Abstract
The accumulation of DNA in the cytosol serves as a key immunostimulatory signal associated with infections, cancer and genomic damage1,2. Cytosolic DNA triggers immune responses by activating the cGAS/STING pathway3. The binding of DNA to the cytosolic enzyme cGAMP synthase (cGAS), activates its enzymatic activity, leading to the synthesis of a second messenger, cyclic[G(2’,5’)pA(3’,5’)] (2’3’-cGAMP)4–8. 2’3’-cGAMP, a cyclic dinucleotide (CDN), activates the protein ‘stimulator of interferon genes’ (STING)9, which in turn activates the transcription factors IRF3 and NF-κB promoting the transcription of genes encoding type I interferons and other cytokines and mediators that stimulate a broader immune response. Exogenous 2’3’-cGAMP and other CDNs, including CDNs produced by bacteria and synthetic CDNs used in cancer immunotherapy, must traverse the cell membrane to activate STING in target cells. How these charged CDNs pass through the lipid bilayer is unknown. Here we used a genome-wide CRISPR interference screen to identify the reduced folate carrier SLC19A1 as the major CDN transporter for uptake of synthetic and naturally occurring CDNs. CDN uptake and functional responses are inhibited by depleting SLC19A1 from cells and enhanced by overexpressing SLC19A1. In both cell lines and primary cells ex vivo, CDN uptake is inhibited competitively by folate and blocked by the SLC19A1 inhibitor sulfasalazine, a medication approved for the treatment of inflammatory diseases. The identification of SLC19A1 as the major transporter of CDNs into cells has far reaching implications for the immunotherapeutic treatment of cancer10, transport of 2’3’-cGAMP from tumor cells to other immune cells to trigger the anti-tumor immune response11, host responsiveness to CDN-producing pathogenic microorganisms12, and potentially in certain inflammatory diseases.
Main text
The cGAS/STING pathway senses cytosolic DNA originating from viruses and bacteria9 as well as CDNs produced by certain bacteria13–15. Notably, the STING pathway is also activated by cytosolic self DNA, which accumulates in cells in certain autoinflammatory disorders, including Aicardi–Goutieres Syndrome and systemic lupus erythematosus16–19. Furthermore, cytosolic DNA accumulates in cells subjected to DNA damage, as occurs in tumor cells, resulting in activation of the cGAS/STING pathway and the initiation of an anti-tumor immune response20. Recently, we revealed that 2’3’-cGAMP can be transferred from tumor cells to immune cells in vivo, prompting the activation of the immune response11. Furthermore, synthetic STING agonists, such as 2’3’-RR CDA, an analogue of 2’3’-cGAMP (Fig. S1)21, can greatly enhance the anti-tumor immune response when delivered directly into the tumor microenvironment in mouse models of cancer, causing tumor regressions10,22. 2’3’-RR CDA and other synthetic CDNs are currently being tested in clinical trials as cancer immunotherapies. However, a critical outstanding question is the mechanism of transport of CDNs into cells of the immune system23. CDNs may be incorporated into cells via gap junctions, membrane fusions, or by incorporation into viral particles24–27,28. but none of these mechanisms explain (systemic) immune activation by extracellular CDNs. To systematically identify the genes involved in cytosolic transport of CDNs, we performed a genome-wide CRISPR interference screen in the monocytic THP-1 cell line.
To visualize STING activation, THP-1 cells were transduced with a CDN-inducible reporter construct (Fig. 1a). The reporter was composed of Interferon Stimulatory Response Elements (ISRE) and a mouse minimal IFN-β promoter that drives the expression of tdTomato upon hIFN-β or CDN exposure (Fig. 1b). In line with previous results, the synthetic CDN 2’3’-RR CDA induced a more potent response than 2’3’-cGAMP10 even when applied at a lower concentration. The response to both CDNs was several fold higher than the response to hIFN-β, and was completely dependent on STING expression (Fig. 1b), implying that the reporter primarily reported cell-intrinsic STING activity. To interrogate the approximately 20,000 human genes for their role in CDN-induced reporter expression, we performed a genome-wide CRISPR interference (CRISPRi) forward genetic screen in THP-1 cells. We generated a stable line of THP-1 cells expressing dCas9-BFP-KRAB, which was validated and expanded before transducing the cells with the CRISPRi v2 library at a low multiplicity of infection (see Methods). The CRISPRi library of cells was stimulated either with 2’3’-RR CDA, or with 2’3’-cGAMP, using concentrations that resulted in 90% reporter-positive cells. The highest expressing 25% and lowest expressing 25% of stimulated cells in each library were sorted by flow cytometry, DNA isolated, and gRNA sequences from each population, and unsorted cells, were amplified and DNA from each of these populations as well as from unsorted cells was deep-sequenced to identify the targeted genes in each population (Fig. 1c and Fig. S2). The fold enrichment and depletion of gRNAs in the hypo-responsive population versus the hyper-responsive population was calculated for each screen (Fig. S3, Table S1 and Table S2). We also integrated multiple gRNAs per gene using Mageck (see Methods) comparing the hyporesponsive and hyperresponsive populations calculated as robust rank aggregations scores and depicted in Fig. 1d and e. Similar results were obtained when each sorted population was compared to unsorted cells (Table S2). The two screens yielded many common hits, but there were some differences, such as numerous hits in the 2’3’-cGAMP screen including STAT2, IRF9, IFNAR1, and IFNAR2 (Table S2). Hence, the 2’3’-RR CDA screen may have been mostly dependent on intrinsic STING signaling, whereas the 2’3’-cGAMP screen may have been partly dependent on autocrine/paracrine IFN-β signaling.
In both CDN screens, the top hits in the hypo-responsive population (i.e. the genes most important for robust responses to CDNs) included the transcription factor IRF3, which acts directly downstream of STING. One of the five gRNAs for STING itself was also enriched in hyporesponsive cells from both screens, though the other STING gRNAs were not, presumably because they were ineffective at interfering with STING expression (Table S1). Other significant hits included genes involved in transcription, splicing, and immune modulation (Table S2).
One of the most significant hits in both screens was the SLC19A1 gene. SLC19A1 is a cell surface transporter known as the reduced folate carrier. SLC19A1 and another transporter, SLC46A1, are responsible for uptake of folate from the extracellular environment29. To validate the role of SLC19A1 in CDN stimulation, the top two enriched SLC19A1-targeting gRNAs from the 2’3’-RR CDA screen were used to stably deplete SLC19A1 in THP-1 cells expressing dCas9-KRAB (Fig. S4a). SLC19A1-depleted cells grew normally and appeared healthy, suggesting that other folate transport mechanisms fully suffice in SLC19A1-deficient cells. SLC19A1-depleted and, for comparison, IRF3-depleted cells (Fig. S4b) were stimulated with 2’3’-cGAMP, 2’3’-RR CDA, cyclic [A(3’,5’)pA(3’,5’)] (3’3’ CDA, a bacterial CDN) or hIFN-β, and reporter induction was measured 20 h later (Fig. 2a). Responses to 2’3’-cGAMP, 2’3’-RR CDA and 3’3’-CDA were each strongly inhibited in IRF3- and SLC19A1-depleted cells (Fig. 2b), whereas stimulation by hIFN-β was not affected (Fig. 2b). Restoration of SLC19A1 expression by transduction of a cDNA expression vector rescued CDN responsiveness without affecting stimulation by hIFN-β (Fig. 2c).
As an alternative approach to corroborate the role of SLC19A1 in CDN responses, the conventional CRISPR/Cas9 system was used to target a coding exon in order to generate loss of function mutations in the SLC19A1 gene. Disruption of the SLC19A1 gene was confirmed by genomic PCR, TA-cloning and sequencing for nine SLC19A1-/- clones (see Methods). These clones were all significantly less sensitive to CDN stimulation when compared to seven control clones that received a non-targeting gRNA (Fig. 2d).
Importantly, SLC19A1 overexpression robustly increased CDN responsiveness in THP-1 cells as well as in cell lines that normally responded poorly or not at all to CDN stimulation, including C1R, K562, 293T (pre-transduced with STING), and RAW macrophage cell lines (Fig. 2e and f). Taken together, our data show reduced responses to CDNs in SLC19A1-deficient cells and much amplified responses in cells overexpressing SLC19A1, as might be expected for a CDN transporter. Together, these data support a central role of the SLC19A1 transporter in responses to several cyclic dinucleotides, including the mammalian CDN 2’3’-cGAMP.
Based on our findings, we tested whether the drug sulfasalazine (SSZ), a non-competitive inhibitor of SLC19A130, would block stimulation by CDNs. THP-1 reporter cells were exposed to various concentrations of SSZ or DMSO vehicle in the presence of 2’3’-cGAMP, 2’3’-RR CDA, or hIFN-β. Responses to both CDNs were robustly inhibited with increasing concentrations of SSZ, whereas responses to hIFN-β stimulation were only modestly inhibited (Fig. 2g). The concentrations required for inhibition were only modestly higher than those that inhibit uptake of folate derivatives in another study30. Surprisingly, at lower concentrations, SSZ modestly enhanced stimulation by 2’3’-RR CDA, but had no effect on stimulation by 2’3’ cGAMP (Fig. 2g).
The effect of SLC19A1 on reporter induction by CDNs led us to test the impact of SLC19A1 deficiency on endogenous transcriptional targets downstream of STING, including the genes encoding the chemokines CCL5 and CXCL10, which are direct targets of IRF331,32. In control cells, CCL5 and CXCL10 gene expression was highly elevated 5h after 2’3’-RR CDA stimulation. In cells depleted of IRF3, SLC19A1 or STING, chemokine expression was strongly inhibited, indicating that SLC19A1 action is necessary for CDN-induced effects, including those downstream of STING (Fig. 3a and b).
To directly assess the effect of SLC19A1 on STING pathway activation (Fig. S5), we evaluated phosphorylation of STING, IRF3 and TBK1 in control (non-targeting gRNA) versus CRISPRi-depleted cells by immunoblotting (Fig. 3c). Within 2 hours after stimulation with 2’3’-RR CDA, phosphorylation of STING, IRF3, and TBK1 were each significantly elevated in control THP-1 cells. In IRF3-depleted cells, phosphorylation of upstream signaling components STING and TBK1 was not affected, whereas in STING-depleted cells phosphorylation of both TBK1 and IRF3 was nearly ablated. SLC19A1-targeted cells showed major defects in phosphorylation of STING, TBK1 and IRF3, supporting the conclusion that SLC19A1 acts upstream of STING. Notably, protein levels of STING, TBK1, and IRF3 were unaltered in SLC19A1-depleted cells, indicating that SLC19A1 does not influence stability or degradation of STING pathway components.
To further exclude a general defect in STING activation caused by SLC19A1-depletion, the cGAS/STING pathway was directly triggered intracellularly by transfecting cells with interferon stimulatory DNA (ISD). ISD transfection of both WT and SLC19A1-depleted THP-1 cells resulted in a strong and equal induction of IFNB gene expression (Fig. 3d). Thus, STING functioned normally in SLC19A1-depleted cells when DNA was introduced directly into the cytosol by transfection.
The finding that SLC19A1 was not essential when DNA was transfected into cells suggested that SLC19A1 may function by transporting CDNs into cells. Therefore, we enzymatically synthesized [32P] 2’3’-cGAMP, which we confirmed by TLC and DRaCALA33 binding analysis (Fig. S6a and b). We next monitored 2’3’-cGAMP uptake by cells expressing different levels of SLC19A1. SLC19A1 overexpression greatly enhanced uptake of [32P] 2’3’-cGAMP by THP-1 cells (Fig. 4a) and C1R cells (Fig. S7a). Conversely, SLC19A1-depletion reduced the uptake of 32P 2’3’-cGAMP in THP-1 cells (Fig. 4a). We next sought to determine the specificity for 2’3’-cGAMP uptake in THP-1 cells. Addition of excess, unlabeled bacterial-derived 3’3’-linked cyclic di-nucleotides as well as host-derived 2’3’-cGAMP to cell culture media completely inhibited [32P] 2’3’-cGAMP uptake by THP-1 cells, suggesting that cyclic di-nucleotide interactions with the transporter are not highly specific for the 2’3’ linkage or the specific nucleotides (Fig. 4b). In quantitative competition ligand uptake assays, unlabeled 2’3’-cGAMP inhibited uptake of [32P] 2’3’-cGAMP with an IC50 of 1.89 ± 0.11 µM, in line with the reported affinity of SLC19A1 for methotrexate and other folates (Fig. 4c)34. As SLC19A1 was first described as a folate transporter, we performed similar competition experiments using excess, unlabeled folic acid, and we also tested an inhibitor of folate uptake by SLC19A1, sulfasalazine. Remarkably, both folic acid and sulfasalazine inhibited [32P] 2’3’-cGAMP uptake with IC50’s of 4.79 ± 0.08 µM and 2.06 ± 0.17 µM, respectively (Fig. 4d and e). We extended the study by asking whether folic acid or sulfasalazine inhibited uptake of [32P] 2’3’-cGAMP in other cell types. We found that the addition of excess folate and sulfasalazine to cell cultures abrogated [32P] 2’3’-cGAMP uptake by U937 monocytes as well as primary murine peritoneal leukocytes and splenocytes (Fig. 4f, S7b). Taken together these results suggested that uptake of the mammalian CDN 2’3’-cGAMP by human and mouse cells, including cell lines and primary cell ex vivo, depends on SLC19A1 expression and function.
If SLC19A1 transports CDNs into cells, CDNs may directly interact with SLC19A1. Consistent with a direct interaction between 2’3’-cGAMP and SLC19A1, His-tagged SLC19A1 was precipitated by 2’3’-cGAMP immobilized on Sepharose beads (Fig. 4g and S8). This interaction was specific, as free, unbound 2’3’-cGAMP competitively disrupted the 2’3’-cGAMP-SLC19A1 interaction (Fig. 4h and S8). As a positive control, His-tagged STING C-terminal domain was also precipitated by 2’3’-cGAMP-Sepharose (Fig 4g and S8). These data suggest that CDNs interact with SLC19A1, consistent with the proposed role of SLC19A1 as a CDN transporter. Taken together, our results demonstrate that SLC19A1 is a mammalian CDN transporter, required for exogenous CDN-mediated type I Interferon activation.
The response to CDNs is weak in most cell lines tested, and can be increased by overexpression of SLC19A1. Indeed, THP-1 cells are near the top of a large set of cell lines in expression of both SLC19A1 and STING (Fig. S9), suggesting that SLC19A1 expression and STING expression may together predict the responsiveness to CDN stimulation by cell lines and tumors.
Both folic acid and sulfasalazine almost completely blocked CDN uptake and/or stimulation, whereas CDN stimulation was not completely inhibited in SLC19A1-null cells. This implied that another transporter sensitive to folic acid competition and sulfasalazine inhibition may play a role in CDN uptake. Overexpression of SLC46A1, which encodes the only other known folate transporter, did increase responses to CDNs (Fig. S10). However, depletion of SLC46A1 had only a modest effect on CDN stimulation, and was not a significant hit in our screen. Furthermore, depleting SLC46A1 and SLC19A1 together was no more effective than depleting SLC19A1 alone (Fig. S11). These data suggest that yet another transporter that is inhibited by folic acid and sulfasalazine may play a partial role in CDN transport. SLC46A3, another transporter, was also a hit in our screen. Overexpression of SLC46A3 increased the response to CDNs (Fig. S10). Depletion of SLC46A3 had a modest effect on reporter induction by both CDNs (Fig. S11). However, depleting both SLC19A1 and SLC46A3 together did not reduce responses more than depletion of SLC19A1 alone (Fig. S11), suggesting that SLC46A3 is not responsible for most of the residual CDN transport in SLC19A1-depleted cells.
Our findings define SLC19A1 as a major transporter of exogenous 2’3’ cGAMP, 2’3’-RR CDA and probably other CDNs into the cytosol. In this context, it likely plays an important role in the anti-tumor and adjuvant effects of injected CDNs. It may also be important in cell-to-cell transport of CDNs in immune responses, both in the context of cancer11 and potentially during viral infections. SLC19A1-mediated uptake of CDNs may also be critical for the pathology of various inflammatory diseases35,36. For example, in mouse models of inflammatory bowel disease (IBD), some evidence suggests that host cells import CDNs produced by intestinal bacteria, activating STING in a cGAS-independent fashion36. SLC19A1-mediated uptake of CDNs from the extracellular environment may thus contribute to the inflammatory profile underlying such diseases. Moreover, the SLC19A1 inhibitor sulfasalazine is a first line treatment in rheumatoid arthritis, and is often used to treat inflammatory bowel diseases (IBD), including ulcerative colitis and Crohn’s disease37,38. Sulfasalazine has an immunosuppressive effect, in part by inhibiting the NF-κB pathway39, but the mechanism of inhibition is unknown. Our results raise the intriguing possibility that sulfasalazine exerts its anti-inflammatory effects in these diseases by inhibiting uptake of CDNs produced endogenously or by commensal bacteria, preventing STING activation. In conclusion, we have identified SLC19A1 as a CDN transporter with potential relevance in the context of cancer immunotherapy, immunosurveillance, and inflammatory disease.
Methods
Cell culture
All cell lines were cultured at 37°C in humidified atmosphere containing 5% CO2 with media supplemented with 100 U/mL penicillin, 100 µg/mL streptomycin, 0.2 mg/mL glutamine, 10 µg/mL gentamycin sulfate, 20 mM Hepes and 10% FCS. THP-1, C1R, and K562 cells were cultured in RPMI medium, and 293T, 293T transfected with hSTING (293T+hSTING), MDA-MBA-453 (MDA), and RAW macrophages were cultured in DMEM medium. THP-1, K562, 293T cells, and RAW macrophages were present in the lab at the time this study began. MDA cells were obtained from the Berkeley Cell Culture Facility. C1R cells were a generous gift from Veronika Spies (Fred Hutchinson Cancer Center, Seattle WA). 293T+hSTING cells were generated at Aduro Biotech Inc.
Antibodies and reagents
The following antibodies were derived from Cell signaling technologies: rabbit-anti-human TBK1 mAb (clone D1B4, used 1:500 for immunoblot [IB]), rabbit-anti-human phospho-TBK1 mAb (clone D52C2, 1:1000 for IB), rabbit-anti-human STING mAb (clone D2P2F, 1:2000 for IB), rabbit-anti-human phospho STING mAb (clone D7C3S, used 1:1000 for IB), rabbit-anti-human phospho-IRF3 mAb (clone 4D4G, 1:1000 for IB). Antibodies derived from LI-COR Biosciences: goat-anti-mouse IgG IRDye 680RD conjugated (cat. #: 926-68070, used 1:5000), donkey-anti-rabbit IgG IRDye 800CW conjugated (cat. #: 926-32213), donkey-anti-rabbit IgG IRDye 680RD (cat. #: 926-68073). Other antibodies: rabbit-anti-human IRF3 mAb (Abcam, cat. #: EP2419Y, used 1:2000 for IB), mouse-anti-human transferrin receptor mAb (Thermo Fischer Scientific, clone H68.4, used 1:1000 for IB), rabbit-ant-human SLC19A1 pAb (Picoband, cat. #: PB9504, used 0.4 µg/ml for IB), APC-conjugated mouse-anti-human CD55 mAb (BioLegend, clone JS11, used 1:50 for flow cytometry), mouse-anti-human CD59 mAb (BioLegend clone p282, used 1:250 for flow cytometry), APC-conjugated goat-anti-mouse IgG (BioLegend, cat. #: 405308, used 1:100 for flow cytometry). Reagents used: Sulfasalazine (Sigma-Aldrich, cat. #: S0883), polybrene (EMD Millipore, cat. #: TR1003G), 3’3’-cyclic-di-AMP (CDA) (Invivogen, cat. #: tlrl-nacda), 2’3’-RR CDA and 2’3’-cyclic-di-GMP-AMP (cGAMP) (generous gift from Aduro Bioscience Inc.), human interferon-β (PeproTech, cat. #: 300-02B), mouse interferon-β1 (BioLegend, cat. #: 581302). Antibiotic selection: puromycin (Sigma-Aldrich, cat. #: P8833), blasticidin (Invivogen, cat. #: ant-bl-1, used at 10 µg/ml), zeocin (Invivogen, cat. #: ant-zn-1, used at 200 µg/ml).
Plasmids
A gBLOCK gene fragment (Integrated DNA Technologies, Inc.) encoding the tdTomato reporter gene driven by the interferon stimulatory response elements (ISREs) and the minimal mouse interferon-β promoter was cloned into a dual promoter lentiviral plasmid by means of Gibson assembly. This lentiviral plasmid co-expressed the Zeocin resistance gene and GFP via a T2A ribosomal skipping sequence controlled by the human EF1A promoter, and was generated as described previously40.
For rescue and overexpression of SLC19A1, SLC46A1, or SLC46A3, a gBLOCK gene fragment encoding SLC19A1 (gene ID 6573, transcript 1), SLC46A1 (gene ID 113235) or SLC46A3 (gene ID 283537) was cloned by Gibson assembly into a dual promoter lentiviral plasmid co-expressing the Blasticidin resistance gene and the fluorescent gene mAmetrine.
For CRISPR interference (CRISPRi)-mediated depletions, cells were transduced with a lentiviral dCas9-HA-BFP-KRAB-NLS expression vector (Addgene plasmid #102244).
For screen validation using individual gRNAs, gRNAs (table S3) were cloned into the same expression plasmid used for the gRNA library (“pCRISPRia-v2”, Addgene plasmid #84832, a gift from Jonathan Weissman). The lentiviral gRNA plasmid co-expresses a puromycin resistance gene and blue fluorescence protein (BFP) via a T2A ribosomal skipping sequence controlled by the human EF1A promoter. The CRISPRi gRNAs introduced into this vector by Gibson assembly were expressed from a murine U6 promoter. For expression of multiple gRNAs, additional gRNAs were introduced in a separate vector that co-expressed the blasticidin resistance gene and mAmetrine via a T2A ribosomal skipping sequence under the control of a human EF1A promoter.
Conventional CRISPR gRNAs (see table S3) were cloned into a selectable lentiviral CRISPR/Cas9 vector. This lentiviral vector includes a human codon-optimized S.pyogenes Cas9 co-expressing puromycin resistance gene via a T2A ribosome skipping sequence under the control of a minimal human EF1A promoter40,41.
Lentiviral production and transduction
Lentivirus was produced by transfecting lentiviral plasmids and 2nd generation packaging/polymerase plasmids into 293T cells using TransIT-LT1 Reagent (Mirus Bio LLC). Virus-containing supernatants were harvested 72h later, centrifuged to remove cell debris, and filtered using a 0.45 µm PES filter. Filtered virus supernatant was used to transduce target cells by spin-infection (800 x g for 90min at 33°C) in the presence of 8 µg/ml polybrene. After spin-infection virus and polybrene containing medium was diluted 1:1 with fresh medium. 72 hours after transduction, cells were sorted based on fluorescence expression using a BD FACSAria cell sorter, or selected with relevant selection reagent.
2’3’-RR CDA and 2’3’-cGAMP screens
THP-1 cells co-expressing the tdTomato reporter, GFP, and dCas9-BFP were single cell sorted to select for a THP-1 cell clone with efficient dCas9-BFP-knockdown capacity. Clonal populations were transduced with lentiviral vectors encoding gRNAs targeting GFP, CD55 or CD59. After 1 week on puromycin (2 µg/ml) selection, CD55, CD59 and GFP expression were quantified using the BD LSR Fortessa flow cytometer. A clonal cell that showed the highest reduction in all three marker genes was selected for the screens. Two cultures of THP-1 cells were separately transduced with the human genome-wide CRISPRi v2 library42. Each library of THP-1 cells was separately screened by treating the cells with 2’3’-RR CDA or 2’3’-cGAMP followed by selection and analysis. Hence, each screen was performed twice, with different CRISPRi library transduced cultures of THP-1 cells.
For each transduction, the THP-1 clone was expanded to 320 million cells and transduced with the human genome-wide CRISPRi v2 library42, which contains approximately 100,000 gRNAs targeting around 20,000 genes. Sufficient cells were transduced and propagated to maintain at least 50 million transduced (BFP+) cells, representing 500x coverage of the gRNA library. The transduction efficiency was around 20% to minimize the chance of multiple lentiviral integrations per cell. Two days after transduction, cells were cultured in the presence of puromycin for two days and one day additional day without puromycin. 400 million cells were seeded to a density of 1 million cells/ml and stimulated with 2’3’-RR CDA (2 µg/ml) or 2’3’-cGAMP (15 µg /ml). 20h later, cells were harvested, washed in PBS, and sorted based on BFP expression (presence of gRNAs), GFP expression (presence of reporter) and tdTomato expression using the BD Influx cell sorter and BD FACSaria Fusion cell sorter. The cells were sorted into two populations based on tdTomato expression: the highest 25% of tdTomato expressing cells (hyper-responsive population) and lowest 25% of tdTomato expressing cells (hypo-responsive population). During sorting, all cells were kept at 4°C. After sorting, cells were counted: the sorted populations contained 15-20 million cells, and the unsorted control contained 100-150 million cells. Cells were washed in PBS, and cell pellets were stored at −80°C until further processing.
gDNA isolation and sequencing
Genomic DNA was isolated from sorted cells using NucleoSpin Blood kits (Macherey-Nagel). PCR was used to amplify gRNA cassettes with Illumina sequencing adapters and indexes as described previously43. Genomic DNA samples were first digested for 18 hours with SbfI-HF (NEB) to liberate a ∼500 bp fragment containing the gRNA cassette. The gRNA cassette was isolated by gel electrophoresis as described previously43. using NucleoSpin Gel and PCR Clean-up kits (Macherey-Nagel), and the DNA was then used for PCR. Custom PCR primers are listed in Supplementary Table 5. Indexed samples were pooled and sequenced on an Illumina HiSeq-2500 for the 2’3’RR CDA screen and an Illumna HiSeq-4000 for the 2’3’-cGAMP screen using a 1:1 mix of two custom sequencing primers (Supplementary Table 5). Sequencing libraries were pooled proportional to the number of sorted cells in each sample. The target sequencing depth was at least 2,000 reads/gRNA in the library for unsorted “background” samples, and at least 10 reads/cell in sorted samples.
Screen data analysis
CRISPRi samples were analyzed using the Python-based ScreenProcessing pipeline (https://github.com/mhorlbeck/ScreenProcessing). Normalization using a set of negative control genes and calculation of phenotypes and Mann-Whitney p-values was performed as described previously42,44. Briefly, Illumina 50bp single end sequencing reads for pooled sublibraries one to four and five to seven were trimmed to 29bp and guides were quantified by counting exact matches to the CRISPRi v2 human library guides. Phenotypes were calculated as the log2 fold change in enrichment of an sgRNA in the high and low samples versus background as well as high versus low, normalized by median subtracting non-targeting sgRNAs44,45. Phenotypes from sgRNAs targeting the same gene were collapsed into a single sensitivity phenotype for each gene using the average of the top three scoring sgRNAs (by phenotype absolute value). For genes with multiple independent transcription start sites (TSSs) targeted by the sgRNA libraries, phenotypes and p-values were calculated independently for each TSS and then collapsed to a single score by selecting the TSS with the lowest Mann-Whitney p-value. Counts from the ScreenProcessing pipeline were then used as input to the MAGeCK program to obtain FDR scores for filtering (see table S2).
Genes were also ranked by individual gRNAs with the greatest enrichment/depletion between the hypo-responsive and hyper-responsive libraries. gRNA read counts were normalized to library sequencing depth by converting to read counts per million total reads. For each gRNA, the ratio between the read counts for the hypo-responsive and hyper-responsive libraries was found and averaged between replicates. For hypo-responsive gene rankings, each gene was ranked by the single corresponding gRNA with the highest hypo-to-hyper ratio (see table S1, ‘highest ratio hypo/hyper’ column). For hyper-responsive gene rankings, each gene was ranked by the single corresponding gRNA with the lowest hypo-to-hyper ratio (see table S1, ‘lowest ratio hypo/hyper’ column). Gene-level phenotypes are available as Supplemental Materials (table S1 and S2).
CDN and IFN-β Stimulation assays
The week prior to stimulation experiments, cells were cultured at the same density. The day before stimulation, cells were seeded to 0.5×105 cells/ml. Cells were stimulated with CDNs or IFN-β in 48W plates using 50,000 cells/well in 300 µl medium. After 18-24h, cells were transferred to a 96W plate and tdTomato expression was measured by flow cytometry using a high throughput plate reader on a BD LSR Fortessa. For stimulations in the presence of sulfasalazine, cells were stimulated in 48W plates using 20,000 cells/well in 300 µl medium. Cells were incubated with sulfasalazine or DMSO as vehicle prior to stimulations with CDNs or IFN-β. 18-24h after stimulation, tdTomato reporter expression was quantified by flow cytometry using a high throughput plate reader on a BD LSR Fortessa.
Production of SLC19A1 knockout cell lines
THP-1 cells expressing the tdTomato reporter were transduced with a CRISPR/Cas9 lentiviral plasmid encoding a control gRNA or a gRNA targeting SLC19A1 at a region critical for transport46 (see table S3). Transduced cells were selected using puromycin for 2 days and single cell sorted using a BD FACSAria cell sorter. Control cells and SLC19A1-targeted cells were selected that had comparable forward and side scatter by flow cytometry analysis. Genomic DNA was isolated from clones using the Qiamp DNA minikit (Qiagen), and the genomic region surrounding the SLC19A1 gRNA target site was amplified by PCR using primers 5’-TTCTCCACGCTCAACTACATCTC-3’ and 5’-CAGCATCCGCGCCAGCACTGAGT-3’. PCR product was cloned into 5-alpha competent bacteria (New England Biolabs, cat. #C2987) using a TOPO TA cloning kit (Thermo Fischer Scientific, cat. # 450641) according to manufacturer’s instructions. After blue/white screening, a minimum of 10 colonies were sequenced per THP-1 clone, and sequences were analyzed using SeqMan (Lasergene DNASTAR). THP-1 clones with out-of-frame mutations at the SLC19A1 gRNA target site were selected for further experiments.
RT-qPCR
Cells were harvested and washed in ice-cold PBS. Cells were transferred to RNase-free microcentrifuge tubes and RNA was isolated using the RNeasy mini kit (Qiagen, cat. #: 74104) including a DNase step (Qiagen, cat. #: 79254). RNA concentration was measured by NanoDrop (Thermo Fischer), and 1 µg of RNA was used as input for cDNA synthesis using the iScript cDNA synthesis kit (Bio-rad, cat. #: 1708890). cDNA was diluted to 20 ng/µl and 2.5 µl/reaction was used as input for the qPCR reaction. qPCR reactions were set up using SSOFast EvaGreen Supermix (Bio-Rad, cat. #: 1725200) according to the manufacturer’s recommendations, using 500 nM of each primer and following cycling conditions on a Bio-Rad C1000 Thermal Cycler: 2 min at 98°C, 40 repeats of 2 sec at 98°C and 5 sec at 55°C. Primers used to amplify the HPRT1, YHWAZ, CCL5, CXCL10, STING, IRF3, SLC19A1, SLC46A1, and SLC46A3-specific PCR products are listed in Table S4. The housekeeping genes HPRT1 and YHWAZ served as endogenous control.
For quantification of IFNB1 mRNA, RNA was extracted with the Nucleospin RNA Isolation Kit (Machery-Nagel) and reverse-transcribed with the iScript cDNA synthesis kit (Bio-Rad). TaqMan real-time qPCR assays were used for quantification of human IFNB1 (Hs01077958_s1). ACTB (Hs01060665_g1) served as an endogenous control.
Synthesis of [32P] cyclic GMP-AMP and [32P] cyclic di-AMP
Radiolabeled 2’3’ cGAMP was enzymatically synthesized by incubating 0.33 µM α-[32P] ATP (Perkin-Elmer) with 250 µM unlabeled GTP, 1 µg of Interferon Stimulatory DNA 100mer (kindly provided by Daniel Stetson), and 1 µM of recombinant His-tagged 2’3’ cGAMP Synthase (cGAS) in binding buffer [40 mM Tris pH 7.5, 100 mM NaCl, 20 mM MgCl2] at 37°C overnight. The reaction was confirmed to have gone to completion by Thin Layer Chromatography (TLC) analysis. Briefly, the 2’3’ cGAMP synthesis reaction was separated on Polygram CEL300 PEI TLC plates (Machery-Nagel) in buffer containing 1:1.5 (vol/vol) saturated (NH4)2SO4 and 1.5 M NaH2PO4 pH 3.6. The TLC plates were then air dried and exposed to a PhosphorImager screen for visualization using a Typhoon scanner (GE Healthcare Life Sciences). Next, the sample was incubated with HisPur Ni-NTA resin (Thermo Scientific) for 30 min in order to remove recombinant cGAS. The resultant slurry was transferred to a minispin column (Thermo Scientific) to elute crude [32P] 2’3’ cGAMP. Recombinant mSTING-CTD protein was used for further purification of synthesized [32P] 2’3’ cGAMP. 100 µM mSTING-CTD was bound to HisPur Ni-NTA resin and incubated with the remaining crude 2’3’ cGAMP synthesis reaction mixture for 30 min on ice. Following removal of the supernatant, the Ni-NTA resin was washed three times with cold binding buffer. The resin was then incubated with 100 µL of binding buffer for 10 min at 95 °C, and transferred to a minispin column to elute [32P] 2’3’ cGAMP. The resulting STING-purified [32P] 2’3’ cGAMP was evaluated by TLC analysis and determined to be ∼99% pure.
Radiolabeled c di-AMP was synthesized as described previously 47. Briefly, 1 µM α-[32P] ATP (Perkin-Elmer) was incubated with 1 µM of recombinant DisA in binding buffer at 37°C overnight. The reaction mixture was boiled for 5 min at 95°C and DisA was removed by centrifugation. Recombinant His-tagged RECON was then used to further purify the c di-AMP reaction mixture. 100 µM His-tagged RECON was bound to HisPur Ni-NTA resin for 30 min on ice. The resin was washed three times with cold binding buffer and then incubated with 100 µL of binding buffer for 5 min at 95°C. The slurry was then transferred to a minispin column to elute [32P] c di-AMP. The purity of the radiolabeled c di-AMP was assessed by TLC and determined to be ∼98%.
Nucleotide-Binding Assays
The ability of radiolabeled 2’3’ cGAMP and c di-AMP to bind recombinant STING was evaluated by DRaCALA (differential radial capillary action of ligand assay) analysis, as previously described48. Briefly, varying concentrations of recombinant STING were incubated with ∼1 nM of radiolabeled cyclic di-nucleotide in binding buffer for 10 min at room temperature. The reaction mixtures were blotted on nitrocellulose membranes and air dried for 15 min. The membranes were then exposed to a PhosphorImager screen and visualized using a Typhoon scanner.
Nucleotide-Uptake Assays
For transport assays, cells were collected by centrifugation and washed in Dulbecco’s Phosphate-Buffered Saline (DPBS) (Life Technologies). The cell pellets were re-suspended in pre-warmed RPMI 1640 medium (GIBCO) containing 10% heat-inactivated FBS (HyClone) and supplemented with 10 mM HEPES, 1 mM sodium pyruvate and 2 mM L-Glutamine (Thermo Fisher) to a final cell density of 1 × 107 cells per ml. Uptake of 1 nM [32P] cGAMP and c di-AMP was assayed in cell suspensions at 37°C over the indicated time points. At the end of each time point, transport was quenched by the addition of cold DPBS. Cells were washed three times with cold DPBS, followed by lysis in 50 µL of cold deionized water. The cell lysates were then transferred to 5 ml of liquid scintillation cocktail (National Diagnostics) and the associated radioactivity was measured by liquid scintillation counting using a LS6500 Liquid Scintillation Counter (Beckman Coulter). For each sample, [32P] cyclic di-nucleotide uptake (counts per minute) was normalized to cell count. For competition experiments, cells were pre-incubated with indicated concentrations of “cold” unlabeled ligand for 15 minutes prior to the addition of 1 nM “hot” [32P] cGAMP. Cells were then collected at the indicated time points and processed as described above.
Protein Expression and Purification
Full-length human SLC19A1 cDNA with a C-terminal 8 X His-tag was subcloned into a dual promoter lentiviral vector (see above). Recombinant His-tagged SLC19A1 was expressed using a FreeStyle 293 Expression System. Briefly, 293F cells (1 × 106 cells per ml) grown in FreeStyle 293 Media supplemented with GlutaMax (GIBCO) were transfected with the SLC19A1 expression construct (1µg plasmid DNA per ml of cells) using PEI transfection reagent. Transfected cells were grown for 72 hours in a shaking incubator at 37°C in 5% CO2. Three days after transfection, the cells were harvested by centrifugation and washed in DBPS. Cell pellets were then re-suspended in lysis buffer [25 mM Tris pH 8.0, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride] supplemented with HALT Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific) and lysed by sonication. The cell lysate was supplemented with 2% (w/v) n-dodecyl-β-D-maltoside (DDM) and rotated for 2h at 4°C. The cell lysates were centrifuged at 15,000 rpm for 1h at 4°C to remove cell debris, and the detergent-soluble fraction was incubated with HisPur Ni-NTA resin for 1h at 4°C. The resin was washed with 100 column volumes of wash buffer [25 mM Tris pH 6.0, 150 mM NaCl, 30 mM imidazole, 5% glycerol (v/v), and 0.05% DDM (w/v)], and bound proteins were eluted in elution buffer [25 mM Tris pH 6.0, 150 mM NaCl, 300 mM imidazole, 5% glycerol (v/v), and 0.05% DDM (w/v)]. The resulting proteins were analyzed by SDS-PAGE followed by Coomasie staining and immunoblotting to confirm expression and purification of His-tagged SLC19A1.
Recombinant cGAS, DisA, mSTING-CTD, and mRECON were expressed and purified as previously described47–49. Briefly, plasmids for cGAS, DisA, mSTING-CTD, and mRECON expression were transformed into Rosetta (DE3) pLysS chemically competent cells. Overnight cultures of the resulting transformed bacteria were inoculated into 1.5 L of LB broth at a 1:100 dilution. Bacterial cultures were grown at 37°C to OD600 0.5 followed by overnight induction at 18°C with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). Cells were harvested and lysed in PBS supplemented with 1 mM PMSF and soluble protein was purified using nickel-affinity chromatography followed by gel filtration chromatography (S-300, GE Healthcare, Piscataway, New Jersey, USA). After SDS-PAGE analysis, the purified proteins were concentrated in storage buffer [40 mM Tris pH 7.5, 100 mM NaCl, 20 mM MgCl2, 25% glycerol (v/v)] and stored at −80°C.
Synthesis of cGAMP Sepharose
2’3’ cyclic GMP-AMP was enzymatically synthesized using recombinant cGAS as described previously8,48. Approximately, 100 mg of purified cGAMP was dissolved in PBS to 200 µM. The pH of the solution was adjusted to 7.5 with NaOH, and the resulting solution was added directly to washed epoxy-activated Sepharose and incubated at 56°C for 2 days. The Sepharose was washed and the absorbance spectrum of 50% slurry was measured to ensure nucleotide coupling. HPLC analysis of the remaining uncoupled nucleotide ensured no degradation of cGAMP occurred during the 2-day incubation. The remaining epoxy groups were blocked with ethanolamine following the instructions provided by GE. In parallel with this blocking step, fresh epoxy-activated Sepharose was also treated with ethanolamine to generate control resin.
cGAMP Pulldowns
Following nickel affinity purification, recombinant His-tagged SLC19A1 was incubated with 100 µL of ethanolamine-or cGAMP-conjugated Sepharose beads for 4h at 4°C with rotation, as described previously (Sureka et. al., 2014; McFarland et. al., 2016). Beads were washed three times with wash buffer [25 mM Tris pH 6.0, 150 mM NaCl, 5% glycerol (v/v), and 0.05% DDM (w/v)], and bound proteins were eluted by boiling in SDS-PAGE sample loading buffer for 5 min at 95°C. The soluble fraction was then removed and analyzed by SDS-PAGE followed by Coomassie Blue staining and immunoblotting. As a control, recombinant His-tagged mSTING-CTD was incubated with ethanolamine-or cGAMP-conjugated sepharose beads, as described above. Beads were washed three times with binding buffer, and then boiled in SDS-PAGE sample loading buffer for 5 min at 95°C. The soluble fraction was then analyzed by SDS-PAGE followed by Coomasie staining.
Cell lysis and immunoblotting
For anti-SLC19A1 immunoblotting, cells were lysed and proteins were separated by SDS-PAGE as described above in the ‘cGAMP pulldowns’ paragraph. SDS-PAGE-separated proteins were transferred onto nitrocellulose membranes (Bio-Rad) at 30V overnight at 4°C. Membranes were then air dried for 1h and blocked in 5% Blotto, non-fat milk (NFM, Santa Cruz Biotechnology) in 1 X TBS. Membranes were probed in 5% Bovine Serum Albumin (Fisher) in 1 X TBS-T with anti-SLC19A1 Picoband antibody (Boster Bio).
For protein detection using all other antibodies, cells were counted, washed with PBS and lysed in RIPA buffer (25 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.1% SDS) including cOmplete ULTRA protease inhibitors (Sigma-Aldrich cat. #: 05892791001), phosphatase inhibitors (Biomake, cat. # B15001) and 50mM DTT. Cells lysates were mixed with 4x NuPage LDS sample buffer (Invitrogen cat. #: NP0007), pulse sonicated and incubated at 75°C for 5min. Lysates were loaded onto Bolt 4-12% Bis-Tris Plus SDS-PAGE gels (Invitrogen cat. #: NW04125BOX). SDS-PAGE separated proteins were transferred onto Immobilon-FL PVDF membranes (EMD Millipore) at 100V for 1h at 4°C. Membranes were blocked in 4% NFM, and probed in 1% NFM overnight at 4°C with primary antibody. Membranes were subsequently washed 3 times in 1x-TBS-T and probed with secondary antibody for 1h at RT protected from light. Membranes were washed 2 times in TBS-T, once in TBS, and blots were imaged using an Odyssey CLx System (LI-COR).
Mice
C57BL/6J mice were obtained from The Jackson Laboratory. All of the mice were maintained in specific pathogen free conditions by the Department of Comparative Medicine at the University of Washington School of Medicine. All experimental procedures using mice were approved by the Institutional Animal Care and Use Committee of the University of Washington and were conducted in accordance with institutionally approved protocols and guidelines for animal care and use.
Isolation of Mouse Peritoneal Cavity Cells and Splenocytes
Mouse peritoneal cavity cells were recovered by peritoneal lavage with 5 ml ice cold PBS supplemented with 3% FCS, as previously described50. The peritoneal cells were cultured in RPMI 1640 medium (GIBCO) supplemented with 10% (v/v) heat-inactivated FBS (HyClone), 10 mM HEPES, 1 mM sodium pyruvate, 2 mM L-Glutamine (Thermo Fisher), 100 U/ml penicillin, 100 µg/ml streptomycin at 37°C in the presence of 5% CO2.
For isolation of murine splenocytes, spleens were removed from mice, strained through a 70 µm cell strainer, and homogenized into a single cell suspension using ice cold PBS supplemented with 3% FCS. Red blood cells were lysed by resuspending spleen cells in Red Blood Cell Lysing Buffer (Sigma) and incubating on ice for 10 min. Splenocytes were washed, resuspended in RPMI 1640 medium (GIBCO) supplemented with 10% (v/v) heat-inactivated FBS (HyClone), 10 mM HEPES, 1 mM sodium pyruvate, 2 mM L-Glutamine (Thermo Fisher), 100 U/ml penicillin, 100 µg/ml streptomycin, and used immediately for [32P] cGAMP uptake assays.
Data availability
Raw sequencing data from the CRISPRi screen will be deposited to NCBI GEO prior to final publication.
Acknowledgements
We thank Lily Zhang and Erik Seidel for lab and technical assistance, Hector Nolla and Alma Valeros for assistance with cell sorting, the UC Berkeley High Throughput Screening Facility for preparation of gRNA lentivirus, Adelle P. McFarland for assistance in the isolation of primary cells from mice, Shana L McDevitt for assistance with deep-sequencing, and Raulet lab members, Russell Vance, Michel DuPage, Jeremy Thorner and Andrea Van Elsas for helpful discussions. RDL is supported by a Cancer Research Institute Irvington Postdoctoral Fellowship. DHR is supported by NIH grants R01-AI113041 and R01-CA093678. BG is supported by the IGI-AstraZeneca Postdoctoral Fellowship, JJW is supported by the Pew Scholars Program in the Biomedical Sciences and 1R21AI137758-01. SAZ is supported by grants from the University of Washington/Fred Hutchinson Cancer Research Center Viral Pathogenesis Training Program (AI083203), the University of Washington Medical Scientist Training Program (GM007266), as well as the Seattle ARCS foundation. JEC is supported by the National Institute Health New Innovator Awards (DP2 HL141006), the Li Ka Shing Foundation and the Heritage Medical Research Institute.
This work used the Vincent J. Coates Genomics Sequencing Laboratory at UC Berkeley, supported by NIH S10 Instrumentation Grants S10 OD018174, S10RR029668 and S10RR027303.
Author contributions
RDL, SAZ, and NG performed and analyzed the experiments, LO, SMM, and GEK assisted with the experiments, SW and BGG analyzed the deep-sequencing data and advised on the screen design, RDL, SAZ, BGG, JEC, JW, and DHR designed the experiments, RDL, SAZ, JJW, and DHR prepared the manuscript. All authors critically read the manuscript.
Competing interests
D.H.R. is a co-founder of Dragonfly Therapeutics and served or serves on the scientific advisory boards of Dragonfly, Aduro Biotech, Innate Pharma, and Ignite Immunotherapy; he has a financial interest in all four companies and could benefit from commercialization of the results of this research. SM is, and GK was, an employee of Aduro Biotech.