ArticlePDF Available

Endosomal lipid signaling reshapes the endoplasmic reticulum to control mitochondrial function

December 2022
Science 378(6625):eabq5209

December 2022
378(6625):eabq5209

DOI:10.1126/science.abq5209

Authors:

Wonyul Jang

Korea Advanced Institute of Science and Technology

Dmytro Puchkov

Leibniz-ForschungsInstitut für Molekulare Pharmakologie

Yongtian Liang

Freie Universität Berlin

Show all 11 authorsHide

Cells respond to fluctuating nutrient supply by adaptive changes in organelle dynamics and in metabolism. How such changes are orchestrated on a cell-wide scale is unknown. We show that endosomal signaling lipid turnover by MTM1, a phosphatidylinositol 3-phosphate [PI(3)P] 3-phosphatase mutated in X-linked centronuclear myopathy in humans, controls mitochondrial morphology and function by reshaping the endoplasmic reticulum (ER). Starvation-induced endosomal recruitment of MTM1 impairs PI(3)P-dependent contact formation between tubular ER membranes and early endosomes, resulting in the conversion of ER tubules into sheets, the inhibition of mitochondrial fission, and sustained oxidative metabolism. Our results unravel an important role for early endosomal lipid signaling in controlling ER shape and, thereby, mitochondrial form and function to enable cells to adapt to fluctuating nutrient environments.

Content uploaded by Yongtian Liang

Content may be subject to copyright.

RESEARCH ARTICLE SUMMARY

◥

CELL BIOLOGY

Endosomal lipid signaling reshapes the endoplasmic

reticulum to control mitochondrial function

Wonyul Jang, Dmytro Puchkov, Paula Samsó, YongTian Liang, Michal Nadler-Holly, Stephan J. Sigrist,

Ulrich Kintscher, Fan Liu, Kamel Mamchaoui, Vincent Mouly, Volker Haucke*

INTRODUCTION: Cells need to react appropri-

ately to nutritional cues. Defects in the rewiring

of metabolism in response to alterations in

nutrient supply have been linked to human

diseases ranging from diabetes to muscle

atrophy. Starvation represses anabolic path-

ways and facilitates catabolic ones, such as the

degradation of macromolecules by autophagy

and endolysosomes. Starvation also promotes

the b-oxidation of fatty acids in mitochondria

to produce adenosine triphosphate (ATP).

Within cells, organelles including lysosomes

and mitochondria undergo changes in shape

and dynamics. These processes are often regu-

lated by phosphoinositide lipids. Phosphoino-

sitides are also involved in the formation of

membrane contacts between organelles and

in the response of cells and tissues to growth

and nutrient signals. How the adaptive changes

that protect mammalian cells and tissues from

starvation-induced damage are coordinated on

a cell-wide scale is unknown.

RATIONALE: Endolysosomal membrane dynam-

ics and function are controlled by phospho-

inositide signaling lipids, most notably by the

synthesis and turnover of phosphatidylinositol

3-phosphate [PI(3)P]. Patients carrying muta-

tions in the gene encoding the lipid phosphatase

MTM1, an enzyme that mediates endosomal

PI(3)P turnover, suffer from X-linked centro-

nuclear myopathy (XLCNM), a severe neuro-

muscular disease characterized by muscle

atrophy, disorganization of mitochondria, and

defects in the organization of the muscle endo-

plasmic reticulum (ER). Given that PI(3)P is

a hallmark of endosomes, we hypothesized

that the control of early endosomal PI(3)P by

MTM1 might serve to orchestrate adaptive

changes in the dynamics of the ER and mito-

chondria in response to altering nutrient

supply.

RESULTS: Working with XLCNM patient–derived

myoblasts and engineered cell lines, we found

that nutrient starvation (for example, lack of

amino acids) induced the hydrolysis of PI(3)P

by endosomal recruitment of MTM1. Concom-

itantly, tubular ER membranes were observed

to be converted into ER sheets by live super-

resolution light microscopy. Mechanistically,

loss of early endosomal PI(3)P upon starvation

was found to reduce membrane contacts be-

tween peripheral ER tubules and early endo-

somes. These contacts function as physical

tethers that may transmit pulling forces from

highly motile peripheral endosomes to the

tubular ER. Using proximity labeling proteomic

and functional cell biological experiments we

demonstrated that the ER–endosome contacts

were mediated by binding of the related ER

membrane proteins RRBP1 and kinectin 1 to

PI(3)P on endosomes. To study the role of

starvation-induced reshaping of tubular ER

membranes into sheets on mitochondrial form

and function, we combined live imaging with

three-dimensional focused ion beam milling

scanning electron microscopy (FIB-SEM) and

proteomic analysis. We found that starvation-

induced ER reshaping by MTM1 reduced the

rate of mitochondrial fission and promoted

the formation of a hyperfused mitochondrial

network. Genetic manipulations that resulted

in ER sheet expansion caused the formation

of an enlarged mitochondrial network even

in fed cells. Conversely, impaired ER reshap-

ing and reduced mitochondrial network for-

mation were observed in starved myoblasts

from XLCNM patients. Mitochondrial net-

work formation appeared to be critical for

the delivery of fatty acids from lipid droplets

to mitochondria and for oxidative ATP pro-

duction to sustain energy supply in nutrient-

deprived cells.

CONCLUSION: Our data unravel a crucial role for

early endosomal lipid signaling in controlling

ER morphology and, thereby, mitochondrial

form and function to orchestrate the adaptive

response of cells to alterations in nutrient (e.g.,

amino acid) supply. This mechanism operates

independent of autophagy, a cellular self-eating

process typically induced by prolonged starva-

tion. Rather, it resembles an organellar con-

veyor belt, in which the tubular ER serves as a

membrane conduit that transmits nutrient-

triggered changes in endosomal PI(3)P levels

to metabolic organelles to enable metabolic

rewiring. How early endosomal PI(3)P levels

and MTM1 function are controlled by cellular

nutrient status is currently unknown. Defects

in ER shape, mitochondrial morphogenesis,

and cellular ATP depletion caused by loss of

MTM1 function can explain the observed myo-

fiber hypotrophy and defective ER organi-

zation in animal models of XLCNM and in

human patients who often appear under-

nourished. We therefore hypothesize that

dysregulated organelle remodeling may un-

derlie XLCNM caused by MTM1 mutations

in humans.▪

RESEARCH

Jang et al., Science 378, 1188 (2022) 16 December 2022 1of1

The list of author affiliations is available in the full article online.

*Corresponding author. Email: haucke@fmp-berlin.de

Cite this article as W. Jang et al., Science 378, eabq5209

(2022). DOI: 10.1126/science.abq5209

READ THE FULL ARTICLE AT

https://doi.org/10.1126/science.abq5209

Role of MTM1-mediated endosomal

PI(3)P signaling in mitochondrial

metabolic rewiring through

reshaping of the ER in response

to starvation. In fed cells, early

endosomes form contacts with ER

tubules. Tubular ER membranes

facilitate mitochondrial fission and

serve as a source for lipid droplet

formation. Nutrient starvation–

induced hydrolysis of endosomal

PI(3)P by MTM1 reduces mem-

brane contacts between the

tubular ER and early endosomes.

The resulting loss of peripheral ER

tubules induces mitochondrial

network formation and the delivery

of fatty acids to mitochondria to

sustain cellular energy supply.

EE, early endosome; MT, microtubule;

FA, fatty acid; LD, lipid droplet.

Sheet ER

Fed

PI3P

Tubular

Starved

LD from

tubular ER

β-oxidation↑

Tubular

Mito

ssion

Sheet

MTM1

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

RESEARCH ARTICLE

◥

CELL BIOLOGY

Endosomal lipid signaling reshapes the endoplasmic

reticulum to control mitochondrial function

Wonyul Jang

, Dmytro Puchkov

, Paula Samsó

, YongTian Liang

, Michal Nadler-Holly

Stephan J. Sigrist

, Ulrich Kintscher

, Fan Liu

1,3

, Kamel Mamchaoui

Vincent Mouly

, Volker Haucke

1,2,3

Cells respond to fluctuating nutrient supply by adaptive changes in organelle dynamics and in metabolism. How

such changes are orchestrated on a cell-wide scale is unknown. We show that endosomal signaling lipid

turnover by MTM1, a phosphatidylinositol 3-phosphate [PI(3)P] 3-phosphatase mutated in X-linked

centronuclear myopathy in humans, controls mitochondrial morphology and function by reshaping the

endoplasmic reticulum (ER). Starvation-induced endosomal recruitment of MTM1 impairs PI(3)P-dependent

contact formation between tubular ER membranes and early endosomes, resulting in the conversion of

ER tubules into sheets, the inhibition of mitochondrial fission, and sustained oxidative metabolism. Our

results unravel an important role for early endosomal lipid signaling in controlling ER shape and, thereby,

mitochondrial form and function to enable cells to adapt to fluctuating nutrient environments.

The ability of cells to react appropriately

to nutritional cues is of fundamental im-

portance for cell physiology, and defects

in the cellular response to altered nutri-

ent supply have been linked to human

diseases ranging from diabetes to muscle

atrophy (1–4). Among the early changes elic-

ited by nutrient stress are the suppression of

anabolic programs such as protein translation

(5,6) and the concomitant induction of cat-

abolic processes involving the proteasome,

autophagy [e.g., lipophagy, reticulophagy (7)],

endolysosomal turnover of proteins (6,8), and

increased mitochondrial b-oxidation of fatty

acids (9). How these adaptive responses that

protect mammalian cells and tissues from

starvation-induced damage and the induction

of apoptotic cell death are coordinated is un-

known. Endolysosomal membrane dynamics

and function are controlled by the spatiotem-

porally regulated synthesis and turnover of

phosphatidylinositol 3-phosphate [PI(3)P] and

related signaling lipids by phosphatidylinositol

(PI) 3-kinases and 3-phosphatases (10,11).

MTM1, the founding member of the myotubu-

larin family of PI 3-phosphatases, is crucially

involved in PI(3)P homeostasis on endosomes

(12–15). Patients carrying mutations in the

MTM1 gene suffer from X-linked centronu-

clear myopathy (XLCNM), a severe, often fatal

disease characterized by muscle weakness

due to myofiber atrophy, disorganization of

mitochondria, and structural defects in the

organization of the sarcoplasmic reticulum, a

specialized form of the endoplasmic reticulum

(ER) found in muscle tissue that is important

for excitation-contraction coupling (16,17).

How XLCNM-linked mutations in endosomal

MTM1 cause such pleiotropic defects in the

organization of the ER and other organelles

has remained elusive but may relate to a thus

far unexplored function of endosomes in

orchestrating adaptive changes in organelle

dynamics.

Results

Nutrient-regulated reshaping of the ER is

controlled by endosomal MTM1

To address the question how XLCNM-linked

mutations in endosomal MTM1 cause defects

in ER organization, we analyzed ER morphol-

ogy (18) in myoblast cell lines from healthy

controls or XLCNM patients (table S1) (19–21)

suffering from pronounced or complete loss of

MTM1 protein (fig. S1A) and a resulting in-

crease in PI(3)P levels (15). The ER membrane

is composed of interconnected uniform flat

cisternal sheets, fenestrated sheets with nano-

holes (22,23), and peripheral dynamic narrow

tubules (∼30 to 60 nm in diameter). ER sheets

and peripheral ER tubules were initially char-

acterized by confocal light microscopy (24)

and have recently been resolved at the nano-

scalebysuper-resolutionimaging(22,25). We

determined the morphology of the peripheral

ER by semiautomated image analysis of cells

expressing mEmerald- or Halo-tagged ver-

sions of the ER membrane protein Sec61bor

cells stained for the ER marker calreticulin

(fig. S1B). This analysis revealed a prominent

accumulation of tubular versus sheet ER in

myoblasts from XLCNM patients compared

with cells from healthy controls (Fig. 1, A and

B). Accumulation of tubular ER in XLCNM

myoblasts was confirmed by time-gated stimu-

lated emission depletion (gSTED) nanoscopy

imagingat50-nmresolution[consistentwith

(22)] (Fig. 1A). Depletion of MTM1 in healthy

controls phenocopied XLCNM myoblasts with

respect to the accumulation of tubular ER and

the concomitant reduction in sheet ER (Fig. 1C

and fig. S1C). These data suggest that the de-

fects in ER morphology observed in XLCNM

patient myoblasts are a consequence of MTM1

loss of function and are consistent with earlier

in vivo data indicating a possible role for MTM1

in the control of ER morphology (17).

A hallmark of MTM1 loss is the accumula-

tion of its substrate lipid PI(3)P (15). We rea-

soned that the observed defects in ER shape in

XLCNM myoblasts are a consequence of ele-

vated PI(3)P levels and, thus, might be rescued

by pharmacological inhibition of PI(3)P syn-

thesis. Consistently, specific inhibition of the

endosomal PI 3-kinase VPS34 in the presence

of VPS34-IN1 or reexpression of active MTM1

sufficed to restore normal ER morphology in

XLCNM patient myoblasts (Fig. 1D and fig. S1,

D and E). An elevated ER sheet-to-tubule ratio

was also observed in VPS34-IN1–treated hu-

man HeLa cells imaged by confocal light mi-

croscopy or gSTED nanoscopy at steady state

(Fig. 1E and fig. S1, F and G). In contrast, de-

pletion of phosphatidylinositol 3,5-bisphosphate

[PI(3,5)P

], another potential MTM1 substrate

lipid (26), did not affect ER shape (fig. S1, F

and H). Previous data show that PI(3)P is

involved in the cellular response to altered

nutrient supply (27–29). We therefore tested

whether PI(3)P metabolism might be regulated

by nutrients. We found endosomal PI(3)P levels

to decline upon cellular nutrient deprivation

[consistent with (29)] (Fig. 1F and fig. S1I),

and this was accompanied by progressive ER

sheet expansion in live or fixed cells imaged

by super-resolution STED or confocal micros-

copy(Fig.1,GandH;fig.S1,JandK;and

Movies 1 and 2). The levels of major ER-

shaping proteins remained unaltered (fig. S3, J

and K). ER sheet expansion was also observed

in cells treated with the potent catalytic mech-

anistic target of rapamycin (mTOR) inhibitor

Torin 1 (fig. S1L), which is often used to mimic

conditions of starvation. Further analysis re-

vealed that deprivation of amino acids (gluta-

mine in particular), rather than growth factors

and glucose, was responsible for ER reshaping

during starvation (Fig. 1I and fig. S1M). In con-

trast, amino acid replenishment failed to rescue

the starvation-induced reduction in cell area

(fig. S1, N and O), suggesting that ER shape and

cell size are controlled by distinct mechanisms.

To probe whether reduced PI(3)P levels

causally underlie the starvation-induced re-

modeling of ER membranes, we depleted cells

of MTM1, a condition under which PI(3)P

accumulates (15), by RNA interference or

RESEARCH

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 1of16

Leibniz-Forschungsinstitut für Molekulare Pharmakologie

(FMP), 13125 Berlin, Germany.

Department of Biology,

Chemistry, and Pharmacy, Freie Universität Berlin, 14195

Berlin, Germany.

Charité-Universitätsmedizin Berlin,

10117 Berlin, Germany.

Centre de Recherche en Myologie,

Institut de Myologie, Inserm, Sorbonne Université, 75013

Paris, France.

*Corresponding author. Email: haucke@fmp-berlin.de

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

CRISPR-Cas9–mediated knockout (KO). Loss

of MTM1 potently antagonized the starvation-

induced conversion of peripheral ER tubules

into sheets (Fig. 1J and fig. S2, A to E), a pheno-

type that was rescued by reexpression of cat-

alytically active mCherry-MTM1 (fig. S2F). Further

ultrastructural analysis by three-dimensional

(3D) focused ion beam scanning electron mi-

croscopy (FIB-SEM) showed that even the peri-

nuclear ER, which appeared as flat uniform

sheets in starved wild-type (WT) cells, was

highly fenestrated in starved cells lacking

MTM1 (Fig. 1K, FIB-SEM; fig. S2G; and movie S1).

The total ER volume fraction was unaltered

(fig. S2H). Accumulation of highly fenestrated

sheet ER and ER tubules in MTM1 KO cells

was further evidenced by the reduced average

length of ER profiles determined by SEM anal-

ysis of 2D cross sections (Fig. 1K, SEM).

We also tested whether other members of

the myotubularin family of MTM1-related phos-

phatases affect ER shape. On the basis of their

mRNA expression levels in HeLa cells, we ana-

lyzed four myotubularin-related proteins (MTMRs)

and found that, of these four, only MTMR1, the

family member most closely related to MTM1,

affected ER shape (fig. S2, I to K). These data

suggest a close functional relationship between

endosomal PI(3)P and ER morphology, in par-

ticular the presence of ER tubules in cells. In-

creased levels of endosomal PI(3)P caused by

MTM1 loss of function prevent the starvation-

induced remodeling of ER membranes, and

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 2of16

Fig. 1. Nutrient-regulated

reshaping of the ER is

controlled by endosomal

MTM1. (A) Confocal and STED

live images of Halo-Sec61bin

fed myoblasts from healthy

and XLCNM patients. (B) Ratio

of sheet/total ER area stained

for calreticulin. AB1190 (n=

206), KM1288 (n= 222), NL15

(n= 102). (C) Ratio of sheet/

total ER area in fed healthy

(AB1190) myoblasts at steady-

state treated with control

(siCo/siCo) or MTM1 siRNA

(siMTM1_1/_SP). siCo/siCo

(n= 181), siMTM1_1/_SP (n=

167). (D) Ratio of sheet/total

ER area in XLCNM patient

myoblasts treated with DMSO

(control) or VPS34-IN1 (5 mM,

2 hours). KM1288, DMSO

(n= 113), VPS34-IN1 (n= 148);

NL15, DMSO (n= 80), VPS34-

IN1 (n= 86). (E) Confocal

and STED images of DMSO or

VPS34-IN1-treated (5 mM,

2 hours) HeLa cells expressing

mEmerald-Sec61b.(F) PI(3)P

levels in fed (n= 820) or

starved (EBSS 2 hours; n=

725) HeLa cells. (G) STED

images of fed or starved live

HeLa cells stably expressing

Halo-Sec61b. See Movies 1

and 2. (H) Ratio of sheet/total

ER area in fed or starved

(EBSS 2 hours) HeLa cells

stained for calreticulin. Fed

(n= 312), Starved (n= 320).

(I) Ratio of sheet/total ER area

of HeLa cells expressing

mEmerald-Sec61bexposed to

different nutrients (2 hours). Full (n= 128), −(i.e., EBSS only, n= 148), +Dialyzed

FBS (n= 160), +Glucose (4.5 g/liter) (n= 111), all amino acids (+ All AA, n= 171).

(J) Ratio of sheet/total ER area in fed or starved HeLa WT or MTM1 KO cells.

Fed WT (n= 136), KO (n= 123); Starved WT (n= 156), KO (n= 207).

(K) (Top left) Electron microscopy images of starved WT or MTM1 KO HeLa cells.

Purple, ER cisternal cross sections. (Right) Length of ER cross sections (# of

objects: WT = 1018, KO = 1070) from three cells. (Bottom left) 3D FIB-SEM

analysis. 3D reconstruction of ER (purple) and mitochondria (brown). ER sheets

are fenestrated in starved KO cells. See fig. S2G and movie S1. Scale bars:

10 mm (white), 2 mm (yellow), and 1 mm (black). n, total number of cells analyzed

from two (H) or three [(B) to (D), (F), (I), and (J)] independent experiments.

One-way ANOVA with Dunnett’s multiple comparisons test (B); two-tailed

Mann-Whitney test [(C), (D), (H), (J), and (K)]; two-tailed unpaired ttest (F);

Kruskal-Wallis test with two-sided Dunn’s multiple comparison test (I).

**P≤0.01, ****P≤0.0001. Data are median ± interquartile range [(B) to (D),

and (H) to (K)] or mean ± SD (F).

Sheet ER ratio

AB1190

KM1288

NL15

0.0

0.2

0.4

0.6

0.8

1.0

Healthy XLCNM

(steady fed)

****

DMSO

VPS34In1

DMSO

VPS34In1

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

KM1288 NL15

**** ****

0.0

0.5

1.0

1.5

Sheet ER ratio

****

AB1190

siCo./siCo.

siMTM1_1/_SP

Fed

Starved

0.6

0.7

0.8

0.9

1.0

1.1

Norm. PI(3)P inte nsity

WTKOWT

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

Fed Starved

****

Fed

Starved

0.0

0.5

1.0

1.5

Sheet ER ratio

****

IJK

AB1190

Healthy myoblast

KM1288NL15

XLCNM myoblast

Halo-Sec61βInset

Confocal STED

Inset

mEmerald-Sec61βInset

Confocal STED

Inset

DMSOVPS34In1

Halo-Sec61βInset

Live-STED

0.0

0.5

1.0

ER length (μm)

****

Starved

Starved-WT Starved-KO

FIB-SEM SEM

Full

+Dialyzed FBS

+Glucose (4.5 g/liter)

+All AA

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

****

EBSS

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

this may underlie the structural defects in ER

morphology found in muscle cells and tissue

from XLCNM patients (16,17,30,31).

Starvation-induced PI(3)P hydrolysis by

MTM1 at ER–early endosome contacts mediates

ER reshaping

We next sought to understand how PI(3)P hy-

drolysis by MTM1 mediates the starvation-

induced conversion of peripheral ER tubules

into flat uniform sheets. As the ER is subject

to turnover by means of autophagy (32), we

probed whether blockade of autophagosome

formation interferes with starvation-induced

ER sheet expansion. However, ER sheet expan-

sion in starved cells was unaffected by phar-

macological or genetic inhibition of autophagy

in the presence of VPS34-IN1 (15)orknock-

down of the essential autophagy factor ATG5

(fig. S3, A to D). Moreover, loss of MTM1, that

is, a condition in which starvation-induced ER

sheet expansion is perturbed, did not alter the

ability of starved cells to form microtubule-

associated protein 1 light chain 3 (LC3)–positive

autophagosomes (fig. S3, E and F), the over-

all levels of the autophagy marker LC3-II (fig.

S3G), or the subcellular localization and ac-

tivation of transcription factor EB (TFEB), a

master regulator of autophagy gene expres-

sion (33) (fig. S3, L and M). MTM1 KO cells

displayed slightly reduced levels of the sheet-

localized ER-phagy receptor FAM134B under

fed and starved conditions (fig. S3, H and I).

Reduced levels of FAM134B have been shown

to result in sheet ER expansion (32), a pheno-

type opposite to the expansion of the tubular

ER observed in MTM1 KO cells (Fig. 1). The

starvation-induced increase in the sheet ER

ratiothusappearstobeindependentofFAM134B-

mediated ER-phagy. Although mechanistic

target of rapamycin complex 1 (mTORC1) ac-

tivity was elevated in fed MTM1 KO cells

[consistent with observations in MTM1 KO

mice (34)], loss of MTM1 did not affect sup-

pression of mTORC1 activity in starved cells

(fig. S3N). Finally, lumenal ER calcium levels

were not significantly altered (fig. S3O), and

no signs of an ER stress response were de-

tected in MTM1 KO cells (fig. S3P). We con-

clude that PI(3)P hydrolysis by MTM1 in starved

cells controls ER shape independently of au-

tophagy, the ER stress response, and ER cal-

cium homeostasis.

Given that PI(3)P is a hallmark of early endo-

somes (10,35) and that the ER makes extensive

contacts with other organelles including the

plasma membrane (36), endosomes, lysosomes,

and mitochondria (37), we hypothesized that

MTM1 specifically acts on peripheral early

endosomes in starved cells and thereby con-

trols ER morphology (Fig. 2A), for example,

through membrane contacts. To test this, we

generated a cell line stably expressing a chimera

between the early endosomal protein Rab5A

and the biotinylating enzyme ascorbate per-

oxidase 2 (APEX2) (38) under the control of

a doxycycline-inducible promoter (fig. S4A).

Proximity labeling and affinity capture re-

vealed a prominent starvation-induced enrich-

ment of endogenous MTM1 on early endosomes,

while early endosomal antigen 1 (EEA1), a PI(3)P-

binding scaffold protein, was depleted in early

endosomes (Fig. 2B and fig. S4B). Nutrient

starvation thus induces the recruitment of

MTM1 to Rab5-containing early endosomes,

likely resulting in PI(3)P hydrolysis and ER

sheet expansion. We further probed this model

by a chemical genetic approach that capitalizes

on the FRB/FKBP system, which enables the

artificial tethering of organelles via rapalog-

induced heterodimerization of chimeras be-

tweenFK506-bindingprotein(FKBP)andthe

FKBP-rapamycin binding (FRB) domain of

mTOR (39). We found that acute rapalog-

induced recruitment of active FKBP-MTM1

to early endosomes tagged with FRB-Rab5A

resulted in massive ER tubule-to-sheet con-

versioninfedcells(Fig.2Candfig.S4C),phe-

nocopying starvation-induced PI(3)P depletion.

Recruitment of active MTM1 to lysosomes (fig.

S4,DandE)orearlyendosomalrecruitment

of catalytically inactive mutant MTM1 (CS) did

not affect ER shape (Fig. 2C). We conclude that

recruitment of catalytically active MTM1 to

early endosomes drives ER sheet expansion

to mount the cellular response to nutrient

starvation.

The role of endosomal MTM1 in controlling

ER shape might depend on the formation of

hitherto molecularly undefined membrane con-

tacts between the tubular ER and highly motile

early endosomes that could provide a force

that aids in keeping ER tubules under tension.

In support of this hypothesis, we observed

PI(3)P and the early endosome marker Rab5A

to tightly colocalize with the tubular ER net-

work in the cell periphery (Fig. 2D). Moreover,

tracking early endosomes and the ER network

in live cells by high-speed spinning disk con-

focal imaging showed that forming ER tubules

are tightly associated with motile early endo-

somes (Fig. 2E and Movie 3), a conclusion fur-

ther supported by the close association of ER

tubuleswithearlyendosomesmarkedbyin-

ternalized bovine serum albumin–gold (BSA-

gold) in electron micrographs (Fig. 2F). Hence,

the large population of peripheral early endo-

somes may serve to promote ER tubules. Con-

sistently, we found that acute depletion of early

endosomes from the cell periphery by rapalog-

induced endosomal recruitment of the micro-

tubule minus end–directed dynein adaptor

BICD2 caused a near complete loss of the tu-

bular ER network and a concomitant accumu-

lation of perinuclear sheet ER (Fig. 2, G and H;

fig. S4F; and movie S2).

We reasoned that MTM1-mediated hydrolysis

of PI(3)P at early endosomes during starvation

maycontrolERshapeeitherby(i)controlling

early endosome motility or (ii) altering the

number or stability of hitherto unknown phys-

ical contacts between early endosomes and

the ER (fig. S4G). As early endosome motility

was unaffected by nutrient starvation (fig. S4,

H and I), we followed the alternative hypoth-

esis that nutrient regulation of early endo-

somal PI(3)P controls membrane contact sites

between early endosomes and the ER and,

thereby, ER shape. To monitor such contacts,

we determined the fractional overlap between

the peripheral ER and early endosomes by

multicolor STED microscopy (fig. S4J). Starva-

tion reduced the fractional overlap of the ER

with early endosomes (Fig. 2I). The total con-

tact area between the ER and early endosomes

marked by internalized BSA-gold was also

found to be significantly reduced in starved

cells analyzed by electron microscopy (Fig. 2J).

To confirm these findings by another inde-

pendent approach, we generated a stable cell

line that coexpresses equimolar ratios of an

ER membrane targeting domain (ERM) fused

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 3of16

Movie 1. Live imaging of ER dynamics in fed or

starved cells. Live-cell spinning disk confocal

imaging of fed or starved HeLa cells expressing

mEmerald-Sec61b. Videos were acquired at

1 frame/5 min for 150 min and correspond to

fig. S1K.

Movie 2. Live STED imaging of tubular or sheet

ER from fed or starved HeLa cells expressing

Halo-Sec61 b.Live-cell STED imaging of starved

HeLa cells expressing Halo-Sec61b. Videos were

acquired at 1 frame/3 s for 30 s and are related

to Fig. 1G.

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 4of16

PI(3)P MergeHalo-Rab5A

Merge with

mEmerald-Sec61β

MTM1

Binding to Rab5A-EE

Fed

Starved

Fed

Starved

EEA1

0.0

0.5

1.0

1.5

****

Binding to Rab5A-EE

BICD2-FRB +

FKBP-Rab5

rapalog

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

****

FKBP-MTM1 WT

FKBP-MTM1 CS

FRB-Rab5A

rapalog -

rapalog

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

**** ns

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

**** ns

rapalog -

rapalog

Fed Starved Fed Starved

Endosomal

FKBP-Rab5

Cytosolic

Rab5-FKBP

FKBP-Rab5

+rapalog

BICD2-

FRB

EE arrest at

perinuclear

Tubular ER

PI3P

Fed

Sheet ER

MTM1

Starved

rapalog

Starved

tarve

Rab5

FRB FKBP

rapalog

Fed

Tubular

PI3P

Fed

GFP 11

GFP1-10

MTM1

Starved

tarve

Norm. split-GFP do ts

(Starved / Fed)

siCo.

siMTM1_1

0.5

0.6

0.7

0.8

0.9

1.0

1.1

# split- GFP dots / Cell

100

150

200

250 *

Fed

Starved

0 s 0.4 s 0.8 s 1.2 s

mEmerald-

Sec61β

mCh-Rab5

Endosomal

FKBP-Rab5 HA

Fed

-rapalog

Starved

-rapalog

Cytosolic

Rab5-FKBP

Fractional overlap between

peripheral ER and EE

Fed

Starved

0.00

0.01

0.02

0.03 *

(STED)

ER-EE contact length (nm)

200

400

600

(EM)

Fed

Starved

Fig. 2. Starvation-induced PI(3)P hydrolysis by MTM1 at ER–early

endosome contacts mediates ER reshaping. (A) PI(3)P-positive early endo-

somes (EE) may control the tubular ER. (B) Starvation-induced recruitment

of MTM1 to EE. EE-associated MTM1 (left) and EEA1 (right) in fed versus starved

HeLa cells. n= 3 independent experiments. (C) Ratio of sheet/total ER in

HeLa cells coexpressing FRB-iRFP-Rab5A and mRFP-FKBP-MTM1 WT (wild-type)

or CS (inactive mutant) ± rapalog. FKBP-MTM1 WT −(n= 162), rapalog (n=

265), FKBP-MTM1 CS −(n= 138), rapalog (n= 187). (D) Confocal images of HeLa

cells stably coexpressing Halo-Rab5A (EE) and mEmerald-Sec61b(ER) and

stained for PI(3)P. (E) Time-lapse confocal images of HeLa cells coexpressing

mEmerald- Sec61b(ER) and mCherry-Rab5 (EE). Yellow arrowheads mark motile

EE. (F) Electron micrographs illustrating tubular ER (blue) contacts with EE

(orange) marked by internalized BSA-gold. (G) Rapalog-induced acute depletion

of EE from the periphery. See movie S2. (H) Ratio of sheet/total ER in mock-

(n= 177) versus rapalog-treated (n= 141) HeLa cells as in (G). (I) Fractional

overlap between the peripheral ER and EE in fed or starved cells determined

by STED microscopy. Fed = 118 ROIs; Starved = 166 ROIs (40 to 50 cells from

four experiments). (J) Morphometric analysis of contact length (nanometers)

between ER and BSA-gold labeled EE in fed (42 endosomes) versus starved

cells (43 endosomes). (K) Reconstitution of split-GFP fluorescence by ER-EE

contacts. (L) Normalized number of ER/EE contacts in fed (n=238)versus

starved (n= 257) HeLa cells. (M)RescueofER/EEcontacts in starved MTM1-

depleted HeLa cells. siCo Fed (n=123),Starved(n=77),siMTM1_1Fed

(n=111),Starved(n=85).(N) ERM-FRB•rapalog•FKBP-Rab5A ER/EE

synthetic tether. (O) Ratio of sheet/total ER in fed or starved cells from

(N). Endosomal FKBP Rab5 Fed −(n= 68), rapalog (n=70),Starved−(n=

92), rapalog (n= 86); Cytosolic Rab5-FKBP Fed −(n= 97), rapalog (n=90),

Starved −(n=104),rapalog(n=120).(P) Confocal images from (O) stained

for HA to mark the ER, gray. Scale bars: 2 mm (yellow), 1 mm(black),and

100 nm (white). n, number of cells analyzed from two [(M) and (O)] or three

[(C), (H), and (L)] independent experiments. Two-tailed unpaired ttest

[(B), (C), (I), (L), and (O);] two-tailed Mann-Whitney test [(H) and (J)];

*P≤0.05, ****P≤0.0001; ns, nonsignificant. Data are mean ± SD [(B), (I),

(L), and (M)] or median ± interquartile range [(C), (H), (J), and (O)].

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

to GFP

(40)alongwithaGFP

1-10

-Rab5A chi-

mera under the control of a doxycycline-inducible

promoter (fig. S4, K and L). Formation of mem-

brane contacts between early endosomes and

the ER in fed cells reconstitutes GFP fluores-

cence (Fig. 2K). ER–early endosome membrane

contacts were greatly reduced in starved cells

(Fig. 2L and fig. S4M), and this effect was re-

verted completely by depletion of MTM1 (Fig.

2M). These data suggest that MTM1-mediated

hydrolysis of PI(3)P at early endosomes reduces

the contacts between these organelles and the

ER (Fig. 2K). We directly tested this model by

examining the effects of semisynthetic tether-

ing of early endosomes to the ER using chem-

ical genetics. Acute rapalog-induced formation

of ER–early endosome tethers mediated by

ERM-FRB•rapalog•FKBP-Rab5A potently sup-

pressed starvation-induced ER tubule-to-sheet

conversion (Fig. 2, N to P; fig. S5B; and movie

S3) but had no effect on early endosome

motility (fig. S5A). In contrast, soluble, non-

lipidated Rab5A (Rab5-mRFP-FKBP) or cyto-

solic mRFP-FKBP had no effect on ER shape

(Fig. 2O and fig. S5, B and C). Extending the

length of the semisynthetic tether well beyond

the typical distance of 20 to 30 nm observed

for organelle contacts in vivo (36,37)didnot

affect its ability to prevent starvation-induced

ER reshaping (fig. S5, D and E), suggesting

that the exertion of a physical pulling force is

critical for the regulation of ER shape by mem-

brane contacts with early endosomes. Further-

more, the capability of ERM-FRB•rapalog•

FKBP-Rab5A tethers to prevent starvation-

induced ER tubule-to-sheet conversion was

unaffected by depletion of PI(3)P, indicating

that PI(3)P acts upstream of ER–early endo-

some membrane contact site formation (fig. S5,

F and G). Other organelles—late endosomes

and lysosomes, in particular—have also been

shown to form numerous contacts with the ER

(41,42).Earlyendosomesdisplayanevendis-

tribution throughout cells and into the periphery

where the tubular ER is located and outnumber

lysosomes by up to an order of magnitude. In

contrast, most lysosomes are concentrated in

the perinuclear area at steady state (fig. S6, A

to D). As a consequence, the total number of

membrane contacts between the ER and early

endosomes greatly exceeds that of the ER with

lysosomes (fig. S6, E to H), in spite of the high

relative fraction of lysosomes in touch with the

ER (42,43). Redistribution of lysosomes to the

perinuclear area either through acute rapalog-

induced dynein adaptor recruitment or deple-

tion of PI(3,5)P

did not affect ER shape (fig.

S6, I to L; fig. S1H; and movie S4). Sustained

loss of the lysosomal kinesin adaptor Arl8b,

a protein essential for lysosome dispersion,

only marginally increased the sheet ER frac-

tion [fig. S6, M to O; see also (44)], whereas

depletion of the ER-lysosome contact site pro-

tein protrudin (41) was without effect. Hence,

the tubular ER is largely controlled by its mem-

brane contacts with early endosomes (as

demonstrated in this study) and a smaller con-

tribution from lysosomes, possibly dependent

on cell type or conditions (44). Taken together,

our findings unravel a crucial role for MTM1-

mediated PI(3)P hydrolysis in the reduction of

membrane contacts between the ER and early

endosomes to reshape the ER in response to

changing nutrient levels.

Contacts formed by ER membrane

protein–mediated recognition of early

endosomal PI(3)P controls the tubular ER

To identify the molecular machinery that teth-

ers ER tubules to early endosomes in fed cells,

wecapitalizedontheobservationthatinWT

but not in MTM1 KO cells ER–early endosome

contacts are reduced under conditions of nu-

trient starvation (Fig. 3A). Starved WT or MTM1

KO HeLa cells inducibly expressing APEX2-

Rab5A were subjected to combined proximity

labeling and affinity purification–mass spec-

trometry to probe the molecular environment

of Rab5A-containing early endosomes. Subse-

quent biochemical fractionation to enrich for

ER membrane proteins (fig. S7, A to C, and

table S2) and comparison with previously iden-

tified ER surface proteins (45) identified sev-

eral putative ER transmembrane proteins that

might serve as early endosome tethers (fig.

S7D). Analysis of the phenotypic consequences

of cellular depletion of these factors revealed

that only loss of ER ribosome-binding protein

1 (RRBP1) promoted ER tubule-to-sheet con-

version in fed cells (fig. S7E). RRBP1 is local-

ized exclusively to ER membranes (Fig. 3, B

andC)andhasbeenassociatedwithchanges

in ER morphology, although no consensus re-

garding its precise function exists (46–48). ER

tubule-to-sheet conversion was exacerbated by

concomitant loss of RRBP1 and its close para-

log kinectin 1 (KTN1) (Fig. 3D and fig. S8, A to

C). Moreover, loss of RRBP1 and KTN1 potent-

ly reduced the number of ER–early endosome

membrane contacts in fed cells (Fig. 3E), sug-

gesting that RRBP1 and KTN1 regulate ER

shapebyactingastethers for early endosomes.

KTN1 and RRBP1 harbor functionally unchar-

acterized lysine-rich regions (LR1, -2, and -3)

in their cytoplasmic domains (Fig. 3F and fig.

S8,AandB)thatcouldconceivablyattachto

PI(3)P. Notably, KTN1 was found to be in con-

tact with PI(3)P-enriched early endosomes by

spatiotemporally resolved interaction proteo-

mics using 2xFYVE-APEX as a probe (49). We

found recombinant RRBP1-LR1-3 (fig. S8D) to

bind to PI(3)P with preference over both PI(4)

P and PI(3,4)P

(Fig.3,GandH).Thesedata

suggest that RRBP1 might recognize PI(3)P on

early endosomes to form tethers with the ER.

Consistent with this model, we observed that

expression of a mini-RRBP1 truncation pro-

tein variant harboring only the lysine-rich

cytoplasmic domain fused to its ER transmem-

brane anchor sufficed to restore a normal

tubular ER network in HeLa cells depleted of

endogenous RRBP1 and KTN1 (Fig. 3I, WT).

Deletion of lysine-rich regions 1 or 3 or of

the transmembrane anchor rendered trun-

cated mini-RRBP1 inactive (Fig. 3I and fig. S8,

E and F). These data suggest that RRBP1 and

KTN1 mediate recognition of early endosomal

PI 3-phosphates and, possibly, additional fac-

tors, to facilitate contact site formation be-

tween the ER and early endosomes, which

control the tubular ER network in human cells.

MTM1-dependent ER reshaping is required

for mitochondrial network formation

during starvation

Several mechanisms have been shown to con-

tribute to mitochondrial morphogenesis, includ-

ing membrane contacts between mitochondria

and lysosomes (50), late Golgi-derived vesicles

(51), and the tubular ER (37,52). The tubular ER

also directly promotes mitochondrial fission

(53) and acts as a donor organelle for the for-

mation of lipid droplets (LDs) (54,55) that

serve as an energy reservoir in fed cells and

tissues. Under conditions of starvation (e.g.,

deprivation of glutamine and other amino

acids), mitochondria undergo hyperfusion into

tubular networks to protect themselves from

mitophagy (56) and to enable efficient utiliza-

tion of fatty acids (57).

On the basis of these prior works, we hy-

pothesized that the MTM1-mediated starvation-

induced reshaping of the tubular ER into

sheets (Fig. 1) might serve to enable cells to

metabolically adapt to altering nutrient envi-

ronments, for example, by altering mitochon-

drial organization. To test this hypothesis, we

examined the effect of MTM1 loss on mito-

chondrial morphology and respiratory func-

tion. We found mitochondria to become

hyperfused in starved WT but not in MTM1 KO

cells or in WT cells depleted of endogenous

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 5of16

Movie 3. Live imaging of early endosome motility and the tubular ER. Live-cell high-speed spinning

disk confocal imaging of fed HeLa cells expressing mEmerald-Sec61b(labeled gray) and mCherry-Rab5

(labeled yellow). Videos were acquired at 1 frame/0.4 s for 1 min and correspond to Fig. 2E.

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

MTM1 (Fig. 4, A and B, and fig. S9A). Further

ultrastructural analysis by 3D FIB-SEM re-

vealed the formation of an extensive cell-wide

network of hyperfused mitochondria in starved

WT cells. In contrast, starved MTM1 KO cells

displayed an accumulation of small spherical

mitochondria suggestive of exacerbated mito-

chondrial fission and defects in cristae mor-

phology (Fig. 4, C to E, and fig. S9, B to D),

whereas the total mitochondrial volume frac-

tion was unchanged (Fig. 4F). Reduced mito-

chondrial network formation and defective

ER reshaping were also observed in starved

XLCNM patient–derived myoblasts (fig. S9,

G and H). Defective mitochondrial morpho-

genesis was not a consequence of the altered

expression of mitochondrial fusion- or fission-

related proteins (58) such as mitofusin 1/2,

OPA1 (59), and DRP1 or its hyperactive form

(pS616-DRP1) in MTM1 KO cells (fig. S3, J

and K). Moreover, acute rapalog-induced for-

mation of ER–early endosome tethers to inhibit

loss of ER tubules prevented the starvation-

induced formation of a hyperfused mitochon-

drial network in WT cells (Fig. 4G and fig. S9,

E and F), thereby phenocopying MTM1 loss.

Conversely, reducing membrane contacts be-

tween the ER and early endosomes by deple-

tion of RRBP1 and KTN1 led to the formation

of hyperfused mitochondrial networks and

expansion of the sheet ER in MTM1 KO cells

(fig. S9, I to K). RRBP1 and KTN1 therefore

act downstream of MTM1-mediated PI(3)P

hydrolysis.

If reshaping of the tubular ER into sheets

causally underlies the formation of a hyper-

fused mitochondrial network, one would expect

experimental manipulations that reshape the

ER to affect mitochondrial morphology. Con-

sistently, we found that cellular depletion of

either Rab10, a factor required for maintenance

of the tubular ER (60); RRBP1 and KTN1; or the

tubular ER-shaping protein reticulon 4 (61)—

conditions that result in ER sheet expansion

(fig.S10,A,C,D,andF)—cause the formation

of an enlarged mitochondrial network in fed

cells(fig.S10,B,C,E,andF).Mitochondrial

hyperfusion was further induced by overex-

pression of the ER sheet–inducing membrane

protein Climp63 (46) (fig. S10, G and H). Final-

ly, we observed increased rates of mitochon-

drial fission in starved MTM1 KO cells (fig.

S10, I to K, and movie S5), that is, conditions

under which ER tubules accumulate. Col-

lectively, these findings establish that the

starvation-induced conversion of ER tubules

to sheets by MTM1-mediated hydrolysis of

PI(3)P at ER–earlyendosomecontactsfacili-

tates the formation of a functional mitochon-

drial network.

Defective ER morphogenesis in absence of

endosomal MTM1 impairs mitochondrial

metabolic rewiring during starvation

To probe the physiological consequences of

defective mitochondrial morphogenesis, we

monitored mitochondrial oxygen consumption

and mitochondria-driven adenosine triphos-

phate (ATP) production in WT and MTM1 KO

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 6of16

Fig. 3. Membrane contacts formed by ER ribosome-binding

protein 1 and kinectin 1–mediated recognition of early

endosomal PI(3)P control the tubular ER. (A) Starvation-

induced dissociation of ER–early endosome (EE) contacts in

WT but not in MTM1 KO cells. (B) Volcano plot of proximity

biotinylated interactome of APEX2-Rab5A isolated from ER

membranes of starved MTM1 KO versus WT HeLa cells. Dark

purple: RRBP1 and KTN1. (C) Confocal images of HeLa cells

coexpressing mEmerald-Sec61b(ER) and RRBP1 full length

(FL)–mCherry. (D) Ratio of sheet/total ER in fed HeLa cells

treated with control (siCo/siCo = 174) or KTN1+RRBP1 siRNAs

(siKTN1/siRRBP1 = 221). (E) Normalized number of ER/EE

contacts in control or KTN1/RRBP1-depleted fed HeLa cells.

n= 3 independent experiments. (F) Schematic illustrating

truncated mini-RRBP1 (amino acids 1 to 150). TM, trans-

membrane anchor; LR1-3, cytoplasmic lysine-rich regions 1 to 3.

(G) GST-RRBP1-LR1-3 binds to PI(3)P liposomes. Supernatant

(S) and liposomal pellet (P) fractions were analyzed by

immunoblotting for GST. No PIP, liposomes lacking phospho-

inositides. (H) Quantified data as in (G) from n= 3 independent

experiments. (I) Ratio of sheet/total ER in control (siCo/siCo)

or KTN1/RRBP1-depleted stable doxycycline-inducible HeLa

cells expressing the indicated truncated RRBP1 protein

(“mini-RRBP1”) variants [see (F)]. WT, wild-type V5-tagged

mini-RRBP1; delLR1, mutant RRBP1 lacking lysine-rich region 1;

delLR3, mutant RRBP1 lacking lysine-rich region 3; delTM,

mutant RRBP1 lacking its transmembrane domain. siCo/siCo −

(n= 111), WT (n= 129), delLR1 (n= 102), delLR3 (n= 120),

delTM (n= 130); siKTN1/siRRBP1 −(n= 274), WT (n= 271),

delLR1 (n= 326), delLR3 (n= 263), delTM (n= 288). Scale bars:

2mm (yellow). nindicates the total number of cells analyzed

from three independent experiments. Two-tailed Mann-Whitney

test (D); one-way ANOVA with Tukey’s multiple comparisons

(E); one-way ANOVA with Dunnett’s multiple comparisons

test (H); Kruskal-Wallis test with two-sided Dunn’s multiple

comparison test (I). **P≤0.01, ****P≤0.0001. Data

are median ± interquartile range [(D) and (I)]; data are mean ±

SD [(E) and (H)].

RRBP1 (FL)-mCh mEmerald-Sec61βMerged

log2 difference (KO / WT)

-log10 p-value

-6 -4 -2 0 2 4 6

KTN1

RRBP1

PI3P

tether

WT MTM1 KO

MTM1

Starved

siCo./siCo.

siKTN1/siRRBP1

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

****

0.0

0.5

1.0

1.5

2.0

2.5

Norm. split-GFP dots

** **

siKTN1/siRRBP1

siCo./siCo.

Starved

Fed

0.0

0.2

0.4

0.6

0.8

1.0

Sheet ER ratio

********

****

delLR1

delLR3

delTM-

delLR1

delLR3

delTM

siCo./siCo. siKTN1/siRRBP1

reconstituted

mini V5-RRBP1:

TM Lysine-rich

ER lumen

58 145

LR1 LR2 LR3

150aa

Norm. liposome bound

0.0

0.5

1.0

1.5

****

no PIP

PI(3)P

PI(4)P

PI(3,4)P2

PI(3)P

no PIP

PI(4)P

PI(3,4)P2

SPSPSPSP

-55

GST-RRBP1

LR1-3 (1-150aa)

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

cells using Seahorse technology. Starved

MTM1 KO cells displayed severely reduced

basal as well as maximal oxygen consumption

rates resulting in significantly impaired mito-

chondrial ATP synthesis (Fig. 5, A and B, and

fig. S11, A and B). As a consequence, starved

MTM1 KO cells suffered from a pronounced

reduction of total cellular ATP (Fig. 5C). Sim-

ilar results were observed in cells, in which the

ER was artificially tethered to early endo-

somes (Fig. 5D). No significant differences in

mitochondrial basal oxygen consumption or

mitochondrial ATP synthesis were detected

in fed MTM1 KO cells, whereas the maximal

oxygen consumption rate was marginally de-

creased (fig. S11, C to E). The mitochondrial

membrane potential was unaffected by MTM1

loss, irrespective of the nutritional status of

the cells (fig. S11F). As defective mitochondrial

morphogenesis and cellular ATP depletion

are associated with reduced cell viability, we

probed whether MTM1 KO cells might be

poised to undergo apoptosis under conditions

of limited nutrient availability. Starved MTM1

KO cells indeed suffered from increased levels

of cleaved caspase 3 and poly(ADP ribose)

polymerase (PARP), common indicators of

apoptotic cell death (fig. S11, G to I). How-

ever, defective mitochondrial morphogenesis

in MTM1 KO cells was not a secondary con-

sequence of increased apoptosis, as treatment

of KO cells with the pan-caspase inhibitor

Z-VAD-FMK effectively blocked apoptosis (fig.

S11J) but failed to rescue defects in mitochon-

drial morphology (fig. S11K). We note that sim-

ilar apoptotic phenotypes have been reported

in starved OPA1 and mitofusin 1/2 KO cells

defective in mitochondrial fusion (62).

These results indicate that MTM1-mediated

reshaping of the tubular ER into sheets is

required for the formation of a functional mito-

chondrial network and mitochondria-driven

ATP production in starved cells. Previous work

has shown that the effective transfer and uti-

lization of fatty acids (FAs), a major substrate

for mitochondrial ATP synthesis via b-oxidation,

in starved cells requires mitochondria to be

organized into a highly tubulated hyperfused

network (56,57), although the exact underly-

ing molecular mechanism is unclear. We there-

fore hypothesized that impaired mitochondrial

ATP production in MTM1 KO cells might be a

consequence of defective FA trafficking. We

directly tested this by monitoring the fate of

FAs in WT and MTM1 KO cells under different

nutrient conditions. Depending on metabolic

state, cytosolic FAs can be metabolized in mito-

chondria (e.g., during starvation) or stored in

LDs (54,57,63), which exclusively form from

the tubular ER (55,64). Consistently, we found

that in WT cells, the number of LDs inter-

mittently declined at the onset of starvation

(≤2 hours), likely as a result of increased mito-

chondrial b-oxidation of FAs and blocked LD

formation upon loss of the tubular ER, before

eventually rising (Fig. 5E and fig. S11, L to N)

owing to autophagy-promoted lipid buildup

during sustained long-term (≥6hours)starva-

tion (57). In contrast, the number and total

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 7of16

0.003 7.280

Volume (μm³)

Fed Starved

siCo.

InsetSegmented TOM20

siMTM1_1

Fed Starved

WT KO

Starved

Mean area / mitochondrion

(A.U.)

100

200

300

400 **** ***

****

Fed

Starved

Fed

Starved

Fed

Starved

siMTM1_1 KO

siCo.

Starved

0.0

0.5

1.0

Individual mitochondrion

volume (μm³)

Top 3%

Sum of mitochondrial

volume (μm³)

mito

310

mito

Top 3%

EtOH

Rapalog

200

400

600

Mean area / mitochondrion

(A.U.)

Starved

(ER-EE tether)

Mitochondrial volume fraction

per cytoplasm

0.00

0.05

0.10

0.15

Starved

(EM)

# of mitochondria

Cumulative

mitochondria volume

(μm³)

0 100 200 300 400

Starved- WT

Starved- KO

Fig. 4. MTM1-dependent ER reshaping is required for mitochondrial

network formation during starvation. (A) Confocal images of mitochondria

(TOM20) in fed or starved control (siCo) and MTM1-depleted (siMTM1_1) HeLa

cells. (Bottom) Segmented mitochondria in ROI (yellow, 15 mmby15mm).

(B) Mean area of individual mitochondrion per ROI from fed or starved WT or

MTM1 KD or KO HeLa cells. siCo Fed = 345 ROIs, Starved = 418 ROIs; siMTM1_1

Fed = 354 ROIs, Starved = 440 ROIs; KO Fed = 343 ROIs, Starved = 322 ROIs

from four independent experiments. Each ROI represents a single cell. (C)3D

rendering of mitochondria by FIB-SEM in starved WT and MTM1 KO HeLa

cells. Heat bar reflects individual mitochondrial volumes. Images are represen-

tative of three cells. (D) Cumulative plot of mitochondrial volume distribution

in starved WT versus MTM1 KO cells derived from FIB-SEM. (E) 3D volumes

of individual mitochondria as in (C) were plotted. (Inset) The largest 3% of

mitochondria occupied 62% of the total mitochondrial volume in WT (gray)

but only 18% in KO (red) cells. The total volume of mitochondria in WT cells

(69 mito = 12.95 mm

) versus KO cells (310 mito = 13.25 mm

) was unchanged.

(F) Mitochondrial volume fraction in starved WT and MTM1 KO HeLa cells (n=

10) analyzed by stereological analysis of thin-sectioned electron micrographs.

(G) Artificial ER–early endosome tethering prevents mitochondrial hyperfusion.

Mean area of individual mitochondria per ROI from starved HeLa cells ±

rapalog (Fig. 2N). EtOH = 251 ROIs, rapalog = 250 ROIs from three independent

experiments. Each ROI represents a single cell. Scale bars: 10 mm (white),

2mm (yellow), 1 mm (black). One-way ANOVA with Tukey’s multiple comparisons

(B); two-tailed unpaired ttest [(F) and (G)]; *P≤0.05, ***P≤0.001,

****P≤0.0001; ns, nonsignificant. Data are mean ± SD [(B) and (G)] or

median ± interquartile range (F).

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

volume of LDs increased in MTM1 KO cells at

the onset of starvation (Fig. 5E and fig. S11O).

LD accumulation persisted upon treatment of

MTM1 KO cells with the pan-caspase inhibitor

Z-VAD-FMK (fig. S11P). These results suggest

that starvation-induced ER tubule-to-sheet

conversion mediated by MTM1 orchestrates the

reflux of FAs to mitochondria for b-oxidation

while counteracting their storage in LDs. We

directly probed this model by monitoring FA

trafficking during early stages of starvation

using the fluorescent FA analog BODIPY 558/

568 C

(Red C12) (Fig. 5F). Pulse-labeled Red

C12 was efficiently transported to mitochondria

in starved WT cells (Fig. 5G and fig. S11Q), con-

sistent with prior data (57). In contrast, in starved

MTM1 KO cells, Red C12 failed to accumulate in

mitochondria and instead was targeted to LDs

(Fig. 5, G to J). Impaired mitochondrial lipid

and fatty acid catabolism was also clearly shown

by the unbiased quantitative proteomic anal-

ysis of fed or starved WT and MTM1 KO cells

using tandem mass tag (TMT) labeling. These

analyses revealed a down-regulation of pro-

teins involved in mitochondrial respiration and

transport (Fig. 5K, fig. S11R, and tables S3 and

S4), for example, mitochondrial very long-chain

specific acyl–coenzyme A dehydrogenase, car-

nitine palmitoyltransferase 2, NADH:ubiquinone

oxidoreductase, and the mitochondrial protein

import factor TIMM17B (Fig. 5L), possibly as

an indirect consequence of the observed struc-

tural mitochondrial defects in starved MTM1

KO cells (Fig. 4). None of these proteins were

altered in KO cells under fed conditions (fig.

S11S), indicative of a specific defect of MTM1 KO

cells to appropriately respond to altered nu-

trient supply. These findings indicate that

MTM1 mediates reshaping of the ER at the

onset of starvation to drive the formation of a

functional mitochondrial network and facili-

tate mitochondrial b-oxidation, which sustains

ATP production. Conversely, defective ER mor-

phogenesis in the absence of endosomal MTM1

impairs mitochondrial metabolic rewiring dur-

ing starvation.

Discussion

How cells and tissues orchestrate adaptive

changes in organelle dynamics and metabo-

lism on a cell-wide scale has remained unclear.

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 8of16

Fig. 5. Defective ER morphogenesis in absence of

endosomal MTM1 impairs mitochondrial metabolic

rewiring during starvation. (A) Basal mitochondrial

oxygen consumption rate (mito-OCR) of starved WT or

MTM1 KO HeLa cells. (B) Mitochondria-dependent

ATP production of starved WT or MTM1 KO HeLa cells.

WT (set to 1) or MTM1 KO HeLa cells. n= 3 independent

experiments. (D) Normalized total cellular ATP

levels of fed versus starved HeLa cells ± rapalog

(Fig. 2N). Data for fed cells (−rapalog) were set

to 1. n= 5 independent experiments. (E) Number

of BODIPY 493/503–labeled lipid droplets (LDs) in

fed or starved (2 hours) WT or MTM1 KO HeLa

cells. WT Fed (n= 157), Starved (n= 181); KO Fed

(n= 153), Starved (n= 148). (F) Schematic depicting

the pulse-chase assay to monitor FA mobilization.

(G) Confocal images of WT and MTM1 KO HeLa cells

pulse-labeled with RedC12 and chased for 2 hours

in EBSS and stained for TOM20. (H) Number of

RedC12-labeled LDs in WT or MTM1 KO HeLa cells

chased for 0 or 2 hours in EBSS. 0h: WT (n= 126),

KO (n= 112); 2h: WT (n= 139), KO (n= 118). (I) Pearson

correlation coefficient of RedC12-labeled FAs and

mitochondria (TOM20) from randomly selected

100 pixel by 100 pixel ROIs in WT or MTM1 KO cells

as in (H). WT = 169 ROIs, KO = 203 ROIs from three

independent experiments. (J) Schematic of trafficking

of pulse-labeled FA RedC12 in WT and MTM1 KO cells

during starvation. (K) Gene Ontology (GO) analysis

of proteins depleted in starved MTM1 KO compared

with WT HeLa cells. Terms related to fatty acid

catabolism and mitochondrial function are highlighted

in red. All of these proteins were unaltered in fed

MTM1 KO cells. (L) Volcano plot of proteins depleted

in starved MTM1 KO compared with WT HeLa cells.

Red dots, proteins enriched from GO analysis shown

in (K). Scale bars: 10 mm. nindicates the total number

of cells analyzed from three independent experiments.

Two-tailed Mann-Whitney test [(B), (E), (H), and (I)];

two-tailed unpaired ttest [(A), (C), and (D)]; ns,

nonsignificant, *P≤0.05, **P≤0.01, ***P≤0.001,

****P≤0.0001. Data are median ± interquartile range

[(A), (B), (E), (H), and (I)] or mean ± SD [(C) and (D)].

100

150

200 ****

# RedC12 LD / Cell

WT KO

0h 2h

Chase

in EBSS:

0.0

0.2

0.4

0.6

0.8

1.0

Pearson coefficient

(TOM20, RedC12)

Chase 2h

in EBSS

**** Fatty acid

RedC12

β-oxidation

Tubular ER

Starvation

(EBSS 2h)

Pulse RedC12

in EBSS (20min)

Fed

condition

Wash out

Chase

in EBSS (0, 2h)

Fed

Starved 2h

Fed

Starved 2h

100

WT KO

****

# Bodipy 493/503 LD / Cell

****

log2 difference (KO / WT)

-4 -2 0 2 4

-log10 p-value

Starved

Proteins from GO list labeled red

COX6B1

ACADVL

NDUFB11

ETFA

TIMM17B

Chase 2h

WTKO

RedC12 TOM20 Merge

WTKO

0.6

0.7

0.8

0.9

1.0

1.1

Norm. total cellular

ATP levels

Starved

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

mito-ATP production

(pmol / min / μg)

WT KO

Starved

-0.5

0.0

0.5

1.0

1.5

2.0

Basal mito-OCR

(pmol / min / μg)

****

WT KO

Starved

(ER-EE tether)

***

0.6

0.7

0.8

0.9

1.0

1.1

1.2

Norm. total cellular

ATP levels

rapalog

Fed

rapalog

Starved

AB CD

EFG

0 4 8 12 16

-log10(P)

GO:0006721: terpenoid metabolic process

R-HSA-6809371: Formation of the cornified envelope

GO:0033539: FA β-oxidation using acyl-CoA DH

GO:0016139: glycoside catabolic process

GO:0006839: mitochondrial transport

R-HSA-6806667: Metabolism of fat-soluble vitamins

WP465: Tryptophan metabolism

GO:0051599: response to hydrostatic pressure

GO:0007029: endoplasmic reticulum organization

ko04142: Lysosome

GO:0061684: chaperone-mediated autophagy

R-HSA-1428517: TCA cycle and ETC

R-HSA-5668914: Diseases of metabolism

GO:0051262: protein tetramerization

R-HSA-9609507: Protein localization

CORUM:387: MCM complex

M5887: NABA BASEMENT MEMBRANES

ko00280: Valine, leucine and isoleucine degradation

GO:0007005: mitochondrion organization

GO:0044242: cellular lipid catabolic process

GO : Down-regulated proteome by KO upon starvation

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

Here we reveal a key role for early endosomal

lipid signaling mediated by MTM1, a lipid

phosphatase mutated in XLCNM in humans

(19,30,31), in controlling the tubular ER and,

thereby, mitochondrial morphology and meta-

bolic function in the acute response to fluctuating

nutrient conditions. This mechanism operates

independent of ER-phagy, a process typically

induced by prolonged starvation (65). We dem-

onstrate that starvation-induced PI(3)P hydro-

lysis by endosomal MTM1 reduces previously

undescribed membrane contacts between pe-

ripheral ER tubules and early endosomes. These

contacts act as physical tethers that may trans-

mit pulling forces from highly motile periph-

eral endosomes to the ER and are mediated by

endosomal PI(3)P binding to RRBP1, a large ER

membrane protein overexpressed in colorectal

cancer (66), and its close paralog kinectin 1.

Whether ER–early endosome membrane con-

tacts also enable material exchange in vivo as

shown for contact sites between the ER and

the plasma membrane, the trans-Golgi network,

or lysosomes (37) remains to be determined.

Loss of ER tubules, possibly in conjunction

with ER-independent mechanisms, drives mito-

chondrial network formation and, directly or

indirectly, facilitates FA transfer to mitochon-

dria to fuel b-oxidation and, thereby, mitochon-

drial ATP production to sustain cellular energy

homeostasis. The precise relationship between

mitochondrial morphogenesis, FA mobilization

to mitochondria, and mitochondrial ATP pro-

duction remains to be defined. Interestingly,

ER sheets are favored over tubules from an

energetic perspective (67,68)and,hence,

should prevail under conditions of nutrient

starvation when cellular energy levels are low.

Consistent with this, it has recently been shown

that the hepatic ER in obese mice is character-

ized by disorganized ER sheets and a predomi-

nance of ER tubules and accompanying defects

in lipid metabolism (69). Our findings thus

identify an organellar conveyor belt, in which

the tubular ERserves as a membrane conduit

that transmits nutrient-triggered changes (i.e.,

in glutamine and other amino acids) in early

endosomal PI(3)P levels to metabolic organ-

elles such as LDs and mitochondria [in agree-

ment with (56)] to enable metabolic rewiring

under conditions of limited nutrient supply

and, possibly, in cancer [e.g., when RRBP1 is

overexpressed (66)]. Defects in ER shape, mito-

chondrial morphogenesis, and cellular ATP de-

pletion caused by loss of MTM1 function can

explain the observed myofiber hypotrophy and

defective sarcoplasmic reticulum organization

in animal models of XLCNM (16,17) and in

human patients who often appear undernour-

ished (19,30,31). Furthermore, it is conceivable

that reduced contact formation between early

endosomesandERtubulesduetoMTM1-

mediated PI(3)P hydrolysis, in addition to its

effects on ER shape and mitochondrial func-

tion, may facilitate endosomal exocytosis of

b-integrins, a mechanism shown to be defec-

tive in XLCNM (15). How precisely early endo-

somal PI(3)P levels and MTM1 function are

controlled by cellular nutrient status remains

poorly understood. VPS34, the main PI(3)P-

synthesizing enzyme on endosomes has been

reported to be stimulated by fed signals (70),

that is, conditions in which MTM1 activity

is repressed.

The endosomal signaling lipid–based path-

way to control oxidative cell metabolism un-

covered in this work may synergize with other

cellular mechanisms that impinge on the dy-

namics of metabolically active organelles. For

example, it has been shown that the function

and localization of lysosomes depend on motor

proteins (8) as well as on their association with

the ER (41) and are regulated by cellular nu-

trient status (71), which in turn affects nutrient

signaling (28,72). Late endosomes (i.e., organ-

elles distinct from the Rab5-positive early en-

dosomes described here) have been shown to

undergo fission at sites of contact with the ER

that are molecularly and functionally distinct

(73)fromtheER–earlyendosomecontacts

identified in this study. Finally, mitochondria–

lysosome contacts have been shown to regu-

late mitochondrial fission (50). Whether any

of these contacts are subject to nutrient regu-

lation and impact on cell metabolism will need

to be addressed in future studies. Conceivably,

lipid phosphatases including other members of

the myotubularin family (26), many of which

are linked to human disease, may play crucial

physiological roles in the regulation of these

and other membrane contacts.

Materials and methods

Materials

Plasmids

Plasmids used: mEmerald-Sec61b-C1 (Addgene

#90992), mCh-Rab5 (Addgene #49201), ER-

GCaMP6-1-150 (Addgene #86918), tdTomato-

BICD2-FKBP (Addgene #64205), and mCh-Climp63

(mouse) (Addgene #136293). HA-BICD2-FRB was

kindly provided by G. G. Farías. Plasmids for

transient transfection (e.g., mCherry MTM1

WT, mRFP-FKBP-empty, mRFP-FKBP-MTM1

WT, mRFP-FKBP-MTM1 C375S, FRB-iRFP-

Rab5A, mRFP-FKBP-Rab5A, TMEM192-3xHA-

FRB, ERM-2xHA-FRB) were generated with

the pcDNA3.1(+) vector and polymerase chain

reaction(PCR) or ligation-based cloning. Note

that ERM is the N-terminal ER membrane

targeting sequence of residues 1 to 27 of ER-

resident P450 oxidase 2C1. Full-length RRBP1

was amplified from pcDNA4 HisMax-V5-GFP-

RRBP1(Addgene#92150)andinsertedinto

pcDNA3.1(+)-based mCherry expression vec-

tor with tags at its C terminus. PCR-amplified

RRBP (amino acids 1–150) was inserted in

pGEX4T-1 vector by ligation-based cloning.

Plasmids for lentivirus transduction [e.g.,

mEmerald-Sec61b, mScarlet-Sec61b,3xHA-

APEX2-Rab5A, Halo-Rab5A, ERM-GFP11-p2a-

GFP1-10-Rab5A, ERM-2xHA-FRB, ERM-2xHA-

(EAAAK)x9-FRB, mRFP-FKBP-Rab5A, Rab5A-

mRFP-FKBP, V5-RRBP1 1-150aa WT, V5-RRBP1

del PB1, V5-RRBP1 del PB3, V5-RRBP1 del TM]

were generated with the pLVX-TetONE puro

vector (Takara, Catalog 631849) by PCR, re-

petitive oligo annealing, or inverse PCR with

Gibson assembly. All constructs were verified

by double-stranded DNA sequencing.

Primary antibodies for immunoblots

Anti-GAPDH (glyceraldehyde-3-phosphate de-

hydrogenase; mouse, Sigma-Aldrich, G8795,

1:1000), anti-MTM1 (goat, Invitrogen, PA5-

17972, 1:250), anti-Calnexin (rabbit, Abcam,

ab75801, 1:2000), anti-calreticulin (rabbit,

Thermo Fisher, PA 3-900, 1:1000), anti-Reticulon4

(Nogo) (mouse, Santa Cruz, sc-271878, 1:1000),

anti-RPL26 (rabbit, Proteintech, 17619-1-AP,

1:1000),anti-RPL7 (rabbit, Proteintech, 14583-

1-AP, 1:1000), anti-EEA1 (mouse, BD biosciences,

610456, 1:500), anti-Histone H1 (rabbit, Abcam,

ab17729, 1:1000), anti-GST (mouse, Thermo

Fisher, MA4-004, 1:1000), anti-KTN1 (rabbit,

Proteintech, 19841-1-AP, 1:500), anti-RRBP1

(rabbit, Proteintech, 22015-1-AP, 1:500), anti-

Clim63 (mouse, Enzo Life Sciences, ENZ-

ABS669-0100, 1:2000), anti-TOM20 (mouse,

Santa Cruz, sc-17764, 1:200), anti-p-DRP1 (S616)

(rabbit, Cell signaling, 3455S, 1:500), anti-

DRP1 (rabbit, Abcam, ab184247, 1:500), anti-

MFN1 (rabbit, Proteintech, 13798-1-AP, 1:500),

anti-MFN2 (rabbit, Proteintech, 12186-1-AP,

1:500), anti-OPA1 (rabbit, Proteintech, 27733-

1-AP, 1:500), anti- GRP78/BIP (rabbit, Proteintech,

11587-1-AP, 1:3000), anti-CHOP (rabbit, Pro-

teintech, 15204-1-AP, 1:500), anti-LC3-II (mouse,

MBL, M152-3, 1:200), anti-Cleaved PARP(Asp214)

(rabbit, Cell Signaling, 9541S, 1:1000), anti-

Cleaved Caspase3(Asp175) (rabbit, Cell Signal-

ing, 9661T, 1:250), anti-V5 (mouse, Invitrogen,

P/N 46-0705, 1:1000), anti-Phospho-p70 S6

Kinase (Thr389) (rabbit, Cell signaling, 9205L,

1:500), anti-p70 S6 KinaseAntibody(rabbit,Cell

Signaling, 9202L, 1:1000), anti-ATG5 (rabbit,

Proteintech, 10181-2-AP, 1:1000).

Primary antibodies for

immunofluorescence

Anti-calreticulin (rabbit, Thermo Fisher, PA

3-900, 1:200), anti-calreticulin (rabbit, Abcam,

ab92516, 1:200), anti-Rab5A (mouse, BD

biosciences, 610724, 1:100), anti-GFP (mouse,

Invitrogen, A-11120, 1:500), anti-LC3-II (mouse,

MBL, M152-3, 1:100), anti-HA (mouse, Santa

Cruz, sc-7392, 1:500), anti-RFP (rabbit, MBL,

PM005, 1:400), anti-LAMP1 (mouse, BD bio-

sciences, 555798, 1:1000), anti-LAMP1 (rabbit,

Cell signaling, 9091P, 1:300), anti-V5 (mouse,

Invitrogen, P/N 46-0705, 1:200), anti-TOM20

(mouse,SantaCruz,sc-17764,1:500),anti-

TOM20 (rabbit, Abcam, ab186734, 1:500),

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 9of16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

anti-TFEB(rabbit, Biomol, A303-673A, 1:200),

Streptavidin, Alexa Fluor647conjugate(Thermo

Fisher, S21374).

siRNAs

The small interfering RNAs (siRNAs) used

were: Scrambled siControl 5′-CGUACGCG-

GAAUACUUCGA-3′, or Sigma MISSION Uni-

versal Negative Control (SIC001), siMTM1-1

5′-GATGCAAGACCCAGCGTAA-3′, siMTM1-2

5′-TATGAGTGGGAAACGAAATAA-3′,siMTM1-

SP (ON-TARGETplus SMARTpool, Dharmacon,

L-008036-00-0005), siMTMR1-1 5′-GAGA-

TAGTGTGCAAGGATA-3′, siMTMR1-2 5′-CG-

CTGATACCAACAAGACAAA-3′,siMTMR2-1

5′-GGACATCGATTTCAACTAA-3′,siMTMR2-2

5′-CGGCCAAGTGTTAATGCTGTT-3′, siMTMR2-

35′-GTAGAAAGTCTTCGGAATTTA-3′,

siMTMR6-1 5′-GGACTACAAGATTTGTGAA-3′,

siMTMR6-2 5′-CGGGACTACAAGATTTGTGAA-

3′,siMTMR12-15′-CCAGGTGAACAGCTGCTTT-3′,

siMTMR12-2 5′-GCAAGAGAATTAGCAAACTTA-

3′, siMTMR12-3 5′- CGCTTCAAACATCAACGA-

CAA-3′, siKTN1 (ON-TARGETplus SMARTpool,

Dharmacon, L-010605-00-0005), siERLIN1

(ON-TARGETplus SMARTpool, Dharmacon,

L-015639-01-0005), siERLIN2 (ON-TARGETplus

SMARTpool, Dharmacon, L-017943-01-0005),

siRRBP1 (ON-TARGETplus SMARTpool,

Dharmacon, L-011891-02-0005), siOSBPL8

(ON-TARGETplus SMARTpool, Dharmacon,

L-009508-00-0005), siITPR2 (ON-TARGETplus

SMARTpool, Dharmacon, L-006208-02-0005),

siARL8B (ON-TARGETplus SMARTpool,

Dharmacon, L-020294-01-0005), siRTN4

(ON-TARGETplus SMARTpool, Dharmacon,

L-010721-00-0005), siRab10 (ON-TARGETplus

SMARTpool, Dharmacon, L-010823-00-0005),

siProtrudin 5′-CTTCTTGATCCAGCTGGAGG-

3′, siATG5 (ON-TARGETplus SMARTpool,

Dharmacon, L-004374-00-0005).

Cell culture

HeLa, human embryonic kidney 293 T

(HEK293T), and Cos7 cells were obtained

from ATCC. Cells were cultured in Dulbecco’s

modified Eagle’smedium(DMEM)high

(4.5 g/liter glucose, Thermo Fisher, 41965062)

containing 10% fetal bovine serum (FBS;

Thermo Fisher, 10270106) and 50 units/ml

penicillin, 50 mg/ml streptomycin (Thermo

Fisher, 15070063). Cells were routinely tested

for mycoplasm contamination. The human

myoblast cell line KM1288 was derived from

the deltoid muscle of a patient with XLCNM

and carries a missense genomic mutation

(c.205<T) in exon 4 of the MTM1 gene (20).

The human myoblast cell strain NL15 was

derived from the quadriceps of a patient with

XLCNM and carries the missense genomic

mutation R241C in the MTM1 gene (21). The

control myoblast cell line AB1190 was derived

from the paravertebral muscle of a healthy

individual. AB1190 and KM1288 were immor-

talized as described before (74). Myoblasts were

cultured in the homemade medium: 1 volume

medium 199 (Thermo Fisher, 41150020) + 4

volumes DMEM high, supplemented with

20% FBS, Fétuin (25 mg/ml, Thermo Fisher,

10344026), human epidermal growth factor

(5 ng/ml, Thermo Fisher, PHG0311), basic

fibroblast growth factor (0.5 ng/ml, Thermo

Fisher, PHG0026), insulin (5 mg/ml, Sigma,

91077C-1G), and dexamethasone (0.2 mg/ml,

Sigma, D4902-100mg). To induce starvation, cells

were washed with prewarmed Earle’s balanced

salt solution (EBSS; Thermo Fisher, 24010043)

five times for 10 s each and then incubated

in EBSS in 5% CO

at 37°C for 2 hours, unless

indicated otherwise. This washing step is im-

portant to fully remove remaining nutrients

and growth factors. For steady-state (fed) con-

ditions, cells were incubated overnight in fresh

complete DMEM medium supplemented with

10% FBS. For the experiment in fig. S1M, EBSS

was supplemented, as indicated, with 5% di-

alyzed FBS (One Shot format, Gibco, A3382001)

or insulin (final at 5 mg/ml) or glucose (final at

4.5 g/liter) or sodium pyruvate (final at 1 mM)

(Gibco, 11360070) or MEM essential amino

acids (50x) solution (Sigma, M5550) (final 1x

dilution) or MEM solution or nonessential

amino acids (100x) (Gibco, 11140050) (final 1x

dilution) or L-glutamine (final at 4 mM) (Gibco,

25030081).

CRISPR-Cas9–mediated genome engineering

Guide RNAs targeting genomic human MTM1

exon 2 (sgMTM1: 5′AGTTGATGCAGAAGC-

CATCC 3′) was cloned into Lenti-CRISPRv2

(Addgene plasmid # 52961). HeLa cells were

transfected using FuGene-6 as a transfection

reagent. Cells were t selected with puromycin

(2 mg/ml) for 72 hours. Surviving cells were

diluted and plated into 96-well plates with a

density of 1 cell per well. Expanded colonies

were screened by immunoblotting using anti-

MTM1 antibodies.

Generation of doxycycline-inducible stable

cell lines

Lentivirus was generated by transient trans-

fection of HEK293T cells seeded in 10-cm cell

culture plates at 80 to 90% confluency with

pCMV delta R8.2 (3.5 mg), VSV-G (0.5 mg), and

pLVX-TetONE puro-based constructs (4 mg)

combined with 16 ml of JetPrime in 400 ml

JetPrime buffer. After 16 hours of transfection,

cells were replenished with 7 ml fresh DMEM.

After 48 hours, the supernatant was collected,

and cellular debris was removed by centri-

fugation (3000 rpm, 20 min). Viral super-

natant (4 to 5 ml) was added to the cells with

polybrene (Merck, Cat.#TR-1003-G) at 10 mg/ml.

Cells were incubated with virus for 16 hours

and then replenished with fresh DMEM con-

taining puromycin (2 mg/ml) followed by selec-

tion for 2 to 3 days. After selection, cells were

stabilized for 3 to 4 days in culture without

puromycin. For the induction of protein ex-

pression, doxycycline (1 mg/ml) was added for

16 hours.

Transient transfection

HEK293T and HeLa cells and myoblast cells

were transfected with plasmids using jetPRIME

(PolyPlus, 101000001) or FuGene-6 (Promega,

E2691) or ViaFect (Promega, E4981) according

to the manufacturer’s protocol, respectively

[e.g., DNA (micrograms): reagent (microliters)

ratio of 1:2]. After 18 to 24 hours of transfection,

cells were further treated and analyzed. For

siRNA transfection, cells were transfected with

the indicated siRNA (20 nM) using jetPRIME

according to the manufacturer’s protocol. After

48 hours of transfection, cells were further

treated and analyzed.

Dyes and pharmacological inhibitors

Dyes

TMRE (tetramethylrhodamine ethyl ester per-

chlorate): Cell signaling, Mitochondrial mem-

brane potential assay kit, #13296, 200 nM).

Lysotracker Red DND-99: Thermo Fisher,

L7528, 1 mMfor45min.Notethatforfixedcell

samples, buffer containing detergents should

not be used, in order to preserve lysotracker

staining. BODIPY 493/503: Thermo Fisher,

D3922, 2 mM for 20 min in serum free media.

No detergent-containing buffers should be used.

Before treatment, cells were washed twice with

serum-free media. BODIPY 558/568 C

:Thermo

Fisher, D3835, 1 mM for 20 min in serum-free

media. Before labeling, cells were washed twice

with serum-free media. For TOM20 antib ody

co-staining, fixed samples were permeabilized

with 20 mM digitonin for 5 min.

MitoTracker Deep Red FM: Thermo Fisher,

M22426, 100 nM for 30 min in full medium.

CellMask Deep Red Plasma membrane Stain:

Thermo Fisher, C10046, fixed cells were stained

for 20 min at 1:2000 dilution. Halo-tag li-

gands (CA-JF646) were chemically synthesized

as described in (75) and used at 100 nM for 10

to 16 hours.

Pharmacological inhibitors

VPS34-IN1 (Selleckchem, S7980, 1mM), apilimod

(Echelon Biosciences, B-0308, 50nM), rapalog

(Takara, A/C Heterodimerizer 635056, 0.5 mM),

nocodazole (Sigma, M1404, 5mM), dimethyl sulf-

oxide (DMSO; Sigma, D2650), thapsigargin

(Sigma, T9033, 3mM), Torin1 (Tocris, Cat. No.

4247, 1 mM), Z-VAD-FMK (Tocris, Cat. No.

2163, 25 mM).

RNA isolation, RT-PCR, and qRT-PCR

Total RNA was isolated using the RNeasy Plus

Mini Kit (Qiagen) according to the manufac-

turer’s instructions. Quantification of RNA

yields was done with a multimode microplate

reader (SPECTROstar Nano from BMG Labtech).

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 10 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

A reverse transcription kit (SuperScript IV;

Invitrogen) was used to reverse transcribe

RNA (800 ng) in a 20 ml reaction using oligo(dT)

and random hexamer Primers. Reverse tran-

scription polymerase chain reaction (RT-PCR)

was performed with specific primers set against

each gene with indicated cycles. For quan-

titative reverse transcription polymerase chain

reaction (qRT-PCR), SYBR Green Master Mix

(BioRad) was used according to the manufac-

turer’s protocol with BioRad CFX Connect.

qRT-PCR primers

GAPDH Fwd: 5′-CTTCGCTCTCTGCTCCTCCT-3′;

Rev: 5-GTTAAAAGCAGCCCTGGTGA-3′,MTM1

Fwd: 5′-GTTTGAGATCCTCACGAGATACG-3′;

Rev:5′-GTCCATCCATCCACGTTAAACTT-3′,

MTMR1 Fwd: 5′-CCTTGATGTTCCCCTTGGAGT-

3′; Rev:5′-GTGCCTGTCCGTTAGAAAGAG-3′,

MTMR2 Fwd: 5′-GTGGAAAGCGAAGCAAA-

GAAG-3′;Rev:5′-CTTGGCCGGGCATCAAATATAA-

3′,MTMR3Fwd:5′-GACTGAACAACGCAATCC-

GAC-3′;Rev:5′-CCTTGAAGTTACATGCTCCCC-3′,

MTMR4 Fwd: 5′-CCAAGCCAAGGATCTGTTCCC-

3′; Rev:5′-GCCGGTAGTTAGAGATGGCAA-3′,

MTMR5 Fwd: 5′-CGACCACACGGAGGTGTTC-

3′;Rev:5′-GGTTCCCAATCACGTTCTCCA-3′,

MTMR6 Fwd: 5′-GTTCCCCGGATAGCAAGCAAA-

3′;Rev:5′-GTGGCTGACTACATCGACAAAT-3′,

MTMR7 Fwd: 5′-TCCGCTTGGTAGATCGAGTGT-

3′;Rev:5′-TTTTCCACGAATATGACATGGGT-3′,

MTMR8 Fwd: 5′-TGCACTCCATCACATTGCCA-

3′;Rev:5′-GGCACACAAGGTCAGAATCTAA-3′,

MTMR9 Fwd: 5′-TGAAGCTCTTCGGAAGG-

TAGC-3′;Rev:5′-GTGGCTGACCACTTCGCATAA-3′,

MTMR10 Fwd: 5′-ATCCACTTGCCTTCCAGAA-

TACA-3′;Rev:5′-CACAAGAGCACTGCCGTTAGA-3′,

MTMR11 Fwd: 5′-GCTGCTCAGAGTTGGTTTTGA-

3′;Rev:5′-CCCCGAATACTGTTGGGCTT-3′,

MTMR12 Fwd: 5′-GGCTCCTAAACTGCTTAAA-

CGA-3′; Rev:5′-GTTGCCTTTGGTCCGTTCCA-3′,

MTMR13 Fwd: 5′-TCATCGTGGTAGGCTAT-

GACC-3′; Rev:5′-CCAGGCTGACAAAACAACTCA-3′,

MTMR14 Fwd: 5′-GGAGTTCTCCCGGACTCAGTA-

3′; Rev:5′-AACAGTAGTCTCGGCCAAACA-3′.

Immunoblot analysis

Cells were lysed with radioimmunoprecipita-

tion assay (RIPA) buffer (50 mM Tris–Cl pH 7.5,

150 mM NaCl, 1% NP-40, 0.5% sodium deoxy-

cholate, 0.1% SDS) containing protease and

phosphatase inhibitors. Lysates were incu-

bated for 10 min on ice before centrifugation

at 17,000gfor 10 min at 4°C. Protein concen-

tration was measured by Bradford or bicin-

choninic acid (BCA) assays. Cell lysates in

Laemmli sample buffer were boiled for 10 min.

Between 20 and 40 mg of protein was resolved

by SDS–polyacrylamide gel electrophoresis

(SDS-PAGE). Immunoblotting was done on

nitrocellulose membranes. Membranes were

incubated with the indicated primary anti-

bodies at 4°C overnight. The next day, bound

primary antibodies were detected by incu-

bation with IRDye 680/800CW-conjugated or

horseradish peroxidase (HRP)–conjugated sec-

ondary antibodies via the Odyssey Fc Imaging

system (LI-COR Biosciences).

Light microscopy

Immunocytochemistry

Cells were seeded on Matrigel-coated coverslips

(BD/Corning), fixed in 4% paraformaldehyde

(PFA) for 15 to 20 min at room temperature (RT).

Cells were washed three times with phosphate-

buffered saline (PBS) before incubation with

3% bovine serum albumin (w/v) in 0.3% PBST

(Triton X-100) for 20 min. Using the same buffer,

the cells were incubated with primary antibodies

for 2 hours at RT, washed three times with 0.3%

PBST, and then incubated with secondary anti-

bodies for 1 hour at RT. After washing with 0.3%

PBST, cells were mounted withHoechst 33258

(Invitrogen, H3569, 1:2000) to counter stain

nuclei. Images were acquired on Zeiss 710 or

780 Laser Scanning Confocal Microscopes

using ZEN.

Note for the specific staining conditions:

For calreticulin staining, cells were fixed with

37°C 4% PFA for 34 min at 37°C. For Rab5A

staining, cells were fixed with 37°C 4% PFA for

8 min at 37°C. For LC3-II staining, cells were

permeabilized with 20 mM digitonin for 5 min

at RT. After permeabilization, all subsequent

procedures were done in PBS.

PI(3)P staining

Cells were fixed in 2% PFA for 15 min at RT,

washed twice in PBS with 50 mM NH

Cl, and

permeabilized with 20 mM digitonin in buffer

A (20 mM PIPES pH 6.8 with NaOH, 137 mM

NaCl, 2.7 mM KCl) for 5 min. Note that per-

meabilization is critical for successful PI(3)P

staining, so depending on the batch of digitonin,

one might need to optimize the concentration

and incubation time.

Cells were washed three times in buffer A

before addition of purified GFP (or mCherry)-

2xFYVE

Hrs

at 0.25 mg/ml in buffer A with 5%

normal goat serum, 50 mM NH

Cl for 1 hour.

Samples were washed and decorated with anti-

bodies against GFP or RFP in buffer A with 5%

normal goat serum, 50 mM NH

Cl for 2 hours.

Cells were washed three times in buffer A, in-

cubated for 1 hour with secondary antibodies

in buffer A with 5% normal goat serum, 50 mM

Cl. Samples were washed three times with

buffer A, cells were postfixed for 5 min in 2%

PFA, washed three times in PBS, and mounted

with Hoechst 33258. Fluorescent sum inten-

sity was measured from individual cells and

normalized to the level of PI(3)P detected in

fed cells set to 1.

Live-cell imaging

Cells (6 × 10

to 8 × 10

) were seeded on

Matrigel-coated 8-well chamber slides (ibidi,

80827). The next day, live-cell imaging was

carried out using Nikon-CSU Yokogawa Spin-

ning disk (CSU-X1) microscope equipped with

an EMCCD Camera (Andor AU-888) at 37°C in

the presence of 5% CO

.Imageswereacquired

using Nikon Elements.

Image analysis and quantification

Acquired images were first segmented to re-

solve individual cells by manually drawing each

cell boundary using the Fiji freehand selec-

tion tools combined with Clear Outside option.

Individual cell image files were saved in the in-

put folder and batch processed in CellProfiler

[v4.1.3 (76)] or in Fiji. CellProfiler pipeline

modules used in this study were posted at

https://cellprofiler.org/published-pipelines.

Sheet-to-tubular ER ratio

Total ER (calreticulin or mEmerald-Sec61b)

segmentation was done using a three-class

Otsu threshold followed by conversion into

binary images. ER area was then measured

using the MeasureObjectSizeShape module

in CellProfiler. From the same image, sheet

ER was segmented in a similar manner except

using the Minimum Cross-Entropy threshold.

Depending on signal-to-noise ratio and im-

age quality, threshold settings (e.g., Thresh-

old smoothing scale, correction factor) were

adjusted using the IdentifyObjects module in

CellProfiler. Sheet ER area was then divided

by total ER area using CalculateMath module

in the same pipeline of CellProfiler. Calculated

sheetERarea/totalERarearatiowasdefined

as “Sheet ER ratio.”

Mean area and number of mitochondria

per ROI

Regions of interest (ROIs; e.g., 15 mmby15mm)

were randomly generated from individual

cells in Fiji, exported to CellProfiler, and

filtered with a Gaussian filter module with

1 or 2 sigma. ROIs were then processed using

the EnhanceOrSuppressFeatures module with

the following options: Operation = “Enhance,”

Feature type = “Neurites,”Enhancement

method = “Tubeness,”Smoothing scale =

“1.”After binarization, mitochondria were

segmented using the Robust Background

Threshold in the IdentifyObjects CellProfiler

module. The number and area of individual

mitochondria per ROI were calculated using

MeasureObjectSizeShape module in CellProfiler.

Lysosome position

Images were processed in CellProfiler using

the EnhanceOrSuppressFeatures module with

the following options: Operation = “Enhance,”

Feature type = “Speckles,”Feature size = “4-6.”

Lysosome were then segmented using Robust

Background threshold in the IdentifyObjects

module. The distance between individually

identified lysosomes from the cell centroid

was calculated using the RelateObjects module

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 11 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

in CellProfiler. The standard deviation of the

distance of lysosomes from the centroid define

the extent of lysosomal dispersion or clustering.

Early endosome distribution

Rectangular ROIs (80 pixels by width pixels rang-

ing from cell center to cell boundary) were ran-

domly selected from individual cells in Fiji and

processed using the EnhanceOrSuppressFeatures

module with the following options: Operation =

“Enhance,”Feature type = “Speckles,”Feature

size = “4-6.”A Gaussian filter with sigma = 1 or

2 was applied in CellProfiler. Early endosomes

and lysosomes were segmented using two-

classes Otsu and robust background thresholds

using the IdentifyObjects module. To compare

the partitioning of early endosomes and lyso-

somes between the cell center and the periphery,

segmented individual objects were shrunken to

a dot using the ExpandOrShrinkObjects module

with the “Shrink objects to a point”option in

CellProfiler and converted into binary image

files. The MeasureObjectIntensityDistribution

module was then used to measure the inten-

sity distribution from each object’s center to

its boundary within a set of bins with the

following options: Bin = “5,”Measurement =

“Fraction at distance.”The relative intensity

fraction among the five bins was calculated

from the binary images in CellProfiler. This

intensity value indirectly represents the frac-

tion of endosomes per bin.

Fluorescent imaging of ER–early

endosome contacts and dynamics

Tubular ER segmentation of 100 pixel by

100 pixel ROIs from the cell periphery were

generated using the Minimum Cross-Entropy

threshold option in the IdentifyObjects mod-

ule, and the ExpandOrShrinkObjects module

with Operation = “Skeletonize each object,”

and subsequent ExpandOrShrinkObjects

module with Operation = “Expand objects

by a specified number of pixels”setting with

1-pixel expansion in CellProfiler. Early endo-

somes and lysosomes from the same image

were segmented as described above. Segmented

individual objects were then shrunken to a dot

using the ExpandOrShrinkObjects module with

“Shrink objects to a point”option, followed

by Operation = “Expand objects by a specified

number of pixels”setting with 1-pixel expansion.

To calculate the number of endosomes that

contact the ER, the MeasureObjectNeighbors

module in CellProfiler was operated with

the following options: Method to determine

neighbors = “Within a specified distance”set-

ting distance = 1 pixel. If the processed tubular

ER and endosomes overlapped with each other

at least by 1 pixel, a contact was scored.

Split GFP ER–early endosome contact sensor

To quantitatively determine the number ER–

earlyendosomemembranecontactsbymeasur-

ing the number of GFP puncta, images were

blurred with a Gaussian filter (sigma = 1 or 2,

depending on the signal-to-noise ratio), back-

ground was subtracted with a rolling ball ra-

dius of 50 pixels, and segmented using the Find

Maxima tool with prominence = 10 in Fiji.

STED image analysis of ER–early

endosome contact

ROIs (8 mmby8mm) were randomly gen-

erated from individual cells in Fiji, ex-

ported to CellProfiler, processed using the

EnhanceOrSuppressFeatures module with

the following options: Operation = “Enhance,”

Feature type = “Speckles,”with subsequent

Gaussian filter module with 1. This allowed the

segmentation of early endosomes via robust

background thresholds using the IdentifyObjects

module. Segmented early endosomes were

then converted to binary images using the

ConvertObjectsToImage module. ER images

were filtered with a Gaussian filter module with

1 and converted to binary images using the

Threshold module. The MeasureImageOverlap

module was used to analyze the fractional

overlap between the ER and early endosomes.

Lipid droplets (BODIPY 493/503, Red C12)

Lipid droplet images were blurred with a

Gaussian filter (sigma = 1), enhanced using

the EnhanceOrSuppressFeatures module

(Operation = “Enhance”, Feature type =

“Speckles”with size = 10), and then segmented

by the three-classes Otsu threshold option in

CellProfiler. Segmented objects were further

filtered on the basis of their measured sizes

and intensities using the FilterObjects module

in CellProfiler to minimize spurious detec-

tions. To measure the volume of lipid droplets,

z-stacked confocal images were analyzed using

the 3D objects counter plug-in in Fiji.

Pearson coefficient

To quantitatively assess colocalization between

two channels, two or three randomly chosen

100 pixel by 100 pixel ROIs from the cell pe-

riphery were blurred with a Gaussian filter

(sigma = 1). Pearson’s coefficients were calcu-

lated using the Coloc2 plugin in Fiji or the

MeasureColocalization module with corre-

lation option in CellProfiler.

Super-resolution microscopy

Stimulated emission depletion (STED) images

were taken with either a Leica SP8 TCS STED

microscope (Leica Microsystems) for fixed

samplesoraSTEDYCON(AbberiorInstruments

GmbH, Göttingen) for live cell imaging. Leica

SP8 TCS STED microscope was equipped with

a pulsed white-light excitation laser [WLL;

∼80-p s p uls e width, 80 MHz repetition rate

(NKT Photonics)] and a STED laser for de-

pletion (775 nm). The system was controlled

by the Leica LAS X software. For images of

mEmerald-Sec61b, fixed cells were further

stained for GFP-Booster nanobody conjugated

to Atto647N (Chromotek, gba647n-100, 1:200).

Fluorophore specific excitation (Ex.) and emis-

sion filter (EmF.) settings: Atto647N (Ex.:

640 nm; EmF.: 650 to 700 nm). Time-gated

detection was set from 0.3 to 6 ns. The fluo-

rescence emission signal was collected by hy-

brid detectors (HyD). Images were acquired

with a HC PL APO CS2 100×/1.40 NA oil ob-

jective (Leica Microsystems) in a pixel size of

18.9 nm by 18.9 nm. STEDYCON was mounted

on the Nikon Eclipse Ti research microscope

equipped with a Plan APO 100x/1.45 NA oil

objective (Nikon). Excitation laser with a wave-

lengthof640nmwasusedfortheHaloligand

JF647 fluorescence. The wavelength of the

STED laser was 775 nm. Live cells expressing

Halo-Sec61bwere imaged at 37°C degree with

a pixel size of 20 nm by 20 nm.

Focused ion beam milling scanning electron

microscopy (FIB-SEM)

Cells were fixed with 2% glutaraldehyde (GA)

in PBS, washed with 0.1 M cacodylate buffer,

and embedded in Durcupan following a mod-

ified rOTO protocol with 1% (w/v) OsO4/

1.5% (w/v) K3Fe(CN)6 in 0.1 M cacodylate buf-

fer (pH 7.4); 0.2% (w/v) thiocarbohydrazide in

water; 1% OsO4 in water, 1% aqueous uranyl

acetate with corresponding washes in be-

tween and subsequent acetone dehydration

and resin infiltration. For polymerization, cov-

erslips were mounted onto pre-polymerized

resin blocks and placed into a heating cupboard

for 48 hours. After polymerization, coverslips

were removed by liquid nitrogen treatment,

and blocks were glued to SEM aluminum stubs

with conductive silver epoxy and carbon sputter

coating (≈30 nm).

Helios 5CX FIB-SEM autoslice and view

workflow was used to section and image the

embedded cells. FIB was run at 30 kV, 0.23 nA,

and 10-nm milling step. Cross section images

were scanned at 1.5 to 2 kV and 86 pA to 0.17 nA.

Dwell time was 5 ms at 3.37-nm pixel resolution

and ICD detection. The resulting 3D stacks

were binned to isotropic 10-nm voxel resolution.

Alignment, segmentation of mitochondr ia, and

analysis of ER networks were performed using

Microscopy Image Browser, 3D visualization

was done using Imaris. To access how con-

tinuous or fenestrated the ER is, the length

of ER cisternae was measured in several dis-

tantly positioned 2D sections from FIB-SEM

stacks. Sections for measurements from each

cell were randomly chosen and were at least

1mm apart from each other.

For 2D analysis of mitochondrial cristae

length, ultrathin sections of cells were collected

onto carbon-coated coverslips and imaged

with a Helios 5CX SEM. The length of cristae

was measured using Image J and normalized

per mitochondrial cross-section perimeter.

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 12 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

For 2D analysis of ER–early endosome mem-

brane contacts, 5 nm BSA-gold was allowed to

internalize into HeLa cells for 5 min to reach

early endosomes. Cells were washed before fix-

ation with 2% GA. Cells were processed for EM

analysis as described above. Ultrathin sections

mounted onto EM grids were imaged using a

Zeiss 900 transmission electron microscope

and ER–early endosome contact sites were

measured in Fiji.

Volume fraction analysis was performed

on ultrathin sections mounted on wafers and

scanned with a directional back-scatter de-

tector at 2 kV, 0.34 nA, 5-msdwelltime.A

measurement grid (1 mm

per point) was super-

imposed over cellular profiles and the refer-

ence volume (cytoplasm), as well as the relative

volumes of the ER, and mitochondria were es-

timated according to stereological principles.

APEX2 proximity labeling experiments

HeLa cells stably expressing doxycycline-

inducible 3xHA-APEX2-Rab5A were pretreated

with doxycycline for one day. Cells were then

incubated with 500 mM phenol-biotin (Sigma,

SML2135) at 37°C for 30 min. To enable ef-

ficient biotinylation, cells were treated with

1 mM hydrogen peroxide and incubated at

RT for 1 min. The reaction was terminated by

removing the medium and the addition of ice-

cold quenching buffer (10 mM sodium azide,

10 mM sodium ascorbate, 5 mM Trolox in PBS).

Cells were washed with PBS, lysed with RIPA

buffer (see above section on immunoblotting),

and centrifuged at 17,000gfor 10 min at 4°C.

The protein concentration of the resulting

supernatant was determined using the Bradford

assay. Then, 150- to 200-mg samples were saved

as whole cell lysate (WCL) control, while 1000

or 1500 mg of lysates were incubated with 40 ml

of Streptavidin Magnetic Beads (Thermo Fisher,

88817) overnight, nutating at 4°C. Magnetic

beads were washed twice with RIPA buffer,

once with 1 M KCl, once with 0.1 M Na

once with 2 M Urea in 10 mM Tris-HCl, pH

8.0, two times with RIPA buffer, and finally

eluted with Laemmli sample buffer in the pres-

ence of 2 mM biotin. WCL and eluate samples

were analyzed by SDS-PAGE gels and immu-

noblotting. For quantification, the band inten-

sities of eluted endogenous MTM1 or EEA1

were normalized to the band intensity of self-

biotinylated Rab5A in Streptavidin-HRP blots

(Fig. 2B and fig. S4B).

ER isolation and biotinylated protein collection

for liquid chromatography combined with mass

spectrometry (LC-MS)

HeLa WT or MTM1 KO cell line stably ex-

pressing doxycycline-inducible 3xHA-APEX2-

Rab5A were seeded in 15-cm dishes with 17 ×

cells (WT = 6 dishes, KO = 3 dishes) in the

presence of doxycycline for 16 hours before

biotinylation. The next day, cells at ∼85% con-

fluencyweretreatedwith500mMphenol-biotin

and hydrogen peroxide to induce biotinylation,

quenched, washed once with isolation buffer

(225 mM mannitol, 75 mM sucrose, 30 mM

Tris-HCl pH 7.4, 0.1 mM EGTA), and, finally,

scraped into 1 ml of the same buffer resulting

in ∼2 ml total volume. Half a milliliter of this

sample was saved (WCL) as a control. The re-

maining 1.5 mlwas homogenized using a 2 ml

Dounce homogenizer (∼100 to 120 strokes) on

ice in the presence of protease and phosphatase

inhibitors. To isolate light membranes con-

taining the ER, the material was centrifuged

twice at 600gfor 2 min each. The pellet (F1)

including nuclei and cell debris was resus-

pended in 150 ml RIPA buffer. The remaining

supernatant was centrifuged twice at 7000g

for 20 min each. The resulting pellet (F2) in-

cluding the mitochondrial fraction was resus-

pended in 150 ml RIPA buffer and further

fractionated by centrifugation at 20,000gfor

60 min to pellet light ER membranes (F3). This

ER-enriched membrane fraction was resus-

pended in150 ml RIPA buffer. Protein concen-

tration was determined using the Bradford

assay. To isolate biotinylated proteins, 130 mg

ofWCLorF3fractionwasrotatedinthepres-

ence of Streptavidin Magnetic Beads. Eluted

biotinylated proteins were reduced (5 mM

dithiothreitol, 30 min at 55°C), alkylated (15 mM

iodacetamide, 20 min at RT in the dark), and

submitted to LC-MS analysis (samples from

two independent experiments).

LC-MS analysis of affinity-purified samples

Proteins were loaded on SDS-PAGE and sub-

jected to in-gel digestion. In brief, gel bands

were excised, and protein digestion was carried

out using trypsin at an enzyme-to-protein ratio

of 1:100 (w/w) at 37°C overnight. LC-MS mea-

surement was achieved by reverse phase high-

performance liquid chromatography (RP-HPLC)

on a Thermo Scientific Dionex UltiMate 3000

system connected to a PepMap C-18 trap-

column [0.075 mm by 50 mm, 3-mm particle

size, 100-Å pore size (Thermo Scientific)] and a

200 cm mPAC column was used (PharmaFluidics,

Ghent, Belgium) with 750 or 350 nl/min flow

rate with a 120-min gradient. Samples were

analyzed on an Orbitrap Fusion mass spectrom-

eter. MS1 scan were acquired in the Orbitrap

with a range of 375 to 1500 m/z, mass resolution

of 120,000, automatic gain control (AGC) target

value of 4 × 10

, and 50-ms maximum injection

time. MS2 scans were acquired in the ion trap

with an AGC target value of 1 × 10

and 35-ms

maximum injection time. Precursor ions with

charge states 2 to 4 were isolated with an iso-

lation window of 1.6 m/z and 40 s dynamic ex-

clusion. Precursor ions were fragmented using

higher-energy collisional dissociation with 30%

normalized collision energy. Analysis of the raw

data was done with MaxQuant (MQ) software

version 1.6.2.6. MaxQuant standard settings

were kept as default. In the search parameters,

two missed cleavage sites were included, the

fixed modification was set to cysteine carba-

midomethyl modification, and variable modifi-

cations to methionine oxidation and N-terminal

protein acetylation. The peptide mass tolerance

was set to 4.5 parts per million (ppm) for MS1

scans and 20 ppm for MS2 scans. Match be-

tween runs option was enabled. The database

search was done using Andromeda against

the Human UniProt/Swiss-Prot database with

common contaminants. The false discovery

rate (FDR) was set to 1% for both peptide and

protein level. Protein quantification was done

on the basis of at least two razor and unique

peptides. Label-free quantification and iBAQ

calculation were enabled. Statistical analysis

was done on the “ProteinGroups”table with

Perseus version 1.6.7.0. Proteomics data have

been deposited to the ProteomeXchange Con-

sortium via PRIDE and are available via

ProteomeXchange with identifier PXD033846.

Quantitative whole-cell proteomics with

TMT labeling

Fed or starved (2 hours) WT or MTM1 KO HeLa

cells (one 10-cm dish at 90% confluency) col-

lected from three independent experiments

were lysed in 300 ml 8 M urea-lysis-buffer in

50 mM triethylammonium bicarbonate (TEAB)–

containing protease inhibitor cocktail (Roche)

and 0.5 ml Benzonase Nuclease HC (Millipore).

Samples were applied to a Bioruptor Pico

(Diagenode) for 10 cycles (30 s on/30 s off) at

4°C.Sampleswerethenincubatedfor30min

at 25°C. Proteins were reduced with 5 mM Tris

(2-carboxyethyl)phosphin-hydrochloride (TCEP)

and alkylated with 40 mM chloroacetamide

(CAA) for 60 min at 37°C in the dark. Protein

digestion was carried out using Lys C at an

enzyme-to-protein ratio of 1:100 (w/w) at 37°C

for 3 hours. After diluting to 2 M urea with

50 mM TEAB buffer, the digestion was con-

tinued with trypsin at an enzyme-to-protein

ratio of 1:100 (w/w) at 37°C and overnight.

Digestion was stopped by adding formic acid

to a final concentration of 1%. Samples were

desalted with C18 Sep-Pak cartridge (Waters)

and quantified with Pierce colorimetric pep-

tide assay (Thermo Fisher Scientific). Peptides

were dried under speed vacuum and stored

at −20°C.

TMT labeling

Peptides were reconstituted in 50 mM TEAB

buffer to a concentration of 2.1 mg/ml. TMT

10-plex reagent (Thermo Fisher Scientific)

was dissolved in 20 ml 100% acetonitrile to

reach 0.2 mg. For each TMT channel, peptides

(100 mg) were labeled with 0.2 mg TMT 10-plex

reagent (two multiplex—10plex for fed and

10plex for starved). For the internal standard

(IS), peptides from all samples were mixed to

reach 100 mg total peptides and labeled with

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 13 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

the TMT 131 channel to be able to compare

between the two plexes. The labeling reac-

tion was carried out for 60 min at RT and

quenched with 55 mM Tris pH 8.0 for 15 min

at RT. All TMT 10-plex labeled samples were

mixed and desalted with C18 Sep-Pak car-

tridge (Waters). The eluted peptides from the

Sep-Pak were dried under speed vacuum and

stored at −20°C.

High-pH prefractionation

The mixed TMT labeled peptides were re-

constituted in 10 mM NH

OH buffer with

1% acetonitrile to reach 200 mg. The Peptides

were fractionated by High-pH chromatogra-

phy using a Gemini column (3 mm, C18, 110 Å,

Phenomenex) on an Agilent 1260 Infinity II

system. An 85-min gradient was applied, and

72 fractions were collected and pooled into

12 fractions. Fractions were dried under speed

vacuum until analysis by LC-MS.

LC-MS analysis

Separation of the labeled TMT samples was

achieved by RP-HPLC on a Thermo Scientific

Dionex UltiMate 3000 system, as described

above. Samples were analyzed on an Orbitrap

Fusion Lumos mass spectrometer with FAIMS

Pro device (Thermo Scientific). MS1 and MS2

scans were acquired in the Orbitrap with a

mass resolution of 120,000 and 50,000, respec-

tively, MS1 scan range was set to between 400

and 1600 m/z, standard AGC target, and maxi-

mum injection time was set to auto. Precursor

ions with charge states 2 to 6 were isolated with

an isolation window of 0.7 m/z and dynamic

exclusion of 60 s. MS2 scans were set to custom

AGC target with normalized AGC target of

250%, and maximum injection time was set

to auto. Precursor ions were fragmented using

higher-energy collisional dissociation with 38%

normalized collision energy. Cycle time was set

to 2 s. An internal stepping of CVs −50, −65,

and −85 was used in all runs. Data acquisition

was done with Xcalibur software 4.4 and Instru-

ment Control Software version 3.4. Data analy-

sis for the TMT labeled samples was done in

Proteome Discoverer version 2.5. The TMT

10-plex was set as the quantification method,

and the 131 mass was set as the control chan-

nel. For the Sequest HT search, the following

parameters were applied: MS1 ion mass toler-

ance of 10 ppm and a MS2 mass tolerance of

0.02 Da. Tryptic digestion allowing two missed

cleavages, minimum peptide length of 6 amino

acids and maximum peptide length of 144

amino acids. The following modifications

were included: cysteine carbamidomethylation

(+57.021 Da) as static modification, methionine

oxidation (+15.995 Da) and N-terminal acet-

ylation (+42.011 Da) were set as dynamic modi-

fications. In addition, TMT 6-plex (229.163 Da)

was set as static modification for peptide

N-terminal and for lysine residue. Strict FDR

was set to 0.01, and relaxed FDR was set to

0.05. The search was performed against the

Human UniProt/Swiss-Prot database. Unique

and razor peptides were used for quantifica-

tion, co-isolation threshold was setto 50, and

average reporter S/N to 10. Data were normal-

ized against total peptide amount, and scaling

was done against the control channel average.

The result “Proteins”output table was exported,

and the statistical analysis was done in Perseus

version 1.6.15.0. Then, Gene Ontology analysis

was conducted using Metascape web tool (77)

(https://metascape.org/gp/index.html#/main/

step1).Theproteomicsdatahavebeendepos-

ited to the ProteomeXchange Consortium via

PRIDE and are available via ProteomeXchange

with identifier PXD033850.

Liposome co-sedimentation assays

Powdered lipids were individually resuspended

in chloroform then mixed together in a glass

sample vial and slowly evaporated with a dry

stream. Liposome composition was 60%

phosphatidylcholine (Avanti #850375P), 19.8%

phosphatidylethanolamine (Avanti #850725P),

0.2% rhodamine-phosphatidylethanolamine

(Avanti #810150P), 10% cholesterol (Avanti

#700000), and 10% phosphatidylinositol-3-

sphosphate (Avanti #850150P) or 10% phos-

phatidylinositol 4-phosphate (Avanti #850151P)

or 10% phosphatidylinositol 3,4-biphosphate

(Avanti #850153P). The mixture was resus-

pended in 100 ml HEPES-buffered salt solution

(20 mM Hepes-NaOH pH 7.5, 150 mM NaCl)

for 40 min at 37°C, occasionally vigorously

vortexed, and sonicated in a 37°C water bath

five times, 45 s on, 1 min off. Liposomes (25 ml)

were gently mixed with 25 ml(3mgsuspended

with the same buffer) recombinant Escherichia

coli BL21 expressed GST-RRBP1 1-150aa WT

and incubated for 45 min at 25°C. Liposomes

were reisolated by centrifugation at 70,000g

for 15min at 25°C. Supernatant and pellet frac-

tions were dissolved in Laemmli sample buffer

andanalyzedbySDS-PAGE.

Seahorse XFe96 analyzer

The Seahorse XFe96 sensor cartridge was hy-

drated, and 1 × 10

cells were seeded on 96-well

plates one day before the assay. The next day,

while cells were fed or starved for 100 min at

37°C incubator without CO

,themitochon-

drial stress kit (Agilent, 103015-100) compound

was prepared in complete DMEM or EBSS and

loaded on the cartridge (OligomycinA1: 2 mM;

FCCP 1.7 mM, Rot/AA 1 mM). Assay media did

not include pyruvate. After instrument cali-

brations, cells were transferred to the XFe96

analyzer to record oxygen consumption rate at

37°C. All values were background-subtracted

(i.e., control well without cells). Finally, cells

were lysed with RIPA buffer and protein con-

tents was measured using the BCA assay for

data normalization.

Luminescent ATP assay

Cells (5 × 10

to 8 × 10

) were seeded in the

black 96-well plate. On the next day, total

cellular ATP levels of fed or starved cells were

analyzed using ATPlite Luminescence Assay

System (PerkinElmer, 6016943) according to

the manufacturer’s protocol. Luminescence

was measured by TECAN Luminescence plate

reader, and the value was normalized by pro-

tein concentration measured via Bradford or

BCA assay.

Software

Cartoons and schematics were generated using

BioRender and Adobe illustrator.

Statistics and reproducibility

Statistical analysis and graphing were carried out

using Prism 8. Normality testing (D’Agostino-

Pearson) was conducted to determine whether

to use parametric or nonparametric statistical

tests. To compare two datasets, normally dis-

tributed data were analyzed by two-tailed un-

paired Student’sttest, whereas non-normally

distributed data were analyzed by two-tailed

Mann-Whitney test. For the comparison of

more than two normally distributed datasets,

we used ordinary one-way analysis of variance

(ANOVA) with either Dunnett’smultiplecom-

parisons (to compare the mean of each column

with a control column) or Tukey’smultiple

comparisons test (to compare the mean of each

column with every other column). To com-

pare more than two non-normally distributed

datasets, we used the Kruskal-Wallis test with

two-sided Dunn’s multiple comparison test. All

statistical analyses were performed on samples

drawn from at least two or three independent

experiments.

REFERENCES AND NOTES

1. P. Bonaldo, M. Sandri, Cellular and molecular mechanisms of

muscle atrophy. Dis. Model. Mech. 6,25–39 (2013).

doi: 10.1242/dmm.010389; pmid: 23268536

2. B. H. Goodpaster, L. M. Sparks, Metabolic flexibility in health

and disease. Cell Metab. 25, 1027–1036 (2017). doi: 10.1016/

j.cmet.2017.04.015; pmid: 28467922

3. M. Sandri et al., Foxo transcription factors induce the atrophy-

related ubiquitin ligase atrogin-1 and cause skeletal muscle

atrophy. Cell 117, 399–412 (2004). doi: 10.1016/S0092-8674

(04)00400-3; pmid: 15109499

4. S. Schiaffino, K. A. Dyar, S. Ciciliot, B. Blaauw, M. Sandri,

Mechanisms regulating skeletal muscle growth and atrophy.

FEBS J. 280, 4294–4314 (2013). doi: 10.1111/febs.12253;

pmid: 23517348

5. H. An, A. Ordureau, M. Körner, J. A. Paulo, J. W. Harper,

Systematic quantitative analysis of ribosome inventory during

nutrient stress. Nature 583, 303–309 (2020). doi: 10.1038/

s41586-020-2446-y; pmid: 32612236

6. A. Efeyan, W. C. Comb, D. M. Sabatini, Nutrient-sensing

mechanisms and pathways. Nature 517, 302–310 (2015).

doi: 10.1038/nature14190; pmid: 25592535

7. H. An et al., TEX264 is an endoplasmic reticulum-resident

ATG8-interacting protein critical for ER remodeling during

nutrient stress. Mol. Cell 74, 891–908.e10 (2019). doi: 10.1016/

j.molcel.2019.03.034; pmid: 31006537

8. A. Ballabio, J. S. Bonifacino, Lysosomes as dynamic regulators

of cell and organismal homeostasis. Nat. Rev. Mol. Cell Biol.

21, 101–118 (2020). doi: 10.1038/s41580-019-0185-4;

pmid: 31768005

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 14 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

9. S. M. Houten, S. Violante, F. V. Ventura, R. J. Wanders, The

biochemistry and physiology of mitochondrial fatty acid

b-oxidation and its genetic disorders. Annu. Rev. Physiol. 78,

23–44 (2016). doi: 10.1146/annurev-physiol-021115-105045;

pmid: 26474213

10. T. Balla, Phosphoinositides: Tiny lipids with giant impact on cell

regulation. Physiol. Rev. 93, 1019–1137 (2013). doi: 10.1152/

physrev.00028.2012; pmid: 23899561

11. Y. Posor, W. Jang, V. Haucke, Phosphoinositides as membrane

organizers. Nat. Rev. Mol. Cell Biol. 23, 797–816 (2022).

doi: 10.1038/s41580-022-00490-x; pmid: 35589852

12. C. C. Campa et al., Rab11 activity and PtdIns(3)P turnover

removes recycling cargo from endosomes. Nat. Chem. Biol. 14,

801–810 (2018). doi: 10.1038/s41589-018-0086-4;

pmid: 29915378

13. C. Cao, J. M. Backer, J. Laporte, E. J. Bedrick,

A. Wandinger-Ness, Sequential actions of myotubularin lipid

phosphatases regulate endosomal PI(3)P and growth factor

receptor trafficking. Mol. Biol. Cell 19, 3334–3346 (2008).

doi: 10.1091/mbc.e08-04-0367; pmid: 18524850

14. C. Cao, J. Laporte, J. M. Backer, A. Wandinger-Ness, M. P. Stein,

Myotubularin lipid phosphatase binds the hVPS15/hVPS34

lipid kinase complex on endosomes. Traffic 8, 1052–1067

(2007). doi: 10.1111/j.1600-0854.2007.00586.x;

pmid: 17651088

15. K. Ketel et al., A phosphoinositide conversion mechanism

for exit from endosomes. Nature 529, 408–412 (2016).

doi: 10.1038/nature16516; pmid: 26760201

16. L. Al-Qusairi et al., T-tubule disorganization and defective

excitation-contraction coupling in muscle fibers lacking

myotubularin lipid phosphatase. Proc. Natl. Acad. Sci. U.S.A.

106, 18763–18768 (2009). doi: 10.1073/pnas.0900705106;

pmid: 19846786

17. L. Amoasii et al., Myotubularin and PtdIns3Premodel the

sarcoplasmic reticulum in muscle in vivo.J. Cell Sci. 126,

1806–1819 (2013). doi: 10.1242/jcs.118505; pmid: 23444364

18. Y. Shibata, G. K. Voeltz, T. A. Rapoport, Rough sheets and

smooth tubules. Cell 126, 435–439 (2006). doi: 10.1016/

j.cell.2006.07.019; pmid: 16901774

19. J. Laporte et al., Mutations in the MTM1 gene implicated in

X-linked myotubular myopathy. Hum. Mol. Genet. 6, 1505–1511

(1997). doi: 10.1093/hmg/6.9.1505; pmid: 9305655

20. C. R. Piers on et al., Modeling the human MTM1 p.R69C

mutation in murine Mtm1 results in exon 4 skipping and a

less severe myotubular myopathy phenotype. Hum. Mol.

Genet. 21,811–825 (2012). doi: 10.1093/hmg/ddr512;

pmid: 22068590

21. O. M. Do rchies et al., Normal innervation and differentiation

of X-linked myotubular myopathy muscle cells in a

nerve-muscle coculture system. Neuromuscul. Disord. 11,

736–746 (2001). doi: 10.1016/S0960-8966(01)00221-8;

pmid: 11595516

22. L. K. Schroeder et al., Dynamic nanoscale morphology of the

ER surveyed by STED microscopy. J. Cell Biol. 218,83–96

(2019). doi: 10.1083/jcb.201809107; pmid: 30442642

23. G. E. Palade, Studies on the endoplasmic reticulum. II.

Simple dispositions in cells in situ. J. Biophys. Biochem.

Cytol. 1, 567–582 (1955). doi: 10.1083/jcb.1.6.567;

pmid: 13278367

24. G. K. Voeltz, W. A. Prinz, Y. Shibata, J. M. Rist, T. A. Rapoport,

A class of membrane proteins shaping the tubular endoplasmic

reticulum. Cell 124, 573–586 (2006). doi: 10.1016/

j.cell.2005.11.047; pmid: 16469703

25. J. Nixon-Abell et al., Increased spatiotemporal resolution

reveals highly dynamic dense tubular matrices in the

peripheral ER. Science 354, aaf3928 (2016). doi: 10.1126/

science.aaf3928; pmid: 27789813

26. M. A. Raess, S. Friant, B. S. Cowling, J. Laporte, WANTED -

Dead or alive: Myotubularins, a large disease-associated

protein family. Adv. Biol. Regul. 63,49–58 (2017). doi: 10.1016/

j.jbior.2016.09.001; pmid: 27666502

27. E. L. Axe et al., Autophagosome formation from membrane

compartments enriched in phosphatidylinositol 3-phosphate

and dynamically connected to the endoplasmic reticulum.

J. Cell Biol. 182, 685–701 (2008). doi: 10.1083/jcb.200803137;

pmid: 18725538

28. Z. Hong et al., PtdIns3P controls mTORC1 signaling through

lysosomal positioning. J. Cell Biol. 216, 4217–4233 (2017).

doi: 10.1083/jcb.201611073; pmid: 29030394

29. M. J. Munson et al., mTOR activates the VPS34-UVRAG

complex to regulate autolysosomal tubulation and cell survival.

EMBO J. 34, 2272–2290 (2015). doi: 10.15252/

embj.201590992; pmid: 26139536

30. J. Laporte et al., A gene mutated in X-linked myotubular

myopathy defines a new putative tyrosine phosphatase family

conserved in yeast. Nat. Genet. 13, 175–182 (1996).

doi: 10.1038/ng0696-175; pmid: 8640223

31. M. W. Lawlor, J. J. Dowling, X-linked myotubular myopathy.

Neuromuscul. Disord. 31, 1004–1012 (2021). doi: 10.1016/

j.nmd.2021.08.003; pmid: 34736623

32. A. Khaminets et al., Regulation of endoplasmic reticulum

turnover by selective autophagy. Nature 522, 354–358 (2015).

doi: 10.1038/nature14498; pmid: 26040720

33. C. Settembre et al., TFEB links autophagy to lysosomal

biogenesis. Science 332, 1429–1433 (2011). doi: 10.1126/

science.1204592; pmid: 21617040

34. K. M. Fetalvero et al., Defective autophagy and mTORC1

signaling in myotubularin null mice. Mol. Cell. Biol. 33,98–110

(2013). doi: 10.1128/MCB.01075-12; pmid: 23109424

35. A. L. Marat, V. Haucke, Phosphatidylinositol 3-phosphates-at

the interface between cell signalling and membrane traffic.

EMBO J. 35, 561–579 (2016). doi: 10.15252/embj.201593564;

pmid: 26888746

36. Y. Wu et al., Contacts between the endoplasmic reticulum

and other membranes in neurons. Proc. Natl. Acad. Sci. U.S.A.

114, E4859–E4867 (2017). doi: 10.1073/pnas.1701078114;

pmid: 28559323

37. H. Wu, P. Carvalho, G. K. Voeltz, Here, there, and everywhere:

The importance of ER membrane contact sites. Science

361, eaan5835 (2018). doi: 10.1126/science.aan5835;

pmid: 30072511

38. S. S. Lam et al., Directed evolution of APEX2 for electron

microscopy and proximity labeling. Nat. Methods 12,51–54

(2015). doi: 10.1038/nmeth.3179; pmid: 25419960

39. A. Fegan, B. White, J. C. Carlson, C. R. Wagner, Chemically

controlled protein assembly: Techniques and applications.

Chem. Rev. 110, 3315–3336 (2010). doi: 10.1021/cr8002888;

pmid: 20353181

40. T. Calì, M. Brini, Quantification of organelle contact sites by

split-GFP-based contact site sensors (SPLICS) in living cells.

Nat. Protoc. 16, 5287–5308 (2021). doi: 10.1038/

s41596-021-00614-1; pmid: 34686857

41. C. Raib org et al., Repeated ER-endosome cont acts promote

endosome translocation and neurite outgrowth. Nature

520,234–238 (2015). doi: 10.1038/nature14359;

pmid: 25855459

42. A. M. Valm et al., Applying systems-level spectral imaging

and analysis to reveal the organelle interactome. Nature

546, 162–167 (2017). doi: 10.1038/nature22369;

pmid: 28538724

43. A. A. Rowland, P. J. Chitwood, M. J. Phillips, G. K. Voeltz, ER

contact sites define the position and timing of endosome

fission. Cell 159, 1027–1041 (2014). doi: 10.1016/

j.cell.2014.10.023; pmid: 25416943

44. M. Lu et al., The structure and global distribution of the

endoplasmic reticulum network are actively regulated

by lysosomes. Sci. Adv. 6, eabc7209 (2020). doi: 10.1126/

sciadv.abc7209; pmid: 33328230

45. V. Hung et al., Proteomic mapping of cytosol-facing outer

mitochondrial and ER membranes in living human cells by

proximity biotinylation. eLife 6, e24463 (2017). doi: 10.7554/

eLife.24463; pmid: 28441135

46. Y. Shibata et al., Mechanisms determining the morphology of

the peripheral ER. Cell 143, 774–788 (2010). doi: 10.1016/

j.cell.2010.11.007; pmid: 21111237

47. P. Zheng et al., ER proteins decipher the tubulin code to

regulate organelle distribution. Nature 601, 132–138 (2022).

doi: 10.1038/s41586-021-04204-9; pmid: 34912111

48. C. S. Janot a et al., Shielding of actin by the endoplasmic

reticulum impacts nuclear positioning. Nat. Commun. 13,

2763 (2022). doi: 10.1038/ s41467-02 2-3038 8-3;

pmid: 35589708

49. B. T. Lobi ngier et al., An approach to spatiotemporally

resolve protein interaction networks in living cells. Cell 169,

350–360.e12 (2017). doi: 10.1016/j.cell.2017.03.022;

pmid: 28388416

50. Y. C. Wong, D. Ysselstein, D. Krainc, Mitochondria–lysosome

contacts regulate mitochondrial fission via RAB7 GTP

hydrolysis. Nature 554, 382–386 (2018). doi: 10.1038/

nature25486; pmid: 29364868

51. S. Nagashima et al., Golgi-derived PI(4)P-containing vesicles

drive late steps of mitochondrial division. Science 367,

1366–1371 (2020). doi: 10.1126/science.aax6089;

pmid: 32193326

52. S. C. Lewis, L. F. Uchiyama, J. Nunnari, ER-mitochondria

contacts couple mtDNA synthesis with mitochondrial division

in human cells. Science 353, aaf5549 (2016). doi: 10.1126/

science.aaf5549; pmid: 27418514

53. J. R. Friedman et al., ER tubules mark sites of mitochondrial

division. Science 334, 358–362 (2011). doi: 10.1126/

science.1207385; pmid: 21885730

54. T. C. Walther, J. Chung, R. V. Farese Jr., Lipid droplet

biogenesis. Annu. Rev. Cell Dev. Biol. 33, 491–510 (2017).

doi: 10.1146/annurev-cellbio-100616-060608; pmid: 28793795

55. A. Santinho et al., Membrane curvature catalyzes lipid droplet

assembly. Curr. Biol. 30, 2481–2494.e6 (2020). doi: 10.1016/

j.cub.2020.04.066; pmid: 32442467

56. A. S. Rambold, B. Kostelecky, N. Elia, J. Lippincott-Schwartz,

Tubular network formation protects mitochondria from

autophagosomal degradation during nutrient starvation.

Proc. Natl. Acad. Sci. U.S.A. 108, 10190–10195 (2011).

doi: 10.1073/pnas.1107402108; pmid: 21646527

57. A. S. Rambold, S. Cohen, J. Lippincott-Schwartz, Fatty acid

trafficking in starved cells: Regulation by lipid droplet lipolysis,

autophagy, and mitochondrial fusion dynamics. Dev. Cell 32,

678–692 (2015). doi: 10.1016/j.devcel.2015.01.029;

pmid: 25752962

58. M. Giacomello, A. Pyakurel, C. Glytsou, L. Scorrano, The cell

biology of mitochondrial membrane dynamics. Nat. Rev. Mol.

Cell Biol. 21,204–224 (2020). doi: 10.1038/s41580-020-0210-7;

pmid: 32071438

59. T. W ai et al., Imbalanced OPA1 processing and mitochondrial

fragmentation cause heart failure in mice. Science 350,

aad0116 (2015). doi: 10.1126/science.aad0116;

pmid: 26785494

60. A. R. English, G. K. Voeltz, Rab10 GTPase regulates ER

dynamics and morphology. Nat. Cell Biol. 15, 169–178 (2013).

doi: 10.1038/ncb2647; pmid: 23263280

61. J. E. Lee, P. I. Cathey, H. Wu, R. Parker, G. K. Voeltz,

Endoplasmic reticulum contact sites regulate the dynamics of

membraneless organelles. Science 367, eaay7108 (2020).

doi: 10.1126/science.aay7108; pmid: 32001628

62. L. C. Gomes, G. Di Benedetto, L. Scorrano, During autophagy

mitochondria elongate, are spared from degradation and

sustain cell viability. Nat. Cell Biol. 13, 589–598 (2011).

doi: 10.1038/ncb2220; pmid: 21478857

63. W. M. Henne, M. L. Reese, J. M. Goodman, The assembly of

lipid droplets and their roles in challenged cells. EMBO J. 37,

e98947 (2018). doi: 10.15252/embj.201898947;

pmid: 29789390

64. M. A. Roberts, A. Segura-Roman, J. A. Olzmann, Organelle

biogenesis: ER shape influences lipid droplet nucleation.

Curr. Biol. 30, R770–R773 (2020). doi: 10.1016/

j.cub.2020.05.027; pmid: 32634419

65. J. R. Liang et al., A genome-wide ER-phagy screen highlights

key roles of mitochondrial metabolism and ER-resident

UFMylation. Cell 180, 1160–1177.e20 (2020). doi: 10.1016/

j.cell.2020.02.017; pmid: 32160526

66. Y. Pan et al., Endoplasmic reticulum ribosome-binding protein

1, RRBP1, promotes progression of colorectal cancer and

predicts an unfavourable prognosis. Br. J. Cancer 113, 763–772

(2015). doi: 10.1038/bjc.2015.260; pmid: 26196185

67. A. H. Bahrami, G. Hummer, Formation and stability of lipid

membrane nanotubes. ACS Nano 11, 9558–9565 (2017).

doi: 10.1021/acsnano.7b05542; pmid: 28873296

68. T. Shemesh et al., A model for the generation and interconversion

of ER morphologies. Proc. Natl. Acad. Sci. U.S.A. 111,

E5243–E5251 (2014). doi: 10.1073/pnas.1419997111;

pmid: 25404289

69. G. Parlakgül et al., Regulation of liver subcellular architecture

controls metabolic homeostasis. Nature 603, 736–742 (2022).

doi: 10.1038/s41586-022-04488-5; pmid: 35264794

70. D. S. Hirsch et al., Insulin activation of vacuolar protein sorting

34 mediates localized phosphatidylinositol 3-phosphate

production at lamellipodia and activation of mTOR/S6K1.

Cell. Signal. 26, 1258–1268 (2014). doi: 10.1016/

j.cellsig.2014.02.009; pmid: 24582588

71. V. I. Korolchuk et al., Lysosomal positioning coordinates

cellular nutrient responses. Nat. Cell Biol. 13, 453–460 (2011).

doi: 10.1038/ncb2204; pmid: 21394080

72. A. L. Marat et al., mTORC1 activity repression by late

endosomal phosphatidylinositol 3,4-bisphosphate. Science

356, 968–972 (2017). doi: 10.1126/science.aaf8310;

pmid: 28572395

73. M. J. Hoyer et al., A novel class of ER membrane proteins

regulates ER-associated endosome fission. Cell 175, 254–265.

e14 (2018). doi: 10.1016/j.cell.2018.08.030; pmid: 30220460

74. K. Mamchaoui et al., Immortalized pathological human

myoblasts: Towards a universal tool for the study of

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 15 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

neuromuscular disorders. Skelet. Muscle 1, 34 (2011).

doi: 10.1186/2044-5040-1-34; pmid: 22040608

75. F. Bottanelli et al., Two-colour live-cell nanoscale imaging of

intracellular targets. Nat. Commun. 7, 10778 (2016).

doi: 10.1038/ncomms10778; pmid: 26940217

76. C. McQuin et al., CellProfiler 3.0: Next-generation image

processing for biology. PLOS Biol. 16, e2005970 (2018).

doi: 10.1371/journal.pbio.2005970; pmid: 29969450

77. Y. Zho u et al., Metascape p rovides a biolog ist-oriented

resource for the an alysis of syste ms-level dataset s. Nat.

Commun. 10,1523(2019).doi:10.10 38/s41467-019-092 34-6;

pmid: 30944313

ACKNOWL EDGME NTS

We thank members of the Haucke lab for discussion and Y. Posor

for critical reading of the manuscript. We thank S. Zillmann,

D. Löwe, M. Mühlbauer, and C. Schmidt for expert technical

assistance and M. Lehmann, C. Schmied, and J. Eichhorst for aid

with microscopy and image analysis. We also thank G. G. Farías

and G. Voeltz for plasmids and protocols and the MyoLine platform

of the Institute of Myology in Paris for aid in the generation

of myoblast cell lines. Funding: This work was funded by a

Leibniz-German Academic Exchange Service (DAAD) Research

Fellowship (57423756) (W.J.), the Postdoctoral Fellowship

Program (Nurturing Next-generation Researchers) of the National

Research Foundation of Korea (NRF) (2018R1A6A3A03010583)

(W.J.), and Deutsche Forschungsgemeinschaft (TRR186/ A08)

(V.H.). Author contributions: Conceptualization: W.J. and

V.H. Investigation: all cell and molecular biology: W.J.; electron

microscopy: D.P. with W.J.; quantitative proteomics: W.J. with

M.N.-H. and F.L.; Seahorse analysis: W.J., Y.L., S.J.S., and U.K.;

XLCNM myoblast patient cells: K.M. and V.M.; CRISPR: W.J. and

P.S. Funding acquisition: W.J. and V.H. Project administration: V.H.

Supervision: V.H., F.L., U.K., and S.J.S. Writing –original draft:

W.J. and V.H. Writing –review & editing: all authors. Competing

interests: The authors declare no competing financial interests.

Data and materials availability: All data are available in the main

text or the supplementary materials. Proteomics data have been

deposited to the ProteomeXchange Consortium via PRIDE and

are available via ProteomeXchange with identifier PXD033846.

Materials and reagents are available from the corresponding author

authors, some rights reserved; exclusive licensee American

Association for the Advancement of Science. No claim to original

US government works. https://www.science.org/about/science-

licenses-journal-article-reuse

SUPPLEMENTARY MATERIALS

science.org/doi/10.1126/science.abq5209

Figs. S1 to S11

Tables S1 to S4

MDAR Reproducibility Checklist

Movies S1 to S5

View/request a protocol for this paper from Bio-protocol.

Submitted 13 April 2022; resubmitted 23 September 2022

Accepted 25 October 2022

10.1126/science.abq5209

Jang et al., Science 378, eabq5209 (2022) 16 December 2022 16 of 16

RESEARCH |RESEARCH ARTICLE

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

Use of this article is subject to the Terms of service

Science (ISSN ) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW, Washington, DC

20005. The title Science is a registered trademark of AAAS.

to original U.S. Government Works

Endosomal lipid signaling reshapes the endoplasmic reticulum to control

mitochondrial function

Wonyul JangDmytro PuchkovPaula SamsóYongTian LiangMichal Nadler-HollyStephan J. SigristUlrich KintscherFan

LiuKamel MamchaouiVincent MoulyVolker Haucke

Science, 378 (6625), eabq5209. • DOI: 10.1126/science.abq5209

A lipid-triggered signal in starvation

Nutrient starvation triggers changes in metabolism that are coordinated across the cell and its organelles. Jang et

al. studied how endosomal signaling lipid turnover through MTM1, a phosphoinositide 3-phosphatase mutated in X-

linked centronuclear myopathy in humans, reshapes the endoplasmic reticulum to control mitochondrial morphology

and oxidative metabolism (see the Perspective by Zanellati and Cohen). A lipid-controlled organellar relay transmits

nutrient-triggered changes in endosomal signaling lipid levels to mitochondria to enable metabolic rewiring. —SMH

View the article online

https://www.science.org/doi/10.1126/science.abq5209

Permissions

https://www.science.org/help/reprints-and-permissions

Downloaded from https://www.science.org at Leibniz-Institut für Molekulare Pharmakologie on December 16, 2022

AMPK targets PDZD8 to trigger carbon source shift from glucose to glutamine

Article

Full-text available

Jun 2024
CELL RES

The shift of carbon utilization from primarily glucose to other nutrients is a fundamental metabolic adaptation to cope with decreased blood glucose levels and the consequent decline in glucose oxidation. AMP-activated protein kinase (AMPK) plays crucial roles in this metabolic adaptation. However, the underlying mechanism is not fully understood. Here, we show that PDZ domain containing 8 (PDZD8), which we identify as a new substrate of AMPK activated in low glucose, is required for the low glucose-promoted glutaminolysis. AMPK phosphorylates PDZD8 at threonine 527 (T527) and promotes the interaction of PDZD8 with and activation of glutaminase 1 (GLS1), a rate-limiting enzyme of glutaminolysis. In vivo, the AMPK-PDZD8-GLS1 axis is required for the enhancement of glutaminolysis as tested in the skeletal muscle tissues, which occurs earlier than the increase in fatty acid utilization during fasting. The enhanced glutaminolysis is also observed in macrophages in low glucose or under acute lipopolysaccharide (LPS) treatment. Consistent with a requirement of heightened glutaminolysis, the PDZD8-T527A mutation dampens the secretion of pro-inflammatory cytokines in macrophages in mice treated with LPS. Together, we have revealed an AMPK-PDZD8-GLS1 axis that promotes glutaminolysis ahead of increased fatty acid utilization under glucose shortage.

A tripartite organelle platform links growth factor receptor signaling to mitochondrial metabolism

Article

Full-text available

Jun 2024

One open question in the biology of growth factor receptors is how a quantitative input (i.e., ligand concentration) is decoded by the cell to produce specific response(s). Here, we show that an EGFR endocytic mechanism, non-clathrin endocytosis (NCE), which is activated only at high ligand concentrations and targets receptor to degradation, requires a tripartite organelle platform involving the plasma membrane (PM), endoplasmic reticulum (ER) and mitochondria. At these contact sites, EGFR-dependent, ER-generated Ca²⁺ oscillations are sensed by mitochondria, leading to increased metabolism and ATP production. Locally released ATP is required for cortical actin remodeling and EGFR-NCE vesicle fission. The same biochemical circuitry is also needed for an effector function of EGFR, i.e., collective motility. The multiorganelle signaling platform herein described mediates direct communication between EGFR signaling and mitochondrial metabolism, and is predicted to have a broad impact on cell physiology as it is activated by another growth factor receptor, HGFR/MET.

Morphodynamical adaptation of the endolysosomal system to stress

Article

Full-text available

May 2024
FEBS J

In eukaryotes, the spatiotemporal control of endolysosomal organelles is central to the maintenance of homeostasis. By providing an interface between the cytoplasm and external environment, the endolysosomal system is placed at the forefront of the response to a wide range of stresses faced by cells. Endosomes are equipped with a dedicated set of membrane‐associated proteins that ensure endosomal functions as well as crosstalk with the secretory or the autophagy pathways. Morphodynamical processes operate through local spatialization of subdomains, enabling specific remodeling and membrane contact capabilities. Consequently, the plasticity of endolysosomal organelles can be considered a robust and flexible tool exploited by cells to cope with homeostatic deviations. In this review, we provide insights into how the cellular responses to various stresses (osmotic, UV, nutrient deprivation, or pathogen infections) rely on the adaptation of the endolysosomal system morphodynamics.

Insulin signaling accelerates the anterograde movement of Rab4 vesicles in axons through Klp98A/KIF16B recruitment via Vps34-PI3Kinase

Preprint

Full-text available

Apr 2024

Rab4 GTPase organizes endosomal sorting essential for maintaining the balance between recycling and degradative pathways. Rab4 localizes to many cargos whose transport in neurons is critical for regulating neurotransmission and neuronal health. Furthermore, elevated Rab4 levels in the CNS are associated with synaptic atrophy and neurodegeneration in Drosophila and humans, respectively. However, how the transport of Rab4-associated vesicles is regulated in neurons remains unknown. Using in vivo time-lapse imaging of Drosophila larvae, we show that activation of insulin signaling via Dilp2 and dInR increases the anterograde velocity, run length, and flux of Rab4 vesicles in the axons. Molecularly, we show that activation of neuronal insulin signaling further activates Vps34, elevates the levels of PI(3)P on Rab4-associated vesicles, recruits Klp98A (a PI(3)P-binding kinesin-3 motor) and activates their anterograde transport. Together, these observations delineate the role of insulin signaling in regulating axonal transport and synaptic homeostasis. Highlights Dilp2-mediated insulin signaling activates anterograde transport of Rab4 vesicles Vps34 regulates anterograde velocity, run length and flux of Rab4 vesicles Local PI(3)P signaling on Rab4 vesicles regulates their motility in the axons PI(3)P production upon acute insulin stimulation recruits Klp98A on Rab4 vesicles

Organelle synergy unleashed: Modulating mitochondrial-endoplasmic reticulum contacts with a self-assembled prodrug amplifies ferroptosis for innovative cancer therapy

Article

Jun 2024

The endoplasmic reticulum connects to the nucleus by constricted junctions that mature after mitosis

Article

Jun 2024
EMBO REP

Junctions between the endoplasmic reticulum (ER) and the outer membrane of the nuclear envelope (NE) physically connect both organelles. These ER–NE junctions are essential for supplying the NE with lipids and proteins synthesized in the ER. However, little is known about the structure of these ER–NE junctions. Here, we systematically study the ultrastructure of ER–NE junctions in cryo-fixed mammalian cells staged in anaphase, telophase, and interphase by correlating live cell imaging with three-dimensional electron microscopy. Our results show that ER–NE junctions in interphase cells have a pronounced hourglass shape with a constricted neck of 7–20 nm width. This morphology is significantly distinct from that of junctions within the ER network, and their morphology emerges as early as telophase. The highly constricted ER–NE junctions are seen in several mammalian cell types, but not in budding yeast. We speculate that the unique and highly constricted ER–NE junctions are regulated via novel mechanisms that contribute to ER-to-NE lipid and protein traffic in higher eukaryotes.

Role of lipids in interorganelle communication

Article

Jun 2024
TRENDS CELL BIOL

Movement of the endoplasmic reticulum is driven by multiple classes of vesicles marked by Rab-GTPases

Preprint

Full-text available

May 2024

Peripheral endoplasmic reticulum (ER) tubules move along microtubules to interact with various organelles through membrane contact sites (MCS). Traditionally, ER moves by either sliding along stable microtubules via molecular motors or attaching to the plus ends of dynamic microtubules through tip attachment complexes (TAC). A recently discovered third process, hitchhiking, involves motile vesicles pulling ER tubules along microtubules. Previous research showed that ER hitchhikes on Rab5- and Rab7-marked endosomes, but it is uncertain if other Rab-vesicles can do the same. In U2OS cells, we screened Rabs for their ability to cotransport with ER tubules and found that ER hitchhikes on post-Golgi vesicles marked by Rab6 (isoforms a and b). Rab6-ER hitchhiking occurs independently of ER-endolysosome contacts and TAC-mediated ER movement. Disrupting either Rab6 or the motility of Rab6-vesicles reduces overall ER movement. Conversely, relocating these vesicles to the cell periphery causes peripheral ER accumulation, indicating that Rab6-vesicle motility is crucial for a subset of ER movements. Proximal post-Golgi vesicles marked by TGN46 are involved in Rab6-ER hitchhiking, while other post-Golgi vesicles (Rabs 8/10/11/13/14) are not essential for ER movement. Our further analysis finds that ER to Golgi vesicles marked by Rab1 are also capable of driving a subset of ER movements. Taken together, our findings suggest that ER hitchhiking on Rab-vesicles is a significant mode of ER movement. SIGNIFICANCE STATEMENT Peripheral endoplasmic reticulum tubules move on microtubules by either attaching to motors (cargo adaptor-mediated), dynamic microtubule-plus ends (tip attachment complexes) or motile vesicles (hitchhiking) but the prevalence of each mode is not clear Post-Golgi vesicles marked by Rab6/TGN46 and ER to Golgi vesicles marked by Rab1 drive ER movements ER hitchhiking on multiple classes of vesicles (endolysosomal, post-Golgi and ER to Golgi) marked by Rabs plays a prominent role in ER movement

Interplay between myotubularins and Ca2+ homeostasis

Article

May 2024
BBA-MOL CELL RES

ER remodeling via lipid metabolism

Article

Feb 2024

Shielding of actin by the endoplasmic reticulum impacts nuclear positioning

Article

Full-text available

May 2022

Nuclear position is central to cell polarization, and its disruption is associated with various pathologies. The nucleus is moved away from the leading edge of migrating cells through its connection to moving dorsal actin cables, and the absence of connections to immobile ventral stress fibers. It is unclear how these asymmetric nucleo-cytoskeleton connections are established. Here, using an in vitro wound assay, we find that remodeling of endoplasmic reticulum (ER) impacts nuclear positioning through the formation of a barrier that shields immobile ventral stress fibers. The remodeling of ER and perinuclear ER accumulation is mediated by the ER shaping protein Climp-63. Furthermore, ectopic recruitment of the ER to stress fibers restores nuclear positioning in the absence of Climp-63. Our findings suggest that the ER mediates asymmetric nucleo-cytoskeleton connections to position the nucleus.

Regulation of liver subcellular architecture controls metabolic homeostasis

Article

Full-text available

Mar 2022
NATURE

Cells display complex intracellular organization by compartmentalization of metabolic processes into organelles, yet the resolution of these structures in the native tissue context and their functional consequences are not well understood. Here we resolved the three-dimensional structural organization of organelles in large (more than 2.8 × 105 µm3) volumes of intact liver tissue (15 partial or full hepatocytes per condition) at high resolution (8 nm isotropic pixel size) using enhanced focused ion beam scanning electron microscopy1,2 imaging followed by deep-learning-based automated image segmentation and 3D reconstruction. We also performed a comparative analysis of subcellular structures in liver tissue of lean and obese mice and found substantial alterations, particularly in hepatic endoplasmic reticulum (ER), which undergoes massive structural reorganization characterized by marked disorganization of stacks of ER sheets3 and predominance of ER tubules. Finally, we demonstrated the functional importance of these structural changes by monitoring the effects of experimental recovery of the subcellular organization on cellular and systemic metabolism. We conclude that the hepatic subcellular organization of the ER architecture are highly dynamic, integrated with the metabolic state and critical for adaptive homeostasis and tissue health. Detailed reconstruction using enhanced focused ion beam scanning electron microscopy imaging and deep-learning-based automated segmentation demonstrates that hepatocyte subcellular organelle architecture regulates metabolism.

ER proteins decipher the tubulin code to regulate organelle distribution

Article

Full-text available

Jan 2022
NATURE

Organelles move along differentially modified microtubules to establish and maintain their proper distributions and functions 1,2 . However, how cells interpret these post-translational microtubule modification codes to selectively regulate organelle positioning remains largely unknown. The endoplasmic reticulum (ER) is an interconnected network of diverse morphologies that extends promiscuously throughout the cytoplasm ³ , forming abundant contacts with other organelles ⁴ . Dysregulation of endoplasmic reticulum morphology is tightly linked to neurologic disorders and cancer 5,6 . Here we demonstrate that three membrane-bound endoplasmic reticulum proteins preferentially interact with different microtubule populations, with CLIMP63 binding centrosome microtubules, kinectin (KTN1) binding perinuclear polyglutamylated microtubules, and p180 binding glutamylated microtubules. Knockout of these proteins or manipulation of microtubule populations and glutamylation status results in marked changes in endoplasmic reticulum positioning, leading to similar redistributions of other organelles. During nutrient starvation, cells modulate CLIMP63 protein levels and p180–microtubule binding to bidirectionally move endoplasmic reticulum and lysosomes for proper autophagic responses.

X-linked myotubular myopathy

Article

Full-text available

Oct 2021
NEUROMUSCULAR DISORD

X-linked myotubular myopathy (XLMTM) is a severe congenital muscle disease caused by mutation in the MTM1 gene. MTM1 encodes myotubularin (MTM1), an endosomal phosphatase that acts to dephosphorylate key second messenger lipids PI3P and PI3,5P2. XLMTM is clinically characterized by profound muscle weakness and associated with multiple disabilities (including ventilator and wheelchair dependence) and early death in most affected individuals. The disease is classically defined by characteristic changes observed on muscle biopsy, including centrally located nuclei, myofiber hypotrophy, and organelle disorganization. In this review, we highlight the clinical and pathologic features of the disease, present concepts related to disease pathomechanisms, and present recent advances in therapy development.

The structure and global distribution of the endoplasmic reticulum network are actively regulated by lysosomes

Article

Full-text available

Dec 2020

The endoplasmic reticulum (ER) comprises morphologically and functionally distinct domains: sheets and interconnected tubules. These domains undergo dynamic reshaping in response to changes in the cellular environment. However, the mechanisms behind this rapid remodeling are largely unknown. Here, we report that ER remodeling is actively driven by lysosomes, following lysosome repositioning in response to changes in nutritional status: The anchorage of lysosomes to ER growth tips is critical for ER tubule elongation and connection. We validate this causal link via the chemo- and optogenetically driven repositioning of lysosomes, which leads to both a redistribution of the ER tubules and a change of its global morphology. Therefore, lysosomes sense metabolic change in the cell and regulate ER tubule distribution accordingly. Dysfunction in this mechanism during axonal extension may lead to axonal growth defects. Our results demonstrate a critical role of lysosome-regulated ER dynamics and reshaping in nutrient responses and neuronal development.

Systematic quantitative analysis of ribosome inventory during nutrient stress

Article

Full-text available

Jul 2020
NATURE

Mammalian cells reorganize their proteomes in response to nutrient stress through translational suppression and degradative mechanisms using the proteasome and autophagy systems1,2. Ribosomes are central targets of this response, as they are responsible for translation and subject to lysosomal turnover during nutrient stress3,4,5. The abundance of ribosomal (r)-proteins (around 6% of the proteome; 10⁷ copies per cell)6,7 and their high arginine and lysine content has led to the hypothesis that they are selectively used as a source of basic amino acids during nutrient stress through autophagy4,7. However, the relative contributions of translational and degradative mechanisms to the control of r-protein abundance during acute stress responses is poorly understood, as is the extent to which r-proteins are used to generate amino acids when specific building blocks are limited⁷. Here, we integrate quantitative global translatome and degradome proteomics⁸ with genetically encoded Ribo–Keima⁵ and Ribo–Halo reporters to interrogate r-protein homeostasis with and without active autophagy. In conditions of acute nutrient stress, cells strongly suppress the translation of r-proteins, but, notably, r-protein degradation occurs largely through non-autophagic pathways. Simultaneously, the decrease in r-protein abundance is compensated for by a reduced dilution of pre-existing ribosomes and a reduction in cell volume, thereby maintaining the density of ribosomes within single cells. Withdrawal of basic or hydrophobic amino acids induces translational repression without differential induction of ribophagy, indicating that ribophagy is not used to selectively produce basic amino acids during acute nutrient stress. We present a quantitative framework that describes the contributions of biosynthetic and degradative mechanisms to r-protein abundance and proteome remodelling in conditions of nutrient stress.

Phosphoinositides as membrane organizers

Article

May 2022

Phosphoinositides are signalling lipids derived from phosphatidylinositol, a ubiquitous phospholipid in the cytoplasmic leaflet of eukaryotic membranes. Initially discovered for their roles in cell signalling, phosphoinositides are now widely recognized as key integrators of membrane dynamics that broadly impact on all aspects of cell physiology and on disease. The past decade has witnessed a vast expansion of our knowledge of phosphoinositide biology. On the endocytic and exocytic routes, phosphoinositides direct the inward and outward flow of membrane as vesicular traffic is coupled to the conversion of phosphoinositides. Moreover, recent findings on the roles of phosphoinositides in autophagy and the endolysosomal system challenge our view of lysosome biology. The non-vesicular exchange of lipids, ions and metabolites at membrane contact sites in between organelles has also been found to depend on phosphoinositides. Here we review our current understanding of how phosphoinositides shape and direct membrane dynamics to impact on cell physiology, and provide an overview of emerging concepts in phosphoinositide regulation. Phosphoinositides are signalling, membrane lipids derived from phosphatidylinositol, whose intracellular distribution and interconversion via phosphoinositide kinases and phosphatases is tightly coupled to membrane dynamics. Accordingly, phosphoinositides are now recognized as key regulators of endocytic and exocytic traffic, the autolysosomal system, and membrane contact site organization and function.

Quantification of organelle contact sites by split-GFP-based contact site sensors (SPLICS) in living cells

Article

Oct 2021

Membrane contact sites between organelles are essential for maintaining cellular homeostasis, which requires the continuous exchange of signaling molecules, ions, nutrients and lipids. Alterations of different contact sites are associated with a wide spectrum of human diseases. However, visualizing and quantifying these contact sites remains a challenge. This protocol describes the use of split-GFP-based contact site sensors (SPLICS) in microscopy applications for mapping organelle contact sites both in fixed and living cells. SPLICS sensors are engineered to express equimolar amounts of the organelle-targeted nonfluorescent β11 and GFP1-10 portions of the split-GFP protein in a single vector, and are capable of reconstituting fluorescence when two opposing membranes come into proximity. Reconstitution will occur only over the cell volume at defined contact sites resulting in a bright signal that can be detected easily and quantified automatically with specific custom-made plugins. The use of minimal targeting sequences facilitates targeting specificity and membrane coverage, avoiding artifacts due to full-length fusion protein overexpression and, thus, possible perturbations of the cell’s physiology. SPLICS sensors engineered to simultaneously detect multiple contact sites within the same cell have been generated by exploiting the ability of the β11 GFP fragment to reconstitute different color-shifted variants of the GFP1-10 fragment. Here, we describe a detailed protocol to set up SPLICS expression in living cells (2–3 d), detection and acquisition (1 d), and automated quantification with custom plugins (1–2 d). We also advise on construct design and characterization for novel organelle contacts. Split-GFP-based contact site sensors (SPLICS) are capable of reconstituting fluorescence when two opposing membranes come into close proximity. SPLICS can be used to quantify membrane contact sites between organelles in both fixed and living cells.

Organelle Biogenesis: ER Shape Influences Lipid Droplet Nucleation

Article

Jul 2020
CURR BIOL

Lipid droplets (LDs) are neutral lipid storage organelles assembled at the endoplasmic reticulum (ER). A new study reveals that the high membrane curvature of ER tubules catalyzes the nucleation of a neutral lipid lens, an early step in LD biogenesis.

Membrane Curvature Catalyzes Lipid Droplet Assembly

Article

May 2020
CURR BIOL

Lipid droplet (LD) biogenesis begins in the endoplasmic reticulum (ER) bilayer, but how the ER topology impacts this process is unclear. An early step in LD formation is nucleation, wherein free neutral lipids, mainly triacylglycerols (TGs) and sterol esters (SEs), condense into a nascent LD. How this transition occurs is poorly known. Here, we found that LDs preferably assemble at ER tubules, with higher curvature than ER sheets, independently of the LD assembly protein seipin. Indeed, the critical TG concentration required for initiating LD assembly is lower at curved versus flat membrane regions. In agreement with this finding, flat ER regions bear higher amounts of free TGs than tubular ones and present less LDs. By using an in vitro approach, we discovered that the presence of free TGs in tubules is energetically unfavorable, leading to outflow of TGs to flat membrane regions or condensation into LDs. Accordingly, in vitro LD nucleation can be achieved by the sole increase of membrane curvature. In contrast to TGs, the presence of free SEs is favored at tubules and increasing SE levels is inhibitory to the curvature-induced nucleation of TG LDs. Finally, we found that seipin is enriched at ER tubules and controls the condensation process, preventing excessive tubule-induced nucleation. The absence of seipin provokes erratic LD nucleation events determined by the abundance of ER tubules. In summary, our data indicate that membrane curvature catalyzes LD assembly.

Endosomal lipid signaling reshapes the endoplasmic reticulum to control mitochondrial function

Abstract

Recommended publications

ER remodeling via lipid metabolism

ER mediates spatial regulation of lysosome-endosome interactions via motion switch at junction sites

Antagonistic control of active surface integrins by myotubularin and phosphatidylinositol 3-kinase C...

A Golgi-derived vesicle potentiates PtdIns4P to PtdIns3P conversion for endosome fission