Sphingomyelin is critical in organizing phosphoinositide dynamics during phagocytic uptake of Mycobacterium tuberculosis

Infection by Mycobacterium tuberculosis (Mtb) represents a significant global health burden. Mtb invades alveolar macrophages through phagocytosis, and cholesterol has been shown as essential for this invasion, but the roles played by other host lipids have remained unstudied. As such, this study aimed to explore the manners in which host sphingolipids play roles in the early stages of Mtb infection. To this end, we use chemical and genetic perturbations of the sphingolipid biosynthetic pathway and fluorescence microscopy to show that sphingolipids are required for efficient phagocytosis of Mtb. We isolate this dependency to the presence of sphingomyelin at the cell surface. We furthermore show that the turnover dynamics of phosphoinositides at the phagosomal synapse are perturbed upon blockade of sphingolipid synthesis. Finally, we show that the aggregation of membrane-bound receptor at the phagocytic synapse is defective in sphingolipid-deficient cells. This work represents an essential step in understanding the coalescence of host-patho-gen interactions that give rise to Mtb as a global pathogen.

: Cellular sphingolipid levels may be manipulated using chemical and genetic tools.

Sphingolipid depletion has no effect on cell morphology or cell division. 148
What effect does the inhibition of sphingolipid synthesis have on basal cellular functions? 149 We hoped to characterize the changes to morphology and cell growth resulting from myriocin and 150 fumonisin B1 treatments, and Sptlc2 knockout. 151 171

Cells deficient in sphingolipid biosynthesis are reduced in the capacity to phagocytose Mtb. 172
To investigate the role of sphingolipids during phagocytic uptake of Mtb, we used Myr and 173 FB1 to block sphingolipid synthesis in, RAW 264.7, THP-1, DC2.4, and U937 cells. We then 174 infected cells with an mCherry-expressing strain of Mtb to quantify rates of phagocytosis. Briefly, 175 cells were infected at a multiplicity of infection (MOI) of 10 for two hours, after which cells were 176 fixed and stained with Alexa Fluor 488-conjugated phalloidin and with DAPI (to visualize F-actin 177 at cell boundary and nuclei, respectively). Fluorescence microscopy was used to quantify uptake 178 efficiency through automated counting of cell number (by nuclei) and internalized bacteria (red 179 fluorescent particles within green cell boundary). We report uptake efficiency as the ratio of the 180 total number of internalized Mtb particles against the number of identified nuclei, and normalized 181 the uptake rate of treated cells to that of untreated cells to compensate for variability between 182 biological replicates. In all cell models analyzed, we observed a statistically significant reduction 183 in Mtb uptake of approximately 50% following treatment with myriocin, and approximately 40% 184 for fumonisin B1 (Figure 2a-h). Similarly, the knockout of Sptlc2 resulted in approximately 50% 185 of wildtype uptake in both RAW 264.7 and DC 2.4 cells (Figure 2a-h). 186 These results suggest that the maintenance of total cellular sphingolipid levels through ac-187 tive sphingolipid biosynthesis is required for efficient uptake of Mtb across four model cell lines 188 and irrespective of the mechanism of inhibition. We next hoped to identify the exact species re-189 quired for efficient Mtb uptake. 190

Sphingolipid depleted cells are defective in activation of small GTPases Cdc42/Rac1. 211
What is the mechanism through which sphingolipids enable phagocytic uptake of Mtb? 212 Phagocytosis is strictly regulated through a signaling network that involves the dynamic activation 213 and recruitment of both enzymes (such as GTPases, kinases, and phosphatases) and lipidspar-214 ticularly several phosphatidylinositol species (Figure 3a). During the phagocytosis of pathogenic are activated through phosphorylation following engagement of particle and aggregation of recep-217 tor, and this activation builds as a particle is fully engaged by phagocytic receptor. Once a particle 218 is engulfed, the phagocytic signaling cascade is terminated by the dephosphorylation of Cdc42 and 219 Rac1 25 . We hypothesized that inhibiting sphingolipid biosynthesis would prohibit the activation of 220 Cdc42 and Rac1 during phagocytosis. 221 To address this hypothesis, we transfected wildtype and Sptlc2-/-RAW 264.7 macrophages 222 with a fluorescent biosensor consisting of the p21 binding domain of the p21 activated kinase, 223 fused to YFP (PBD(PAK)-YFP)which is recruited from the cytosol to specifically bind to phos-224 phorylated Cdc42 and Rac1 at the site of phagosome formation 26,27 . We then infected cells using 225 the phagocytic model particle Zymosan A at an MOI of 10 for short timepoints before performing 226 live imaging to track the localization of these GTPases. 227 Due to the BSL3 requirements of Mtb, we used the model phagocytic particle Zymosan A 228 in these live imaging experiments. Zymosan A particles are composed of chains of fungus-derived 229 β-glucanwhich engages the mannose receptor on professional phagocytes 28 . We have observed 230 that sphingolipid depletion reduces uptake of Zymosan A to a similar degree as to Mtb (data not 231 shown). Established literature suggests that Dectin-1 is a major phagocytic receptor engaged by We first demonstrate in wildtype RAW 264.7 cells that Cdc42/Rac1 activity builds upon 234 particle recognition and peaks as a particle is engulfed (Figure 3b and Supplementary Movie 1). 235 Quantifying the enrichment of reporter signal to the site of particle contact versus cytosol signal 236 across at least four cells shows a sharp peak of biosensor recruitment as a particle is engulfed 237 (Figure 3c). In contrast, this activation initially builds in Sptlc2-/-cells at a rate indistinguishable 238 from wildtype, but Cdc42/Rac1 activity fails to continue accumulating at later timepoints ( Figure  239 3b+c and Supplementary Movie2). 240 From these results, we conclude that in sphingolipid-deficient cells, the initial activation of 241 Cdc42/Rac1 are unaffectedbut that this activation fails to peak at high enough levels to fully 242  Next, we transfected RAW 264.7 cells with a PI(3,4,5)P3 biosensor consisting of the PH do-285 main of Akt fused to GFP (PH(AKT)-GFP) 32 . We use this biosensor construct to show that 286 wildtype cells begin synthesizing PI(3,4,5)P3 early in phagocytosis, but that this synthesis strongly 287 peaks as a particle is fully engulfed (Figure 4f+g and Supplementary Movie 5). However, Sptlc2 288 knockout cells fail to produce a trailing wave of PI(3,4,5)P3, and thus never reach the levels of syn-289 thesis observed in wildtype cells (Figure 4f+g and Supplementary Movie 6). 290 Together, these findings suggest that the inefficiency of Mtb uptake in sphingolipid-defi-291 cient cells is due to defects in phagocytic signaling resulting from defective phosphoinositide dy-292 namics. At early stages of phagocytosis, there are minimal differences between wildtype and 293 Sptlc2-/-cellsbut the knockout cells fail to reach critical thresholds of phosphoinositide signal 294 molecules (particularly PI(3,4,5)P3) to drive the full propagation of phagocytosis and thereby fail to 295 uptake particles. 296

Sphingomyelin in particular is required for efficient uptake of Mtb. 311
Which sphingolipid confers this phenotype? Because sphingomyelin is highly enriched to 312 the exoplasmic leaflet of the plasma membranes, we hypothesized that this lipid species is respon-313 sible for the observed phenotype of reduced Mtb uptake in cells depleted of total sphin-314 golipids 17,22,33 . Because sphingomyelin is a terminal branch of the sphingolipid biosynthetic path-315 way, and its synthesis at the plasma membrane is mediated by the enzyme sphingomyelin synthase 316 1 (Sgms1), cells may be depleted of this specific lipid species through CRISPR/Cas9-mediated 317 knockout of this gene. We isolated two U937 Sgms1 knockout clones and assessed their capacity 318 to phagocytose Mtb through infection as described above. We found that, despite minor clonal 319 differences, these cells are significantly reduced in their capacity to uptake Mtb to a similar degree 320 as to cells deficient in the biosynthesis of all sphingolipids (Figure 5a+b). 321 Together with the above results, these data suggest that sphingomyelin is particularly cru-322 cial for the phagocytic uptake of particles such as Mtb.

327
Representative confocal images were collected at 63x, and non-confocal quantitative images were collected at either  To address this hypothesis, we transiently transfected wildtype and Sptlc2-/-RAW 264.7 344 macrophages with a GFP-conjugated Dectin-1 construct, and assessed the co-localization of this 345 fluorescent protein with Zymosan A particles at short timepoints. We find that Dectin-1 in wildtype 346 cells strongly accumulates at the site of contact with Zymosan A, and that this contact area spreads 347 with the engulfment of the particle (Figure 6a+b). In contrast, Sptlc2 knockout cells fail to accu-348 mulate Dectin-1 at the site of engagement (Figure 6c+d). We believe that this failure to aggregate 349 phagocytic receptor suggests that fluidity at the phagocytic synapse is significantly perturbed by 350 the depletion of sphingolipids.

ATATTAGTACAAAATACGTGACGTAGAAAGTAATAATTTCTTGGGTAGTTTGCAG-467
TTTTAAAATTATGTTTTAAAATGGACTATCATATGCTTACCGTAACTTGAAAGTATTT 468 For transient transfection, Raw264.7 cells were seeded into a SensoPlate™ 96-Well Glass-533 Bottom Plates at a cell number of 1.0x10 4 cells/well 24 hours prior transfection. Transfection was 534 conducted with Lipofectamine 3000 (Thermo Fisher) according to manufacturer's instructions. 535 Briefly, 100ng of plasmid DNA and 0.2µl of P3000 reagent were mixed in 5µl Opti-MEM and 536 0.33µl Lipofectamine 3000 was mixed in 5µl Opti-MEM. Both reagent dilutions were mixed to-537 gether and allowed to sit for 15 min. The mix was then added directly to the appropriate well and 538 mixed by swirling the plate gently. 539 For imaging, the growth medium was aspirated and replaced with RPMI without phenol 540 red containing Zymosan A Bioparticles A595 (Thermo Fisher) at an MOI of 20. The 96-well plate 541 were centrifuged at 1000rpm for one minute. The cells were imaged with the SDC microscope 542 (Nikon) and 3-4 different spots per well were imaged at the same time for 60 min with capture 543 intervals of 15 sec. The parameters for imaging were kept the same for each sample. 544

Live imaging analysis 545
Of live cell images, individual images were selected for further analysis if they captured a 546 cell bound to a zymosan particle while maintaining a healthy morphology throughout the imaging 547 timecourse.. Imaging analysis was performed using FIJI software 43 . The parameters for image pro-548 cessing were kept constant when comparing different data sets. The area of the phagosome was 549 selected, cropped and opened in a new window. For the original image, the threshold was set au-550 tomatically on the GFP-channel and for the threshold of the GFP-phagosome area, the threshold 551 of the whole cell was multiplied by 1.5 to ensure that only an increase in signal is counted (GFP-552 threshold cell x 1.5 = GFP-threshold phagosome area). The threshold area at different time points 553 was measured by the analyze-measure function of FIJI and the results were exported into Excel (Microsoft). The values for the GFP-threshold of the phagosome area were divided by the values 555 for the GFP-threshold of the whole cell at the corresponding time points. 556

Dectin-1 transfection and co-localization 557
Raw264.7 WT and Sptlc2 -/cells were seeded on glass coverslips in 12-well plates and 558 grown to 70-90% confluency overnight. The next day, cells were transfected with pMXsIP Dec-559 tin-1-GFP using Lipofectamine 3000 (Invitrogen) according to the manufacturer's instructions. 560 After 24 hrs, cells were washed three times with ice-cold PBS, kept on ice for 5 min, and inocu-561 lated with Alexa Fluor 594-conjugated zymosan in chilled serum-free DMEM at a ratio of 10 562 beads per cell. After centrifugation for 5 min at 250 x g and 4°C, cells were incubated for 5 min 563 at 37°C, washed three times with ice-cold PBS to remove unbound zymosan, then fixed for 15 564 min with 4% paraformaldehyde. 565