Article
Triaging of
a
-helical proteins to the mitochondrial
outer membrane by distinct chaperone machinery
based on substrate topology
Graphical abstract
Highlights
d
Genome-wide screens identify biogenesis factors for diverse
⍺
-helical OMM proteins
d
Parallel pathways mediate cytosolic targeting based on
substrate topology
d
The NAC complex selectively targets polytopic proteins to
the outer membrane
d
TTC1 is a cytosolic chaperone involved in OMM signal-
anchored protein biogenesis
Authors
Gayathri Muthukumar,
Taylor A. Stevens, Alison J. Inglis, ...,
Rebecca M. Voorhees, Alina Guna,
Jonathan S. Weissman
Correspondence
aguna@wi.mit.edu (A.G.),
weissman@wi.mit.edu (J.S.W.)
In brief
Muthukumar et al. use genome-wide and
systematic arrayed CRISPRi screens to
identify
⍺
-helical OMM protein biogenesis
and quality control factors that triage
substrates by topology. They establish
NAC as a targeting factor for polytopic
proteins and identify TTC1 as a cytosolic
chaperone for signal-anchored proteins,
using a C-terminal hydrophobic groove.
Muthukumar et al., 2024, Molecular Cell
84
, 1–19
March 21, 2024
ª
2024 The Authors. Published by Elsevier Inc.
https://doi.org/10.1016/j.molcel.2024.01.028
ll
Article
Triaging of
a
-helical proteins
to the mitochondrial outer membrane by distinct
chaperone machinery based on substrate topology
Gayathri Muthukumar,
1
,
3
Taylor A. Stevens,
2
Alison J. Inglis,
2
Theodore K. Esantsi,
1
,
4
Reuben A. Saunders,
1
,
4
,
5
,
6
Fabian Schulte,
1
Rebecca M. Voorhees,
2
Alina Guna,
1
,
2
,
*
and Jonathan S. Weissman
1
,
3
,
4
,
7
,
8
,
*
1
Whitehead Institute for Biomedical Research, Cambridge, MA 02142, USA
2
Division of Biology and Biological Engineering, California Institute of Technology, 1200 East California Avenue, Pasadena, CA 91125, USA
3
Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02142, USA
4
Howard Hughes Medical Institute, Massachusetts Institute of Technology, Cambridge, MA 02142, USA
5
Tetrad Graduate Program, University of California, San Francisco, San Francisco, CA 94158, USA
6
Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, CA 94158, USA
7
David H. Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02142, USA
8
Lead contact
*Correspondence:
aguna@wi.mit.edu
(A.G.),
weissman@wi.mit.edu
(J.S.W.)
https://doi.org/10.1016/j.molcel.2024.01.028
SUMMARY
Mitochondrial outer membrane
⍺
-helical proteins play critical roles in mitochondrial-cytoplasmic communi-
cation, but the rules governing the targeting and insertion of these biophysically diverse proteins remain
unknown. Here, we first defined the complement of required mammalian biogenesis machinery through
genome-wide CRISPRi screens using topologically distinct membrane proteins. Systematic analysis of
nine identified factors across 21 diverse
⍺
-helical substrates reveals that these components are organized
into distinct targeting pathways that act on substrates based on their topology. NAC is required for the effi-
cient targeting of polytopic proteins, whereas signal-anchored proteins require TTC1, a cytosolic chaperone
that physically engages substrates. Biochemical and mutational studies reveal that TTC1 employs a
conserved TPR domain and a hydrophobic groove in its C-terminal domain to support substrate solubiliza-
tion and insertion into mitochondria. Thus, the targeting of diverse mitochondrial membrane proteins is
achieved through topological triaging in the cytosol using principles with similarities to ER membrane protein
biogenesis systems.
INTRODUCTION
⍺
-helical integral membrane proteins are found in prokaryotic
plasma membranes and across eukaryotic membranes.
In eukaryotes, this class of proteins plays critical roles in main-
taining cellular homeostasis, primarily in the secretory pathway
and mitochondrial membranes.
1–3
Their functional diversity is
enabled by a wide range of physical properties, such as hydro-
phobicity, charged residues, topology, and variable helix lengths
of transmembrane domains (TMDs), posing challenges for
biosynthesis and folding.
4–6
In particular, the aggregation-prone
TMDs must be kept soluble in the cytosol before delivery to the
correct compartment.
7
Ultimately, the cytosolic and membrane-
embedded biosynthetic machinery must coordinate to facilitate
the insertion and folding of
⍺
-helical membrane proteins in the
lipid bilayer.
Eukaryotic membrane protein biogenesis has been most
intensely studied at the endoplasmic reticulum (ER).
8
Here, the
cell has developed parallel pathways to accommodate the high
diversity of substrate types. Initially, it was thought that all pro-
teins relied on co-translational targeting mediated by the ribo-
some-associated signal recognition particle (SRP), which shields
nascent TMDs
9–11
during delivery to the SEC61 translocon for
membrane insertion.
12
,
13
However, SRP and SEC61 cannot
accommodate all membrane-bound proteins. For example,
tail-anchored (TA) proteins, defined by a single TMD close to
the stop codon, are targeted post-translationally as the nascent
chain is released before the emergence of the TMD.
14–16
Post-
translation, cytosolic factors such as GET3, via SGTA
17
and
calmodulin (CaM),
18
capture TAs based on hydrophobicity and
target them to either the EMC or GET complexes for inser-
tion.
19–22
In addition to these strategies, ER biogenesis also re-
lies on TMD assembly factors
23
and quality-control factors
24–26
to degrade aberrant TMDs. Whether a similar set of strategies
exists for
⍺
-helical protein biogenesis at the mitochondria re-
mains unknown.
Molecular Cell
84
, 1–19, March 21, 2024
ª
2024 The Authors. Published by Elsevier Inc.
1
This is an open access article under the CC BY license (
http://creativecommons.org/licenses/by/4.0/
).
ll
OPEN ACCESS
Please cite this article in press as: Muthukumar et al., Triaging of
a
-helical proteins to the mitochondrial outer membrane by distinct chaperone ma-
chinery based on substrate topology, Molecular Cell (2024), https://doi.org/10.1016/j.molcel.2024.01.028
The outer mitochondrial membrane (OMM) proteome is critical
for mitochondrial-cytoplasmic communication. There are two
central classes of integral OMM proteins:
b
-barrel proteins,
which are also found in bacterial outer membranes, and
⍺
-helical
TMD-containing proteins, an evolutionarily newer and broader
class. The OMM
⍺
-helical proteins resemble the ER integral
membrane proteins in their biophysical and topological diversity
and can be broadly categorized by type as signal-anchored (SA,
single TMD at the N-terminus), TA (single TMD at the C-termi-
nus), and polytopic proteins. This diversity suggests the exis-
tence of multiple distinct biogenesis pathways, but mechanistic
knowledge, including the full range of factors involved in their
biogenesis, remains incomplete.
All OMM proteins are nuclear-encoded and must be specif-
ically targeted to the membrane in an insertion-competent
manner.
27–29
Unlike the proteins targeted to the intermembrane
space (IMS), inner membrane, or matrix, OMM proteins do not
have canonical mitochondrial targeting sequences (MTSs).
Elegant studies have demonstrated that MTS proteins are im-
ported into the mitochondria through the central TOM40 pore
30
in yeast and mammals, after recognition by the TOM20 and
TOM22 receptors.
29,31
Hydrophobic internal carrier proteins
are recognized by the alternate TOM70 receptor
32
before import
through TOM40.
b
-barrel OMM proteins are also imported
through TOM40 and subsequently inserted from the IMS by
the SAM complex, using a mechanism evolutionarily conserved
from bacteria.
33
,
34
By contrast,
⍺
-helical OMM proteins do not use either the
TOM40 pore or the SAM complex,
35
and structural studies
have established that, unlike the ER insertases, TOM40
does not provide an energetically favorable lateral gate for
insertion.
36–39
Instead, distinct insertases for
⍺
-helical OMM
proteins are employed in yeast, trypanosomes, and mam-
mals.
35
,
40
,
41
The MIM complex was identified as an
⍺
-helical
insertase in yeast
35
,
42
,
43
and later shown to be functionally
interchangeable with the trypanosome multi-spanning protein
pATOM36,
40
despite a lack of structural or sequence similar-
ity, as an example of convergent evolution. More recently, we
identified the multi-spanning
⍺
-helical proteins MTCH1/2
41
as
being necessary and sufficient for inserting
⍺
-helical proteins
in the OMM. However, MTCH2 has minimal exposure to the
cytosol, and
⍺
-helical OMM proteins have varying depen-
dencies on these insertases. Thus, there remain open ques-
tions regarding the cytosolic targeting machinery and the rules
used by the biogenesis machinery overall to support the full
range of OMM proteins.
Here, we employ a combination of systematic large-scale ge-
netic screens and biochemical characterization to decipher the
logic involved in coordinating diverse
⍺
-helical substrate biogen-
esis at the OMM. First, we use genome-wide CRISPRi screens in
human cells for four topologically distinct substrates to uncover
varied mammalian biogenesis and quality control machinery.
Then, we systematically test the effects of nine key identified fac-
tors across twenty-one diverse substrates to determine distinct
biogenesis pathways for specific substrate classes. Finally, we
characterize modes of cytosolic targeting to the outer membrane
using biochemical and mutational studies based on high-confi-
dence AlphaFold2 models and cell biological assays, showing
that distinct chaperone networks triage
⍺
-helical proteins based
on their topologies.
RESULTS
Systematic genome-wide CRISPRi screens identify
factors required for
⍺
-helical OMM protein biogenesis
We selected representative topologically distinct
⍺
-helical sub-
strates to use as reporters in genome-wide CRISPRi screens:
the iron-sulfur cluster biogenesis factor CISD1 as a SA pro-
tein,
44–46
the 3-TMD mitophagy effector FUNDC1 with its C-ter-
minus in the IMS and N-terminus in the cytosol,
47
and the 5-TMD
cholesterol translocator TSPO with its N-terminus in the IMS and
C-terminus in the cytosol
48
,
49
(
Figure 1
A). For comparison, we
used the recently described TA protein OMP25,
41
which is
post-translationally targeted to the membrane and requires
MTCH2 for mitochondrial insertion. We adapted a split GFP re-
porter system
41
,
50
to a fluorescence-activated cell sorting
(FACS)-based screening platform, expressing GFP1–10 in the
IMS and fusing GFP11 onto the reporter terminus with predicted
IMS localization (
Figure 1
A). Thereby, we could specifically
monitor OMM insertion rather than mislocalization into other or-
ganelles or cytosolic retention. We verified reporter mitochon-
drial localization (
Figure S1
A) and topologically correct insertion
by showing that only the reporters with GFP11 at the IMS-pre-
dicted terminus result in fluorescence (
Figure S1
B).
We engineered cell lines stably expressing the endogenous re-
porter sequences along with an expression normalization marker
(RFP), IMS GFP1–10, and the CRISPRi machinery (
Figure S1
C).
Cells with perturbed GFP/RFP fluorescence after genome-scale
sgRNA library transduction
51
and selection were isolated using
FACS and the associated sgRNAs identified with next-genera-
tion deep sequencing. We analyzed sgRNAs enriched in either
high or low populations of GFP/RFP fluorescence using this ra-
tiometric approach to identify putative quality control and
biogenesis factors, respectively.
We first identified several ‘‘hits’’ previously known to impact
our reporters, validating our screening approach. Primarily,
MTCH2 depletion impacted all our substrates to varying de-
grees, confirming its role in the insertion of diverse
⍺
-helical
proteins beyond TAs
41
(
Figure 1
B). Additionally, knockdown of
the TOMM70 receptor specifically impacted the polytopic
TSPO reporter, consistent with its role in mediating TSPO
biogenesis
52
(
Figure 1
B). Further support of our screen design
came from the CISD1 screen, where CISD1’s partner iron-sul-
fur-biogenesis factors (FAM96A, NARFL, NFS1, ABCB7) were
specific hits (
Figure S1
D).
53
Importantly, we also identified putative biogenesis and quality
control factors not previously associated with
⍺
-helical OMM
proteins, and many of these factors showed strong differential
impacts on the four reporters. Cytosolic factors were prominent
among these, presumably mediating chaperoning and targeting
to the outer membrane. First, depletion of the poorly character-
ized tetratricopeptide repeat (TPR) domain-containing protein
TTC1, implicated in protein folding and trafficking at the mito-
chondria and contact sites,
54
,
55
specifically impacted biogenesis
of our SA reporter CISD1 (
Figure 1
B) without impacting topolog-
ically distinct OMM reporters or insertion into the IMS (
Figure 1
C).
ll
OPEN ACCESS
Article
2
Molecular Cell
84
, 1–19, March 21, 2024
Please cite this article in press as: Muthukumar et al., Triaging of
a
-helical proteins to the mitochondrial outer membrane by distinct chaperone ma-
chinery based on substrate topology, Molecular Cell (2024), https://doi.org/10.1016/j.molcel.2024.01.028
Figure 1. Identifying biogenesis factors for topologically distinct
⍺
-helical mitochondrial outer membrane proteins
(A) Split GFP reporter system querying insertion of
⍺
-helical OMM proteins (SA, signal-anchored; poly, polytopic; and TA, tail-anchored). GFP11-fused reporters
expressed in K562 IMS GFP 1–10 cells with an RFP translation control.
(B) Volcano plots for the four genome-wide CRISPRi FACS screens (two replicates each). Specific genes of interest are highlighted across all screens.
The
OMP25 screen was conducted in a previous study
41
and is replotted to allow for direct comparison.
(C and D) Integration of GFP11-fused reporters in K562 IMS GFP1–10 cells under (C) TTC1 or (D) MARCHF5 depletion. Flow data were normalized based on the
GFP:RFP ratio and plotted as histograms.
See
Figure S1
.
ll
OPEN ACCESS
Article
Molecular Cell
84
, 1–19, March 21, 2024
3
Please cite this article in press as: Muthukumar et al., Triaging of
a
-helical proteins to the mitochondrial outer membrane by distinct chaperone ma-
chinery based on substrate topology, Molecular Cell (2024), https://doi.org/10.1016/j.molcel.2024.01.028
Given that TTC1 has a TPR domain, which is known to recruit
cytosolic HSP90 and/or HSP70,
56
it is an attractive chaperone
candidate. Similarly, depletion of J-domain HSP40 chaperone
DNAJC24
57
impacted polytopic and SA reporters, but not
OMP25, a TA. Both members of the co-translational chaperone
NAC (NACA and BTF3)
58
were required for efficient biogenesis of
polytopic reporters TSPO and FUNDC1, but not the TA reporter
OMP25 (
Figure 1
B). This suggests that, like ER polytopic pro-
teins, polytopic OMM proteins need to be shielded soon after
emergence from the ribosome.
13
We also identified potential quality control factors whose
depletion causes increased reporter OMM levels. For example,
depletion of membrane-resident E3 ligase MARCHF5
59
,
60
spe-
cifically impacted the polytopic reporter TSPO (
Figure 1
B) but
not SA reporter CISD1 or IMS control TIMM9A (
Figure 1
D). Pre-
sumably, MARCHF5 ubiquitinates aberrant TMDs of TSPO for
proteasomal degradation, as outlined for other known sub-
strates.
59
,
60
UBQLN1 depletion specifically impacted the TA re-
porter OMP25 (
Figure 1
B),
41
consistent with its role in facilitating
degradation of mitochondrial TA proteins that fail to be in-
serted.
61
A cross-comparison of gene-level phenotypes across
the four screens showed differential effects (
Figure S1
D;
Table S1
for complete screen data), supporting our initial hypoth-
esis that a complex network of factors would be required to co-
ordinate diverse OMM protein biogenesis. This analysis also
ruled out hits falling along the diagonal with more general effects
on the OMM or the IMS. For example, we noted that canonical
TOM complex receptors TOMM20 and TOMM22
31
affected all
reporter types similarly and attributed this to impacts on the
IMS GFP1–10 system (
Figure S1
E).
Substrate type dictates factor dependence
We next sought to determine whether factors identified in our
genome-wide CRISPRi screens had reporter-specific effects or
more general roles in
⍺
-helical OMM protein biogenesis. To
this end, we designed a comprehensive panel of functionally
and biophysically diverse SA, TA, and polytopic reporters as
our limited screen reporter set made it hard to resolve mecha-
nistic patterns (
Figure 2
A). Polytopic reporters MUL1 and
MARCHF5 were fused to full-length GFP at the C-terminus due
to having both termini in the cytosol and being incompatible
with the split GFP system (
Figure S1
F). As controls we used
IMS proteins with canonical targeting sequences, such as
MICU1 and LACTB, and those with internal targeting sequences,
such as TIMM9A and TIMM10, which are dependent on known
machinery such as TOMM40 and MIA40.
62
,
63
Due to their differential effects in the screens, we chose to test
the cytosolic factors TTC1, NACA, BTF3, and DNAJC24, puta-
tive quality control factors MARCHF5 and UBQLN1, MTCH1/2
insertases, and TOMM70 across the entire reporter panel. This
was performed as a large arrayed screen in duplicate, with the
data depicted as a heatmap colored by the fold change in re-
porter biogenesis for each factor perturbation relative to a non-
targeting sgRNA control condition (
Figure S2
A; see also
Figures 2
C and
S2
B for individual data points). Factor depletion
was confirmed using RT-qPCR (
Figure S2
C). All factors
impacted the reporters differentially, showing that their effects
were reporter-specific and not a consequence of generalized
outer membrane dysregulation. Further, the effects we saw
were likely not due to impacts on the IMS GFP 1–10 system as
the IMS controls remained relatively unchanged across all factor
perturbations (heatmap in
Figure S2
A;
Table S2
); only minor de-
fects in MICU1 and LACTB GFP11 biogenesis were seen under
DNAJC24 depletion.
Unbiased hierarchical clustering on the raw data (
Figure S2
A)
first revealed that the reporters were largely organized by type
(i.e., SA, TA, polytopic, and IMS control) as well as topology
(orientation of the termini with respect to the membrane); for
example, FUNDC1 and MTCH2 (both polytopic) share the
same topology and cluster closest to each other. The SA pro-
teins were partitioned by their strong differential dependency
on TTC1, with an
50% reduction in OMM integration upon
TTC1 loss (
Figure S2
A). NAC complex members NACA and
BTF3 primarily impacted the polytopic reporters TSPO,
FUNDC1, and MTCH2, distinguishing them from other substrate
types. We also noted factors whose effects were not type- or to-
pology-specific. Specifically, MARCHF5 depletion impacted
TSPO and other single-pass reporters such as RHOT1 (TA)
and PLD6 (SA), showing no topology-specific pattern. Similarly,
TOMM70 depletion reduced TSPO OMM integration by
70%
but no other reporter to this extent; this, combined with
MARCHF5’s unique effect on TSPO, separated it from other pol-
ytopic reporters (
Figures S2
A and S2B). Finally, we saw that
NACA separated from all factors in the heatmap, likely due to
previously observed toxicity in its absence
64
by stronger ZIM3-
mediated CRISPRi depletion
65
(
Figures S2
A and S2C). This pre-
sented as a wider distribution of the GFP signal peak due to
significantly fewer recorded live cell events, inflating the
GFP:RFP ratios across the reporter set and causing inaccurate
biogenesis measurements. To resolve this, we used weaker
KOX1-mediated CRISPRi depletion of NACA (
Figures 4
A and
S3
G) and showed that this primarily impacted polytopic proteins.
We next organized the factors by constructing a correlation
matrix followed by hierarchical clustering (
Figure 2
B), revealing
putative biogenesis ‘‘modules.’’ The correlation values were
highly reproducible (
Figure S3
A) and the matrix could be clus-
tered into three broad categories: (1) a quality control module
comprising MARCHF5 and UBQLN1; (2) a cytosolic targeting
module comprising TTC1, DNAJC24, NACA, and BTF3; and (3)
a membrane biogenesis module comprising MTCH1, MTCH2,
and TOMM70 (
Figure 2
B). Within the targeting module, both
members of the co-translational NAC complex clustered closest
to each other, segregating from TTC1 and DNAJC24, which have
been implicated in HSP co-chaperoning
56
,
57
(
Figure 2
B). The
closest correlating factor pair in the entire dataset was NACA
and BTF3 (
Figures 2
B and
S3
A), validating our clustering analysis
as they form a complex.
66
,
67
We also noted interesting individual
strong correlations, such as TTC1 and MTCH2 (
Figure 2
B). This
results from their impacts on a largely similar set of reporters (
Fig-
ure S2
A) and raises the intriguing possibility that substrates tar-
geted by TTC1 are recruited to MTCH2 for subsequent insertion.
Similarly, we clustered all the reporters by their patterns of cor-
relation with each other (
Figures 2
C and
S3
B). Once again, we
saw that the reporters were largely organized by type and topol-
ogy within type, as seen for the raw data cluster map (
Fig-
ure S2
A). Importantly, some of the IMS controls (particularly
ll
OPEN ACCESS
Article
4
Molecular Cell
84
, 1–19, March 21, 2024
Please cite this article in press as: Muthukumar et al., Triaging of
a
-helical proteins to the mitochondrial outer membrane by distinct chaperone ma-
chinery based on substrate topology, Molecular Cell (2024), https://doi.org/10.1016/j.molcel.2024.01.028
Figure 2. Systematically exploring biogenesis pathways for diverse
⍺
-helical OMM proteins
(A) Table describing reporters to query the specificity of screen hits (21 total). ‘‘LACTB’’ and ‘‘MICU1’’ indicate just the respective targeting seq
uence fused to
GFP11. MUL1 and MARCHF5 have full-length GFP fused to the C termini.
(B) Arrayed test for depletion of 9 screen hits against each reporter in (A) in K562 IMS GFP 1–10 ZIM3 CRISPRi cells. Data shown as GFP:RFP ratios normali
zed to
the control (non-targeting sgRNA) in a heatmap; individual squares are color-coded by the reporter type, as indicated in (A) (see
Figure S3
A for complete details).
Hierarchical clustering of a factor correlation matrix assigns putative biogenesis pathways.
(C) Clustering of reporter correlation matrix assigns groups largely defined by type and topology. Phenotypes from factor depletion that are predict
ive of SA and
TA reporter clustering patterns are shown here: (i) responses to TTC1 depletion, (ii) responses to MTCH2 depletion, (iii) responses to UBQLN1 deplet
ion.
See
Figures S2
and
S3
.
ll
OPEN ACCESS
Article
Molecular Cell
84
, 1–19, March 21, 2024
5
Please cite this article in press as: Muthukumar et al., Triaging of
a
-helical proteins to the mitochondrial outer membrane by distinct chaperone ma-
chinery based on substrate topology, Molecular Cell (2024), https://doi.org/10.1016/j.molcel.2024.01.028
Figure 3. TTC1 and TOMM70 impact
⍺
-helical outer membrane protein integration independent of their effects on the MTCH2 insertase
(A) Genetic interaction assay of TTC1 and MTCH2, using reporters dependent on TTC1 and MTCH2 (MIEF1, CISD1, MTCH2), MTCH2 alone (USP30, FUNDC1,
OMP25), and neither (TSPO, TIMM9A). Reporters were expressed in K562 ZIM3 CRISPRi IMS GFP1–10 cells and sgRNAs targeting either a combination of
factors or a factor with a non-targeting control. Data are represented as a heatmap colored by change in reporter integration (GFP:RFP ratio) relativ
e to the non-
targeting control.
(B) Flow data for representative reporters from (A) plotted as cdf (cumulative distribution function) plots.
(C) Genetic interaction of TOMM70A and MTCH2 assessed in a similar manner to (A), with reporters dependent on TOMM70A and MTCH2 (OMP25, MAVS),
TOMM70A alone (TSPO), and neither (MICU1, LACTB, TIMM9A).
(D) Flow data for representative reporters from (C) plotted as cdf plots.
(legend continued on next page)
ll
OPEN ACCESS
Article
6
Molecular Cell
84
, 1–19, March 21, 2024
Please cite this article in press as: Muthukumar et al., Triaging of
a
-helical proteins to the mitochondrial outer membrane by distinct chaperone ma-
chinery based on substrate topology, Molecular Cell (2024), https://doi.org/10.1016/j.molcel.2024.01.028