Local Fatty Acid Channeling into Phospholipid Synthesis ...

[Pages:39]Article

Local Fatty Acid Channeling into Phospholipid Synthesis Drives Phagophore Expansion during Autophagy

Graphical Abstract

Authors

Maximilian Schu? tter, Patrick Giavalisco, Susanne Brodesser, Martin Graef

Correspondence

martin.graef@age.mpg.de

In Brief

Expansion of the phagophore membrane during autophagy is driven by localized de novo phospholipid biosynthesis.

Highlights

d Mechanisms of membrane assembly during autophagosome biogenesis were discovered

d Acyl-CoA synthetase Faa1 channels fatty acids into de novo phospholipid synthesis

d Newly synthesized phospholipids specifically drive phagophore expansion

d Description of the phospholipid composition of purified autophagic membranes

Schu? tter et al., 2020, Cell 180, 135?149 January 9, 2020 ? 2019 Elsevier Inc.

Article

Local Fatty Acid Channeling into Phospholipid Synthesis Drives Phagophore Expansion during Autophagy

Maximilian Schu? tter,1 Patrick Giavalisco,2 Susanne Brodesser,3 and Martin Graef1,3,4,* 1Max Planck Research Group of Autophagy and Cellular Ageing, Max Planck Institute for Biology of Ageing, 50931 Cologne, Germany 2Metabolomics Core Facility, Max Planck Institute for Biology of Ageing, 50931 Cologne, Germany 3Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, 50931 Cologne, Germany 4Lead Contact *Correspondence: martin.graef@age.mpg.de

SUMMARY

Autophagy is a conserved catabolic homeostasis process central for cellular and organismal health. During autophagy, small single-membrane phagophores rapidly expand into large double-membrane autophagosomes to encapsulate diverse cargoes for degradation. It is thought that autophagic membranes are mainly derived from preformed organelle membranes. Instead, here we delineate a pathway that expands the phagophore membrane by localized phospholipid synthesis. Specifically, we find that the conserved acyl-CoA synthetase Faa1 accumulates on nucleated phagophores and locally activates fatty acids (FAs) required for phagophore elongation and autophagy. Strikingly, using isotopic FA tracing, we directly show that Faa1 channels activated FAs into the synthesis of phospholipids and promotes their assembly into autophagic membranes. Indeed, the first committed steps of de novo phospholipid synthesis at the ER, which forms stable contacts with nascent autophagosomes, are essential for autophagy. Together, our work illuminates how cells spatially tune synthesis and flux of phospholipids for autophagosome biogenesis during autophagy.

INTRODUCTION

Macroautophagy, hereafter referred to as autophagy, is a central catabolic process essential for cell homeostasis under basal and stress conditions with broad implications for aging and age-associated diseases (Dikic and Elazar, 2018; Hansen et al., 2018; Leidal et al., 2018). A defining feature of autophagy is the de novo formation of specialized transient double-membrane organelles, termed autophagosomes, which enable cells to target an unprecedented scope of substrates for degradation (Gatica et al., 2018). Autophagosome biogenesis is initiated by hierarchical assembly

of conserved autophagy protein machinery at phagophore assembly sites or pre-autophagosomal structures (PASs) in close association with the endoplasmic reticulum (ER) (Hurley and Young, 2017; Mizushima et al., 2011; Noda and Inagaki, 2015). Here, precursor membranes likely comprised of tubulovesicular Atg9 compartments and COPII vesicles are fused to form a small single-membrane phagophore (or isolation membrane) (Davis et al., 2016; Ge et al., 2013; Ge et al., 2017; Ge et al., 2014; Graef et al., 2013; Ishihara et al., 2001; Karanasios et al., 2016; Longatti et al., 2012; Mari et al., 2010; Puri et al., 2018; Shima et al., 2019; Suzuki et al., 2013; Yamamoto et al., 2012; Young et al., 2006; Zoppino et al., 2010). Subsequently, the phagophore enters a stage of dramatic membrane expansion and, within minutes, engulfs its substrates as a large cup-shaped structure. The cargo is topologically isolated from the cytosol upon closure of the phagophore, which generates the inner and outer vesicle of the double-membrane autophagosome. The outer autophagosomal membrane fuses with lysosomes in mammals or vacuoles in yeast and plants exposing the inner vesicle and its enclosed cargo to resident hydrolases for degradation (Abada and Elazar, 2014; Lamb et al., 2013; Wen and Klionsky, 2016). It is thought that a number of organelles including the ER, Golgi, endosomes, mitochondria, plasma membrane, and lipid droplets contribute mainly preformed membranes to the formation of autophagosomes (Axe et al., 2008; Dupont et al., 2014; Geng et al., 2010; Hailey et al., 2010; Longatti et al., 2012; Nascimbeni et al., 2017; Puri et al., 2013, 2018; Ravikumar et al., 2010; Shpilka et al., 2015; van der Vaart et al., 2010). However, it has been a major challenge to conclusively determine at which stage and to which extent these organelles may supply membranes directly to autophagosome biogenesis. In particular, the mechanisms and membrane sources assembling the vast majority of autophagic membranes during the rapid expansion of the phagophore into the large autophagosome have not been clearly defined.

The ER, in conjunction with mitochondria, is the central organelle for de novo lipid synthesis. Specified by dedicated protein tethers, the ER network forms membrane contact sites (MCSs) with virtually all other organelles (Lahiri et al., 2015; Wu et al., 2018). The ER is also in close spatial and physical contact with nascent autophagosomes, an apparent universal feature of autophagy (Axe et al., 2008; Biazik et al., 2015; Go? mez-Sa? nchez et al.,

Cell 180, 135?149, January 9, 2020 ? 2019 Elsevier Inc. 135

2018; Graef et al., 2013; Hayashi-Nishino et al., 2009; Suzuki et al., 2013; Yla? -Anttila et al., 2009). The conserved Atg2-18 complex physically tethers phagophores to the neighboring ER organizing ER-phagophore MCSs in yeast and mammals (Chowdhury et al., 2018; Go? mez-Sa? nchez et al., 2018; Kotani et al., 2018; Zheng et al., 2017). Similar to other membrane tethers, Atg2 mediates phospholipid transport between opposing membranes in vitro (Maeda et al., 2019; Osawa et al., 2019; Valverde et al., 2019), and, in the absence of Atg2, cells accumulate small Atg8-positive phagophores close to the ER and fail to form autophagosomes (Kotani et al., 2018; Tamura et al., 2017; Velikkakath et al., 2012). These findings suggest that phospholipid transfer across the ER-phagophore MCS may promote autophagic membrane assembly. However, if and how a net transfer of ER lipids to the phagophore is achieved in vivo and whether it may involve the de novo phospholipid synthesis activity of the ER is unknown.

One level at which cells regulate lipid fluxes is constituted by a conserved protein family of long-chain acyl-CoA synthetases (ACSs), which catalyzes the activation of fatty acids (FAs) by ATP hydrolysis-driven thioesterification of free FAs with coenzyme A (CoA). ACSs form complex networks composed of peripheral and transmembrane proteins (6 in yeast and 13 in mammals). Based on their dynamic and differential subcellular localization, ACSs are thought to channel activated FAs into diverse pathways including lipid synthesis, membrane editing, protein acylation, and vesicular fusion (Black and DiRusso, 2007; Coleman, 2019; Digel et al., 2009; Grevengoed et al., 2014; Watkins et al., 2007). Here, we define a pathway composed of the conserved ACS Faa1 and de novo phospholipid synthesis required for autophagy. We show that Faa1 localizes to nucleated phagophores and channels activated FAs locally into de novo phospholipid synthesis to specifically drive autophagic membrane expansion. Together, our work provides mechanistic insights into how cells rewire their phospholipid metabolism for autophagy and cellular function.

RESULTS

Acyl-CoA Synthetases Faa1 and Faa4 Localize to Forming Autophagosomes To test for a potential role in autophagy, we examined the ACS network in budding yeast for spatial association with autophagic membranes. We genomically modified yeast strains to coexpress C-terminally green fluorescent protein (GFP)-tagged ACS family members Faa1, Faa3, Faa4, or Fat1 along with N-terminally red fluorescent protein (mCherry)-tagged Atg8, which is covalently attached to autophagic membranes and serves as a highly selective marker (Duronio et al., 1992; Faergeman et al., 1997; Huang et al., 2000; Johnson et al., 1994a; Johnson et al., 1994b; Kirisako et al., 1999). Using live-cell imaging, ACSs showed differential spatial organization in untreated cells consistent with previous data; Faa1 and Faa3 mainly localized to the plasma membrane (PM) and ER, whereas Faa4 and Fat1 were found predominantly on lipid droplets (Figure 1A) (Black and DiRusso, 2007; Kohlwein et al., 2013; Zou et al., 2003). Next, we induced autophagy by either inhibiting target of rapamycin complex I (TORC1) with rapamycin or nitrogen starvation (starvation). As previously described, we morphologically distinguished

punctate Atg8 structures and Atg8-positive ring-like autophagosomes, which were clearly labeled by PAS and autophagosome marker Atg1 (Figures 1A, 1B, and S1A) (Black and DiRusso, 2007; Graef et al., 2013). Notably, Faa1 and Faa4 significantly localized to Atg8 puncta and quantitatively marked autophagosomes, whereas Faa3 and Fat1 were barely detectable on any autophagic structures (Figures 1A, 1B, and S1A). These data reveal differential but redundant association of ACSs with autophagic membranes. Consistently, Faa1 and Faa4, but not Faa3 or Fat1, were degraded in an autophagy-dependent manner under these conditions similar to autophagy components Atg8 and Atg1, which bind to autophagic membranes (Figure S1B) (Huang et al., 2000; Kirisako et al., 1999; Kraft et al., 2012).

To decipher at which stage of autophagosome biogenesis ACSs are recruited, we focused on Faa1 during rapamycininduced autophagy. Genetic analysis revealed that Faa1 colocalization with Atg8 puncta depended on the integrity of the hierarchical assembly of the core autophagy machinery (Figure 1C) (Suzuki et al., 2007). Next, we resorted to time-lapse imaging in living cells. We found that a cycle of autophagosome biogenesis, from the formation of an Atg8 punctum until vacuolar fusion, took on average 9 min (Figures 1D?1F), consistent with prior estimates (Xie et al., 2008). Faa1 became detectable around 1 min downstream of Atg8 puncta formation and progressively accumulated on autophagic membranes when both signal intensity and circumference of Atg8 puncta started to increase, consistent with the expansion of nucleated phagophores (Figures 1D?1F). Notably, Faa1 remained on autophagosomes, even after Atg8 signals decreased due to deconjugation from the outer autophagosomal membrane, until autophagosomes abruptly disappeared upon vacuole fusion (Figures 1D?1F) (Kirisako et al., 2000). Together, our genetic and imaging analyses reveal that ACS Faa1 and Faa4 localize to forming autophagosomes and indicate that Faa1 recruitment to nucleated phagophores downstream of the core autophagy machinery spatiotemporally coincides with phagophore expansion.

Faa1 Is Critical for Autophagy Flux Our observations raised the possibility of a direct function of ACSs in autophagy. In yeast, ACSs and de novo fatty acid synthesis by fatty acid synthase (FAS) constitute two parallel pathways producing activated FAs (acyl-CoA; Figure 2A). In contrast, mammalian FAS generates free FAs that have to be activated by downstream ACSs (Black and DiRusso, 2007; Leibundgut et al., 2008; Vance et al., 1972). To test for conserved ACS functions for autophagy in yeast, we inhibited FAS using the specific inhibitor cerulenin (Vance et al., 1972). We examined autophagy flux by employing the established GFP-Atg8 assay, in which cleaved free GFP is a measure of the turnover of GFP-Atg8-marked autophagosomes in vacuoles (Shintani and Klionsky, 2004). Single ACS gene-knockout cells, Dfaa1, Dfaa3, Dfaa4, or Dfat1, possessed wild-type (WT)-like autophagy flux independent of FAS activity, consistent with the functional redundancy suggested by autophagosomal localization of Faa1 and Faa4 (Figures S2A and S2B). Precluding their functional analysis, Dfaa1D faa4 and Dfaa1Dfaa3Dfaa4 deletion cells showed markedly reduced viability during starvation and FAS-inhibition, demonstrating the essential nature of FA activation (Figure S2C). To

136 Cell 180, 135?149, January 9, 2020

A

untreated

X-3GFP

Atg1

Faa1

Faa3

Faa4

Fat1

rapamycin X-3GFP mCherry-Atg8 merge

Atg8 structures colocalizing (%)

Atg8 puncta/cell

B

puncta AP 100

75 50 25

0 Atg1 Faa1 Faa3 Faa4 Fat1

C

Faa1-negative Faa1-positive 1.5

1 0.5

0 atg1 atg9 atg14 atg5

Figure 1. ACS Faa1 and Faa4 Localize to Forming Autophagosomes (A) Fluorescence imaging of indicated strains in logphase (untreated) or after rapamycin treatment (1 h). Arrowheads: autophagosomes. Arrows: Atg8 puncta. (B) Quantification of data shown in (A). Data are means ? SD (n = 3; R 150 cells/strain). AP, autophagosomes. (C) Faa1-3GFP recruitment to mCherry-Atg8 puncta in indicated deletion strains. Data are means ? SD (n = 3; 105 cells/strain). (D and E) Time-lapse analysis of cells expressing mCherry-ATG8 and FAA1-3GFP after rapamycin treatment (3 h). Images are maximum intensity Z stack projections. (D) Arrowheads indicate an autophagosome biogenesis event labeled by Atg8 (red), Faa1 (green), or both (yellow). Representative timelines of autophagosome biogenesis in (E). Data are means ? SD (n = 3; 75 events). (F) Illustration of Faa1 recruitment to nascent autophagosomes. Dashed lines indicate cell boundaries. Scale bars, 1 mm. See also Figure S1.

D rapamycin

1'

2'

3'

4'

5'

6'

7'

8'

9'

merge

mCherry- Faa1Atg8 3GFP

E

time (min)

F

0 1 2 3 4 5 6 7 8 9 10

Atg8 Faa1

0.9?0.2 6.7?0.2 0.8?0.1

~1 min

~7 min

overcome this limitation and to focus our analysis on Faa1, we engineered Dfaa3Dfaa4 cells to express FAA1 (FAA1 cells hereafter) or a FAA1 variant encoding for plasma membrane-tethered Faa1 (PM-FAA1 cells) from the genomic FAA1 locus (Figure 2B). Both, FAA1 and PM-FAA1 expression fully supported cell survival upon FAS-inhibition during starvation (Figure S2D). As expected, GFP-tagged Faa1, but not PM-Faa1, accumulated on autophagic membranes during rapamycin-induced autophagy (Figure 2B). With this tool in hand, we examined the functional role of Faa1 on autophagic membranes for autophagy. Whereas FAA1 and PM-FAA1 cells displayed indistinguishable autophagy responses during starvation, strikingly, PM-tethered Faa1 failed to support autophagy flux in starving cells upon FAS-inhibition when monitored by western blot or cytological analysis of vacuolar GFP-signals as a consequence of autophagy (Figures 2C and S2E). Next, we measured autophagic turnover of cytosolic cargo by following vacuolar cleavage of cytosolic tandem GFP (2GFP). Similar to autophagy-deficient Datg7 cells, PM-FAA1

cells displayed a drastic defect in bulk

autophagy compared with WT or FAA1

9'

cells during starvation and FAS-inhibition

(Figure 2D). Using electron microscopy,

we observed that PM-FAA1 cells were

strongly impaired in accumulating inner

autophagosomal vesicles in hydrolase-

deficient vacuoles (Dpep4) in contrast to

FAA1 cells after 3 h of starvation and

vacuole

FAS-inhibition (Figure 2E) (Takeshige

et al., 1992). In addition, when we lowered

the glucose concentration during starva-

~1 min

tion in the absence of cerulenin, we identified metabolic conditions in which auto-

phagy flux is predominantly dependent

on Faa1, because we found significantly

reduced autophagy flux in PM-FAA1 cells compared with

FAA1 cells (Figure 2F). To exclude the possibility that these se-

vere autophagy defects were caused by general vacuole

dysfunction in PM-FAA1 cells, we tested the integrity of the mul-

tivesicular body (MVB) pathway, using integral PM Mup1-GFP as

an established substrate (Menant et al., 2006). WT, Datg7, FAA1,

and PM-FAA1 cells showed significant vacuolar GFP-signal and

conversion of Mup1-GFP to free GFP during starvation and FAS-

inhibition, demonstrating that the MVB pathway and vacuolar

activity were intact in these cells in contrast to MVB pathway

mutant Dvps4 (Figure S2F) (Teis et al., 2008). Taken together,

these data show that Faa1 fulfills critical functions for autophagy.

Human ACSL4 Localizes to Autophagic Membranes and Drives Autophagy in Yeast We asked whether ACS function in autophagy is evolutionarily conserved. High complexity of mammalian ACS systems, consisting of up to 13 protein isoforms, poses a major challenge to

Cell 180, 135?149, January 9, 2020 137

A

glucose

free fatty acids acetyl-CoA

S CoA

ATP

ACC

CoA

AMP PPi

ACS

FAS cerulenin

(yeast-specific)

acyl-CoA

B

Faa1

1

mCherry-Atg8 Faa1-GFP PM-Faa1-GFP

699

PM-Faa1

468 581 2

Slm1

699

ATP/AMP FA

S CoA

GFP Cherry GFP Cherry

C

FAA1

PM-FAA1

-

+

-

+ cerulenin

0 3 6 0 3 6 0 3 6 0 3 6 time (h)

2GFP-Atg8

free/total GFP (%)

GFP

**

40

*

30

20

**

10

* **

0

free/total GFP (%)

D

wt atg7 FAA1 PM-FAA1 0 3 6 0 3 6 0 3 6 0 3 6 time (h)

2GFP (cytosolic)

GFP

40

***

***

30

20

10

***

******

0

***

autophagic bodies/cell

E

FAA1pep4 PM-FAA1 pep4 4 3 2 *** 1 0 FAA1 PM-FAA1

G

merge

human ACSL4v1-GFP

mCherry-Atg8

ACSL4v1-positive Atg8 puncta (66?2%) and AP (88?1%)

F

FAA1PM-FAA1

2GFP-Atg8

free/total GFP (%)

GFP

40

30

20

***

10

0 0.01 % Glc

PM-FAFAAA11ACSL4v1

H

2GFP

-Atg8

GFP

rel. autophagy flux

1.25

***

1

***

0.75

0.5

0.25

0

Figure 2. Autophagy Depends on Local Fatty Acid Activation by Evolutionarily Conserved ACSs (A) Two pathways produce activated fatty acids (acyl-CoA) in S. cerevisiae. ACC, acetyl-CoA carboxylase; ACSs, acyl-CoA synthetases; FAS, fatty acid synthase. (B) PM-tethered Faa1 does not localize to autophagic membranes. Schematics of Faa1 and PMFaa1 proteins including ATP- and AMP- and FAbinding motifs. Right panel, fluorescence imaging of WT cells coexpressing mCherry-ATG8 and plasmid-borne FAA1-GFP or PM-FAA1-GFP (1 h rapamycin treatment). Lower panels show magnifications of boxed areas. (C and D) PM-restricted FA activation impairs autophagy flux. (C) Autophagy flux of indicated strains expressing 2GFP-ATG8 during starvation ? cerulenin. Data are means ? SD (n = 4). (D) Degradation of cytosolic 2GFP in indicated strains during starvation + cerulenin. Data are means ? SD (n = 6). (E) EM analysis of autophagic bodies in indicated strains after 3 h of starvation + cerulenin. Data are means ? SD (n = 3; 50 cells/strain). (F) Autophagy flux in FAA1 or PM-FAA1 cells expressing 2GFP-ATG8 after 6 h starvation in the presence of 0.01% (w/v) glucose. Data are means ? SD (n = 6). (G and H) Human ACSL4v1 localizes to autophagic membranes and restores autophagy flux in yeast. (G) Plasmid-encoded ACSL4v1-GFP localizes to autophagosomes (arrowheads) in mCherry-ATG8expressing Dfaa1Dfaa3Dfaa4 cells after 1 h of rapamycin treatment. (H) Autophagy flux in PM-FAA1 or Dfaa1Dfaa3Dfaa4 cells expressing 2GFP-ATG8 and plasmid-encoded ACSL4v1-mCherry or FAA1mCherry, respectively, after 6 h of starvation + cerulenin. Data are means ? SD (n = 3). Dashed lines indicate cell boundaries. Scale bars, 1 mm. t tests: * ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download