Abstract
Tumor-derived exosomes carry programmed death-ligand 1 (PD-L1), which
binds programmed cell death protein 1 (PD-1) on T cells, suppressing
immune responses locally and systemically. However, the mechanisms
governing exosomal PD-L1 sorting and secretion remain elusive. Here, we
identify Munc13-4 as a crucial regulator of this process. Deletion of
Munc13-4 in breast tumors enhances T cell-mediated anti-tumor immunity,
suppresses tumor growth, and improves the efficacy of immune checkpoint
inhibitors. Mechanistically, Munc13-4 collaborates with hepatocyte
growth factor-regulated tyrosine kinase substrate (HRS), Rab27, and
SNAREs to facilitate PD-L1 sorting and secretion via exosomes.
Cryogenic electron microscopy (cryo-EM) analysis of the Munc13-4–Rab27a
complex provide structural insights into exosome secretion.
Importantly, PD-L1 sorting relies on a ternary complex composed of
Munc13-4, PD-L1 and HRS, which is regulated by interferon gamma (IFNγ)
signaling. A designed peptide that disrupts Munc13-4–PD-L1 interaction
impedes PD-L1 sorting, enhances antitumor immunity, and suppresses
tumor growth, highlighting the therapeutic potential of targeting this
pathway.
Subject terms: Membrane trafficking, Cancer
__________________________________________________________________
Munc13-4 is known to regulate a Ca^2+- stimulated exosome release
pathway. Here, the authors discover that Munc13-4 promotes breast
cancer immune evasion through regulation of PD-L1 sorting in exosomes.
Introduction
Tumor cells evade immune surveillance by increasing the surface
expression of PD-L1, which interacts with the PD-1 receptor on T cells
to trigger the immune checkpoint response and leading to T cell
inhibition^[44]1,[45]2. Inhibitors targeting the PD-1/PD-L1 have shown
promise in cancer therapy by restoring T cell function and enhancing
anti-tumor immunity^[46]3. Nonetheless, a significant proportion of
patients do not respond to anti-PD-L1/PD-1 therapies^[47]3,[48]4. A
primary reason for this resistance is the secretion of PD-L1 into the
bloodstream via extracellular vesicles (EVs), particular exosomes,
where it can disrupt immune function remotely^[49]5–[50]7. Recent
studies have demonstrated that genetic blockade of exosomal PD-L1
biogenesis and/or secretion not only suppresses local tumor growth but
also elicits a durable systemic anti-tumor immune response^[51]8.
Therefore, the role of exosomal PD-L1 in immune modulation highlights
the urgent need for targeted strategies to inhibit exosomal PD-L1
biogenesis and secretion, which could significantly improve the
efficacy of cancer immunotherapies.
The secretion of tumor-derived exosomes begins with the formation of
early endosomes, which mature into multivesicular bodies (MVBs)
containing intraluminal vesicles (ILVs). MVBs fuse with the plasma
membrane, releasing ILVs as exosomes into the extracellular
space^[52]9. Endosomal sorting complexes required for transport
(ESCRTs) are critical for cargo sorting within the endosomal system.
HRS, a key component of the ESCRT machinery, mediates cargo recognition
and sorting into MVBs^[53]10. In addition, the Rab family of small
GTPases (Rabs) is essential for membrane trafficking, with Rab27
specifically controlling various steps of the exosome secretion
pathway, particularly the docking of MVBs to the plasma
membrane^[54]11. Moreover, the SNARE machinery mediates the fusion of
MVBs with the plasma membrane, with syntaxin-4, SNAP-23, and VAMP-7
forming the SNARE complex to drive the secretion of exosomes in various
tumor cells^[55]12. Despite these advances, the mechanisms governing
PD-L1 sorting onto exosomes and their secretion remain elusive.
Understanding these processes is essential for developing targeted
molecules that inhibit the PD-L1 secretion pathway, and it is urgent to
identify and investigate key players in the endosomal sorting pathway
that mediate PD-L1 recognition and sorting onto exosomes.
In this study, we identified Munc13-4, a member of the Munc13 protein
family known for its regulatory role in membrane trafficking, as
upregulated in various tumor tissues, where it mediates tumor immune
evasion by regulating exosomal sorting and secretion of PD-L1. Deleting
Munc13-4 in breast tumors significantly enhances T cell-mediated
anti-tumor immunity, suppresses tumor growth, and boosts the efficacy
of immune checkpoint inhibitors. We elucidated a coherent mechanism
whereby Munc13-4 collaborates with HRS, Rab27, and SNAREs to regulate
PD-L1 sorting and secretion via exosomes. Notably, we discovered a
critical function of Munc13-4 in PD-L1 sorting that depends on its
direct interaction with PD-L1; disrupting this interaction with a
specifically designed peptide markedly impaired PD-L1 sorting, leading
to enhanced T cell-mediated anti-tumor responses in vivo. These
findings position Munc13-4 as a promising therapeutic target for
boosting immune responses against tumors.
Results
Munc13-4 deficiency in tumor cells inhibits tumor growth in an
immunity-dependent way
The Munc13 protein family functions as critical regulators of vesicle
trafficking and exocytosis across various cell types. While Munc13-1,
Munc13-2, and Munc13-3 are involved in the exocytosis of synaptic
vesicles and dense-core vesicles in neurons and neuroendocrine
cells^[56]13–[57]15, Munc13-4 has specialized roles in cytotoxic
granule exocytosis in immune cells^[58]16–[59]19 and has recently been
implicated in exosome secretion in tumor cells^[60]20. To explore the
role of Munc13-4 in tumor progression, we assessed its expression in
tumor and adjacent normal tissues using TIMER2.0
(cistrome.shinyapps.io/timer)^[61]21,[62]22, a platform that analyzes
genomic data from The Cancer Genome Atlas (TCGA). Our analysis revealed
significant upregulation of Munc13-4 across various tumor types
(Supplementary Fig. [63]1A). Immunohistochemical staining of tumor and
adjacent normal tissues on tissue microarrays demonstrated increased
Munc13-4 expression in tumors, including breast cancer, thyroid cancer,
cholangiocarcinoma, gastrointestinal stromal tumor, pancreatic cancer,
and hepatocellular carcinoma (Fig. [64]1A and Supplementary
Fig. [65]1B–E), suggesting a crucial role for Munc13-4 in tumor
progression. Given the high global incidence of breast cancer, we
focused on investigating the specific role of Munc13-4 in breast
cancer.
Fig. 1. Munc13-4 deficiency in tumor cells inhibits tumor growth in an
immunity-dependent way.
[66]Fig. 1
[67]Open in a new tab
A Representative immunohistochemical images showing Munc13-4 expression
in breast cancer (n = 30 patients), thyroid cancer (n = 6 patients),
cholangiocarcinoma (n = 5 patients), gastrointestinal stromal tumors
(n = 6 patients), pancreatic cancer (n = 6 patients), and
hepatocellular carcinoma (n = 9 patients) tissues, along with their
corresponding adjacent normal tissues, assessed using a multi-organ
carcinoma tissue array. Scale bar, 500 μm. B–D Tumor growth in BALB/c
mice inoculated with wild-type (WT), control, or Munc13-4 knockout (KO)
4T1 cells (n = 9 mice). B Schematic of experimental design. C Tumor
growth curves following mammary gland inoculation. D Percentage change
in tumor volume, normalized to WT group. E–G Tumor growth in BALB/Nude
mice inoculated with WT, control, or Munc13-4 KO 4T1 cells (n = 8
mice). E Schematic of experimental design. F Tumor growth curves
following mammary gland inoculation. G Percentage change in tumor
volume, normalized to WT group. H–J Tumor growth in NOD/SCID mice
inoculated with WT, control, or Munc13-4 KO 4T1 cells (n = 6 mice). H
Schematic of experimental design. I Tumor growth curves following
mammary gland inoculation. J Percentage change in tumor volume,
normalized to WT group. Data are presented as means ± SEM, and p-values
were calculated by one-way ANOVA with multiple comparisons (C, F, I),
ns not significant. Source data are provided as a [68]Source data file.
Using the CRISPR-Cas9 system, we generated Munc13-4 knockout 4T1 murine
mammary carcinoma cells, with control cells infected with a lentivirus
carrying Cas9 without sgRNA (Supplementary Fig. [69]2A). The deletion
of Munc13-4 did not influence the proliferation of 4T1 cells in vitro
(Supplementary Fig. [70]2B). We then conducted in vivo studies by
creating orthotopic mouse models of breast cancer using wild-type (WT),
control and Munc13-4 knockout 4T1 cells (Fig. [71]1B). Mice inoculated
with Munc13-4 knockout 4T1 cells showed a significant delay in tumor
growth compared to those implanted with WT or control cells
(Fig. [72]1C and Supplementary Fig. [73]2C, D), with average tumor
volume in the knockout group stabilizing at approximately 17.17% of
that in the WT group (Fig. [74]1D). These data demonstrate that
Munc13-4 knockout substantially impairs the oncogenic potential of 4T1
cells, indicating a pivotal role for Munc13-4 in breast tumor
progression.
Next, we identified differentially expressed proteins in the proteomes
of control and Munc13-4 knockout 4T1 cells, and conducted enrichment
analysis based on the KEGG database to pinpoint relevant cellular
processes and organismal systems affected by Munc13-4 knockout. KEGG
enrichment analysis suggests significant involvement in transport and
catabolic pathways, along with a critical association with the immune
system (Supplementary Fig. [75]2E). Given that immune evasion is a
hallmark of cancer, we explored the relationship between Munc13-4 and
immune evasion by assessing the oncogenicity of Munc13-4 knockout 4T1
cells in immunodeficient mouse models. In BALB/Nude mice lacking T
cells, the difference in tumor growth between those with Munc13-4
knockout cells and WT or control cells became less pronounced
(Fig. [76]1E, F and Supplementary Fig. [77]2F, G), with Munc13-4
knockout tumors approaching 60.19% of the size of WT tumors
(Fig. [78]1G). Moreover, in NOD/SCID mice with severe combined
immunodeficiency, tumors in the Munc13-4 knockout group grew comparably
to those in the WT group (Fig. [79]1H, I and Supplementary Fig. [80]2H,
I), with the Munc13-4 knockout tumors reaching 83.67% of the size of
the WT tumors (Fig. [81]1J). These results suggest that the role of
Munc13-4 in breast tumor progression is closely linked to its capacity
to modulate immune responses within the tumor microenvironment.
Munc13-4 deficiency in tumor cells enhances T cell infiltration and
activation
We next explored whether deletion of Munc13-4 in breast tumor cells
influences the quantity and activity of T cells within tumors, spleens,
and lymph nodes of tumor-bearing mice inoculated with either control or
Munc13-4 knockout 4T1 cells. Flow cytometry showed a significant
increase in the infiltration of both CD4^+ and CD8^+ T cells within the
tumors of mice implanted with Munc13-4 knockout 4T1 cells (Fig. [82]2A
and Supplementary Fig. [83]3A), which was corroborated by
immunofluorescent staining (Supplementary Fig. [84]3B). In addition,
the populations of both CD4^+ and CD8^+ cells in the spleens and lymph
nodes of these mice were notably elevated compared to the control group
(Fig. [85]2B, C). These results indicate that the deletion of Munc13-4
in tumor cells enhances T cell infiltration in the tumor, spleen, and
lymph nodes.
Fig. 2. Munc13-4 deficiency in tumor cells enhances T cell infiltration and
activation.
[86]Fig. 2
[87]Open in a new tab
Flow cytometric quantification of the percentage of CD45^+CD3^+CD4^+
and CD45^+CD3^+CD8^+ T cells among total cells in the tumors (A),
spleens (B), and draining lymph nodes (C) of BALB/c mice (n = 5 mice),
21 days after mammary gland injection with 3 × 10^5 control or Munc13-4
KO 4T1 cells per mouse. Quantification of the percentage of granzyme
B^+ (GzmB^+) (D), Ki67^+ (E), and IFNγ^+ (F) cells among
CD45^+CD3^+CD4^+ and CD45^+CD3^+CD8^+ T cells within tumors from
orthotopic mouse models of breast cancer generated by control or
Munc13-4 KO 4T1 cells (n = 5 mice). Quantification of the percentage of
granzyme B^+ (G), Ki67^+ (H), and IFNγ^+ (I) cells among
CD45^+CD3^+CD4^+ and CD45^+CD3^+CD8^+ T cells within spleens from
orthotopic mouse models of breast cancer generated by control or
Munc13-4 KO 4T1 cells (n = 5 mice). Quantification of the percentage of
granzyme B^+ (J), Ki67^+ (K), and IFNγ^+ (L) cells among
CD45^+CD3^+CD4^+ and CD45^+CD3^+CD8^+ T cells within the draining lymph
nodes from orthotopic mouse models of breast cancer generated by
control or Munc13-4 KO 4T1 cells (n = 5 mice). Box plots show the
median (center line), interquartile range (box), minima and maxima
(whiskers), and all individual data points (dots). All p-values were
calculated by two-tailed Multiple t-tests. Source data are provided as
a [88]Source data file.
We further assessed T cell activation markers, including the cytotoxic
molecule granzyme B, the proliferation marker Ki67, and the cytokine
interferon gamma (IFNγ), in CD4^+ and CD8^+ T cell populations from
tumors, spleens, and lymph nodes of tumor-bearing mice. Mice implanted
with Munc13-4 knockout 4T1 cells exhibited a significant increase in
the expression of granzyme B, Ki67, and IFNγ in both CD4^+ and CD8^+ T
cells across all examined tissues compared to those inoculated with
control 4T1 cells (Fig. [89]2D–L and Supplementary Fig. [90]3C–E).
Collectively, these results demonstrate that Munc13-4 deficiency in
breast tumor cells augments T cell infiltration and activation, thereby
enhancing the systemic T cell-mediated immune response.
Facilitating PD-L1 secretion by Munc13-4 suppresses anti-tumor efficacy of T
cells
PD-L1 binding to PD-1 on T cells is a crucial mechanism for tumor
evasion of immune surveillance. To explore whether reduced
tumor-induced immunosuppression in Munc13-4-deficient models was linked
to PD-L1 changes, we examined the effect of Munc13-4 knockout on PD-L1
expression by western blot. Our results showed no impact on PD-L1
protein levels in 4T1 and SUM159 breast tumor cells (Fig. [91]3A).
Fig. 3. Facilitating PD-L1 secretion by Munc13-4 suppresses anti-tumor
efficacy of T cells.
[92]Fig. 3
[93]Open in a new tab
A Western blot (WB) analysis of total PD-L1 level in control and
Munc13-4 KO SUM159 or 4T1 cells (n = 3 biological replicates). B
Representative TEM images of EVs secreted by control and Munc13-4 KO
SUM159 or 4T1 cells (n = 3 biological replicates). Scale bar, 50 nm. C
Quantification of exosomes secreted by equal numbers of control and
Munc13-4 KO SUM159 (left) or 4T1 (right) cells through NTA (n = 3
biological replicates). D Schematic of experimental design for (E, F).
WB analysis of PD-L1, Alix, CD63, and CD81 in EVs secreted by equal
numbers of control and Munc13-4 KO SUM159 (E) or 4T1 (F) cell,
collected from factions 1–6 (F1–6) (n = 3 biological replicates). WB
analysis of PD-L1, Alix, CD63, and CD81 abundance on equal numbers of
exosomes secreted by control and Munc13-4 KO SUM159 cells (G) and
corresponding quantification of blot band intensities (H) (n = 4
biological replicates). WB analysis of PD-L1, Alix, CD63, and CD81
abundance on equal numbers of exosomes secreted by control and Munc13-4
KO 4T1 cells (I) and corresponding quantification of blot band
intensities (J) (n = 3 biological replicates). K Schematic of
experimental design for (L). L Quantification of killing efficiency
against control and Munc13-4 KO 4T1 cells (n = 3 biological
replicates). M Schematic of experimental design for (N). N Tumor growth
curves following mammary gland inoculation of control or Munc13-4 KO
4T1 cells, with subsequent injection of PBS or the indicated exosomes
(n = 6 mice). In (A), for each cell line, all samples were run on the
same gel; in (E, F, G, I), samples were run on the same gel. Data are
presented as means ± SEM (C, H, J, N), and p-values were calculated by
unpaired two-tailed t-test (C), two-way ANOVA (H, J) with multiple
comparisons, paired two-tailed t-test (L), and one-way ANOVA with
multiple comparisons (N). Source data are provided as a [94]Source data
file.
PD-L1 transport to the plasma membrane and sorting onto exosomes
enables its interaction with PD-1 on T cells, thus inhibiting T
cell-mediated anti-tumor immunity. Proteomics analysis suggests the
involvement of Munc13-4 in transport and catabolic pathways
(Supplementary Fig. [95]2E), we thus examined the effect of Munc13-4
knockout on the transport of PD-L1 to the plasma membrane and to
exosomes. Flow cytometry analyses revealed that no change in PD-L1 on
the plasma membrane was observed after Munc13-4 knockout (Supplementary
Fig. [96]4A, B). Transmission electron microscopy (TEM), nanoparticle
tracking analysis (NTA) and tunable resistive pulse sensing (TRPS)
showed that EVs isolated from control and Munc13-4 knockout cell
culture supernatants displayed exosome-like morphology (Fig. [97]3B)
and were within the 50–200 nm size range characteristic of exosomes
(Supplementary Fig. [98]4C–F). However, NTA and TRPS indicated a
significant decrease in the total number of exosomes secreted by
Munc13-4 knockout cells compared to control cells (Fig. [99]3C and
Supplementary Fig. [100]4C–F). Western analysis showed a marked
reduction in PD-L1 and exosome marker proteins (Alix, CD63, CD81) in
EVs from Munc13-4 knockout cells (Supplementary Fig. [101]4G, H).
Isolation of EVs via Optiprep^TM density gradient centrifugation
confirmed the association of PD-L1 with exosomes and indicated
disrupted exosome secretion in Munc13-4 knockout cells
(Fig. [102]3D–F).
To assess PD-L1 levels on exosomes, we collected equivalent numbers of
exosomes from both control and knockout cells. Western analysis showed
that PD-L1 levels were significantly lower in exosomes from Munc13-4
knockout cells, while levels of Alix, CD63, and CD81 remained unchanged
(Fig. [103]3G–J). These results indicate that Munc13-4 deletion does
not affect the total PD-L1 levels or its presence on the plasma
membrane but significantly reduces secreted PD-L1 by inhibiting
exosomes secretion and decreasing its enrichment onto exosomes.
We further assessed the role of exosomes from control and Munc13-4
knockout cells in T cell suppression by flow cytometry. Again,
equivalent numbers of exosomes from both control and knockout cells
were collected. Exosomes from Munc13-4-deficient 4T1 cells showed
reduced inhibitory effects on the cytotoxicity of CD8^+ T cells
compared to exosomes from control cells (Supplementary Fig. [104]4I,
J). Notably, anti-PD-L1 treatment significantly decreased the
suppression of CD8^+ T cell activation by control exosomes, while it
had minimal effect on exosomes from Munc13-4 knockout cells
(Supplementary Fig. [105]4I, J). Taken together, the reduced PD-L1
presence on exosomes due to Munc13-4 knockout in tumor cells leads to
decreased immunosuppressive capacity.
To directly characterize the anti-tumor efficacy of T cells influenced
by Munc13-4 deficiency in tumor cells, we examined the cytotoxicity of
T cells primed by either control or Munc13-4-deficient 4T1 cells in
vitro. Mouse spleen lymphocytes activated with anti-CD3 and anti-CD28
antibodies displayed significantly enhanced cytotoxicity against
Munc13-4 knockout 4T1 cells compared to control cells (Fig. [106]3K,
L). Meanwhile, we explored the relationship between decreased
oncogenicity and impaired exosomal PD-L1 secretion in
Munc13-4-deficient 4T1 cells in vivo (Fig. [107]3M). Infusion of
exosomes from control 4T1 cells markedly accelerated tumor growth in
mice bearing Munc13-4 knockout 4T1 cell transplants (Fig. [108]3N and
Supplementary Fig. [109]4K, L). In contrast, exosomes pre-treated with
anti-PD-L1 antibody had minimal impact on tumor growth (Fig. [110]3N
and Supplementary Fig. [111]4K, L). Collectively, these findings
underscore the essential role of Munc13-4 in T cell suppression and
tumor progression through its regulation of PD-L1 secretion via
exosomes.
Munc13-4 deficiency in tumor cells boosts immune checkpoint blockade therapy
effectiveness
Immune checkpoint blockade (ICB) therapy, which targets the PD-1/PD-L1
interaction using antibodies, has become a common approach in cancer
treatment. However, it faces challenges such as limited durability of
remission and a low overall response rate, restricting its benefits to
a small subset of patients^[112]3,[113]4. Recent studies indicate that
PD-L1 on exosomes secreted by tumor cells may antagonize ICB
therapy^[114]5–[115]7. Given our findings that Munc13-4 promotes PD-L1
secretion and inhibits immune surveillance in vivo, we investigated
whether Munc13-4 deletion in tumor cells could enhance ICB therapy
efficacy (Supplementary Fig. [116]5A).
In experiments, neither anti-PD-1 nor anti-PD-L1 treatment slowed tumor
growth in mice inoculated with control 4T1 cells compared to IgG
isotype controls (Supplementary Fig. [117]5B–D). In contrast, tumor
growth was significantly delayed in mice implanted with
Munc13-4-deficient 4T1 cells, and this effect was further enhanced by
anti-PD-1 and anti-PD-L1 treatments (Supplementary Fig. [118]5B–D).
These results suggest that Munc13-4 depletion in tumor cells improves
the therapeutic efficacy of immune checkpoint inhibitors.
Munc13-4 does not influence MVB biogenesis
The above findings indicate that Munc13-4 deletion in breast tumor
cells has two detrimental effects on PD-L1 secretion: (i) reduced PD-L1
sorting on exosomes and (ii) impaired exosome secretion. This may
result from impaired MVB biogenesis and/or MVB fusion with the plasma
membrane. To explore the underlying mechanisms, we utilized the human
breast tumor cell line SUM159 and examined whether Munc13-4 deficiency
disrupts MVB biogenesis. TEM analysis of control and Munc13-4 knockout
SUM159 cells revealed a significant accumulation of MVBs in the
knockout cells (Supplementary Fig. [119]6A, B). This was corroborated
by immunofluorescence and western analyses showing a marked increase in
CD63, a well-established MVB marker (Supplementary Fig. [120]6F–H). In
addition, TEM analysis indicated an increase in ILVs within MVBs in
Munc13-4 knockout cells (Supplementary Fig. [121]6C), suggesting that
reduced PD-L1 secretion is not due to impaired MVB biogenesis.
Further TEM analysis showed a significant increase in hybrid structures
formed between MVBs and lysosomes in Munc13-4 knockout cells,
characterized by electron-dense compartments and double-membrane
autophagosomes (Supplementary Fig. [122]6A, D). Immunofluorescence
analysis confirmed increased colocalization of CD63 with the lysosomal
marker LAMP1 (Supplementary Fig. [123]6I, J). Moreover, TEM data
indicated that the diameter of MVBs in Munc13-4 knockout cells was
significantly enlarged (Supplementary Fig. [124]6E). Together, these
observations suggest an increased prevalence of both homotypic fusion
among MVBs and heterotypic fusion between MVBs and lysosomes in the
absence of Munc13-4.
Munc13-4 facilitates MVB docking and fusion with the plasma membrane
Since Munc13-4 deletion does not affect MVB biogenesis, we investigated
its role in downstream processes related to the docking and fusion of
MVBs with the plasma membrane. Rab GTPase Rab27a plays a critical role
in various stages of the exosome secretion pathway, particularly in the
transport and docking of MVBs to the plasma membrane^[125]11. As an
effector of Rab27a, Munc13-4 collaborates with Rab27a to regulate
exocytosis in various immune cells^[126]19,[127]23,[128]24. Consistent
with this role, our in vitro binding experiment detected significant
binding between Munc13-4 and Rab27a (Supplementary Fig. [129]6K). In
contrast, only minimal interactions were detected with Rab5 or Rab7
(Supplementary Fig. [130]6K), which are primarily associated with early
and late endosomes, respectively. To study the role of the
Munc13-4–Rab27a complex in MVB docking, we expressed CD63 tagged with
orange fluorescent protein in both control and Munc13-4 (or Rab27a)
knockout SUM159 cells. By using total internal reflection fluorescence
(TIRF) microscopy, we tracked the movement of MVBs near the plasma
membrane^[131]11. Either Munc13-4 or Rab27a deficiency increased MVB
mobility (Supplementary Fig. [132]6L, M), indicating the functional
significance of Munc13-4 and Rab27a in MVB docking.
To investigate the molecular mechanisms by which Munc13-4 and Rab27a
contribute to MVB docking, we determined the cryo-EM structure of the
Munc13-4–Rab27a complex (Fig. [133]4A and Supplementary Fig. [134]7).
Note that the non-hydrolyzable GTP analog GppNHp was added to maintain
Rab27a in its active state for efficient binding to Munc13-4
(Supplementary Fig. [135]8A–C). The structure was resolved to 3.4 Å in
the core region of the complex, while the two C[2] domains at the N-
and C-termini of Munc13-4 exhibited a resolution of 4.4 Å due to their
inherent flexibility (Supplementary Fig. [136]8D–H). Compared to the
solved structure of core domain (C[1]-C[2]B-MUN) of Munc13-1^[137]25,
the overall architecture of Munc13-4 is more curved (Supplementary
Fig. [138]8I). The binding interface of the complex comprises residues
F46, W73, and F88 on Rab27a, along with N739, T740, V660, and K661 on
Munc13-4 (Fig. [139]4A). Mutations at these sites significantly
impaired the interaction (Fig. [140]4B, C and Supplementary
Fig. [141]8J, K). Consistent with these findings, such mutations
resulted in increased MVB mobility (Fig. [142]4D, E) and a marked
decrease in the total number of exosomes (Fig. [143]4F). Hence, the
structure information provides mechanistic insight into how Munc13-4
works with Rab27a to promote MVB docking.
Fig. 4. Munc13-4 facilitates MVB docking and fusion with the plasma membrane.
[144]Fig. 4
[145]Open in a new tab
A Cryo-EM structure of the Munc13-4–Rab27a complex (upper panel) and
detailed interface view (lower panel). The corresponding cryo-EM
density map is shown as a semi-transparent surface. GST pull-down
assays examining the effects of different mutations in Munc13-4 (B),
and diverse mutations in Rab27a (C), on the formation of the
Munc13-4–Rab27a complex (n = 3 biological replicates). Quantification
of the mean diffusion coefficient (D) detected by TIRF microscopy in
SUM159 cells with the indicated mutations in Munc13-4 (D) or Rab27a (E)
(n = 3 biological replicates). F Quantification of exosomes secreted by
equal numbers of indicated SUM159 cells through NTA (n = 3 biological
replicates). G Illustration of FRET assay for SNARE complex assembly
detection. VAMP-7 SNARE motif (V7 SNARE) labeled with donor dye BODIPY
FL, SNAP-23 (SN-23) labeled with acceptor dye 5-TAMRA, and syntaxin-4
(transmembrane domain deleted, Syx-4 ΔTM) form a SNARE complex leading
to FRET. H Representative graph of time-dependent SNARE complex
assembly measured by the development of FRET (n = 3 biological
replicates). I Illustration of the liposome fusion experiment.
Syntaxin-4 (Syx-4) was incorporated into DiD-labeled liposomes, and
VAMP-7 was incorporated into DiI-labeled liposomes. Munc13-4
accelerates liposome fusion mediated by SNARE complex, leading to FRET.
J Time-dependent liposome fusion measured from the development of FRET
(n = 3 biological replicates). K Quantification of the FRET efficiency
at the end of the detection (n = 3 biological replicates). Data are
represented as means ± SEM (D, E, F, J, K), and p-values were
calculated by one-way ANOVA with multiple comparisons (D, E, F, K).
Source data are provided as a [146]Source data file.
Our previous studies have identified a SNARE complex composed of
syntaxin-4, SNAP-23, and VAMP-7 as mediators of MVB fusion with the
plasma membrane for exosome release in various tumor cells^[147]12. Our
binding assays demonstrated a direct interaction between Munc13-4 and
both SNAP-23 and VAMP-7, and with the assembled SNARE complex
(Supplementary Fig. [148]6N). We further assessed the regulatory role
of Munc13-4 in SNARE complex assembly and membrane fusion using
FRET-based assembly and fusion assays. Our results indicated that
Munc13-4 significantly facilitates the assembly of syntaxin-4, SNAP-23,
and VAMP-7 into the SNARE complex (Fig. [149]4G, H) and promotes the
fusion between liposomes bearing syntaxin-4/SNAP-23 and liposomes
containing VAMP-7 (Fig. [150]4I–K), underscoring its critical role in
the fusion of MVBs with the plasma membrane. Altogether, these results
suggest that Munc13-4 works with Rab27a and SNAREs to complete exosome
secretion by facilitating the docking and fusion of MVBs with the
plasma membrane.
Exosomal sorting of PD-L1 by Munc13-4 and HRS
Consistent with our finding that Munc13-4 knockout reduces PD-L1
abundance on exosomes, PD-L1 showed decreased colocalization with CD63
and increased colocalization with LAMP1 in Munc13-4 knockout SUM159
cells (Fig. [151]5A, B), indicating improper translocation of PD-L1
from MVBs to lysosomes and strengthening the role of Munc13-4 in
sorting PD-L1 to exosomes.
Fig. 5. Exosomal sorting of PD-L1 by Munc13-4 and HRS.
[152]Fig. 5
[153]Open in a new tab
A Representative confocal images of SUM159 cells co-expressing
GFP-PD-L1 with Orange-CD63 or Orange-LAMP1 (3 independent experiments).
Scale bar, 20 μm. B Quantification of the mean Pearson’s correlation
coefficient between PD-L1 and CD63 or LAMP1 (n = 3 independent
experiments). C WB analysis of PD-L1, Alix, CD63, and CD81 on equal
numbers of exosomes from SUM159 cells (n = 2 biological replicates). D
WB analysis of total HRS in SUM159 cells (n = 3 biological replicates).
E Co-IP/immunoblotting (IB) analysis in SUM159 cells transfected with
indicated constructs (n = 3 biological replicates). F Representative
confocal images of SUM159 cells in the PLA (3 independent experiments).
Scale bar, 10 μm. G Quantification of the mean number of PLA puncta per
cell (n = 3 independent experiments). H Co-IP/IB analysis in Munc13-4
KO SUM159 cells transfected with indicated constructs (n = 3 biological
replicates). I Co-IP/IB analysis in HEK293T cells transfected with
indicated constructs (n = 3 biological replicates). J Representative
PLA image of SUM159 cells (3 independent experiments). Scale bar,
10 μm. K Co-IP/IB analysis in HEK293T cells transfected with indicated
constructs (n = 3 biological replicates). L Representative PLA image of
SUM159 cells (3 independent experiments). Scale bar, 10 μm. M Schematic
of in vitro liposome co-flotation assay. N WB analysis of Munc13-4 and
PD-L1 in top three fractions and bottom fraction (n = 3 biological
replicates). O Co-IP/IB analysis in different SUM159 cells transfected
with indicated constructs (n = 3 biological replicates). In (C, D, N),
samples were run on the same gel; in (E, H, I, K, O), IP and Input
samples were derived from the corresponding same experiment, but
different gels for IP and Input samples were processed in parallel.
Data are represented as means ± SEM (B, G), and p-values were
calculated by two-way ANOVA with multiple comparisons (B) and unpaired
two-tailed t-test (G). Source data are provided as a [154]Source data
file.
HRS is a key component of ESCRT-0 and has been found to mediate PD-L1
sorting^[155]26,[156]27. This function was verified by the significant
reduction in PD-L1 abundance on exosomes following the deletion of HRS
(Fig. [157]5C). To further explore this, we examined the relationship
between HRS and PD-L1 in both the presence and absence of Munc13-4. In
Munc13-4 knockout SUM159 cells, the expression of PD-L1 and HRS was
unaffected (Figs. [158]3A and [159]5D); while the interaction between
HRS and PD-L1 was markedly diminished compared to control cells,
regardless of whether they were exogenously or endogenously expressed
(Fig. [160]5E–G). Reintroducing Munc13-4 into knockout cells
significantly restored the HRS–PD-L1 interaction (Fig. [161]5H),
suggesting that this interaction relies on Munc13-4. Supporting these
findings, co-immunoprecipitation (co-IP) assay showed a marked binding
preference of Munc13-4 for HRS over STAM (another ESCRT-0 component)
under exogenous expression conditions (Fig. [162]5I). In addition,
proximity ligation assay (PLA) indicated an interaction between
endogenous expressed Munc13-4 and HRS in SUM159 cells (Fig. [163]5J).
We also examined whether Munc13-4 directly interacts with PD-L1. Both
co-IP and PLA assays revealed this interaction in SUM159 cells
(Fig. [164]5K, L). This interaction was further verified through in
vitro liposome co-flotation assay (Fig. [165]5M, N). Importantly, the
deletion of HRS in SUM159 cells did not influence the interaction
between Munc13-4 and PD-L1 (Fig. [166]5O), indicating that the
Munc13-4–PD-L1 interaction is independent of HRS. Altogether, these
results suggest the formation of a ternary complex comprising HRS,
Munc13-4, and PD-L1, which is critical for sorting of PD-L1 to
exosomes.
IFNγ-induced modifications of Munc13-4 and HRS exert opposing effects on
PD-L1 sorting
IFNγ, a cytokine produced by NK and T cells, contributes substantially
to immunosurveillance against tumors by activating immune cells and
inducing apoptosis in tumor cells^[167]28,[168]29. Conversely, tumor
cells can exploit IFNγ signaling to evade immune destruction through
the elevation of PD-L1 expression, which inhibits immune cell
activity^[169]28,[170]29. Indeed, IFNγ stimulation dramatically
increased the protein level of PD-L1 (Supplementary Fig. [171]9A, B),
leading to an elevated presence of PD-L1 on both the plasma membrane
(Supplementary Fig. [172]9C, D) and exosomes (Supplementary
Fig. [173]9E, F), without altering the total quantity of secreted
exosomes (Supplementary Fig. [174]9E–G). Considering the significant
effect of exosomal PD-L1 on the suppression of T cell activity, we
hence explored whether and how HRS and Munc13-4 regulate PD-L1 sorting
onto exosomes in response to IFNγ.
Unlike PD-L1, the overall levels of HRS and Munc13-4 remained unchanged
under IFNγ treatment in SUM159 cells (Supplementary Fig. [175]9A, B).
Strikingly, IFNγ stimulation significantly decreased the ubiquitylation
of HRS (Supplementary Fig. [176]9H), without affecting its acetylation
and phosphorylation (Supplementary Fig. [177]9I–K). Meanwhile, IFNγ
stimulation significantly increased the acetylation of Munc13-4
(Supplementary Fig. [178]9L), without influencing its phosphorylation
and ubiquitylation (Supplementary Fig. [179]9M–O). These results
indicate that both HRS ubiquitylation and Munc13-4 acetylation induced
by IFNγ may cooperate to regulate PD-L1 sorting onto exosomes.
To explore the mechanism that regulates Munc13-4 acetylation, we
individually expressed six common acetyltransferases—GCN5, PCAF, CBP,
P300, TIP60, and HBO1—with Munc13-4 in HEK293T cells and found that CBP
and P300 are able to acetylate Munc13-4 (Supplementary Fig. [180]10A).
Knockout of CBP—rather than P300—reduced the acetylation of endogenous
Munc13-4 in SUM159 cells, both in the absence and presence of IFNγ
(Supplementary Fig. [181]10B). In addition, IFNγ stimulation promoted
the translocation of CBP from the nucleus to the cytoplasm
(Supplementary Fig. [182]10C, D). These data indicate that CBP acts as
a physiological acetyltransferase for Munc13-4. Then, we screened for
the deacetylase mediating Munc13-4 deacetylation using deacetylase
inhibitors. We found that trichostatin A (TSA), an inhibitor of histone
deacetylases (HDACs)^[183]30, increased Munc13-4 acetylation, while
nicotinamide (NIC), an inhibitor of class-III sirtuin deacetylases
(SIRTs)^[184]30, had minimal effect (Supplementary Fig. [185]10E).
Among the HDAC1–8 isoforms, expression of HDAC3 and HDAC4, but not the
other HDACs, reduced Munc13-4 acetylation in HEK293T cells
(Supplementary Fig. [186]10F–H). Notably, knockout of HDAC3, rather
than HDAC4, increased the acetylation level of endogenous Munc13-4,
regardless of IFNγ treatment (Supplementary Fig. [187]10I), suggesting
that HDAC3 serves as a physiological deacetylase for Munc13-4. It is
noteworthy that the deletion of CBP and HDAC also influenced Munc13-4
expression, as evidenced by a reduction in Munc13-4 transcription
(Supplementary Fig. [188]10J, L) and a corresponding decrease in the
total amount of Munc13-4 (Supplementary Fig. [189]10K, M). We clarify
that to ensure the accuracy of the data, our evaluation of the
acetylation level of Munc13-4 in the experiments mentioned above was
performed with the condition that the total amounts of Munc13-4 samples
remained consistent (see “Methods”).
We proceeded to investigate the impact of Munc13-4 acetylation on the
sorting of PD-L1 onto exosomes. In the presence of IFNγ, the knockout
of CBP and HDAC3 did not influence the total amount of PD-L1
(Supplementary Fig. [190]10N, O) but displayed opposing effects on
PD-L1 sorting. Specifically, the knockout of CBP, which reduces
Munc13-4 acetylation, led to a significant increase in the abundance of
PD-L1 on exosomes (Fig. [191]6A, B). In contrast, the knockout of
HDAC3, which enhances Munc13-4 acetylation, resulted in a remarkable
decrease in PD-L1 abundance on exosomes (Fig. [192]6C, D). Hence, these
results consistently suggest that Munc13-4 acetylation inhibits the
sorting of PD-L1 onto exosomes.
Fig. 6. IFNγ-induced modifications of Munc13-4 and HRS exert opposing effects
on PD-L1 sorting.
[193]Fig. 6
[194]Open in a new tab
WB analysis on equal numbers of exosomes from control and CBP KO SUM159
cells under IFNγ treatment (A) and quantification of blot band
intensities (B) (n = 3 biological replicates). WB analysis on equal
numbers of exosomes from control and HDAC3 KO SUM159 cells under IFNγ
treatment (C) and quantification of blot band intensities (D) (n = 3
biological replicates). WB analysis on equal numbers of exosomes from
control and NEDD4L knockdown (KD) SUM159 cells under IFNγ treatment (E)
and quantification of blot band intensities (F) (n = 3 biological
replicates). WB analysis on equal numbers of exosomes from SUM159 cells
co-treated with IFNγ and PR-619 or DMSO (G) and quantification of blot
band intensities (H) (n = 3 biological replicates). I Representative
confocal images of indicated SUM159 cells in the PLA (3 independent
experiments). Scale bar, 10 μm. J Quantification of the mean number of
PLA puncta per cell (n = 3 independent experiments). WB analysis on
equal numbers of exosomes from indicated SUM159 cells (K) and
quantification of blot band intensities (L) (n = 3 biological
replicates). M Co-IP/IB analysis in HEK293T cells transfected with
indicated constructs (n = 3 biological replicates). N IP/IB analysis in
different SUM159 cells ± IFNγ treatment (n = 3 biological replicates).
In (A, C, E, G, K), samples were run on the same gel; in (M, N), IP and
Input samples were derived from the corresponding same experiment, but
different gels for IP and Input samples were processed in parallel.
Data are represented as means ± SEM (B, D, F, H, J, L), and p-values
were calculated by two-way ANOVA with multiple comparisons (B, D, F, H,
J, L). Source data are provided as a [195]Source data file.
On the other hand, IFNγ stimulation induces HRS deubiquitylation, and
we next investigate the effect of HRS ubiquitylation on PD-L1 sorting.
To identify the E3 ligases responsible for HRS ubiquitylation, we
utilized the UbiBrowser 2.0 database
([196]http://ubibrowser.bio-it.cn/ubibrowser_v3/)^[197]31. Among the
identified and predicted E3 ligases, including NEDD4^[198]32,
NEDD4L^[199]33, SH3RF1^[200]34, ITCH, CBL, and PRKN, we found that
NEDD4L efficiently catalyzed HRS ubiquitylation in HEK293T cells
(Supplementary Fig. [201]10P). In SUM159 cells, NEDD4L knockdown led to
a significant reduction in HRS ubiquitylation, regardless of IFNγ
stimulation (Supplementary Fig. [202]10Q). Notably, under IFNγ
treatment, NEDD4L knockdown resulted in a marked increase in the amount
of PD-L1 on exosomes (Fig. [203]6E, F) without altering total amount of
PD-L1 (Supplementary Fig. [204]10Q). Furthermore, expression of several
common deubiquitinases, including STAMBPL1, CYLD, USP11, USP7, and
USP8, was found to reduce NEDD4L-mediated HRS ubiquitylation in HEK293T
cells (Supplementary Fig. [205]10R), suggesting that the regulation of
HRS deubiquitylation involves multiple deubiquitinases. Treatment with
PR-619, a broad-spectrum reversible inhibitor of ubiquitin
isopeptidases^[206]35,[207]36, induced a substantial increase in HRS
ubiquitylation in SUM159 cells independent of IFNγ (Supplementary
Fig. [208]10S). However, PR-619 treatment resulted in a significant
reduction in the abundance of PD-L1 on exosomes under IFNγ stimulation
(Fig. [209]6G, H), without influencing the overall PD-L1 level
(Supplementary Fig. [210]10S). Collectively, these findings demonstrate
that HRS deubiquitylation facilitates the sorting of PD-L1 onto
exosomes.
IFNγ-induced modifications of Munc13-4 and HRS regulate PD-L1 binding
The finding that Munc13-4 acetylation inhibits PD-L1 sorting suggests
that it impairs the interaction between Munc13-4 and PD-L1. Indeed, the
enhanced acetylation of Munc13-4, driven by the expression of CBP, led
to a reduction in its interaction with PD-L1 (Supplementary
Fig. [211]11A), as well as a weakened association of PD-L1 with HRS
(Supplementary Fig. [212]11B), while not affecting the interaction
between HRS and Munc13-4 (Supplementary Fig. [213]11B). This suggests
that Munc13-4 serves as a central hub for the formation of the
HRS–Munc13-4–PD-L1 complex. Interestingly, the mutant deleting the
C-terminus of Munc13-4 (termed Munc13-ΔC, lacking residues 1049–1090)
significantly diminished the acetylation by CBP (Supplementary
Fig. [214]11C, D), indicating that the acetylation sites are located
within this region. Furthermore, Munc13-ΔC failed to bind PD-L1
(Supplementary Fig. [215]11E), suggesting that the residues within
1049–1090 are crucial for both acetylation and PD-L1 binding. However,
Munc13-ΔC retained its ability to bind HRS. Further screening revealed
that residues 546–782 within the MUN domain of Munc13-4 mediate HRS
interaction (Supplementary Fig. [216]11F).
We aimed to identify the acetylation sites on Munc13-4. The region
between residues 1049 and 1090 contains two lysine (K) residues, and
mass spectrometry analysis revealed that both K1062 and K1079 were
acetylated in the presence of CBP (Supplementary Fig. [217]11G, H).
Single mutations of either K1062 or K1079 to arginine (K1062R or
K1079R) preserved the positive charge but impaired acetylation.
Moreover, the double mutant (K1062R/K1079R, termed KKRR) nearly
completely abolished CBP-mediated acetylation of Munc13-4
(Supplementary Fig. [218]11I), confirming our mass spectrometry
findings. Detected by co-IP, mutating either K1062 or K1079 in Munc13-4
to glutamine (Q), which mimics acetylation, did not affect its
interaction with PD-L1 (Supplementary Fig. [219]11J). However, double
mutations (K1062Q/K1079Q, termed KKQQ) significantly impaired this
interaction (Supplementary Fig. [220]11J, K). In contrast, the KKRR
mutant, which abolishes acetylation, did not affect the Munc13-4–PD-L1
interaction (Supplementary Fig. [221]11K).
In Munc13-4 knockout SUM159 cells, where the endogenous interaction
between HRS and PD-L1 was severely damaged, expression of Munc13-4 WT
rescued their endogenous interaction to the level comparable to that in
control cells (Fig. [222]6I, J). In contrast, Munc13-4 KKQQ only
slightly restored the HRS–PD-L1 interaction (Fig. [223]6I, J), while
Munc13-4 KKRR substantially rescued this interaction (Fig. [224]6I, J).
Similar results were observed in HEK293T cells (Supplementary
Fig. [225]11L). Furthermore, the decrease in the PD-L1 abundance on
exosomes observed upon Munc13-4 deletion was fully restored by the
expression of either Munc13-4 WT or the KKRR mutant, but not by the
KKQQ mutant (Fig. [226]6K, L). Overall, CBP-mediated acetylation of
Munc13-4 at K1062 and K1079 disrupts its interaction with PD-L1, which
in turn impairs the HRS–PD-L1 interaction and hinders the sorting of
PD-L1 onto exosomes.
We next investigated the mechanism by which HRS deubiquitylation
promotes PD-L1 sorting onto exosomes. Expression of NEDD4L in HEK293T
cells, which significantly enhanced HRS ubiquitylation, notably
suppressed the interaction between HRS and PD-L1 (Fig. [227]6M). In
contrast, additional expression of USP11, which decreased HRS
ubiquitylation, restored the HRS–PD-L1 interaction (Fig. [228]6M).
Notably, the ubiquitylation status of HRS did not affect its
association with Munc13-4 (Supplementary Fig. [229]11M). These findings
suggest that HRS ubiquitylation negatively regulates its interaction
with PD-L1. In SUM159 cells, deletion of Munc13-4 also led to a
reduction in HRS ubiquitylation, independent of IFNγ stimulation
(Fig. [230]6N), which likely compensates for the diminished HRS–PD-L1
interaction caused by Munc13-4 deficiency, suggesting a potential
compensatory mechanism in PD-L1 sorting in the absence of Munc13-4.
A peptide that disrupts PD-L1–Munc13-4 interaction inhibits tumor growth
Given the critical role of Munc13-4 in sorting PD-L1 to exosomes, we
sought to disrupt their interaction in tumor cells to mitigate immune
evasion. Targeting the PD-L1–Munc13-4 interaction might be a promising
therapeutic strategy. Our co-IP analysis identified both the
cytoplasmic motif and the transmembrane domain of PD-L1 as essential
for binding to Munc13-4 (Supplementary Fig. [231]12A). To develop a
peptide inhibitor targeting this interaction interface, we selected an
18-residue segment (residues 256–273) spanning the
transmembrane–cytoplasmic junction of PD-L1. This region encompasses
key residues from both domains identified by co-IP and was chosen for
its suitable length and favorable biochemical properties, such as
predicted solubility.
Next, we tested whether the 256–273 sequence could competitively
inhibit the interaction between PD-L1 and Munc13-4. Overexpressing this
sequence in HEK293T cells significantly disrupted their interaction
(Supplementary Fig. [232]12B). This disruption also affected the
interaction between HRS and PD-L1, but did not impact the HRS–Munc13-4
interaction (Supplementary Fig. [233]12C). Further exploration utilized
cell-penetrating peptide (CPP)^[234]37,[235]38 to deliver the PD-L1
256–273 peptide (P-pep) and a scrambled version (S-pep) (Fig. [236]7A).
Treatment of HEK293T cells with P-pep, compared to S-pep, significantly
inhibited the ectopic interaction between PD-L1 and Munc13-4
(Fig. [237]7B) and consequently disrupted the interaction between HRS
and PD-L1 (Fig. [238]7C), while the HRS–Munc13-4 interaction remained
unaffected (Fig. [239]7C). In addition, P-pep treatment of SUM159 cells
led to a marked decrease in PD-L1 levels on exosomes (Fig. [240]7D),
suggesting that competitive inhibition of PD-L1 sorting is a viable
approach.
Fig. 7. A peptide disrupting PD-L1–Munc13-4 interaction inhibits tumor
growth.
[241]Fig. 7
[242]Open in a new tab
A Diagram of the sequences for P-pep and S-pep. P-pep comprises a
cell-penetrating peptide (CPP) fused to the human PD-L1 256–273 motif,
whereas S-pep consists of a CPP linked to a scrambled sequence
containing the same amino acid composition as the human PD-L1 256–273
motif. B Co-IP/IB analysis in HEK293T cells transfected with indicated
constructs and incubated with P-pep or S-pep to examine the effect of
P-pep on Munc13-4–PD-L1 interaction (n = 3 biological replicates). C
Co-IP/IB analysis in HEK293T cells transfected with indicated
constructs and incubated with P-pep or S-pep to assess the effect of
P-pep on the interactions of HRS with PD-L1 and Munc13-4 (n = 3
biological replicates). D WB analysis on the same number of exosomes
secreted from equal number of SUM159 cells treated with S-pep or P-pep
(n = 3 biological replicates). E Schematic of experimental design for
the assessment of in vivo anti-tumor efficacy of P-pep. F Tumor growth
curves of orthotopic mouse models of breast cancer treated with P-pep
or S-pep (n = 9 mice). Flow cytometric quantification of the percentage
of CD45^+CD3^+CD4^+ (G) and CD45^+CD3^+CD8^+ (H) T cells among total
cells in tumors (n = 5 mice). Representative contour plots depicting
CD45^+CD3^+CD4^+ (I) and CD45^+CD3^+CD8^+ (K) T cell populations within
tumors, showing the expression of granzyme B (n = 5 mice).
Quantification of the percentage of granzyme B^+ cells among
CD45^+CD3^+CD4^+ (J) and CD45^+CD3^+CD8^+ (L) T cells within tumors
(n = 5 mice). In (B, C), IP and Input samples were derived from the
corresponding same experiment, but different gels for IP and Input
samples were processed in parallel; in (D), all samples were run on the
same gel. Data are represented as means ± SEM (F), box plots show the
median (center line), interquartile range (box), minima and maxima
(whiskers), and all individual data points (dots) (G, H, J, L), and
p-values were all calculated by unpaired two-tailed t-test. Source data
are provided as a [243]Source data file.
We also evaluated the in vivo effects of P-pep on tumor growth using
orthotopic breast cancer mouse models (Fig. [244]7E and Supplementary
Fig. [245]12D). Mice treated with P-pep exhibited a significant delay
in tumor growth compared to those receiving S-pep (Fig. [246]7F and
Supplementary Fig. [247]12E, F). Immunophenotyping revealed increased
infiltration of both CD4^+ and CD8^+ T cells in the tumors of
P-pep-treated mice (Fig. [248]7G, H), along with enhanced cytotoxicity
of tumor-infiltrating CD8^+ T cells (Fig. [249]7K, L), while CD4^+ T
cell cytotoxicity remained unchanged (Fig. [250]7I, J). Comprehensive
assessments, including body weight, blood routine tests, biochemical
analysis, and histological evaluations of major organs (Supplementary
Fig. [251]12G–P), indicated no noticeable toxicity from either P-pep or
S-pep. Collectively, these results demonstrate that P-pep effectively
targets the PD-L1–Munc13-4 interaction, reducing tumor-induced
immunosuppression and inhibiting tumor growth without systemic side
effects.
Discussion
Tumor-derived exosomes carry PD-L1, which engages with PD-1 receptors
on T cells, influencing immune responses within the tumor
microenvironment and in distant sites, resulting in a more extensive
immunosuppressive environment^[252]5,[253]8,[254]39. Therapeutic
strategies aimed at counteracting the immunosuppressive effects of
exosomal PD-L1 require the development of molecules that can
effectively suppress its extracellular secretion. Hence, it is
essential to identify the regulatory factors governing the secretory
pathway of exosomal PD-L1 and to elucidate the underlying mechanisms.
In this study, we reveal a critical role for Munc13-4 in modulating the
immunosuppressive effects of exosomal PD-L1 by influencing its sorting
and secretion. Specifically, deleting Munc13-4 in breast tumor cells
significantly reduces both the number of secreted exosomes and the
abundance of PD-L1 on these exosomes, which systematically enhances T
cell-mediated anti-tumor responses, suppresses tumor progression, and
improves the efficacy of immune checkpoint inhibitors. The underlying
mechanisms involve i) formation of the Munc13-4–PD-L1–HRS ternary
complex, which promotes efficient sorting of PD-L1 to MVBs and loading
onto exosomes; ii) assembly of the Munc13-4–Rab27a complex, which
enables proper MVB docking; and iii) cooperation between Munc13-4 and
the SNARE complex comprising syntaxin-4, SNAP-23 and VAMP-7, which
facilitates MVB fusion with the plasma membrane to release
PD-L1-containing exosomes. Notably, employing a specially designed
peptide to disrupt the Munc13-4–PD-L1 interaction, thereby impairing
PD-L1 sorting, significantly enhances anti-tumor immunity and slows
tumor growth in vivo (Fig. [255]8). This underscores the potential of
targeting the Munc13-4–PD-L1 axis as an effective approach to augment
the efficacy of immune checkpoint inhibitors.
Fig. 8. Mechanistic model of Munc13-4-mediated tumor immune evasion through
the regulation of PD-L1 sorting and secretion via exosomes.
[256]Fig. 8
[257]Open in a new tab
Schematic illustration showing that Munc13-4 collaborates with HRS,
Rab27, and SNAREs to regulate PD-L1 sorting and secretion via exosomes.
Loss of Munc13-4 in breast tumors enhances T cell-mediated anti-tumor
immunity, suppresses tumor growth, and improves the efficacy of immune
checkpoint inhibitors. Mechanistically, Munc13-4 regulates PD-L1
sorting by forming a ternary complex with PD-L1 and HRS. IFNγ
stimulation modifies Munc13-4 and HRS, establishing a dynamic
regulatory mechanism that enables tumor cells to adapt to immune
pressure by modulating exosomal PD-L1 sorting. Downstream of sorting,
Munc13-4 engages Rab27a to regulate MVB tethering and promotes SNARE
complex assembly, thereby facilitating MVB–plasma membrane fusion and
exosome release. Therapeutically, a designed peptide that disrupts the
Munc13-4–PD-L1 interaction impairs PD-L1 sorting, resulting in enhanced
anti-tumor immunity and reduced tumor growth in vivo.
Munc13-4 is ubiquitously expressed in various types of cells and
contains a MUN domain flanked by two C[2] domains^[258]40. Munc13-4 has
attracted significant attention for its role in the exocytosis of
secretory granules in immune cells. Inherited variants of Munc13-4 have
been associated with Familial Hemophagocytic Lymphohistiocytosis Type 3
(FHL3), a rare autosomal recessive disorder characterized by impaired
granule exocytosis^[259]16,[260]41. Munc13-4 is also involved in
exosome secretion in tumor cells^[261]20. Building upon this existing
knowledge, our work contributes important insights by revealing that
Munc13-4 is upregulated in various tumor cells and regulates multiple
steps toward MVB exocytosis, leading to exosome secretion, including
MVB docking and fusion with the plasma membrane. This emphasizes its
broader role in exocytosis across different cell types and highlights
its potential as a biomarker for cancer diagnosis and prognosis. By
regulating exosome secretion in tumor cells, Munc13-4 impacts the tumor
microenvironment and facilitates communication between tumor and immune
cells, influencing tumor growth, metastasis, and anti-tumor immunity.
The accumulation of MVBs in Munc13-4 knockout cells suggests that
Munc13-4 is crucial for the docking of MVBs with the plasma membrane.
While previous studies have underscored the importance of the
Munc13-4–Rab27a complex in regulating secretory granule
docking^[262]19,[263]23,[264]24, the underlying mechanisms remain
unclear due to the absence of structural information. In this study, we
present the cryo-EM structure of the Munc13-4–Rab27a complex, revealing
a previously unexplored binding interface essential for complex
stability. Specifically, residues F46, W73, and F88 in Rab27a, along
with residues N739/T740 and V660/K661 in Munc13-4, maintain this
stability and mediate MVB docking with the plasma membrane. Notably,
the Rab27a-binding surface on Munc13-4 is located within the middle of
the MUN domain (spanning residues 651–778), rather than in an
N-terminal sequence connecting the C[2]A and MUN domain as reported
previously^[265]19,[266]24. Indeed, mutations in F46 and W73 in Rab27a
and T740 in Munc13-4 have been previously linked to Griscelli Syndrome
type 2 (GS2)^[267]42,[268]43 and FHL3^[269]44. Consistent with the
structural data, introducing the mutations that disrupt the binding
surface between Munc13-4 and Rab27a significantly increases the
mobility of MVBs beneath the plasma membrane, which leads to impaired
exosome secretion. Therefore, our results support the notion that
Munc13-4 serves as the effector of Rab27a, stabilizing MVB docking
close to the plasma membrane, thereby promoting exosome secretion.
Together, Munc13-4–Rab27a-mediated MVB docking model may represent a
common mechanism governing exosome secretion in various tumor cells.
Our previous study has identified that the SNARE complex that mediates
MVB fusion with the plasma membrane in tumor cells is composed of
syntaxin-4, SNAP-23, and VAMP-7^[270]12. Here, we observed an
accelerated effect of Munc13-4 on SNARE complex assembly and
SNARE-mediated membrane fusion, which is dependent on its interaction
with SNAP-23 (Q[bc]-SNARE) and VAMP-7 (R-SNARE). This suggests an
important role for Munc13-4 in chaperoning the proper conformation of
SNAREs and/or in stabilizing the SNARE complex. The SNARE-chaperoning
role of Munc13s, particularly the well-studied isoform Munc13-1, has
been extensively documented^[271]45–[272]49. While Munc13-4 shares
similar domain structures with Munc13-1, their mechanisms in assisting
SNARE complex assembly and membrane fusion may differ significantly.
Syntaxin-4 is known to adopt a closed conformation similar to
syntaxin-1^[273]50,[274]51. However, Munc13-4 lacks the hydrophobic
core (the NF pocket) in its MUN domain^[275]46, making it unlikely to
catalyze the opening of syntaxin-4. In this regard, the activation of
syntaxin-4 may depend on the interaction between its N-peptide and the
corresponding SM protein, Munc18-3^[276]52,[277]53. Our observation
that Munc13-4 does not interact with syntaxin-4 further corroborates
the notion. As a member of the CATCHR protein family, our finding that
Munc13-4 binds to Q[bc]- and R-SNAREs is expected. For example,
Munc13-1 binds to Q[bc]-SNARE SNAP-25^[278]54 and R-SNARE
VAMP2^[279]48, aiding their assembly into the SNARE complex. The Dsl1
complex interacts with Q[b]-SNARE Sec20 and Q[c]-SNARE Use1,
stabilizing the complex’s conformation^[280]55. The GARP complex
subunit Vps51 binds to Q[c]-SNARE Tlg1, likely promoting its connection
to SNARE bundles^[281]56,[282]57. Similarly, the Exocyst complex
subunit Sec6 interacts with Q[bc]-SNARE Sec9, facilitating both binary
and ternary SNARE complex formation^[283]58,[284]59. Together with our
findings, these evidences suggest that Munc13-4 promotes SNARE complex
formation during MVB fusion through mechanisms shared by various CATCHR
family members.
The observation that Munc13-4 deletion does not alter the overall
levels of PD-L1 on the plasma membrane but reduces the abundance of
PD-L1 on secreted exosomes indicates a previously unrecognized role for
Munc13-4 in cargo sorting within the endosomal system. HRS recognizes
ubiquitinated proteins through its ubiquitin-interacting motif (UIM),
which is essential for initiating cargo sorting during ESCRT-mediated
MVB biogenesis^[285]60,[286]61. While HRS was reported to sort cargoes
such as interleukin-2 receptor beta (IL-2Rβ)^[287]62 and PD-L1^[288]27
independently of the UIM-ubiquitin interaction, the mechanism is not
well understood. Our findings show that Munc13-4 independently binds
both PD-L1 and HRS, but HRS cannot bind to PD-L1 without Munc13-4,
highlighting Munc13-4’s critical role in recognizing PD-L1 before HRS
interaction. This suggests that Munc13-4 mediates the recruitment of
HRS to PD-L1, initiating a series of events that involve other ESCRT
complexes and proteins required for the formation of PD-L1-containing
ILVs. Therefore, the formation of a ternary complex consisting of
Munc13-4, HRS, and Munc13-4-binding cargoes (e.g., PD-L1) may represent
a ubiquitin-independent cargo sorting mechanism, where Munc13-4 and HRS
work together to enable the proper sorting and packaging of cargoes
into exosomes, facilitating its subsequent secretion. Further research
is required to identify the full spectrum of exosomal cargo proteins
sorted by Munc13-4. Overall, Munc13-4 works in conjunction with HRS,
Rab27a, and SNAREs to establish an “assembly line” that effectively
manages the processes of cargo sorting, packaging, trafficking, and
release. This collaborative mechanism is critical for ensuring the
efficient secretion of various proteins, including those involved in
immune responses, highlighting the importance of Munc13-4 in exosomal
biology.
In the tumor microenvironment, tumor-infiltrating lymphocytes can
secrete IFNγ to stimulate anti-tumor immune responses and induce tumor
cell apoptosis^[289]28,[290]29. However, tumor cells can exploit IFNγ
to reduce anti-tumor immunity by increasing PD-L1 levels on exosomes.
Our findings reveal that IFNγ exerts dual and opposing effects on PD-L1
sorting onto exosomes by modulating Munc13-4 acetylation and HRS
deubiquitylation. IFNγ triggers the translocation of the
acetyltransferase CBP from the nucleus to the cytoplasm, leading to
acetylation of K1062/K1079 at the C-terminal end of Munc13-4. The
acetylation of K1062/K1079 disrupts the Munc13-4–PD-L1 interaction,
thereby reducing PD-L1 sorting. Notably, these two lysine residues are
unique to Munc13-4 and not conserved among Munc13s, underscoring
Munc13-4’s distinct functional role in cargo recognition. Meanwhile,
IFNγ stimulates HRS deubiquitylation, which enhances the HRS–PD-L1
interaction to increase PD-L1 sorting, likely due to a conformational
change that leads to the activation of HRS^[291]63. These opposing
effects indicate a nuanced regulatory mechanism where tumor cells may
adapt to immune pressure by manipulating the sorting of PD-L1. Besides
ubiquitylation, ERK-mediated phosphorylation of HRS also influences
PD-L1 sorting^[292]27. Understanding these subtle regulatory mechanisms
can help develop targeted therapies, such as modulating Munc13-4
acetylation through CBP or HDAC3, or altering HRS ubiquitylation and/or
phosphorylation via NEDD4L, deubiquitinases, or ERK. Recent studies
suggest that exosomal PD-L1 levels in the blood of cancer patients
could serve as a potential biomarker for predicting responses to immune
checkpoint blockade (ICB) therapies^[293]5–[294]7. The variability in
PD-L1 levels on circulating exosomes among different patients may be
linked to differences in Munc13-4 acetylation and HRS
ubiquitylation/phosphorylation. Our findings could provide valuable
insights for guiding ICB treatment decisions by analyzing Munc13-4
acetylation and HRS ubiquitylation in pathological tissue sections.
Tumor-derived exosomes containing PD-L1 have systemic immunosuppressive
effects^[295]8,[296]39, making the reduction of exosomal PD-L1
secretion a promising therapeutic strategy. Genetic blockade of overall
exosome secretion has proven effective in slowing tumor growth and
enhancing immunotherapy outcomes. For instance, loss of Rab27a in tumor
cells inhibits exosome secretion, decreasing the release of all
exosomal cargoes, which leads to reduced tumor growth and improved T
cell anti-tumor activity^[297]8,[298]64, similar to the effects
observed with Munc13-4 in our study. However, while reducing overall
exosome secretion can lessen their immunosuppressive effects, some
evidence suggests that exosomes may also offer therapeutic benefits in
immunotherapy. For example, exosomes can deliver tumor-associated
antigens (TAAs) to dendritic cells, boosting T cell activation and
promoting anti-tumor responses^[299]65. Exosomes carrying TAA-MHC
complexes can directly stimulate antigen-specific T cell
activation^[300]66. Therefore, selectively targeting PD-L1 sorting onto
exosomes could eliminate their immunosuppressive effects while
retaining their immunostimulatory potential. Our findings pinpoint the
binding sites necessary for the Munc13-4–PD-L1 interaction that is
specific for PD-L1 sorting. Disrupting this interaction with a designed
peptide effectively reduces PD-L1 enrichment on exosomes without
affecting overall exosome secretion. In vivo, this peptide treatment
significantly enhances T cell function and inhibits tumor growth
without major side effects. The peptide targets the PD-L1 motif that
interacts with Munc13-4, allowing for selective PD-L1 targeting and
minimizing the risk of immune dysfunction from non-specific peptide
uptake by immune cells. Overall, this peptide holds great promise as a
therapeutic agent for modulating immune responses.
In conclusion, we elucidate the Munc13-4-dependent mechanisms that
govern the secretion of PD-L1 via exosomes in breast cancer cells,
highlighting the functional complexity of Munc13-4 and enhancing our
understanding of exosome biogenesis and secretion. Our in vivo findings
offer valuable insights into the regulation of PD-L1 secretion,
suggesting promising therapeutic strategies to improve patient outcomes
in cancer treatment.
Methods
Ethics statement
For mice: BALB/c, BALB/Nude, and NOD/SCID mice (female, 5–7-week-old)
were purchased from Beijing Vital River Laboratory Animal Technology
Co., Ltd. (Beijing, China). The mice were housed in a controlled animal
facility under consistent environmental conditions, including a room
temperature of 22 ± 1 °C, relative humidity of 40–70%, and a 12-h
light/dark cycle. Food and water were provided ad libitum. Mice were
randomly assigned at the start of each experiment. All experimental
procedures were conducted in compliance with the guidelines and with
approval from the Institutional Animal Care and Use Committee (IACUC)
of Tongji Medical College, Huazhong University of Science and
Technology (Wuhan, China).
For patients’ samples: formalin-fixed, paraffin-embedded human tissue
arrays (HOrgC180PG01-2 and HBreD180Bc01-2) were obtained from Shanghai
Outdo Biotech Co., Ltd. (China). The HOrgC180PG01-2 array comprises 180
tissue cores derived from 91 patients across 14 tumor types, and the
HBreD180Bc01-2 array includes 180 tissue cores from 150 patients with
triple-negative breast cancer. Both tissue arrays were subjected to
immunohistochemical (IHC) analysis, with detailed clinical information
retrieved from the company’s website
([301]https://www.superchip.com.cn/). All the human tissue samples were
collected with informed consent from the donors, and their use was
approved by the Ethics Committee of Shanghai Outdo Biotechnology Co.,
Ltd., in accordance with relevant ethical guidelines and regulations.
Bacterial strains
Escherichia coli strains BL21(DE3) (Thermo), DH10Bac (Gibco), and DH5α
(Thermo) were cultured in Luria-Bertani (LB) broth at 37 °C with
shaking at 200 rpm. The media was supplemented with appropriate
antibiotics: ampicillin (100 µg/ml), kanamycin (50 µg/ml), gentamicin
(7 µg/ml), or tetracycline (10 µg/ml), as necessary.
Cell culture
HEK293T and 4T1 cells were obtained from American Type Culture
Collection (ATCC). SUM159 were obtained from Pcocell. HEK293T and
SUM159 cells were cultured in DMEM medium (Gibco, 11965092) with 10%
FBS (Gibco, A5670701) and 1% penicillin-streptomycin solution
(Proteintech, PR40022). 4T1 cells were cultured in RPMI 1640 medium
(Gibco, 11875093) supplemented with 10% FBS (Gibco, A5670701) and 1%
penicillin-streptomycin (Proteintech, PR40022). These cells were
cultured in a humidified incubator (Thermo) at 37 °C with 5% CO2. Sf9
cells (Gibco) were cultured in SIM-SF Expression Medium
(SinoBiological, MSF1) at 27 °C, 125 rpm.
Online data acquisition and analysis
The differential expression of Munc13-4 between tumor and normal
tissues was analyzed utilizing TIMER2.0
([302]http://timer.cistrome.org/)^[303]21,[304]22. The reported and
predicted ubiquitin ligase (E3) of HRS was searched using UbiBrowser
2.0 ([305]http://ubibrowser.bio-it.cn/ubibrowser_v3/)^[306]31.
Plasmids
The coding sequences for full-length human CD63, PD-L1, and Munc13-4
were inserted into the NPY-td-Orange2 vector. Full-length human
Munc13-4, PD-L1, and various mutants of Munc13-4, including
K1062R/K1079R, K1062Q/K1079Q, V660A/K661A, and N739G/T740G, along with
truncated forms of Munc13-4 (residues 1–1048, 1–910, 1–782, 1–546,
1–287, and 1–108), were cloned into the pEGFP-N3 vector (Clontech).
Similarly, full-length human HRS, STAM, and Rab27a, along with Rab27a
mutants (F46S, W73S, and F88S), were cloned into the pEGFP-C1 vector
(Clontech).
Constructs for full-length human Munc13-4, PD-L1, HRS, and mutant
Munc13-4 (K1062R/K1079R, K1062Q/K1079Q), as well as truncated PD-L1
(residues 256–273), histone acetyltransferase domains of human GCN5
(residues 503–656), PCAF (residues 503–651), CBP (residues 1323–1700),
P300 (residues 127–1663), TIP60 (residues 227–504), and HBO1 (residues
332–607), and full-length human NEDD4, ITCH, CBL, SH3RF1, PRKN, NEDD4L,
STAMBPL1, CYLD, USP11, USP7, USP8, USP36, were generated in the
pcDNA3.1- vector (Invitrogen) with an N-terminal Flag-tag. Histone
deacetylase domains of human HDAC1 (residues 9–321), HDAC2 (residues
9–322), HDAC3 (residues 3–316), HDAC4 (residues 655–1084), HDAC5
(residues 684–1028), HDAC6 (residues 87–404 and 482–800), HDAC7
(residues 518–865), and HDAC8 (residues 14–324) were also cloned into
the pcDNA3.1- vector with an N-terminal HA-tag. Coding sequence of
full-length human NEDD4L was cloned into the pcDNA3.1- vector
(Invitrogen) with an N-terminal Strep-tag.
Full-length human Munc13-4, Rab27a, and their respective mutants,
including Munc13-4 (K1062R/K1079R, K1062Q/K1079Q, V660A/K661A,
N739G/T740G) and Rab27a (F46S, W73S, F88S), were cloned into the
pLV-EF1α-IRES-Hygro vector (Addgene). Full-length Munc13-4 and its
mutants (V660A/K661A, N739G/T740G) were subcloned into the
pFastBac^TMHT B vector (Invitrogen).
Coding sequences for full-length human Rab27a, syntaxin-4, truncated
syntaxin-4 (residues 1–275, Syx-4 ΔTM), SNAP-23 mutant (All cysteine
residues were mutated to serine, SN-23-6CS), SN-23-6CS (S161C), VAMP-7
SNARE motif (residues 123–187, A131C), VAMP-7 SNARE-TM motif (residues
123–208), and PD-L1 (residues 19–290) were cloned into the pET-28a
vector (Novagen). Finally, full-length Rab27a, Rab5, Rab7, and their
mutants (Rab27a (F46S, W73S, F88S)), Syx-4 ΔTM (1–275), SN-23-6CS, and
VAMP-7 SNARE motif (123–187) were subcloned into the pGEX-6P-1 vector
(Cytiva).
Transfection
Recombinant bacmid was transfected into Sf9 cells using the
X-tremeGENE™ 9 DNA Transfection Reagent (Roche). For all other plasmid
transfections, Hieff Trans® Liposomal Transfection Reagent (Yeasen,
40802ES08) was employed, following the manufacturer’s instructions.
Cells were prepared for subsequent experiments 24–36 h
post-transfection.
Cell viability assay
Cell proliferation of control and Munc13-4 knockout 4T1 cells was
assessed using a CCK-8 kit (Vazyme, A311-01) according to the
manufacturer’s instructions. Briefly, the same number of control and
Munc13-4 knockout cells were plated onto 96-well culture plates, and
cell viability was measured at 0, 24, and 48 h using the CCK-8 kit.
Histological analyses
Tissue arrays were deparaffinized with heat at 60 °C for 30 min
followed by two 15-min washes with xylene. Then, the paraffin sections
were rehydrated by washing for 5 min in absolute ethanol I, absolute
ethanol II, 85% alcohol, 75% alcohol, and distilled water in sequence.
Following the procedures outlined in the “PTLink Quick Operation Guide”
(Dako), slides were subjected to antigen retrieval using the specified
instrument. Upon completion, slides were immersed in distilled water at
room temperature for natural cooling for a minimum of 10 min.
Subsequently, the slides were rinsed with PBST buffer. The diluted
Munc13-4 primary antibody working solution (1:50, Santa Cruz,
sc-271300) was applied, and the slides were incubated overnight at
4 °C. The next day, the slides were removed from refrigeration and
allowed to equilibrate to room temperature for 45 min before being
washed with PBST buffer. Automated staining, including blocking,
secondary antibody binding, and DAB color development, was performed
using the DAKO automated immunohistochemistry staining system according
to the “Autostainer Link 48 User Guide.” Counterstaining was conducted
using hematoxylin for 1 min, followed by immersion in 0.25%
hydrochloric acid alcohol (prepared with 400 ml of 70% ethanol and 1 ml
of concentrated hydrochloric acid) for no less than 2 s. The slides
were rinsed under running water for a minimum of 2 min, air-dried at
room temperature, and mounted using neutral resin. Digitization of the
slides was performed at ×20 magnification using the Aperio XT Scanner
(Leica). To evaluate the protein expression levels of Munc13-4 in
breast cancer tissues and adjacent normal tissues, immunohistochemistry
(IHC) images were analyzed using IHC Profiler, an open-source plugin
for ImageJ. The staining intensity for Munc13-4 was scored using a
four-tier scoring system: 0 (negative), 1 (low positive), 2 (positive),
and 3 (high positive).
Samples of mouse heart, liver, spleen, lung, and kidney were fixed
overnight in 4% formalin, embedded in paraffin, and cut into 4 mm
consecutive sections. The paraffin sections were sequentially immersed
in Environmental-Friendly Dewaxing Transparent Liquids I and II
(Servicebio, G1128) for 20 min each, followed by treatment with
anhydrous ethanol I and II for 5 min each. Subsequently, the sections
were immersed in 75% ethanol for 5 min and thoroughly rinsed with tap
water. Hematoxylin and eosin (H&E) staining was performed using the
Hematoxylin-Eosin (H&E) HD Constant Dye Kit (Servicebio, G1076)
according to the manufacturer’s instructions. The sections were then
dehydrated through a graded series of absolute ethanol solutions (I,
II, and III) for 2 min each, followed by sequential immersion in normal
butanol I and II for 2 min each and clearing in xylene I and II for
2 min each. Finally, the sections were sealed with neutral gum and
scanned using the NanoZoomer S360 Digital Slide Scanner (Hamamatsu).
For tissue immunofluorescence assays, paraffin-embedded tumor tissues
from orthotopic mouse models of breast cancer sections were
deparaffinized and rehydrated through a graded ethanol series, followed
by washing in distilled water. Antigen retrieval was performed using
EDTA Antigen Retrieval Solution (Beyotime, P0085) under high
temperature and pressure conditions, and the sections were allowed to
cool to room temperature before washing in Tris-buffered saline with
0.05% Tween-20, pH 7.4 (TBST). Endogenous peroxidase activity was
blocked using 3% H2O2, followed by washing in distilled water. The
sections were then encircled with a hydrophobic pen and incubated with
10% goat serum (Boster, AR1009) at 37 °C for blocking. For the first
staining, a CD4 primary antibody (Abcam, RM1013, 1:50) diluted in TBST
was applied, and the sections were incubated overnight at 4 °C. After
washing, a secondary antibody, Goat Anti-Rabbit IgG H&L (HRP) (Abcam,
ab205718, 1:4000) was added and incubated at 37 °C. Tyramide signal
amplification (TSA) staining was performed using iFluor^® 488 tyramide
working solution (AAT Bioquest, 45100), followed by washing in TBST.
The slides underwent a second round of antigen retrieval in Improved
Citrate Antigen Retrieval Solution (Beyotime, P0083) using a microwave,
were cooled to room temperature, and washed again. Blocking was
repeated with 10% goat serum at 37 °C. For the second staining, a CD8
primary antibody (Abcam, RM1129, 1:100) diluted in TBST was applied,
and the sections were incubated overnight at 4 °C. Following washing,
the HRP-conjugated secondary antibody was added and incubated, and TSA
staining was conducted using Cy3 tyramide working solution (AAT
Bioquest, 11065). Nuclear staining was performed with DAPI (Solarbio,
C0060) in the dark, followed by washing in TBST. Finally, the sections
were mounted with Fluoromount-G^® (SouthernBiotech, 0100-01) and stored
at 4 °C in the dark. The sections were imaged using Pannoramic SCAN II
(3D HISTECH).
iTRAQ-based quantitative proteomics
4T1 cells (control and Munc13-4 knockout) were plated in triplicate for
proteomics. To extract proteins, an appropriate volume of SDS-free L3
buffer supplemented with final concentration of 1 × Cocktail (EDTA
contained) was added to the sample. The mixture was incubated on ice
for 5 min, followed by the addition of DTT to achieve a final
concentration of 10 mM. Ultrasonic disruption was performed to lyse the
sample, and the lysate was centrifuged at 25,000 × g and 4 °C for
15 min to remove insoluble debris. The supernatant was collected and
further treated with DTT (final concentration of 10 mM), followed by
incubation in a water bath at 56 °C for 1 h. Subsequently,
iodoacetamide was added to the solution to a final concentration of
55 mM, and the mixture was incubated in the dark for 45 min. A second
centrifugation at 25,000 × g and 4 °C for 15 min was performed, and the
supernatant, containing the extracted protein solution, was collected
for downstream analyses.
Protein samples (100 µg each) were digested with Trypsin Gold (Promega,
V5280) at a protein-to-trypsin ratio of 20:1 (w/w) at 37 °C for 16 h.
The resulting peptides were dried via vacuum centrifugation and
reconstituted in 0.5 M TEAB. iTRAQ labeling was conducted following the
manufacturer’s protocol for the 4-plex iTRAQ reagent kit
(Sigma-Aldrich, 4374321). The labeled samples were combined in equal
proportions and fractionated using high-performance liquid
chromatography (HPLC) on a Thermo DIONEX Ultimate 3000 BioRS system
equipped with a Durashell C18 column (5 µm, 100 Å, 4.6 × 250 mm, Welch
Materials). A total of 20 fractions were collected for further
analysis.
Peptides separated by liquid chromatography were ionized using a
nanoESI source and analyzed on a Q-Exactive HF X mass spectrometer
(Thermo Fisher Scientific) operating in data-dependent acquisition
(DDA) mode. Instrument parameters were configured as follows: the ion
source voltage was set to 1.9 kV; the MS1 scan range was 350–1,500 m/z
with a resolution of 60,000; and the MS2 scan range started at a fixed
m/z of 100 with a resolution of 15,000. Precursor ion selection
criteria included charge states between 2+ and 6+ and the top 20 most
intense ions with signal intensities exceeding 10,000. Fragmentation
was performed using higher-energy collisional dissociation (HCD), and
the resulting fragments were detected in the Orbitrap analyzer. The
dynamic exclusion duration was set to 30 s, and the automatic gain
control (AGC) targets were 3 × 10^6 for MS1 and 1 × 10^5 for MS2.
Protein identification was conducted utilizing the Mascot search engine
(version 2.3.02; Matrix Science) against the Mus musculus subset of the
NCBI non-redundant (NR) sequence databases. The search parameters were
configured as follows: monoisotopic mass, peptide mass tolerance of
20 ppm, fragment mass tolerance of 0.05 Da, trypsin as the digestion
enzyme, allowance for one missed cleavage, and charge states of +2 and
+3 for peptides. Variable modifications included Gln-> pyro-Glu
(N-terminal Q), oxidation (M), and deamidation (NQ), while fixed
modifications comprised carbamidomethylation (C) and iTRAQ8plex
labeling (N-terminal and K). Protein quantification was performed using
the automated software IQuant. Peptides with a confidence interval of
95% were filtered based on a 1% false discovery rate (FDR), and
confident proteins were required to include at least one unique
peptide. Quantitative protein ratios were weighted and normalized using
the median ratio in Mascot. Differentially expressed proteins (DEPs)
between the control and Munc13-4 knockout groups were identified via
t-tests, with results subjected to a 5% FDR correction. Proteins with
expression fold changes ≥1.2 or ≤0.83 were classified as DEPs.
Additionally, KEGG pathway enrichment analysis was performed for DEPs,
and a heatmap was generated using an online platform for data analysis
and visualization ([307]https://www.bioinformatics.com.cn/)^[308]67.
Cell treatment with IFNγ, PR-619, and peptide
To explore the effects of IFNγ on the expression and exosomal sorting
of PD-L1, 100 ng/ml Recombinant Human IFN-gamma Protein (Abclonal,
RP01038) was added to SUM159 cells, and 100 ng/ml Recombinant Mouse
IFN-gamma Protein (Abclonal, RP01070) was added to 4T1 cells,
incubating for 24 h. To examine the effect of HRS ubiquitylation on the
sorting of PD-L1 onto exosomes under IFNγ stimulation, SUM159 cells
were co-treated with 100 ng/ml Recombinant Human IFN-gamma Protein
(Abclonal, RP01038) and 8 μM PR-619 (MCE, HY-13814) for 24 h. To
investigate the effects of the designed peptide on protein interactions
and exosomal PD-L1 sorting, HEK293T and SUM159 cells were incubated
with 10 μg/ml P-pep (Homo sapiens) or S-pep.
Western blot
Cells or exosomes were lysed on ice in RIPA buffer (50 mM Tris-HCl, pH
7.5, 150 mM NaCl, 1% Triton X-100) supplemented with a protease
inhibitor cocktail (Topscience, C0001). Following a 20-min incubation,
the lysates were centrifuged at 4 °C, 12,000 × g for 10 min. The total
protein concentration in the supernatant was determined using a BCA
Protein Quantification Kit (Yeasen, 20201ES76) to ensure consistent
loading of different samples. Protein samples were then denatured by
heating in diluted 1× SDS-PAGE Sample Loading Buffer (Yeasen,
20315ES05) for 10 min at 100 °C. Proteins were separated by SDS-PAGE
and subsequently transferred to PVDF membranes (Millipore, ISEQ00010).
The membranes were blocked with 5% non-fat bovine milk in Tris-Cl
buffer (150 mM NaCl, pH 7.2) containing 0.1% Tween-20, followed by
incubation with the indicated primary antibody and a subsequent
incubation with an HRP-conjugated secondary antibody. Immunodetection
was carried out using the Super Sensitive ECL Luminescence Reagent
(Meilunbio, MA0186-2). The integrated density of the blot bands was
quantified and analyzed using ImageJ and Prism 6.0 software to assess
the relative protein levels. The primary antibodies used in western
blot assays are as follows: Munc13-4 Antibody (C-2) (Santa Cruz,
sc-271300, 1:1000, for samples of human origin), UNC13D Monoclonal
antibody (Proteintech, 67193-1-Ig, 1:5000, for samples of mouse
origin), PD-L1/CD274 Rabbit mAb (Abclonal, A19135, 1:2000, for samples
of human origin), Anti-PD-L1 antibody (abcam, ab213480, 1:2000, for
samples of mouse origin), HGS Polyclonal antibody (Proteintech,
10390-1-AP, 1:10000, for samples of both human and mouse origin),
β-Actin Rabbit mAb (High Dilution) (Abclonal, AC026, 1:100000, for
samples of both human and mouse origin), Alix Monoclonal antibody
(Proteintech, 67715-1-Ig, 1:5000, for samples of both human and mouse
origin), CD63 Antibody (MX-49.129.5) (Santa Cruz, sc-5275, 1:1000, for
samples of both human and mouse origin), CD81 Antibody (B-11) (Santa
Cruz, sc-166029, 1:1000, for samples of both human and mouse origin),
Mouse anti-GFP-Tag mAb (Abclonal, AE012, 1:10000, species independent),
Rabbit anti-GFP-Tag pAb (Abclonal, AE011, 1:10000, species
independent), DYKDDDDK tag Polyclonal antibody (Proteintech,
20543-1-AP, 1:10000, species independent), StrepII Tag Mouse Monoclonal
Antibody (Beyotim, AF2924, 1:1000, species independent), Mouse anti
HA-Tag mAb (Abclonal, AE008, 1:5000, species independent), Pan
Acetylation Monoclonal antibody (Proteintech, 66289-1-Ig, 1:1000,
species independent), Ubiquitin Antibody (P4D1) (Santa Cruz, sc-8017,
1:1000, for samples of both human and mouse origin), pan
Phospho-Serine/Threonine Rabbit Polyclonal Antibody (Beyotim, AF5725,
1:1000, species independent), Pan Phospho-Tyrosine Mouse mAb (Abclonal,
AP0973, 1:1000, for samples of both human and mouse origin),
CBP/KAT3A/CREBBP Antibody (C-1) (Santa Cruz, sc-7300, 1:1000, for
samples of both human and mouse origin), P300 Antibody (F-4) (Santa
Cruz, sc-48343, 1:1000, for samples of both human and mouse origin),
Histone Deacetylase 3 (HDAC3) Antibody (A-3) (Santa Cruz, sc-376957,
1:1000, for samples of both human and mouse origin), Histone
Deacetylase 4 (HDAC4) Antibody (B-5) (Santa Cruz, sc-365093, 1:1000,
for samples of both human and mouse origin), NEDD4L Polyclonal antibody
(Proteintech, 13690-1-AP, 1:3000, for samples of both human and mouse
origin). The secondary antibodies used in western blot experiments are
as follows: HRP-conjugated Goat anti-Rabbit IgG (H+L) (Abclonal, AS014,
1:10000), HRP-conjugated Goat anti-Mouse IgG (H+L) (Abclonal, AS003,
1:10000). The integrated density of blot strips was analyzed by ImageJ
software to characterize the relative protein level.
Generation of gene-edited cell lines
To generate gene knockout cell lines, the CRISPR-Cas9 system was
employed. HEK293T cells were transfected with the lentiCRISPR v2
plasmid, which contained a single-guide RNA (sgRNA) targeting the gene
of interest, along with the psPAX2 and pMD2.G plasmids to produce
lentivirus. After 36–48 h of transfection, the culture medium of the
HEK293T cells was collected, centrifuged to remove cell debris, and the
supernatant containing the lentivirus-sgRNA was used to infect SUM159
or 4T1 cells. Following 48 h of infection, cells were selected with
puromycin (MCE, HY-B1743) at a concentration of 2.0 μg/mL for 5–7 days.
Limiting dilution was then performed to isolate single-cell clones from
the infected SUM159 or 4T1 cells. Cells infected with lentivirus
produced by HEK293T cells co-transfected with an empty lentiCRISPR v2
vector, psPAX2, and pMD2.G plasmids served as controls. Successful
knockout of the target SNARE protein was verified by western blot
analysis.
The sgRNAs used in this study are as follows:
Mus-Munc13-4: GTGGCCTTCAGGCAAAATAC
Hs-Munc13-4: TGAAGGTCTCGTCCCAGACG
Hs-Rab27a: CCAAAGCTAAAAACTTGATG
Hs-HRS: CTGCCTGCAGAGACAAGTGG
Hs-P300: GTTCAATTGGAGCAGGCCGA
Hs-CBP: CGCGTGACCAGTCATTTGCG
Hs-HDAC3: GGTGAAGCCTTGCATATTGG
Hs-HDAC4: GGAGCCCATTGAGAGCGATG
Hs-NEDD4L: GGAGCCCATTGAGAGCGATG
For the generation of gene-complemented cell lines, the
pLV-EF1a-IRES-Hygro plasmid containing the full-length sequence of the
gene of interest was utilized. HEK293T cells were transfected with the
pLV-EF1a-IRES-Hygro plasmid along with the psPAX2 and pMD2.G plasmids
to produce lentivirus. At 36–48 h post-transfection, the culture medium
was collected, centrifuged to remove cell debris, and the resulting
lentivirus-containing supernatant was used to infect the gene knockout
SUM159 cells. Infected cells were selected using Hygromycin B (Sangon
Biotech, A100607) at a final concentration of 500 μg/ml for 5–7 days.
Single-cell clones were subsequently isolated through limiting
dilution.
Immunoprecipitation (IP) and co-IP
To investigate the post-transcriptional modifications of Munc13-4 and
HRS, IP assays were conducted. Equal numbers of SUM159 cells were
seeded onto 15-cm culture dishes, with one plate treated with 100 ng/ml
IFNγ (ABclonal, RP01038) for 24 h. Following the incubation, cells were
lysed on ice using RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1%
Triton X-100) supplemented with a protease inhibitor cocktail
(Topscience, C0001). The lysates were then centrifuged at 4 °C for
10 min at 12,000 × g. The supernatant was incubated overnight at 4 °C
with Munc13-4 Antibody (C-12) (Santa Cruz, sc-271301, 1:100) or HGS
Polyclonal antibody (Proteintech, 10390-1-AP, 1:500), in conjunction
with Protein A/G magnetic beads (Biolinkedin, L-1004), on a rotator. To
identify the acetyltransferase of Munc13-4, GFP-tagged Munc13-4 was
co-expressed with Flag-labeled GCN5 503–656, PCAF 503–651, CBP
1323–1700, P300 127–1663, TIP60 227–504, or HBO1 332–607 in HEK293T
cells. To investigate the type of deacetylase, HEK293T cells
co-expressing GFP-Munc13-4 and Flag-CBP 1323–1700 were treated with
0.2% DMSO, 5 mM nicotinamide (NIA) (MCE, HY-B0150) or 1 μM trichostatin
A (TSA) (MCE, HY-15144) for 24 h. To identify the deacetylase of
Munc13-4, GFP- Munc13-4 and Flag- CBP 1323–1700 was co-expressed with
HA-fused HDAC1 9–321, HDAC2 9–322, HDAC3 3–316, HDAC4 655–104, HDAC5
684–1028, HDAC6 87–404, HDAC6 482–800, HDAC7 518–865 or HDAC8 14–324 in
HEK293T cells. To identify the E3 ligase of HRS, GFP-HRS was
co-expressed with Flag-tagged NEDD4, ITCH, CBL, SH3RF1, PRKN, or NEDD4L
in HEK293T cells. To identify the deubiquitinase of HRS, GFP-HRS and
strep-NEDD4L were co-expressed with Flag-fused STAMBPL1, CYLD, USP11,
USP7, USP8, or USP36 in HEK293T cells. The transfection of plasmids
mentioned above was performed using Hieff Trans^® Liposomal
Transfection Reagent (Yeasen, 40802ES03) according to the
manufacturer’s protocol.GFP-Munc13-4 or GFP-HRS was immunoprecipitated
from cell lysates using anti-GFP magnetic beads (Biolinkedin, L-1016)
according to the manufacturer’s guidance. After incubation, the beads
were washed three times with PBS. Subsequently, 1× SDS-PAGE Sample
Loading Buffer (Yeasen, 20315ES05) was added, and the samples were
heated at 100 °C for 10 min. The protein samples were then analyzed by
western blotting.
The deletion of either CBP or HDAC3 in SUM159 cells resulted in a
significant decrease in Munc13-4 expression. To examine the effects of
CBP and HDAC3 knockout on Munc13-4 acetylation, Munc13-4 was enriched
from a cell population three times larger in the knockout groups
compared to the control cells. The amount of immunoprecipitated
Munc13-4 used for acetylation analysis was standardized based on the
results of preliminary western blot analysis.
To investigate protein-protein interactions, co-IP assays were
conducted. Recombinant plasmids were transfected into SUM159 or HEK293T
cells using Hieff Trans^® Liposomal Transfection Reagent (Yeasen,
40802ES03). GFP-fused proteins were then enriched from cell lysates
using anti-GFP magnetic beads (Biolinkedin, L-1016). Following
enrichment, the beads were washed three times with PBS and subsequently
heated in 1× SDS-PAGE Sample Loading Buffer (Yeasen, 20315ES05) at
100 °C for 10 min. Next, the samples were analyzed by western blotting.
Isolation of EVs/exosomes
For extracellular vesicle (EV)/exosome preparation, equal numbers of
cells with the indicated genotypes were seeded onto 15-cm dishes. Once
the cells adhered, they were rinsed with PBS and cultured for 24 h in
DMEM supplemented with 10% exosome-depleted FBS (prepared by
ultracentrifugation at 100,000 × g overnight to remove bovine
vesicles). Conditioned medium was collected and subjected to stepwise
centrifugation to eliminate contaminants. Specifically, the medium was
centrifuged at 300 × g for 10 min at 4 °C to remove intact cells,
followed by 2000 × g for 20 min to eliminate cellular debris, and then
10,000 × g for 30 min to deplete larger vesicles. The clarified
supernatant was ultracentrifuged at 100,000 × g for 70 min (Beckman
Type 70Ti rotor) to pellet exosomes. The pellet was washed once with
cold PBS and centrifuged again at 100,000 × g for 70 min to enhance
purity. The final exosome pellet was resuspended in PBS or RIPA buffer
for downstream assays.
Optiprep^TM density gradient centrifugation
Pellets of EVs, obtained through ultracentrifugation from cell culture
supernatants, were washed and resuspended in 200 µL of buffer
containing 0.25 M sucrose, 10 mM Tris-Cl, and 1 mM EDTA (pH 7.4). The
suspension was then transferred to an SW55Ti rotor tube (Beckman,
344090), mixed in a 1:1 ratio with a 60% (wt/vol) Optiprep™ stock
solution, and sequentially layered with 160 µL of a 20% (wt/vol)
Optiprep™ solution and 150 µL of a 10% (wt/vol) Optiprep™ solution.
Tubes were centrifuged for 1 h at 4 °C, 350,000 × g in an SW55Ti rotor
(stopping without break). Following centrifugation, six 100 µL
fractions were collected from the top of the gradient. These fractions
were diluted with 600 µL of PBS and subjected to a second round of
centrifugation for 1 h at 4 °C, 100,000 × g. The resulting pellets from
the concentrated fractions were resuspended in 20 µL of PBS and
analyzed by western blotting.
Transmission electron microscopy (TEM)
The morphology of EVs was characterized by TEM. 20 μL of EVs suspension
was carefully deposited onto a copper grid (EMCN, BZ11262a) and
incubated for 3–5 min. Excess liquid was then removed with filter
paper. Subsequently, 2% uranyl acetate was applied to the copper grid
for 2–3 min, after which the excess solution was absorbed using filter
paper, and the sample was allowed to air-dry at room temperature. The
samples were then observed with TEM (HITACHI, HT7800). The number of
ILVs and MVBs and the percentage of MVB-lysosome hybrids among total
MVBs were scored manually, and the diameter of MVBs was measured by
ImageJ software.
Nanoparticle tracking analysis (NTA)
Exosome size distribution and concentration were assessed using a
NanoSight NS300 system (Malvern) equipped with a high-sensitivity sCMOS
camera. Samples were recorded in triplicate, with three 60-s videos
collected for each preparation. The captured data were analyzed using
NanoSight NTA software, which determines particle size and
concentration based on tracking Brownian motion and calculating the
corresponding diffusion coefficients.
Tunable resistive pulse sensing (TRPS)
Tunable resistive pulse sensing (TRPS) was performed using the
Nanocoulter G system (Resun Technology, China) equipped with a nanopore
chip capable of detecting particles in the 50–250 nm size range. This
technique was employed to measure both the concentration and size
distribution of exosomes.
nanoLCMS/MS analysis
nanoLCMS/MS analysis was performed for the identification of Munc13-4
acetylation. 24 h following the co-expression of GFP-Munc13-4 and
Flag-CBP 1323–1700 in HEK293T cells, GFP-Munc13-4 was
immunoprecipitated from cell lysates using anti-GFP magnetic beads
(Biolinkedin, L-1016), following the manufacturer’s protocol. The beads
were subsequently washed three times with PBS, after which diluted 1×
SDS-PAGE Sample Loading Buffer (Yeasen, 20315ES05) was added, and the
samples were heated at 100 °C for 10 min. Proteins separated by
SDS-PAGE were subjected to trypsin digestion (Promega, V5280) in 100 mM
NH4HCO3 overnight at 37 °C. The resulting peptides were extracted with
extraction buffer (1:2, vol/vol, 5% formic acid/acetonitrile) and then
vacuum-dried.
A total of 200 ng of peptides were separated and analyzed using a
nano-UPLC system (Evosep One) coupled to a timsTOF Pro2 mass
spectrometer (Bruker) equipped with a nano-electrospray ionization
source. Peptide separation was achieved on a reversed-phase column
(PePSep C18, 1.9 µm, 150 µm × 15 cm, Bruker) with mobile phases
consisting of H2O containing 0.1% formic acid (phase A) and
acetonitrile (ACN) with 0.1% formic acid (phase B). A 44-min gradient
was used for the separation. Data acquisition was performed in DDA
PaSEF mode, with the mass spectrometer scanning within a range of 100
to 1700 m/z for MS1. During PASEF MS/MS acquisition, the collision
energy was linearly increased in correlation with ion mobility, ranging
from 20 eV (1/K0 = 0.6 Vs/cm²) to 59 eV (1/K0 = 1.6 Vs/cm²).
The raw MS data files provided by the vendor were processed using
SpectroMine software (version 4.2.230428.52329) in conjunction with the
integrated Pulsar search engine. The MS spectra were queried against
the species-specific UniProt FASTA database for Homo sapiens
(uniprot_Homo sapiens_9606_reviewed_2023_09.fasta), with
carbamidomethylation of cysteine (C) set as a fixed modification, and
oxidation (M) and acetylation at the protein N-terminus as variable
modifications. Trypsin was employed as the protease, with a maximum
allowance of two missed cleavages. A false discovery rate (FDR) of 0.01
was applied at both the peptide-spectrum match (PSM) and peptide
levels. Peptide identification was performed with an initial precursor
mass tolerance of 20 ppm. All other parameters were left at their
default settings.
Proximity ligation assay (PLA)
SUM159 cells were plated onto the wells of 29-mm glass-bottom dishes
(Cellvis, D29-10-0-N). After fixation with 4% (wt/vol) paraformaldehyde
at room temperature for 15 min, the cells were permeabilized with 0.2%
(vol/vol) Triton X-100 in PBS for 10 min at room temperature, followed
by three washes with PBS. Subsequently, in situ proximity ligation
assays were performed using the Duolink^® In Situ Red Starter Kit
Mouse/Rabbit (Sigma-Aldrich, DUO92101), following the manufacturer’s
protocol. This included blocking, primary antibody incubation,
Duolink^® PLA probe incubation, ligation, amplification, final washing,
and nuclear staining in sequence. The cell samples were imaged using an
FV3000 Confocal Laser Scanning Microscope (Olympus) equipped with a 60×
oil-immersion objective (NA 1.42). DAPI was excited using a 405 nm
laser, and Duolink^® In Situ Detection Reagents Red were excited with a
594 nm laser. To determine the interaction between Munc13-4 and PD-L1,
Munc13-4 Antibody (C-2) (Santa Cruz, sc-271300, 1:200) and PD-L1/CD274
(C-terminal) Polyclonal antibody (Proteintech, 28076-1-AP, 1:200) were
used. To explore the interaction between HRS and PD-L1, HGS Polyclonal
antibody (Proteintech, 10390-1-AP, 1:500) and PD-L1/CD274 Monoclonal
antibody (Proteintech, 66248-1-Ig, 1:200) were employed. Data were
processed by home-written MATLAB script
([309]https://github.com/shenwang3333/PLA_Counting).
Orthotopic mouse models of breast cancer and treatments
Mice were anesthetized via intraperitoneal injection of pentobarbital
sodium at a dose of 40 mg/kg. Orthotopic breast cancer models were
established by injecting control or Munc13-4 knockout 4T1 cells
(3 × 10^5 cells per mouse) into the right fourth mammary fat pad of
BALB/c, BALB/c Nude, or NOD/SCID female mice. Tumor dimensions were
measured every other day starting on either day 4 or day 6
post-inoculation using a digital caliper, and tumor volume was
calculated using the formula: (width² × length × 0.5). At the end of
the observation period, mice were euthanized, and tumors were
harvested, weighed, and photographed for further analysis. According to
the IACUC guidelines, the maximal allowable tumor size in adult mice is
20 mm in diameter in any direction. In this study, the tumor
size/burden in all experimental animals did not exceed this limit.
Exosomes (1 × 10^9 particles) secreted by control 4T1 cells were
pre-incubated with either an IgG isotype control antibody (Bioxcell,
BE0090, 1:100) or InVivoMAb anti-mouse PD-L1 antibody (Bioxcell,
BE0101, 1:100). To remove unbound antibodies, the exosome-antibody
complexes were subjected to ultracentrifugation at 100,000 × g for
70 min at 4 °C. The purified exosomes were then intravenously injected
into mice via the tail vein. Injections were administered every other
day for a total of nine treatments.
Peptide treatments were performed using P-pep (Mus musculus) or S-pep,
administered via intraperitoneal injection at a dosage of 100 µg per
mouse. Once the tumor volume reached approximately 100 mm³, peptides
were administered every other day, with a total of six injections.
For immune checkpoint blockade (ICB) studies, mice were treated with
IgG isotype control antibody (Bioxcell, BE0090, 100 µg/mouse),
InVivoMAb anti-mouse PD-L1 antibody (Bioxcell, BE0101, 100 µg/mouse),
or InVivoMAb anti-mouse PD-1 antibody (Bioxcell, BE0146, 100 µg/mouse).
Antibodies were administered via intraperitoneal injection every 3 days
for a total of six treatments.
Immune profiling
To analyze T cell infiltration and activation, tumors, spleens, and
tumor-draining lymph nodes (TDLNs) were excised from orthotopic mouse
models of breast cancer. Tumor tissues were minced into small pieces
and incubated with RPMI 1640 medium containing 1 mg/ml collagenase D
(Roche, COLLD-RO) and 0.2 mg/ml DNase I (BioFroxx, 112MG010) at 37 °C
for 1 h, followed by mechanical dissociation using a mesh cell
strainer. The cells were then centrifuged at 500 × g for 5 min, washed
with PBS, and treated with red blood cell (RBC) lysis buffer (BD
Biosciences, 555899) to remove RBCs. The resulting cell suspension was
filtered twice through a 70 μm nylon mesh to obtain single-cell
suspensions. Lymphocytes from the spleens and TDLNs were isolated by
mechanically squashing the tissues through a 70 μm mesh and removing
RBCs.
For stimulation, the single-cell suspensions were incubated with RPMI
1640 medium containing Leukocyte Activation Cocktail (BD Biosciences,
550583, 1:1000) at 37 °C for 6 h. After staining with viability dye FVS
575V (BD Biosciences, 565694, 1:1000) to exclude dead cells, the cells
were stained with the following antibodies. For surface marker
analysis, cells were stained with anti-CD45-APC-Cy7 (BD Biosciences,
557659, 1:50), anti-CD3-BV421 (BD Biosciences, 562600, 1:50),
anti-CD4-Alexa Fluor 700 (BD Biosciences, 557956, 1:50), and
anti-CD8-Percp-Cy5.5 (BD Biosciences, 551162, 1:50). For intracellular
cytokine staining, cells were fixed and permeabilized using Fix/Perm
buffer (BD Biosciences, 562574) and Perm/Wash buffer (BD Biosciences,
562574), then re-stained with anti-KI67-BV510 (BD Biosciences, 563462,
1:50), anti-IFN-γ-BV650 (BD Biosciences, 563854, 1:50), or
anti-Granzyme B-FITC (Invitrogen, 11-8898-82, 1:500). Flow cytometric
analysis was performed using the CytoFLEX flow cytometer.
CD8^+ T cell suppression assay
To block PD-L1 on the exosome surface, equal quantities of exosomes
isolated from control and Munc13-4 knockout 4T1 cells were incubated
with PD-L1 blocking antibodies (Bioxcell, BE0101, 1:100) or IgG isotype
control antibody (Bioxcell, BE0090, 1:100) at room temperature for 5 h.
After incubation, the exosomes were washed with 25 ml PBS and subjected
to ultracentrifugation to remove unbound antibodies.
Mouse CD8^+ T cells were purified from splenocytes using the Mouse CD8
T Cell Isolation Kit (Vazyme, CS103-01) and stimulated for 24 h with
anti-CD3 (Biolegend, 300301, 1 μg/ml) and anti-CD28 (Biolegend, 117003,
1 μg/ml) antibodies. Post-stimulation, the CD8⁺ T cells were incubated
with the pre-processed exosomes for 16 h in the continued presence of
anti-CD3 and anti-CD28 antibodies.
Following treatment, CD8^+ T cells were harvested, stained with
anti-CD8-Percp-Cy5.5 (BD Biosciences, 551162, 1:50), and permeabilized
using Fix/Perm buffer and Perm/Wash buffer (BD Biosciences, 562574).
Fixed cells were subsequently stained with anti-Granzyme B-FITC
(Invitrogen, 11-8898-82, 1:500) and analyzed by flow cytometry.
T cell-mediated tumor cell killing assay
Spleens were aseptically harvested from BALB/c mice and placed in
sterile petri dishes containing cold PBS. The spleens were gently
disrupted by grinding against a 70 μm cell strainer using a sterile
syringe plunger, and the resulting cell suspension was centrifuged at
500 × g for 5 min at 4 °C. The pellet was resuspended in 5 ml of PBS,
and the suspension was carefully layered onto 5 ml of Ficoll-Paque™
PLUS (Cytiva, 17144002) in a 15 ml conical tube without mixing. The
gradient was centrifuged at 1000 × g for 20 min at room temperature
with the centrifuge brake turned off. The mononuclear cell layer at the
interface was carefully collected using a pipette, transferred to a
fresh tube, and washed twice with PBS by centrifugation at 500 × g for
5 min at 4 °C to remove residual Ficoll. The purified lymphocytes were
then resuspended and stimulated with anti-CD3 (Biolegend, 300301,
1 μg/ml) and anti-CD28 (Biolegend, 117003, 1 μg/ml) antibodies for
24 h. Subsequently, the stimulated lymphocytes were co-cultured with
adherent control or Munc13-4 knockout 4T1 cells in 96-well plates at an
effector-to-target (E: T) ratio of 1:1 for 24 h. The viability of
control and Munc13-4 knockout 4T1 cells was assessed using the Cell
Counting Kit-8 (Vazyme, A311-01) following the manufacturer’s
instructions.
Quantitative reverse transcription (qRT)-PCR assay
Total RNA was extracted from each sample using the AFTSpin Tissue/Cell
Fast RNA Extraction Kit for Animal (Abclonal, RK30120). The isolated
RNA was eluted in nuclease-free water and reverse-transcribed into
complementary DNA (cDNA) using the ABScript II cDNA First-Strand
Synthesis Kit (Abclonal, RK20400). The resulting cDNA samples were
subjected to quantitative PCR on a QuantStudio™ 6 Pro Real-Time PCR
system using SYBR Green (Abclonal, RK21203) for detection.
The primers used for qPCR were as follows:
Hs-Munc13-4 qPCR forward primer: CCCTTTGTCCAGCTGACCTT
Hs-Munc13-4 qPCR reverse primer: AGCAGGCACCAGGAATTCAA
Hs-Actin qPCR forward primer: GCCGCCAGCTCACCAT
Hs-Actin qPCR reverse primer: AGGAATCCTTCTGACCCATGC
Fluorescence imaging
To analyze the colocalization between PD-L1 and CD63/LAMP1, Munc13-4
knockout and control SUM159 cells were seeded onto glass-bottom dishes
and transfected with the pEGFP-N3-PD-L1 plasmid along with either
NPY-td-Orange2-CD63 or NPY-td-Orange2-LAMP1 plasmids. Following 24–36 h
of transfection, cells were fixed with 4% paraformaldehyde (PFA) for
15 min. For immunofluorescence staining, cells were permeabilized with
0.2% Triton X-100 for 10 min and subsequently blocked with 5% bovine
serum albumin (BSA) for 1 h at room temperature. After blocking, cells
were incubated with primary antibodies overnight at 4 °C and washed
three times with PBS. Secondary antibody staining was performed for 2 h
at room temperature, followed by three PBS washes. Imaging was
conducted using a Nikon confocal microscope equipped with a 60×
oil-immersion objective lens (NA 1.40). The primary antibodies used in
immunofluorescence experiments are as follows: CD63 Antibody
(MX-49.129.5) (Santa Cruz, sc-5275, 1:100), LAMP1/CD107a Rabbit mAb
(Abclonal, A21194, 1:200) and CBP/KAT3A/CREBBP Antibody (C-1) (Santa
Cruz, sc-7300, 1:200). The secondary antibodies used in
immunofluorescence experiments are as follows: Goat anti-Mouse IgG
(H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488 (Invitrogen,
A-11001, 1:500), Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed
Secondary Antibody, Alexa Fluor™ Plus 647 (Invitrogen, A-32733, 1:500).
Fluorescence intensity was quantified using NIS-Elements AR 4.40
software, and colocalization was assessed by calculating the Pearson’s
correlation coefficient, also employing NIS-Elements AR 4.40 software.
TIRF microscopy for monitoring MVB mobility
The mobility of exosomes in live cells was assessed using total
internal reflection fluorescence (TIRF) microscopy (Nikon). To evaluate
the effects of mutations in Munc13-4 on MVB mobility, control and
Munc13-4 knockout SUM159 cells were seeded onto glass-bottom dishes.
These cells were subsequently co-transfected with NPY-td-Orange2-CD63
and pEGFP-N3-Munc13-4 WT or Munc13-4 mutants. Similarly, to examine the
effects of Rab27a mutations on MVB mobility, control or Rab27a knockout
SUM159 cells were plated and co-transfected with NPY-td-Orange2-CD63
and pEGFP-C1-Rab27a WT or Rab27a mutants. After 24–48 h
post-transfection, live-cell imaging was performed using a Nikon Ti
inverted TIRF microscopy system equipped with a 100× oil-immersion
objective (NA 1.49) and an EMCCD camera (Andor DU897). Orange
fluorescence was excited using a 532 nm laser with an exposure time of
300 ms. The diffusion coefficient (D), representing MVB mobility, was
quantified using the Python-based trackpy library
([310]https://soft-matter.github.io/trackpy/dev/tutorial/walkthrough.ht
ml).
Protein expression and purification
Protein expression of human Munc13-4 and its mutants was performed
using the Bac-to-Bac™ baculovirus expression system (Invitrogen).
Briefly, recombinant bacmid DNA extracted from DH10Bac was transfected
into Spodoptera frugiperda clone 9 (Sf9) cells using X-tremeGENE™ 9 DNA
Transfection Reagent (Roche) to produce P1 baculovirus. Sequential
infection of Sf9 cells with P1 baculovirus generated P2 and P3
baculoviruses. For protein expression, Sf9 cells were infected with P3
baculovirus and cultured for 48 h. Harvested cells were resuspended in
lysis buffer (20 mM Tris-HCl, pH 8.1, 150 mM NaCl) supplemented with
protease inhibitors (2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 μg/ml
pepstatin, and 1 mM PMSF). Cells were lysed using an AH-1500 Nano
Homogenizer (ATS Engineering Inc.) at 800 bar under 4 °C. The lysate
was clarified by centrifugation at 16,000 rpm using a JA-25.50 rotor
(Beckman Coulter) at 4 °C. Supernatants were incubated with
nickel-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen) for 1 h at 4 °C,
followed by two washes with wash buffer (20 mM Tris-HCl, pH 8.1, 150 mM
NaCl, 20 mM imidazole). Bound proteins were eluted using buffer
containing 20 mM Tris-HCl, pH 8.1, 150 mM NaCl, and 300 mM imidazole.
Eluted proteins were further purified by size-exclusion chromatography
using a Superdex™ 200 10/300 GL column (Cytiva).
Other proteins were expressed in Escherichia coli BL21 (DE3). Protein
expression was induced with isopropyl β-D-1-thiogalactopyranoside
(IPTG). Harvested cells were lysed as described above. For GST-tagged
proteins, clarified lysates were incubated with glutathione Sepharose
4B (GE Healthcare), washed, and eluted using buffer containing 20 mM
glutathione (neoFroxx, 1392GR025), 20 mM Tris-HCl, pH 8.1, and 150 mM
NaCl. His-tagged proteins were purified as described for Munc13-4.
Purified proteins were further processed using ion exchange and
size-exclusion chromatography. For transmembrane proteins, 1.5% sodium
deoxycholate was included throughout the purification process. Proteins
were used immediately after affinity purification.
For cryo-EM complex preparation, Munc13-4 and Rab27a were mixed at a
molar ratio of 1:2 in the presence of 1 mM GppNHp (Aladdin, G276465)
and incubated at room temperature for 1 h. The mixture was subjected to
size-exclusion chromatography to isolate the stable protein complex.
GST pull-down assay
For all GST pull-down assays, 2 μM of the GST-tagged protein was
incubated with 3 μM of the target protein at room temperature for 2 h.
Subsequently, the protein mixture was combined with glutathione
Sepharose 4B resin (GE Healthcare) and incubated at 4 °C for 1 h. The
resin was then washed four times with wash buffer (20 mM Tris-HCl, pH
8.1, 150 mM NaCl, 0.02% Triton X-100) to remove unbound proteins. Bound
proteins were eluted using an elution buffer containing 20 mM Tris-HCl
(pH 8.1), 150 mM NaCl, 0.02% Triton X-100, and 50 mM glutathione.
Eluted proteins were analyzed by SDS-PAGE.
SNARE assembly assay
For SNARE assembly assay, VAMP-7 SNARE motif (A131C) was labeled with
the Förster resonance energy transfer (FRET) donor dye BODIPY FL
(Molecular Probes, [311]B10250), and SN-23-6CS (S161C) was labeled with
the FRET-acceptor dye 5-tetramethylrhodamine (5-TAMRA) (Molecular
Probes, T6027). During the experiment, both the donor protein and
acceptor protein were at a concentration of 0.5 µM. The concentration
of Syx-4 ΔTM (residues 1–275) was 2 µM, and Munc13-4 was at 10 µM. The
experiments were performed using a PTI QM-40 spectrophotometer, with an
excitation wavelength of 485 nm and an emission wavelength of
513/580 nm, at room temperature. SNARE complex formation signals were
interpreted as the FRET proximity ratio (E[PR]) between the donor
(BODIPY) and acceptor (5-TAMRA). The E[PR] was determined using
Eq. [312]1:
[MATH:
EPR=I5−
TAMRAI5−T
AMRA+IBODIPY :MATH]
1
where I[5-TAMRA] and I[BODIPY] represent the fluorescence intensities
of 5-TAMRA and BODIPY FL, respectively, measured under the 485/10
excitation filter.
Liposome fusion assay
All lipids were dissolved at an initial concentration of 10 mg/ml in
chloroform, except for PI(4,5)P₂, which was dissolved in a chloroform:
methanol: water mixture (20:9:1) at 1 mg/ml. For Syx-4-liposome
preparation, 52% 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC;
Avanti Polar Lipids, 850457), 20%
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE; Avanti
Polar Lipids, 850757), 15% 1,2-dioleoyl-sn-glycero-3-phospho-L-serine
(DOPS; Avanti Polar Lipids, 840035), 10% cholesterol (Avanti Polar
Lipids, 700000), 1% PI(4,5)P₂ (Avanti Polar Lipids, 840046), and 2% DiD
(Molecular Probes, D307) were mixed to a final lipid concentration of
1 mM. For VAMP-7-liposomes, 38% POPC, 11% POPE, 7%
1,2-dioleoyl-sn-glycero-3-phospho-(1’-myo-inositol) (PI; Avanti Polar
Lipids, 850149), 30% cholesterol, 15% sphingomyelin (Avanti Polar
Lipids, 860584), and 3% DiI (Molecular Probes, D282) were mixed to the
same final concentration. The lipid mixtures were vacuum-dried and
resuspended in 2% sodium deoxycholate. Full-length Syx-4 protein (5 µM)
and VAMP-7 SNARE-TM protein (5 µM) were incorporated into corresponding
liposomes, respectively, by incubation at room temperature for 20 min.
Detergent removal was performed using PD-10 desalting columns (GE
Healthcare). The resulting liposomes were combined in a 1:1 ratio with
a 60% (w/v) OptiPrep™ stock solution (Serumwerk Bernburg AG), layered
with 2 ml of 20% (w/v) OptiPrep™ solution, and 200 µL of 20 mM Tris-HCl
(pH 8.1), 150 mM NaCl. Liposomes were centrifuged at 120,000 × g for
5 h at 18 °C using an SW55Ti rotor (Beckman). The top fraction was
collected and dialyzed in a Slide-A-Lyzer™ dialysis cassette (Thermo
Fisher, 66383) for 12 h before use. For the fusion assay, 10 µM SN-23
was pre-incubated with Syx-4-liposomes at 37 °C for 2 h, while a
negative control group was prepared without SN-23. Equal volumes of
Syx-4- and VAMP-7-liposomes were mixed to a total volume of 60 µL, and
0.4 µM Munc13-4 protein was added to the experimental group. Fusion was
monitored using a FluoDia T70 fluorescence plate reader (Photon
Technology Incorporated) at 37 °C, with excitation at 530 nm and
emission at 580 nm and 667 nm. Liposome fusion signals were quantified
by calculating the FRET proximity ratio (E[PR]) between the donor (DiI)
and acceptor (DiD). The E[PR] was determined using Eq. [313]2:
[MATH:
EPR=IDi
DI
DiD+IDiI :MATH]
2
where I[DiD] and I[DiI] represent the fluorescence intensities of DiD
and DiI, respectively, measured under the 530/10 excitation filter.
Co-flotation experiment of PD-L1-containing liposomes with Munc13-4
For the preparation of PD-L1-containing liposomes, 80%
1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC; Avanti Polar
Lipids, 850457) and 20% 1,2-dioleoyl-sn-glycero-3-phospho-L-serine
(DOPS; Avanti Polar Lipids, 840035) were combined to achieve a total
lipid concentration of 1 mM. Lipid mixtures were vacuum-dried and
resuspended in 2% sodium deoxycholate. PD-L1 protein (5 µM) was
incorporated into the liposomes, followed by detergent removal using
PD-10 desalting columns (GE Healthcare). The resulting liposome
suspension (200 µL) was mixed in a 1:1 ratio with an 80% (w/v)
Histodenz™ stock solution (Thermo Fisher, D2158). The mixture was
layered sequentially with 350 µL of 30% (w/v) Histodenz™ and 20 µL of
20 mM Tris-HCl (pH 8.1), 150 mM NaCl. The prepared gradient was
centrifuged at 240,000 × g for 1.5 h at 18 °C using an SW55Ti rotor
(Beckman). Fractions (20 µL) were sequentially collected from the top
three layers of the gradient, with an additional 20 µL sample retrieved
from the bottom layer. These fractions were analyzed by Western blot to
assess the co-flotation of PD-L1-containing liposomes with Munc13-4.
Cryo-EM sample preparation
The Munc13-4–Rab27a complex, bound to GppNHp, was prepared at a
concentration of 0.5 mg/ml for cryo-EM analysis. Samples (3.5 µL) were
applied to glow-discharged cryo-EM grids (Quantifoil, Cu, R1.2/R1.3,
300 mesh) in an environment of 100% humidity at 4 °C. Grids were
blotted for 2 s with a blotting force of 4 and subsequently vitrified
by plunging into liquid ethane using a Vitrobot Mark IV (Thermo Fisher
Scientific). Prepared grids were either screened immediately or stored
in liquid nitrogen for future use.
Cryo-EM data acquisition and image processing
The Munc13-4–Rab27a complex with GppNHp datasets were collected by
300 kV Titan Krios electron microscope (Thermo Fisher Scientific)
equipped with a Falcon4 direct electron detector coupled with a
SelectrisX energy filter (10 eV slit width). The automated collection
was performed using the EPU software in electron event representation
(EER) mode, and all micrographs were recorded at a nominal
magnification of 165,000× with a raw pixel size of 0.73 Å on the image
plane. The micrographs were recorded in a −0.8 μm to −2.4 μm defocus
range, with an electron dose rate of 11.47 e^– /Å^2 /s and a total dose
of 50 e^– /Å^2. All the EER movies were pre-processed by CryoSPARC
(version 4.5.3)^[314]68 to perform the motion correction and CTF
estimation. 1,424,230 particles were selected by Blob Picking and
subsequently subjected to three rounds of 2D classification, and
317,449 particles from the selected classes were subjected to
ab-initial 3D reconstruction. The initial volume was further refined by
heterogeneous refinement and non-uniform refinement to yield a
consensus map with 3.36 Å global resolution. To acquire a map with
improved characteristics of the C2A and C2B domains, a total of 102,381
particles with apparent features were selected from the 3D
classification, yielding a reconstruction at 3.42 Å resolution.
Particle subtraction was subsequently applied to these particles,
focusing on the regions surrounding the C2A and C2B domains,
respectively. Local refinement of the resulting datasets produced
focused maps with resolutions of 4.39 Å and 7.2 Å. These two local
volume maps were then combined during model building, producing a
composite map with a resolution of 4.38 Å. All reported resolutions
were estimated using the gold-standard Fourier shell correction 0.143
criterion^[315]69. Data collection and refinement statistics are
summarized in Supplementary Table [316]1.
Model building and refinement of the Munc13-4–Rab27a complex
Initial models of Munc13-4 (AF-[317]Q70J99-F1) and Rab27a
(AF-[318]P51159-F1) were generated using AlphaFold 2
([319]https://colab.research.google.com/github/deepmind/alphafold/blob/
main/notebooks/AlphaFold.ipynb)^[320]70. The predicted structures were
fitted into the cryo-EM density map through rigid-body docking using
UCSF ChimeraX (v1.7.1)^[321]71. Further manual adjustments were
performed using COOT (v0.9.6)^[322]72. Subsequent real-space refinement
of the models was carried out through multiple iterative rounds using
PHENIX (v1.20)^[323]73, followed by final model validation. Figures
were prepared using PyMOL ([324]https://pymol.org/2/) and UCSF
ChimeraX. Data validation statistics are summarized in Supplementary
Table [325]1.
Statistics and reproducibility
Statistical analyses were conducted using GraphPad Prism software
(version 9.3). All key findings were reproduced in at least three
independent experiments. Specific statistical tests employed for each
experiment are detailed in the corresponding figure legends. For
comparisons between two groups, either a two-tailed unpaired or paired
t-test was applied, as appropriate. For multiple group comparisons,
one-way analysis of variance (ANOVA) followed by Tukey’s multiple
comparisons test, multiple t-tests, or two-way ANOVA with Sidak’s
multiple comparisons test was utilized. A p-value of less than 0.05
(P < 0.05) was considered indicative of statistical significance.
Additional information on the study design, the number of replicates,
and the statistical methods used is shown in the figure legends.
Reporting summary
Further information on research design is available in the [326]Nature
Portfolio Reporting Summary linked to this article.
Supplementary information
[327]Supplementary Information^ (7.9MB, pdf)
[328]Reporting Summary^ (5.2MB, pdf)
[329]Transparent Peer Review file^ (1.4MB, pdf)
Source data
[330]Source data^ (38.2MB, xlsx)
Acknowledgements