A bioorthogonal system reveals antitumour immune function of pyroptosis

Article abstract of DOI:10.1038/s41586-020-2079-1

Bioorthogonal chemistry capable of operating in live animals is needed to investigate biological processes such as cell death and immunity. Recent studies have identified a gasdermin family of pore-forming proteins that executes inflammasome-dependent and

Full text of DOI:10.1038/s41586-020-2079-1

Article
A bioorthogonal system reveals antitumour
immune function of pyroptosis
https://doi.org/10.1038/s41586-020-2079-1
Received: 26 July 2019
Qinyang Wang1,7, Yupeng Wang2,3,7, Jingjin Ding3,4, Chunhong Wang1, Xuehan Zhou1,
1,6
Wenqing Gao3, Huanwei Huang3, Feng Shao2,3,5 & Zhibo Liu
ꢀ✉
ꢀ✉
Accepted: 4 February 2020
Published online: xx xx xxxx
Check for updates
Bioorthogonal chemistry capable of operating in live animals is needed to investigate
biological processes such as cell death and immunity. Recent studies have identifed a
gasdermin family of pore-forming proteins that executes infammasome-dependent
and -independent pyroptosis1–5. Pyroptosis is proinfammatory, but its efect on
antitumour immunity is unknown. Here we establish a bioorthogonal chemical
system, in which a cancer-imaging probe phenylalanine trifuoroborate (Phe-BF3) that
can enter cells desilylates and ‘cleaves’ a designed linker that contains a silyl ether.
This system enabled the controlled release of a drug from an antibody–drug
conjugate in mice. When combined with nanoparticle-mediated delivery, desilylation
catalysed by Phe-BF3 could release a client protein—including an active gasdermin—
from a nanoparticle conjugate, selectively into tumour cells in mice. We applied this
bioorthogonal system to gasdermin, which revealed that pyroptosis of less than 15%
of tumour cells was sufcient to clear the entire 4T1 mammary tumour graft. The
tumour regression was absent in immune-defcient mice or upon T cell depletion, and
was correlated with augmented antitumour immune responses. The injection of a
reduced, inefective dose of nanoparticle-conjugated gasdermin along with Phe-BF3
sensitized 4T1 tumours to anti-PD1 therapy. Our bioorthogonal system based on
Phe-BF3 desilylation is therefore a powerful tool for chemical biology; our application
of this system suggests that pyroptosis-induced infammation triggers robust
antitumour immunity and can synergize with checkpoint blockade.
As a positron emission tomography (PET) imaging probe for cancers,
18F]Phe-BF3 has a comparable sensitivity to—but a higher specificity but suffer from an uncontrolled and inefficient release of the drug. We
Antibody–drug conjugates hold promise in anticancer therapy10,
[
than—[18F]fluorodeoxyglucose6–8. Free fluoride, which is a deprotec- applied Phe-BF3 desilylation to an antibody–drug conjugate system.
tion agent in organic synthesis9, catalyses efficient desilylation. We Specifically, we used silyl-ether (TBS)-containing carbamate (SEC) to
wondered whether Phe-BF3 could also induce desilylation, and whether conjugate monomethyl auristatin E (MMAE) to trastuzumab (anti-HER2
it could be developed into a bioorthogonal system for tumour-selective antibody) (Fig. 1d, Extended Data Figs. 1e, 2d–f). Phe-BF3 incubation
manipulation. To this end, we designed an ortho-carbamoylmethylene caused about 90% MMAE release from the MMAE–SEC–trastuzumab
silyl-phenolic ether system attached to a client molecule, coumarin conjugate (Extended Data Fig. 2g, h). Trastuzumab–SEC–MMAE
(Fig. 1a, Extended Data Fig. 1a). Desilylation of this system will cause incorporating trastuzumab labelled with zirconium-89 (89Zr) showed
the release of free coumarin, which becomes fluorescent (Extended expected tumour-targeting in nude mice grafted with HER2-positive
Data Fig. 1b). We began with the tert-butyldimethyl silyl (TBS) group, BGC823 cells (Fig. 1e, Extended Data Fig. 2i, j). As previously reported7,8,11
,
and synthesized TBSO–coumarin with incorporation of a para-N,N- [18F]Phe-BF3 exhibited tumour uptake and retention; the concentration
dimethylaminoacetamide for conjugation to a potential carrier (Fig. 1a, of [18F]Phe-BF3 in the tumour was much higher than in other tissues, and
Extended Data Fig. 1a–c). Upon incubation with Phe-BF3, TBSO– was also higher than the concentration of trastuzumab–SEC–MMAE in
coumarin released fluorescent coumarin and the reaction-rate constant the tumour (Fig. 1e, Extended Data Fig. 2k, l). Marked release of MMAE in
was comparable to that of sodium fluoride (Fig. 1b, Extended Data the tumour was detected when trastuzumab–SEC–MMAE and Phe-BF3
Fig. 2a–c). TBSO–coumarin did not react with hydrogen peroxide, were injected into the same mouse (Fig. 1f). Thus, our bioorthogonal
glutathione or other biologically relevant anions (Extended Data system based on Phe-BF3 desilylation enables the controlled and effi-
Fig. 2b). In human BGC823 cells transfected with TBSO–coumarin, treat- cient release of the drug from an antibody–drug conjugate.
ment with Phe-BF3 resulted in strong coumarin fluorescence (Fig. 1c)
Thus, desilylation catalysed by Phe-BF3 can operate within live cells.
To optimize this system, we diversified the TBS group in the coumarin
conjugate into triethylsilyl (TES) or triisopropyl silyl (TIPS) (Extended
1Beijing National Laboratory for Molecular Sciences, Radiochemistry and Radiation Chemistry Key Laboratory of Fundamental Science, College of Chemistry and Molecular Engineering, Peking
University, Beijing, China. 2Research Unit of Pyroptosis and Immunity, Chinese Academy of Medical Sciences, Beijing, China. 3National Institute of Biological Sciences, Beijing, China. 4National
Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China. 5Tsinghua Institute of Multidisciplinary Biomedical Research, Tsinghua University,
7
Beijing, China. 6Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China. These authors contributed equally: Qinyang Wang, Yupeng Wang. e-mail: shaofeng@nibs.ac.cn;
✉
zbliu@pku.edu.cn
Nature | www.nature.com | 1

Article
Conjugate to
trastuzumab
Phe-BF3
responsive linker
d
a
O
OTBS
MMAE
NH3
HN
O
–
BF3
Phe-BF3
cleavable motif
OTBS
O
O
N
HN
H
N
O
O
O
N
H2N
O
O
O
N
O
H
N
O
N
O
N
O
TBSO–coumarin
O
O
Coumarin
mAb
O
OH
O
N
H
n = 4
H
Desilylation
b
[1,4]Fragmentation
TBS-F
OH
OH
HN
H
N
0
0.3
1
10
10
NaF
Phe -BF3
N
O
O
O
O
O
O
O
N
OH
N
H
O
N
N
H
H
Blank
TBSO–coumarin TBSO–coumarin + Phe-BF3
c
mAb
O
n = 4
MMAE
200
150
e
f
[
18F]Phe-BF3
[
89Zr]ADC
High
P < 0.001
100
50
0
t
t
l
k
mAb-SEC-MMAE
Phe-BF3
+
−
+
+
Low
Fig. 1 | Desilylation bioorthogonal chemistry and its application to an
antibody–drug conjugate. a, Schematic of the release of fluorescent
BGC823 cells. e, PET-computed tomography images of mice intravenously
injected with [18F]Phe-BF3 or [89Zr]trastuzumab–SEC–MMAE (marked as ADC).
t, tumour; l, liver; k, kidney. f, Mice were injected with trastuzumab (mAb)–
SEC–MMAE alone (4.5 mg per kg body weight (kg−1)) or additionally with Phe-
BF3 (15 mg kg−1). Tumour concentrations of MMAE are mean s.e.m. n = 4 mice,
two-tailed unpaired Student’s t-test. Data are representative of two (c, f) or
three (b, e) independent experiments.
coumarin from TBSO–coumarin, induced by Phe-BF3 desilylation. b, TBSO–
coumarin was reacted with the indicated relative amounts of Phe-BF3 or NaF.
c, Confocal images of BGC823 cells that were transfected with TBSO–coumarin
and then treated with Phe-BF3. Scale bars, 20 μm. d, Design of the
trastuzumab (mAb)–SEC–MMAE conjugate, and MMAE release induced by
Phe-BF3 desilylation. e, f, Nu/Nu mice were implanted subcutaneously with
NP–GFP
a
–
Phe-BF3
NaF
S
FDG
S
Leꢀunomide Capecitabine
5-FU
S
T
S
P
T
S
P
T
P
T
P
T
S
P
T
S
P
T
P
d
e
Actin
Hoechst
P < 0.0001
40
30
20
10
0
*
P < 0.0001
2,000
1,000
NP–GFP
Hoechst
b
c
[
89Zr]GFP NP–[89Zr]GFP
[
18F]Phe-BF3
GFP
Merge
High
mNeonGreen
Merge
0
t
l
t
Phe-BF3
–
+
–
+
t
gb
NP–mNeonGreen–NLS
l
k
Low
Fig. 2 | Phe-BF3 desilylation releases GFP from TES-linked NP–GFP with
tumour selectivity in mice. a, NP–GFP was treated with Phe-BF3 or another
injected with [89Zr] G F P, N P–[89Zr]GFP or [18F]Phe-BF3. gb, gallbladder. d, e, Mice
were injected with NP–mNeonGreen–NLS. d, Confocal fluorescence images of
a tumour section, magnified from Extended Data Fig. 4i. Scale bars, 5 μm.
e, Numbers and percentages of mNeonGreen–NLS-positive cells in the tumour.
n = 22 fields from 2 mice injected with NP–mNeonGreen–NLS or n = 30 fields
from 4 mice injected with NP–mNeonGreen–NLS and Phe-BF3. Data shown as
mean s.e.m. (two-tailed unpaired Student’s t-test) (e) and representative of
two (b–e) or three (a) independent experiments.
indicated fluorine compound. The reactions were centrifuged (Extended Data
Fig. 4b). P, pellet; S, supernatant; T, total;*, loading control sampled before
centrifugation. b, Confocal fluorescence images of primary BMDMs treated
with NP–GFP (1 mg ml-1) for 24 h, followed by Phe-BF3 (100 μM) for another 24 h.
Scale bars, 20 μm. c–e, BALB/c mice were implanted subcutaneously with 4T1
cancer cells. c, PET-computed tomography images of mice intravenously
2 | Nature | www.nature.com

a
Pyroptosis
Gasdermin
(GA3) pore
LAT1
NP
Gasdermin
(GA3)
Phe-BF3
NP–GA3
Nucleus
Control
NP–GA3
Phe-BF3
Control
NP–GA3
Phe-BF3
b
NP + GA3
HeLa
NP–GA3 + Phe-BF3 NP–GA3(mut) + Phe-BF3
NP + GA3
NP–GA3 + Phe-BF3 NP–GA3(mut) + Phe-BF3
P < 0.0001
P < 0.001
c
P < 0.0001
25
20
15
10
5
100
80
60
40
20
60
40
20
0
40
30
20
10
0
priBMDM
4T1
EMT6
P < 0.001
0
Fig. 3 | Phe-BF3 desilylation releases gasdermin from NP–GSDMA3 to induce
pyroptosis. a, Cartoon of the experimental design. Purified GSDMA3(N + C)
was conjugated to a nanoparticle to obtain NP–GSDMA3. b, c, HeLa, EMT6 or
4T1 cells, or primary BMDMs, were treated as indicated. NP + GSDMA3,
nanoparticles mixed with the GSDMA3(N + C) complex. b, Confocal images of
the treated HeLa and EMT6 cells. Scale bars, 20 μm. Propidium iodide (PI) and
annexin V–fluorescein isothiocyanate (FITC) were added to the cells 15 min
before imaging. c, Flow-cytometry measurements of cells positive for
propidium iodide and annexin V. priBMDM, primary BMDM. Mean s.d. from
three independent replicates, two-tailed unpaired Student’s t-test.
GA3, GSDMA3(N + C). All data are representative of three independent
experiments.
Data Figs. 1a, c, 3a). Although TIPSO–coumarin resisted desilylation GFP completely and the release from TBS-linked NP–GFP was partial
catalysed by Phe-BF3, TESO–coumarin was desilylated much more effi- (Fig. 2a, Extended Data Fig. 4b, c). In HeLa cells, NP–GFP accessed
ciently (Extended Data Fig. 3b–e). Differential reactivity of the three silyl the cytosol after damaging the endosomal membrane (Supplemen-
groups indicates a steric hindrance effect, echoing desilylation cata- tary Videos 1, 2, Supplementary Fig. 1). HeLa cells, mouse mammary
lysed by sodium fluoride12 (Extended Data Fig. 3e). At 10-mM concen- carcinoma EMT6 and 4T1 cells and primary mouse bone-marrow-
tration, the desilylation induced by Phe-BF3 was completed instantly. derived macrophages (BMDMs) all expressed the LAT1 transporter for
Such a fast velocity suggests that the desilylation induced by Phe-BF3 Phe-BF3 (Extended Data Fig. 4d). Following treatment with NP–GFP, the
may proceed through direct fluoride transfer13, as the generation of a addition of Phe-BF3 induced evident GFP release in all the four types
free-fluoride intermediate from Phe-BF3 features an extremely slow of cell; this did not occur with sodium fluoride, because it is not cell-
kinetics14. TESO–coumarin, TBSO–coumarin and TIPSO–coumarin all permeable (Extended Data Fig. 4e). Owing to the loss of quenching by
showed no spontaneous desilylation. TESO–coumarin was not desi- the nanoparticle18, the released GFP became fluorescent (Fig. 2b). Thus,
lylated by other agents, including fluorine-derived drugs (Extended desilylation catalysed by Phe-BF3 can release a functional protein from
Data Fig. 3f, g). Thus, of the groups we tested, TES is the best choice for nanoparticles in mammalian cells.
the bioorthogonal system based on Phe-BF3 desilylation.
Tumour enrichment of [18F]Phe-BF3 was also observed in BALB/c
Gold nanoparticles are a widely used drug-delivery vehicle15. A nano- mice that were engrafted subcutaneously with 4T1 carcinoma
particle has a tumour preference owing to an enhanced permeability cells (Fig. 2c). The remainder of the Phe-BF3 was rapidly cleared via
and retention effect in tumours, and a more-efficient internalization by renal excretion. Unlike 89Zr-labelled GFP ([89Zr]GFP), nanoparticle-
cancer cells16,17. We attached the silyl-phenolic ether to nanoparticles conjugated [89Zr]GFP (NP–[89Zr]GFP) started to accumulate in
(Extended Data Fig. 1d) and replaced the coumarin with green fluo- the tumour at 6 h after injection; its level reached a plateau at 18 h
rescence protein (GFP) to generate nanoparticle (NP)–GFP (Extended (Extended Data Fig. 4f, g). Although NP–[89Zr]GFP was highly enriched
Data Fig. 4a). Upon treatment with Phe-BF3 or sodium fluoride (but not in the liver, the distributions of NP–GFP and Phe-BF3 overlap only in
other fluorine-containing compounds), TES-linked NP–GFP released the tumour. To detect the desilylation of nanoparticle conjugates,
Nature | www.nature.com | 3

Article
a
c
e
PBS (iv)
NP–GA3 +
Phe-BF3 (iv) + Phe-BF3 (iv)
NP–GA3(mut)
P = 0.0171
P = 0.3270
89Zr]GA3
NP–[89Zr]GA3
[
18F]Phe-BF3
1,000
500
[
High
P < 0.0001
P = 0.0002
3
P < 0.0001
2
1
0
t
l
t
t
gb
0
l
NP–GA3 (iv)
Phe-BF3 (iv)
NP–GA3 +
Phe-BF3 (it)
k
1,000
500
0
Low
0
20
40
0
20
40
0
20
40
Day
Day
Day
b
Day
0
6
7
8
9
10
11
12
13
14
4T1/EMT6 cell implantation
NP–GA3 Phe-BF3
Phe-BF3 NP–GA3 Phe-BF3
Phe-BF3 NP–GA3 Phe-BF3
Phe-BF3
4T1 tumour model
NP–GA3(mut)
+ Phe-BF3
NP–GA3
+ Phe-BF3
d
f
PBS
NP–GA3
1,500
PBS (iv)
1,000
500
0
NP–GA3 +
Phe-BF3 (iv)
NP–GA3(mut)
+ Phe-BF3 (iv)
0
20
Day
40
0
20
Day
40
0
20
Day
40
0
20
Day
40
NP–GA3 (iv)
g
1,000
500
P < 0.0001
P = 0.013
PBS
NP–GA3
Phe-BF3 (iv)
NP–GA3(mut) +
Phe-BF3
P = 0.0036
NP–GA3 +
Phe-BF3 (it)
NP–GA3 + Phe-BF3
0
Day
0
5
20
10
15
Day
Fig. 4 | Treatment with NP–GSDMA3 and Phe-BF3 causes tumour regression
in mice. a, PET-computed tomography images of mice bearing 4T1 tumours
that intravenously injected with [89Zr]GSDMA3, NP–[89Zr]GSDMA3 or [18F]Phe-
BF3. b–g, BALB/c mice were implanted subcutaneously with 4T1 (c–e) or EMT6
(f, g) cells, followed by intravenous (iv) or intratumoural (it) injection of
NP–GSDMA3 or Phe-BF3 alone, or in combination. b, Treatment scheme.
In c–e, n = 7 mice per group. In f, g, n = 8 mice for PBS and
NP–GSDMA3(mut) + Phe-BF3, 6 mice for NP–GSDMA3 and 7 mice for
NP–GSDMA3 + Phe-BF3. c, f, Tumour volume of an individual mouse.
d, Photographs of representative tumours on day 26. e, g, Average tumour
weight or volume. Mean s.e.m., two-tailed unpaired Student’s t-test. GA3,
GSDMA3(N + C). Data are representative of three (c–e) or two (a, f, g)
independent experiments.
we replaced GFP with the brighter fluorescent protein mNeonGreen was released from NP–GSDMA3 and was capable of forming pores on
and introduced a nuclear-localization signal (NLS) to concentrate liposome membranes (Extended Data Fig. 5c, d). When HeLa, EMT6 or
the released mNeonGreen fluorescence (Extended Data Fig. 4h). The 4T1 cells, or BMDMs, were treated with NP–GSDMA3 and Phe-BF3, we
4T1 tumours in control mice showed no mNeonGreen fluorescence; observed membrane enrichment of gasdermin N domains (Extended
about 15% of cells in tumours injected with Phe-BF3 showed nuclear Data Fig. 5e); the cells showed pyroptotic morphology (Fig. 3b). Cells
mNeonGreen signal (Fig. 2d, e, Extended Data Fig. 4i). Thus, the Phe- treated with NP–GSDMA3 and Phe-BF3 released lactate dehydro-
BF3 desilylation system is compatible with nanoparticle-mediated genase, which was insensitive to ferroptosis, necroptosis and pan-
delivery, and shows tumour selectivity.
caspase inhibitors (Extended Data Fig. 5f). Treatment with Phe-BF3
The pore-forming N-terminal domain (N domain) of gasdermin or NP–GSDMA3 alone caused no pyroptosis. GSDMA3 with E14K and
translocates to the plasma membrane to induce pyroptosis1–3,5. The L184D mutations (hereafter, GSDMA3(mut))19,20, which is deficient
effect of this proinflammatory cell death on tumours is unknown. Our in pore-forming, blocked the pyroptosis induced by treatment with
desilylation system provides an ideal approach for investigating this NP–GSDMA3 and Phe-BF3 (Fig. 3b, c, Extended Data Fig. 5g). About
question (Fig. 3a). We used gasdermin A3 (GSDMA3)19,20, and mutated 40%, 35% and 20% of HeLa, EMT6 and 4T1 cells, respectively, treated
both cysteine residues in the C-terminal domain (C domain) of gasder- with NP–GSDMA3 and Phe-BF3 underwent pyroptosis (Fig. 3c). Primary
min (thus, conjugation to nanoparticles will occur only via the gasder- BMDMs died more extensively, owing to their strong phagocytic capac-
min N domain). The mutation did not affect the pyroptosis-inducing ity (Extended Data Fig. 5h).
activity of the complex formed by the N and C domains of GSDMA3
Similar to NP–[89Zr]GFP, NP–[89Zr]GSDMA3 showed the desired bio-
(GSDMA3(N + C)) (Extended Data Fig. 5a, b). GSDMA3(N + C) was conju- compatibility and selective tumour targeting (Extended Data Fig. 6a–d).
gated to 60-nm nanoparticles through the TES ether linker, generating Transmission electron microscopy confirmed the uptake and cytosolic
NP–GSDMA3 (Fig. 3a). Upon incubation with Phe-BF3, GSDMA3(N + C) distribution of NP–GSDMA3 in 4T1 tumours (Extended Data Fig. 6e).
4 | Nature | www.nature.com

800
600
400
200
0
0% 4T1
10% 4T1
GA3dox-on
30% 4T1
GA3dox-on
a
PBS
b
c
NP–GA3 + Phe-BF3 NP–GA3(mut) + Phe-BF3
GA3dox-on
PBS
Dox
0
0
0
5
10 15 20
Day
5
10 15 20
Day
5
10 15 20
Day
600
400
200
0
d
Nu/Nu mice
P = 0.015
P = 0.017
30
20
10
0
PBS
NP_GA3
NP–GA3(mut) + Phe-BF3
NP–GA3 + Phe-BF3
0
5
10 15 20 25
P = 0.037
P = 0.037
Day
20
15
10
5
PBS
Anti-CD4
Anti-CD8
e
f
NP–GA3 + Phe-BF3
PBS
NP–GA3 + Phe-BF3
1,250
Macrophage
Macrophage
NK
Isotype
NK
MDSC
MDSC
1,000
750
500
250
0
CD8
CD4
CD8
0
8
Density
Max
CD4
P = 0.042
P = 0.044
Monocyte
Monocyte
Neutrophil
Neutrophil
6
4
2
Min
0
5
10 15 20 25 30
Day
t-SNE dimension 1
PBS
Anti-PD1
NP–GA3 + Phe-BF3
P = 0.44
P = 0.78
g
h
4T1 tumour model
0
P = 0.040
P = 0.039
PBS
NP–GA3 + Phe-BF3
+ anti-PD1
P < 0.0001
50
40
30
20
1,000
500
0
Anti-PD1
NP–GA3 +
Phe-BF3
NP–GA3 + Phe-BF3
+ anti-PD1
0
5
10 15 20 25
Day
i
6
7
8
9
11
13
15
Day
0
4T1 cell implantation
NP–GA3 Phe-BF3 Phe-BF3 Anti-PD1
Anti-PD1
Anti-PD1
Anti-PD1
Fig. 5 | A low level of tumour-cell pyroptosis induces effective antitumour
immunity and synergizes with anti-PD1 blockade. a, BALB/c mice bearing 4T1
tumours were treated with PBS (n = 3), NP–GSDMA3 and Phe-BF3 (n = 4) or NP–
GSDMA3(mut) and Phe-BF3 (n = 4), as in Fig. 4b. Propidium iodide was
intravenously injected into the mice before assay. Representative tumour-
section images are shown. Scale bars, 100 μm. b, Normal 4T1 cells mixed with
0%, 10% or 30% 4T1-GSDMA3dox-on cells were inoculated into BALB/c mice (n = 7),
followed by intratumoural injection of doxycycline (dox) or PBS on day 6. c, Nu/
Nu mice bearing 4T1 tumours were treated as in Fig. 4b. n = 7 mice for PBS, NP–
GSDMA3 and NP–GSDMA3(mut) + Phe-BF3, 6 mice for NP–GSDMA3 + Phe-BF3.
d, Quantification of 4T1-tumour-infiltrating lymphocytes from BALB/c mice.
n = 6 mice for PBS and NP–GSDMA3 + Phe-BF3, 7 mice for NP–
antibodies or an isotype control (n = 9 mice) were intraperitoneally injected
into BALB/c mice bearing 4T1 tumours before treatment with NP–GSDMA3 and
Phe-BF3. Mice treated with PBS (n = 8) serve as a control. f, Single-cell RNA
sequencing of CD45+ immune cells from 4T1 tumours treated with PBS or NP–
GSDMA3 and Phe-BF3. t-distributed stochastic neighbour embedding (t-SNE)
density plots of 7,000 CD45+ cells randomly sampled from each group are
shown. NK, natural killer. g–i, Mice bearing 4T1 tumours were treated with PBS
(n = 7 mice), anti-PD1 (n = 8 mice) or NP–GSDMA3 and Phe-BF3 (n = 7 mice) alone,
or in combination (n = 8 mice). g, Photographs of representative tumours at
day 26. h, Tumour growth curve. i, Treatment scheme. In b, c, e, h, the average
tumour volume at indicated time points is shown. GA3, GSDMA3(N + C). Data
shown as mean s.e.m. (two-tailed unpaired Student’s t test) (b–e, h) and
representative of two (a–c, e–h) and three (d) independent experiments.
GSDMA3(mut) + Phe-BF3. e, Anti-CD4 (n = 9 mice) or -CD8 (n = 7 mice)
Thus, distributions of NP–GSDMA3 and Phe-BF3 also converged onto or of NP–GSDMA3(mut) and Phe-BF3, did not cause tumour shrinkage.
the tumour (Fig. 4a). The 4T1 cells had no endogenous gasdermin D Tumour weight measurements confirmed the antitumour effect caused
(GSDMD), gasdermin E (GSDME) and IL-1β (Extended Data Fig. 6f, g). by GSDMA3 released from nanoparticles (Fig. 4e). Treatment with
In mice treated with PBS, the volume of 4T1 tumours increased by NP–GSDMA3 and Phe-BF3 induced a similar tumour regression in the
about 20-fold in 2 weeks (Fig. 4b–d). After three rounds of treatment EMT6 tumour model (Fig. 4f, g). Mice treated with NP–GSDMA3 and
with NP–GSDMA3 and Phe-BF3 (Fig. 4b), massive tumour shrinkage Phe-BF3 showed neither weight loss nor abnormalities in major organs,
occurred and the tumour burden became negligible by day 25 (Fig. 4c, d, including the liver and kidneys (Extended Data Fig. 7a, b). Serum con-
Extended Data Fig. 6h, i). The injection of NP–GSDMA3 or Phe-BF3 alone, centrations of alkaline phosphatase, calcium and phosphate—as well
Nature | www.nature.com | 5

Article
as the skeletons—of mice treated with Phe-BF3 were similar to those of within the tumour microenvironment is low or ineffective for sufficient
mice treated with PBS (Extended Data Fig. 7c–f).
infiltration and/or activation of T lymphocytes; thus, the tumours
Intratumoural injection with NP–GSDMA3 and Phe-BF3 caused the resistant to checkpoint blockade are deemed to be ‘cold’25. The 4T1
same regression of 4T1 tumours as that caused by intravenous injection tumour—which is known to be cold—did not respond to anti-PD1 treat-
(Fig. 4c–e), which suggests that the antitumour effect is the result of ment (Fig. 5g, h). One round of injection with NP–GSDMA3 and Phe-BF3
GSDMA3 activation at the tumour site. Consistently, mice treated with also did not prevent tumour growth (Fig. 5g, h). Notably, when the
NP–GSDMA3 and Phe-BF3 contained a higher number of propidium- one-round injection was followed by anti-PD1 treatment (Fig. 5i), a
iodide-positive pyroptotic cells in the tumour, compared to control marked shrinkage in tumour volume and reduction in tumour weight
mice (Fig. 5a, Extended Data Fig. 6j). Propidium-iodide-positive cells occurred (Fig. 5g, h, Extended Data Fig. 10c). Thus, pyroptosis-induced
were only a small fraction of the entire tumour-cell population, which inflammation within the tumour microenvironment can synergize with
is consistent with the 20% pyroptosis that we detected in cultured checkpoint blockade in inducing antitumour immunity.
4T1 cells (Fig. 3c). Furthermore, we generated 4T1 cells that contain
Here we show that the Phe-BF3 desilylation bioorthogonal system
a doxycycline-inducible PreScission protease along with PreScission- operates efficiently in the animal, which potentially can be combined
protease-cleavable GSDMA3. These cells (which we term 4T1-GSD
-
with other methods for new applications. The application of our sys-
MA3dox-on cells) resembled normal 4T1 cells in forming tumour grafts tem to gasdermin activation uncovers an antitumour immune effect
(Fig. 5b). Notably, injecting doxycycline into the tumour completely of pyroptosis. Thus, a gasdermin agonist may improve the efficacy of
stopped tumour growth—despite the fact that only 10–30% of 4T1-GSD- cancer immunotherapy. An immune-dependent anticancer effect has
MA3dox-on cells were mixed into the inoculated 4T1 cells (Fig. 5b).
long been known for cell-killing chemotherapeutic drugs26,27; this effect
The fact that only a fraction of the tumour cells underwent pyroptosis might be due to gasdermin activation. Direct gasdermin activation,
but the entire tumour was eliminated suggests a role of the immune sys- via desilylation mediated by Phe-BF3, provides a powerful system for
tem. Supporting this, athymic Nu/Nu mice (which lack mature T cells) developing mechanistic understandings of the antitumour immunity
did not show tumour regression after treatment with NP–GSDMA3 stimulated by immunogenic cell death such as pyroptosis.
and Phe-BF3 (Fig. 5c). Markedly increased CD3+ T cell infiltration was
recorded in tumours treated with NP–GSDMA3 and Phe-BF3 (Fig. 5d,
Extended Data Fig. 8a–c). Both the CD4+ and CD8+ cell populations
Online content
were increased, whereas the percentage of CD4+FOXP3+ T regulatory Any methods, additional references, Nature Research reporting sum-
cells decreased; these changes were absent in tumours treated with maries, source data, extended data, supplementary information,
NP–GSDMA3(mut) and Phe-BF3 (Fig. 5d, Extended Data Fig. 8a, d, e). acknowledgements, peer review information; details of author con-
Depletion of the CD4+ or CD8+ cell population blocked tumour regres- tributions and competing interests; and statements of data and code
sion induced by NP–GSDMA3 and Phe-BF3 (Fig. 5e, Extended Data availability are available at https://doi.org/10.1038/s41586-020-2079-1.
Fig. 8f). Thus, pyroptosis-induced tumour regression requires both
cytotoxic T cells and CD4+ T helper cells.
We performed single-cell RNA sequencing on CD45+ leukocytes
from 4T1 tumours of two control and two therapeutically treated
mice (Extended Data Fig. 9a). Sequenced cells were clustered into ten
immune-cell subsets (Fig. 5f, Extended Data Fig. 9b–d). Compared with
tumours treated with PBS, tumours treated with NP–GSDMA3 and Phe-
BF3 had increased CD4+, CD8+ and natural killer cell populations, but
decreased monocyte, neutrophil and myeloid-derived suppressor cell
populations (Fig. 5f, Extended Data Fig. 9d). The M1 macrophage popu-
lation also increased whereas the percentage of M2 marker (Arg1)-pos-
itive cells decreased (Fig. 5f, Extended Data Fig. 9e). Genes that encode
chemotactic cytokines (Ccl5, Cxcl9 and Cxcl10) or are important for
lymphocyte activation (Cd69 and Klrk1) were upregulated in tumours
treated with NP–GSDMA3 and Phe-BF3 (Extended Data Fig. 9f, g).
Antitumour effector genes (Ifng, Gzma, Gzmb, Gzmk and Fasl) were
also upregulated, whereas the protumoural or immunosuppressive
genes Csf1, Vegfa, Arg1, Cd274 (which encodes PD-L1) and Pdcd1lg2
(which encodes PD-L2) were downregulated, in tumours treated with
NP–GSDMA3 and Phe-BF3 (Extended Data Fig. 9e, g). These immuno-
logical changes corroborate the potent antitumour effect of injections
of NP–GSDMA3 and Phe-BF3.
1.
Kayagaki, N. et al. Caspase-11 cleaves gasdermin D for non-canonical inflammasome
signalling. Nature 526, 666–671 (2015).
Shi, J. et al. Cleavage of GSDMD by inflammatory caspases determines pyroptotic cell
death. Nature 526, 660–665 (2015).
Shi, J., Gao, W. & Shao, F. Pyroptosis: gasdermin-mediated programmed necrotic cell
death. Trends Biochem. Sci. 42, 245–254 (2017).
2.
3.
4. Wang, Y. et al. Chemotherapy drugs induce pyroptosis through caspase-3 cleavage of a
gasdermin. Nature 547, 99–103 (2017).
5.
6.
7.
Broz, P., Pelegrín, P. & Shao, F. The gasdermins, a protein family executing cell death and
inflammation. Nat. Rev. Immunol. https://doi.org/10.1038/s41577-019-0228-2 (2019).
Liu, Z. et al. Preclinical evaluation of a high-affinity 18F-trifluoroborate octreotate
derivative for somatostatin receptor imaging. J. Nucl. Med. 55, 1499–1505 (2014).
Liu, Z. et al. Boramino acid as a marker for amino acid transporters. Sci. Adv. 1, e1500694
(2015).
8.
9.
Liu, Z. et al. An organotrifluoroborate for broadly applicable one-step 18F-labeling. Angew.
Chem. Int. Edn Engl. 53, 11876–11880 (2014).
David Crouch, R. Selective monodeprotection of bis-silyl ethers. Tetrahedron 60,
5833–5871 (2004).
10. Chau, C. H., Steeg, P. S. & Figg, W. D. Antibody–drug conjugates for cancer. Lancet 394,
793–804 (2019).
11. Perrin, D. M. [18F]-Organotrifluoroborates as radioprosthetic groups for PET imaging: from
design principles to preclinical applications. Acc. Chem. Res. 49, 1333–1343
(2016).
12. Nelson, T. D. & Crouch, R. D. Selective deprotection of silyl ethers. Synthesis 1996,
1031–1069 (1996).
13. Handy, C. J., Lam, Y. F. & DeShong, P. On the synthesis and NMR analysis of
tetrabutylammonium triphenyldifluorosilicate. J. Org. Chem. 65, 3542–3543 (2000).
14. Liu, Z. et al. From minutes to years: predicting organotrifluoroborate solvolysis rates.
Chemistry 21, 3924–3928 (2015).
15. Papasani, M. R., Wang, G. & Hill, R. A. Gold nanoparticles: the importance of physiological
principles to devise strategies for targeted drug delivery. Nanomedicine 8, 804–814
(2012).
16. Wilhelm, S. et al. Analysis of nanoparticle delivery to tumours. Nat. Rev. Mater. 1, 16014
(2016).
17. Blanco, E., Shen, H. & Ferrari, M. Principles of nanoparticle design for overcoming
biological barriers to drug delivery. Nat. Biotechnol. 33, 941–951 (2015).
18. Swierczewska, M., Lee, S. & Chen, X. The design and application of fluorophore–gold
nanoparticle activatable probes. Phys. Chem. Chem. Phys. 13, 9929–9941 (2011).
19. Ding, J. et al. Pore-forming activity and structural autoinhibition of the gasdermin family.
Nature 535, 111–116 (2016).
20. Ruan, J., Xia, S., Liu, X., Lieberman, J. & Wu, H. Cryo-EM structure of the gasdermin A3
membrane pore. Nature 557, 62–67 (2018).
21. Ghiringhelli, F. et al. Activation of the NLRP3 inflammasome in dendritic cells induces IL-
1β-dependent adaptive immunity against tumors. Nat. Med. 15, 1170–1178 (2009).
IL-1β has an established function in T-cell-mediated antitumour
immunity21,22. Levels of IL-1β, as well as IL-18, in the serum and the
tumour were increased in mice injected with NP–GSDMA3 and Phe-
BF3 (Extended Data Fig. 10a). As expected, levels of HMGB1 also
increased. An injection of anti-IL-1β antibody into the tumour-bearing
mice severely inhibited the tumour regression induced by treatment
with NP–GSDMA3 and Phe-BF3, whereas anti-IL-18 and anti-HMGB1
antibodies showed partial and no effects, respectively (Extended Data
Fig. 10b). Thus, IL-1β—probably derived from activated macrophages—is
required for antitumour immunity induced by pyroptosis of tumour
cells. Immune checkpoint blockade (particularly using anti-PD1) is
highly successful in the treatment of cancer, but has a low response
rate23,24. One main cause of this low response rate is that inflammation
6 | Nature | www.nature.com

22. Lee, P. H. et al. Host conditioning with IL-1β improves the antitumor function of adoptively
transferred T cells. J. Exp. Med. 216, 2619–2634 (2019).
23. Zou, W., Wolchok, J. D. & Chen, L. PD-L1 (B7-H1) and PD-1 pathway blockade for cancer
therapy: mechanisms, response biomarkers, and combinations. Sci. Transl. Med. 8,
328rv4 (2016).
26. Schwartz, H. S. & Grindey, G. B. Adriamycin and daunorubicin: a comparison of antitumor
activities and tissue uptake in mice following immunosuppression. Cancer Res. 33,
1837–1844 (1973).
27. Casares, N. et al. Caspase-dependent immunogenicity of doxorubicin-induced tumor cell
death. J. Exp. Med. 202, 1691–1701 (2005).
24. Ribas, A. & Wolchok, J. D. Cancer immunotherapy using checkpoint blockade. Science
359, 1350–1355 (2018).
25. Sharma, P. & Allison, J. P. The future of immune checkpoint therapy. Science 348, 56–61
(2015).
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2020
Nature | www.nature.com | 7

Article
Methods
1H NMR (400 MHz, DMSO-d6) δ 10.51 (s, 1H), 10.31 (s, 1H), 7.73–7.64
(m, 2H), 7.56 (d, J = 1.9 Hz, 2H), 7.43 (dd, J = 8.6, 1.9 Hz, 1H), 6.88 (d, J = 8.7
No statistical methods were used to predetermine sample size. The Hz, 1H), 6.24 (s, 1H), 5.13 (s, 2H), 3.74 (s, 2H), 3.35 (s, 6H), 2.38 (s, 3H),
experiments were not randomized and investigators were not blinded 0.95 (s, 9H), 0.23 (s, 6H). 13C NMR (101 MHz, DMSO) δ 160.06, 153.83,
to allocation during experiments and outcome assessment.
153.23, 153.22, 149.51, 142.79, 131.89, 126.32, 126.06, 121.74, 121.23, 118.67,
114.38, 114.30, 111.91, 104.46, 61.98, 45.26, 43.96, 25.53, 18.02, 17.88, 8.41,
−4.44 (Supplementary Fig. 3). TIPSO–coumarin: 1H NMR (400 MHz,
Chemical reagents, plasmids and antibodies
5-Fluorouracil (5-FU) (920052), leflunomide (448506), capecitabine DMSO-d6) δ 10.28 (s, 1H), 7.70 (d, J = 8.7 Hz, 1H), 7.64 (d, J = 1.9 Hz, 1H),
(392078) and fluorodeoxyglucose (D234500) were obtained from 7.57–7.49 (m, 2H), 7.43 (dd, J = 8.7, 1.8 Hz, 1H), 6.86 (d, J = 8.8 Hz, 1H),
J&K Scientific. Synthesis and characterization of compounds 1 to 6 6.27–6.20 (m, 1H), 5.19 (s, 2H), 2.55 (s, 5H), 2.41–2.37 (m, 3H), 1.07
(Extended Data Fig. 1c, d) are described and shown in the Supplemen- (d, J = 1.3 Hz, 16H), 1.04 (s, 3H) (Supplementary Fig. 4). High-perfor-
tary Methods, Supplementary Figs. 2–27. Other chemical reagents and mance liquid chromatography (HPLC)–mass spectrometry was used
solvents used in this study were purchased from Sigma-Aldrich, J&K to analyse the stability of the synthesized TESO–coumarin, TBSO–cou-
Scientific, Energy Chemical or Thermo Fisher Scientific. All nuclear marin and TIPSO–coumarin (Supplementary Figs. 22–24).
magnetic resonance (NMR) spectra were recorded at room temperature
Synthesis and isolation of the TESO–linker (compound 7) and
the TBSO–linker (compound 7′) for NP conjugation
on a Bruker Avance 400-MHz or 600-MHz spectrometer. Signals are
presented as ppm, and multiplicity is identified as single (s), broad (br),
doublet (t), triplet (t), quartet (q) or multiplet (m); coupling constants Compound 6 (1.0 equivalent, 1 mmol) and DBTL (0.05 equivalent,
are in Hz. The concentration of the compounds was performed by 0.05 mmol) were dissolved in acetone (10 ml) under N2 atmosphere.
rotary evaporation without heating at an appropriate reduced pressure. The solution was stirred under refluxing THF for 48 h. Without washing,
Chemistry yields refer to the isolated pure chemicals.
the organic layer was removed under reduced pressure. The residue
Complementary DNA (cDNA) for mouse Gsdma3 was synthesized by was purified by flash chromatography to obtain compound 7. This
our in-house gene-synthesis facility. cDNA for eGFP and mNeonGreen synthetic route is shown in Extended Data Fig. 1d. 1H NMR (400 MHz,
were gifts from P. Xu. The cDNA was cloned into a modified pET vector chloroform-d) δ 7.75–7.68 (m, 1H), 7.53 (dd, J = 5.6, 3.3 Hz, 2H), 7.35
with an N-terminal 6×His-SUMO tag for recombinant expression in (s, 1H), 6.74 (d, J = 8.6 Hz, 1H), 5.08 (s, 2H), 3.81 (d, J = 6.0 Hz, 2H), 3.57
Escherichia coli. cDNA for mNeonGreen-NLS was generated by standard (s, 2H), 3.46–3.38 (m, 3H), 3.21 (q, J = 4.8, 3.5 Hz, 2H), 2.67 (s, 6H), 1.71
PCR cloning strategy, with a reverse primer containing the SV40 NLS (dd, J = 14.7, 6.8 Hz, 2H), 1.43 (t, J = 7.2 Hz, 3H), 0.98 (t, J = 7.9 Hz, 11H),
sequence (5′-CCGAAAAAACGTAAAGTT-3′). The cDNA was inserted into 0.76 (q, J = 7.9 Hz, 8H) (Supplementary Fig. 9). HPLC and high-resolution
a modified pCS2-3×Flag vector for transient expression in HeLa cells. mass spectrometry were used to analyse the quality of the synthesized
Primers used for generating point mutations were designed using an TESO–linker (Supplementary Figs. 25, 27).
online program (https://www.agilent.com/store/primerDesignPro-
gram.jsp). All plasmids were verified by DNA sequencing.
TBSO–linker (compound 7′) 1H NMR (600 MHz, DMSO-d6) δ 7.51–7.50
(m, 1H), 7.29–7.25 (m, 1H), 6.81–6.79 (m, 1H), 6.66 (s, 2H), 4.92 (s, 1H), 4.45
For fluorescence-activated cell sorting (FACS) analyses of tumour- (s, 2H), 3.74–3.72 (t, 2H), 3.44–3.41 (m, 2H), 3.17–3.11 (t, 2H), 2.45–2.33
infiltrating lymphocytes, PE-conjugated anti-mouse CD3 (clone 17A2), (s, 6H), 2.03–1.95 (m, 2H), 1.88–1.71 (m, 4H), 0.97 (s, 9H), 0.20 (s, 6H)
FITC-conjugated anti-mouse CD4 (clone RM4.5) and APC-conjugated (Supplementary Fig. 10).
anti-mouse CD8 (clone 53-6.7) were purchased from BioLegend.
eFluor-450-conjugated anti-mouse FOXP3 antibody (clone FJK-16s)
was obtained from Invitrogen. The trastuzumab used to generate the
Chromatography analysis to determine the efficiency of the
desilylation reaction
antibody–drug conjugate was a gift from the Beijing People’s Hospi- TESO–coumarin, TBSO–coumarin and TIPSO–coumarin (150 μM)
tal. The PD1 antibody used for treating 4T1 tumours was a gift from were treated with Phe-BF3 (150 μM) in PBS (including 5% DMSO) at
BeiGene. For immune-cell depletion, anti-mouse CD4 (clone GK1.5) and 37ꢀ°C. HPLC analysis was performed at 5 and 240 min after incubation
isotype control (clone LTF-2) antibodies were produced by BioXcell, (Extended Data Fig. 3d, e). For bioorthogonality evaluation (Extended
and anti-mouse CD8 was a gift from J. Sui. Anti-LAT1 antibody (D-10) Data Fig. 3f), TESO–coumarin (150 μM) was treated with H2O2, GSH or
was obtained from Santa Cruz Biotechnology. Anti-GFP (11814460001) other biologically relevant anions, including Cl−, I−, NO3−, PO43− and
and anti-Na, K-ATPase α1(2047-1) were obtained from Roche and SO42− (150 μM) in PBS (including 5% DMSO) at 37ꢀ°C. HPLC analysis was
Epitomics, respectively. Anti-GAPDH and anti-Flag antibodies were performed at 240 min after incubation. To compare the reaction effi-
purchased from Sigma-Aldrich. Anti-GSDME (ab215191) and anti- ciency of Phe-BF3 with those of other fluorine donors (Extended Data
GSDMD (ab219800) were from Abcam. Anti-IL-1β (GTX74034) was Fig. 3g), TESO–coumarin (150 μM) was treated with Phe-BF3 or another
obtained from Genetex.
organofluorines in PBS (including 5% DMSO) at 37ꢀ°C. HPLC analysis
was performed at 240 min after incubation. An Agilent Eclipse XDB-C18
5-μm, 4.6 × 250-mm analytical column was used for the HPLC analyses,
and leflunomide (c = 20 μM) was added as the internal standard. Sol-
Synthesis and isolation of TESO–coumarin, TBSO–coumarin-
and TIPSO–coumarin
The synthetic routes for TESO–coumarin, TBSO–courmarin and TIPSO– vent A was water (0.1% TFA); solvent B was MeCN; 0 to 2 min: 5% B, 2 to
coumarin are essentially the same, and are illustrated in Extended Data 10 min: 5% to 95% B, 10 to 15 min: 95% B, 15 to 17 min: 95% to 5% B; flow
Fig. 1c. In brief, coumarin–NCO (500 mg, 2.48 mmol) was dissolved in rate: 0.6 ml/min; column temperature: 19 to 21ꢀ°C. The reaction yields
dry THF (15 ml), and then mixed with compound 5 (1.0 equivalent, 2.37 were determined by HPLC peak integration at the given wavelength
mmol) and dibutyltin dilaurate (DBTL, 0.05 equivalent, 0.12 mmol) (254 nm) (Supplementary Figs. 16–21). In addition, the desilylation rate
under N2 atmosphere. The solution was stirred under refluxing THF for constant was determined by liquid chromatography analysis as previ-
24 h. Without washing, the organic layer was removed under reduced ously described28. All the measurements were performed in triplicate
pressure. The residue was purified by flash chromatography to obtain and the average numbers are shown.
silyl-ether-conjugated coumarin as the white solid (40% yield). TESO–
coumarin: 1H NMR (600 MHz, DMSO-d6) δ 7.75–7.65 (m, 2H), 7.62–7.51
Evaluation of MMAE release efficiency in vitro and in vivo
(m, 2H), 7.44 (dt, J = 8.8, 2.6 Hz, 1H), 6.85 (dd, J = 8.5, 3.9 Hz, 1H), 6.24 (s, To evaluate the desilylation efficiency in vitro, trastuzumab–SEC–
1H), 5.14 (d, J = 15.1 Hz, 2H), 2.48–2.30 (m, 9H), 0.92 (dt, J = 25.9, 7.8 Hz, MMAE stock solution (100 nM, 3 ml) and Phe-BF3 stock solution (20
9H), 0.75 (q, J = 8.0 Hz, 3H) (Supplementary Fig. 2). TBSO–coumarin: μM, 3 ml) were mixed and incubated at 37ꢀ°C in PBS for 3 h, 15 h, 20 h or

24 h. One hundred microlitres of acetonitrile was added to the crude NP–mNeonGreen–NLS, 100 μl of N3–OTES–maleimide linker (com-
product to remove the protein, and the crude product was then centri- pound 7, TESO–linker) (PBS, 500 μM) was added to 100 μ; of GFP or
fuged at 10,000 rpm for 3 min. Ten microlitres of the supernatant was mNeonGreen–NLS solution (PBS, 100 μM) at 4ꢀ°C for 24 h. Excess linker
analysed, and the content of MMAE released was measured by liquid was removed from the solution by ultrafiltration centrifugation 4 times,
chromatography–mass spectrometry analysis. A Hypersil GOLD 1.9- and the modified GFP or mNeonGreen–NLS protein was obtained.
μm 2.1 × 100-mm analytical column was used for the HPLC analyses. One hundred microlitres of each of the modified proteins (20 μM)
Solvent A was water (0.1% FA); solvent B was MeCN; 0 to 2 min: 5% B, 2 was reacted with 100 μl of DBCO–PEG–NP solution (0.2 mg/ml) with
to 10 min: 5% to 95% B, 10 to 14 min: 95% B, 14 to 15 min: 95% to 5% B, 15 stirring at room temperature for 30 min. After overnight storage at
to 17 min: 5% B; flow rate: 0.3 ml/min; column temperature: 19 to 21ꢀ°C. 4ꢀ°C, the reaction was centrifuged at 5,000g for 5 min, decanted and
The reaction yields were determined by HPLC peak integration at the resuspended in water to remove excess unconjugated proteins. To pre-
given mass spectrum signal. All the measurements were performed in pare NP–GSDMA3, all operations were carried out at 4ꢀ°C. One hundred
triplicate and the average numbers are shown. To evaluate the desilyla- microlitres of N3–OTES–maleimide linker (PBS, 500 μM) was added
tion efficiency in vivo, the tumour-bearing mice that had been intrave- slowly to 100 μl of the GSDMA3(N+C) nonvalent complex (60 μM)
nously injected with trastuzumab–SEC–MMAE were divided into two solution, and the reaction proceeded for 6 h. Excess linker was removed
groups, which were left untreated or treated with Phe-BF3 every two by overnight dialysis. After centrifugation to remove the precipitate,
days. The tumours were collected on the fifth day (n = 4), frozen with the modified GSDMA3 protein was obtained and resuspended in PBS.
liquid nitrogen, ground into powder and then extracted by methanol. One hundred microlitres of the modified GSDMA3 protein (20 μM) was
After removing the methanol by rota-evaporation, the precipitates incubated with 100 μl of DBCO–PEG–NP solution (0.2 mg/ml) for 6 h.
were redissolved in PBS, in which MMAE was quantified by methods The mixture was used directly to treat cells or mice.
described in previous publications29,30
.
Transmission electron microscopy and negative-staining
electron microscopy
Preparation and characterization of trastuzumab–SEC–MMAE
Trastuzumab in 4.8 ml of PBS (10 mg/ml) was treated with tris(2-car- 4T1 cells (1 × 106) were implanted into the right flank of BALB/c female
boxyethyl)phosphine (2.3 equivalents) at 37ꢀ°C for 30 min to make mice (6–8 weeks old). Mice were intravenously injected with NP–
the reduction reaction was complete. The reaction crude product was GSDMA3 on day 6. The 4T1 tumours (about 100 mm3) were collected
diluted with PBS until the antibody concentration was 2.5 mg/ml, and on day 8. The freshly isolated tumours were minced into tissue blocks
cooled to 4ꢀ°C. The reduced antibody was then reacted with maleimide– (about 2 mm3) and fixed with 2% paraformaldehyde and 2% glutaralde-
SEC–MMAE (about 8.0 equivalents) in dimethyl sulfoxide stock solu- hyde immediately, followed by secondary fixing with 1% OsO4 and en
tion for 1 h on ice. The conjugation reaction was quenched by a 20-fold bloc staining with 2% uranyl acetate. Samples were then dehydrated
excess of cysteine over maleimide. The reaction crude products were through a progressive lowering of temperature method (an ascending
concentrated by centrifugal ultrafiltration and the trastuzumab–SEC– acetone series 15–100% in 7 steps). The samples were then embed-
MMAE conjugation product was purified by elution through a PD10 ded in SPI-PON 812 resin (Sigma-Aldrich) and sectioned to 90 nm. The
column in PBS and then stored at −80ꢀ°C for analyses and experiments. ultrathin sections were stained with 3% uranyl acetate for 7 min and
The drug-to-antibody ratio was determined by UV–visible spectrum Sato’s lead for 2 min before imaging on a 120-kV Tecnai G2 Spirit elec-
and matrix-assisted laser desorption/ionization time-of-flight (MALDI tron microscope (FEI). Reconstitution and imaging of the GSDMA3 pore
TOF) mass spectrometry as previously described31.
by negative-staining electron microscopy were performed following
the same protocol as previously described20.
Radiochemistry and PET imaging
[
18F]Phe-BF3 was radiosynthesized via the one-step 18F–19F isotope
Side-effect analysis in mouse tumour model
exchange (IEX) reaction. The labelling method and purification pro- For the histological assessment in Extended Data Fig. 7b, tumour-
cedure have previously been described7,32. PET scans were obtained, and bearing mice were killed after the indicated treatments, and the kid-
image analysis was performed, using a Mediso 122 s PET scanner. About ney and the liver were removed and fixed overnight with 4% formalin.
3.7 MBq of [18F]Phe-BF3, 89Zr-labelled antibody–drug conjugates and After dehydration by gradient ethanol treatment, the tissue samples
89Zr-labelled protein–nanoparticle conjugates were administered via were embedded in paraffin and sectioned for haematoxylin and eosin
tail-vein injection under isoflurane anaesthesia. Standard data acquisi- staining. Independent experiments were performed with three mice
tion and image reconstruction of the PET data were performed.
for each experimental group. Blood samples of mice treated with Phe-
BF3 or PBS were also collected for serum isolation. The serum levels of
alkaline phosphatase (P0321, Beyotime), calcium (S1063S, Beyotime)
Preparation of DBCO–PEG-decorated nanoparticles
The water-soluble spherical gold nanoparticles (A68095) were generated and phosphate (DIPI007, BioAssay Systems) were determined follow-
by 3A Chemicals. PEG-decorated nanoparticles were prepared according ing the manufacturers’ instructions.
to published methods33,34. In brief, a solution of nanoparticles (60 nm)
Cell culture and transfection
containing citric acid was centrifuged at 14,000g for 5 min, decanted
and resuspended in water to remove excess citric acid. Ten microlitres of HeLa, EMT6 and CT26 cells were obtained from the American Type
10 mM DBCO–PEG3400–SH were added to 1 ml of 0.2 mg/ml nanoparticle Culture Collection. The 4T1 cells were obtained from the China Infra-
solution. The mixture was stirred for 30 min at room temperature, and structure of Cell Line Resources. Primary BMDMs were prepared and
40 μl of 10 mM MeO–PEG5000–SH were then added, stirring for 24 h at cultured by following a standard protocol, as previously described2.
4ꢀ°C. The reaction crude products were centrifuged at 14,000gfor 5 min, The cells were frequently checked by their morphological features and
decanted and resuspended in water to remove excess PEG. The size of the functionalities, but were been subjected to authentication by short
DBCO–PEG-decorated nanoparticles (DBCO–PEG–NPs) was confirmed tandem repeat profiling. All cell lines were tested for mycoplasma regu-
by transmission electron microscopy (Supplementary Fig. 28).
larly, using the commonly used PCR strategy. The 4T1, EMT6 and CT26
cells and primary BMDMs were cultured in RIPM 1640 medium and
HeLa cells were cultured in Dulbecco’s modified Eagle’s medium. The
media were supplemented with 10% (v/v) fetal bovine serum (FBS) and
Preparation of NP–GFP, NP–mNeonGreen–NLS and NP–GSDMA3
materials
The desired nanoparticle–protein conjugates were prepared 2 mM L-glutamine. All cells were cultured in a 5% CO2 incubator at 37ꢀ°C.
following previous publications35,36. In brief, to obtain NP–GFP and Transient transfection was performed with the JetPRIME (Polyplus

Article
Transfection) by following the manufacturer’s instructions. Full-length sequentially for further purification. To obtain the GSDMA3(N + C)
GSDMA3 or GSDMA3(N + C) proteins were electroporated into CT26 noncovalent complex, the purified engineered GSDMA3 protein was
cells using the Neon Transfection System (Life Technologies).
digested overnight with homemade PPase at 4ꢀ°C. The GSDMA3(N + C)
complex was further purified by Superdex G75 gel-filtration chroma-
tography in the absence of reductants. 2-Mercaptoethanol present
Immunostaining and fluorescence microscopy
Frozen 4T1 tumour sections were used for anti-CD3 immunostaining. At in the mNeonGreen–NLS protein was also removed at Superdex G75
the end of indicated treatments on day 16, 4T1 tumours were dissected gel-filtration chromatography step.
from mice and embedded into Cryomold moulds filled with the OCT
Stimulation of cells with nanoparticle–protein conjugates and
Phe-BF3
compound (SAKURA 4583). The tissue samples were frozen on dry ice
for 30 min and stored at −20ꢀ°C. For immunostaining, the frozen tumour
tissues were sectioned into 10-μm pieces using cryostat (Lecia CM1950) HeLa, EMT6 and 4T1 cells, and primary BMDMs, seeded in 6-well plates
and mounted on slides. After washing out the OCT with PBS, the slices were treated with the NP–GSDMA3 conjugates for 24 h. Subsequently,
were fixed with 4% paraformaldehyde for 30 min and then blocked and the medium containing NP–GSDMA3 was replaced with medium con-
permeabilized with PBS containing 1% FBS and 0.2% Triton X-100 for 1 h. taining Phe-BF3, and incubated for another 24 h. The cells were then
The slides were stained with PE-conjugated anti-mouse CD3 antibody subjected to flow cytometry analyses. For confocal microscopy imag-
for 1 h and Hoechst for 1 min. Zeiss LSM800 confocal laser scanning ing assay of GFP release from NP–GFP, primary BMDMs were seeded
microscope was used to acquire images on the 10× or 40× objectives. onto glass coverslips in 24-well plates and treated sequentially with the
The 4T1 tumour model was used for quantifying Phe-BF3-medi- NP–GFP conjugate for 24 h and Phe-BF3 for another 24 h. The treated
ated release of mNeonGreen–NLS from NP–mNeonGreen–NLS. The cells were washed with PBS and fixed with 4% paraformaldehyde. The
NP–mNeonGreen–NLS conjugates were administered into 4T1 tumour- nucleus was stained with Hoechst. Fluorescence images were captured
bearing mice through intravenous injections on day 6, 9 and 12 fol- by using the Zeiss LSM800 confocal microscope on a 20× objective.
lowed by intravenous injections of Phe-BF3 on day 7, 8, 10, 11, 13 and
In vivo propidium iodide staining assay of tumour cell
pyroptosis
14. On day 15, mice were killed and tumours were dissected from the
surrounding fascia, embedded in the Cryomold moulds filled with
the OCT compound and frozen. Processing of the tumour tissue was For propidium iodide labelling of pyroptotic cells in vivo, mice bear-
performed as described for anti-CD3 staining and the slides were ing 4T1 tumours were administered propidium iodide (2.5 mg/kg) via
stained with Alexa-Fluor-647-conjugated phalloidin (A22287, Thermo intravenous injection at 24 h after the last round of treatment with
Fisher Scientific). Zeiss LSM850 confocal microscope was used to NP–GSDMA3 and Phe-BF3 (Fig. 4b). Ten minutes later, the mice were
record the images on a 60× oil objective.
killed and the tumours were collected, placed into OCT-containing
Cryomold moulds and frozen. After slicing and mounting, the slides
were scanned directly and immediately on the Zeiss LSM800 confocal
Pyroptosis assays
HeLa, EMT6 and 4T1 cells, and primary BMDMs, were seeded into a 6- or microscope.
96-well plate 24 h before being subjected to the indicated treatments.
Tumour model and FACS analyses of tumour-infiltrating
lymphocytes
To examine cell morphology, annexin V–FITC and propidium iodide
were added to the cell-culture medium before subjected to imaging on
an LSM800 confocal microscope. To examine cell morphology, static All mice used were purchased from the Vital River Laboratories. To
bright-field images of pyroptotic cells were captured using an Olympus construct the tumour model, BGC823 (1 × 106), 4T1 (1 × 106) or EMT6
IX71 microscope. The image data shown are representative of at least (2 × 106) cells in 100 μl of PBS were implanted into the right flank of Nu/
three randomly selected fields. For flow cytometry analysis, all cells in Nu nude female mice (for BGC823 cells) or BALB/c female mice (6–8
each 6-well plate were collected and washed twice with PBS, stained weeks old). Mice were intravenously injected with NP–GSDMA3 or an
by using the annexin V–FITC and PI staining kit (Abmaking) (annexin indicated control on day 6, 9 and 12, and each of the 3treatments was
V–FITC for 10 min and propidium iodide for 5 min). The sample volume followed by 2 intravenous injections of Phe-BF3 (on day 7, 8, 10, 11, 13
was increased to 500 μl by adding more PBS, and the samples were ana- and 14), as illustrated in Fig. 4b. For the doxycycline tumour model in
lysed on a BD FACS Aria III flow cytometer. Data were processed using Fig. 5b, normal 4T1 cells mixed with 0%, 10% or 30% 4T1-GSDMA3dox-on
FlowJo software. Cell viability and LDH release assays were determined cells (total cell number of 1.5 × 106) were implanted into the right
by using the CellTiter-Glo Luminescent Cell Viability Assay (Promega) flank of BALB/c female mice (6–8 weeks old). Doxycycline (20 μg for
and LDH-Glo Cytotoxicity Assay (Promega), respectively. A membrane each mouse) or PBS were intratumourally injected into the mice at
protein isolation kit (SM-005, Invent Biotechnologies) was used to day 6. For the combinatory therapy in Fig. 5g–i, anti-PD1 antibody
detect the translocation of the GSDMA3 N domain to the cell membrane. (5 mg/kg) was intraperitoneally injected into the mice on day 9, 11, 13
and 15 after a single round of treatment with NP–GSDMA3 and Phe-BF3
Purification of recombinant proteins
from day 6 to day 8. When the tumour was palpable, its long diameter
To obtain engineered GSDMA3 protein containing the PreScission (L) and short diameter (W) were measured every 3–5 days with a cal-
protease (PPase) cleavage site and the mNeonGreen–NLS proteins, liper; the tumour volume was determined using the volume formula
E. coli BL21 (DE3) cells containing pET28a-6×His-SUMO-GSDMA3 or for an ellipsoid (that is, 1/2 × L × W2). Mice were killed by CO2 inhalation
mNeonGreen–NLS were grown in Luria–Bertani (LB) medium supple- when the tumour reached 2,000 mm3. The same protocol was used
mented with 30 μg/ml kanamycin. After the optical density at 600 nm to construct the 4T1 tumour model in the Nu/Nu mice. For FACS of
of the culture reached 0.8, 0.4 mM isopropyl-B-D-thiogalactopyrano- tumour-infiltrating lymphocytes, tumours were collected on day 16
side (IPTG) was added to induce protein expression at 18ꢀ°C overnight. after inoculation. The tumours were dissected from the surrounding
Bacteria were collected and sonicated in the lysis buffer containing fascia, weighted, minced into pieces by sterile scissors and ground. Cell
20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 20 mM imidazole and 10 mM clumps were removed through a 70-μm cell strainer to obtain single-cell
2-mercaptoethanol. The fusion protein was first affinity-purified by suspensions. The suspension was centrifuged and the cell pellets were
Ni-Sepharose beads (GE Healthcare Life Sciences). The homemade washed twice with PBS containing 1% BSA (FACS buffer). Lymphocytes
ULP1 protease was used to remove the 6×His–SUMO tag by overnight were isolated by Percoll density-gradient centrifugation, washed and
cleavage at 4ꢀ°C. HiTrap Q ion-exchange and Superdex G75 gel-filtration resuspended in the FACS buffer, blocked with anti-mouse CD16/CD32
chromatography (GE Healthcare Life Sciences) were then performed (clone 93, BioLegend) for 30 min, and finally stained with the indicated

antibodies for another 1 h. The LIVE/DEAD Fixable Near-IR Dead Cell dimensions (according to the limits defined by the IACUC protocol),
Stain Kit (L10119, Invitrogen) was used to determine cell viability during and no mouse had severe abdominal distension (≥10% increase of origi-
FACS analysis. The FOXP3 Fixation/Permeabilization Kit (00-5521-00, nal body weight).
Invitrogen) was used to stain the intracellular FOXP3, following the
manufacturer’s instructions.
Reporting summary
Further information on research design is available in the Nature
Research Reporting Summary linked to this paper.
Single-cell RNA sequencing
The tumour-infiltrating lymphocytes isolated as in ‘Tumour model and
FACS analyses of tumour-infiltrating lymphocytes’ were stained with
PE-conjugated anti-mouse CD45 antibody (clone 30-F11, Biolegend).
Data availability
CD45+ immune cells were then enriched by using a BD FACS Aria III flow All data supporting the findings of this study are included in the Article
cytometer. Cell viability was monitored in real time during preparation and its Supplementary Information. Single-cell RNA sequencing of
of the single CD45+ immune-cell suspension. Ten thousand cells (about tumour-infiltrating immune cells were also deposited at the National
600 single cells per microlitre) from each experimental group were Genomics Data Center (NGDC) under accession number PRJCA002149
barcoded and pooled using the 10x Genomics device. Samples were (https://bigd.big.ac.cn/bioproject/browse/PRJCA002149). Source Data
prepared following the manufacturer’s protocol and sequenced on an for Figs. 1–5, Extended Data Figs. 2–7, 10 are available with the paper.
Illumina NextSeq sequencer. The Cell Ranger analysis pipeline (v.3.0.2)
was used for sample demultiplexing, barcode processing, alignment,
filtering, UMI counting and aggregation of the sequencing runs. For
quality control of the single-cell RNA-sequencing procedure, cells with
less than 300 genes detected —as well as cells with transcript counts for
mitochondria-encoded genes of more than 15% of the total transcript
counts—were removed from subsequent analyses. Genes detected in
fewer than three cells across the dataset were also excluded, yielding a
preliminary expression matrix of 18,069 cells. After obtaining the digital
gene-expression data matrix, Seurat (v.3.0.0.9000) was used for dimen-
sion reduction, clustering and differential gene-expression analyses.
28. Tu, J., Xu, M., Parvez, S., Peterson, R. T. & Franzini, R. M. Bioorthogonal removal of
3-isocyanopropyl groups enables the controlled release of fluorophores and drugs
in vivo. J. Am. Chem. Soc. 140, 8410–8414 (2018).
29. Rossin, R. et al. Chemically triggered drug release from an antibody–drug conjugate
leads to potent antitumour activity in mice. Nat. Commun. 9, 1484 (2018).
30. Zheng, Y. et al. Enrichment-triggered prodrug activation demonstrated through
mitochondria-targeted delivery of doxorubicin and carbon monoxide. Nat. Chem. 10,
787–794 (2018).
31. Lyon, R. P. et al. Reducing hydrophobicity of homogeneous antibody–drug conjugates
improves pharmacokinetics and therapeutic index. Nat. Biotechnol. 33, 733–735
(2015).
32. Liu, Z. et al. One-step 18F labeling of biomolecules using organotrifluoroborates. Nat.
Protocols 10, 1423–1432 (2015).
33. Cheng, Y. et al. Highly efficient drug delivery with gold nanoparticle vectors for in vivo
photodynamic therapy of cancer. J. Am. Chem. Soc. 130, 10643–10647 (2008).
34. Duncan, B., Kim, C. & Rotello, V. M. Gold nanoparticle platforms as drug and
biomacromolecule delivery systems. J. Control. Release 148, 122–127 (2010).
35. De, M. et al. Sensing of proteins in human serum using conjugates of nanoparticles and
green fluorescent protein. Nat. Chem. 1, 461–465 (2009).
Immune-cell depletion and cytokine neutralization
Each tumour-bearing mouse was intraperitoneally administered 200 μg
of anti-mouse CD4, anti-mouse CD8 or the isotype-control antibody
on day 5, 8, 11 and 14 after inoculation of the 4T1 cells. To verify the
depletion efficiency, the percentage of CD4+ or CD8+ T lymphocytes
in the spleen was determined on day 16 by using a BD FACSAria III flow
cytometer. For cytokine neutralization experiments, 100 μg of anti-
IL-1β (BE0246, BioXcell) or anti-IL-18 (BE0237, BioXcell) was intraperi-
toneally injected into each tumour-bearing mouse 3 times (once every
3 days). Two hundred micrograms of anti-HMGB1 (a gift from F. Zheng)
was intraperitoneally injected into the mouse five times every two days
to neutralize the HMGB1 protein.
36. Rana, S., Yeh, Y. C. & Rotello, V. M. Engineering the nanoparticle–protein interface:
applications and possibilities. Curr. Opin. Chem. Biol. 14, 828–834 (2010).
Acknowledgements We thank J. Sui, Z. Shen, F. Zheng and P. Xu for reagents; J. Chen, F. Wang,
X. Jia, Z. Jiang and W. He for technical assistance; M. Shi for the cartoon illustration; and X. Liu,
P.R. Chen, T. Luo and W. Wei for advice. The work was supported by NSFC grants U1867209 and
21778003 to Z.L., NSFC Basic Science Center Project (81788104), the National Key Research
and Development Program of China (2017YFA0505900 and 2016YFA0501500), Chinese
Academy of Medical Sciences Initiative for Innovative Medicine (2019-I2M-5-084) to F.S.
Author contributions Z.L. and F.S. conceived the study; Q.W., assisted by C.W. and X.Z.,
performed material synthesis, characterization and chemical analysis; Y.W. and Q.W.
performed most of the experiments; C.W. performed radiosynthesis and PET imaging; J.D., W.G.
and H.H. provided technical assistance and suggestions. Q.W., Y.W., F.S. and Z.L. analysed the
data. F.S. and Z.L. wrote the manuscript with input from all authors. All authors discussed the
results and commented on the manuscript.
Animal ethics and general protocols of animal studies
All mouse studies were conducted in accordance with the principles and
procedures outlined in the Guide for the Care and Use of Laboratory
Animals (Ministry of Health, China) and the protocol was approved by
the Institutional Animal Care and Use Committee (IACUC) of Peking
University (IACUC ID: CCME-LiuZB-1). The tumour-bearing mice were
subjected to indicated treatments and small-animal PET studies when
the tumour volume reached 100 mm3 (about 1 week after inoculation)
and 100–300 mm3 (2–3 weeks after inoculation), respectively. Ethi-
cal compliance with the IACUC protocol was maintained. In none of
the experiments did the size of tumour graft surpass 2 cm in any two
Competing interests The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41586-020-
2079-1.
Correspondence and requests for materials should be addressed to F.S. or Z.L.
Peer review information Nature thanks Dmitri Krysko, Andreas Linkermann and the other,
anonymous, reviewer(s) for their contribution to the peer review of this work.
Reprints and permissions information is available at http://www.nature.com/reprints.

Article
Extended Data Fig. 1 | Phe-BF3 catalyses desilylation of carbamate linkers
containing silyl ether, and synthetic routes of related compounds.
a, Chemical structures of silyl-phenolic-ether-conjugated coumarin
derivatives. b, Proposed mechanism for Phe-BF3-catalysed desilylation of the
silyl ether that triggers decarboxylation on the carbamate and consequent
release of the coumarin. c, Synthetic route of TESO–coumarin. TBSO–
coumarin and TIPSO–coumarin were synthesized via a similar strategy.
d, Synthetic route of the silyl-ether-containing carbamate linker used for
nanoparticle conjugation. e, Synthetic route of Phe-BF3-responsive SEC–MMAE
for antibody conjugation.

Extended Data Fig. 2 | Establishment of bioorthogonal chemistry based on
desilylation that can achieve efficient and controlled MMAE release from
the trastuzumab–SEC–MMAE antibody–drug conjugate system. a, Liquid
chromatography (LC) assay of cleavage of TBSO–coumarin induced by Phe-BF3
desilylation. Red and blue mark the maximum absorbance wavelength of free
coumarin (365 nm) and the mixtures of Phe-BF3 and TBSO–coumarin (254 nm),
respectively. b, Assay of possible desilylation of TBSO–coumarin (10 μM) by
various biologically relevant nucleophilic anions (5 mM for GSH, 20 mM for
H2O2 and 10 mM for others). The fluorescence intensity of coumarin
MMAE is approximately 4. g, h, HPLC–QTOF mass spectrometry analysis of
MMAE released from trastuzumab–SEC–MMAE after incubation with Phe-BF3
for 3 h (magenta), 15 h (blue), 20 h (brown) or 24 h (orange) in 50% serum.
h, Quantification of the released MMAE using calibration curves. Data shown
are mean s.e.m. n = 3 independent replicates. i–l, Nu/Nu mice were implanted
subcutaneously with human BGC823 cancer cells. i, Dynamic PET computed
tomography 3D projection images of mice bearing BGC823 tumours at the
indicated time points after intravenous injection of [89Zr]trastuzumab–SEC–
MMAE. j, Time–activity curve (TAC) of [89Zr]trastuzumab–SEC–MMAE in the
blood, liver and tumour. k, Local concentration of Phe-BF3 and trastuzumab–
SEC–MMAE in BGC823 tumours after tail-vein injection of the corresponding
agents. l, Ex vivo biodistribution of [18F]Phe-BF3 in mice bearing BGC823
tumours at 75 min after injection. Mean s.d., n = 4 mice. Data shown are
representative of three (a–i) or two (l) independent experiments.
(mean s.e.m., n = 4 independent replicates) was determined at λex = 370 nm
and λem = 435 nm. c, Determination of the rate constant of desilylation mediated
by Phe-BF3, by chromatography assay. Data shown are mean s.e.m.
n = 3 independent replicates. d, Purity of the trastuzumab–SEC–MMAE
conjugate determined by size-exclusion HPLC. e, f, UV–visible (e) and MALDI–
TOF (f) analyses reveal that the drug-to-antibody ratio of trastuzumab–SEC–

Article
Extended Data Fig. 3 | Optimization of the silyl-ether-containing carbamate
linker for higher sensitivity to Phe-BF3. a, Schematic of Phe-BF3-catalysed
desilylation that can release the ‘caged’ coumarin from a designed silyl-ether-
containing carbamate linker. b–d, ‘Decaging’ and liquid chromatography
assays of Phe-BF3-catalysed desilylation of TESO–coumarin (TESO–C),
TBSO–coumarin (TBSO–C) or TIPSO–coumarin (TIPSO–C). b, c, Photographs
(b) and quantification (c) of coumarin fluorescence. d, Blue and magenta mark
the maximum absorbance wavelength of TESO–coumarin or TBSO–coumarin
(254 nm) and free coumarin (365 nm), respectively. e, Summary of the
desilylation efficiency of Phe-BF3 and NaF towards TESO–coumarin,
TBSO–coumarin or TIPSO–coumarin. f, Fluorescence-emission assay of
possible desilylation of TESO–coumarin (10 μM) by various biologically
relevant nucleophilic anions (5 mM for GSH, 20 mM for H2O2 and 10 mM for
others). The coumarin fluorescence intensity was determined at λex = 370 nm
and λem = 435 nm. g, Summary of desilylation efficiency of TESO–coumarin
(10 μM) by FDA-approved organofluorines. e, g, Reaction efficiency was
determined by HPLC. ND, not detected. Data shown are representative of two
(c, f) or three (b, d, e, g) independent experiments.

Extended Data Fig. 4 | See next page for caption.

Article
Extended Data Fig. 4 | Release of GFP from NP–GFP induced by Phe-BF3
desilylation, and biodistribution of [89Zr]GFP and NP–[89Zr]GFP in mice.
a, Design of Phe-BF3 desilylation of the silyl-ether-carbamate-linked NP–GFP
for releasing GFP from the nanoparticle. GFP fluorescence is quenched in NP–
GFP; desilylation-induced cleavage of the linker releases the GFP. b, Workflow
of assaying release of GFP from NP–GFP induced by Phe-BF3 desilylation
in vitro. The samples were subjected to immunoblotting in c and Fig. 2a.
c, Comparison of desilylation-induced release of GFP from TESO- or TBSO-
linked NP–GFP by Phe-BF3 and NaF. The loading control (*) was sampled before
centrifugation. d, Expression of LAT1 transporter in the four cell types assayed
in Extended Data Fig. 4e. e, Assays of release of GFP from NP–GFP induced by
Phe-BF3 desilylation in cells. HeLa, EMT6 or 4T1 cells, or primary BMDMs, were
treated with NP–GFP (1 mg ml−1) for 24 h and then with Phe-BF3 (100 μM) or NaF
for another 24 h. Centrifuged lysates were subjected to anti-GFP and anti-
GAPDH immunoblotting. f, g, BALB/c mice were implanted subcutaneously
with 4T1 cancer cells. f, Representative PET computed tomography 3D
projection images of mice bearing 4T1 tumours at 1, 6, 12 and 18 h after
intravenous injection of [89Zr]GFP or NP–[89Zr] G FP. g, Time–activity curve of
[
89Zr]GFP and NP–[89Zr]GFP in the blood, liver and tumour of mice.
h, Representative confocal images of HeLa cells transfected with a plasmid
expressing mNeonGreen–NLS. Scale bars, 20 μm. i, Assay of release of
mNeonGreen–NLS from the NP–mNeonGreen–NLS conjugates induced by
Phe-BF3 desilylation in mice bearing 4T1 tumours. Representative confocal
images of tumour sections are shown. Scale bars, 20 μm. Image in the box is
magnified in Fig. 2d. Data shown in c–i are representative of two independent
experiments.

Extended Data Fig. 5 | See next page for caption.

Article
Extended Data Fig. 5 | Phe-BF3 desilylation of NP–GSDMA3 releases the
gasdermin N domain that is capable of forming pores and inducing
pyroptosis. a, b, Preparation of the GSDMA3(N + C) used for conjugation onto
the nanoparticle. Purified GSDMA3 containing a engineered PPase cleavage
site between the gasdermin N and C domain was cleaved in vitro to obtain
GSDMA3(N + C). a, Coomassie blue staining of the prepared GSDMA3 proteins.
b, ATP-based viability of CT26 cells electroporated with the prepared GSDMA3
protein. Mean s.d., n = 3 independent replicates. c, d, Pore-forming activity of
the gasdermin N domain released from NP–GSDMA3 by desilylation catalysed
by Phe-BF3. c, Representative negative-stain electron microscopy images of the
gasdermin pores formed by purified native GSDMA3(N + C) protein or the
GSDMA3(N + C) protein released from NP–GSDMA3. Scale bars, 100 nm.
d, Representative 2D averages of the gasdermin pores viewed from the top
(left) and the side (right). Scale bar, 10 nm. e, f, HeLa or EMT6 cells were treated
with NP–GSDMA3 alone or in combination with Phe-BF3. e, Total cell lysates or
their membrane fractions were immunoblotted with indicated antibodies.
Flag–GSDMA3(N + C) proteins (the Flag tag was fused C-terminal to the
gasdermin N domain, before the PPase cleavage site) were conjugated to the
nanoparticle. The GSDMA3 N domain was probed by anti-Flag
immunoblotting. f, LDH assay of the effect of various inhibitors on cell death
induced by treatment with NP–GSDMA3 and Phe-BF3. Mean s.d.,
n = 3 independent replicates. Fer1, ferrostatin 1; Nec-1, necrostatin 1. g, HeLa,
EMT6 or 4T1 cells, or primary BMDMs, were treated as indicated. NP + GSDMA3,
nanoparticles mixed with the noncovalent GSDMA3(N + C) complex.
GSDMA3(Mut), the pore forming-deficient E14K/L184D mutant version of
GSDMA3(N + C). Flow-cytometry plots of PI- and annexin V–-FITC-stained cells
are shown. h, Effect of phagocytosis inhibitor cytochalasin D on cell death
induced by treatment with NP–GSDMA3 and Phe-BF3 in primary BMDMs. Flow-
cytometry plots of PI- and annexin V–Alexa-647-stained cells are shown. Data
are representative of two (c–f, h) or three (a, b, g) independent experiments.

Extended Data Fig. 6 | See next page for caption.

Article
Extended Data Fig. 6 | Biodistribution and cytodistribution of NP–[89Zr]
GSDMA3, and effect of treatment with NP–GSDMA3 and Phe-BF3 on mice
bearing 4T1 tumours. a–e, BALB/c mice were implanted subcutaneously with
4T1 cancer cells, and intravenously injected with [89Zr]GSDMA3 or NP–[89Zr]
GSDMA3. a, b, PET computed tomography 3D projection images of mice
bearing 4T1 tumours at the indicated time points after injection of [89Zr]
GSDMA3 or NP–[89Zr]GSDMA3. c, Time–activity curve of NP–[89Zr]GSDMA3 in
the blood, liver and tumour of injected mice. d, Systemic biodistribution of
NP–[89Zr]GSDMA3 in mouse tissues. Data shown are mean s.e.m. n = 4 mice.
e, Representative transmission electron microscopy images showing the
cytodistribution of NP–GSDMA3 in the 4T1 tumour tissue. Nu, nucleus; M,
mitochondrion; NP, nanoparticle; RBC, red blood cell. Scale bars, 5 μm (A1),
500 nm (A2–A4). f, g, Immunoblotting assay of endogenous IL-1β, GSDMD and
GSDME expression. f, 4T1 or MH-S cells were stimulated with LPS (2 μg ml−1) for
12 h and the cell lysates were subjected to anti-IL-1β or anti-tubulin
immunoblotting. g, Lysates of EMT6 and 4T1 cells were blotted with anti-
GSDMD, anti-GSDME or anti-tubulin antibodies. h, i, Mice bearing 4T1 tumours
were treated with NP–GSDMA3 and Phe-BF3, as depicted in Fig. 4b (n = 7 mice
for each group). Average tumour volumes of each group of mice at the
indicated time points after the injection are shown as mean s.e.m. (in h) with
P values in i (two-tailed unpaired Student’s t-test). j, Propidium iodide staining
of tumour-cell pyroptosis induced by treatment with NP–GSDMA3 and Phe-BF3
in mice. An independent experiment from that shown in Fig. 5a. Scale bars,
100 μm. All data shown are representative of two (a–g, j) or three (h, i)
independent experiments.

Extended Data Fig. 7 | See next page for caption.

Article
Extended Data Fig. 7 | Evaluation of the possible side effects of treatment
with NP–GSDMA3 and Phe-BF3 in mice. a, Records of mouse body weight after
the indicated treatments. Data shown are mean s.e.m. n = 8 mice for PBS, NP–
GSDMA3, Phe-BF3 and intratumourally injected NP–GSDMA3 + Phe-BF3, 9 mice
for NP–GSDMA3 + Phe-BF3 and NP–GSDMA3(mut) + Phe-BF3. b, Representative
haematoxylin and eosin (H & E) staining of liver and kidney tissues from mice
with indicated treatments. Three mice per group. c–e, Assay of serum
concentrations of alkaline phosphatase, calcium and phosphate in mice
treated with PBS (n = 4 mice) or Phe-BF3 (n = 5 mice). Data shown are
and high-resolution (right) computed tomography scans of the joint from mice
treated with Phe-BF3 or PBS. For the normal-resolution scan, a 3D
reconstruction image of the whole joint is shown on the left; representative 2D
images of the sagittal plane (top) and axial plane near the femur growth plate
(bottom) are shown on the right. For the high-resolution computed
tomography scan that images a voxel near the femur growth plate, the whole
voxel is shown on the left and representative images of the sagittal plane (top)
and the axial plane (bottom) are shown on the right. Data are representative of
three (a) or two (b–f) independent experiments.
mean s.e.m. Two-tailed unpaired Student’s t test. f, Normal-resolution (left)

Extended Data Fig. 8 | Pyroptosis induced by NP–GSDMA3 and Phe-BF3
stimulates inflammation and increases the tumour-infiltrating
lymphocytes. a, d, e, Gating strategy (a) and representative flow-cytometry
plots for assessing 4T1 tumour-infiltrating CD3+ T cells (d) or
following treatment with PBS (n = 4 mice), NP–GSDMA3 and Phe-BF3 (n = 4 mice)
or NP–GSDMA3(mut) and Phe-BF3 (n = 6 mice). Scale bars, 200 μm (b), 50 μm
(c, left), 20 μm (c, right). f, Flow-cytometry analysis of CD4+ or CD8+ T cells upon
depletion by their corresponding antibody. Data shown are representative of
two (b, c, f) or three (a, d, e) independent experiments.
FOXP3+CD4+ regulatory T cells (e) following the indicated treatments.
b, c, Representative fluorescence images of CD3–PE-stained 4T1 tumours

Article
Extended Data Fig. 9 | Tumour-infiltrating immune-cell-subtype analysis by
single-cell RNA sequencing. a, Gating strategy and representative flow-
cytometry plots for the enrichment of 4T1 tumour-infiltrating single
CD45+ immune cells. b, Heat map of ten immune-cell clusters with unique
signature genes. Colours on top of the map indicate the immune-cell clusters.
The three or four marker genes used for each cluster are listed alongside the
cluster. c, Signature gene-expression patterns for the corresponding cell
clusters on the t-SNE plot (n = 4, 18,069 cells). d, t-SNE plots of tumour-
infiltrating single CD45+ immune cells of 4T1 tumours from mice treated with
PBS (n = 2 mice, 10,171 cells) or NP–GSDMA3 and Phe-BF3 (n = 2 mice, 7,898 cells)
(left), and the relative frequencies of different clusters (right). e–g, Expression
levels of protumoural and immunosuppressive genes (e), proinflammatory
chemokine (f) and T and/or natural killer cell activation or effector (g) genes in
immune cells. Paired quantile–quantile (Q–Q) plots were used to compare the
gene-expression levels in CD45+ immune cells between 4T1 tumours treated
with PBS or NP–GSDMA3 and Phe-BF3. P values were calculated using a two-
sided Wilcoxon rank-sum test. All data are representative of two independent
experiments.

Extended Data Fig. 10 | Antitumour immunity activated by NP–GSDMA3 and
Phe-BF3 requires IL-1β, and can synergize with the anti-PD1 therapy. a,
Enzyme-linked immunosorbent assay measurements of IL-1β, IL-18 and HMGB1
concentration in the serum (top) and the tumour homogenates (bottom) of
mice treated with PBS, NP–GSDMA3 and Phe-BF3 or NP–GSDMA3(mut) and
Phe-BF3 (n = 5 or 7, as shown in the figure for each group). b, Anti-IL-1β, IL-18 or
HMGB1 antibodies were intraperitoneally injected into mice bearing 4T1
tumours, before treatment with NP–GSDMA3 and Phe-BF3 (n = 7 mice each
group). Mice bearing 4T1 tumours were also treated with PBS alone as the
control group (n = 8). Average tumour volumes at the indicated time points
after implantation are shown. c, Mice bearing 4T1 tumours were treated with
PBS (n = 7 mice), anti-PD1 (n = 8 mice) or NP–GSDMA3 and Phe-BF3 (n = 7 mice)
alone, or in combination (n = 8 mice) as shown in Fig. 5i. All data are shown as
mean s.e.m. (two-tailed unpaired Student’s t-test) and are representative of
two independent experiments.

Corresponding author(s):
Feng Shao
Life Sciences Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form is intended for publication with all accepted life
science papers and provides structure for consistency and transparency in reporting. Every life science submission will use this form; some list
items might not apply to an individual manuscript, but all fields must be completed for clarity.
For further information on the points included in this form, see Reporting Life Sciences Research. For further information on Nature Research
policies, including our data availability policy, see Authors & Referees and the Editorial Policy Checklist.
Please do not complete any field with "not applicable" or n/a. Refer to the help text for what text to use if an item is not relevant to your study.
For final submission: please carefully check your responses for accuracy; you will not be able to make changes later.
` Experimental design
1. Sample size
No statistical methods were used to predetermine the sample sizes. It is impossible to predict
the magnitude of experimental variation between animals based on our current knowledge.
The group sizes (at least three animals per treatment group) represents the minimum
number animals needed to reach statistical significance (p < 0.05) between experimental
groups.
Describe how sample size was determined.
2. Data exclusions
Describe any data exclusions.
3. Replication
There were no data exclusions.
All attempts at replication are successful. Experiments results were robust and reproducible.
The difference between the treated group and control group were statistically significant in
the repetitive experiments.
Describe the measures taken to verify the reproducibility
of the experimental findings.
4. Randomization
No randomization was used in this study. For all in vivo experiments, animals were randomly
assigned into a treatment group after tumour inoculation. The starting tumour burden in the
treatment and control groups was similar before treatment.
Describe how samples/organisms/participants were
allocated into experimental groups.
5. Blinding
No blinding was done in this study. Most of the studies contained multiple steps (including
the material preparation, mouse tumour treatment, and so on) and the scientists must keep
careful track of conditions. It would be exceedingly difficult to blind such studies.
Describe whether the investigators were blinded to
group allocation during data collection and/or analysis.
Note: all in vivo studies must report how sample size was determined and whether blinding and randomization were used.
1

6. Statistical parameters
For all figures and tables that use statistical methods, confirm that the following items are present in relevant figure legends (or in the
Methods section if additional space is needed).
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement (animals, litters, cultures, etc.)
A description of how samples were collected, noting whether measurements were taken from distinct samples or whether the same
sample was measured repeatedly
A statement indicating how many times each experiment was replicated
The statistical test(s) used and whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of any assumptions or corrections, such as an adjustment for multiple comparisons
Test values indicating whether an effect is present
Provide confidence intervals or give results of significance tests (e.g. P values) as exact values whenever appropriate and with effect sizes noted.
A clear description of statistics including central tendency (e.g. median, mean) and variation (e.g. standard deviation, interquartile range)
Clearly defined error bars in all relevant figure captions (with explicit mention of central tendency and variation)
See the web collection on statistics for biologists for further resources and guidance.
` Software
Policy information about availability of computer code
7. Software
Videos were processed using the Volocity software (PerkinElmer). All the flow cytometry data
were processed using FlowJo (version 10.4.0, Becton, Dickinson & Company).
Describe the software used to analyze the data in this
study.
For manuscripts utilizing custom algorithms or software that are central to the paper but not yet described in the published literature, software must be made
available to editors and reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). Nature Methods guidance for
providing algorithms and software for publication provides further information on this topic.
` Materials and reagents
Policy information about availability of materials
8. Materials availability
All unique materials used are readily available from the authors.
Indicate whether there are restrictions on availability of
unique materials or if these materials are only available
for distribution by a third party.
9. Antibodies
PE-conjugated anti-mouse CD3 (Cat # 100206, clone 17A2, Lot: B257196, 1:100), FITC-
conjugated anti-mouse CD4 (Cat # 100510, clone RM4.5, Lot: B245229, 1:100), and APC-
conjugated anti-mouse CD8 (Cat # 100712, clone 53-6.7, Lot: B244174, 1:100) were
purchased from BioLegend. eFluor 450-conjugated anti-mouse Foxp3 antibody (Cat #
14-5773-82, clone FJK-16s, Lot: 4332193, 1:100) was obtained from Invitrogen. Trastruzumab
antibody used to generate the ADC was a gift from Beijing People’s Hospital. The PD1
antibody used for treating 4T1 tumours was a gift from BeiGene. For immune cell depletion,
anti-mouse CD4 (Cat # BE0003-1, clone GK1.5, Lot: 699918O1B) and isotype control (Cat #
BE0090, clone LTF-2, Lot: 695618J2) antibodies were produced by BioXcell, and anti-mouse
CD8 was a gift from Dr. J. Sui (National Institute of Biological Sciences, Beijing). Anti-LAT1
antibody (Cat # sc-374232, D-10, Lot: C0518, 1:1000) was obtained from Santa Cruz
Biotechnology. Anti-GFP was a gift from Dr. Ting Han (National Institute of Biological Sciences,
Beijing), originally purchased from Roche (Cat # 11814460001, 1:5000). Anti-Na, K-ATPase α1
was a gift from Dr. Xiaodong Wang (National Institute of Biological Sciences, Beijing),
originally purchased from Epitomics (Cat # 2047-1, 1:1000). Anti-GAPDH (Cat # G8795,
1:5000) and anti-Flag (Cat # F1804, D-10, Lot: 124K6106, 1:5000) antibodies were purchased
from Sigma-Aldrich. Anti-GSDME (Cat # ab215191, 1:1000) and anti-GSDMD (Cat # ab219800,
1:1000) were from Abcam. Anti-IL-1β (Cat # GTX74034, Lot: 42900, 1:1000) was obtained
from Genetex.
Describe the antibodies used and how they were validated
for use in the system under study (i.e. assay and species).
All antibodies for FACS and immune cell depletion were well-recognized clones in the field
and validated by the manufacturers. These antibodies are further validated and routinely
used in our lab. Antibodies targeting GFP, LAT1, Na, K-ATPase α1, GAPDH, GSDMD, GSDME,
IL-1β, and Flag were validated by immunoblotting and all the detected-bands were matched
with the predicted molecular weight.
2

10. Eukaryotic cell lines
a. State the source of each eukaryotic cell line used.
HeLa and mouse EMT6 and CT26 cells were obtained from the American Type Culture
Collection (ATCC). Mouse mammary carcinoma 4T1 cells were obtained from the China
Infrastructure of Cell Line Resources (Chinese Academy of Medical Sciences, Beijing, China).
Primary bone marrow-derived macrophage (priBMDM) cells were prepared and cultured by
following a standard protocol as previously described (Reference 4)
b. Describe the method of cell line authentication used.
Identity of the cell lines were frequently checked by their morphological features but have
not been authenticated by the short tandem repeat (STR) profiling.
c. Report whether the cell lines were tested for
mycoplasma contamination.
All cell lines were tested to be mycoplasma-negative by the standard PCR method.
No commonly misidentified cell lines are used in this study.
d. If any of the cell lines used are listed in the database
of commonly misidentified cell lines maintained by
ICLAC, provide a scientific rationale for their use.
` Animals and human research participants
Policy information about studies involving animals; when reporting animal research, follow the ARRIVE guidelines
11. Description of research animals
6-8 week-old female mice (BALB/c background or the Nu/Nu nude mice) were used for all
animal experiments described in this study.
Provide all relevant details on animals and/or
animal-derived materials used in the study.
Policy information about studies involving human research participants
12. Description of human research participants
There are no human subjects involved in this study.
Describe the covariate-relevant population
characteristics of the human research participants.
3