Amino acid starvation inhibits autophagy in lipid droplet-deficient cells through 1 mitochondrial dysfunction

Lipid droplets are ubiquitous organelles in eukaryotes that act as storage sites for neutral lipids. Under normal growth conditions they are not required in the yeast Saccharomyces cerevisiae. However, recent works have shown that lipid droplets are required for autophagy to proceed in response to nitrogen starvation and that they play an essential role in maintaining ER homeostasis. Autophagy is a major catabolic pathway that helps degradation and recycling of potentially harmful proteins and organelles. It can be pharmacologically induced by rapamycin even in the absence of lipid droplets. Here, we show that amino acid starvation is responsible for autophagy failure in lipid droplet-deficient yeast.  It not only fails to induce autophagy but also inhibits rapamycin-induced autophagy. The general amino acid control pathway is not involved in this paradoxical effect of amino acid shortage. We correlate the autophagy failure with mitochondria aggregation and we show that amino acid starvation-induced autophagy is restored in lipid droplet-deficient yeast by increasing mitochondrial biomass physiologically (respiration) or genetically (REG1 deletion). Our results establish a new functional link between lipid droplets, ER and mitochondria during nitrogen starvation-induced autophagy.

Introduction 4 cells is accompanied by the formation of membrane tangles initially observed in electron 67 microscopy [7,8]. The autophagy failure in lipid droplet-deficient cells was originally 68 interpreted as a possible requirement for fatty acids generated by TAG or STE hydrolysis in 69 autophagosome membrane biogenesis [6,8]. This view was, however, challenged by two 70 observations. Firstly, rapamycin, a potent inhibitor of TORC1 (Target of Rapamycin complex 71 1), was very efficient at inducing autophagy in the lipid droplet-deficient mutant, a result 72 indicating that autophagy is not strictly dependent on lipid droplets [7,9]. Secondly, nitrogen 73 starvation -but not rapamycin -caused an increase in fatty acid biosynthesis and an alteration 74 of phospholipid composition that proved detrimental to autophagy [7,9]. In fact, the partial 75 inhibition of fatty acid synthesis by cerulenin, or experimental elevation of the PtdCho 76 (phosphatidylcholine)/PtdIns (phosphatidylinositol) ratio, restored autophagy in response to 77 nitrogen starvation and also eliminated the membrane tangles [7,9]. Thus, the inability of the 78 lipid droplet-deficient mutant to buffer free fatty acid into lipid droplets or its inadequate 79 phospholipid composition appeared more likely to be responsible for the autophagy defect. Restoration of autophagy by cerulenin or by PtdCho/PtdIns clamp was, however, only partial 81 when compared to the intense autophagy triggered by rapamycin, suggesting that the 82 mechanism of impairment had not been fully addressed. To gain access to this mechanism, a 83 better definition of the origin of the problem appeared to be required. One question that arose 84 was whether the absence of autophagy upon nitrogen starvation was a simple failure to 85 activate the process or if it reflected an inhibitory mechanism capable of blocking the 86 autophagic response to other kinds of stress (i.e. rapamycin). Another aspect of the situation 87 that had not been examined is the fact that nitrogen starvation is a composite stress in a 88 medium lacking both ammonium ion and amino acids. Previous studies in lipid droplet-89 proficient yeast have shown that ammonium ion deficiency and amino acid deficiency induce 90 autophagy through different mechanisms and have different requirements for the General 91 Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200551/892531/bcj-2020-0551.pdf by guest on 20 September 2020 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20200551 5 Amino Acid Control (GAAC) pathway [15]. To obtain a better definition of the situation 92 encountered in lipid droplet-deficient yeast, we sought to examine separately the effects of 93 ammonium ion or amino acid deficiencies on autophagy.

94
Here, we show that ammonium ion deficiency induces autophagy in lipid droplet-95 deficient yeast (RS4Δ strain). In contrast, amino acid deficiency not only fails to activate 96 autophagy but also inhibits the responses to ammonium deficiency and to rapamycin. Amino 97 acid starvation also alters mitochondrial morphology. We report that this phenotype is 98 dependent upon the carbon source present in the medium since it is observed in glucose-99 grown cells, but not in respiratory medium. Finally, we show that this phenotype can also be 100 suppressed by deletion of the REG1 gene which is necessary for carbon catabolite repression 101 in glucose-grown cells. cultures in selective medium were inoculated in YPD medium and maintained in exponential 122 growth phase for at least five generations; cells were harvested by centrifugation, washed with 123 distilled water and autophagy was induced by inoculation in YNB medium without 124 ammonium sulfate and amino acid (-NH 3 -aa) for nitrogen starvation, or YNB medium without 125 amino acid (+NH 3 -aa) for amino acid starvation. For ammonium starvation experiments, (-126 NH 3 + aa), yeast cells were inoculated in YNB without ammonium sulfate but supplemented 127 with histidine (10mg/L), tryptophane (10mg/L), leucine (50mg/L), methionine (10mg/L) and 128 lysine (15mg/L) (-NH 3 +aa). Autophagy induction with rapamycin was achieved with 129 rapamycin 200nM.

130
To obtain rho 0 mutants, a culture of RS4Δ grown to saturation on YPD medium 131 containing ethidium bromide (10μg/mL) was diluted and re-inoculated at low density in the 132 same medium. The resulting respiratory deficiency was confirmed by a complete lack of 133 growth on YPLactate medium which is an obligatory respiratory medium. Microscopy was performed on a FV1000 Olympus confocal microscope using a 100X oil 179 immersion lens (NA 1.40) coupled to a 2.0X numerical zoom (0.08μm per pixel). The 180 excitation wavelengths for GFP or FM4-64 were set to 488nm and 543nm respectively. Image 181 acquisition and conversion were performed separately for green (520nm) and red (603nm) 182 channels and processed with the Olympus Fluoview version 4.1 software. Transmission 183 images were acquired with differential interference contrast optics. Results and discussion 188 Amino acid starvation inhibits autophagy in lipid droplet-deficient cells 189 Previous reports have shown that rapamycin efficiently induced autophagy in RS4Δ cells,190 whereas nitrogen starvation failed to do so [7,9]. Here, both stimuli were combined in our 191 experiments and autophagy completion was assessed by the GFP-Atg8 cleavage assay [21]. 192 We observed that rapamycin-induced autophagy was completely inhibited by nitrogen  The mitochondrial respiratory chain plays a role when autophagy is induced by amino acid 250 starvation [33]. We sought to examine mitochondrial shape following autophagy induction in 251 RS4Δ cells. Using plasmid pSu9-GFP as a reporter of mitochondrial shape, we observed that 252 mitochondrial morphology in RS4Δ cells was differentially affected in response to amino acid 253 deficiency and/or ammonium limitation. As shown in Fig.3A, 90% of the cells in the RS4Δ 254 strain contained compact mitochondrial aggregates, clustered in one or two spots, when the 255 cells were challenged by amino acid starvation for 2 hours. This shape is strikingly similar to 256 the mdm (mitochondrial distribution and morphology) mutants, some of which are respiratory 257 deficient [34]. The mitochondrial aggregates are clearly excluded from the vacuole ruling out 258 the possibility that they result from mitophagy (Supplemental Figure 2). In response to 259 ammonium starvation, RS4Δ displayed a less severe phenotype since aggregated 12 mitochondria were observed in only 43% of the cells. None of these starved conditions altered 261 the typical tubular morphology of mitochondria in wild type cells (Fig. 3A). To substantiate 262 the correlation between mitochondria integrity and autophagic ability in RS4Δ, we generated 263 a RS4Δ rho 0 mutant and analyzed autophagy in response to nitrogen starvation. Rho 0 mutants 264 lack all the components encoded by mitochondrial DNA and are thus unable to produce 265 mitochondrial ATP. Interestingly, mitochondrial DNA ablation exacerbated the phenotype of 266 RS4Δ since RS4Δ rho 0 proved unable to accomplish autophagy even in response to 267 ammonium starvation (Fig. 3B). Importantly, autophagy was still observed in the rho 0 lipid 268 droplet-proficient strain in response to ammonium starvation as reported before [33]. We that autophagy in response to ammonium ion deficiency was obliterated by the addition of the 283 ROS scavenger N-acetylcysteine (Fig. 3D). The same concentrations of N-acetylcysteine did 284 not abolish autophagy in lipid droplet-proficient cells (Fig. 3D). Together, the data are in keeping with our previous report indicating that RS4Δ displays a high GSH (glutathione 286 reduced form)/GSSG (glutathione oxidised form) ratio [38]. This increased ROS buffering 287 capacity in RS4Δ cells might dampen the burst of ROS production that is required for 288 autophagy [36]. The fact that the addition of H 2 O 2 facilitated autophagy but did not fully 289 restore it suggests that ROS production was not the only contribution of mitochondria to the 290 autophagic process. This prompted us to turn to experimental protocols that allow full 291 restoration of mitochondrial function. 294 We then examined whether autophagy could be restored in RS4Δ cells by culture conditions 295 that favour mitochondria proliferation and activity. For this purpose, cells were grown in 296 lactic acid media. In this fully respiratory medium, lipid droplet-deficient cells showed 297 essentially the same growth-rate as wild type cells, although the growth rate of both strains 298 was reduced by 50% as compared to fermentative conditions. Taking into account this growth 299 delay, autophagy was monitored over 8 hours. Western blot analysis indicated the RS4Δ 300 mutant was able to execute autophagy in response to the different combinations of ammonium 301 and/or amino acid starvation (Fig. 4A, left panel). Comparison with the wild type strain (Fig.   302 4A, right panel) revealed that the autophagy at 8 hours was nearly as intense in the RS4Δ 303 mutant. A delayed autophagy could be observed in the mutant at 4 hours in the groups lacking 304 amino acids (Fig. 4A, compare left and right panels). To obtain additional evidence, GFP-305 Atg8-expressing RS4Δ cells were examined by fluorescence microscopy after 6h of 306 ammonium and/or amino acid starvation in respiratory growth conditions. In 70-100% of the 307 cells, the GFP signal was either restricted to the vacuole (vacuolar) or encompassing the 308 vacuole and the cytosol (uniform) (Fig. 4B). We interpreted these results as indicative of a 309 completed autophagy or an autophagic flux in progress, respectively (Fig. 4B). For comparison, effective autophagy in response to rapamycin in glucose medium resulted in 90% 311 of the cells with GFP entirely or partially vacuolar, whereas autophagy blockade in glucose 312 medium lacking both ammonium and amino acids yielded only 10% of that score (Fig. 4B).

313
Taken together, our results indicated that autophagy was restored to near wild type levels in 314 lactate-grown RS4Δ cells. Consistent with these results, the tubular morphology of the 315 mitochondria was not affected by amino acid shortage in respiratory medium (Supplemental 316 Figure 3). Because previous studies have indicated that RS4Δ cells generate ER membrane 317 tangles during autophagy failure [7-9], we examined the effect of respiration on this process.

318
To observe accurately the ER membrane, the ER-resident protein Sec63 was epitope-tagged 319 with GFP (Supplemental Figure 4 and 5). Using Sec63-GFP, fluorescence membrane tangles 320 could be observed in nearly all the cells upon nitrogen starvation in glucose medium (Fig. 4C,   321 left panel), whereas they were extremely rare in respiratory medium (Fig. 4C, right panel).

322
Therefore, respiration was also effective in resolving this other dysfunction of nitrogen-323 starved RS4Δ cells. In fermentable medium, mitochondria production is limited by the 324 catabolic repression pathway, which is highly dependent on the phosphatase regulatory 325 subunit Reg1 [39]. Deletion of the REG1 gene in RS4Δ cells efficiently abolished the 326 inhibitory effect of amino acid starvation on autophagy induced by ammonium starvation in 327 glucose-grown cells (Fig. 4D). Moreover, amino acid starvation per se proved capable of 328 inducing autophagy in this mutant (Fig. 4D). To further substantiate this conclusion, we 329 analyzed GFP-Atg8 localisation in the RS4Δ-reg1Δ by fluorescence microscopy. As 330 illustrated in Figure 4B, upon ammonium or amino acid starvation, GFP fluorescence reached 331 the vacuole in more than 80% of the cells. Thus, REG1 deletion restored the autophagic 332 turnover of GFP-Atg8 in glucose medium. Because REG1 deletion causes both an increase in 333 mitochondria biomass and ATP production by respiration even in fermentative medium, we 334 sought to separate the structural and metabolic contributions of mitochondria to autophagy restoration. As illustrated in Fig. 4E, RS4Δ-reg1Δ rho 0 cells remained capable of activating 336 autophagy in all conditions of ammonium and/or amino acid starvation. A slight decrease in 337 GFP-Atg8 cleavage (compare Fig. 4D and E) may represent the loss of ROS contribution to 338 autophagy restoration, although it also lies within the range of western blot variability.

340
In this report, we clarify the origin of the autophagy failure observed in lipid droplet-341 deficient cells by showing that it is singularly due to a toxic effect of amino acid starvation.

342
Despite our best efforts, the amino acid starvation sensor that triggers this toxic effect has not 343 been identified, although we were able to exonerate the GAAC pathway. We have been more 344 successful at identifying the role of aggregated mitochondria in the mechanism of autophagy 345 inhibition. This new information complements previous reports indicating that a rise in fatty 346 acid synthesis was responsible for the autophagy blockade in lipid droplet-deficient cells 347 challenged by nitrogen starvation [7,9]. It could be anticipated that excess fatty acid was not 348 the only cause of autophagy inhibition, because pharmacological correction of fatty acid 349 synthesis merely achieved partial restoration of autophagy [7,9]. In contrast, we show here 350 that autophagy can be restored to near wild type level by increasing mitochondria activity and 351 biomass (i.e. respiration or ablation of the REG1 gene). Interestingly, although we observed 352 that ROS production along the respiratory chain facilitated autophagy, the RS4Δ-reg1Δ rho 0 353 cells remained capable of activating autophagy upon amino acid starvation. Therefore, it 354 would appear that mitochondria biomass and structural integrity are more meaningful than 355 respiration for the restoration of autophagy. A noticeable correlate of autophagy inhibition in 356 lipid droplet-deficient cells is an accumulation of tangled ER membranes whose mechanism 357 remains to be elucidated [7,9]. We would like to speculate that these tangled ER membranes