Re: Sulfur starvation-induced autophagy .. 2024년 nature

[문형철]

1. nature  
2. nature communications  
3. articles  
4. article

• Article
• Open access
• Published: 13 August 2024

Sulfur starvation-induced autophagy in Saccharomyces cerevisiae involves SAM-dependent signaling and transcription activator Met4

• Magali Prigent, 
• Hélène Jean-Jacques, 
• Delphine Naquin, 
• Stéphane Chédin, 
• Marie-Hélène Cuif, 
• Renaud Legouis & 
• Laurent Kuras 

Nature Communications volume 15, Article number: 6927 (2024) Cite this article

• 3408 Accesses

• 1 Altmetric

• Metricsdetails

Abstract

Autophagy is a key lysosomal degradative mechanism allowing a prosurvival response to stresses, especially nutrient starvation. Here we investigate the mechanism of autophagy induction in response to sulfur starvation in Saccharomyces cerevisiae. We found that sulfur deprivation leads to rapid and widespread transcriptional induction of autophagy-related (ATG) genes in ways not seen under nitrogen starvation. This distinctive response depends mainly on the transcription activator of sulfur metabolism Met4. Consistently, Met4 is essential for autophagy under sulfur starvation. Depletion of either cysteine, methionine or SAM induces autophagy flux. However, only SAM depletion can trigger strong transcriptional induction of ATG genes and a fully functional autophagic response. Furthermore, combined inactivation of Met4 and Atg1 causes a dramatic decrease in cell survival under sulfur starvation, highlighting the interplay between sulfur metabolism and autophagy to maintain cell viability. Thus, we describe a pathway of sulfur starvation-induced autophagy depending on Met4 and involving SAM as signaling sulfur metabolite.

요약

자가포식은 스트레스, 특히 영양 결핍에 대한 생존 반응에 관여하는 핵심적인 리소좀 분해 메커니즘입니다.

여기에서는 사

카로마이세스 세레비시아의 황 결핍에 대한 자가포식 유도 메커니즘을 조사합니다.

황 결핍은 질소 결핍에서는 볼 수 없는 방식으로

자가포식 관련 (ATG) 유전자의 신속하고 광범위한 전사 유도를 유도한다는 것을 발견했습니다.

이 독특한 반응은 주로 유황 대사의 전사 활성화 인자 Met4에 달려 있습니다. 일관되게, Met4는 유황 결핍 상태에서의 자가포식에 필수적입니다. 시스테인, 메티오닌 또는 SAM의 고갈은 자가포식 흐름을 유도합니다. 그러나, 오직 SAM의 고갈만이 ATG 유전자의 강력한 전사적 유도와 완전한 기능의 자가포식 반응을 유발할 수 있습니다. 게다가, Met4와 Atg1의 결합된 불활성화는 유황 결핍 상태에서 세포 생존을 극적으로 감소시킵니다. 따라서, 우리는 Met4에 의존하고 SAM을 신호 전달 유황 대사 산물로 포함하는 유황 결핍에 의한 자가포식 경로를 설명합니다.

Similar content being viewed by others

Autophagy sustains glutamate and aspartate synthesis in Saccharomyces cerevisiae during nitrogen starvation

Article Open access04 January 2021

Autophagy facilitates adaptation of budding yeast to respiratory growth by recycling serine for one-carbon metabolism

Article Open access07 October 2020

Selectivity of mRNA degradation by autophagy in yeast

Article Open access19 April 2021

Introduction

Sulfur is a chemical element critical to life present in a number of indispensable organic compounds, including the two amino acids methionine and cysteine, the ubiquitous methyl-group donor S-adenosylmethionine (SAM) and the antioxidant tripeptide glutathione (GSH, γ-glutamyl-cysteinyl-glycine). In mammals, methionine is an essential amino acid serving as a precursor for the biosynthesis of SAM, cysteine and GSH1. The clinical features associated with known rare inherited disorders in sulfur amino acid metabolism underline the importance of the sulfur compounds in human health. For instance, homocystinuria, usually caused by mutations in the biosynthetic pathway leading from methionine to cysteine, manifests itself in the form of symptoms affecting multiple organs/systems, including the eyes, the skeleton, the vascular system, and the central nervous system2. Disruption of sulfur amino acid metabolism could cause pathophysiological states through epigenetic dysregulation3. In particular, SAM depletion in mouse liver and human colon cancer cells coincides with global changes in histone H3 lysine 4 trimethylation and affects expression‎ of genes involved in specific cellular processes4,5. Disruption of sulfur amino acid metabolism can also be beneficial. Methionine restriction obtained through dietary limitation or metabolic mutations extends lifespan in different species including yeast6,7,8, worms9,10, and mammals11,12. Although the exact underlying mechanisms are still unknown, several possibilities, including modification of gene expression‎ and induction of antioxidant defense have been proposed13.

Sulfur amino acid metabolism is conserved between yeast and mammals, except that yeast can assimilate inorganic sulfur through the sulfate assimilation pathway14,15. Expression‎ of S. cerevisiae sulfur amino acid biosynthesis genes (referred to as MET genes) are controlled by complex regulatory mechanisms that involve several transcriptional factors, including the transcriptional activator Met416,17,18. MET genes and a number of sulfur compound-specific transporter genes are rapidly and coordinately activated in response to cellular shortage of sulfur amino acids18.

Autophagy is a key nutrient starvation-protective mechanism, in which cytoplasmic components are recycled through degradation in the lysosome (the vacuole in yeast) to maintain pools of basic biosynthetic precursors19,20. Several types of autophagy exist, differing in their mode of substrate delivery to the lysosome/vacuole and their substrate selectivity. Macroautophagy (hereafter referred to as autophagy) involves the sequestering of cytoplasmic material into an isolation membrane, the phagophore, which after expansion and closure generates a double-membrane vesicle called the autophagosome. Fusion of the autophagosome with the lysosome/vacuole leads to degradation of the sequestered material. The process generally ends with the release of breakdown products into the cytosol21. Autophagy is also involved in the removal of harmful cytoplasmic constituents (including proteotoxic aggregates, damaged organelles, and pathogens) as well as in various developmental processes; consequently autophagy dysregulation is associated with a number of pathologies such as cancer, neurodegeneration and microbial infection22.

Autophagy-related (Atg) proteins were first identified in S. cerevisiae23,24,25. More than forty Atg proteins have been described, almost all functionally conserved among eukaryotes26,27. Autophagosome biogenesis is a multi-step process that necessitates cargo recognition and coordinated transfer of lipids from various reservoirs. The core machinery for autophagosome biogenesis include six main functional entities: the Atg1 (ULK1/2 in mammals) protein kinase complex, Atg9-containing lipid vesicles, the class III phosphatidylinositol 3-kinase (PI3K) complex, and the Atg12 and Atg8 (LC3/GABARAP in mammals) conjugation systems28,29,30.

Autophagy regulation takes place at transcriptional and posttranscriptional levels and involves various mechanisms, including epigenetic changes, transcriptional repression and activation, and diverse types of protein modification (phosphorylation ubiquitylation…)31,32,33. In S. cerevisiae, phosphorylation of Atg1 complex by target of rapamycin complex (TORC) 1 was shown to impede autophagy initiation in nutrient-rich conditions, so that TORC1 inactivation is viewed as a key trigger promoting autophagy upon starvation34,35. Transcription of some ATG genes is also upregulated under certain starvation conditions, such as nitrogen starvation31,32. However, the significance of the transcriptional regulation of autophagy has not been much explored and key questions remain unanswered, including: What is the exact raison d’être of this particular level of regulation in the response to starvation? How and to what extent is the autophagic process affected by transcriptional changes?

In this report, we focus on the transcriptional regulation of autophagy in response to sulfur starvation. Our results establish Met4 as a main regulator of autophagy ensuring coordination between the metabolic and autophagic responses to sulfur starvation, with SAM serving as key signaling metabolite.

소개

유황은 생명에 필수적인 화학 원소로서, 두 가지 아미노산인 메티오닌과 시스테인, 어디에나 존재하는 메틸기 기증체인 S-아데노실메티오닌(SAM), 항산화 트리펩타이드인 글루타티온(GSH, γ-글루타밀-시스테인-글리신)을 포함한 여러 가지 필수 유기 화합물에 존재합니다. 포유류에서 메티오닌은 필수 아미노산으로, SAM, 시스테인, GSH1의 생합성을 위한 전구물질 역할을 합니다. 유전성 황 아미노산 대사 장애와 관련된 임상적 특징은 인간 건강에 있어 황 화합물의 중요성을 강조합니다. 예를 들어, 호모시스테인뇨증은 일반적으로 메티오닌에서 시스테인으로 이어지는 생합성 경로의 돌연변이에 의해 발생하며, 눈, 골격, 혈관계, 중추신경계를 포함한 여러 기관/체계에 영향을 미치는 증상의 형태로 나타납니다2. 황 아미노산 대사의 중단은 후성유전학적 조절 장애를 통해 병리생리학적 상태를 유발할 수 있습니다3. 특히, 생쥐의 간과 인간 대장암 세포에서 SAM이 고갈되면 히스톤 H3 리신 4 트리메틸화(histone H3 lysine 4 trimethylation)의 전역적 변화와 일치하며, 특정 세포 과정에 관여하는 유전자의 발현에 영향을 미칩니다4,5. 황 아미노산 대사의 붕괴도 도움이 될 수 있습니다. 식이 제한이나 대사적 돌연변이를 통해 얻어지는 메티오닌 제한은 효모6,7,8, 벌레9,10, 포유류11,12를 포함한 다양한 종의 수명을 연장시킵니다. 정확한 기전은 아직 밝혀지지 않았지만, 유전자 발현의 변형과 항산화 방어의 유도를 포함한 몇 가지 가능성이 제시되었습니다13.

황 아미노산 대사는 효모와 포유류 사이에서 보존되어 있지만, 효모는 황산염 동화 경로를 통해 무기 황을 동화할 수 있다는 점을 제외하면14,15. S. cerevisiae 황 아미노산 생합성 유전자(MET 유전자라고도 함)의 발현은 전사 활성화 인자 Met4를 포함한 여러 전사 인자를 포함하는 복잡한 조절 메커니즘에 의해 제어됩니다16,17,18. MET 유전자와 다수의 황 화합물 특이적 수송체 유전자는 세포 내 황 아미노산 부족에 반응하여 신속하고 협응력 있게 활성화됩니다18.

자가포식은 영양 결핍으로부터 세포를 보호하는 핵심 메커니즘으로, 세포질 성분이 리소좀(효모의 액포)에서 분해 과정을 통해 재활용되어 기본적인 생합성 전구체의 풀을 유지합니다19,20. 여러 종류의 자가포식 작용이 존재하며, 리소좀/액포로의 기질 전달 방식과 기질 선택성이 다릅니다. 거대 자가포식 작용(이하, 자가포식 작용)은 세포질 물질을 격리막인 파고포어로 격리하는 것을 포함하며, 파고포어는 확장 및 폐쇄 후에 오토파지솜이라고 불리는 이중막 소포를 생성합니다. 자가포식체와 리소좀/액포의 융합은 격리된 물질의 분해를 유도합니다. 이 과정은 일반적으로 분해 산물이 세포질로 방출되면서 끝납니다21. 자가포식은 또한 유해한 세포질 구성 요소(단백질 독성 응집체, 손상된 세포기관, 병원체 포함)의 제거와 다양한 발달 과정에도 관여합니다. 따라서 자가포식 조절 장애는 암, 신경 퇴행, 미생물 감염과 같은 여러 병리와 관련이 있습니다22.

오토파지 관련(Atg) 단백질은 S. cerevisiae23,24,25에서 최초로 확인되었습니다. 40개 이상의 Atg 단백질이 설명되었으며, 거의 모든 Atg 단백질이 진핵생물 사이에서 기능적으로 보존되어 있습니다26,27. 오토파지소체 생성은 화물 인식과 다양한 저장소에서 지질의 협응력 있는 전달을 필요로 하는 다단계 과정입니다. 자가포식체 생성의 핵심 기계장치에는 6가지 주요 기능적 실체, 즉 Atg1(포유류에서는 ULK1/2) 단백질 키나아제 복합체, Atg9 함유 지질 소포, 클래스 III 포스파티딜이노시톨 3-키나제(PI3K) 복합체, 그리고 Atg12와 Atg8(포유류에서는 LC3/GABARAP) 결합 시스템이 포함됩니다28,29,30.

자가포식 조절은 전사 및 전사 후 수준에서 이루어지며, 후성유전학적 변화, 전사 억제 및 활성화, 다양한 유형의 단백질 변형(인산화, 유비퀴틸화 등)을 포함한 다양한 메커니즘을 포함합니다31,32,33. S. cerevisiae에서, 라파마이신 복합체(TORC) 1에 의한 Atg1 복합체의 인산화 작용은 영양이 풍부한 조건에서 자가포식 개시를 방해하는 것으로 나타났습니다. 따라서 TORC1의 비활성화는 기아 시 자가포식을 촉진하는 핵심적인 방아쇠로 간주됩니다34,35. 일부 ATG 유전자의 전사는 질소 결핍과 같은 특정 기아 조건 하에서 조절됩니다31,32. 그러나, 자가포식의 전사적 조절의 중요성은 아직 충분히 연구되지 않았고, 다음과 같은 핵심 질문들에 대한 답이 아직 없습니다: 기아에 대한 반응에서 이 특정 수준의 조절이 존재하는 이유는 무엇입니까? 자가포식 과정은 전사적 변화에 의해 어떤 방식으로 그리고 어느 정도까지 영향을 받습니까?

이 보고서에서는 유황 결핍에 대한 자가포식의 전사적 조절에 초점을 맞춥니다. 우리의 연구 결과는 유황 결핍에 대한 대사적 반응과 자가포식 반응 사이의 협응력을 보장하는 자가포식의 주요 조절자로 Met4를 확립하고, SAM을 핵심 신호 대사 산물로 사용합니다.

Results

Sulfur starvation leads to transcriptional induction of ATG genes

Ohsumi and colleagues reported in the early 1990s that sulfur-starved yeast cells accumulate autophagic bodies in the vacuole36; however, sulfur starvation-induced autophagy in yeast has not been further studied since then and the underlying mechanisms remain unknown. To monitor the autophagic flux under sulfur starvation, we utilized the GFP-Atg8 processing assay37, which quantifies the levels of free GFP moiety resulting from the cleavage of GFP-Atg8 delivered to the vacuole. Starvation was performed as follows: cells were grown to exponential phase in synthetic sulfur-free (SF) medium supplemented with methionine as unique sulfur source, collected by filtration, washed, and then transferred into fresh SF-medium. We observed low GFP-Atg8 expression‎ prior starvation but no free GFP (Fig. 1a). Following sulfur depletion, GFP-Atg8 expression‎ increased significantly and free GFP accumulated gradually (Fig. 1a), demonstrating GFP-Atg8 processing. We also performed live-cell microscopy analysis to follow the cellular localization of GFP fluorescence (Fig. 1b). The results showed an overall increase of cellular fluorescence in the first 2 h after sulfur depletion, accompanied by an increase of the number of GFP foci, and followed by progressive accumulation of GFP into the vacuole. Altogether, these experiments confirmed that sulfur depletion results in autophagy induction.

Fig. 1: Sulfur starvation induces autophagy.

a GFP-Atg8 processing assay. WT cells expressing GFP-Atg8 from ATG8 endogenous promoter (Y1408) were grown to exponential phase in sulfur free (SF)-medium supplemented with 0.1 mM Met and starved as described in Methods. Cells were collected before (t0), and 2, 4, 6 and 8 h after starvation. Protein extracts were resolved by SDS-polyacrylamide gel electrophoresis and analyzed by western blot using antibodies against GFP and Pgk1 (loading control). Molecular weight markers are in kDa. Quantification was performed as described in Methods. GFP-Atg8 expression‎ and GFP release are relative to the WT at 8 h. Data are mean of two independent cultures. b Live-cell microscopy. The strain and the starvation conditions are the same as above. Representative images are shown. The graphs indicate the percentage of cells showing GFP dots (left) and accumulating GFP fluorescence in the vacuole (right) Data are mean from two independent experiments using each time two different mutant clones, with in total 200–400 cells scored/ time point. Scale bar, 5 μm. c Cell viability assay. Indicated strains (BY4742, Y1397, Y1424, Y1425, Y1426 & Y1406) were grown in SF-medium supplemented with 0.1 mM Met before sulfur starvation. Viability was determined as described in Methods. Data are mean ± SD of n independent cultures, n = 4 in the case of WT and atg8Δ, and 3 in the case of atg1Δ, atg3Δ, atg7Δ, and atg9Δ. d RNA-sequencing analysis. Cells (BY4742) were grown overnight in SF-medium supplemented with 0.1 mM Met and starved as described in Methods. Cells were collected by centrifugation before (t0) and 40, 80 and 120 min after the shift. RNA was extracted and processed for RNA-sequencing as described in Methods. The upper graph indicates maximum log2-fold change (FC) values (plain blue circle) and mean normalized counts (gray bars) calculated by the DESeq2 package for the 36 ATG genes. The lower graphs represent log2-FC values and normalized counts at the different time points for the top induced ATG genes (log2-FC > 3). Data are mean ± SD (log2-FC) or ± SEM (normalized counts), n = 3 RNA preparations from independent cultures. Source data are provided with this paper.

Full size image

We then asked whether a defect of autophagy would affect the survival capacity of cells deprived of sulfur. For this, a wild-type (WT) strain and several mutants of core components of the autophagy machinery were submitted to sulfur starvation and the number of cells able to resume growth when placed back on medium containing sulfur was measured over time (Fig. 1c). The number of cells increased slightly during the first day of starvation, presumably because the cells did not stop dividing immediately and contained enough sulfur to complete their cycle. Then this number remained quite stable during the 10-day starvation period in the case of the WT strain, whereas it dropped by more than 10-fold in the case of the autophagy-defective strains. These results demonstrate that autophagy is critical for cell survival under conditions of sulfur starvation.

The increase in GFP-Atg8 expression‎ let us hypothesize a regulation at the transcriptional level. To determine how ATG genes respond to sulfur depletion, we performed transcriptional profiling, using RNA-sequencing, before and at several time points after sulfur depletion (Fig. 1d; Supplementary Fig. 1 and Table 5). Out of the 36 ATG genes, 16 (44%) had maximal log2-fold change (LFC) > 2 and only one had negative maximal LFC, indicating a quite homogeneous response. For comparison, out of the 6663 ORFs included in our analysis, 11 % had maximal LFC > 2 and 35% had negative maximal LFC while 45% of the 44 genes belonging to the sulfur amino acid biosynthesis pathway had maximal LFC > 2 and 30% had negative maximal LFC (see Supplementary Fig. 1a and Table 6). RT-qPCR analysis of selected ATG genes gave results in line with the RNA-Seq results (Supplementary Fig. 1c). These analyses demonstrate that the vast majority of ATG genes are upregulated following sulfur depletion.

Even though ATG genes responded quite homogeneously following sulfur depletion, ATG41 was induced noticeably more strongly than the others (Fig. 1d). Several reports have presented evidence supporting a role for Atg41 in autophagy, possibly during autophagosome formation, but its molecular function remains unknown38,39,40. To establish further that Atg41 was required for the autophagic response to sulfur depletion, we monitored GFP-Atg8 processing and cellular localization in atg41Δ cells. GFP-Atg8 was expressed prior starvation in the atg41Δ mutant but expression‎ did not increase following sulfur depletion (Supplementary Fig. 2a). Moreover, only 12% of GFP-Atg8 was processed after 8 hours of starvation in the atg41Δ cells (75% in the WT cells), resulting in very low levels of GFP release. Under the microscope, the two strains showed similar proportions of cells with GFP foci; however, the atg41Δ mutant did not show accumulation of GFP fluorescence in the vacuole contrary to the WT strain (Supplementary Fig. 2b). We also carried out the Pho8Δ60 assay, which measures the alkaline phosphatase (ALP) activity resulting from Pho8Δ60 delivery to the vacuole through autophagy. In the WT strain, Pho8Δ60 ALP activity was strongly increased in the hours following sulfur depletion (Supplementary Fig. 2c). In contrast, there was almost no increase of activity in the atg41Δ strain, similarly to the atg1Δ strain. Therefore, ATG41 inactivation caused an important reduction of the autophagic activity induced by sulfur starvation, thereby demonstrating that Atg41 plays a substantial role in the autophagic response under this starvation condition. The lack of increase of GFP-Atg8 expression‎ in the atg41Δ mutant prompted us to carry out transcription analysis (Supplementary Fig. 2d). Atg41 inactivation caused a decrease in transcription of ATG8 but not ATG1, ATG9 or MET3, a representative sulfur amino acid biosynthesis gene. Therefore, Atg41 seems to be involved in ATG8 transcriptional induction, but our results do not support a general role in the transcriptional induction of ATG genes under sulfur starvation.

Autophagy induction upon sulfur and nitrogen starvation involves distinct mechanisms

To investigate whether sulfur depletion-induced autophagy had distinctive characteristics, we analyzed autophagy flux and ATG gene transcription in cells exposed in parallel to sulfur or nitrogen starvation (Fig. 2). We used here prototrophic strains to exclude interferences from auxotrophic mutations. While GFP-Atg8 induction was similar between both conditions, processing was significantly more efficient under nitrogen depletion compared to sulfur depletion, especially in the first 2 hours (Fig. 2a). Transcription analysis showed more striking differences (Fig. 2b): following nitrogen depletion, only ATG8 was substantially induced, whereas following sulfur depletion, ATG1, ATG8, ATG9 and ATG41 were all strongly induced. Therefore, global induction of ATG gene transcription appears to be a distinctive characteristic of sulfur depletion not observed with nitrogen depletion.

Fig. 2: Autophagy induction upon sulfur and nitrogen starvation involves distinct mechanisms.

a GFP-Atg8 processing assay. A prototrophic strain expressing GFP-Atg8 (Y1727) was grown into SF-medium supplemented with 0.1 mM Met and transferred into SF-medium without sulfur supplementation or into a derived medium lacking ammonium and amino acids except for 0.1 mM Met. Cells were collected before (t0), and 2, 4, and 6 h after starvation. GFP-Atg8 expression‎ and GFP release are relative to the 6-h time point in no sulfur. Data are mean of n = 2 independent experiments. b Transcription analysis. A prototrophic strain (Y1725) was subjected to starvation as above and transcription levels were measured by RT-qPCR. MET3 and MEP2 were used as specific control genes responsive to sulfur and nitrogen starvation, respectively. Fold induction is relative to t0. Data are mean ± SEM from n = 3 independent experiments. Statistical significance between conditions was determined by multiple two-sided t test comparisons using Holm-Sidak method. P values: **p < 0.0001; *p = 0.00037 and 0.00086; ns = 0.51 and 0.17 (from left to right). Source data are provided with this paper.

Full size image

Met4 is essential to induce autophagy in response to sulfur starvation

Using a candidate approach, we next looked for the transcription factor(s) responsible for ATG gene induction upon sulfur depletion. Gcn4 is the only transcription activator that was involved in induction of certain ATG genes in response to amino acid starvation in S. cerevisiae41. On the other hand, the only transcriptional activator known to respond to sulfur availability in S. cerevisiae is Met442. Whereas Gcn4 can bind its target promoters on its own, Met4 recruitment requires the basic helix-loop-helix factor Cbf1 and the two homologous zinc-finger factors Met31 and Met3216,17. Sequences matching the consensus binding sites for Gcn4, Cbf1 and Met31/32 are present upstream of several ATG genes, including ATG1, ATG8, ATG9 and ATG41 (Fig. 3a and Supplementary Table 7). To determine whether Gcn4 or Met4 bind these genes, we performed chromatin immunoprecipitation (ChIP) experiments using strains expressing C-terminal Myc-tagged proteins. In the case of Met4 (Fig. 3b, upper graph), we measured a strong enrichment of ATG1 and ATG41 promoter regions following sulfur depletion (by as much as 6-fold and 20-fold). We also observed a 2- to 3-fold enrichment of ATG8 and ATG9, and no significant enrichment with the negative control IME2. In the case of Gcn4 (Fig. 3b, lower graph), enrichment of ATG1 and ATG41 promoter regions also increased quite strongly following depletion (by 19- and 21-fold, respectively). By contrast, there was no significant enrichment of ATG8 and ATG9. These results strongly support that Met4 and Gcn4 bind ATG1 and ATG41 promoters upon sulfur starvation and suggest association of Met4 with ATG8 and ATG9.

Fig. 3: The transcriptional activators Met4 and Gcn4 bind ATG genes in response to sulfur depletion.

a Upper drawing: schematics depicting the main mechanism of recruitment of Met4 and Gcn4 to their target promoters. Black boxes represent DNA-binding consensus sequences sites. Lower drawing: colored boxes indicate the positions of putative DNA-binding sites for Cbf1 (purple), Met31/Met32 (dark blue) and Gcn4 (orange) upstream of the initiation codon (0) of the indicated genes. Light gray boxes represent ORFs. Black arrows represent the primers used in the ChIP experiment. b Met4 and Gcn4 association with promoters. Cells expressing Myc-tagged Met4 or Gcn4 (Y1526 & Y1578) were grown and starved as in Fig. 1. Association with the indicated genes was measured by ChIP with antibodies against the Myc-tag. DNA was quantified by qPCR using primers specific for the promoter regions of ATG1, ATG8, ATG9, ATG41, and MET16 and for the middle region of IME2 ORF. MET16 and IME2 were used as positive and negative controls, respectively. Data are mean ± SEM (n = 4). Statistical significance compared with t0 was determined by two-way ANOVA with Geisser-Greenhouse correction followed by Dunnett’s multiple comparisons test. P values for Met4: ****p < 0.0001; ***p = 0.0005; **p = 0.0026, 0.0013, 0.0013, 0.0085, and 0.0096; *p = 0.0426, 0.0146, 0.0363, 0.0278 and 0.0136; ns = 0.16, 0.95, 0.07 and 0.11 (from left to right). For Gcn4: **p = 0.0076, 0.0093, 0.0048, 0.0029 and 0.0029; *p = 0.0417, 0.0234, 0.043, 0.0126; ns =0.481, 0.133, 0.556, 0.169, 0.999, 0.222 and 0.094. Source data are provided with this paper.

Full size image

To determine whether Met4 and Gcn4 are required for transcriptional induction of ATG1, ATG8, ATG9 and ATG41, we performed RT-qPCR in met4 and gcn4 single and double mutants subjected to sulfur starvation (Fig. 4a). Inactivation of Met4 or Gcn4 separately altered the transcription profiles of the four ATG genes; however, Met4 inactivation had a stronger effect compared to Gcn4 inactivation, especially in the case of ATG9 and ATG41. The strongest effect was observed in the case of the double met4Δ gcn4Δ mutant, in which maximal induction of ATG1, ATG8, ATG9 and ATG41 were 5-, 3-, 6- and 9-fold lower, respectively, compared to the WT strain. Together with the ChIP data, these results support the hypothesis that Met4 and Gcn4 are both involved in the transcriptional induction of the four ATG genes under sulfur starvation, with Met4 having a preponderant role. The residual transcription in the double mutant may suggest the existence of additional players.

Fig. 4: Met4 is essential to induce autophagy in response to sulfur starvation.

a Transcription analysis. Indicated strains (BY4742, Y1437, Y1439 & Y1571) were grown to exponential phase in SF-medium supplemented with 0.1 mM Met and 0.01 mM SAM and starved as in Fig. 1. Transcription was measured by RT-qPCR on samples collected at the indicated times. CCW12 was used as positive control. Fold induction is relative to WT at t0. Data are mean ± SEM (WT, n = 6; met4Δ, n = 4; gcn4Δ and gcn4Δ met4Δ, n = 3; n are independent experiments). Statistical significance compared with WT was determined by two-way ANOVA with Geisser-Greenhouse correction followed by Dunnett’s multiple comparisons test. P values: ****p < 0.0001; ***p = 0.0006 and 0.0004; ns = 0.45, 0.76, 0.97, 0.99 and 0.99. b GFP-Atg8 processing assay. Indicated strains expressing GFP-Atg8 from ATG8 endogenous promoter (Y1408, Y1611, Y1583 & Y1608) were grown and starved as above. Cells were collected at the indicated times and processed as described in Methods. Molecular weight are in kDa. GFP-Atg8 expression‎ and GFP release are relative to the value in the WT at the 6-h time point. Data are mean ± SEM (n = 3 independent experiments). Statistical significance compared with WT was determined as above. P-values, in each case from left to right: ****p < 0.0001; **p = 0.0012, 0.0058 and 0.0016; ns = 0.99, 0.79, 0.96 and 0.99. c Pho8Δ60 assay. Indicated strains expressing Pho8Δ60 under the control of ADH1 promoter (Y1628, Y1718 or 1720, 1722 or 1765 & Y1723) were grown and subjected to sulfur starvation as above. Vacuolar Pho8Δ60 alkaline phosphatase (ALP) activity was measured as described in Methods and is relative to the WT at the 6-h time point. Data are mean ± SEM (WT, n = 5 and gcn4Δ, met4Δ, gcn4Δ met4Δ, n = 3). Statistical significance compared with WT was determined as above. P values: **p = 0.0070 and 0.0029; ns = 0.73. Source data are provided with this paper.

Full size image

We next used the GFP-Atg8 and the Pho8Δ60 assays to assess the consequences of the transcriptional defects observed in the met4Δ and gcn4Δ mutants on the overall autophagic process. Gcn4 inactivation had no significant effect on neither expression‎ nor processing of GFP-Atg8 (Fig. 4b) and caused only a slight decrease in ALP activity after 6 h of starvation (Fig. 4c). In contrast, Met4 inactivation caused a notable 5-fold decrease in GFP-Atg8 expression‎ as well as free GFP levels (Fig. 4b), suggesting reduced autophagic activity, as further confirmed by the ALP activity measures shown in Fig. 4c. However, the persistence of low amount of free GFP indicated functional autophagy flux. In the gcn4Δ met4Δ double mutant, GFP-Atg8 expression‎ was as low as in the met4Δ single mutant, free GFP levels and ALP activity were further reduced, and GFP-Atg8 processing was significantly diminished (Fig. 4b, c). Similar results were observed in a strain carrying the gcn4Δ and met4Δ mutations in a prototrophic background (Supplementary Fig. 3). We also exposed the met4Δ and gcn4Δ prototrophic strains to nitrogen depletion and assessed autophagy flux using the GFP-Atg8 assay (Supplementary Fig. 4). We observed no significant effect of met4Δ and gcn4Δ mutations on GFP-Atg8 expression‎ and GFP-Atg8 processing compared to the WT strain, suggesting that Met4 role in autophagy regulation under starvation does not extend to other nutrients besides sulfur. Finally, we measured the survival capacity under sulfur starvation of met4Δ, atg1Δ and met4Δ atg1Δ cells (Supplementary Fig. 3c). The three mutants showed a severe loss of viability in the 8 days following starvation compared to the WT strain. Strikingly, the amounts of atg1Δ met4Δ cells did not increase during the first day of starvation, contrary to the other strains. Moreover, the combined atg1Δ met4Δ mutations caused a significantly stronger decrease in survival than the simple atg1Δ mutation, indicating negative epistasis.

Altogether, our results show that Met4 plays a critical role in the autophagic process under sulfur starvation, whereas Gcn4 participates but is dispensable. They also support the notion that the cell capacity to overcome sulfur depletion involves interactions between Met4 and autophagy.

Transcription of ATG genes is induced by depletion of organic sulfur

Depletion of sulfur from the growth medium causes deficiency of several key sulfur-containing metabolites, including methionine, cysteine, SAM and GSH. Therefore, we investigated which metabolite was involved in signaling the induction of ATG genes. To test whether inorganic sulfate has a role by itself, we used the met17Δ mutant, which is blocked in the last step of the sulfur assimilation pathway and is therefore unable to synthesize organic sulfur compounds from sulfate (Fig. 5). Shifting WT cells from a medium containing methionine as sulfur source to a medium containing sulfate caused only moderate and transient activation of ATG1, ATG8, ATG9 and ATG41 (Supplementary Fig. 5). In contrast, transfer of met17Δ cells from methionine to sulfate caused a strong and sustained transcriptional activation of the four ATG genes, indicating that ATG genes respond to the depletion of organic sulfur under sulfur starvation.

Fig. 5: Schematics of the sulfur-containing amino acid biosynthesis pathway showing the metabolic mutants used in this study.

The met17Δ mutation prevents assimilation of sulfate. Combination of met6Δ, mht1Δ and sam4Δ prevents methylation of homocysteine into methionine. The sam1Δ and sam2Δ mutations inactivate the two genes encoding methionine-adenosyltransferase (also known as SAM synthetase), preventing SAM biosynthesis from methionine. The cys4Δ mutation inactivates cystathionine-β-synthase and blocks the transsulfuration pathway ensuring conversion of homocysteine (Hcy) into cysteine. The gsh1Δ mutation prevents GSH biosynthesis.

Full size image

Cysteine or methionine depletion is not a major determinant of ATG gene induction

To achieve selective depletion of individual organic sulfur compounds, we used metabolic mutants (Fig. 5) placed in SF-medium supplemented with chosen sulfur sources. Inactivation of GSH1 blocks the first step of GSH biosynthesis. The gsh1Δ cells were grown prior to depletion in a medium containing in addition to methionine GSH at very low concentration (5 μM) to sustain normal growth while avoiding GSH hyperaccumulation. RT-qPCR analyses showed that none of the ATG genes were activated in the 2-h period following the transfer in the medium with no GSH, whereas strong activation occurred when all sulfur was depleted (Supplementary Fig. 6). Therefore, GSH depletion alone does not induce transcription of the ATG genes.

Inactivation of CYS4 blocks the transsulfuration pathway responsible for methionine conversion into cysteine. Transfer of cys4Δ cells into a medium containing only methionine as sulfur source leads to cysteine depletion and rapid growth arrest (Supplementary Fig. 7b). Compared to the medium with no sulfur, transcription of the four ATG genes was induced with a delay and at significantly lower levels in this medium (Fig. 6a), indicating that cysteine depletion had only a limited effect on ATG gene expression‎. Western blot and fluorescence microscopy analyses of cys4Δ cells expressing GFP-Atg8 showed that cysteine depletion could induce autophagy flux, however GFP-Atg8 expression‎ and free GFP levels were significantly lower in this condition compared to total sulfur depletion (Fig. 6b, c).

Fig. 6: Depletion of cysteine or methionine is not a major determinant of ATG gene induction.

Transcription analysis. Simplified schematics of the sulfur-containing amino acid biosynthesis pathway indicating the steps interrupted in the cys4Δ (a, upper panel) and met6Δ mht1Δ sam4Δ (d, upper panel) mutants. The sulfur compounds becoming depleted after transfer into the indicated starvation media are written in white on gray. Sulfur supplements are marked with a thick colored outline. The cys4Δ strain (Y1559 or 1677) was grown in SF-medium supplemented with 0.1 mM Met and 0.5 mM Cys, and shifted into SF-medium alone or containing 0.1 mM Met. The met6Δ mht1Δ sam4Δ strain (Y1533 or 1534) was grown in SF-medium supplemented with 0.1 mM Cys and Met, and shifted to SF-medium alone or containing 0.1 mM Cys and SAM. Samples were collected at the indicated times. Transcript levels (graphs) were measured by RT-qPCR. Fold induction is relative to t0. Data are mean of n = 4 in (a) or n = 2 in (d); n are independent experiments. Error bars indicate SEM. Statistical significance between conditions in (a) was determined by multiple two-sided t test comparisons using Holm-Sidak method. ***p < 0.0001. b, e GFP-Atg8 processing assay. cys4Δ (Y1558) and met6Δ mht1Δ sam4Δ (Y1535) mutants expressing GFP-Atg8 from ATG8 endogenous promoter were grown and shifted to starvation medium as above. Samples were collected at the indicated times. GFP-Atg8 expression‎ and GFP release are relative to the 6-h time point in the no sulfur condition. Data are mean of n = 3 in (b) or n = 2 in (e); n are independent experiments. Error bars indicate SEM. Statistical significance between the two conditions was determined by multiple two-sided t test comparisons using Holm-Sidak method. P values: ***p < 0.0001; **p = 0.00095, 0.00017 and 0.00016; *p = 0.0072 and 0.0018; ns = 0.15, 0.43 and 0.25 (from left to right). c, f Live-cell imaging by fluorescence microscopy. The strains and starvation conditions are the same as above. Representative images are shown. The graphs indicate the percentage of cells showing GFP dots (left) and accumulating GFP fluorescence in the vacuole (right). Data are mean of n = 2 independent experiments using each time two different mutant clones, with in total 200–400 cells scored/ time point. Scale bar, 5 μm. Source data are provided with this paper.

Full size image

To block methionine biosynthesis, we deleted MET6 together with MHT1 and SAM4. MET6 encodes methionine synthase, the main enzyme that catalyzes the conversion of homocysteine to methionine, using 5-methyltetrahydrofolate as methyl donor. The products of MHT1 and SAM4 can also methylate homocysteine to give methionine but using SAM as methyl donor43,44. The met6Δ mht1Δ sam4Δ cells arrested growth similarly upon transfer in the medium deprived of methionine and the medium deprived of all sulfur supplements (Supplementary Fig. 7c). However, we observed a marked difference in ATG1, ATG8, ATG9 and ATG41 transcription between the two starvation conditions (Fig. 6d). ATG1, ATG8 and ATG41 very poorly induced and ATG9 was not, under methionine starvation, whereas the four genes were strongly induced under total sulfur starvation. In line with these results, GFP-Atg8 expression‎ and free GFP levels in the met6Δ mht1Δ sam4Δ mutant were significantly lower under methionine starvation than in the complete absence of sulfur, even though GFP-Atg8 was processed with similar efficiency in the two starvation conditions (Fig. 6e). At individual cell level, we observed an increase in the number of GFP dots as well as vacuolar accumulation of GFP under methionine starvation, but fluorescence signals were lower compared to sulfur starvation (Fig. 6f).

We further measured the effect of cysteine and methionine depletion on nxxonselective bulk autophagy specifically by using the 3-phosphoglycerate kinase (Pgk1)-GFP processing assay45. As expected, sulfur depletion in the WT cells induced autophagy-dependent processing of Pgk1-GFP (Supplementary Fig. 8a). Depletion of cysteine in the cys4Δ cells induced similar levels of Pgk1-GFP processing as depletion of total sulfur (Supplementary Fig. 8b). By contrast, depletion of methionine in the met6Δ mht1Δ sam4Δ cells induced very weak levels of Pgk1-GFP processing, 5 times lower compared to the depletion of total sulfur (Supplementary Fig. 8c). Curiously, the met6Δ mht1Δ sam4Δ mutant showed in the latter condition a delay in Pgk1-GFP processing and slightly lower percentage of free GFP compared to the other strains.

Altogether, our results indicate that individual depletion of the two sulfur amino acids fails to induce full transcription of ATG genes and triggers an incomplete autophagic response.

SAM depletion strongly activates transcription of ATG genes in a Met4-dependent manner

To deplete SAM, we constructed a sam1Δ sam2Δ double mutant devoid of the two SAM synthetases found in S. cerevisiae. This mutant stopped growing rapidly in the medium which lacked SAM but still contained cysteine and methionine (Supplementary Fig. 7d). ATG1, ATG8 and ATG9 transcription was in this latter growth condition induced as much as in the total absence of sulfur (Fig. 7a). ATG41 transcription was also strongly induced, albeit at a lower level. Therefore, SAM depletion was sufficient to trigger high transcriptional induction of the four ATG genes. Moreover, the sam1Δ sam2Δ cells subjected to SAM starvation showed levels of GFP-Atg8 expression‎, free GFP release and GFP-Atg8 processing comparable to the cells subjected to total sulfur starvation (Fig. 7b). We also observed similar increases between the two starvation conditions in the proportion of cells with GFP foci and in the proportion of cells accumulating GFP in the vacuole (Fig. 7c). Finally, Pgk1-GFP was processed almost as efficiently in the absence of SAM as in the complete absence of sulfur (Supplementary Fig. 8d). Therefore, SAM depletion, in addition to activate the ATG genes, also induces strong autophagic activity, even in conditions where the two sulfur amino acids are not limiting.

Fig. 7: SAM depletion fully activates transcription of ATG genes.

a Transcription analysis. Simplified schematics of the sulfur-containing amino acid biosynthesis pathway indicating the steps interrupted in the sam1Δ sam2Δ mutant (upper panel). The sulfur compounds that become depleted after transfer of the strain into the indicated starvation media are written in white on gray backgrounds. Sulfur supplements are marked with a thick outline. The sam1Δ sam2Δ (Y1504) mutant was grown in SF-medium supplemented with 0.1 mM Cys, Met and SAM, and shifted into SF-medium alone or containing 0.1 mM Cys and Met. Samples were collected at the indicated times. Transcript levels (graphs) were measured by RT-qPCR. Fold induction is relative to WT at t0. Data are mean of n = 2 independent experiments. b GFP-Atg8 processing assay. The sam1Δ sam2Δ mutant expressing GFP-Atg8 from ATG8 endogenous promoter (Y1506) was grown and starved as above. Cells were collected at the indicated times. GFP-Atg8 expression‎ and free GFP are relative to the 6-h time point in the no sulfur condition. Data are mean of two independent cultures. c Live-cell imaging by fluorescence microscopy. The strain sam1Δ sam2Δ GFP-ATG8 was grown and starved as above. Samples were observed at the indicated times. Representative images are shown. The graphs indicate the percentage of cells showing GFP dots (left) and accumulating GFP fluorescence in the vacuole (right). Data are mean of n = 2 independent experiments with in total 200-300 cells scored/ time point. Scale bar, 5 μm.

Full size image

To establish whether Met4 was required upon SAM depletion, we carried out RT-qPCR in a sam1Δ sam2Δ met4Δ triple mutant. The sam1Δ sam2Δ double mutant and the WT strain were also included in the experiment for comparison. Met4 inactivation had a strong effect on the transcription of the four ATG genes upon transfer to the medium containing no SAM (Fig. 8). ATG1, ATG8, ATG9, and ATG41 induction levels were on average 4, 7, 9 and 30 times lower, respectively, in the sam1Δ sam2Δ met4Δ cells than in the sam1Δ sam2Δ cells. We conclude that transcription induction of the ATG genes in response to SAM depletion depends strongly on Met4.

Fig. 8: Met4 has a key role in the transcriptional regulation of ATG genes by SAM.

sam1Δ sam2Δ (1Δ 2Δ), sam1Δ sam2Δ met4Δ (1Δ 2Δ 4Δ) and WT strains (Y1504, Y1601 & BY4742), were grown in SF-medium supplemented with 0.1 mM Cys, Met and SAM, and shifted to SF-medium supplemented with 0.1 mM Cys and Met. Samples were collected before (t0), and 40, 80 and 120 min after the shift. Transcript levels (graphs) were measured by RT-qPCR. SCR1 is used as positive control. Data are mean ± SEM of n = 3 independent experiments. Statistical significance compared with sam1Δ sam2Δ was determined by two-way ANOVA with Geisser-Greenhouse correction followed by Dunnett’s multiple comparisons test. ****p < 0.0001. Source data are provided with this paper.

Full size image

SAM is sufficient to ensure full repression of ATG genes

To establish further the essential contribution of SAM in the transcriptional regulation of ATG genes, we performed depletion-repletion experiments with three combinations of mutations blocking different parts of the sulfur amino acid biosynthesis pathway: SAM synthesis, methionine synthesis, and both cysteine and methionine synthesis (Fig. 9). Cells were first starved for 90 min in SF-medium to induce autophagy and then cysteine, methionine and SAM were individually added. Repletion of the sam1Δ sam2Δ cells with methionine had no effect on the transcription levels of the four ATG genes, whereas repletion with SAM caused a rapid drop so that transcription levels were back to pre-starvation levels after 40 min (Fig. 9a). This drop in transcription levels could be due to SAM but also cysteine given that SAM conversion into cysteine is still possible in sam1Δ sam2Δ. However, addition of cysteine to the triple mutant met6Δ mht1Δ sam4Δ, which can transform cysteine into homocysteine but not into methionine and SAM, had only a limited effect on the transcription levels of the four ATG genes, far from the strong decrease obtained when adding methionine (Fig. 9b). Addition of cysteine to the quadruple mutant met6Δ mht1Δ sam4Δ cys4Δ had also a modest effect; by contrast addition of SAM to this mutant caused a strong decrease in the transcription levels of the four ATG genes (Fig. 9c), demonstrating that ATG gene repression by SAM does not require its conversion into cysteine or methionine. Altogether, these results emphasize the central role of SAM in the transcriptional regulation of autophagy in response to sulfur availability.

Fig. 9: Replenishment with SAM is sufficient to repress transcription of ATG genes in sulfur-starved cells.

The sam1Δ sam2Δ (a), met6Δ mht1Δ sam4Δ (b) and met6Δ mht1Δ sam4Δ cys4Δ (c) mutants (Y1504, Y1533 or 1534 & Y1660) were grown to exponential phase in SF-medium supplemented with 0.5 mM Cys, 0.1 mM Met and 0.1 mM SAM, and shifted to medium containing no sulfur. After 90 min, cells were replenished with 0.5 mM Cys, 0.1 mM Met or 0.1 mM SAM, as indicated. The schematics indicate: the sulfur component added after the 90-min starvation period (thick outline), the components that could be synthetized (thin outline), and components that could not be synthetized due to the mutation and therefore remained depleted (written in white on gray backgrounds). The Samples were collected before the shift (+), following the starvation period (–) and at the indicated time points following repletion. Transcript levels (graphs) were measured by RT-qPCR, and are relative to the 90-min total starvation time point. Fold induction is relative to t0. Data are mean ± SD of n = 3 independent experiments. Source data are provided with this paper.

Full size image

Discussion

We show here that the regulation of autophagy in response to sulfur availability occurs in large part at the transcriptional level. Interestingly, comparison of cells deprived of either sulfur or nitrogen, revealed a clear difference in the way ATG genes respond in each case, in particular ATG1 and ATG9, which are significantly more induced in the sulfur-deprived cells. This difference emphasizes the role of ATG gene transcriptional activation in the autophagic response to sulfur deprivation. By combining autophagy and transcription analyses upon depletion of individual sulfur compounds, we were able to assess the importance of ATG gene transcriptional induction in this particular autophagic response.

Individual depletions of methionine, cysteine and SAM are sufficient to trigger autophagy flux. However, the resulting autophagic activity, as assessed using the GFP-Atg8 and Pgk1-GFP cleavage assays, differ strongly between the three depletion conditions. Strikingly, methionine depletion has almost no effect on ATG gene transcription and autophagic activity, while SAM depletion causes strong transcriptional induction and mimics alone the high levels of autophagic activity obtained upon depletion of all sulfur sources. By contrast, cysteine depletion has only a limited effect on ATG gene transcription but still induces some autophagic activity. Interestingly, levels of GFP-Atg8 expression‎ and GFP release are lower upon cysteine depletion compared to sulfur depletion, while levels of cleaved Pgk1-GFP are similar. This suggests that cysteine depletion would preferentially signal the induction of bulk autophagy whereas sulfur depletion would signal the induction of both bulk and selective autophagy, raising the possibility that the type of cargoes sequestered by autophagosomes might vary depending on the sulfur compound depleted. Moreover, our results as a whole suggest a model in which sulfur nutrients would regulate autophagy activity by affecting different steps. We propose that the two sulfur-containing amino acids would regulate autophagy mostly at a posttranscriptional step, for instance through protein modifications such as already described in the case of nitrogen starvation35. SAM would act above all at the transcriptional level, especially by regulating transcription of core ATG genes required for formation of autophagosome such as ATG1, ATG8 and ATG9, resulting in a boost of autophagy levels.

Only a limited number of transcriptional regulators of ATG genes have been identified in S. cerevisiae, most of which are negative regulators41,46,47,48,49,50,51. We provide in this report evidence that Met4, the master transcriptional activator of S. cerevisiae sulfur amino acid metabolism, is directly involved in the regulation of autophagy. We show that Met4 binds and activates transcription of four ATG genes, including ATG1, ATG8 and ATG9, which encode components fulfilling critical functions in autophagosome initiation. The fact that most ATG genes show an increase in transcription after sulfur depletion suggests that Met4 might bind and directly control more ATG genes than the four described in the present study. Using the GFP-Atg8 and Pho8Δ60 assays, we also show that Met4 inactivation does not block autophagy flux under sulfur starvation but decreases strongly the autophagic degradation activity, presumably because components of the autophagy machinery are still present and functional but in lesser amounts, which should result in smaller autophagosomes52. By contrast, Met4 does not seem to be required under nitrogen starvation conditions, indicating that its role does not extend to nutrient starvation in general. Met4 involvement in the regulation of both autophagy and sulfur metabolism may be seen as a way to allow coordination of the two processes, for example to sustain transformation of the sulfur amino acids generated by autophagy into other most needed sulfur compounds such as SAM and GSH.

How do cells sense and signal to the autophagy machinery that sulfur nutrients are insufficient? We provide evidence of a SAM-sensing and signaling pathway that connects sulfur nutrients availability to autophagy. First, we show that SAM depletion is sufficient to trigger induction of several key ATG genes even when sulfur amino acids are not limiting. Second, addition of external SAM to cells subjected to sulfur starvation rapidly stops transcription of the ATG genes even in genetic backgrounds where SAM cannot be transformed neither into methionine nor into cysteine. Importantly, SAM depletion does not only affect transcription of ATG genes but also leads to induction of the complete autophagic degradation process. Our results also provide the demonstration that SAM synthetase deficiency, which is in human the most common cause of persistent hypermethioninemia53, can trigger autophagy induction.

Several types of mechanisms can be considered to explain how SAM regulates transcription of ATG genes. The positively charged sulfur atom of SAM is attached to three groups that can each be involved in enzymatic reactions: a methyl group, a 3-amino-3-carboxypropyl group and a 5’-deoxyadenosyl group54. The vast majority of reactions using SAM involves transfer of the methyl group55. The 3-amino-3-carboxypropyl group is required for synthesis of spermidine and spermine, and the 5’-deoxyadenosyl group serves as a source of radical intermediates in a number of biosynthetic reactions56,57. However, spermidine supplementation was shown to induce autophagy in several model systems including S. cerevisiae58, indicating an antagonistic effect compared to SAM. Moreover, no radical SAM enzyme has been implicated in signaling so far56. Therefore, it is most probable that the transcriptional regulation of autophagy by SAM involves the methylation of some targets by a SAM-dependent methyltransferase. In particular conditions of induction of autophagy, Tu and colleagues59 identified the carboxyl-methyltransferase Ppm1 as a regulator of Protein Phosphatase 2 A (PP2A) in response to intracellular methionine/SAM levels. However, in our conditions, inactivation of Ppm1 has no effect on ATG gene activation and repression following sulfur depletion and then repletion (Supplementary Fig. 9), suggesting another type of mechanism. SAM could also be sensed by a protein possessing a SAM-binding domain but no enzymatic activity, like in the case of SAMTOR in mammals60. The SAM signal probably acts by interfering with Met4 since no ATG gene activation occurs in a mutant deficient for SAM synthetase and Met4 together. Possible mechanisms include downregulation of Met4 stability or downregulation of its nuclear localization, its recruitment to promoters or its activation capacity61,62,63,64. This could occur through direct methylation of Met4, or indirectly through methylation of the DNA-binding cofactors assisting Met4 recruitment or the ubiquitin-ligase SCFMet30 known to downregulate Met4 activity and stability61,62,65,66. The SAM signal could also interfere with Met4 recruitment to ATG genes by affecting promoter accessibility, for example through changes in the methylation status of histones, as methylation levels of certain histones are linked to SAM levels67. Further investigation will be necessary to unveil the different actors involved in this SAM-sensing and signaling pathways.

Methods

Yeast strains and growth media

S. cerevisiae strains used in this study (Supplementary Table 1) derive from BY4742 and contain auxotrophic markers, unless otherwise stated in the text and figure legends. The individual deletion strains were obtained from the yeast knockout collection. Y1407 was generated by one-step integration at BY4741 chromosomal ATG8 locus of pP1KGFP-ATG8(406)52, which contains 990 base-pairs of ATG8 promoter region in front of GFP-ATG8 CDS. Other GFP-ATG8-URA3 strains were constructed from Y1407 by crossing followed by diploid sporulation and tetrad dissection. Y1526 and Y1578 were obtained by genomic integration into Y1539 of PCR fragments amplified from Met4 and Gcn4 Myc-tagged strains previously published68, followed by sporulation and tetrad dissection. Y1439 was obtained by sporulation and tetrad dissection on YPD medium supplemented with SAM of Y614. The double mutant gcn4Δ::KanMX4 met4Δ::KanMX4 (Y1571) was obtained likewise using a diploid resulting from the crossing of two single mutants. Y1628 was obtained by genomic integration at the PHO8 locus of BY4742 of a PCR fragment containing: PHO8 upstream region, followed by K. lactis URA3 gene, followed by ADH1 promoter fused to a fragment of PHO8 CDS starting at nucleotide position 181 (plasmid was kindly provided by Pr. Fulvio Reggiori). The mutant strains expressing Pho8Δ60 were derived by transformation with deletion cassettes and/or crossing, sporulation and tetrad dissection. Y1504 and Y1506 were generated by two successive crosses. First, Y1490 and Y1491 were crossed with Y1407 to give diploids that produced, after sporulation and tetrad dissection, the haploids MATα sam1Δ GFP-ATG8-URA3 and MATa sam2Δ. Second, these two haploids were crossed and put to sporulate. Y1533 and Y1535 were constructed in three steps. First, Y1489 was crossed with Y1407 to produce, after sporulation of the diploid and tetrad dissection, a strain of genotype MATa met6Δ GFP-ATG8-URA3. In parallel Y1511 was transformed with a mht1Δ::S.kluy.HIS3 PCR fragments amplified from CY51-1A43 to create a strain of genotype MATα mht1Δ::S.kluy.HIS3 sam4Δ::KanMX4. Finally, the two strains were crossed and put to sporulate. Y1660 was derived from Y1533 by genomic integration of a cys4Δ::LEU2 PCR fragment amplified from pUG7369. Y1681 was obtained by sporulation of a diploid resulting from the crossing of BY4741 with BY4742. The prototrophic strains Y1724 and Y1727 were obtained by sporulation of a diploid that was constructed by crossing Y1407 with Ura+ transformants obtained by transformation of BY4720 (lys2Δ0 trp1Δ63 ura3Δ0) with a PCR fragment containing the wild-type URA3 gene. The prototrophic deletion mutants containing gcn4Δ, met4Δ and atg1Δ null alleles were obtained by transformation with PCR fragments and/or crossing, sporulation and tetrad dissection.

Y1813, Y1831, Y1834 and Y1866 were constructed by one-step genomic integration in BY4742, Y1533, Y1677 and Y1504, respectively, of a GFP-hphNT1 cassette amplified by PCR from pYM2570. Y1819 was constructed by one-step genomic integration in Y1813 of atg1Δ::KanMX4 fragments amplified by PCR from Y1397.

YPD contained 1% yeast extract, 2% bacto-peptone and 2% glucose. YNB contained 0.7% yeast nitrogen base, 0.5% ammonium sulfate and 2% glucose. Solid medium also contained 2% agar. Sulfur-free (SF) medium was prepared from individual components based on YNB formula but replacing all sulfate salts by chloride salts (see Supplementary Table 2). YPD and YNB media were sterilized by autoclaving and SF medium by filtration through 0.2-μm-pore-size filters.

Starvation protocol

YPD pre-cultures, inoculated with cells taken from agar plates freshly streaked from −80 °C stocks, were grown to late exponential phase during the day. A 1-mL aliquot was centrifuged, supernatant was eliminated by aspiration and cells were resuspended in 1 mL of fresh SF-medium. The suspension was then used to inoculate at low cell density 25 ml (120 mL for ChIP experiments) of SF-medium supplemented with the appropriate sulfur sources. Cultures were incubated overnight at 30 °C with orbital agitation (180 RPM; INFORS HT Multitron Standart) until OD600 = 1–1.25 (around 107 cells/mL). To change medium, cultures were collected by filtration through Millipore 0.45-μm-pore-size HA-type membranes (except for viability assays; see below), SF-medium was passed through the membranes to wash the cells, and the membranes were transferred to SF-medium and incubation was resumed.

Cell viability assay

Cells from 10-mL of culture were collected by 2-min centrifugation at 4500 RPM using a swinging bucket rotor, the supernatant was eliminated by aspiration, cells were then washed twice with 1 mL of SF-medium and finally transferred into 10 mL of SF-medium. Cell viability, i.e. the ability to divide and proliferate71, was assessed by quantification of colony forming units (CFU). Appropriate dilutions were spread out on YPD-plates (supplemented with SAM when necessary) with glass beads, and Colony-forming units (CFU) per mL were counted after 4 days of incubation at 30 °C.

RNA isolation

Cells from 5 mL of culture at cell density 1–2 ×107/mL were collected by centrifugation and quickly frozen in liquid nitrogen. Cell pellets were kept at −80 °C until RNA extraction. Cells were resuspended in 400 μl of cold AE buffer (50 mM sodium acetate pH 5.3, 10 mM EDTA pH 8, 10% SDS) and the suspension was mixed with 400 μl of cold phenol saturated with 0.1 M citrate buffer pH 4.3 (Sigma). The mixture was incubated for 8 min at 65 °C with agitation, quickly frozen in liquid nitrogen, and incubated for another 8 min at 65 °C with agitation. Tubes were cooled down in ice and centrifuged at room temperature, 12000 rpm, for 10 min. The aqueous phase was then extracted once with 1 vol of acidic phenol:chloroform (1:1) and once with 1 vol of chloroform. Finally, RNA was precipitated with 0.1 vol of 3 M LiCl and 2.5 vol of absolute ethanol. After centrifugation at 12000 rpm and 4 °C for 10 min, RNA pellets were washed with absolute ethanol, air-dried and resuspended in autoclaved milli-Q water.

RNA-sequencing

RNA quality was assessed on Agilent Bioanalyzer 2100 using RNA 6000 Pico kit (Agilent Technologies). 500 ng of RNA were treated with Baseline-ZERO DNAse (Epicenter) prior to ribosomal RNA depletion with the Ribo-Zero Yeast magnetic Kit (Illumina). Directional RNA-seq libraries were constructed using the Stranded Total RNA Preparation kit (Illumina). Libraries were pooled in equimolar proportions and sequenced (2 ×80 bp paired-end) on an Illumina NextSeq550 instrument using NextSeq 550 Mid Output kit.

Demultiplexing was done with the bcl2fastq2 conversion software (v2.18.12) and adapter sequences were removed with Cutadapt (v1.15). Only reads longer than 10pb were considered for further analysis. Between 12,890,040 and 17,482,126 reads were mapped to S288C genome (update R64.2.1) using the TopHat software (v2.1.1). Mapped reads were counted using featuresCounts from the Subread package (v2.0.1). For each sample, between 75 and 80% of sequencing reads were successfully assigned. Differential analysis of read counts across time points was performed with the DESeq2 package72, which uses the Walt test and the Benjamini-Hochberg procedure for hypothesis testing and adjusted p value (padj) calculation.

RT-qPCR

Reverse Transcription (RT)-quantitative PCR was conducted following a two-step protocol. RevertAid reverse transcriptase (Thermo Scientific) was used, with random hexamers for priming, for the RT step, and Luna Universal qPCR Master Mix (Biolabs) and the LightCycler 480 instrument (Roche) for the qPCR step. Data were analyzed with the LightCycler 480 software by performing Absolute Quantification using an in-run standard curve and the Second Derivative Maximum analysis method. Sequence of primers is given in Supplementary Table 3.

Chromatin immunoprecipitation (ChIP)

Cells in 40 mL of culture at 1–2 ×107 cells/mL were fixed with 1% formaldehyde for 15 min at 30 °C. Fixation was stopped by addition of 0.4 M glycine, and cells were collected by centrifugation, washed with cold Tris-Hcl (20 mM, pH 8.0) and kept at -80 °C. Cell pellets were resuspended in 0.5 mL of FA-lysis buffer (50 mM Hepes-KOH, pH7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% deoxycholic acid sodium salt, 0.1% SDS) containing 1 mM Pefablock SC (Roche), and disrupted in the presence of 425-600 μm glass beads (Sigma-Aldrich) using a FastPrep FP120 instrument (Qbiogene). Lysates were recovered and centrifuged (20,100 g, 60 min, 4 °C) to pellets the crosslinked chromatin. Pellets were subjected to sonication in 1.8 mL of FA-lysis buffer using a 500 W Vibra-Cell ultrasonic processor (Sonics and Materials, Inc.) equipped with a 3-mm stepped microtip (8 cycles of 20 s at amplitude 30%, 4 °C), yielding DNA fragments ranging from 100 to 1000 base pairs with an average of 400 base pairs. The insoluble debris was finally eliminated by centrifugation (13,400 g, 10 min, 4 °C), and chromatin extracts were kept at -80 °C.

Immunoprecipitation was performed by overnight incubation at 4 °C of 400 μL of chromatin extract with 2 μg of c-Myc mouse monoclonal antibody (clone 9E10, Santa Cruz Biotechnology, ref. sc-40). Immune complexes were collected with protein A-Sepharose (GE healthcare, ref. 17–0780.) and washed five times in FA buffer containing 0.5 M NaCl, followed by one time in TE (25 mM Tris-HCl pH 8.0, 5 mM EDTA). Immunoprecipitates were eluted from the protein A-Sepharose beads by incubating for 20 min at 65 °C in a solution containing 25 mM Tris-HCl pH7.5, 5 mM EDTA and 0.5% SDS. Eluates were decrosslinked overnight at 65 °C after addition of 1 mg/mL of Pronase (Roche). DNA was purified using GeneJET PCR purification kit (Thermo Scientific), and eluted in 100 μl of Tris-HCl (10 mM, pH 8.0).

DNA was quantified by real time PCR using the LightCycler 480 instrument (Roche) and Luna Universal qPCR Master Mix (Biolabs). Primer sequence is given in Supplementary Table 4. A typical run included 2 μl of immunoprecipitated (IP) and input (total) DNA in duplicate and serial dilutions of one input DNA to create a standard curve and determine PCR efficiency. Data were analyzed with the LightCycler 480 software by performing Absolute Quantification using the in-run standard curve and the Second Derivative Maximum analysis method. To calculate enrichment, we divided the qPCR quantification value for IP by the qPCR quantification value for the corresponding input.

GFP-Atg8 processing assay

Cells from 5 mL of culture at 1–2 × 107 cells/mL were collected by centrifugation and quickly frozen in liquid nitrogen. Cell pellets were kept at -80 °C. Extract preparation and western blotting were carried out according to Guimaraes et al. 73. Cells were broken by shaking in 200 µL of cold 10% trichloroacetic acid (TCA) in the presence of glass beads on a Vortex for 10 min at 4 °C, and the lysates were centrifuged at 13000 RPM for 5 min at 4 °C. Pellets were resuspended in 2X Laemmli sample buffer and heated for 5 min at 95 °C. A volume corresponding to 1.5 × 107 cells was separated on 12% SDS-polyacrylamide gel, transferred onto nitrocellulose membrane, and probed with antibodies against GFP (clones 7.1 and 13.1, Roche ref. 11814460001, dil. 1:1,000) and Pgk1 antibodies (clone 22C5D8, Invitrogen ref. 459250, dil. 1:1,000), followed by peroxidase-conjugated anti-mouse antibodies. Detection was performed with Amersham ECL Western Blotting System, and signals were captured on films. Scans were quantified using ImageJ.

Pgk1-GFP processing assay

Cells from 5 mL of culture at 1–2 × 107 cells/mL were collected by centrifugation and quickly frozen in liquid nitrogen. Total cell protein extracts were prepared by TCA precipitation. Cells were resuspended in cold 10% TCA and broken in a Precellys Evolution (Bertin Technologies, France) bead beater equipped with a Cryolys cooling system (settings: 6 × 15 s at 6,500 RPM, with 30 s of pause between cycles, 4 °C). Cell extracts were recovered, centrifuged at 13,000 RPM for 10 min, 4 °C, and pellets were resuspended in 2X Laemmli sample buffer. Western blots were performed as described above using mouse monoclonal antibodies against GFP (clones 7.1 and 13.1, Roche ref. 11814460001, dil. 1:1,000) and GAPDH (clone 1E6D9, Proteintech ref. 60004-1.-Ig, dil. 1:10,000), goat anti-mouse IgG HRP-conjugated antibodies (Santa Cruz Biotechnology ref. sc-2005, dil. 1:20,000), SuperSignal West Pico PLUS detection reagents (Thermo Fisher Scientific), and the ChemiDoc Imaging System (Bio-Rad Laboratories). Bands were quantified by densitometric analysis using the ImageJ software.

Pho8Δ60 assay

Cells from 5 mL of culture at 1–2 × 107 cells/mL were collected by centrifugation and quickly frozen in liquid nitrogen. Cell lysates were prepared, and alkaline phosphatase (ALP) activity measured, according to Araki et al. 74. Briefly, cells were resuspended in ice‐cold assay buffer (250 mM Tris‐HCl, pH 9; 10 mM MgSO4 and 10 μM ZnSO4) containing 1 mM PMSF and broken by vortexing in the presence of glass beads for 15 min at cold temperature. For the assay, 10 μl of the clarified lysate were added to 500 μl of ice‐cold assay buffer and placed at 30 °C for 5 min before adding 50 μl of 55 mM α‐naphthyl phosphate disodium salt (Merck). The reaction was stopped after 20 min with 500 μl of 2 M glycine‐NaOH, pH 11 and the fluorescence measured at wavelength of 345 nm for excitation and 472 nm for emission using Tecan’s Infinite 200 Pro plate reader. For normalization, the protein concentration of cell lysates was determined using the Bradford reagent from Sigma-Aldrich. The ALP activity corresponds to the emission per the amount of protein in the reaction (mg) and the reaction time (min).

Fluorescent microscopy

Small (1-mL) cell aliquots were transferred into 1.5 mL tubes and left on the bench at 30 °C for 15 min before preparing the slides to allow cells to settle. Cells were observed using a three-dimensional deconvolution microscope (DMIRE2, Leica Microsystems) equipped with a 100x oil objective (HC PL APO 100x/1.4 oil CS, Leica Microsystems) and an incubation chamber set to 30 °C. Images were captured using a 20-MHz CoolSNAP HQ2 CDD camera (Roper Technologies) with a z-optical spacing of 0.2 μm. Z-series acquisition (21 images per stack) and raw image deconvolution was performed using Metamorph sofware (Molecular Devices). Images were processed using ImageJ.

Statistical analyses

Graphs and statistics were performed using GraphPad Prism 8.0 software. The number of independent biological replicates (n) is indicated in the figure legends. Error bars indicate standard error of the mean (SEM) when individual values are shown and standard deviation (SD) when they are not. Statistical significance was assessed as detailed in the figure legends by two-way ANOVA followed by the recommended post hoc test or by multiple two-sided t test comparisons. Statistical significance is given in figure legend.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Numerical data and Western blots are provided as a Source data file with this paper. RNA-sequencing data were deposited in GEO under accession code GSE204733. Source data are provided with this paper.

References

1. Finkelstein, J. D. Methionine metabolism in mammals. J. Nutr. Biochem. 1, 228–237 (1990).

   Article CAS PubMed Google Scholar