Transmembrane protein CD69 acts as an S1PR1 agonist

The activation of Sphingosine-1-phosphate receptor 1 (S1PR1) by S1P promotes lymphocyte egress from lymphoid organs, a process critical for immune surveillance and T cell effector activity. Multiple drugs that inhibit S1PR1 function are in use clinically for the treatment of autoimmune diseases. Cluster of Differentiation 69 (CD69) is an endogenous negative regulator of lymphocyte egress that interacts with S1PR1 in cis to facilitate internalization and degradation of the receptor. The mechanism by which CD69 causes S1PR1 internalization has been unclear. Moreover, although there are numerous class A GPCR structures determined with different small molecule agonists bound, it remains unknown whether a transmembrane protein per se can act as a class A GPCR agonist. Here, we present the cryo-EM structure of CD69-bound S1PR1 coupled to the heterotrimeric Gi complex. The transmembrane helix (TM) of one protomer of CD69 homodimer contacts the S1PR1-TM4. This interaction allosterically induces the movement of S1PR1-TMs 5–6, directly activating the receptor to engage the heterotrimeric Gi. Mutations in key residues at the interface affect the interactions between CD69 and S1PR1, as well as reduce the receptor internalization. Thus, our structural findings along with functional analyses demonstrate that CD69 acts in cis as a protein agonist of S1PR1, thereby promoting Gi-dependent S1PR1 internalization, loss of S1P gradient sensing, and inhibition of lymphocyte egress.

purified human S1PR1 protein alone to validate its activation in the presence of S1P using 81 the GTPase-Glo assay (Fig. 1a). We then tested the effect on S1PR1 of adding the CD69 82 homodimer in the absence of S1P. Remarkably, addition of CD69 alone caused a similar 83 amount of Gi activation as addition of S1P indicating that CD69 functions as a protein 84 agonist of S1PR1 (Fig. 1a). 85

86
To perform structural studies, we mixed lysates from HEK293 cells that independently 87 expressed human CD69 and human S1PR1. The CD69-S1PR1 complex was then 88 incubated with Gαiβ1γ2 heterotrimer and scFv16 23 at 1:1.2:1.4 molar ratio. After gel-89 filtration purification, the resulting complex was concentrated for cryo-EM analysis 90 (Extended Data Fig. 1a). We obtained over 1 million particles from ~4,000 cryo-EM 91 images. The overall structure of the CD69-bound S1PR1 coupled to heterotrimeric Gi was 92 determined at 3.15 Å resolution by 293,516 particles (Extended Data Fig. 1b-f and Table  93 1). The structure shows that one S1PR1 binds one CD69 homodimer and one Gi 94 heterotrimer. It also revealed well-defined features for the canonical seven transmembrane 95 helices (7-TMs) of S1PR1, the Gai Ras-like domain, the Gβ and Gγ subunits and scFv16 96 (Fig. 1b and Extended Data Fig. 2). The intracellular loop 3 (ICL3) and the C-terminus of 97 S1PR1 and the intracellular and extracellular domains of CD69 were not found in the cryo-98 EM map indicating their flexibility in the complex. In contrast, the TMs of the CD69 99 homodimer were clearly defined in the map owing to their interactions with S1PR1 ( Fig.  100 1b and Extended Data Fig. 2; the interacting TM helix is referred to as CD69-a). Because 101 no lipid was supplemented into the protein during the expression and purification, there is 102 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 Another type of intramembrane interaction observed for GPCRs is the formation of either 125 homodimers or heterodimers. The metabotropic glutamate receptor 2 (mGlu2), a Class-C 126 GPCR, employs TM4 to maintain its inactive dimeric state or TM6 to assemble as a 127 homodimer in the presence of its agonist 28 (Extended Data Fig. 4b). The structure of 128 inactive mGlu2-mGlu7 heterodimer shows that TM5 plays a key role in the complex 129 assembly 28 (Extended Data Fig. 4c). Moreover, TM1 of the class D GPCR Ste2 is 130 responsible for engaging the TM1 of another Ste2 to form a homodimer 29 (Extended Data 131 The TM of one protomer of CD69 homodimer interacts with the TM4 of S1PR1 (Fig. 1c). 135 The interface area between TMs is about 600 Å 2 . Structural analysis shows that residues 136 V41, V45, V48, V49, T52, I56, I59, A60 of CD69 mediate its extensive interactions with 137 the receptor (Fig. 2a). Residues L160 4.42 , F161 4.43 , I164 4.46 , W168 4.50 , V169 4.51 , L172 4.54 , 138 I173 4.55 , G176 4.58 , I179 4.61 and M180 4.62 of S1PR1-TM4 contribute to the interaction with 139 CD69 (Fig. 2b). However, the TM of another CD69 does not have any interactions with 140 the receptor and heterotrimeric Gi protein (Fig. 1c). Further structural comparison with the 141 S1P-bound S1PR1-Gi complex indicates that the heterotrimeric Gi proteins in both 142 complexes exhibit a similar state with a RMSD of 0.45 Å. Also, the intracellular regions 143 of the heptahelical domain adopt a similar conformation to accommodate the Gi proteins. 144 This finding implies that S1P and CD69 stimulate the receptor to engage the heterotrimeric 145 Gi proteins in an analogous fashion. 146 147 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 To validate our structural observations, we performed the co-immunoprecipitation (co-IP) 148 assay using S1PR1 and CD69 variants. Compared to the wild-type S1PR1, two mutants 149 (V169 4.51 Y and M180 4.62 Y) present reduced binding to CD69 (Fig. 2c). The TGFa 150 shedding assay showed that these two mutants retained normal activity in response to S1P 151 ( Fig. 2d). We also tested two CD69 double mutations (V48F/V49F and I56F/I59F) for 152 their association with S1PR1. The co-IP results show that the interaction between S1PR1 153 and either mutant is considerably attenuated, thus directly supporting the role of CD69-TM 154 in the complex assembly (Fig. 2c). Moreover, we have purified CD69(V48F/V49F) and 155 CD69(I56F/I59F) individually and mixed with S1PR1 and Gαiβ1γ2 to conduct a GTPase-156 Glo assay (Extended Data Fig. 5). Consistent with the results of our co-IP assays, the 157 activation of Gi proteins in the presence of either variant was decreased (Fig. 2e). To 158 further validate the physiological role of the CD69-S1PR1-Gi complex, we tested the two 159 CD69 variants, for their influence on CD69-mediated S1PR1 internalization in WEHI231 160 B lymphoma cells. In accord with the biochemical data, CD69(V48F/V49F), and 161 CD69(I56F/I59F) were both reduced in their ability to downregulate S1PR1 (Fig. 2f). 162

163
The structures of the S1PR1 complex with its small molecule modulators (including S1P, 164 FTY720-P, BAF312 and ML056) uncover that the TMs 3, 5, 6, and 7 contribute to 165 accommodate the modulators in the orthosteric site 18-22 . In contrast, S1PR1 employs its 166 TM4 to associate with CD69 which functions as a protein agonist for triggering receptor 167 activation. Structural comparison with the inactive state of ML056 bound S1PR1 reveals 168 a unique mechanism of CD69-mediated S1PR1 activation (Fig. 3a). The binding of CD69 169 induces a 4-Å shift at the intracellular end of TM4 causing the residues C167 4.49 , I170 4.52 170 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 and L174 4.56 in TM4 to face TM3 (Fig. 3b). C167 4.49 and I170 4.52 have hydrophobic 171 contacts with the F133 3.41 in TM3, and L174 4.56 pushes the F210 5.47 in TM5 towards the 172 edge of the receptor to form the hydrophobic interactions with W269 6.48 and F273 6.52 in 173 TM6 (Fig. 3c). These interactions trigger the notable movement of TM5 and TM6 allowing 174 the opening of intracellular regions to engage the heterotrimeric Gi proteins ( Fig. 3a and  175 d). Residues A137 3.45 , I170 4.52 , L174 4.56 , F210 5.47 , W269 6.48 and F273 6.52 are conserved 176 among S1PR1, S1PR2 and S1PR3, but not F133 3.41 and C167 4.49 . Remarkably, further 177 comparison shows that the key residues, which are crucial for the S1P binding and receptor 178 activation, present similar conformations in the structures of S1P-bound S1PR1 and CD69-179 bound S1PR1, although S1P and CD69 have different structural natures and completely 180 distinct binding sites in the receptor (Fig. 3e). 181

182
To date, five S1PRs have been identified. These receptors have different tissue 183 distributions, and they also function via distinct kinds of G proteins (including Gi, Gq and 184 G12/13) 3 . Previous work showed that CD69 specifically binds to S1PR1, and it does not 185 associate with S1PR2, S1PR3 or S1PR5 5,6,12 . To dissect the binding specificity of CD69, 186 we carried out the co-IP assays to show a very weak interaction between S1PR2 and CD69 187 (Extended Data Fig. 6a). Although the sequence homology among five S1PRs is high, 188 residues L157 4.51 and L168 4.62 in S1PR2-TM4 are not conserved with those in S1PR1 and 189 are determinants for specific recognition of CD69 (Extended Data Fig. 6b). We speculated 190 that converting these two residues to those in S1PR1 may prompt the interaction between 191 S1PR2 variant and CD69. Our co-IP result clearly shows that S1PR2(L157 4.51 V/L168 4.62 M) 192 could interact with CD69 albeit the interactions are weaker than that between S1PR1 and 193 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 CD69 (Extended Data Fig. 6a). This finding further demonstrates the essential role of 194 S1PR1-TM4 in the CD69 mediated S1PR1 signaling. 195 196 All the known small molecule S1PR1 agonists or antagonists bind to the orthosteric site in 197 the heptahelical domain [18][19][20][21][22] . Interestingly, the CD69 binding site is akin to that of the 198 allosteric agents which attach to receptors 30-32 , although the nature of these agents and 199 CD69 is quite different. The diversity of the allosteric modulator binding sites in GPCRs 200 has been revealed by numerous structures (Extended Data Fig. 7). When the orthosteric 201 site is occupied, the positive allosteric modulator attaches to the receptor and then increases 202 agonist affinity and/or efficacy. CD69 binds to the edge of S1PR1, but it acts as a protein 203 agonist to directly activate the receptor in the absence of any agonists in the orthosteric site. 204 Thus, our finding suggests CD69 is different from other S1PR1 agonists in that it functions 205 via a direct binding to the edge of the receptor. 206 207 It remains unknown whether the antagonist of S1PR1 bound to the 7-TMs will affect the 208 CD69-mediated regulation of S1PR1. We co-transfected S1PR1-GFP and CD69-mCherry 209 into HEK293 cells in a lipid depleted medium. After 24 hours, the fluorescence images 210 show that substantial receptors (~80%) have been internalized with CD69. However, when 211 we added the Ex26, a potent S1PR1 antagonist 33 , into the cells 6 hours after transfection, 212 the images show that just ~50% receptors have been internalized ( Fig. 4a and b). This 213 finding indicates that the CD69-mediated S1PR1 activation could be reversed when the 7-214 TMs pocket is preoccupied by an antagonist. 215 216 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 It has been known that S1P, FTY720-P and CD69, could promote the internalization of 217 S1PR1. However, the mechanisms of S1P-and FTY720-P-mediated internalization appear 218 to be different. While both S1P and FTY720-P activate Gi-signaling, FTY720-P is 219 considered as a β-arrestin-biased agonist, and FTY720-P-induced S1PR1 internalization is 220 b-arrestin-dependent 20,34 . The pathway of S1P-mediated internalization can be b-arrestin-221 dependent or independent 35,36 . To test the mechanism of how CD69 induces the receptor 222 internalization, we performed a fluorescence imaging assay to check the internalization of 223 S1PR1 in the presence of either Gi inhibitor Pertussis toxin (PTX) or β-arrestin inhibitor 224 Barbadin. The plasmids encoding S1PR1-GFP and CD69-mCherry were co-transfected 225 into HEK293 cells in a lipid depleted medium. After 6 hours, we added PTX or Barbadin. 226 On day 2, we calculated the fraction of internalized S1PR1 in each group by fluorescence 227 imaging. The results show that Barbadin does not interfere with CD69-induced receptor 228 internalization ( Fig. 4c and d), but PTX could prevent half of the receptors from 229 internalization ( Fig. 4e and f). Our finding also supports that Barbadin was effective in 230 reducing FTY70-P-induced S1PR1 internalization (Extended Data Fig. 8). Thus, CD69 231 agonism of S1PR1 induces Gi-dependent internalization of the complex. 232

Discussion 234
Our studies provide a model for understanding how the lymphocyte activation marker 235 CD69 controls lymphocyte egress and thus augments adaptive immunity. As an immediate 236 early gene, CD69 is strongly transcriptionally induced in lymphocytes within an hour of 237 exposure to type I IFN, toll-like receptor (TLR) ligands, or antigen receptor engagement 238 5,11,37 . Following induction, CD69 protein engages S1PR1 as an agonist, causing S1PR1 239 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 internalization and loss of the ability to sense S1P gradients. We speculate that even prior 240 to internalization, CD69 disrupts S1PR1's egress promoting function by acting as a high 241 concentration agonist and thus making the receptor 'blind' to S1P distribution. 242 Consistently, our functional analysis reveals that CD69 could not synergize with S1P to 243 trigger S1PR1 activation (Extended Data Fig. 9). 244 245 Previous work has shown the critical importance of correctly distributed S1P and thus 246 correctly localized S1PR1 activation for effective lymphocyte egress 38 . As well as 247 promoting egress, S1PR1, transmits signals needed for maintaining T cell survival 39 and 248 CD69 has been implicated in transmitting signals that influence T cell differentiation 40,41 . 249 Whether the CD69-S1PR1 complex contributes to these signals before undergoing 250 degradation merits further study. GRK2 34,42,43 and dynamin 44 participate in S1PR1 251 internalization in response to S1P. In accord with these factors possibly having a role in 252 CD69-mediated S1PR1 internalization, they have been shown to promote internalization 253 of some receptors independently of b-arrestins 45 . The selectivity of CD69 for S1PR1 is 254 important for allowing activated CD69 + lymphocytes and natural killer cells to employ 255 other S1PRs, such as S1PR2 and S1PR5, to carry out functions without interruption by 256 CD69 12,46,47 . The lack of conservation of key residues that mediate the S1PR1-CD69 257 interaction in TM4 of S1PR2, S1PR5 and the other S1PRs provides an explanation for this 258 selectivity (Extended Data Fig. 6b). In summary, we provide the first example of GPCR 259 activation by interaction in cis with a transmembrane ligand and thereby explain the 260 mechanism of lymphocyte egress shutdown. The structure also offers insights that may 261 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ; https://doi.org/10.1101/2023.02.13.528406 doi: bioRxiv preprint

enable introduction of transcriptionally inducible GPCR switches into CAR-T cells and 262
other engineered cell types. 263 264 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ; https://doi.org/10.1101/2023.02.13.528406 doi: bioRxiv preprint
For co-270 immunoprecipitation assay, the full-length wild-type human S1PR1 fused with a C- After further incubation for 64 hours at 30°C, cells expressing S1PR1-Flag and CD69-285 StrepII were mixed together and resuspended in buffer A (20 mM HEPES, pH 7.5, 150 286 mM NaCl) supplemented with protease inhibitors and then homogenized by sonication. 287 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Tactin XT resin (IBA Cat# 2-5030-025) for batch binding. The resin was washed with 20 291 column volumes (CV) of buffer A containing 0.01% LMNG/0.001% CHS. The protein 292 complex was eluted with 6 CVs of buffer A containing 0.01% LMNG/0.001% CHS and 293 50 mM biotin, followed by a second affinity purification by anti-Flag M2 resin (Sigma-294 Aldrich). The excessive CD69-StrepII was washed off with 20 CVs of buffer A containing 295 0.01% LMNG/0.001% CHS, and the complex was eluted with 5 CVs of 3×Flag peptide 296 (0.1 mg/ml; ApexBio). The eluted protein was further purified by gel filtration using a 297 Superose 6 Increase 10/300 GL column (Cytiva) with 20 mM HEPES, pH 7.5, 150 mM 298 NaCl, 0.001% L-MNG/0.0001% CHS, and 0.0025% glyco-diosgenin (GDN). The peak 299 fractions were collected for complex assembly. 300

301
To assemble the CD69-S1PR1-Gi-scFv16 complex, purified CD69-S1PR1 was mixed with 302 the Gi heterotrimer at a 1:1.2 molar ratio. This mixture was incubated on ice for 1 hour, 303 followed by the addition of apyrase to catalyze the hydrolysis of unbound GDP on ice for 304 1 hour. Then, scFv16 was added at a 1.4:1 molar ratio (scFv16: CD69-S1PR1) followed 305 by 30-min incubation on ice. The mixture was diluted 10-fold by gel filtration column 306 buffer. To remove excess Gi and scFv16 proteins, the mixture was purified by anti-Flag 307 M2 affinity chromatography. The complex was eluted and concentrated using an Amicon 308 Ultra Centrifugal Filter (molecular weight cutoff 100 kDa). The complex was further 309 purified by gel filtration (Superose 6 Increase 10/300 GL) with buffer 20 mM HEPES, pH 310 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ; https://doi.org/10.1101/2023.02.13.528406 doi: bioRxiv preprint 7.5, 150 mM NaCl, 0.001% L-MNG/0.0001% CHS, and 0.0025% GDN. Peak fractions 311 consisting of CD69-S1PR1-Gi complex were concentrated to ~10 to 12 mg/ml for cryo-312 EM studies. 313 314 Cryo-EM sample preparation and data acquisition 315 The freshly purified CD69-S1PR1-Gi-scFv16 complex was added to Quantifoil R1.

Imaging processing and 3D reconstruction 324
A total of 4,239 dose-fractionated image stacks of CD69-S1PR1-Gi complex were 325 collected and subjected to single particle analysis using RELION-3.1 50 and cryoSPARC 326 v3.3 51 . MotionCor2 52 was used for motion correction and dose weighting, CTFFIND-4.1 327 53 for contrast transfer function (CTF) estimation, and crYOLO 54 for particle picking with 328 a general model. A total of 1,113,446 particles were extracted with a pixel size of 1.66 Å 329 in RELION and imported to cryoSPARC. The imported particles were subjected to ab 330 initio model reconstruction and several rounds of alternating 2D classification and 331 heterogeneous refinement. Then 336,669 particles from the best class were re-extracted at 332 full pixel size (0.83 Å) in RELION and imported to cryoSPARC again. Two heterogeneous 333 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ; https://doi.org/10.1101/2023.02.13.528406 doi: bioRxiv preprint refinements were performed in parallel and the resulting particles from the two best classes 334 were combined with duplicates removed. These 293,516 particles were subjected to CTF 335 refinement and Bayesian polishing followed by masked 3D auto refinement. RELION 336 postprocessing was used for sharpening of the final map. 337 338

Model construction and refinement 339
The cryo-EM structure of the S1PR1-Gi bound to S1P (PDB: 7TD3) 19 was used as initial 340 models and manually docked into cryo-EM density map with UCSF Chimera-1.15 55 . The 341 transmembrane helix of CD69 was manually built using Coot-0.9.6 56 . Due to the limited 342 local resolution, the TM of CD69-b was built as polyalanine. The resulting model was 343 subjected to iterative rounds of manual adjustment and rebuilding in Coot and real-space 344 refinement in Phenix-1.16 57 . MolProbity 58 was used to validate the geometries of the 345 model. Structural figures were generated using UCSF Chimera-1.15, ChimeraX-1.5 59 , and 346 PyMOL-2.3 (https://pymol.org/2/). 347 348 GTP turnover assay 349 GTP turnover was analyzed using GTPase-Glo Assay kit (Promega Cat# V7681). Briefly, 350 the purified S1PR1 was first incubated with purified CD69 and/or S1P followed by mixing 351 with isolated Gi protein in an assay buffer containing 20 mM HEPES, pH7.5, 150 mM 352 NaCl, 0.01% LMNG/0.001% CHS, 10 mM MgCl2, 100 µM TCEP, 10 μM GDP and 5 μM 353 GTP. After incubation for 60 min, the reconstituted GTPase-Glo reagent was added to the 354 sample and incubated for 30 min at room temperature. The amount of remaining GTP was 355 assessed by measuring luminescence after adding and incubation with the detection reagent 356 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 for 10 min at room temperature. The luminescence signal was normalized in each case to 357 that of G-protein alone. Data were analyzed using GraphPad Prism 9. 358 359 Co-immunoprecipitation and immunoblotting assay 360 HEK293 GnTIcells were transfected with plasmids encoding CD69-StrepII and S1PR1-361 (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The agonist activity of S1P for the mutant S1PR1s was determined by the TGFα 383 shedding assays 60 . Briefly, three pCAGGS plasmids encoding the human full-length 384 S1PR1 variant (empty vector as negative control), the chimeric Gαq/i1 subunit and alkaline 385 phosphatase-fused TGFα (AP-TGFα) were co-transfected into HEK293 cells using 386 Then, the cells were seeded into a 96-well culture plate and treated with S1P, which was 390 serially diluted in HEPES-containing HBSS with 0.01% fatty acid-free bovine serum 391 albumin. After incubation with S1P, the cell plate was spun, and conditioned media was 392 transferred to an empty 96-well plate. AP reaction solution (120 mM Tris-HCl, pH 9.5, 393 40 mM NaCl, 10 mM MgCl2, and 10 mM p-nitrophenyl phosphate) was added into the 394 cell plates and the conditioned media plates. The absorbance at 405 nm was measured 395 using a microplate reader (Synergy Neo2, BioTek) before and after 2-h incubation at 396 37°C. Ligand-induced AP-TGFα release was calculated as described previously 60 . AP-397 TGFα release signal of empty vector-transfected cells were subtracted from that of 398 S1PR1 cells at the corresponding S1P concentration points. Then, the vehicle-treated 399 AP-TGFα release signal was set as a baseline and ligand-induced AP-TGFα release 400 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made followed by transfection with S1PR1-GFP and/or CD69-mCherry using FuGene 6 406 reagent on the next day. 24 h post transfection, the cells were stained with Hoechst 33342 407 reagent (Thermo Fisher Cat# R37605) and fluorescence images were acquired using a 408 Zeiss LSM 800 microscope system with ZEN imaging software (Zeiss). N-terminal tagged human S1PR1 containing an IRES-hCD4 reporter and either empty 421 vector or constructs encoding wildtype, V48F/V49F or I56F/I59F human CD69 and an 422 IRES-GFP reporter using methods previously described 62 . After 3-5 days, the cells were 423 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The data were collected at the UT Southwestern Medical Center Cryo-EM Facility 429 (funded in part by the CPRIT Core Facility Support Award RP170644). We thank L. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 accession numbers XXXX. All other data needed to evaluate the conclusions in the 450 paper are present in the paper and/or the supplementary materials. 451 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 Figures Chen et al. Fig. 1 Overall structure of human CD69-S1PR1-Gi complex. a, S1PR1-induced GTP turnover for Gi1 in the presence of purified CD69 or S1P. Luminescence signals were normalized relative to the condition with Gi1 only. Data are mean ± s.e.m. of three independent experiments. One-way ANOVA with Tukey's test; ****P<0.0001. Experiments were repeated at least three times with similar results. b, Cryo-EM map of human CD69 bound S1PR1-Gi complex. c, Cartoon presentation of the complex in the same view and color scheme as shown in b. Slab view of S1PR1 from the extracellular side showing that the orthosteric binding pocket is vacant.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 Fig. 2 The binding interface between CD69 and S1PR1. a and b, Detailed interactions between CD69-a and TM4 of S1PR1. Residues that contribute to complex formation are labeled. CD69 is shown in green and S1PR1 in slate. c, S1PR1-Flag and CD69-Strep co-immunoprecipitation assay in transfected HEK293 GnTIcells. d, Dose-response curves of S1PR1WT, S1PR1V169Y and S1PR1M180Y for the TGFα shedding assay using S1P. Data are mean ± s.d. (n=3). e, S1PR1-induced GTP turnover for Gi1 in the presence of purified wildtype and mutant CD69. Luminescence signals were normalized relative to the condition with Gi1 only. Data are mean ± s.e.m. of three independent experiments. One-way ANOVA with Tukey's test; ***P<0.001, ****P<0.0001. Experiments in (c)-(e) were repeated at least twice with similar results. f, Flow cytometric analysis of S1PR1 surface expression on WEHI231 lymphoma cells transduced with S1PR1 and CD69 wild-type and mutant constructs as indicated. From one experiment that is representative of three.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101/2023 Comparison between CD69-bound S1PR1 and ML056-or S1P-bound S1PR1. a, Overall structures of S1PR1 binding with CD69 and ML056. The CD69-bound S1PR1 structure was aligned to ML056-bound inactive S1PR1 (PDB code: 3V2Y). ML056-bound receptor is shown in brown, CD69-bound receptor in blue, and the TM of CD69 in green. The same color scheme is used c-d. b, The movements of TM4 and TM6 of CD69-bound S1PR1 compared with ML056-bound inactive S1PR1. c, Residues involved in the TM movements. d, TM6 movement around W269 6.48 and F273 6.52 . e, Comparison between CD69-bound S1PR1 and S1P bound S1PR1 (PDB code: 7TD3). Residues in the ligand binding pocket are shown. CD69-bound receptor in blue and S1P-bound S1PR1 in cyan. S1P is shown as balls and sticks in yellow.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 Fig. 4 CD69 induced S1PR1 internalization. a, HEK293 cells were treated with 2 µM Ex26 or vehicle for 12 h and imaged using confocal microscopy. Scale bar, 10 µm. b, Quantification of intracellular S1PR1 of the cells in (a). c, HEK293 cells were treated with 20 µM Barbadin for 12 h and imaged for analysis. Scale bar, 10 µm. d, Quantification of intracellular S1PR1 of the cells in (c). e, HEK293 cells were treated with 200 ng/ml pertussis toxin (PTX) for 12 h and imaged for analysis. Scale bar, 10 µm. f, Quantification of intracellular S1PR1 of the cells in (e). Data are mean ± s.e.m. Two-sided Welch's t-test; ns, not significant, **P<0.01, ****P<0.0001. All experiments were repeated at least three times with similar results.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Fig. 2 The cryo-EM density map of CD69-bound S1PR1-Gi complex.

Extended Data
Representative regions of CD69, S1PR1, and Gαi1 are shown. The density map is drawn by ChimeraX.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Fig. 3 Structural comparison between CD69-bound S1PR1 and S1P-bound S1PR1. a, Structures of S1PR1 binding with CD69 and S1P. CD69-bound receptor in blue and S1P-bound S1PR1 (PDB code: 7TD3) in cyan. S1P is shown as balls and sticks in yellow. b, The flipping of F161 4.43 of the structures. c, A model of 2:2 CD69-S1PR1 complex with Gi heterotrimer. The steric clash between the modeled S1PR1 and the N-helix of Gai is indicated by an arrow.

Extended Data
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  CD69 is a dimer in size exclusion buffer, but partially dissociated to be monomer when resolved in SDS-PAGE gel.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Fig. 6 S1PR1 specificity for CD69 binding. a, Co-immunoprecipitation assay in transfected HEK293 GnTIcells from one experiment that is representative of three. b, Sequence alignment of S1PR homologues. The different residues among S1PRs which are crucial for CD69 binding are indicated by blue stars.

Extended Data
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Fig. 8 Barbadin alters the FTY720-P mediated S1PR1 internalization. a, HEK293 cells were treated with 20 µM Barbadin for 2 h followed by vehicle-or FTY720-Ptreatment for 1 h and imaged using confocal microscopy. Scale bar, 10 µm. b, Quantification of intracellular S1PR1 of the cells in (a). Data are mean ± s.e.m. Two-sided Welch's t-test; ****P<0.0001. Experiments were repeated at least twice with similar results.

Extended Data
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  and S1P.
Luminescence signals were normalized relative to the condition with Gi1 only. Data are mean ± s.e.m. of three independent experiments. Two-sided Student's t-test; ns, not significant.
Experiments were repeated at least twice with similar results.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 15, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023