Building new compartments for unconventional protein secretion from the early and late Golgi membranes

CUPS, a compartment for unconventional secretion of signal sequence lacking proteins, is built during starvation. CUPS, lacking the Golgi specific glycosyltransferases, form by COPI independent extraction of membranes from the early Golgi cisterna, require PI4P for their biogenesis and PI3P for stability. We now show that a PI4P effector Drs2 of the trans-Golgi network, relocates to a new compartment monikered TCUPS because it touches CUPS. Although localized to TCUPS, Drs2 is required for CUPS formation specifically by interacting with Rcy1, and this process is essential for unconventional secretion. Visualizing cells by 4D SCLIM technology revealed that tubules emanating from TCUPS are often collared by CUPS and severed. Incidentally, while CUPS are stable, TCUPS are vesiculated at late stages of starvation. This mirrors the dynamics of the early and late Golgi during conventional protein secretion. TCUPS and CUPS thus emerge as the functional equivalent of early and late Golgi of the conventional secretory pathway, thus representing key compartments in unconventional secretion.


Introduction 33
The problem of how proteins that cannot enter the endoplasmic reticulum (ER)-Golgi pathway of 34 secretion are released to the extracellular space remains a fascinating challenge. This is an There might indeed be different routes for this mode of transport, but we have focussed 49 on the pathway for secretion of Acb1 and Sod1. The export of these proteins has the following 50 essential requirements. 1, Their secretion is triggered upon carbon and nitrogen starvation, and 51 growth in potassium acetate; 2, intracellular production of reactive oxygen species (ROS); 3, the 52  We now report that that Drs2, a PI4P effector that functions as an aminophospholipid 69 flippase and localized at the TGN in growth, is essential for CUPS biogenesis. This requirement 70 in CUPS biogenesis is dependent on its binding partner Rcy1. Our data reveal that during 71 unconventional secretion, cells create a new, TGN-derived compartment that is enriched in Drs2, 72 The multi-spanning transmembrane protein Drs2, aminophospholipid flippase, is a PI4P 95 effector localized at the TGN membranes. We examined location of genomically expressed 96 Grh1-2xmCherry and Drs2-3xGFP in growth and throughout the time course of starvation by 97 confocal live spinning disk microscopy. In growth, Drs2 labelled 4-6 punctae per cell that were 98 often apposed to but not colocalized with Grh1 ( Figure 1A). Upon starvation, Grh1 re-localized 99 to 1-3 larger foci, which we have shown previously to be the CUPS. Curiously, Drs2 also re-100 localized to 1-3 larger foci per cell and in addition displayed faint diffuse localization throughout 101 the cytoplasm ( Figure 1A). The foci of Grh1 and Drs2 were never observed to be stably 102 localized, however, transient co-localization of the foci was frequently observed, particularly 103 early in starvation (26% of cells). The rate of transient co-localization decreased throughout the 104 time course of starvation (10% of cells), however this could be a reflection of the fact that the 105 overall number of Drs2 compartments also decreased throughout starvation. The average number 106 of Drs2 structures per cell decreased from 2.3/cell in the first 45 min to 1.5/cell in the last 30 min 107 of starvation, while the average percentage of cells with no Drs2 structures increased from 7.9% 108 to 13.8% in the same time period (Figure 1). 109 Next, we asked if Drs2 contributed to the process of CUPS formation. We examined 110 localization of Grh1-2xGFP in cells lacking Drs2 and found a strong defect in CUPS biogenesis 111 as observed by Grh1 localized to numerous smaller structures ( Figure 1B). After 2.5-3 hours of 112 starvation, CUPS were still unable to form in the absence of Drs2. TGN upon starvation, we therefore did not expect a role for these players in CUPS formation. 124 Regardless, Gea2, Arl1 and clathrin (clathrin heavy chain or adaptor proteins) mutant strains 125 lacking the function of these proteins were tested and revealed no effect on CUPS formation 126 ( Figure S1). Drs2, via its interaction to Rcy1, also regulates a retrograde pathway required for 127 recycling of exocytic v-SNAREs Snc1 to the TGN. This interaction is also via the C-terminal and therefore could not be tested (data not shown). However, rcy1 cells do not 166 exhibit as many defects associated with loss of Drs2 function, therefore we tested their capacity 167 to secrete unconventional cargoes such as Acb1, and the antioxidants Sod1 and Trx2. Wild type 168 and rcy1 cells were starved for 2.5 hours after which the secreted material was extracted from 169 the cell wall, as described previously (Curwin et al., 2016). The intracellular and secreted 170 fractions were probed by western blot for the various cargoes, Cof1-which is used to monitor cell 171 lysis, and the known cell wall protein Bgl2. Loss of Rcy1 led to a strong defect in release of 172 Acb1, Sod1 and Trx2 without causing release of cytoplasmic content measured by the lack of 173 Cof1 presence ( Figure 3A). Similarly, the v-SNARE double mutant defective in CUPS formation 174 ( Figure 2) was tested and also exhibited a reduction in secretion of Acb1, Sod1 and Trx2 (50-175 60% compared to control cells) ( Figure 3B). Therefore, we can conclude that unconventional 176 secretion in starvation requires Rcy1 (presumably in concert with Drs2) and v-SNARE activity, 177 but the precise function of these players remains to be determined. Upon starvation the Snc1 signal rapidly became diffuse in most cells (less than 5% 201 retained 1-2 faint foci) indicating Snc1 is the preferred v-SNARE for exocytosis. In contrast, the 202 Snc2 signal remained high in all cells, labelling fewer and larger punctate elements ( Figure 4B). 203 Tlg2 also labelled fewer and larger structures immediately upon starvation, in the same manner 204 as Drs2. Both Snc2 and Tlg2 structures could be found transiently co-localized with Grh1 205 ( Figure 4B). In the case of Snc2, this was observed on average in 13% of cells at any particular 206 time point in starvation, whileTlg2 co-localization with Grh1 was more frequent early in 207 starvation (14% of cells early and 7% of cells later in starvation), similar to Drs2. Examination of 208 GFP-Tlg2 with Drs2-3xCherry revealed that they are indeed contained in the same compartment 209 in starvation ( Figure S2). Drs2 could be predicted to be in both the early and late TGN 210 membranes due its function in anterograde and retrograde transport, while Tlg2 is specific to 211 early TGN. We observed a partial co-localization in growth conditions, as expected if this were 212 true ( Figure S2). In starvation, the number of TGN membranes was reduced and the co-213 localization of Drs2 and Tlg2 was greatly increased ( Figure S2). The same was also observed 214 when mCherry-Snc2 and GFP-Tlg2 were tested for their location during starvation. The signal of 215 mCherry-Snc2 was very weak compared to the GFP version, but larger Snc2 structures were 216 observed to co-localize with Tlg2 ( Figure S2). Therefore, starvation induces the formation of a 217 new compartment derived from the early TGN that is enriched Drs2, Tlg2 and Snc2. We now 218 discuss data showing this new compartment contacts CUPS transiently and based on this feature 219 we have called it TCUPS for Touching CUPS. 220

SCLIM reveals the process of CUPS formation in 4D 222
We previously presented the ultra-structure of CUPS using CLEM (correlative light electron 223 microscopy) as a spherical tubulovesicular structure that grows in overall size during starvation Drs2, Tlg2 or Snc2 as the marker of TCUPS. In Figure 7A and Movie 9 an example with Drs2 is 262 provided, where the Drs2-positive membranes inserted into CUPS and fragments were produced. 263 In another remarkable movie, with Tlg2 as the TCUPS marker, the CUPS collar and appear to 264 sever a tubule derived from TCUPS, although direct cutting of TCUPS by CUPS cannot be 265 conclusively stated by this analysis alone ( Figure 7B and Movie 10). In addition to TCUPS, 266 Snc2, and to a lesser extent Drs2, also labelled numerous smaller structures that also often 267 contacted with or were in the vicinity of CUPS ( Figure S3 and Movies S4-8). These are likely 268 vesicles/tubules, as Drs2 itself has been shown to be packaged into vesicles during its activity 269 proteins that function with Drs2 in trafficking at the TGN that includes Gea2, Arl1 and Rcy1, 302 only Rcy1 is involved in unconventional secretion. We also report a requirement for v-SNARE 303 function (Snc1 and Snc2 orthologous pair) in CUPS formation and unconventional secretion. 304 Together, the starving yeast generate a single new compartment from the TGN that contains 305 Drs2, the v-SNARE Snc2 (but not Snc1) and the t-SNARE Tlg2. This compartment that we have 306 called TCUPS contacts Grh1 containing CUPS. 307 308 Building and remodeling TCUPS during unconventional protein secretion. 309

Live cell 4D imaging (SCLIM) has revealed membrane contacts between CUPS and TCUPS. 310
The contacts are transient and highly dynamic, with CUPS membranes often observed to enwrap 311 or encircle TCUPS. We have captured a fascinating event in the contact of CUPS to TCUPS: a 312 tubule emerging from TCUPS is collared by CUPS and appears to be severed. This event is 313 reminiscent of the contact of ER and the endosomes and the fission of the latter compartment 314 only obvious difference being that they do not contain the Golgi specific glycosylating enzymes. 332 We have shown previously that CUPS contain Acb1, but this has only been observed with 333 immunoelectron microscopy. We admit that we do not know if Acb1 enters directly into CUPS. 334

CUPS and TCUPS transiently contact and it is possible that Acb1 is transferred from CUPS to 335
TCUPS during their transient connection. TCUPS, we suggest are the sorting station for the 336 cargo. How this is achieved is also not known. We have seen that CUPS collar a tubule 337 emanating from TCUPS. Snf7 of the ESCRT pathway that is also recruited transiently to CUPS

Construction of N-terminally tagged SNAREs 367
The plasmid pYM-N9 (PCR toolbox) was used to generate a new template vector for PCR-based 368 integration containing the NatNT2 selection cassette, the promoter of Sed5, followed by 2 369 tandem yeGFP. The promoter of Sed5 was amplified from genomic DNA with primers "SacI 370 PrSed5 Fw1": ATAGAGCTCTTACCATGTCCTCCAGAATTACGA and "XbaI PrSed5 Rv1": 371 TCATCTAGAGGGAGTTGTGTGGTATGGTG to generate a 658 bp fragment and was cloned 372 into pYM-N9, replacing the high expression ADH1 promoter. Subsequently a second yeGFP 373 fragment was generated using primers "XbaI ATG yeGFP": 374 TGATCTAGAAAAAATGTCTAAAGGTGAAGAATTATTCACTGG and "EcoRV non-stop 375 yeGFP": TCTGATATCAGGCCTCATCGATGAATTCTCTGTCGGA and cloned downstream 376 of the first yeGFP. Finally, standard S1/S4 primers were used to generate the N-terminal 377 integration fragment targeting the Snc1, Snc2 and Tlg2 loci. Strains were confirmed as positive 378 by microscopy and PCR to confirm the presence of 2 yeGFP. In most cases, however only one 379 GFP integrated and the resulting 1xyeGFP strains were used. In the case of Snc1, 2xyeGFP was 380 initially analyzed and subsequently a single yeGFP version was generated and found to behave in 381 an identical manner as the 2xyeGFP version. SnapGene software (from GSL Biotech, Chicago, 382 IL; available at www.snapgene.com) was used for molecular cloning design.

Cell wall extraction assay 397
Yeast cells were inoculated at a density of 0.003-0.006 OD600/mL in SC medium at 25C. The 398 following day, when cells had reached OD600 of 0.4-0.7 equal numbers of cells (16 OD600 units) 399 were harvested, washed twice in sterile water, resuspended in 1.6 mL of 2% potassium acetate 400 and incubated for 2.5 hours. When growing cells were to be analyzed 16 OD600 units were 401 directly harvested. The cell wall extraction buffer (100mM Tris-HCl, pH 9.4, 2% sorbitol) was 402 always prepared fresh before use and kept on ice. To ensure no loss of cells and to avoid cell 403 contamination in the extracted buffer, 2mL tubes were siliconized (Sigmacote) prior to 404 collection. Cells were harvested by centrifugation at 3000xg for 3 minutes at 4C, medium or 405 potassium acetate was removed and 1.6 mL of cold extraction buffer was added. Cells were 406 resuspended gently by inversion and incubated on ice for 10 minutes, after which they were 407 centrifuged as before, 3000xg for 3 minutes at 4C, and 1.3 mL of extraction buffer was removed 408 to ensure no cell contamination. The remaining buffer was removed and the cells were 409 resuspended in 0.8 mL of cold TE buffer (Tris-HCl, pH 7.5, EDTA) with protease inhibitors 410 (aprotinin, pepstatin, leupeptin (Sigma)) and 10 L was boiled directly in 90 L of 2x sample 411 buffer (lysate). For western blotting analysis, 30 g of BSA (bovine serum albumin (Sigma)) 412 carrier protein and 0.2 mL of 100% Trichloroacetic acid (Sigma) was added to the extracted 413 protein fraction. Proteins were precipitated on ice for 1 hour, centrifuged 16,000xg for 30 414 minutes and boiled in 50 L 2x sample buffer. For detection, proteins (10 L each of lysate or 415 wall fractions) were separated in a 12% polyacrylamide gel before transfer to 0.2 m 416 nitrocellulose (GE Healthcare) for detection by western blotting. For preparation of cell wall 417 extracts for mass spectrometry analysis, no BSA carrier protein was added and the proteins were 418 precipitated with acetone and not TCA. 419

Epifluorescence microscopy 420
After incubation in the appropriate medium cells were harvested by centrifugation at 3,000 g for 421 3 min, resuspended in a small volume of the corresponding medium, spotted on a microscopy 422 slide, and imaged live with a DMI6000 B microscope (Leica) equipped with a DFC 360FX 423 camera (Leica) using an HCX Plan Apochromat 100x 1.4 NA objective. Images were acquired 424 using LAS AF software (Leica) and processing was performed with ImageJ 1.47n software.   washed twice and cultured in 2% potassium acetate for 2.5 hours. The cell wall proteins were 657 extracted from equal number of cells followed by precipitation with TCA ("secreted"). Lysates 658 and secreted proteins were analyzed by western blot and the ratio of the secreted/lysate for the 659 indicated protein was determined and compared to that of wild type in each experiment. 660 Statistical analyses were performed for the indicated unconventional cargo proteins and the 661 reduction in secretion compared to wild type is indicated ± standard deviation. release essential proteins to cell's exterior. A cis Golgi membrane produces small fragments, that 707 do not contain glycosylation enzymes, in a COPI independent manner to synthesize CUPS 708 (magenta). The early TGN produces small membranes to generate a compartment that we have 709 called TCUPS (green). Our data show that tubules emanating from TCUPS are collared by 710 CUPS, which is followed by severing of the tubule. We suggest that these contacts, over a 711 period, lead to the consumption of TCUPS to produce smaller elements (vesicles + tubules). 712 These smaller elements are likely used for delivering essential proteins to other compartments of 713 the cell and release proteins like SOD1 and Acb1 to cell's exterior. This mode of TGN 714 consumption is common to both the conventional and unconventional protein secretion 715 processes. Upon shifting cells to growing conditions, components of the CUPS are delivered by 716 COPI vesicles to the ER, which then traffic the respective components to the Golgi, thereby 717 restoring the Golgi to restart the conventional mode of protein secretion. 718 719 Figure S1. No CUPS defect in cells lacking Gea2, Arl1, Chc1, Apl6, Aps1, Snc1 or Snc2 720 The indicated deletion strains expressing Grh1-2xGFP were grown to log phase and starved for 721 2.5h. Scale bar = 2m 722