RESEARCH ARTICLE | CELL BIOLOGY OPEN ACCESS

TMED10 mediates the trafficking of insulin-­like growth factor

2 along the secretory pathway for myoblast differentiation
Tiantian Lia,1 , Feng Yanga,1 , Youshan Henga , Shaopu Zhoua , Gang Wanga , Jianying Wangb, Jinhui Wanga, Xianwei Chena ,
Zhong-­Ping Yaob,c,2 , Zhenguo Wua,2 , and Yusong Guoa,d,e,2

Edited by David Ginsburg, University of Michigan, Ann Arbor, MI; received September 7, 2022; accepted October 2, 2023

The insulin-­like growth factor 2 (IGF2) plays critical roles in cell proliferation, migra-
tion, differentiation, and survival. Despite its importance, the molecular mechanisms Significance
mediating the trafficking of IGF2 along the secretory pathway remain unclear. Here,
we utilized a Retention Using Selective Hook system to analyze molecular mechanisms Insulin-­like growth factor 2 (IGF2)
that regulate the secretion of IGF2. We found that a type I transmembrane protein, is a key regulator of skeletal
TMED10, is essential for the secretion of IGF2 and for differentiation of mouse myo- myogenesis during development.
blast C2C12 cells. Further analyses indicate that the residues 112-­140 in IGF2 are Currently, mechanisms
important for the secretion of IGF2 and these residues directly interact with the GOLD regulating IGF2 expression and
domain of TMED10. We then reconstituted the release of IGF2 into COPII vesicles. the signal transduction pathway
This assay suggests that TMED10 mediates the packaging of IGF2 into COPII vesicles induced by IGF2 have been
to be efficiently delivered to the Golgi. Moreover, TMED10 also mediates ER export of
extensively investigated.
TGN-­localized cargo receptor, sortilin, which subsequently mediates TGN export of
However, how IGF2 proteins,
IGF2. These analyses indicate that TMED10 is critical for IGF2 secretion by directly
regulating ER export and indirectly regulating TGN export of IGF2, providing insights after synthesized from
into trafficking of IGF2 for myoblast differentiation. ribosomes, are secreted to
perform their functions remains
secretion | IGF2 | TMED10 | sorting | COPII elusive. We demonstrate that a
cargo receptor, transmembrane
emp24 domain-­containing
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

Insulin-­like growth factor 2 (IGF2) is involved in various physiological activities, especially

protein 10 (TMED10), promotes
for skeletal myogenesis (1). It is an embryonic regulator of myogenesis and an autocrine
factor that promotes myoblast differentiation in vitro (2–4). Knockdown of IGF2 in mouse endoplasmic reticulum (ER)
muscle myoblasts impaired the differentiation progress, demonstrating that IGF2 regulates export of IGF2 via recognizing an
muscle stem cell differentiation (3). Despite its importance, molecular mechanisms that export signal on IGF2. Moreover,
mediate the secretion of IGF2 proteins from producing cells are poorly understood. TMED10 also mediates ER export
After synthesized from ribosomes, IGF2 needs to be delivered along the secretory of sortilin, which is important to
transport pathway to perform its physiological functions. IGF2 is first synthesized as a regulate the export of IGF2 from
precursor hormone containing 180 amino acids. After being imported into the ER, an the trans-­Golgi network. These
N-­terminal signal peptide is cleaved, generating pro-­IGF2 (IGF225-­180). The correctly analyses provide insight into the
folded pro-­IGF2 proteins are then packaged into transport vesicles to be delivered to the mechanisms regulating IGF2
Golgi apparatus. At the Golgi, pro-­IGF2 undergoes O-­glycosylation modifications and secretion, filling an important
endoproteolysis, generating IGF2 peptides IGF225-­128, IGF225-­111, and the mature IGF2
gap in our understanding of IGF2
(IGF225-­91) (5).
Coat protein complex II (COPII) is the key player that regulates the packaging of cargo from its expression to its
proteins into vesicles at the ER. In the conventional secretory transport pathway, soluble function.
cargo proteins in the lumen of the ER cannot be directly recognized by the COPII coat;
instead, these cytosolic proteins are thought to be transported under the recognition of
transmembrane cargo receptors (6). ERIGIC53 is a major cargo receptor that recruits a Author contributions: T.L., F.Y., and Y.G. designed
research; T.L., F.Y., Y.H., S.Z., G.W., Jianying Wang, Jinhui
variety of soluble cargo proteins to COPII vesicles in mammals (6). In addition, the p24 Wang, X.C., and Y.G. performed research; Z.-­P.Y., Z.W.,
proteins play crucial roles in ER-­Golgi bidirectional transport. Some p24 family members and Y.G. contributed new reagents/analytic tools; T.L.,
F.Y., Y.H., S.Z., G.W., Jianying Wang, Jinhui Wang, X.C.,
function as cargo receptors to regulate ER export of specific GPI-­anchored proteins and Z.-­P.Y., Z.W., and Y.G. analyzed data; and T.L., F.Y., and Y.G.
autotaxin in mammalian cells (6, 7) and to mediate secretion of Wnt proteins in Drosophila wrote the paper.

(8, 9). Soluble cargo proteins are also proposed to enter the nascent COPII vesicles by The authors declare no competing interest.

default, a process referred to as bulk flow (10). It is currently unknown whether the traf­ This article is a PNAS Direct Submission.

ficking of IGF2 is passively regulated by bulk flow or is actively mediated by cargo Copyright © 2023 the Author(s). Published by PNAS.
This open access article is distributed under Creative
receptors. Commons Attribution License 4.0 (CC BY).
The trans-­Golgi network (TGN) is another important station in the secretory transport 1
T.L. and F.Y. contributed equally to this work.
pathway. At the TGN, various cargo adaptors and receptors have been shown to capture 2
To whom correspondence may be addressed. Email:
cargo molecules into nascent vesicles (11). Sortilin is one of the Golgi-­localized cargo recep­ [email protected], [email protected], or zhongping.yao@
polyu.edu.hk.
tors that is crucial for the sorting of numerous proteins in the anterograde and retrograde
This article contains supporting information online at
pathways and is broadly involved in multiple physiological activities, including lipid metab­ https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.​
olism, neuronal development, immune system, myogenesis and diabetes (12–16). Particularly, 2215285120/-­/DCSupplemental.
the noncoding genetic variants at a locus near gene encoding sortilin were significantly Published November 6, 2023.

PNAS 2023 Vol. 120 No. 46 e2215285120 https://doi.org/10.1073/pnas.2215285120 1 of 12

 associated with LDL cholesterol and coronary artery disease in standard concentration we used in our study. We observed that as the
humans (17, 18). Further analysis indicates that sortilin interacts DNA concentration for transfection increased, so did the expression
with apolipoprotein B100 (ApoB100) and modulates the secretion level of RUSH-­IGF2-­HA in the cell lysates (SI Appendix, Fig. S1A).
of ApoB100-­containing lipoproteins in hepatocytes, but whether Across all conditions and expression levels, RUSH-­IGF2-­HA were
sortilin up-­regulates or down-­regulates lipoprotein secretion remains detected in the juxta-­nuclear Golgi area in around 80% of cells
controversial (19, 20). The cargo receptors that deliver IGF2 from 20 min after biotin treatment (SI Appendix, Fig. S1B). Notably, the
the TGN to the cell surface still remain unclear. percentage showed no significant difference across all conditions.
Here, we analyzed IGF2 intracellular transport utilizing the Furthermore, we identified a positive correlation between the expres­
Retention Using Selective Hook (RUSH) assay. We also reconsti­ sion level of RUSH-­IGF2-­HA and the abundance of secreted
tuted the release of IGF2 into COPII vesicles through an in vitro RUSH-­IGF2-­HA, two hours after biotin treatment (SI Appendix,
vesicle formation assay to quantify the efficiency of cargo packag­ Fig. S1C). A similar positive correlation was also found between the
ing. Through these approaches, we found that a p24 family pro­ abundance of secreted ShhN-­HA and the expression level of
tein, TMED10, functions as a cargo receptor to mediate ER export ShhN-­HA (21). These findings suggest that these cargo molecules,
of IGF2 for myoblast differentiation. Moreover, we found that irrespective of their expression levels, can be efficiently transported to
TMED10 regulates ER export of sortilin, and sortilin is important the Golgi and secreted, without reaching saturation. One explanation
for TGN export of IGF2. These studies shed light on molecular for the ability of a limited number of cargo receptors to mediate
mechanisms regulating IGF2 secretion along the secretory path­ trafficking of a large volume of clients is the recycling mechanism of
way for muscle stem cell differentiation. these cargo receptors. After delivering their clients to the destination,
the cargo receptors disassociate and return to their original location
Results to initiate the next round of trafficking.
We then performed an siRNA knockdown (KD) experiment
TMED10 Mediates ER-­to-­Golgi Trafficking and the Secretion to test whether TMED10 is required for IGF2 secretion. Given
of IGF2. We have recently performed coimmunoprecipitation that the abundance of secreted proteins generally correlates posi­
(co-­IP) experiments to reveal proteins that interact with the HA-­ tively with their expression levels, we normalized the quantity of
tagged N-­terminal fragment of sonic hedgehog (ShhN-­HA) or secreted IGF2 against the abundance found in the cell lysate in
HA-­tagged IGF2 (IGF2-­HA) (21). Label-­free quantitative mass the absence of biotin. This normalized value serves as an indicator
spectrometry analysis of the immunoprecipitated proteins identified of the efficiency of IGF2 secretion. The expression of TMED10
a transmembrane protein, TMED10, which binds stronger to was greatly reduced in HeLa cells transfected with siRNA against
IGF2-­HA than to ShhN-­HA (21). TMED10 is an ER-­and Golgi-­ TMED10 (Fig. 1F). Remarkably, the secretion efficiency of
localized transmembrane protein that belongs to the p24 family. RUSH-­IGF2-­HA was significantly decreased in TMED10 KD
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

To test whether TMED10 is important for IGF2 secretion, we cells (Fig. 1G, compare lanes 4 and 8, and Fig. 1H), indicating
utilized a RUSH transport assay (21–23). In this assay, HeLa cells that TMED10 plays an important role in IGF2 secretion. In con­
were transfected with plasmids encoding human IGF2 (aa: 25-­ trast, the secretion of another protein, ShhN, was not affected in
180, removed signal peptide) tagged with EGFP, the streptavidin TMED10 KD cells (Fig. 1 I and J), demonstrating that TMED10
binding peptide (SBP) and an HA tag at the C terminus (referred is a specific regulator for IGF2 rather than a common regulator.
to as RUSH-­IGF2-­HA or RUSH-­IGF225-­180-­HA) (Fig. 1A). This We then generated TMED10 knockout (KO) HeLa cells to test
plasmid also encodes streptavidin fused to a C-­terminal ER retention the effect of depleting TMED10 on IGF2 secretion. Western blot
signal (Lys-­Asp-­Glu-­Leu; Str-­KDEL). Due to the binding between analysis indicates that TMED10 was completely depleted in
streptavidin and SBP, RUSH-­IGF2-­HA was retained at the ER TMED10 KO HeLa cells (SI Appendix, Fig. S2A). The efficiency
upon expression (Fig. 1B, 0 min). It is noted that the RUSH system of RUSH-­IGF2-­HA secretion in TMED10 KO cells was also
does not halt ER export but rather creates an imbalance, favoring significantly reduced (SI Appendix, Fig. S2B, compare lanes 4 and
retrieval over export. Consequently, most of the cargo remains at the 8, and SI Appendix, Fig. S2C), demonstrating that TMED10 is
ER in the absence of biotin. When cells were incubated with biotin, essential for IGF2 secretion.
SBP was uncoupled with streptavidin, thereby releasing RUSH-­ We noticed that the size of the secreted RUSH-­IGF2-­HA
IGF2-­HA from the ER retrieval process (Fig. 1 C–E′). Over 80% detected by anti-­HA antibodies is similar to the size of
of cells showed Golgi-­localized IGF2 when the cells were incubated RUSH-­IGF2-­HA in cell lysates, suggesting that the secreted
with biotin for 20 min (Fig. 1C). After biotin treatment for 120 RUSH-­IGF2-­HA we detected is not the cleaved form. A possible
min, the signal of RUSH-­IGF2-­HA was greatly reduced (Fig. 1 explanation is that we used antibodies detecting the HA tag at the
E and E′), suggesting that IGF2 was secreted out of the cells or C terminus of IGF2 for the immunoblot analyses. We repeated
delivered to lysosomes for degradation. this assay by using N-­terminal HA-­tagged RUSH-­IGF2 (referred
We measured the efficiency of ER-­to-­Golgi trafficking of to as RUSH-­HA-­IGF2) to monitor the secretion of both big IGF2
RUSH-­IGF2-­HA by quantifying the percentage of cells showing and mature IGF2. We detected two bands in the medium group,
juxta-­nuclear located RUSH-­IGF2-­HA following biotin treatment. and their molecular weights match the predicted molecular
This method does not differentiate between cells in which RUSH weights of the pro-­IGF2 and mature IGF2 (Fig. 1K, lane 4).
cargo proteins are partly located in the ER and partly in the perinu­ Consistent with the previous analyses, the secretion of both
clear Golgi region, and those in which all RUSH cargo proteins are pro-­IGF2 and mature IGF2 was reduced after TMED10 KD
located in the Golgi. Despite this limitation, this quantification (Fig. 1K, compare lanes 4 and 8, and Fig. 1L).
approach has been efficiently employed previously to measure the We found that the abundance of RUSH-­IGF2-­HA in cell lysates
efficiency of ER-­to-­Golgi trafficking of RUSH cargo proteins (21, from TMED10 KD or KO groups was decreased after biotin treat­
24, 25). To test whether the trafficking of the cargo protein reaches ment (Fig. 1K, lanes 5–6, SI Appendix, Fig. S2B, lanes 5–6). We
saturation, we performed experiments to analyze the trafficking of hypothesize that some of the RUSH-­IGF2-­HA have been degraded
RUSH-­IGF2-­HA at varying expression levels. For this, we transfected rather than secreted. To test this, we performed the RUSH assay
HeLa cells with different concentrations (0.8 μg/mL, 1.6 μg/mL, 3.2 using TMED10 KO cells and treated the cells with the lysosomal
μg/mL) of plasmids encoding RUSH-­IGF2-­HA. 1.6 μg/mL is the inhibitor (bafilomycin A1), or the proteasome inhibitor (MG132)

2 of 12 https://doi.org/10.1073/pnas.2215285120 pnas.org
 Fig. 1. Knockdown of TMED10 causes defects in secretion
of IGF2. (A) A diagram demonstrating the design of the
RUSH-­IGF2-­HA construct and the RUSH assay. (B–E′) HeLa
cells were transfected with plasmids encoding Str-­KDEL and
full-­length RUSH-­IGF2-­HA. Day 1 after transfection, cells
were preincubated with cycloheximide for 2 h. Then, the
localization of RUSH-­IGF2-­HA was analyzed after incubating
with biotin and cycloheximide for the indicated time. The
view of the indicated area in panel E at a higher exposure
was shown in panel E′. (F) HeLa cells were transfected
with control siRNA or siRNA against TMED10. Day 2 after
transfection, the level of TMED10 and β actin in cell lysates
was analyzed by immunoblot. (G) Day 1 after transfection
with siRNAs, cells were retransfected with plasmids encoding
Str-­KDEL and RUSH-­IGF2-­HA. On day 3 after knockdown,
cells were preincubated with cycloheximide for 2 h. Then,
cells were incubated with biotin and cycloheximide for 2 h.
After biotin incubation, the level of RUSH-­IGF2-­HA in the
medium and in cell lysates was analyzed by immunoblot.
(H) Quantification of the abundance of secreted IGF2
normalized to the abundance detected in the cell lysate
group in the absence of biotin (mean ± SD; n = 3). (I) Day
1 after transfection with siRNAs, cells were retransfected
with plasmids encoding Str-­KDEL and RUSH-­ShhN-­HA.
On day 3 after knockdown, cells were preincubated with
cycloheximide for 2 h. Then cells were incubated with biotin
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

and cycloheximide for 2 h. After biotin incubation, the level

of RUSH-­ShhN-­HA in the medium and in cell lysates was
analyzed by immunoblot. (J) Quantification of the abundance
of secreted ShhN normalized to the abundance detected
in the cell lysate group in the absence of biotin (mean ±
SD; n = 3). (K) Day 1 after transfection with siRNAs, cells
were transiently transfected RUSH-­HA-­IGF2. On day 3 after
knockdown, cells were preincubated with cycloheximide for
2 h. Then, cells were incubated with biotin and cycloheximide
for 2 h. After biotin incubation, the abundance of RUSH-­
HA-­IGF2 in the medium and in cell lysates was analyzed
by immunoblot. (L) Quantification of the abundance of
secreted IGF2 normalized to the abundance detected in
the cell lysate in the absence of biotin (mean ± SD; n = 3).
In each experimental group of each replicated experiment,
the value was normalized to the average value of the Mock
group and the TMED10 KD group (H, J, and L). ***P < 0.001;
****P < 0.0001; n.s., not significant.

in the presence of biotin. We found that bafilomycin A1 treatment biotin, it was located at the ER in the majority of the TMED10-­
enhanced the abundance of RUSH-­IGF2-­HA in cell lysates after and IGF2-­coexpressing cells (SI Appendix, Fig. S4 A–C). Under
biotin treatment, whereas MG132 treatment only showed a modest this condition, only around 4% of the coexpressing cells showed
effect (SI Appendix, Fig. S2D), indicating that the reduced RUSH-­ detectable localizations of TMED10-­FLAG at the juxtanuclear area
IGF2-­HA protein levels in cell lysates after biotin treatment are (SI Appendix, Fig. S4G). After biotin treatment for 10 min, RUSH-­
mainly due to the lysosomal degradation. IGF2-­HA was located at the juxta-­nuclear Golgi area (SI Appendix,
Fig. S4 D–F). Under this condition, TMED10-­FLAG was colocalized
TMED10 Is Important for Packaging IGF2 into COPII Vesicles. Next, with RUSH-­IGF2-­HA at the juxta-­nuclear Golgi area in ∼60% of
we analyzed which step TMED10 is involved in the secretion of IGF2. the coexpressing cells (SI Appendix, Fig. S4 D–F and quantification
We found that knockdown of TMED10 caused a defect in ER-­to-­ in SI Appendix, Fig. S4G). The percentage of the coexpressing cells
Golgi trafficking of RUSH-­IGF2-­HA (Fig. 2 A–G and quantification showing juxta-­nuclear TMED10-­FLAG was significantly higher in
in Fig. 2H). The defect was also observed in TMED10 KO cells the presence of biotin than that detected in the absence of biotin
(SI Appendix, Fig. S3 A–R and quantification in SI Appendix, Fig. S3S). (SI Appendix, Fig. S4G). These analyses indicate that ER retention of
This defect was rescued by the expression of TMED10-­FLAG in IGF2 causes accumulations of TEMD10 at the ER. These analyses
TMED10 KO cells (SI Appendix, Fig. S3 T–AB and quantification also indicate that TMED10 traffics together with RUSH-­IGF2-­HA
in SI Appendix, Fig. S3AC), suggesting that TMED10 is important from the ER to the Golgi. We then performed a similar analysis in cells
for ER-­to-­Golgi trafficking of RUSH-­IGF2-­HA. When TMED10-­ coexpressing the RUSH construct of ShhN (RUSH-­ShhN-­HA) and
FLAG was coexpressed with RUSH-­IGF2-­HA in the absence of TMED10-­FLAG (SI Appendix, Fig. S4 H–N). Our findings revealed

PNAS 2023 Vol. 120 No. 46 e2215285120 https://doi.org/10.1073/pnas.2215285120 3 of 12

 Fig. 2. TMED10 mediates the release of IGF2 into COPII vesicles.
(A–F) HeLa cells were transfected with control siRNA or siRNA against
TMED10. Day 1 after transfection, cells were retransfected with
plasmids encoding Str-­KDEL and RUSH-­IGF2-­HA. On day 3 after
knockdown, cells were preincubated with cycloheximide for 2 h.
Then cells were incubated with biotin and cycloheximide for 20
min and the localization of IGF2 was analyzed (Size bar, 10 μm).
(G) Concurrently, the abundance of TMED10 and β actin in cell
lysates before biotin treatment was analyzed by immunoblot. A
representative example of three biological repeats was shown in
this panel. (H) Quantification of the percentage of cells showing
juxtanuclear-­localized RUSH-­IGF2-­HA at the indicated time point
after biotin treatment (mean ± SD; n = 3; >100 cells counted in each
experiment). (I) A diagram demonstrating the vesicle formation
assay. (J, K, and M) The vesicle formation assay was performed using
HEK293T cells (J) or HEK293T cells transfected with control siRNA or
siRNA against TMED10 (K and M). Vesicle fractions were analyzed
by immunoblot. (L and N) Quantification of the budding efficiency
of the indicated proteins from the vesicle formation assay (mean
± SD; n = 3). The budding efficiency was quantified by calculating
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

the abundance of RUSH-­IGF2-­HA in the vesicle fraction normalized

to the level of the cargo protein in 1% loading. The value was then
normalized to the average value of all experimental groups in each
replicated experiment. *P < 0.05; **P < 0.01; ***P < 0.001.

that TMED10-­FLAG did not co-­traffic with RUSH-­ShhN-­HA (with or without biotin, Fig. 2 K and L), suggesting that TMED10
from the ER to the Golgi following biotin treatment (SI Appendix, is important for packaging RUSH-­IGF2 into COPII vesicles.
Fig. S4N). These analyses indicate that TMED10 is cotransported In yeast, ERV25 (the yeast homolog of human TMED10) forms
with IGF2 but not ShhN from the ER to the Golgi. a heteromeric complex with EMP24, ERP1, and ERP2 (the yeast
We hypothesize that TMED10 functions as a cargo receptor to homologs of human TMED2, 4, and 7, respectively) (27). Their
regulate packaging IGF2 into COPII vesicles. To test this hypoth­ protein levels are interdependent, and these proteins function in
esis, we reconstituted the release of IGF2 into COPII vesicles using a cooperative manner (27). In mammals, TMED10 exists in a
the vesicle formation assay (24, 26). HEK293T cells transfected hetero-­oligomeric complex with TMED2, TMED7, and TMED9
with RUSH-­IGF2-­HA were permeabilized by digitonin. After (28, 29). We then analyzed the efficiency of budding of two p24
permeabilization, the semi-­intact cells were washed with cold family proteins, TMED2 and TMED7, and another cargo receptor,
KOAc buffer to remove cytosolic proteins. The semi-­intact cells ERGIC53, in control cells and in cells knockdown of TMED10.
were then incubated at 32 °C with GTP and an ATP regeneration Consistent with previous reports, the abundances of TMED7 and
system (ATPrS) in the presence or absence of biotin, rat liver TMED2 in cell lysates were markedly reduced in TMED10 KD
cytosol (RLC) and a GTPase defective mutant of SAR1A, SAR1A cells (Fig. 2M), indicating interdependence among TMED family
(H79G) (Fig. 2I). The released vesicles after incubation were then proteins for their stability. Consequently, the abundance of these
isolated by centrifugation and analyzed by immunoblot (Fig. 2I). two cargo proteins in the vesicle fraction was also reduced (Fig. 2M).
RUSH-­IGF2-­HA was detected in the vesicle fraction when the Interestingly, the packaging efficiency of ERGIC53 was found to
vesicle formation assay was performed in the presence of RLC be enhanced in TMED10 KD cells (Fig. 2 M and N).
(Fig. 2J, lane 3). We detected biotin-­independent budding of We next analyzed the colocalization between IGF2 and
RUSH-­IGF2-­HA suggesting that the RUSH system does not TMED10 utilizing a digitonin-­permeabilized cell assay. We have
block ER export. The abundance of IGF2 in the vesicle fraction previously demonstrated that this assay locks the ER export pro­
was enhanced when the assay was performed in the presence of cess at the sorting step (21), providing a convenient way to analyze
biotin likely due to the release from retrieval (Fig. 2J, compare the colocalization between cargo receptors and cargo molecules.
lanes 3 and 4). Adding SAR1A (H79G) blocked the vesicular Cells coexpressing RUSH-­IGF2-­HA and TMED10-­FLAG were
release of RUSH-­IGF2-­HA (Fig. 2J, compare lanes 4 and 5). permeabilized by digitonin and washed with high salt buffer to
These results indicate that this assay successfully reconstituted the remove the endogenous cytosolic proteins. Then cells were incu­
release of IGF2 into COPII vesicles. Remarkably, knockdown of bated with rat liver cytosol, biotin, and GTPγS for 15 min. After
TMED10 caused a significant reduction in the abundance of incubation, RUSH-­IGF2-­HA and TMED10-­FLAG showed
RUSH-­IGF2-­HA in transport vesicles under both conditions punctate localization patterns (SI Appendix, Fig. S5 A–C). Many

4 of 12 https://doi.org/10.1073/pnas.2215285120 pnas.org
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

Fig. 3. The residues 112-­140 in IGF2 are req­

uired for ER-­to-­Golgi trafficking of IGF2. (A)
Sequence alignment of IGF2 from different
species. (B–E and G–X) HeLa cells transfected
with plasmids encoding the indicated RUSH
constructs were incubated with biotin and
cycloheximide for the indicated period of time.
The localizations of the indicated proteins were
analyzed by immunofluorescence (Size bar, 10
μm). (F and Y) The percentage of cells showing
juxtanuclear localization patterns of the RUSH
construct after biotin treatment was quantified
(mean ± SD; n = 3; >100 cells counted in each
experiment). **P < 0.01; ***P < 0.001; ****P <
0.0001; n.s., not significant.

IGF2 punctate structures overlapped with the punctate structures and F). In contrast, RUSH-­IGF298-­180-­HA localized at the Golgi
of TMED10 (SI Appendix, Fig. S5 A–C, magnified views in apparatus in around 80% of the cells after biotin treatment for
SI Appendix, Fig. S5 D–I). This analysis indicates that IGF2 colo­ 10 min (Fig. 3 C and F), suggesting that IGF298-­180 is the critical
calizes with TMED10 upon exiting the ER. part of IGF2 trafficking from the ER to the Golgi.
Sequence alignment indicates that the residues between positions
Residues 112-­140 in IGF2 Are Important for ER-­to-­Golgi 112 and 140 of IGF2 are conserved across species (Fig. 3A, highlighted
Trafficking of IGF2. The next question we want to address is which in the green box). To test whether these residues are important for ER
motif of IGF2 is the major determinant for IGF2 trafficking and export of IGF2, we generated a RUSH construct of IGF298-­180 frag­
secretion. We performed sequence alignment of different IGF2 ment depleted these residues (RUSH-­IGF298-­180, Δ112-­140-­HA). Upon
orthologues and synthesized different IGF2 truncated proteins; biotin treatment for 10 min or 20 min, RUSH-­IGF298-­180-­HA showed
each contains one or several distinct highly conserved regions juxtanuclear localization in the majority of cells (Fig. 3 G–O and Y).
(Fig. 3A). Utilizing the RUSH assay, we tested ER-­to-­Golgi In contrast, the majority of cells expressing RUSH-­IGF298-­180, Δ112-­140-­
trafficking of these mutant constructs. Interestingly, we found that HA showed an ER pattern (Fig. 3 P–Y). Further analyses indicate
RUSH-­IGF225-­48-­HA was located at the ER in over 90% of cells that residues 112 to 140 in IGF2 are sufficient for SBP-­EGFP to be
after 10 min biotin treatment (Fig. 3 E and F). RUSH-­IGF225-­97-­ delivered from the ER to the Golgi with an efficiency that is similar
HA also showed a defect of ER-­to-­Golgi trafficking (Fig. 3 D to full-­length IGF2 (SI Appendix, Fig. S6 A–M). In summary, these

PNAS 2023 Vol. 120 No. 46 e2215285120 https://doi.org/10.1073/pnas.2215285120 5 of 12

 Fig. 4. Residues 112-­140 in IGF2 directly interact with the GOLD
domain of TMED10. (A, C, and E) HEK293T cells were cotransfected
with plasmids encoding the indicated constructs. Day 1 after
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

transfection, the cells were treated with 2 mM DSP, and the cell
lysates were incubated with beads conjugated with anti-­FLAG
antibodies. The bound proteins were analyzed by immunoblot.
(B, D, and F) Quantification of relative levels of indicated proteins
that coimmunoprecipitated with the FLAG-­tagged proteins (mean
± SD; n = 3). The quantification was performed by calculating the
abundance of the bound protein normalized to the abundance of
the protein in the loading. The value was then normalized to the
average value of all of the experimental groups in each biological
repeat. (G) Peptides corresponding to the 112-­140 residues in IGF2
were covalently linked to thiopyridone sepharose 6B, incubated
with purified GST or GST-­tagged human TMED10 GOLD domain
(residues 1-­130). After incubation, the bound proteins were
analyzed by immunoblot. (H) Levels of GST-­TMED101-­130 bound
to the IGF2 peptides were quantified (mean ± SD; n = 3). The
quantification is normalized to the average level of GST and GST-­
TMED10 (1-­130) that bound to the IGF2 peptides in each biological
repeat. *P < 0.05; **P < 0.01; ****P < 0.0001.

observations provide evidence demonstrating that IGF2112-­140 is the region (30, 31) (Fig. 4C). The cytosolic portion of many TMED
ER-­to-­Golgi transport motif of IGF2. proteins contains two C-­terminal hydrophobic residues that promote
ER export (32) and dilysine motifs (KK) that are important for ER
IGF2 Interacts with the GOLD Domain of TMED10, and This retrieval (33). TMED10, while it possesses a dilysine motif within its
Interaction Depends on Residues 112-­140 in IGF2. We then cytosolic domain, lacks hydrophobic residues at its C terminus. As
performed co-­IP experiments using HEK293T cells cotransfected TMED10 forms a complex with other proteins from the TMED10
with plasmids encoding FLAG-­tagged TMED10 (TMED10-­ family, TMED10 might be enriched into COPII vesicles through the
FLAG) and HA-­tagged IGF2 or ShhN (IGF2-­HA or ShhN-­ ER export motif found in other members of the TMED family. The
HA). A cross-­linker DSP was used in the co-­IP experiments to GOLD domain is implicated in recognizing cargo proteins (34). We
stabilize the interaction. The co-­IP assay revealed that IGF2-­HA then generated FLAG-­tagged TMED101-­130, which contains the SS
bound TMED10-­FLAG in cell lysates (Fig. 4A). The percentage motif and the GOLD domain, to test whether the GOLD domain
of IGF2-­HA in cell lysates that bound to TMED10-­FLAG was is sufficient for the interaction. Strikingly, the abundance of IGF2-­HA
significantly higher than the percentage of ShhN-­HA in cell lysates that bound to TMED101-­130-­FLAG was significantly higher than
that bound to TMED10-­FLAG (Fig. 4 A and B), indicating that bound to the full-­length TMED10-­FLAG (Fig. 4C, compare
TMED10 specifically interacts with IGF2. lanes 4 and 5, and Fig. 4D), suggesting that TMED10 binds IGF2
TMED family proteins have highly conserved structures. The through its luminal GOLD domain.
luminal part of TMED proteins consists of the signal sequence (SS), Since the ER-­to-­Golgi transport of IGF2 depends on its resi­
the Golgi dynamics (GOLD) domain, and the coiled-­coil (CC) dues between 112 and 140, we next tested whether this motif is

6 of 12 https://doi.org/10.1073/pnas.2215285120 pnas.org
 important for the IGF2-­TMED10 interaction. We found that Sortilin Is Another Cargo Client of TMED10. Next, we sought to
depleting this motif significantly reduced the abundance of identify other cargo proteins that depends on TMED10 to be
IGF2-­HA that bound to TMED10-­FLAG (Fig. 4E, compare enriched into transport vesicles. We have previously developed
lanes 3 and 4, and Fig. 4F). We then performed a peptide b­ inding a vesicle formation assay in combination with a label-­free
assay to study whether this interaction is direct. Synthesized pep­ quantitative mass spectrometry approach and this approach
tides corresponding to the 112-­140 residues of IGF2 (IGF2112-­140) revealed the cargo clients of two ER cargo receptors, ERGIC53
were covalently linked to beads. The beads were then incubated and SURF4 (26). Here, we utilized a similar approach to uncover
with purified GST or GST-­tagged TMED10 GOLD domain the cargo clients of TMED10. A large-­scale vesicle formation
(GST-­TMED101-­130). The result shows that the abundance of assays were performed using donor membranes provided by wild-­
GST-­TMED101-­130 that bound to IGF2112-­140 was significantly type (WT) or TMED10 KO HeLa cells. A label-­free quantitative
higher than the abundance of GST that bound to the peptides mass spectrometry analysis was then conducted to compare the
(Fig. 4 G and H), suggesting that the ER export motif of IGF2 protein profiling of vesicles produced from these two experimental
interacts with the GOLD domain of TMED10 directly. groups (the WT group and the TMED10 KO group, Dataset S2).
We detected peptides that match TMED10 in the vesicle fraction
TMED10 Is Important for the Secretion of IGF2 from C2C12 Cells generated by the TMED10 KO cells. A possible explanation is that
for Muscle Stem Cell Differentiation. We then tested whether a negligible quantity of TMED10 may persist in rat liver cytosol
TMED10 is important for the secretion of IGF2 from mouse prepared from rat livers, which may associate with vesicles after
C2C12 myoblasts. We collected the medium incubated with the vesicle formation assay. Although these residual amounts of
C2C12 cells transfected with control siRNA or siRNA against proteins are not detectable through immunoblotting, they may be
TMED10. The proteins in the medium were TCA precipitated detected using mass spectrometry that has the capacity to detect
and then analyzed by label-­free quantitative mass spectrometry to proteins in the low picogram range. We found that the abundance
compare the abundances of proteins detected in the medium from of a series of transmembrane proteins is greatly reduced in the
the two experimental groups (Dataset S1, sheet 1). The abundance vesicle fraction in the TMED10 KO group based on two biological
of IGF2 was at least 1.9-­fold higher in the medium of control cells repeats (Fig. 6A, average fold change of TMED10 KO/WT < 0.5).
compared to the medium of TMED10 knockdown cells in each of These identified transmembrane proteins including several p24
the two replicated experiments (Dataset S1, sheet 3, highlighted family proteins: TMED1, TMED2, TMED3, TMED4, TMED5,
in Red). Along with IGF2, we identified 283 proteins that also TMED7, and TMED9 (Fig. 6A). This decrease is presumably
exhibited a similar increase in abundance in the medium of control caused by the degradation of these TMED proteins induced by
cells (Dataset S1, sheet 2). Of these proteins, 53 are secretory the depletion of TMED10, thereby reducing their presence not
proteins (Dataset S1, sheet 3), while the others are transmembrane, only within the cells but also in the vesicle fraction.
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

GPI-­anchored, cytoplasmic, peripheral membrane, or nuclear In addition, we identified a Golgi-­and plasma-­membrane local­
proteins. We suspect that those nonsecretory proteins including ized transmembrane protein, sortilin, that depends on TMED10
cytoplasmic, peripheral membrane, transmembrane, GPI-­anchored, to be enriched into transport vesicles (Fig. 6A, highlighted in red,
and nuclear proteins might originate from dying cells during the Fig. 6B). The total level of sortilin was also greatly decreased in both
cell culturing process or from extracellular vesicles secreted by cells. cell lysates and vesicle fraction of the KO group (Fig. 6B). To inves­
The abundance of these proteins in the medium may be indirectly tigate whether TMED10 is important for the ER-­to-­Golgi traffick­
influenced by the knockdown of TMED10 in C2C12 cells. ing of sortilin, we generated a RUSH construct of sortilin and
Expression of the myoblast differentiation marker, myogenin, performed the RUSH transport assay in WT or TMED10 KO
was significantly reduced in TMED10 knockdown cells incu­ HeLa cells. We found that the ER-­to-­Golgi transport of sortilin was
bated with the differentiation medium (DM) when compared strongly impaired in KO cells (Fig. 6 C and D): after biotin treat­
to the control cells incubated at the same condition (Fig. 5A). ment for 40 min, the majority of sortilin was trapped at the ER.
Adding purified IGF2 into the differentiation medium rescues This defect was rescued by transfecting TMED10-­FLAG in KO
the expression of myogenin in TMED10 knockdown cells cells (Fig. 6 C and D). These results indicate that TMED10 is also
(Fig. 5A and quantification in Fig. 5B). These results indicate essential for ER-­to-­Golgi trafficking of sortilin. We then performed
that TMED10 plays an important role in myoblast differenti­ the vesicle formation assay and immunoisolated vesicles enriched
ation by functioning as a cargo receptor to enrich IGF2 into with TMED10-­HA (Fig. 6E). Our findings revealed that these iso­
COPII vesicles, a process that is crucial for the secretion of lated vesicles contained RUSH-­sortilin-­Myc and TMED2 (Fig. 6
IGF2. To further analyze the rescue effects, we analyzed the F and G), but not SURF4 and ERGIC53 (Fig. 6 F and G). This
myotube formation of C2C12 cells by staining the myosin analysis indicates that sortilin and TMED2 reside in the same ves­
heavy chain (MHC), a marker protein of the myotubes. Con­ icles as TMED10, unlike SURF4 and ERGIC53.
sistent with our western blot result, the myotube became much
shorter and thinner after knockdown of TMED10 (Fig. 5 F– Sortilin Regulates TGN Export of IGF2. Our previous analyses
H). After incubation with purified IGF2, myotube formation indicate that ER-­to-­Golgi trafficking of IGF2 and sortilin is
was rescued, and myotubes became longer and thicker (Fig. 5 mediated by TMED10. Sortilin mediates insulin-­dependent
I–K). We then quantified the differentiation index after the glucose transport in myocytes and is crucial for myogenesis (16,
C2C12 differentiation assay. The differentiation index was sig­ 35). Interestingly, we found that knockdown of sortilin significantly
nificantly reduced in TMED10 KD cells and this defect was reduced the efficiency of RUSH-­HA-­IGF2 secretion after biotin
recused by purified IGF2 (Fig. 5L). These analyses indicate that treatment (Fig. 7 A–C), indicating that sortilin plays a crucial role
TMED10 regulates C2C12 differentiation through an auto­ in IGF2 secretion. To study whether sortilin is required for ER-­to-­
crine manner. Taken together, we revealed that TMED10 is Golgi trafficking or TGN-­to-­plasma membrane transport of IGF2,
important for packaging IGF2 into COPII vesicles to deliver we analyzed the RUSH-­HA-­IGF2 trafficking in sortilin KD HeLa
IGF2 from the ER to the Golgi, and this step is critical for cells at different time points after biotin treatment. The percentage
myoblast differentiation. of cells showing juxta-­nuclear localized RUSH-­HA-­IGF2 was

PNAS 2023 Vol. 120 No. 46 e2215285120 https://doi.org/10.1073/pnas.2215285120 7 of 12

 Fig. 5. TMED10 plays important roles in the secretion of IGF2 in
C2C12 cells for muscle stem cell differentiation. (A) C2C12 cells
transfected with control siRNA or siRNA against TMED10 were
incubated with the differentiation medium (DM) with or without
100 ng/mL purified IGF2. After incubation for 3 d, the expression
of the indicated proteins was analyzed by immunoblot. (B)
Relative levels of the indicated proteins after cell differentiation
assay were quantified (mean ± SD; n = 3). The abundance of
myogenin in each experimental group was normalized to the
average abundance of myogenin in all of the three experimental
groups in each biological repeat. (C–K) C2C12 cells transfected
with control siRNA or siRNA against TMED10 were incubated with
the DM with or without 100 ng/mL purified IGF2 for 3 d. The
morphology of the myotube that labeled with MHC was analyzed
by immunofluorescence (Size bar, 200 μm). (L) Quantification of
the differentiation index in each of the indicated experimental
groups (mean ± SD; n = 3). The differentiation index was quantified
by calculating the percentage of the number of nuclei in myosin
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

heavy chain (MHC)-­positive cells versus the total number of nuclei

after the C2C12 differentiation assay. ***P < 0.001; **P < 0.01.

similar in Mock and sortilin KD group after biotin treatment for IGF2 and is recycled to the ER by COPI vesicles. In addition,
20 min (Fig. 7 D, E, G, H, and J), suggesting that knockdown of TMED10 also mediates ER export of sortilin. After reaching TGN,
sortilin did not affect the ER-­to-­Golgi trafficking of RUSH-­HA-­ sortilin regulates TGN-­to-­cell surface delivery of IGF2 (Fig. 7L).
IGF2. RUSH-­HA-­IFG2 showed punctate structures in the cell TMED10 is a member of the p24 family. A null mutation in
periphery in over 60% of cells after biotin treatment for 30 min. TMED10 results in early embryonic lethality in mice (39). The
We hypothesize that these punctate structures are TGN-­derived inactivation of one allele of TMED10 in mice causes dilation of
vesicles enriched with RUSH-­HA-­IGF2. The percentage of cells Golgi cisternae (39) and knockdown of TMED9 in HeLa cells
showing punctate structures of RUSH-­HA-­IGF2 was significantly caused dispersal of the Golgi (40). TMED10 negatively regulates
decreased in sortilin KD group compared to the control group autophagy, and the expression of TMED10 is reduced in Alzheimer’s
(Fig. 7 F, I, and K). These analyses demonstrated that sortilin is disease patients (41). The yeast homologue of TMED10 forms a
important for TGN export but not ER export of IGF2. complex with the yeast homologue of TMED2 and this complex
regulates ER-­to-­Golgi transport of a GPI-­anchored protein, Gas1p
Discussion (42). In mammalian cells, TMED10 is also shown to regulate sur­
face delivery of a GPI-­anchored proteins (43). GTP-­bound form
Secretion of soluble signaling proteins from the producing cells is of Rab21 was shown to interact with TMED10 and regulate local­
tightly related to the downstream signaling pathway in targeted izations of TMED10 at the Golgi (44). Immunoprecipitation
cells. Although a series of factors are identified to regulate the results revealed that both TMED10 and TMED2 showed a pref­
expression of IGF2 such as mTOR, PLD1, and miR-­125b (4, erence to interact with Sec24C and Sec24D, indicating that
36–38), the cargo receptors that regulate the biosynthetic trafficking Sec24C and Sec24D are two Sec24 isoforms involved in
of IGF2 remain largely unknown. Here, we found that two trans­ TMED10 mediated protein trafficking at the ER (45). The
membrane proteins, TMED10 and sortilin, cooperatively mediate majority of p24 family proteins are primarily located within the
IGF2 secretion along the secretory pathway. Based on our study, luminal side of organelle membranes. Their asymmetric nature
we propose that the secretion of IGF2 is achieved by several steps imposes a curvature that is opposite to the curvature needed for
(Fig. 7L). First, the correctly folded ER-­localized pro-­IGF2 is cap­ vesicle budding, thereby changing the physical characteristics of
tured into COPII vesicles by the direct interaction between the membranes (46). It has been shown that the scaffolding function
IGF2112-­140 motif and TMED10 GOLD domain. Second, vesicles of the cargo adaptor Lst1p, the yeast homologue of Sec24, and
containing pro-­IGF2 and TMED10 are delivered to the Golgi the outer COPII coat Sec13p are essential to counter the resist­
apparatus where the pro-­IGF2 receives O-­glycosylation modifica­ ance caused by the p24 proteins and facilitate vesicle formation
tions and cleavage. At the Golgi, TMED10 is disassociated from at the ER (46, 47).

8 of 12 https://doi.org/10.1073/pnas.2215285120 pnas.org
 Fig. 6. TMED10 regulates ER-­to-­Golgi transport of
sortilin. (A) Table showing the list of transmembrane
proteins that are less represented in the vesicle
fraction in the TMED10 KO group compared to
the WT group (average fold change < 0.5). (B) The
vesicle formation assay was performed using
HeLa WT or HeLa TMED10 KO cells. The indicated
proteins were analyzed by immunoblot. (C) WT
or TMED10 KO HeLa cells were transfected with
SBP-­EGFP-­sortilin and Str-­KDEL in the presence
or absence of TMED10-­FLAG. Twenty-­four hours
after transfection, cells were preincubated with
cycloheximide for 2 h. Then, cells were incubated
biotin with cycloheximide for the indicated time
points. The localizations of the indicated proteins
were then analyzed by immunofluorescence (Size
bar, 10 μm). (D) Quantification of the percentage
of cells showing juxta-­nuclear-­located sortilin
(mean ± SD; n = 3; >100 cells counted for each
experiment). ***P < 0.001; ****P < 0.0001; n.s.,
not significant. (E) A diagram demonstrating
the vesicle immunoprecipitation assay. (F and
G) The vesicle formation assay was performed
in untransfected HeLa cells or in cells co-­
transfected with plasmids encoding TMED10-­
HA and plasmids encoding RUSH-­sortilin-­Myc.
Subsequently, vesicles enriched with TMED10-­
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

HA were immunoisolated, and the abundance

of the indicated proteins in the immunoisolated
vesicles (bound on beads) and the abundance of
the indicated proteins in the vesicles that were
not immunoisolated (flow through) were analyzed
by immunoblot. Data shown in panels F and G
are representative example of three biological
repeats.

Upon reaching the Golgi, cargo molecules dissociate from their interactions between UPS cargoes and TMED10 C-­terminal tail (31).
clients. We have previously demonstrated that proteoglycans compete In the UPS pathway, TMED10, triggered by the production of UPS
with SURF4 to interact with Shh at the Golgi, thereby causing cargoes, is found to form a higher-­order mono-­oligomer which sta­
SURF4 to be dissociated from its client (2). P24 family proteins have bilizes the TMED10 protein channel on the ERGIC membranes for
been shown to interact with the remodeled GPI-­APs to enrich them UPS protein translocation (31). If the oligomeric form of TMED10
into COPII vesicles (48, 49). This interaction is pH-­dependent, sug­ forms such a channel, the pore of this channel would be highly hydro­
gesting that they may be dissociated from each other at the Golgi due phobic as the transmembrane residues of TMED10 are predominantly
to pH changes (48). Moreover, p24 proteins have been demonstrated hydrophobic. This would create an energy barrier for the passage of
to retrieve escaped, unremodeled GPI-­anchored proteins from the IL1β and other unconventional cargo proteins. Thus, it remains to be
Golgi, returning them to the ER within COPI vesicles (49). This elucidated how a homo-­oligomer of TMED10 can translocate IL1β
suggests that p24 proteins play a crucial role in monitoring anchor into the luminal side of membranes. In addition, IL1β and other
remodeling to ensure accurate trafficking of GPI-­APs (49). The trans­ unconventional secretory proteins are found to be efficiently secreted
membrane domain of TMED2 (but not TMED10) has been found upon infection-­induced permeabilization of the plasma membrane of
to interact specifically with a sphingomyelin, SM18. This interaction immune cells (52), suggesting a substantial fraction of these cargo
facilitates efficient retrograde COPI-­dependent trafficking and is molecules are located in the cytoplasm of immune cells. Further anal­
implicated to modulate the equilibrium between monomeric and ysis is needed to determine the proportion of IL1β that resides within
oligomeric states of TMED2 (50). An interesting future direction is the cytoplasm versus the luminal side of the ERGIC. It remains
to investigate how TMED10 is dissociated from IGF2 at the Golgi unclear whether TMED10 forms mono-­oligomer or hetero-­oligomer
and whether ER retrieval of TMED10 depends on the TMED2-­SM18 with other TMED proteins to regulate IGF2 trafficking. Deleting the
interaction. GOLD domain abolished TMED10 mono-­oligomerization, indicat­
Intriguingly, TMED10 is also shown to function as a protein chan­ ing that the integrity of the GOLD domain is crucial for the
nel to mediate the unconventional protein secretion (UPS) of a group mono-­oligomerization of TMED10 (31). The direct interaction
of leaderless proteins including IL1β, IL-­1α, HSPB5, Tau, and between IGF2 and TMED10 GOLD domain may interfere with
Annexin A1 (51). Unlike the TMED10-­mediated conventional secre­ TMED10 mono-­oligomerization. In addition, the quantitative mass
tory pathway, TMED10-­channeled UPS requires direct or indirect spectrometry analysis indicates that TMED1, 2, 3, 7, and 9 showed

PNAS 2023 Vol. 120 No. 46 e2215285120 https://doi.org/10.1073/pnas.2215285120 9 of 12

Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

Fig. 7. Sortilin regulates TGN export of IGF2. (A) HeLa cells were transfected with control siRNA or siRNA against sortilin. Day 2 after transfection, the level of the
indicated proteins in cell lysates was analyzed by immunoblot. (B) HeLa cells were transfected with control siRNA or siRNA against sortilin. Day 1 after transfection,
cells were retransfected with plasmids encoding Str-­KDEL and RUSH-­HA-­IGF2. On day 3 after knockdown, cells were preincubated with cycloheximide for 2 h. Then,
cells were incubated with biotin and cycloheximide for 2 h. After biotin incubation, the level of RUSH-­HA-­IGF2 in the medium and in cell lysates was analyzed by
immunoblot. (C) Quantification of the abundance of secreted IGF2 normalized to the abundance detected in the cell lysate group (mean ± SD; n = 3). The value
in each experimental group was normalized to the average value in the Mock group and the sortilin KD group in each biological repeat. (D–I″) HeLa cells were
transfected with control siRNA or siRNA against sortilin. Day 1 after transfection, cells were retransfected with plasmids encoding Str-­KDEL and RUSH-­HA-­IGF2.
On day 3 after knockdown, cells were preincubated with cycloheximide for 2 h. Then, cells were incubated with cycloheximide and biotin for the indicated time,
and the localization of RUSH-­HA-­IGF2 was analyzed (Size bar, 10 μm). The magnified view of the indicated area in panels F and I is shown in panels F′, F″, I′, and
I″. (J) Quantification of the percentage of cells showing juxtanuclear-­located RUSH-­HA-­IGF2 (mean ± SD; n = 3; >100 cells counted in each experimental group).
(K) Quantification of the percentage of cells showing punctate patterns of RUSH-­HA-­IGF2 (mean ± SD; n = 3; >100 cells counted in each experimental group). (L)
The proposed model demonstrating the dual functions of TMED10 in mediating IGF2 trafficking along the secretory pathway: 1) the GOLD domain of TMED10
recognizes the 112-­140 residues of IGF2 to enrich IGF2 into COPII vesicles; 2) TMED10 also regulates ER export of sortilin, which is important for TGN-­to-­plasma
membrane trafficking of IGF2. ***P < 0.001; n.s., not significant.

a large decrease in TMED10 KO vesicles (Fig. 6A). Thus, we hypoth­ TMED10 GOLD domain and the IGF2112-­140 motif, suggesting
esized that TMED10 and other p24 family proteins form hetero- that TMED10 directly mediates the ER export of IGF2.
­oligomers to package IGF2 into COPII vesicles. Sortilin is a single-­pass transmembrane protein belonging to
In addition to the selective capture mechanism, bulk flow is the vacuolar protein sorting 10 protein (Vps10p) family (12).
another approach that exports soluble or membrane-­associated pro­ Although functional roles of sortilin have been extensively studied,
teins from the ER (10). Export by bulk flow does not rely on cargo the molecular mechanisms mediating the biosynthetic trafficking
receptors or export motifs on cargoes. Instead, proteins exported by of sortilin remain largely unclear. Here, we revealed that TMED10
bulk flow were packaged into COPII vesicles by default. Utilizing functions as a cargo receptor that mediates the ER-­to-­Golgi trans­
the RUSH assay, we found that RUSH-­IGF298-­180-­HA without the port of sortilin. In addition, we demonstrate that sortilin is impor­
112-­140 aa motif showed a kinetic delay in ER-­to-­Golgi transport tant for the post-­Golgi trafficking of IGF2. These findings suggest
(Fig. 3 P–Y). In contrast, fusing the SBP-­EGFP tag with IGF2112-­140 that TMED10 also indirectly mediates TGN export of IGF2 by
motif efficiently brings SBP-­EGFP to the Golgi (SI Appendix, regulating the ER-­to-­Golgi trafficking of sortilin.
Fig. S6), indicating that bulk flow is not an efficient approach in In summary, our study provides insights into the molecular
mediating ER-­to-­Golgi trafficking of IGF2. Cross-­linking and pep­ machinery that mediates the trafficking of IGF2 along the secre­
tide binding experiments revealed a direct interaction between tory pathway to perform its physiological functions. Dysreg­ulation

10 of 12 https://doi.org/10.1073/pnas.2215285120 pnas.org
 of IGF2 activities is a candidate risk factor for tumo­rigenesis and Sample Preparation for Label-­Free Quantitative MS Analysis. This proce-
is related to multiple disorders, such as Beckwith–Wiedemann dure was described in SI Appendix.
syndrome, Silver–Russell syndrome, and Doege–Potter syndrome
(53–55). The uncovered cellular factors and protein interactions Data, Materials, and Software Availability. All study data are included in the

that are important for the secretion of IGF2 provide therapeutic tar­ article and/or supporting information.
gets to down-­regulate IGF2 signaling by blocking IGF2 secretion.
ACKNOWLEDGMENTS. We thank Dr. Randy Schekman (University of
California, Berkeley) for providing antibodies against SEC22B and ERGIC53.
Materials and Methods
This work was supported by grants from the National Natural Science
Constructs, Reagents, Cell Culture, Immunofluorescence, and Transfection. Foundation of China (NSFC32070699 to Y.G. and 81874306 to Z.Y.). This
Cell lines, cDNAs, siRNAs, antibodies, cell culture, immunofluorescence, and trans- work was also supported by the Hong Kong Research Grants Council Grants
fection were described in SI Appendix. (16104020, 16102921, 16103622, 16103319, C4002-­20W, C6012-­22G,
and T13-­602/21-­N to Y.G. and 15304020, 15306421, 15304022, R5013-­19F,
Immunoprecipitation, Protein Purification, RUSH Assay, Binding Assay,
R4005-­18, and C4002-­20WF to Z.Y.). In addition, this project was supported
and Vesicle Formation Assay. Immunoprecipitation, protein purification,
in part by the Innovation and Technology Commission (ITCPD/17-­9) to Y.G.
RUSH assay, binding assay, and vesicle formation were performed as described
previously (21, 56–58). Vesicle immunoprecipitation assay was performed as
described previously (56).
Author affiliations: aDivision of Life Science and State Key Laboratory of Molecular
Neuroscience, The Hong Kong University of Science and Technology, Hong Kong, China;
Muscle Stem Cell Differentiation Assay. Undifferentiated C2C12 cells were b
State Key Laboratory of Chemical Biology and Drug Discovery, Research Institute for
cultured with DMEM containing 20% FBS and 1% penicillin–streptomycin mix. Future Food, Research Centre for Chinese Medicine Innovation, and Department of
To induce differentiation, C2C12 cells transfected with control siRNA or siRNA Applied Biology and Chemical Technology, The Hong Kong Polytechnic University, Hong
Kong, China; cState Key Laboratory of Chinese Medicine and Molecular Pharmacology
against TMED10 were incubated with DMEM containing 2% horse serum and 1% (Incubation) and Shenzhen Key Laboratory of Food Biological Safety Control, Hong Kong
penicillin–streptomycin in the presence or absence of 100 ng/mL purified IGF2 Polytechnic University, Shenzhen Research Institute, Shenzhen 518057, China; dHong
Kong University of Science and Technology, Shenzhen Research Institute, Shenzhen
(R&D Systems, Catalog number: 792-­MG) for 3 d. Then, the cells were analyzed 518057, China; and eThrust of Bioscience and Biomedical Engineering, Hong Kong
by immunoblot or immunofluorescence. University of Science and Technology, Guangzhou 511453, China

1. W. Chao, P. A. D’Amore, IGF2: Epigenetic regulation and role in development and disease. Cytokine 25. X. Tang, T. Wang, Y. Guo, Export of polybasic motif-­containing secretory proteins BMP8A and SFRP1
Growth Factor. Rev. 19, 111–120 (2008). from the endoplasmic reticulum is regulated by surfeit locus protein 4. J. Biol. Chem. 298, 102687
2. J. R. Florini, D. Z. Ewton, S. L. Roof, Insulin-­like growth factor-­I stimulates terminal myogenic (2022).
differentiation by induction of myogenin gene expression. Mol. Endocrinol. 5, 718–724 26. Y. Huang et al., An in vitro vesicle formation assay reveals cargo clients and factors that mediate
(1991). vesicular trafficking. Proc. Natl. Acad. Sci. U.S.A. 118, e2101287118 (2021).
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

3. J. R. Florini et al., “Spontaneous” differentiation of skeletal myoblasts is dependent upon autocrine 27. M. Marzioch et al., Erp1p and Erp2p, partners for Emp24p and Erv25p in a yeast p24 complex. Mol.
secretion of insulin-­like growth factor-­II. J. Biol. Chem. 266, 15917–15923 (1991). Biol. Cell 10, 1923–1938 (1999).
4. Y. Ge, Y. Sun, J. Chen, IGF-­II is regulated by microRNA-­125b in skeletal myogenesis. J. Cell Biol. 28. D. Gommel et al., p24 and p23, the major transmembrane proteins of COPI-­coated transport
192, 69–81 (2011). vesicles, form hetero-­oligomeric complexes and cycle between the organelles of the early secretory
5. S. J. Duguay et al., Post-­translational processing of the insulin-­like growth factor-­2 precursor. pathway. FEBS Lett. 447, 179–185 (1999).
Analysis of O-­glycosylation and endoproteolysis. J. Biol. Chem. 273, 18443–18451 (1998). 29. J. Fullekrug et al., Localization and recycling of gp27 (hp24gamma3): Complex formation with
6. J. Dancourt, C. Barlowe, Protein sorting receptors in the early secretory pathway. Annu. Rev. other p24 family members. Mol. Biol. Cell 10, 1939–1955 (1999).
Biochem. 79, 777–802 (2010). 30. N. Pastor-­Cantizano, J. C. Montesinos, C. Bernat-­Silvestre, M. J. Marcote, F. Aniento, p24 family proteins: Key
7. L. Lyu et al., Selective export of autotaxin from the endoplasmic reticulum. J. Biol. Chem. 292, players in the regulation of trafficking along the secretory pathway. Protoplasma 253, 967–985 (2016).
7011–7022 (2017). 31. M. Zhang et al., A translocation pathway for vesicle-­mediated unconventional protein secretion. Cell
8. T. Buechling, V. Chaudhary, K. Spirohn, M. Weiss, M. Boutros, p24 proteins are required for secretion 181, 637–652.e615 (2020).
of Wnt ligands. EMBO Rep. 12, 1265–1272 (2011). 32. N. Nakamura et al., Identification of potential regulatory elements for the transport of Emp24p. Mol.
9. X. Li et al., Drosophila p24 and Sec22 regulate Wingless trafficking in the early secretory pathway. Biol. Cell 9, 3493–3503 (1998).
Biochem. Biophys. Res. Commun. 463, 483–489 (2015). 33. R. Aber, W. Chan, S. Mugisha, L. A. Jerome-­Majewska, Transmembrane emp24 domain proteins in
10. C. Barlowe, A. Helenius, Cargo capture and bulk flow in the early secretory pathway. Annu. Rev. Cell development and disease. Genet. Res. (Camb) 101, e14 (2019).
Dev. Biol. 32, 197–222 (2016). 34. V. Anantharaman, L. Aravind, The GOLD domain, a novel protein module involved in Golgi function
11. Y. Guo, D. W. Sirkis, R. Schekman, Protein sorting at the trans-­Golgi network. Annu. Rev. Cell Dev. and secretion. Genome Biol. 3, research0023 (2002).
Biol. 30, 169–206 (2014). 35. G. Huang et al., Insulin responsiveness of glucose transporter 4 in 3T3-­L1 cells depends on the
12. S. Ouyang, B. Jia, W. Xie, J. Yang, Y. Lv, Mechanism underlying the regulation of sortilin expression presence of sortilin. Mol. Biol. Cell 24, 3115–3122 (2013).
and its trafficking function. J. Cell Physiol. 235, 8958–8971 (2020). 36. E. Erbay, I. H. Park, P. D. Nuzzi, C. J. Schoenherr, J. Chen, IGF-­II transcription in skeletal myogenesis is
13. D. M. Conlon, Role of sortilin in lipid metabolism. Curr. Opin. Lipidol. 30, 198–204 (2019). controlled by mTOR and nutrients. J. Cell Biol. 163, 931–936 (2003).
14. S. Y. Xu, J. Jiang, A. Pan, C. Yan, X. X. Yan, Sortilin: A new player in dementia and Alzheimer-­type 37. M. S. Yoon, J. Chen, PLD regulates myoblast differentiation through the mTOR-­IGF2 pathway. J. Cell
neuropathology. Biochem. Cell Biol. 96, 491–497 (2018). Sci. 121, 282–289 (2008).
15. H. Talbot et al., Regulatory roles of Sortilin and SorLA in immune-­related processes. Front. 38. A. Polesskaya et al., Lin-­28 binds IGF-­2 mRNA and participates in skeletal myogenesis by increasing
Pharmacol. 9, 1507 (2018). translation efficiency. Genes. Dev. 21, 1125–1138 (2007).
16. M. Ariga, T. Nedachi, H. Katagiri, M. Kanzaki, Functional role of sortilin in myogenesis and 39. A. Denzel et al., The p24 family member p23 is required for early embryonic development. Curr.
development of insulin-­responsive glucose transport system in C2C12 myocytes. J. Biol. Chem. Biol. 10, 55–58 (2000).
283, 10208–10220 (2008). 40. S. Mitrovic, H. Ben-­Tekaya, E. Koegler, J. Gruenberg, H. P. Hauri, The cargo receptors Surf4,
17. T. M. Teslovich et al., Biological, clinical and population relevance of 95 loci for blood lipids. Nature endoplasmic reticulum-­Golgi intermediate compartment (ERGIC)-­53, and p25 are required to
466, 707–713 (2010). maintain the architecture of ERGIC and Golgi. Mol. Biol. Cell 19, 1976–1990 (2008).
18. N. J. Samani et al., Genomewide association analysis of coronary artery disease. N. Engl. J. Med. 41. J. H. Shin et al., Down-­regulated TMED10 in Alzheimer disease induces autophagy via ATG4B
357, 443–453 (2007). activation. Autophagy 15, 1495–1505 (2019).
19. K. Musunuru et al., From noncoding variant to phenotype via SORT1 at the 1p13 cholesterol locus. 42. W. J. Belden, C. Barlowe, Erv25p, a component of COPII-­coated vesicles, forms a complex with
Nature 466, 714–719 (2010). Emp24p that is required for efficient endoplasmic reticulum to Golgi transport. J. Biol. Chem. 271,
20. M. Kjolby et al., Sort1, encoded by the cardiovascular risk locus 1p13.3, is a regulator of hepatic 26939–26946 (1996).
lipoprotein export. Cell Metab. 12, 213–223 (2010). 43. S. Takida, Y. Maeda, T. Kinoshita, Mammalian GPI-­anchored proteins require p24 proteins for their
21. X. Tang et al., A SURF4-­to-­proteoglycan relay mechanism that mediates the sorting and efficient transport from the ER to the plasma membrane. Biochem. J. 409, 555–562 (2008).
secretion of a tagged variant of sonic hedgehog. Proc. Natl. Acad. Sci. U.S.A. 119, e2113991119 44. T. Del Olmo et al., RAB21 interacts with TMED10 and modulates its localization and abundance. Biol.
(2022). Open 8, bio045336 (2019).
22. G. Boncompain et al., Synchronization of secretory protein traffic in populations of cells. Nat. 45. C. Bonnon, M. W. Wendeler, J. P. Paccaud, H. P. Hauri, Selective export of human GPI-­anchored
Methods 9, 493–498 (2012). proteins from the endoplasmic reticulum. J. Cell Sci. 123, 1705–1715 (2010).
23. Y. Mao et al., The exocyst functions in niche cells to promote germline stem cell differentiation by 46. A. Copic, C. F. Latham, M. A. Horlbeck, J. G. D’Arcangelo, E. A. Miller, ER cargo properties specify a
directly controlling EGFR membrane trafficking. Development 146, dev174615 (2019). requirement for COPII coat rigidity mediated by Sec13p. Science 335, 1359–1362 (2012).
24. L. Niu et al., Atlastin-­mediated membrane tethering is critical for cargo mobility and exit from the 47. J. G. D’Arcangelo et al., Traffic of p24 proteins and COPII coat composition mutually influence
endoplasmic reticulum. Proc. Natl. Acad. Sci. U.S.A. 116, 14029–14038 (2019). membrane scaffolding. Curr. Biol. 25, 1296–1305 (2015).

PNAS 2023 Vol. 120 No. 46 e2215285120 https://doi.org/10.1073/pnas.2215285120 11 of 12

 48. M. Fujita et al., Sorting of GPI-­anchored proteins into ER exit sites by p24 proteins is dependent on 54. A. Y. Kalebi, M. J. Hale, M. L. Wong, T. Hoffman, J. Murray, Surgically cured hypoglycemia secondary
remodeled GPI. J. Cell Biol. 194, 61–75 (2011). to pleural solitary fibrous tumour: Case report and update review on the Doege–Potter syndrome.
49. G. A. Castillon et al., The yeast p24 complex regulates GPI-­anchored protein transport and quality J. Cardiothorac. Surg. 4, 45 (2009).
control by monitoring anchor remodeling. Mol. Biol. Cell 22, 2924–2936 (2011). 55. C. Gicquel et al., Epimutation of the telomeric imprinting center region on chromosome 11p15 in
50. F. X. Contreras et al., Molecular recognition of a single sphingolipid species by a protein’s Silver–Russell syndrome. Nat. Genet. 37, 1003–1007 (2005).
transmembrane domain. Nature 481, 525–529 (2012). 56. X. Tang et al., Molecular mechanisms that regulate export of the planar cell-­polarity protein
51. T. A. Nguyen, J. Debnath, Unconventional secretion: Cargo channeling by TMED10. Cell Res. 30, Frizzled-­6 out of the endoplasmic reticulum. J. Biol. Chem. 295, 8972–8987 (2020), 10.1074/jbc.
713–714 (2020). RA120.012835.
52. C. L. Evavold, J. C. Kagan, Diverse control mechanisms of the interleukin-­1 cytokine family. Front. 57. Y. Guo, G. Zanetti, R. Schekman, A novel GTP-­binding protein-­adaptor protein complex responsible
Cell Dev. Biol. 10, 910983 (2022). for export of Vangl2 from the trans Golgi network. eLlife 2, e00160 (2013).
53. A. J. Ping et al., Genetic linkage of Beckwith–Wiedemann syndrome to 11p15. Am. J. Hum. Genet. 58. X. Tang, F. Yang, Y. Guo, Cell-­free reconstitution of the packaging of cargo proteins into vesicles at
44, 720–723 (1989). the trans Golgi Network. Bio. Protoc. 10, e3537 (2020).
Downloaded from https://www.pnas.org by 201.227.133.86 on November 15, 2023 from IP address 201.227.133.86.

12 of 12 https://doi.org/10.1073/pnas.2215285120 pnas.org