP-triggered NOD receptor activation in human epithelial cells.

Drosophila S2* cells were maintained in Schneider’s Drosophila media (Life Technologies) supplemented with 10% FBS, 1% GlutaMAX (Life Technologies) at 27°C. HEK293 cells were maintained in DMEM (Corning CellGro) supplemented with 10% heat-inactivated FBS at 37°C, 5% CO₂. HCT-116 cells (American Type Culture Collection) were maintained in McCoy’s 5A (Iwakata & Grace Modification; Corning CellGro) supplemented with 10% heat-inactivated FBS at 37°C, 5% CO₂.

HEK293 cells were seeded in DMEM (Corning CellGro) supplemented with 10% heat-inactivated FBS on Nunc Glass Bottom Dishes (Thermo Fisher) coated with Poly-L-Lysine (Sigma) for 24 h at 37°C (50,000 cells per plate). Cells were transfected with pEF-Slc46a2-EGFP-V5 construct using GeneJuice Transfection Reagent (Millipore) and OptiMEM media (Life Technologies), per the manufacturer’s instructions. Transfected cells were kept at 37°C for 24 h. To label the late endosomes and lysosomes, 20 μl of 10 μM LysoTracker (Thermo Fisher) was added to each dish and cells were incubated at 37°C for 30 min in the presence or absence of 10 μM TCT. After incubation, the medium was removed and cells were washed twice with HBSS (Corning CellGro) and immersed in HBSS for imaging. Imaging was performed using a Leica TCS SP8. To measure the perimeter of the SLC46A2-EGFP–expressing, LysoTracker-positive puncta, Z-stacks of optical sections spanning each individual cell were acquired, and the focal point with the greatest surface area per puncta was measured. Each representative image was generated with three to five optical sections that were reconstructed into three-dimensional images and flattened into two dimesions. All images were acquired on the same day, using the same microscope settings and exposure times. FIJI/ImageJ and Adobe Photoshop CS6 were used to process and format images.

For quantitative real-time PCR (qRT-PCR), 1 × 10⁶ S2* cells were plated in 1 ml media and transfected with 2 μg of dsRNA using the calcium phosphate method the following day. For Western blotting, 1 × 10⁷ S2* cells were transfected with 20 μg of dsRNA. In both cases, cells were split into 1:3 in fresh medium 48 h after transfection and treated with 1 μM 20-hydroxyecdysone for 24 h before stimulation. All the RNAi target sequences were retrieved from Harvard DRSC/TRiP Functional Genomics Resources (www.flyrnai.org). For RNAi in human epithelial cell lines, 20,000 cells were cotransfected with 5 pmole of ON-TARGETplus Human NOD1 small interfering RNA (siRNA), SMARTpool or ON-TARGETplus Non-Targeting Pool (Dharmacon) and plasmids in a 96-well plate using Lipofectamine 2000 (Invitrogen).

Total RNA was isolated with Trizol (Invitrogen). Then 0.5 μg of total RNA was treated with amplification-grade DNaseI (Invitrogen) and used for cDNA synthesis (iScript cDNA synthesis kit; Bio-Rad). SYBR Green Supermix (Bio-Rad) was used for real-time PCR using diluted first strand cDNA. PCR primer sequences are as follows: GAPDH1_fw; 5′-ATCGTCGAGGGTCTGATGAC-3′, GAPDH1_rev; 5′-CGGACGGTAAGATCCACAAC-3′, Dpt_fw; 5′-TAGGTGCTTCCCACTTTCCA-3′, Dpt_rev; 5′-CATTGCCGTCGCCTTACTT-3′, yin_fw; 5′-AATGAGTTCTGCGAGCGATT-3′, yin_rev; 5′-TCCCGATCGCAATTAGTAGG-3′, CG2930_fw; 5′-CCAGCGAGTTCTTCCTGTTC-3′, CG2930_rev; 5′-CTTGCCCTTGTTCTCGACTC-3′, CG9444_fw; 5′-GGCAATCTGATCGTGGTTCT-3′, CG9444_rev; 5′-TGTAATCGAAGGCCAAAAGG-3′, CG8046_fw; 5′-CTGTGCCATGTACACCCAAG-3′, CG8046_rev; 5′-AGCCACGGAATAGGTCACAC-3′. All the experiments were repeated at least twice.

For Attacin A-luciferase assay, 1 × 10⁵ S2* cells were plated in 96-well plates and transfected with 50 ng of Attacin A-luciferase (10), 5 ng of pCopia-Renilla luciferase, and 50 ng of pAc5.1 plasmid using Effectene (Qiagen) for 48 h. For the NF-κB luciferase assay, 20,000 cells (HEK293 or HCT-116) were plated in 96-well plates and transfected with 50 ng NF-κB luciferase, 5 ng of pRL-TK (Promega), and 5 ng of pEF plasmid using GeneJuice (Millipore) for 24 h. In all cases, cells were stimulated with indicated ligands for 6 h and subject to dual-luciferase assay. All transfections were performed in triplicate and relevant firefly luciferase activity was normalized to Renilla luciferase activity. All the experiments were repeated at least twice.

All statistical analyses were performed using GraphPad Prism. Two-way ANOVA followed by Tukey multiple comparison or unpaired two-tailed t test was used as indicated. SD was presented as an estimate of variation. A p value <0.05 was considered significant. For survival experiments, the log-rank test was used for statistical analysis.

For the Imd protein cleavage experiment, 50 μg of total protein from whole lysate was separated by SDS-PAGE (10% acrylamide gel) and immunoblotted with anti-Imd Ab (11). After probing with anti-Imd Ab, the blot was stripped and reprobed with chicken anti-GFP Ab (ab13970; Abcam) to detect YFP-PGRP-LE.

To generate deletion mutants for yin, pBac{WH}roX1[f07388] (Exelixis Collection at Harvard University), P{XP}yin[d02176] (Bloomington Drosophila Stock Center, No. 19172), and hs-FLP (Bloomington Drosophila Stock Center, No. 279) were used to induce FRT/FLP-mediated recombination as previously described (12). Male progenies were screened for the loss of the mini-white marker resulting from recombination between two transposons and three positive hits were recovered and balanced. All mutants were homozygous-viable and fertile. Genomic DNA PCR confirmed resulting hybrid elements.

Deletion mutants for CG8046 were generated by homology-directed repair combined with CRISPR-Cas9 method. Two guide sequences targeting both 5′ upstream and 3′ downstream regions of CG8046 coding DNA sequence were selected and cloned into a pCFD4 vector (Plasmid #49411; Addgene) (5′-AACTCAACTGAGTCTTGAAA-3′ for 5′-GGTTTTTAAATGATTTATGG-3′). pCFD4 harboring two guide RNA sequences and pHD-DsRed plasmid (Plasmid #51434; Addgene) with homology arms flanking 3XP3 DsRed transgene were coinjected into w¹¹¹⁸; PBac{y[+mDint2]=vas-Cas9}VK00027 strain (Bloomington Drosophila Stock Center, No. 51324). Injected animals were crossed to w¹¹¹⁸ strain and male progeny were screened for eye-specific DsRed expression. Over 20 DsRed-positive hits were recovered and balanced. Genomic DNA PCR and RT-PCR further confirmed deletion of CG8046 locus.

For in vivo overexpression assays, a PGRP-LC knockdown strain was generated first with a transgene for PGRP-LC RNAi (Vienna Drosophila Resources Center, No. 51968) and the C564 Gal4 driver in a single stock (13). This line failed to induce robust Diptericin (Dpt) induction upon challenge with Escherichia coli, as expected. EP elements driving expression of CG8046 (Bloomington Drosophila Stock Center, No. 27107) or CG30344 (Bloomington Drosophila Stock Center, No. 28425) were crossed to PGRP-LC knockdown strain, and resulting progenies were analyzed for Dpt induction by qRT-PCR following TCT injection (9).

E. coli 1106 was grown overnight in Luria-Bertani broth at 37°C to an OD of 2.0 and was pelleted. Flies were pricked with a needle inoculated with E. coli 1106 or clean pricked (as a control). Infections were administered between the haltere and wing joint in the metathorax of flies aged 3–7 d. Flies were kept at 25°C over the course of the experiment. Survival was measured every 24 h. Seven independent survival experiments were performed and one representative result is presented in this report.

All SLC46 and SLC15 expression constructs are C-terminally fused to the V5 tag. To check their expression, transfected cells were treated with trypsin/EDTA to obtain a single-cell suspension. After fixation and permeabilization using FoxP3 Staining Buffer Set (eBioscience), V5 tag was stained with anti–V5-FITC or anti–V5–Alexa Fluor 647 (Invitrogen) in 1× permeabilization buffer. LSR II (BD Biosciences) and FlowJo (Tree Star) were used for flow cytometry and analysis. Dead cells were excluded using the LIVE/DEAD fixable aqua dead cell stain kit (Invitrogen). Live, singlet cells were gated for the detection of the V5 tag.

Sequences were retrieved from Ensembl and aligned with Clustal Ω version 1.2.0 (14). Phylogenetic trees were reconstructed using the neighbor-joining method and BLOSUM 50 matrix in PFAAT version 2 (15). To assess the reliability of internal nodes, the bootstrap method was performed for 1000 replicates and displayed as a percentage on each node. We confirmed that these neighbor-joining trees had the same topologies as the corresponding gene trees in Ensembl, which are based on a maximum likelihood method.

Earlier studies have suggested that members of the SLC15 family facilitate translocation of NOD1 and NOD2 agonists into the epithelial cells as well as in dendritic cells (16–20). Interestingly, Yin, a Drosophila SLC15 identified in Drosophila S2* cell phagosomes, was reported to enhance MDP recognition in HEK293 cells (21). Therefore, we first generated a deletion allele of yin to determine if Yin is required for TCT-triggered activation of the cytosolic PGN receptor PGRP-LE. However, TCT still triggered robust systemic Dpt induction in PGRP-LC, yin double mutants (Supplemental Fig. 1A–D). Note that PGRP-LC was eliminated in this experimental design to force all recognition through the cytosolic PGRP-LE–dependent pathway.

To facilitate further studies, we established an S2* cell-based assay. Because PGRP-LE expression is very low in S2* cells, Dpt induction in response to TCT is completely PGRP-LC dependent (Fig. 1A) (5). To ask if S2* cells have TCT transporter activity, S2* cells were transfected with an Attacin A (AttA)-luciferase reporter and expression plasmids for different PGRP-LE alleles, including wild type and three point mutants. These mutants were previously shown to interrupt TCT binding and/or TCT-induced multimerization (22). Wild type PGRP-LE significantly promoted AttA-luciferase activity in response to TCT, but the mutant alleles failed to do so (Fig. 1B). These data indicate that S2* cells can transport TCT across the plasma membrane for cytosolic recognition by PGRP-LE.

Based on this result, a stable S2* cell line expressing YFP-tagged PGRP-LE (PGRP-LE stable cells hereafter) was established. When PGRP-LC was knocked down, PGRP-LE stable cells failed to induce Dpt upon polymeric PGN stimulation, but continued to respond to TCT, albeit somewhat reduced, in an imd- and Relish-dependent manner (Fig. 1C). Together these data demonstrate that S2* cells are capable of transporting TCT to the cytosol for detection by PGRP-LE, and we reasoned that double knockdown of PGRP-LC and potential TCT transporters should make PGRP-LE cells less responsive to TCT stimulation.

As earlier studies suggested SLC15s function as Tri-DAP and MDP transporters, all three Drosophila SLC15s were tested for their potential role in TCT transport in this cell-based assay (Supplemental Fig. 1E). Although all three Drosophila SLC15s were readily detectable and subject to efficient knockdown by RNAi, these transporters were not individually or jointly required for TCT activation of the cytosolic PGRP-LE pathway (Supplemental Fig. 1F–O). Furthermore, AttA-luciferase assay revealed that transient expression of Drosophila SLC15s did not enhance TCT-stimulated reporter activity (Supplemental Fig. 1P–R). Therefore, we concluded that Drosophila SLC15s are not associated with cytosolic recognition of TCT.

As the phagosome is the site of microbial degradation, PGN transporters are likely to be localized to this compartment. Therefore, we next focused on candidate transporters associated with the S2* cell phagosome (Fig. 2A) (21). In particular, SLCs, rather than ABC transporters, are typically linked to solute influx in eukaryotic cells (23, 24). PGRP-LC as well as each of five additional SLC transporters were silenced in PGRP-LE stable cells. RNAi of either JhI-21 or CG8046 caused a ∼10-fold decrease in Dpt induction in response to TCT, suggesting that these two transporters may be associated with the TCT/PGRP-LE pathway (Fig. 2B). To determine if the decrease in Dpt induction was specific to the TCT/PGRP-LE pathway, JhI-21 or CG8046 were knocked down, although PGRP-LC was left intact. In this case, JhI-21 RNAi still significantly impaired Dpt induction in response to either TCT or polymeric PGN, whereas CG8046 RNAi did not (Fig. 2C). These results argue that CG8046 functions specifically in the cytosolic TCT/PGRP-LE pathway, whereas JhI-21 has a more general role in regulating the Imd pathway or S2* cell physiology. To validate this finding, an independent dsRNA targeting a different region of CG8046 was synthesized. Again, double knockdown of CG8046 and PGRP-LC severely blocked Dpt induction (∼12-fold decrease) upon TCT stimulation, whereas CG8046 single knockdown had no effect (Fig. 2D). Knockdown efficiency of CG8046 was >60%, and expression of CG8046 was immune-inducible (Fig. 2E).

The most receptor-proximal signaling event characterized in the Imd pathway is the rapid proteolytic cleavage of Imd by the Drosophila Caspase-8 homolog Dredd (25, 26). In PGRP-LE stable cells treated with either LacZ or PGRP-LC RNAi, TCT induced robust Imd cleavage as expected. However, when both PGRP-LC and CG8046 were depleted, TCT-induced cleavage of Imd was lost (Fig. 2F, top). CG8046 RNAi did not change the YFP-PGRP-LE protein level (Fig. 2F, bottom). These results demonstrate that CG8046 participates in the activation of the Imd pathway upstream of receptor-associated Imd cleavage, further supporting the model that CG8046 functions prior to the interaction of TCT with PGRP-LE as expected for a TCT transporter. Altogether, these results imply that CG8046 is a TCT transporter in Drosophila.

Phylogenetic analyses indicated that CG8046 is one of eight Drosophila homologs of SLC46A transporters (Fig. 3A). To further analyze the activity of CG8046 in comparison with other paralogs, the AttA-luciferase assay was performed in PGRP-LE stable cells transiently expressing each of eight Drosophila SLC46 paralogs. Expression of CG8046 had the most robust effect on TCT-stimulated AttA reporter activity, with a ∼2.3-fold increase (Fig. 3B, empty vector versus CG8046, 1 μM TCT, p < 0.0001). With the exception of CG15553, expression of CG8046 and other SLC46s was robust in these S2* transfection assays (Supplemental Fig. 2A, 2B), suggesting that among these seven SLC46s, only CG8046 has the ability to deliver TCT to the cytosol. However, CG15553 expression was very weak, and CG15553 is not naturally expressed in S2* cells. Therefore, the ability of CG15553 to transport TCT is unknown.

Next, the role of CG8046 in recognition of TCT was investigated in vivo. CG8046 was overexpressed in the fat body while simultaneously silencing PGRP-LC with a hairpin RNA, using the C564 Gal4 driver. The resulting flies were challenged by TCT injection. Dpt expression was significantly higher in animals with overexpressed CG8046 compared with the control animals at 6 h after injection (Fig. 3C, 3D). For comparison, CG30344, another SLC46 paralog that showed no activity when expressed in S2* cells, was similarly overexpressed in flies. In this case, Dpt induction was lower following TCT challenge. These results confirm that CG8046 is sufficient to promote TCT recognition by PGRP-LE in adult flies as well as S2* cells.

To further characterize the role of CG8046 in vivo, we generated a null allele of CG8046 by CRISPR/Cas9. Upon injection with TCT, the double PGRP-LC, CG8046 mutant adult flies exhibited robust systemic induction of Dpt (Supplemental Fig. 2C–F). However, Malpighian tubules, the highly immune responsive Drosophila renal system (27), failed to respond to TCT challenge (p < 0.0001) in the double PGRP-LC, CG8046 mutant flies (Fig. 3E). Notably, modEncode data suggest that CG8046 is only weakly expressed in the Malpighian tubule (28), however, it is immune-inducible and readily detectable in this tissue, as in S2* cells, after immune challenge (Supplemental Fig. 2G). In addition, the PGRP-LC, CG8046 double mutant flies succumbed to systemic E. coli infection as observed in the PGRP-LC, PGRP-LE double mutant (Fig. 3F). This finding suggests that CG8046 is a critical component of PGRP-LE-dependent cytosolic host defense. Altogether, these results suggest that CG8046 supports cytosolic recognition of monomeric PGN in vivo, in the Malpighian tubules, and plays an essential role for protection against certain systemic gram-negative infection. Our results also imply that Drosophila may use redundant mechanisms for the cytosolic import of TCT, and the roles of specific transporters may vary depending on the tissue.

Next, we asked whether human and mouse SLC46 transporters also promote TCT recognition. To this end, NF-κB luciferase assays were performed in HEK293 cells, which express a low but functional amount of NOD1. The SLC46 transporter family has three paralogs in mice and humans (Fig. 3A). SLC46A1/proton-coupled folate transporter is responsible for the intestinal absorption of folate and antifolates (29). SLC46A2/TSCOT was first identified because of its abundant expression in mouse thymic cortical epithelial cells, but has not been functionally characterized (30). Likewise, SLC46A3 is not yet characterized. Expression of human SLC46A2 resulted in a ∼4.6-fold increase (empty vector versus hSLC46A2, p < 0.000; 1) in NF-κB reporter activity in response to TCT. Mouse Slc46a2 also showed a significant increase in TCT-triggered NF-κB reporter activity (Fig. 4A). All six mammalian transporters were robustly expressed in transfected HEK293 cells, as monitored by FACS (Fig. 4B, 4C). HCT-116, a human colorectal cancer cell line, also showed enhanced responses to TCT, with either human or mouse SLC46A2 as well as with mouse Slc46a3 (Fig. 4D). Expression of SLC46 transgenes was comparable, except human SLC46A3, which was significantly lower in HCT-116 cells (Fig. 4E, 4F). Therefore, we could not draw any conclusion on the activity of hSLC46A3 in the HCT-116 assays.

Next, we tested if the SLC46 transporters also facilitate recognition of MDP, a NOD2 agonist. Similar dual-luciferase assays were repeated in HCT-116 cells and the same SLC46 homologs that promoted TCT recognition also enhanced NF-κB activation in response to MDP (Fig. 4G). These results suggest that members of the SLC46 family are conserved transporters for muropeptides in general.

The enhanced NF-κB activation observed in response to TCT was not observed with expression of SLC15A1, SLC15A2, or SLC15A4, which were previously associated with Tri-DAP and MDP uptake (Fig. 4H, 4I) (18, 20, 21). In contrast, SLC46A2 did not enhance the response to TNF, which indicates that SLC46A2 does not generally enhance NF-κB activation (Fig. 4J).

NOD1 and NOD2 signal through RIP2 kinase to activate the MAPK and the IKK signaling pathways. To determine if NF-κB activation supported by SLC46 and TCT is NOD1 dependent, NOD1 was silenced by siRNA in HEK293 cells expressing human SLC46A2. NF-κB luciferase assays showed that NOD1 RNAi significantly decreased NF-κB activation compared with the non-targeting siRNA upon TCT stimulation (Fig. 4K). Next, HEK293 cells expressing SLC46A2 were stimulated with TCT and TNF in the absence or presence of gefitinib, an effective RIP2 tyrosine kinase inhibitor (31). NF-κB activation supported by SLC46A2 and TCT decreased with increasing concentrations of gefitinib in a dose-dependent manner, whereas there was no effect on TNF stimulation (Fig. 4L).

Lastly, the subcellular localization of SLC46A2 was examined in HEK293 cells. Confocal microscopy revealed that mouse Slc46a2-EGFP was localized to acidic subcellular organelles labeled by LysoTracker and distributed evenly throughout the cytoplasm without TCT (Fig. 5A). However, upon exposure to TCT, mSlc46a2-EGFP–expressing LysoTracker-positive organelles aggregated to form, in significant quantities, large LysoTracker-positive clusters in the cytosol (Fig. 5B, 5C). Human SLC46A2-EGFP behaved similarly, forming large LysoTracker-positive clusters in response to TCT (Fig. 5D). These data suggest that SLC46A2 facilitates NF-κB activation triggered by NOD1/RIP2 in response to TCT from acidic organelles, most likely late-endosomes and/or endolysosomes, in human cells.

Notwithstanding that mammalian NOD1 and Drosophila PGRP-LE lack any sequence homology, both receptors promote cytosolic recognition of monomeric DAP-type PGNs, like TCT, and trigger conserved NF-κΒ signaling pathways leading to profound changes in target gene expression. By applying a targeted RNAi approach in engineered S2* cells, we identified CG8046, a Drosophila SLC46 family transporter, as a putative TCT transporter. In S2* cells, knockdown of CG8046 profoundly hampered PGRP-LE–mediated recognition of TCT including downstream signaling events as well as AMP gene induction. On the other hand, overexpression of CG8046 enhanced responses to TCT in S2* cells and adult flies. A CG8046 deletion allele exhibited normal activation of the intracellular TCT/PGRP-LE pathway in adult flies at a systemic level, which is primarily fat-body mediated. However, the Malpighian tubules significantly relied on CG8046 for robust induction of Dpt upon cytosolic TCT stimulation, and mutant adult flies lacking both PGRP-LC and CG8046 succumbed to E. coli infection similar to PGRP-LC and PGRP-LE double mutants. Our results argue that Drosophila utilizes redundant mechanisms to present TCT to the cytoplasmic receptor PGRP-LE, and the relative contribution of the redundant transporters varies between different tissues; CG8046 is a major component of this pathway at least in the Malpighian tubules as well as hemocyte-derived S2* cells, but plays little role in the fat body.

In our efforts to determine the functional relevance of human SLC46A2 in innate immune recognition, we examined a number of cell lines, including HEK293T, HCT-116, HT-29, THP-1, and MCF-7. However, SLC46A2 was negligible in all, making loss of function analysis impractical. In vivo, the expression of human SLC46A2 is highest in the thymus, similar to the mouse ortholog (data not shown) (30). With no published information on the role of NLRs in thymic function, it is not yet possible to postulate the role of SLC46A2 in this tissue. A publicly available database (http://biogps.org/) further indicated that human and mouse SLC46A2 are commonly expressed in skin. In addition, human, but not mouse, CD14-positive monocytes, endometrium, and lungs also show moderate SLC46A2 expression. Human keratinocytes express SLC46A2 and respond to the NOD1 agonist (32). Therefore, keratinocytes are of primary interest to further characterize the role of SLC46A2 in human cells.

Previously, Magalhaes et al. (33) suggested that mouse Nod1, but not human NOD1, recognized TCT based on their dual-luciferase assays where human or mouse NOD1 and TCT were cotransfected into HEK293 cells. However, our data argue that human NOD1 is a sensitive sensor of TCT, but requires SLC46A2 for proper delivery of TCT to this endogenous cytosolic receptor (Fig. 4).

The human genome encodes over 390 SLC proteins (347 SLCs in Drosophila), which are grouped into 52 families (24, 34). Although both the SLC15 and SLC46 family are proton-driven cotransporters, they are distantly related. Whereas SLC15 family members have been described to transport oligopeptides and amino acids, SLC46A1, the founding member of the SLC46 family, has been reported to transport folic acid and its derivatives, and heme with lower affinity. In fact, the SLC46 family is most similar in its major facilitator superfamily fold with SLC2, SLC16, SLC17, SLC18, SLC22, SLC37, SLC43, and SLCO, which are organic ion transporters based on their sequences (35), but not SLC15. Nonetheless, some of the SLC46s, in flies and mammals, appear to promote the delivery and recognition of muropeptides. The overlapping but distinct expression pattern of these SLC46s may allow the host to perform optimal surveillance for pathogens as well as commensals in different tissues.

We thank Anni Kleino for critical reading of the manuscript, Maninjay Atianand for technical assistance and discussion, Jean Luc Imler for Attacin A-luciferase construct, Stephen Girardin for SLC15A4 construct, and Michael Brodsky for gene editing by CRISPR.

The authors have no financial conflicts of interest.