Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Article
Potentiating anti-tumor immunity by re-engaging
immune synapse molecules
Graphical abstract
Highlights
dAbsence of ICAM-1 drives tumor resistance to T and NK cell-
mediated killing
dTumor ICAM-1 is repressed by UHRF1/DNMT1-mediated
DNA methylation
dLFA-1 engager restores ICAM-1 signaling and enhances T cell
cytotoxicity
dLFA-1 engager boosts anti-tumor immunity and synergizes
with ICB therapies
Authors
Xindi Zhou, Tian Xu, Changhe Li, ...,
Yangxin Fu, Zexian Zeng, Deng Pan
Correspondence
zexianzeng@pku.edu.cn (Z.Z.),
dpan@tsinghua.edu.cn (D.P.)
In brief
Zhou et al. reveal that DNA methylation-
driven ICAM-1 loss confers tumor
resistance to T and NK cell killing.
Restoring ICAM-1/LFA-1 signaling with
‘‘LFA-1 engager’’ enhances anti-tumor
immunity and boosts immune checkpoint
blockade efficacy, presenting a
promising therapeutic strategy for ICAM-
1-deficient tumors.
Zhou et al., 2025, Cell Reports Medicine 6, 101975
March 18, 2025 ª2025 The Authors. Published by Elsevier Inc.
https://doi.org/10.1016/j.xcrm.2025.101975 ll
Article
Potentiating anti-tumor immunity
by re-engaging immune synapse molecules
Xindi Zhou,
1
Tian Xu,
2
Changhe Li,
1
Yufeng He,
2
Yuanzhi Hu,
1
Hao Gong,
1
Jiahui Li,
1
Haitao Jiang,
1
Liang Wen,
5
Yangxin Fu,
1
Zexian Zeng,
2,4,
*and Deng Pan
1,3,6,
*
1
State Key Laboratory of Molecular Oncology, School of Basic Medical Sciences, Tsinghua University, Beijing 100084, China
2
Peking-Tsinghua Center for Life Sciences, Academy for Advanced Interdisciplinary Studies, Peking University, Beijing 100871, China
3
Tsinghua-Peking Center for Life Science (CLS), Beijing 100084, China
4
Center for Quantitative Biology, Academy for Advanced Interdisciplinary Studies, Peking University, Beijing 100871, China
5
Chinese People’s Liberation Army (PLA) Medical School, Beijing 100850, China
6
Lead contact
*Correspondence: zexianzeng@pku.edu.cn (Z.Z.), dpan@tsinghua.edu.cn (D.P.)
https://doi.org/10.1016/j.xcrm.2025.101975
SUMMARY
The formation of immune synapses (ISs) between cytotoxic T cells and tumor cells is crucial for effective tu-
mor elimination. However, the role of ISs in immune evasion and resistance to immune checkpoint blockades
(ICBs) remains unclear. We demonstrate that ICAM-1, a key IS molecule activating LFA-1 signaling in T and
natural killer (NK) cells, is often expressed at low levels in cancers. The absence of ICAM-1 leads to significant
resistance to T and NK cell-mediated anti-tumor immunity. Using a CRISPR screen, we show that ICAM-1 is
epigenetically regulated by the DNA methylation pathway involving UHRF1 and DNMT1. Furthermore, we
engineer an antibody-based therapeutic agent, ‘‘LFA-1 engager,’’ to enhance T cell-mediated anti-tumor
immunity by reconstituting LFA-1 signaling. Treatment with LFA-1 engagers substantially enhances
immune-mediated cytotoxicity, potentiates anti-tumor immunity, and synergizes with ICB in mouse models
of ICAM-1-deficient tumors. Our data provide promising therapeutic strategies for re-engaging immune
stimulatory signals in cancer immunotherapy.
INTRODUCTION
Cytotoxic T cells (cytotoxic T lymphocytes [CTLs]) and natural
killer (NK) cells are key effectors of the immune system that
play a central role in the elimination of tumor cells.
1–3
CTL activa-
tion is primarily driven by the interaction of the T cell receptor
(TCR) with the major histocompatibility complex (MHC)-peptide
complex presented on tumor cells. Similarly, the activation of NK
cells is driven by the engagement of major NK-activating recep-
tors, including B7-H6 and natural killer group 2 member D
(NKG2D).
4
However, tumors have developed various intrinsic
mechanisms to evade recognition and killing by CTLs and NK
cells. These mechanisms include the downregulation of MHC
expression,
5,6
upregulation of immune inhibitory molecules
(e.g., programmed death-ligand 1 [PD-L1] and SPERPINB9),
7–9
downregulation or shedding of major NK-activating ligands
(e.g., MHC Class I chain-related protein A/B [MICA/B]),
10
and
secretion of immune-suppressive cytokines (e.g., transforming
growth factor b) to reprogram the tumor microenvironment,
11
al-
lowing tumors to persist and grow despite the presence of CTLs
and NK. The development of strategies to overcome these
evasion mechanisms is therefore essential for the success of
cancer immunotherapy.
The effective elimination of tumor cells by CTLs and NK cells is
critically dependent on the formation of the immune synapse, a
specialized interface between the target cells or antigen-pre-
senting cells and T cells. The formation of the immune synapse
involves the interaction between lymphocyte function-associ-
ated antigen 1 (LFA-1) on T cells and intercellular adhesion mole-
cule 1 (ICAM-1) on target cells. Upon ICAM-1/LFA-1 binding,
T cells extend pseudopodia to scan the surface of target cells
for specific MHC-peptide complexes, which then activate the
TCR.
12–14
This interaction not only facilitates the formation and
stabilization of immune synapses between T cells and their
target cells but also plays a crucial role in lymphocyte migration
to lymphoid tissues and sites of infection and inflammation.
15
Additionally, ICAM-1/LFA-1 interaction could lower the density
of antigens and TCR signals necessary for T cell activation and
functional response.
16
Despite their vital roles in immune recog-
nition and trafficking, the contributions of ICAM-1/LFA-1 to can-
cer immune evasion and the potential to manipulate ICAM-1/
LFA-1 signaling for cancer therapy remain unclear.
In this study, we show that ICAM-1 expression is frequently
low or absent in human and murine cancer cells, presenting
an emerging immune evasion mechanism that enables tumor
cells to evade eradication by CTLs and NK cells. Through a
genome-wide CRISPR-Cas9 screen, we demonstrate that
ICAM-1 is regulated through a variety of epigenetic mechanisms,
particularly the DNA methylation inheritance pathway orches-
trated by Ubiquitin-Like containing PHD and RING fingers
Cell Reports Medicine 6, 101975, March 18, 2025 ª2025 The Authors. Published by Elsevier Inc. 1
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
ll
OPEN ACCESS
Figure 1. Silencing of ICAM-1 confers resistance to CTL and NK cell-mediated killing
(A) The mRNA expression level of MHC and co-stimulatory molecules among 1,086 human cancer cell lines from the CCLE database.
(B) Pan-cancer analysis of ICAM-1 expression level on different cancer types from TCGA database. ICAM-1 median expression value is show in red line.
(C and D) FACS analysis of surface ICAM-1 level on indicated human (C) and murine (D) cancer cell lines. Murine cells were either untreated or treated with IFN-g
(50 ng/mL) for 24 hours.
(E) In vitro competition assay of tumor and CTL co-culture. Control (sgControl) SW480 cells were either mixed with tdTomato-labeled control (sgControl) cells or
ICAM-1 KO cells. These mixture cells were then co-cultured with NY-ESO-1-specific T cells or control T cells without the expression of TCR against NY-ESO-1.
Log
2
fold changes of the percentage of mixture SW480 cells upon co-culture with NY-ESO-1-specific CTLs as compa red with that co-cultured with control T cells
were shown (n= 3).
(legend continued on next page)
2Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
domain 1 (UHRF1) and DNA methyltransferase 1 (DNMT1). In
response to the diminished ICAM-1 expression on tumor cells,
we developed an innovative antibody-based therapeutic agent,
termed ‘‘LFA-1 engager,’’ which reconstitutes ICAM-1 signaling
on tumor cells. The use of LFA-1 engagers effectively compen-
sates for the absence of ICAM-1, substantially enhancing im-
mune-mediated cytotoxicity and boosting anti-tumor immunity
in mouse models. Thus, our results present a promising cancer
therapy approach by reactivating the ICAM-1/LFA-1 signaling.
RESULTS
The absence of tumor-intrinsic ICAM-1 confers
resistance to both CTL and NK cell-mediated killing
To identify the immune evasion mechanisms that would lead to
comprised immune stimulatory signals and the formation of the
immune synapse, we analyzed the mRNA expression levels of
MHC-I molecules (B2M, histocompatibility leukocyte antigen
[HLA]-A,HLA-B, and HLA-C) and the co-stimulatory/adhesion
molecules (CD58 and ICAM1), among 1,086 human cancer cell
lines using the Cancer Cell Line Encyclopedia (CCLE) database.
Notably, as compared with HLA molecules and CD58, a signifi-
cant proportion of cancer cells exhibited markedly low expres-
sion of ICAM-1, a key immune synapse molecule interacting
with the integrin receptor LFA-1 on CTLs and NK cells (Figure 1A).
To gain a comprehensive understanding of ICAM-1 expression
across different cancers, we performed a pan-cancer analysis
using The Cancer Genome Atlas (TCGA) database. Cancer types
were categorized into three groups—ICAM-1 high, intermediate,
and low expression—based on the median ICAM-1 expression
across all samples. Our analysis revealed that many cancer
types exhibited high ICAM-1 expression, while others showed
intermediate or low levels, reflecting significant variability in
ICAM-1 expression across cancer types (Figure 1B). To validate
these findings, we assessed ICAM-1 expression by flow cytom-
etry (fluorescence-activated cell sorting [FACS]) using a panel of
human cancer cell lines available in our laboratory. We observed
variable ICAM-1 expression across these lines. The A549 non-
small cell lung cancer (NSCLC) cell line showed detectable
ICAM-1 expression, while the MDA-MB-231 triple-negative
breast cancer (TNBC) cell line exhibited high levels of ICAM-1
(Figure 1C). In contrast, the SKBR3 breast cancer cell line
showed no detectable ICAM-1 expression (Figure 1C). Interest-
ingly, ICAM-1 was undetectable in several murine cell lines,
including B16F10, EMT6, Lewis lung carcinoma (LLC), MC38,
and 4T1 cells (Figure 1D). Treatment with interferon-gamma
(IFN-g) failed to induce ICAM-1 expression in these murine lines,
with the exception of 4T1 cells, which showed an induced
response (Figure 1D). These findings suggest that ICAM-1
expression is highly variable both among and within cancer
types, with absent or low ICAM-1 expression observed in multi-
ple cancer cell lines.
To investigate the functional relevance of ICAM-1 in immune
evasion, we used CRISPR-Cas9 to knock out the ICAM1 gene
in A498 and SW480 cells, both of which express high levels of
ICAM-1 (Figures 1C and S1A). Importantly, knockout (KO) of
ICAM-1 on A498 and SW480 tumor cells did not result in alter-
ations in tumor growth, morphology, or MHC-I levels compared
to control tumor cells (Figures S1B–S1F). We also assessed key
differentiation markers, including CD44 and Ep-CAM,
17–19
which
showed comparable expression levels between ICAM-1 KO and
control cells (Figure S1G). These ICAM1 KO and control cells
were then engineered to express tumor antigen NY-ESO-1 for
specific killing mediated by CTLs expressing NY-ESO-1-specific
TCR. By using CTLs derived from independent healthy donors,
we showed that ICAM-1 KO cells were substantially more resis-
tant to CTL-mediated killing (Figures 1E and S2A–S2E). Consis-
tent with reduced killing phenotype, we also observed a sub-
stantial reduction in IFN-glevels in T cells when co-cultured
with ICAM-1 KO cells as compared to control, indicating that
ICAM-1/LFA-1 signaling plays a critical role in the effector func-
tion of CTLs upon their interaction with tumor cells (Figure 1F). To
determine whether restoring ICAM-1 expression could enhance
tumor cell sensitivity to CTL-mediated killing, we ectopically ex-
pressed Icam1 in B16F10 cells by utilizing a minimal cytomega-
lovirus promoter, ensuring its expression remained within a
physiologically relevant expression level (Figure S2F). Overex-
pression of Icam1 did not change the level of MHC-I (Figure S2G).
As expected, overexpression of Icam1 rendered B16F10 tumors
much more sensitive to killing mediated by OT-I and Pmel-1
T cells, respectively (Figures S2H and S2I).
As LFA-1 is also expressed on NK cells, we examined
whether ICAM-1 silencing could confer resistance to NK cell-
mediated killing. In both A498 and SW480 cell lines, ICAM-1
KO cells were much more resistant to killing mediated by NK-
92MI cells (Figures 1G and S2J). Consistently, the levels of
CD107a in NK cells were also reduced upon co-culture with
ICAM-1 KO cells as compared with control (Figure S2K).
Together, these findings collectively suggest that the silencing
of ICAM-1 in tumors confers resistance to both CTL and NK
cell-mediated killing.
To assess the clinical relevance of ICAM-1 expression and its
association with the response to immune checkpoint blockade
(ICB), we analyzed published bulk RNA sequencing (RNA-seq)
data obtained from tumors collected from patients before they
received ICB treatment. We found that in 3 different cohorts
from various cancer types,
20–22
the expression of ICAM-1 is
significantly lower in non-responders as compared to re-
sponders (Figure 1H), suggesting that low expression of
ICAM-1 is relevant to resistance of ICB in human cancers.
(F) FACS analysis of intracellular IFN-gproduction in NY-ESO-1-specific CTLs upon co-culture with SW480 tumor cells with indicated genotype. Representative
FACS (left) and summary (right) showing the percentage of IFN-g-producing CTLs in the indicated conditions (n= 3).
(G) In vitro competition assay of tumor and NK-92-MI co-culture. Log
2
fold changes of the percentage of mixture SW480 cells upon co-culture with NK-92MI cells
as compared with that without NK-92MI co-culture were shown (n= 3).
(H) Tumor ICAM-1 expression levels of the responders and non-responders of ICB were shown in the indicated clinical cohorts.
Data are presented as means ±SEM (E–G). *p< 0.05, **p< 0.01, ***p< 0.001, and ****p< 0.0001 by one-way ANOVA (E–G). Data are representative of at least two
independent experiments (C–G).
Cell Reports Medicine 6, 101975, March 18, 2025 3
Article
ll
OPEN ACCESS
Tumor-intrinsic ICAM-1 is critical for immune evasion
for both MHC-I-sufficient and deficient tumors
To determine the significance of ICAM-1 in anti-tumor immunity
in vivo, we inoculated mice with ICAM-1-overexpressing cells
(Icam1
OE
)(Figure S2F). In both B16F10 and 4T1 models, tumors
overexpressed with Icam1 showed significantly slower growth
as compared to control tumors in immune-competent mice (Fig-
ure 2A). However, both types of tumors grew at similar rates in
immune-deficient NSG mice (Figure 2B), suggesting that tu-
mor-intrinsic Icam1 expression is important for immune-medi-
ated control of tumor growth. Since 4T1 cells express ICAM-1
upon stimulation with IFN-g(Figure 1D), we also generated
Icam1 KO 4T1 cells and assessed their impact on in vivo tumor
growth (Figures S3A and S3B). Consistently, Icam1 KO 4T1 tu-
mors exhibited much faster growth in immune-competent mice
as compared to control tumors, but the effect was abolished in
NSG mice (Figure 2C). We then analyzed the number of tumor-
infiltrating immune cells in Icam1
OE
,Icam1 KO, and control
Figure 2. Tumor-intrinsic ICAM-1 is critical for immune evasion for both MHC-I-sufficient and deficient tumors
(A and B) Vector-transduced or Icam1
OE
B16F10 and 4T1 tumors were inoculated in wild-type mice (A) and NSG mice (B), respectively. Tumor growth curves were
recorded and shown. n= 5–6 mice per group.
(C) Control (sgControl) and Icam1 KO (sgIcam1) 4T1 tumors were inoculated in the wild-type and NSG mice, respectively. Tumor growth curves were recorded
and shown. n= 4–5 mice per group.
(D) Summary of FACS analysis comparing the number of indicated tumor-infiltrating immune cells between control and Icam1 KO 4T1 tumors on day 16 after
tumor inoculation (n= 5–6).
(E) B2m/Icam1 double KO (sgB2m + sgIcam1)orB2m single KO (sgB2m + sgControl) 4T1 tumors were inoculated in the wild-type and NSG mice, respectively.
Tumor growth curves were recorded and shown. n= 4–6 mice per group.
(F) Summary of FACS analysis comparing the number of indicated tumor-infiltrating immune cells between B2m/Icam1 double KO and B2m single KO 4T1 tumors
on day 15 after tumor inoculation (n= 6).
Data are presented as means ±SEM (A–F). *p< 0.05, **p< 0.01, ***p< 0.001, and ****p< 0.0001 by two-way ANOVA (A–C and E) and one-way ANOV A (D and F).
ns, not significant. Data are representative of at least two independent experiments (A–F).
4Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
tumors. The numbers of CD45
+
immune cells, including CD8
+
T cells and granzyme B+ cytotoxic T cells, were substantially
decreased in Icam1 KO 4T1 tumors compared to control
tumors (Figures 2D and S3C). Consistently, these immune cells
were significantly increased in Icam1
OE
tumors (Figure S3D).
Together, these data suggest that tumor-intrinsic ICAM-1
expression is crucial for anti-tumor immune response.
Since ICAM-1 is also crucial for NK-mediated killing, we
generated B2m/Icam1 double KO (dKO) 4T1 cells (Figure S3E).
This allowed us to assess whether ICAM-1 expression is neces-
sary for the immune-mediated control of MHC-I-deficient tu-
mors, which are resistant to CTL-mediated killing but sensitive
to NK-mediated killing.
23
Interestingly, even in the absence of
MHC-I, the Icam1 KO tumors showed accelerated progression
compared to control tumors in immune-competent mice (Fig-
ure 2E). This phenotype was not observed in immunodeficient
NSG mice (Figure 2E), indicating that the expression of ICAM-1
is also crucial for anti-tumor immunity against MHC-I-deficient
tumors. Importantly, the infiltration of CD45
+
immune cells,
including NK cells, was substantially reduced in B2m/Icam1
dKO tumors compared to control tumors (Figure 2F). To assess
the relevance of our findings in human cancer, we conducted an
analysis of TCGA RNA-seq data. We observed a strong and pos-
itive correlation between ICAM-1 expression levels and the esti-
mated levels of CTL and NK cell infiltration in many cancer types,
such as NSCLC, thyroid carcinoma, and metastatic melanoma
(Figure S3F). Taken together, these results highlight the critical
role of tumor-intrinsic ICAM-1 in anti-tumor immunity.
ICAM-1 is epigenetically co-regulated with a wide range
of pro-inflammatory genes in tumor cells
We next investigated the mechanism by which ICAM-1 is regu-
lated in tumor cells. We used two complimentary approaches to
elucidate the regulatory mechanisms associated with ICAM-1
expression. We first examined the genes whose expressions
are co-regulated with ICAM-1. To this end, we sorted A549 cells
into ICAM-1
high
and ICAM-1
low
populations, followed by RNA-seq
and differential gene expression analysis (Figure S4A). Remark-
ably, ICAM-1
high
tumors exhibited elevated expression of genes
involved in type I and II interferon responses (e.g., IRF7,ISG15,
DDX58,IFIT1,andMX1), inflammatory cytokines and chemokines
(e.g., IL6,CCL2,CCL5,CXCL10,andCXCL11), antigen presenta-
tion pathway (e.g., B2M,HLA-A,TAP1,TAP2,andNLRC5), and
cancer antigens (e.g., CT83 and MAGEA1)(Figures S4Band
S4C), suggesting that ICAM-1 is co-regulated with a pro-inflam-
matory gene expression program.
To identify genes and pathways that directly regulate ICAM-1
expression, we conducted a genome-wide CRISPR screen in
A549 cells, in which ICAM-1 expression is low. A549-Cas9 cells
were transduced with a genome-wide single-guide RNA (sgRNA)
library and subsequently sorted into ICAM-1
high
and ICAM-1
low
populations (Figure 3A). The abundance of sgRNAs in these
ICAM-1
high
and ICAM-1
low
groups was compared to that in un-
sorted control cells. To identify the positive regulators of
ICAM-1, we focused on sgRNAs that were depleted in the
ICAM-1
high
population compared to control cells (Table S1). As
expected, sgRNAs targeting ICAM-1 itself were the most
depleted hits in this screen (Figure 3B). Interestingly, sgRNAs
targeting positive regulators of the nuclear factor kB (NF-kB)
pathway, such as RELA,TRAF6, and MAP3K7, were also signif-
icantly depleted in the ICAM-1
high
group (Figures 3B and 3C).
Conversely, sgRNAs targeting deubiquitinases known to nega-
tively regulate NF-kB, such as CYLD,TNFAIP3, and OTUD5,
were enriched in the ICAM-1
high
fraction (Figures 3B and 3C).
This suggests that NF-kB signaling is essential for ICAM-1
expression, which aligns with the observation that ICAM-1 is
co-expressed with many pro-inflammatory genes.
Subsequently, we shifted our focus to the sgRNAs enriched in
the ICAM-1
high
group, as they likely target negative regulators of
ICAM-1. Gene ontology analysis of the top 100 genes with en-
riched sgRNAs revealed a predominance of pathways related
to histone modification and chromatin binding (Figure 3D). This
suggests that ICAM-1 expression might be repressed through
epigenetic mechanisms. These pathways included components
of major epigenetic complexes such as DNA methylation pro-
cesses (UHRF1 and DNMT1), the polycomb repressive
complex (PRC) 1 (RNF2,RING1, and BMI1), PRC 2 (EZH2,
SUZ12, and EED), the Spt-Ada-Gcn5 acetyltransferase (SAGA)
complex (TAF5L,TADA1,TADA2B,MBD2, and TAF6L), and
the nucleosome remodeling factor (NURF) complex (BPTF and
SMARCA5)(Figures 3C and S4D). To validate our screening re-
sults, we used CRISPR-Cas9 to target the components involved
in DNA methylation (UHRF1 and DNMT1), the PRC (EED), the
NURF complex (BPTF), and cohesion regulation (STAG2)in
A549 cells (Table S2;Figure S5A). Indeed, the loss of these
genes consistently led to increased ICAM-1 expression in
A549 cells (Figure 3E), supporting the idea that ICAM-1 expres-
sion is suppressed through various epigenetic mechanisms.
Additionally, RNA-seq analysis reveals that alongside ICAM-1,
a pro-inflammatory gene expression signature is co-upregulated
upon KO of epigenetic regulators (Figures S5B and S5C).
Notably, knocking out components of the DNA methylation in-
heritance pathway, particularly UHRF1 and DNMT1, results in
the most significant upregulation of ICAM-1 and other inflamma-
tory genes (Figures 3E and S5B and S5C).
Additionally, we also validated the role of these epigenetic reg-
ulators in SKBR3 cells (Table S2), in which ICAM-1 expression is
completely silenced. Interestingly, KO of DNA methylation regu-
lators (UHRF1 and DNMT1) led to a significant increase in
ICAM-1 expression in SKBR3 cells (Figure S5D), as well as in
B16F10 and 4T1 murine tumor cells (Figure S5E), confirming
DNA methylation as a major silencing mechanism across
different cell models. However, KO of other regulators, such as
PRC and NURF components, did not induce ICAM-1 expression
in SKBR3 cells (Figure S5D), suggesting that their regulation of
ICAM-1 may be context dependent. These findings suggest
that the DNA methylation inheritance pathway, mediated by
UHRF1-DNMT1, acts as a potent inhibitor of ICAM-1 expression.
UHRF1-DNMT1-mediated methylation is a major ICAM-1
silencing mechanism in cancer cells
As UHRF1 is responsible for DNA methylation by recognizing
hemi-methylated CpGs through its SET and RING-associated
(SRA) domain or specific histone marks, such as H3K9me2/
me3, through its tandem tudor domain (TTD),
24,25
we aimed to
determine which signal and domain are responsible for the
Cell Reports Medicine 6, 101975, March 18, 2025 5
Article
ll
OPEN ACCESS
methylation of ICAM-1. We reconstituted UHRF1 KO cells with
mutant UHRF1, in which the functions of key domains were dis-
rupted by introducing point mutations that have been previously
characterized (Figure 4A).
26–28
Our results showed that reconsti-
tution with a dysfunctional SRA domain, the key domain for
UHRF1 to maintain DNA methylation on the CpG dinucleotides
during DNA replication,
24
failed to suppress ICAM-1 expression
compared to wild-type UHRF1 (Figures 4B and 4C). These data
suggest that UHRF1-mediated recognition of hemi-methylated
CpGs is required to suppress the expression of ICAM-1.
We then performed whole-genome bisulfite sequencing
(WGBS) on UHRF1 KO and control A549 cells, respectively. As
Figure 3. ICAM-1 is co-expressed with a wide range of pro-inflammatory genes and is epigenetically regulated in tumor cells
(A) Workflow of CRISPR screen to identify regulators of ICAM-1 expression. Cas9-expressing A549 cells were transduced with a genome-wide sgRNA library.
CRISPR-edited A549 cells were then sorted into ICAM-1
high
and ICAM-1
low
fractions, followed by genomic DNA extraction and sequencing to determine the
sgRNA abundance.
(B) Volcano plot showing the log
2
fold change and pvalues of ICAM-1 regulators identified from CRISPR screen. The left graph shows the depleted hits (KO of the
gene reduced ICAM-1 expression) and the right graph shows the enriched hits (KO of the gene enhanced ICAM-1 expression). Annotated genes represent the NF-
kB pathway (blue) and epigenetic regulators (red).
(C) Log
2
fold change of sgRNAs against indicated genes in ICAM-1
high
A549 cells as compared with control. Depleted sgRNA (KO leads to reduced ICAM-1) and
enriched sgRNAs (KO leads to enhanced ICAM-1) are labeled in blue and red bars, respectively. The control sgRNAs are indicated by gray bars.
(D) Gene ontology (GO) analysis in top 100 enriched hits from ICAM-1
high
A549 cells of CRISPR screen.
(E) FACS analysis of ICAM-1 level on A549-Cas9 cells expressing control sgRNA or sgRNAs targeting UHRF1, DNMT1, EED, BPTF, and STAG2. The same control
sample was used for all comparisons shown in the panel. Data are representative of two independent experiments (E).
6Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
expected, the control cells showed a normal bi-modal distribu-
tion of CpG methylation pattern, in which most of the CpGs are
either in the methylated or non-methylated state (Figure S6A).
In contrast, UHRF1 KO cells showed a substantial loss of fully
methylated CpGs (Figure S6A), suggesting global hypomethyla-
tion. We next analyzed the differentially methylated sites be-
tween UHRF1 KO and control A549 cells. The majority of hypo-
methylated sites in UHRF1 KO cells were located at intergenic
and intronic regions, whereas about 4% of the hypomethylated
sites were located at the promoter region (Figure S6B). Notably,
we found that the methylation status on the ICAM1 CpG region
(blue shade) was hypomethylated in UHRF1 KO cells as
Figure 4. UHRF1-DNMT1-mediated methylation is a major ICAM-1 silencing mechanism in cancer cells
(A) Illustration of functional domains in UHRF1. The indicated point mutations abolish the corresponding functions of the domains.
(B) Western blot analysis of UHRF1 protein level in control and UHRF1 KO A549 cells expressing indicated UHRF1 mutants.
(C) Mean fluorescence intensity (MFI) of surface ICAM-1 level determined by flow cytometry in cells expressing indicated UHRF1 mutants (n= 3).
(D) RNA-seq and WGBS profiles of ICAM1 in UHRF1 KO and control A549 cells. CpG region is shaded in blue. One of representative biological replicates is shown
for each sample.
(E) Bisulfite sequencing of the ICAM1 CpG region in control (left) and UHRF1 KO (right) A549 cells. Each line represents a single clone (n= 20). Methylated CpG
sites are shown in black circles and unmethylated sites in blank circles. The percentages of overall methylated CpGs are indicated.
(F) Pearson’s correlation of tumor ICAM-1 expression and ICAM1 promoter methylation score from the CCLE datab ase.
(G and H) Control or UHRF1 KO A549 cells co-cultured either with NY-ESO-1-specific CTLs (G) or NK-92MI cells (H) in the presence of isotype (mouse IgG1 kappa
antibodies) or anti-ICAM-1-blocking antibodies (5 mg/mL). Specific lysis percentage was determined by FACS, counting the number of alive cells after co-culture
with NY-ESO-1-specific CTLs or NK-92MI cells, as compared with control group (n=3).
Data are presented as means ±SEM (C and G and H). *p< 0.05 and ****p< 0.0001 by one-way ANOVA (C) and two-way ANOVA (G and H). ns, not significant. Data
are representative of at least two independent experiments (B, C, G, and H).
Cell Reports Medicine 6, 101975, March 18, 2025 7
Article
ll
OPEN ACCESS
Figure 5. Reconstitution of ICAM-1/LFA-1 signaling through fusion protein Cet3ICAM1-D1
(A) Schematic structure of Cet3ICAM1-D1 fusion protein (left) and working hypothesis (right). Cet3ICAM1-D1 is composed of Fab fragment of cetuximab and
murine natural D1 domain of ICAM-1, fused to a ‘‘LALA-PG’’ human Fc fragment. Working hypothesis: in the absence of ICAM-1, the fusion protein could interact
and activate with LFA-1 signaling through the ICAM-1 D1 domain.
(B) Binding affinity of cetuximab and Cet3ICAM1-D1 to EGFR in MC38 cells (n= 3).
(C and D) OT-1 T cells were co-cultured with MC38 (C) and B16F10 (D) tumor cells with serial dilutions of Cet3ICAM1-D1 or cetuximab. FACS analysis showing
the percentage of intracellular IFN-g-producing OT-I T cells (n=3).
(E) OT-1 cells were co-cultured with SIINFEKL-pulsed or unpulsed MC38 tumor cells in the presence of 10 nM Cet3ICAM1-D1. FACS analysis showing the
percentage of intracellular IFN-g-producing OT-I T cells (n= 3).
(legend continued on next page)
8Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
compared to controls (Figure 4D). The hypomethylation of the
ICAM1 CpG region in UHRF1 KO cells was also confirmed by
bisulfite PCR analysis in both A549 and SKBR3 cells
(Figures 4E and S6C). Additionally, we found that the ICAM1 pro-
moter methylation status also highly correlated with ICAM-1
expression in 1,086 human cancer cell lines in the CCLE data-
base (Figure 4F). These data suggest that UHRF1-mediated
DNA methylation is responsible for the silencing of ICAM-1
expression in cancer cells.
To determine whether targeting the DNA methylation pathway
could enhance T and NK cell-mediated killing through ICAM-1 up-
regulation, we conducted in vitro co-culture experiments to
assess the susceptibility of UHRF1-deficient A549 cells to im-
mune cell-mediated killing. We observed that UHRF1 KO signifi-
cantly increased the sensitivity of A549 cells to NY-ESO-1-spe-
cific CTL-mediated killing and NK-92MI cell-mediated killing in
an ICAM-1-dependent manner, as blocking antibodies against
ICAM-1 abolished the sensitive phenotype observedin the co-cul-
ture experiments (Figures 4G and 4H). In line with our genetic find-
ings, treatment with the DNA methylation inhibitor GSK3685032,
29
which targets DNMT1, robustly induced ICAM-1 expression (Fig-
ure S6D) and enhanced the sensitivity of both A549-NY-ESO-1
and SKBR3-NY-ESO-1 cells to NY-ESO-1-specific T cell and
NK92-MI-mediated killing, again in an ICAM-1-dependent manner
(Figures S6E and S6F). Taken together, these data suggest that
UHRF1/DNMT1-mediated DNA methylation is a major mecha-
nism that suppresses ICAM-1 expression and its pro-killing effect
in tumor cells.
Reconstitution of ICAM-1/LFA-1 signaling through
fusion protein Cet3ICAM1-D1
Next, we explored therapeutic strategies for boosting T cell-
mediated killing by specifically reconstituting ICAM-1/LFA-1
signaling in tumor cells in the absence of ICAM-1 expression.
Previous structural studies suggest that the dimerization of the
first Ig-like domain (D1) of ICAM-1 is essential for binding to
LFA-1.
30
Therefore, we engineered a fusion protein, referred as
Cet3ICAM1-D1, that comprises (1) Fab fragments of cetuximab
(Cet) that target epidermal growth factor receptor (EGFR) as a tu-
mor-associated antigen (TAA) and (2) a D1 domain of ICAM-1
fused to each Fc fragment (Figure 5A). This design could facili-
tate the potential dimerization of the D1 domain for effective
LFA-1 binding (Figure S7A). Additionally, the CH2 domains of
the antibodies were engineered with ‘‘LALA-PG’’ mutations
31
that reduce the binding to Fcgreceptor to avoid unwanted anti-
body-dependent cellular cytotoxicity (ADCC) or antibody-
dependent cellular phagocytosis (ADCP) effect mediated by ce-
tuximab. We first confirmed the purity and molecular weight of
the fusion protein by gel electrophoresis (Figure S7B). We then
engineered murine tumor cell lines MC38 and B16F10, both of
which are ICAM-1 negative, to express a mutant version of
chimeric murine EGFR (Figure S7C). In this mutant, six amino
acids were altered to enable the binding of the anti-human
EGFR antibody cetuximab,
32
while maintaining low immunoge-
nicity to ensure successful tumor engraftment. We confirmed
that Cet3ICAM1-D1 binds to the EGFR on the MC38 tumor
cell line as effectively as cetuximab does (Figure 5B).
To assess the ability of Cet3ICAM1-D1 to enhance T cell-
mediated cytotoxicity, we performed in vitro co-culture experi-
ments, in which tumor cells were co-incubated with OT-1 cells
in the presence of either bispecific Cet3ICAM1-D1 or cetuxi-
mab. In both the MC38 and B16F10 cell lines, Cet3ICAM1-D1
substantially increased IFN-gexpression when co-cultured
with SIINFEKL-pulsed tumor cells—a response not observed
with cetuximab (Figures 5C and 5D). Notably, IFN-gproduction
by T cells in response to Cet3ICAM1-D1 only occurs when tu-
mor cells are pulsed with the SIINFEKL antigen, indicating that
Cet3ICAM1-D1 functions in a signal 1 (TCR-MHC-I)-dependent
manner (Figure 5E). To further confirm that the effect is specif-
ically mediated by LFA-1 signaling on T cells, we employed
CRISPR-Cas9 to knock out CD11a, which encodes a subunit
of LFA-1, in OT-I T cells (Figure 5F). As expected, the KO of
CD11a completely abolished the Cet3ICAM1-D1-mediated up-
regulation of IFN-gexpression in OT-I T cells when co-cultured
with SIINFEKL-pulsed MC38 tumor cells, suggesting that Cet3
ICAM1-D1 re-engages the LFA-1 signaling in T cells (Figure 5G).
We then tested the human version of fusion protein Cet3hI-
CAM1-D1, in which the human Ig-like domain (D1) ICAM-1 is
fused to cetuximab (Figure 5H). In our in vitro co-culture experi-
ments, ICAM-1 KO and NY-ESO-1-positive A498 or SW480 cells
with endogenous expression of EGFR were co-incubated with
NY-ESO-1-specific T cells, in the presence of either Cet3hI-
CAM1-D1 or cetuximab (Figure S7D). In both A498 and SW480
cells, the presence of Cet3hICAM1-D1 significantly enhances
IFN-gproduction in CD8
+
T cells (Figures 5I and 5J), consistent
with our findings using mouse tumor cells. Together, these find-
ings indicate that the LFA-1 engager offers a promising strategy
to restore ICAM-1/LFA-1 signaling, thereby enhancing T cell-
mediated cytotoxicity.
LFA-1 engager potentiates anti-tumor immunity in
mouse models
We evaluated the in vivo therapeutic effect of Cet3ICAM1-D1 in
B16F10 and MC38 mouse models by administering the treat-
ment or a cetuximab control on day 7 post tumor inoculation,
with subsequent doses every 3 days. Treatment with Cet3
(F) OT-I T cells were transduced with a control sgRNA (sgControl) or sgRNA targeting Cd11a. Representative FACS plot (left) and summary of mean fluorescence
intensity (MFI) (right) demonstrate high knockout efficiency of Cd11a, as determined by CD11a staining (n= 3).
(G) CD11a KO or control OT-1 cells were co-cultured with MC38 tumor cells in the presence of 10 nM Cet3ICAM1-D1 or cetuximab. Representative FACS (left)
and summary (right) showing the percentage of intracellular IFN-g-producing OT1 cells in the indicated conditions (n= 3).
(H) Structure of the human version of the ‘‘LFA-1 engager,’’ Cet3hICAM1-D1 fusion protein. Cet3hICAM1-D1 comprises the Fab fragment of Cetuximab and the
human natural D1 domain of ICAM-1, fused to a mutant ‘‘LALA-PG’’ human Fc fragment.
(I and J) FACS analysis of intracellular IFN-gproduction in NY-ESO-1-specific CTLs upon co-culture with A498 (I) and SW480 (J) tumor cells with serial dilutions of
Cet3hICAM1-D1 or cetuximab (n= 3).
Data are presented as means ±SEM (B–G and I and J). ***p< 0.001 and ****p< 0.0001 by two-way ANOVA (B–D, G, and I and J) and unpaired Student’s t test
(E and F). ns, not significant. Data are representative of at least two independent experiments (B–G and I and J).
Cell Reports Medicine 6, 101975, March 18, 2025 9
Article
ll
OPEN ACCESS
(legend on next page)
10 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
ICAM1-D1 demonstrated a substantial therapeutic effect in both
models compared to the cetuximab control (Figures 6A and
S7E). Both tumor types exhibited comparable growth rates in im-
mune-deficient NSG mice under Cet3ICAM1-D1 or cetuximab
treatment, indicating that the anti-tumor effects of the LFA-1 en-
gager are immune-mediated regulation (Figure 6B). FACS anal-
ysis of tumor-infiltrating immune cells from B16F10 tumors
showed that Cet3ICAM1-D1 treatment significantly increased
the infiltration of immune cells (CD45
+
), including CD8
+
, CD4
+
T cells, and NK cells (Figure S7F), indicating enhanced immune
cell recruitment to the tumor microenvironment. Additionally,
we assessed the combined in vivo efficacy of Cet3ICAM1-D1
and anti-PD-1 treatment in the B16F10 model. While individual
treatments with either Cet3ICAM1-D1 or anti-PD-1 showed
therapeutic efficacy of the cetuximab control, the combination
of Cet3ICAM1-D1 and anti-PD-1 resulted in much more sub-
stantial tumor growth inhibition and survival benefits
(Figures 6C and S7G). These findings suggest a synergistic
enhancement of anti-tumor immunity when Cet3ICAM1-D1 is
paired with anti-PD-1 therapy.
To further characterize the potential impact of Cet3ICAM1-D1
on the tumor microenvironment, we conducted single-cell RNA-
sequencing (scRNA-seq) on CD45
+
immune cells isolated from
tumors treated with either cetuximab or Cet3ICAM1-D1 in the
MC38 model. Through transcriptomic profiling, we identified
CD8
+
T cells under different functional states, including naive-
like/progenitor (Tcf7
high
), proliferating (Mki67
high
), and effector-
like T cells (Gzmb
high
;Pdcd1
high
)(Figures 6D and 6E). Notably,
treatment with Cet3ICAM1-D1 led to a substantial shift from
naive-like/progenitor to effector-like CD8
+
T cells (Figures 6F
and S7H), while no significant changes were observed in the
myeloid compartment (Figures S8A and S8B).
To validate the scRNA-seq results, we analyzed T cell immu-
noglobulin domain and mucin domain-containing protein 3
(TIM-3, [Havcr2]) and Ly108 (Slamf6), a surrogate of TCF1
(Tcf7) expression, by flow cytometry in CD8
+
T cells. We
observed an increased proportion of TIM3
+
Ly108
effector-like
and granzyme B+ CD8
+
T cells in the Cet3ICAM1-D1-treated tu-
mors, alongside a decreased proportion in TIM3
Ly108
+
naive-
like/progenitor CD8
+
T cells compared to the cetuximab-treated
tumors (Figures 6G and 6H), consistent with scRNA-seq anal-
ysis. Collectively, our data demonstrate that the fusion protein
Cet3ICAM1-D1 enhances the activation and promotes effector
function of CD8
+
T cells within the tumor microenvironment.
DISCUSSION
Immune evasion involves a complex interplay of various intrinsic
factors within tumors. One common strategy for immune evasion
revolves around the downregulation of critical genes pivotal for
T cell-mediated cytotoxicity, including MHC-I molecules and an-
tigen presentation pathways.
33
Nevertheless, genetic mutations
in MHC and antigen-presenting pathways remain infrequent
among patients with cancer. The scarcity of MHC-I loss-of-func-
tion mutations might be attributed to heightened NK-mediated
elimination of MHC-I-deficient cells. Hence, the concurrent
loss of function in pivotal pathways governing both CTLs and
NK cell-mediated killing could strongly favor immune evasion.
In this context, our study underscores that many human and mu-
rine tumors exhibit low or absent expression of ICAM-1, a critical
molecule for immune synapse formation and immune co-stimu-
latory signal. Remarkably, the presence of ICAM-1 is essential
for the effective elimination of these cells by both CTLs and NK
cells. Our findings align with a study demonstrating that reduced
ICAM-1 expression in solid tumors contributes to the inefficacy
of chimeric antigen receptor T cell therapy.
34,35
Furthermore,
leveraging murine models, we demonstrated that reinstating
the ICAM-1 signal within tumor cells significantly enhances
anti-tumor immune responses. Hence, our results indicate that
the absence or reduced levels of ICAM-1 could serve as a pivotal
mechanism for immune evasion.
The process of immune evasion involves multiple steps
involving diverse tumor-intrinsic pathways, such as transcrip-
tional, epigenetic, and metabolic regulations.
36–38
Through
comprehensive genome-wide CRISPR screens, we demon-
strated that ICAM-1 levels could be upregulated by targeting
various epigenetic regulators in A549 cells, including the DNA
methylation inheritance pathway, the PRC, the NURF complex,
and the mammalian SAGA complex. However, in SKBR3 cells,
KO of UHRF1 or DNMT1 resulted in the upregulation of ICAM-1,
whereas KO of other regulators did not have a significant effect,
suggesting that the DNA methylation inheritance pathway is the
primary mechanism suppressing ICAM-1 expression. Given that
DNA methylation is widely dysregulated in cancer cells,
39
the
UHRF1-DNMT1-mediated silencing of ICAM-1 could represent
a widely relevant immune evasion mechanism in human cancers.
Indeed, we showed that using the DNMT1 inhibitor GSK3685032
could restore the sensitivity of tumor cells to T and NK cell-medi-
ated killing in an ICAM-1-dependent manner. However, given the
Figure 6. Fusion protein Cet3ICAM1-D1 potentiates anti-tumor immunity in vivo
(A and B) Wild-type C57BL/6J mice (A) and NSG mice (B) were inoculated with MC38 (left) and B16F10 (right) tumor cells, respectively. The arrows indicate
administrations of Cet3ICAM1-D1 or cetuximab intratumorally (5 mg/kg) to each group of mice. Tumor growth curves were recorded and shown. n= 5–7 mice per
group.
(C) Wild-type C57BL/6J mice were inoculated with B16F10 tumor cells. Mice were treated with Cet3ICAM1-D1, cetuximab, or anti-PD-1 (5 mg/kg) on day 7, 10,
and 13. Tumor growth curves were recorded and shown. n= 6 mice per group.
(D) Equal number of alive CD45
+
cells from cetuximab and Cet3ICAM1-D1-treated MC38 tumors were sorted and pooled for scRNA-seq analysis (n= 7 per
group). Uniform manifold approximation and projection (UMAP) plot showing the distribution of different T cell subsets from scRNA-seq.
(E) UMAP feature plots showing expression of key markers in T cell subsets.
(F) Density plot showing the relative frequency of indicated T cell subsets in cetuximab and Cet3ICAM1-D1-treated MC38 tumors, respectively.
(G and H) Representative FACS plot (left) and summary of the percentage (right) of indicated T cell subsets in cetuximab and Cet3ICAM1-D1-treated MC38
tumors isolated on day 15 after tumor inoculation (n= 5).
Data are presented as means ±SEM (A and B and G and H). **p< 0.01, ***p< 0.001 and ****p< 0.0001 by two-way ANOVA (A–C and G) and unpaired Student’s t
test (H). ns, not significant. Data are representative of at least two independent experiments (A–C and G and H).
Cell Reports Medicine 6, 101975, March 18, 2025 11
Article
ll
OPEN ACCESS
extensive study of the DNMT1 pathway and its druggabilityin can-
cer,
40,41
our study focuses on therapeutic approaches that specif-
ically reconstitute ICAM-1/LFA-1 signaling.
To reconstitute ICAM-1/LFA-1 signaling in ICAM-1 absent tu-
mor cells, we developed antibody-based approaches using an
LFA-1 engager. This engager comprises two modules: one tar-
geting a TAA and another activating LFA-1 signaling. The fusion
protein, Cet3ICAM1-D1, incorporates either the murine or hu-
man D1 domain of ICAM-1,
42,43
which is essential for binding
to LFA-1 and provides a naturalistic engagement of LFA-1
signaling. Our data suggest that this design could effectively
restore ICAM-1/LFA-1 signaling and potentiate anti-tumor im-
munity in mouse models. A key feature of our LFA-1 engager is
its dependency on signal 1 (TCR-MHC-I) signaling for T cell acti-
vation, as the antibodies alone are insufficient to induce activa-
tion. This specificity ensures that without the presence of tumor
antigen, the engager does not affect T cell killing, aligning with
the dual co-stimulatory and adhesion roles of LFA-1 signaling.
Additionally, the use of ICAM-1’s natural D1 domain in the
Cet3ICAM1-D1 construct activates LFA-1 signaling in a more
physiological manner, reducing non-specific binding and poten-
tial unintended effects. This is in contrast to T cell engagers like
CD3 BiTEs, which, although potent in inducing tumor lysis, can
inadvertently bind TAA-expressed normal cells, leading to
toxicity. Thus, our LFA-1 engager presents a promising and safer
approach to potentiate anti-tumor immunity by leveraging the
specificity and natural signaling pathways of immune cells.
In terms of identifying the potential target population for LFA-1
engagers, ICAM-1 expression levels in cancer subtypes need to
be carefully evaluated. For instance, ICAM-1 is generally highly
expressed in NSCLC but is expressed at low levels in small-cell
lung cancer,
44
a subtype that responds poorly to ICB.
45
Similarly,
in breast cancer, ICAM-1 is expressed at relatively high levels in
TNBC but expressed at low levels in non-TNBC.
46–48
These ob-
servations underscore the need to consider specific cancer sub-
types in the context of ICAM-1-mediated immune evasion and the
potential clinical development of the LFA-1 engager.
Limitations of the study
While we have demonstrated that LFA-1 engagers effectively
potentiate anti-tumor immunity in multiple mouse models where
ICAM-1 expression is silenced, it remains unclear whether this ef-
fect can be consistently observed in tumors with high ICAM-1
expression or if the efficacy of LFA-1 engagers is restricted to tu-
mors with low or absent ICAM-1 expression. Clarifying this
distinction is critical for identifying the optimal clinical indications
where LFA-1 engagers would be most effective. Furthermore,
future studies should explore the design of LFA-1 engagers tar-
geting alternative TAAs, such as human epidermal growth factor
receptor 2 (HER2) or Claudin18.2, to expand their applicability
across a broader range of tumor types.
RESOURCE AVAILABILITY
Lead contact
Further information and requests for resources and reagents should be
directed to and will be fulfilled by the lead contact, Deng Pan (dpan@
tsinghua.edu.cn).
Materials availability
Cell lines generated in this study will be provided by the lead contact under a
material transfer agreement.
Data and code availability
dAll datasets in this publication have been deposited in the NCBI Gene
Expression Omnibus (GEO: GSE268920).
dThis paper does not report original code.
dAny additional information related to this paper is available from the lead
contact upon reasonable request.
ACKNOWLEDGMENTS
We thank all the members of the Pan and Zeng labs for their comments and
suggestions. This work was supported by National Natural Science Founda-
tion of China (NSFC) grant (82341026 [D.P.], 82073163 [D.P.], and 12226005
[Z.Z.]), the National Key Research and Development Program of China no.
2022YFC2505400 (D.P.), Tsinghua University Initiative Scientific Research
Program (D.P.), and the Tsinghua-Peking University Center of Life Science
(D.P. and Z.Z.).
AUTHOR CONTRIBUTIONS
Conceptualization, X.Z., C.L., Z.Z., and D.P.; methodology, X.Z., T.X., C.L.,
J.L., L.W., H.J., and Y.F.; validation, X.Z., C.L., Y. Hu, H.G., and J.L.; investiga-
tion, X.Z., C.L., Y. Hu, H.G., and J.L.; formal analysis, X.Z., C.L., Y. Hu, and
H.G.; data curation, X.Z., T.X., Y. He, L.W., Z.Z., and D.P.; writing – original
draft, Z.Z. and D.P.; writing – review and editing, X.Z., T.X., C.L., Y. Hu, and
D.P.; funding acquisition, Z.Z. and D.P.; supervision, Z.Z. and D.P.
DECLARATION OF INTERESTS
Tsinghua University has filed a PCT patent related to this work, on which X.Z.
and D.P. are inventors. D.P. received sponsored research funding from Bayer
AG and Boehringer Ingelheim. These grants were not related to the research
reported in this study.
STAR+METHODS
Detailed methods are provided in the online version of this paper and include
the following:
dKEY RESOURCES TABLE
dEXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
BCell lines
BPrimary cultures
BMice
dMETHOD DETAILS
BGeneration of cell lines
BFACS-based genome wide CRISPR screen
BIn vitro tumor with NY-ESO-1 specific CTLs or NK-92MI cells co-
culture
BCell proliferation assay
BWestern blot
BTracking of Indels by DEcomposition (TIDE) analysis for gene edit-
ing efficiency
BWhole genome bisulfite sequencing (WGBS) and focused bisulfite
sequencing of ICAM1 promoter region
BIn vivo animal studies xenograft mouse models
BFlow cytometry and analysis of tumor infiltrating lymphocytes
BBulk RNA-seq
BSingle cell RNA-seq of immune cells isolated from Cetuximab and
Cet3ICAM1-D1 treated tumors
BSingle cell RNA-seq analysis
BAnalysis of ICB cohorts
BExperiments related to fusion protein
dQUANTIFICATION AND STATISTICAL ANALYSIS
12 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
SUPPLEMENTAL INFORMATION
Supplementary data related to this article can be found online at https://doi.
org/10.1016/j.xcrm.2025.101975.
Received: August 14, 2024
Revised: November 22, 2024
Accepted: January 28, 2025
Published: February 24, 2025
REFERENCES
1. Patel, S.J., Sanjana, N.E., Kishton, R.J., Eidizadeh, A., Vodnala, S.K.,
Cam, M., Gartner, J.J., Jia, L., Steinberg, S.M., Yamamoto, T.N., et al.
(2017). Identification of essential genes for cancer immunotherapy. Nature
548, 537–542. https://doi.org/10.1038/nature23477.
2. Pan, D., Kobayashi, A., Jiang, P., Ferrari de Andrade, L., Tay, R.E., Luoma,
A.M., Tsoucas, D., Qiu, X., Lim, K., Rao, P., et al. (2018). A major chromatin
regulator determines resistance of tumor cells to T cell–mediated killing.
Science 359, 770–775. https://doi.org/10.1126/science.aao1710.
3. Pech, M.F., Fong, L.E., Villalta, J.E., Chan, L.J., Kharbanda, S., O’Brien,
J.J., McAllister, F.E., Firestone, A.J., Jan, C.H., and Settleman, J. (2019).
Systematic identification of cancer cell vulnerabilities to natural killer
cell-mediated immune surveillance. Elife 8, e47362. https://doi.org/10.
7554/eLife.47362.
4. Shimasaki, N., Jain, A., and Campana, D. (2020). NK cells for cancer
immunotherapy. Nat. Rev. Drug Discov. 19, 200–218. https://doi.org/10.
1038/s41573-019-0052-1.
5. Zaretsky, J.M., Garcia-Diaz, A., Shin, D.S., Escuin-Ordinas, H., Hugo, W.,
Hu-Lieskovan, S., Torrejon, D.Y., Abril-Rodriguez, G., Sandoval, S.,
Barthly, L., et al. (2016). Mutations Associated with Acquired Resistance
to PD-1 Blockade in Melanoma. N. Engl. J. Med. 375, 819–829. https://
doi.org/10.1056/nejmoa1604958.
6. Lee, J.H., Shklovskaya, E., Lim, S.Y., Carlino, M.S., Menzies, A.M., Stew-
art, A., Pedersen, B., Irvine, M., Alavi, S., Yang, J.Y.H., et al. (2020). Tran-
scriptional downregulation of MHC class I and melanoma de- differentia-
tion in resistance to PD-1 inhibition. Nat. Commun. 11, 1897. https://doi.
org/10.1038/s41467-020-15726-7.
7. Burr, M.L. , Sparbier, C.E., Chan, Y.-C., Williamson, J.C., Woods, K., Bea-
vis, P.A., Lam, E.Y.N., Henderson, M.A., Bell, C.C., Stolzenburg, S., et al.
(2017). CMTM6 maintains the expression of PD-L1 and regulates anti-
tumour immunity. Nature 549, 101–105. https://doi.org/10.1038/
nature23643.
8. Mezzadra, R., Sun, C., Jae, L.T., Gomez-Eerland, R., de Vries, E., Wu, W.,
Logtenberg, M.E.W., Slagter, M., Rozeman, E.A., Hofland, I., et al. (2017).
Identification of CMTM6 and CMTM4 as PD-L1 protein regulators. Nature
549, 106–110. https://doi.org/10.1038/nature23669.
9. Jiang, P., Gu, S., Pan, D., Fu, J., Sahu, A., Hu, X., Li, Z., Traugh, N., Bu, X.,
Li, B., et al. (2018). Signatures of T cell dysfunction and exclusion predict
cancer immunotherapy response. Nat. Med. 24, 1550–1558. https://doi.
org/10.1038/s41591-018-0136-1.
10. Ferrari de Andrade, L., Tay, R.E., Pan, D., Luoma, A.M., Ito, Y., Badrinath,
S., Tsoucas, D., Franz, B., May, K.F., Harvey, C.J., et al. (2018). Antibody-
mediated inhibition of MICA and MICB shedding promotes NK cell–driven
tumor immunity. Science 359, 1537–1542. https://doi.org/10.1126/sci-
ence.aao0505.
11. Tauriello , D.V.F., Sancho, E., and Batlle, E. (2022). Overcoming TGFb-me-
diated immune evasion in cancer. Nat. Rev. Cancer 22, 25–44. https://doi.
org/10.1038/s41568-021-00413-6.
12. Grakoui, A., Bromley, S.K., Sumen, C., Davis, M.M., Shaw, A.S., Allen,
P.M., and Dustin, M.L. (1999). The Immunological Synapse: A Molecular
Machine Controlling T Cell Activation. Science 285, 221–227. https://doi.
org/10.1126/science.285.5425.221.
13. Huppa, J.B., and Davis, M.M. (2003). T-cell-antigen recognition and the
immunological synapse. Nat. Rev. Immunol. 3, 973–983. https://doi.org/
10.1038/nri1245.
14. Xie, J., Tato, C.M., and Davis, M.M. (2013). How the immune system talks
to itself: the varied role of synapses. Immunol. Rev. 251, 65–79. https://doi.
org/10.1111/imr.12017.
15. Sligh, J.E., Ballantyne, C.M., Rich, S.S., Hawkins, H.K., Smith, C.W., Brad-
ley, A., and Beaudet, A.L. (1993). Inflammatory and immune responses are
impaired in mice deficient in intercellular adhesion molecule 1. Proc. Natl.
Acad. Sci. USA 90, 8529–8533. https://doi.org/10.1073/pnas.90.18.8529.
16. Ma, V.P.-Y., Hu, Y., Kellner, A.V., Brockman, J.M., Velusamy, A., Blan-
chard, A.T., Evavold, B.D., Alon, R., and Salaita, K. (2022). The magnitude
of LFA-1/ICAM-1 forces fine-tune TCR-triggered T cell activation. Sci.
Adv. 8, eabg4485. https://doi.org/10.1126/sciadv.abg4485.
17. Che n, S., Zhang, S., Chen, S., and Ma, F. (2023). The prognostic value and
immunological role of CD44 in pan-cancer study. Sci. Rep. 13, 7011.
https://doi.org/10.1038/s41598-023-34154-3.
18. Munz, M., Baeuerle, P.A., and Gires, O. (2009). The Emerging Role of
EpCAM in Cancer and Stem Cell Signaling. Cancer Res. 69, 5627–5629.
https://doi.org/10.1158/0008-5472.CAN-09-0654.
19. Gires, O., Pan, M., Schinke, H., Canis, M., and Baeuerle, P.A. (2020).
Expression and function of epithelial cell adhesion molecule EpCAM:
where are we after 40 years? Cancer Metastasis Rev. 39, 969–987.
https://doi.org/10.1007/s10555-020-09898-3.
20. Hwang, S., Kwon, A.-Y., Jeong, J.-Y., Kim, S., Kang, H., Park, J., Kim, J.-
H., Han, O.J., Lim, S.M., and An, H.J. (2020). Immune gene signatures for
predicting durable clinical benefit of anti-PD-1 immunotherapy in patients
with non-small cell lung cancer. Sci. Rep. 10, 643. https://doi.org/10.1038/
s41598-019-57218-9.
21. Kim, S.T., Cristescu, R., Bass, A.J., Kim, K.-M., Odegaard, J.I., Kim, K.,
Liu, X.Q., Sher, X., Jung, H., Lee, M., et al. (2018). Comprehensive molec-
ular characterization of clinical responses to PD-1 inhibition in metastatic
gastric cancer. Nat. Med. 24, 1449–1458. https://doi.org/10.1038/
s41591-018-0101-z.
22. Miao , D., Margolis, C.A., Gao, W., Voss, M.H., Li, W., Martini, D.J., Norton,
C., Bosse
´, D., Wankowicz, S.M., Cullen, D., et al. (2018). Genomic corre-
lates of response to immune checkpoint therapies in clear cell renal cell
carcinoma. Science 359, 801–806. https://doi.org/10.1126/science.
aan5951.
23. Guillerey, C., Huntington, N.D., and Smyth, M.J. (2016). Targeting natural
killer cells in cancer immunotherapy. Nat. Immunol. 17, 1025–1036.
https://doi.org/10.1038/ni.3518.
24. Bostick, M., Kim, J.K., Este
`ve, P.-O., Clark, A., Pradhan, S., and Jacob-
sen, S.E. (2007). UHRF1 Plays a Role in Maintaining DNA Methylation in
Mammalian Cells. Science 317, 1760–1764. https://doi.org/10.1126/sci-
ence.1147939.
25. Fang, J., Cheng, J., Wang, J., Zhang, Q., Liu, M., Gong, R., Wang, P.,
Zhang, X., Feng, Y., Lan, W., et al. (2016). Hemi-methylated DNA opens
a closed conformation of UHRF1 to facilitate its histone recognition. Nat.
Commun. 7, 11197. https://doi.org/10.1038/ncomms11197.
26. Liu, X., Gao, Q., Li, P., Zhao, Q., Zhang, J., Li, J., Koseki, H., and Wong, J.
(2013). UHRF1 targets DNMT1 for DNA methylation through cooperative
binding of hemi-methylated DNA and methylated H3K9. Nat. Commun.
4, 1563. https://doi.org/10.1038/ncomms2562.
27. Harri son, J.S., Cornett, E.M., Goldfarb, D., DaRosa, P.A., Li, Z.M., Yan, F.,
Dickson, B.M., Guo, A.H., Cantu, D.V., Kaustov, L., et al. (2016). Hemi-
methylated DNA regulates DNA methylation inheritance through allosteric
activation of H3 ubiquitylation by UHRF1. Elife 5, e17101. https://doi.org/
10.7554/eLife.17101.
28. Kong, X., Chen, J., Xie, W., Brown, S.M., Cai, Y., Wu, K., Fan, D., Nie, Y.,
Yegnasubramanian, S., Tiedemann, R.L., et al. (2019). Defining UHRF1
Domains that Support Maintenance of Human Colon Cancer DNA
Cell Reports Medicine 6, 101975, March 18, 2025 13
Article
ll
OPEN ACCESS
Methylation and Oncogenic Properties. Cancer Cell 35, 633–648.e7.
https://doi.org/10.1016/j.ccell.2019.03.003.
29. Pappalardi, M.B., Keenan, K., Cockerill, M., Kellner, W.A., Stowell, A.,
Sherk, C., Wong, K., Pathuri, S., Briand, J., Steidel, M., et al. (2021). Dis-
covery of a first-in-class reversible DNMT1-selective inhibitor with
improved tolerability and efficacy in acute myeloid leukemia. Nat. Cancer
2, 1002–1017. https://doi.org/10.1038/s43018-021-00249-x.
30. Jun, C.-D., Shimaoka, M., Carman, C.V., Takagi, J., and Springer, T.A.
(2001). Dimerization and the effectiveness of ICAM-1 in mediating LFA-
1-dependent adhesion. Proc. Natl. Acad. Sci. USA 98, 6830–6835.
https://doi.org/10.1073/pnas.121186998.
31. Saunders, K.O. (2019). Conceptual Approaches to Modulating Antibody
Effector Functions and Circulation Half-Life. Front. Immunol. 10, 1296.
https://doi.org/10.3389/fimmu.2019.01296.
32. Liu, L., Chen, J., Bae, J., Li, H., Sun, Z., Moore, C., Hsu, E., Han, C., Qiao,
J., and Fu, Y.-X. (2021). Rejuvenation of tumour-specific T cells through
bispecific antibodies targeting PD-L1 on dendritic cells. Nat. Biomed.
Eng. 5, 1261–1273. https://doi.org/10.1038/s41551-021-00800-2.
33. Jhunjhunwala, S., Hammer, C., and Delamarre, L. (2021). Antigen presen-
tation in cancer: insights into tumour immunogenicity and immune
evasion. Nat. Rev. Cancer 21, 298–312. https://doi.org/10.1038/s41568-
021-00339-z.
34. Kantari-Mimoun, C., Barrin, S., Vimeux, L., Haghiri, S., Gervais, C., Joa-
quina, S., Mittelstaet, J., Mockel-Tenbrinck, N., Kinkhabwala, A., Damotte,
D., et al. (2021). CAR T-cell Entry into Tumor Islets Is a Two-Step Process
Dependent on IFNgand ICAM-1. Cancer Immunol. Res. 9, 1425–1438.
https://doi.org/10.1158/2326-6066.CIR-20-0837.
35. Larson, R.C., Kann, M.C., Bailey, S.R., Haradhvala, N.J., Llopis, P.M.,
Bouffard, A.A., Scarfo
´, I., Leick, M.B., Grauwet, K., Berger, T.R., et al.
(2022). CAR T cell killing requires the IFNgR pathway in solid but not liquid
tumours. Nature 604, 563–570. https://doi.org/10.1038/s41586-022-
04585-5.
36. Spranger, S., and Gajewski, T.F. (2018). Mechanisms of Tumor Cell–
Intrinsic Immune Evasion. Annu. Rev. Cancer Biol. 2, 213–228. https://
doi.org/10.1146/annurev-cancerbio-030617-050606.
37. Wellenstein, M.D., and De Visser, K.E. (2018). Cancer-Cell-Intrinsic Mech-
anisms Shaping the Tumor Immune Landscape. Immunity 48, 399–416.
https://doi.org/10.1016/j.immuni.2018.03.004.
38. Cao, J., and Yan, Q. (2020). Cancer Epigenetics , Tumor Immunity, and
Immunotherapy. Trends Cancer 6, 580–592. https://doi.org/10.1016/j.tre-
can.2020.02.003.
39. Nishiyama, A., and Nakanishi, M. (2021). Navigating the DNA methylation
landscape of cancer. Trends Genet. 37, 1012–1027. https://doi.org/10.
1016/j.tig.2021.05.002.
40. Yoo, C.B., and Jones, P.A. (2006). Epigenetic therapy of cancer: past, pre-
sent and future. Nat. Rev. Drug Discov. 5, 37–50. https://doi.org/10.1038/
nrd1930.
41. Bates, S.E. (2020). Epigenetic Therapies for Cancer. N. Engl. J. Med. 383,
650–663. https://doi.org/10.1056/NEJMra1805035.
42. Staunton, D.E., Dustin, M.L., Erickson, H.P., and Springer, T.A. (1990). The
arrangement of the immunoglobulin-like domains of ICAM-1 and the bind-
ing sites for LFA-1 and rhinovirus. Cell 61, 243–254. https://doi.org/10.
1016/0092-8674(90)90805-O.
43. Shimaoka, M., Xiao, T., Liu, J.-H., Yang, Y., Dong, Y., Jun, C.-D., McCor-
mack, A., Zhang, R., Joachimiak, A., Takagi, J., et al. (2003). Structures of
the aL I Domain and Its Complex with ICAM-1 Reveal a Shape-Shift ing
Pathway for Integrin Regulation. Cell 112, 99–111. https://doi.org/10.
1016/S0092-8674(02)01257-6.
44. Schardt, C., Heymanns, J., Schardt, C., Rotsch, M., and Havemann, K.
(1993). Differential expression of the intercellular adhesion molecule-1
(ICAM-1) in lung cancer cell lines of various histological types. Eur. J. Can-
cer 29, 2250–2255. https://doi.org/10.1016/0959-8049(93)90217-4.
45. Konen, J.M., Wu, H., and Gibbons, D.L. (2024). Immune checkpoint
blockade resistance in lung cancer: emerging mechanisms and therapeu-
tic opportunities. Trends Pharmacol. Sci. 45, 520–536. https://doi.org/10.
1016/j.tips.2024.04.006.
46. Guo, P., Huang, J., Wang, L., Jia, D., Yang, J., Dillon, D.A., Zurakowski, D.,
Mao, H., Moses, M.A., and Auguste, D.T. (2014). ICAM-1 as a molecular
target for triple negative breast cancer. Proc. Natl. Acad. Sci. USA 111,
14710–14715. https://doi.org/10.1073/pnas.1408556111.
47. Figenschau, S.L., Knutsen, E., Urbarova, I., Fenton, C., Elston, B.,
Perander, M., Mortensen, E.S., and Fenton, K.A. (2018). ICAM1 expres-
sion is induced by proinflammatory cytokines and associated with TLS for-
mation in aggressive breast cancer subtypes. Sci. Rep. 8, 11720. https://
doi.org/10.1038/s41598-018-29604-2.
48. Wei, H., Wang, Z., Kuang, Y., Wu, Z., Zhao, S., Zhang, Z., Li, H., Zheng, M.,
Zhang, N., Long, C., et al. (2020). Intercellular Adhesion Molecule-1 as
Target for CAR-T-Cell Therapy of Triple-Negative Breast Cancer. Front.
Immunol. 11, 573823. https://doi.org/10.3389/fimmu.2020.573823.
49. Dobin, A., Davis, C.A., Schlesinge r, F., Drenkow, J., Zaleski, C., Jha, S.,
Batut, P., Chaisson, M., and Gingeras, T.R. (2013). STAR: ultrafast univer-
sal RNA-seq aligner. Bioinformatics 29, 15–21. https://doi.org/10.1093/
bioinformatics/bts635.
50. Akali n,A., Kormaks son, M., Li, S., Garrett-Bakelman, F.E., Figueroa, M.E.,
Melnick, A., and Mason, C.E. (2012). methylKit: a comprehensive R pack-
age for the analysis of genome-wide DNA methylation profiles. Genome
Biol. 13, R87. https://doi.org/10.1186/gb-2012-13-10-r87.
51. Hao, Y., Stuart, T., Kowalski, M.H., Choudhary, S., Hoffman, P., Hartman,
A., Srivastava, A., Molla, G., Madad, S., Fernandez-Granda, C., and Satija,
R. (2024). Dictionary learning for integrative, multimodal and scalable sin-
gle-cell analysis. Nat. Biotechnol. 42, 293–304. https://doi.org/10.1038/
s41587-023-01767-y.
52. Li, W., Xu, H., Xiao, T., Cong, L., Love, M.I., Zhang, F., Irizarry, R.A., Liu,
J.S., Brown, M., and Liu, X.S. (2014). MAGeCK enables robust identifica-
tion of essential genes from genome-scale CRISPR/Cas9 knockout
screens. Genome Biol. 15, 554. https://doi.org/10.1186/s13059-014-
0554-4.
53. Brinkman, E.K., Chen, T., Amendola, M., and van Steensel, B. (2014). Easy
quantitative assessment of genome editing by sequence trace decompo-
sition. Nucleic Acids Res. 42, e168. https://doi.org/10.1093/nar/gku936.
54. Kumaki, Y., Oda, M., and Okano, M. (2008). QUMA: quantification tool for
methylation analysis. Nucleic Acids Res. 36, W170–W175. https://doi.org/
10.1093/nar/gkn294.
55. Abramson, J., Adler, J., Dunger, J., Evans, R., Green, T., Pritzel, A., Ron-
neberger, O., Willmore, L., Ballard, A.J., Bambrick, J., et al. (2024). Accu-
rate structure prediction of biomolecular interactions with AlphaFold 3.
Nature 630, 493–500. https://doi.org/10.1038/s41586-024-07487-w.
14 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
anti-human-CD3 Biolegend Cat# 317315, RRID: AB_1877070
anti-human-CD28 BD Cat# 555725, RRID: AB_396068
InVivoPlus anti-mouse-CD3εBio X Cell Cat# BE0015-1, RRID: AB_1107624
InVivoMab anti-mouse-CD28 Bio X Cell Cat# BE0015-1, RRID: AB_1107624
InVivoPlus anti-mouse-PD-1 (CD279) Bio X Cell Cat# BE0273, RRID: AB_2687796
Isotype control mouse IgG1 kappa Antibody BioLegend Cat# 401402, RRID: AB_2801451
Anti-ICAM1 monoclonal Antibody Thermo Fisher Scientific Cat# MA5407, RRID: AB_223596
APC anti-human-CD54 Antibody BioLegend Cat# 353112, RRID: AB_10916103
Human TruStain FcX
TM
BioLegend Cat# 422302, RRID: AB_2818986
APC anti-human-CD8a Antibody BioLegend Cat# 301014, RRID: AB_314132
APC anti-human-CD56 Antibody BioLegend Cat# 318309, RRID: AB_604098
APC anti-human-IFN-gAntibody BioLegend Cat# 502511, RRID: AB_315236
FITC anti-human-CD107a Antibody BioLegend Cat# 328605, RRID: AB_1186058
eFluor
TM
450, CD44 Monoclonal Antibody
(IM7) Antibody
Thermo Fisher Scientific Cat# 48-0441-82, RRID: AB_1272246
APC anti-human-CD326 (Ep-CAM) Antibody BioLegend Cat# 369810, RRID: AB_2650907
APC anti-human-EGFR Antibody BioLegend Cat# 352905, RRID: AB_11148943
APC anti-human IgG Fc Antibody BioLegend Cat# 410712, RRID: AB_2565790
anti-mouse-CD16/32 Antibody BioLegend Cat# 101320, RRID: AB_1574975
APC anti-mouse-CD54 Antibody BioLegend Cat# 116119, RRID: AB_10613645
APC anti-mouse-H2Kd Antibody BioLegend Cat# 116619, RRID: AB_10640118
APC anti-mouse-H2Kb Antibody BioLegend Cat# 114614, RRID: AB_2750194
APC anti-mouse-CD11a Antibody BioLegend Cat# 153109, RRID: AB_2716218
Brilliant Violet 421 anti-mouse-CD45 Antibody BioLegend Cat# 103133, RRID: AB_10899570
FITC anti-mouse-CD3 Antibody BioLegend Cat# 100204, RRID: AB_312661
PE anti-mouse-CD8a Antibody BioLegend Cat# 100708, RRID: AB_312747
PerCP/Cyanine5.5 anti-mouse-CD8a Antibody BioLegend Cat# 100734, RRID: AB_2075238
Brilliant Violet 510
C
anti-mouse-CD4 Antibody BioLegend Cat# 100559, RRID: AB_2562608
PerCP-eFluor
TM
710-CD335 (NKp46)
Monoclonal Antibody (29A1.4)
Thermo Fisher Scientific Cat# 46-3351-80, RRID: AB_1834442
PE/Cyanine7 anti-human/mouse-Granzyme B
Recombinant Antibody
BioLegend Cat# 372214, RRID: AB_2728381
APC anti-human/mouse-Granzyme B
Recombinant Antibody
BioLegend Cat# 372203, RRID: AB_2687027
FITC anti-mouse-CD62L Antibody BioLegend Cat# 161211, RRID: AB_2941440
BV421 Hamster Anti-Mouse-CD279 BD Cat# 562584, RRID: AB_2737668
PE-Cyanine7, CD366 (TIM3) Monoclonal
Antibody (RMT3-23)
Thermo Fisher Scientific Cat# 25-5870-82, RRID: AB_2573483
APC, Ly-108 Monoclonal Antibody
(eBio13G3-19D (13G3-19D))
Thermo Fisher Scientific Cat# 17-1508-82, RRID: AB_10717668
HA-Tag (C29F4) Rabbit mAb Cell Signaling Technology Cat# 3724, RRID: AB_1549585
Alexa Fluor 488 Conjugate anti-rabbit
IgG (H + L), F(ab’)2 Fragment Antibody
Cell Signaling Technology Cat# 4412, RRID: AB_1904025
UHRF1 (D6G8E) Rabbit mAb Cell Signaling Technology Cat# 12387, RRID: AB_2715501
(Continued on next page)
Cell Reports Medicine 6, 101975, March 18, 2025 e1
Article
ll
OPEN ACCESS
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Lamin B1 (D9V6H) Rabbit mAb Cell Signaling Technology Cat# 13435, RRID: AB_2737428
Goat Anti-Rabbit IgG (H&L)-HRP Conjugated Easybio Cat# BE0101-100
Bacterial and virus strains
Stbl3 Competent Cell AlpaLife Cat# KTSM110
Chemicals, peptides, and recombinant proteins
Dulbecco’s Modified Eagle Medium (DMEM)
basic (1X)
Thermo Fisher Scientific Cat# C11995500BT
Roswell Park Memorial Institute (RPMI)
1640 basic (1X)
Thermo Fisher Scientific Cat# C11875500BT
RPMI Medium 1640 (1X) Thermo Fisher Scientific Cat# 11875-093
McCoy’s 5A Medium (Modified) (1X) Thermo Fisher Scientific Cat# 16600082
FreeStyle
TM
293 Expression Medium Thermo Fisher Scientific Cat# 12338018
Opti-MEM Thermo Fisher Scientific Cat# 11058021
Alpha MEM without ribonucleosides Thermo Fisher Scientific Cat# 12000063
Sodium Bicarbonate Sigma-Aldrich Cat# S5761
Myo-inositol Sigma-Aldrich Cat# I7508
Folic acid Sigma-Aldrich Cat# F8758
SMM 293-TII Sino Biological Cat# M293TII
PBS, pH 7.4 Thermo Fisher Scientific Cat# 10010049
Trypsin-EDTA (0.25%), phenol red Thermo Fisher Scientific Cat# 25200072
Penicillin-Streptomycin Thermo Fisher Scientific Cat# 10378016
L-Glutamine (200mM) Thermo Fisher Scientific Cat# 25030081
Sodium Pyruvate (100 mM) Thermo Fisher Scientific Cat# 11360070
MEM Non-Essential Amino Acids Solution
(100x)
Thermo Fisher Scientific Cat# 11140050
HEPES (1M) Thermo Fisher Scientific Cat# 15630130
2-Mercaptoethano Thermo Fisher Scientific Cat# 21985023
Fetal Bovine Serum (FBS) Thermo Fisher Scientific Cat# A5669701
Horse serum Thermo Fisher Scientific Cat# 26050-070
Human Fibronectin BD Cat# 354008
Recombinant Human IL-2 (carrier-free) R&D system Cat# 10453-IL
Recombinant Mouse IL-2 (carrier-free) Biolegend Cat# 575406
Recombinant Mouse IFN-g(carrier-free) Biolegend Cat# 575306
Blasticidin InvivoGen Cat# ant-bl-1
Puromycin InvivoGen Cat# ant-pr-1
hygromycin B InvivoGen Cat# ant-hg-1
SIINFEKL/OVA Peptide (257–264) GenScript Cat# RP10611
Dimethyl Sulfoxide (DMSO) Sigma Aldrich Cat# D2650
GSK3685032 MCE Cat# HY-139664
Collagenase type IV Sigma Aldrich Cat# C5138
DNAse type IV Sigma Aldrich Cat# D5025
Hyaluronidase type V Sigma Aldrich Cat# H6254
Cell Lysis Buffer (10X) Cell Signaling Technology Cat# 9803S
ACK Lysis Buffer LEAGENE Cat# CS0001
DAPI Cell Signaling Technology Cat# 4083
Fixable Viability Dye eFluor
TM
506 Thermo Fisher Scientific Cat# 65-0866-18
Brefeldin A Solution (1,000X) Biolegend Cat# 420601
EDTA-free Protease Inhibitor Cocktail Sigma Aldrich Cat# 11873580001
(Continued on next page)
e2 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
SurePAGE, Bis-Tris, 10x8, 4–12%, 15 wells GenScript Cat# M00654
Tris-MOPS-SDS Running Buffer Powder GenScript Cat# M00138
PageRuler Prestained Protein Ladder Thermo Fisher Scientific Cat# 26616
Tris-buffered saline with 0.1% Tween 20
detergent (TBST)
Beyotime Cat# ST673
Skim Milk powder BD Cat# 232100
SuperSignal
TM
West Pico PLUS Chemiluminescent
Substrate
Thermo Fisher Scientific Cat# 34580
TRIzol Thermo Fisher Scientific Cat# 15596026
Critical commercial assays
EasySep
TM
Human CD8
+
T cell Enrichment
Isolation Kit
STEMCELL Cat# 19053
MojoSort Mouse CD8 T cell Isolation Kit Biolegend Cat# 480007
NucleoSpin Blood XL, Maxi kit for DNA from
blood
MACHEREY-NAGEL Cat# 740954.20
NucleoSpin Tissue, Mini kit for DNA from
cells and tissue
MACHEREY-NAGEL Cat# 740952.50
NucleoSpin Gel and PCR Clean-up MACHEREY-NAGEL Cat# 740609.250
NucleoSpin Plasmid Transfection-grade MACHEREY-NAGEL Cat# 740490.250
CellTiter-Glo Luminescent Cell Viability
Assay
Promega Cat# G7570
Live & Dead Zombie NIR Fixable
Viability Kit
Biolegend Cat# 423106
eBioscience Foxp3/Transcription Factor
Staining Buffer
Invitrogen Cat# 2344986
Nuclear and Cytoplasmic Protein
Extraction Kit
Beyotime Cat# P0028
Pierce BCA Protein Assay Kit Thermo Scientific Cat# 23225
T4 DNA Ligase kit Thermo Scientific Cat# EL0012
EZ DNA Methylation-Gold
TM
Kit ZYMO research Cat# D5005
TA/Blunt-Zero Cloning Kit Vazyme Cat# C601-01
PrimerSTAR
TM
Max TAKARA Cat# R045A
EpiTaq
TM
HS TAKARA Cat# R110Q
Deposited data
Raw and analyzed data This paper GEO: GSE268920
Human NSCLC ICB RNA-seq dataset 1 Hwang et al.
20
GEO: GSE136961
Human Gastric ICB RNA-seq dataset 2 Kim et al.
21
PRJEB25780
Human ccRCC ICB RNA-seq dataset 3 Miao et al.
22
NIHMS978739
Experimental models: Cell lines
PBMC from healthy doners SailyBio Technologies,
Beijing, China
N/A
Homo sapiens: A498 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_1056
Homo sapiens: SW480 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_0546
Homo sapiens: A549 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_0023
(Continued on next page)
Cell Reports Medicine 6, 101975, March 18, 2025 e3
Article
ll
OPEN ACCESS
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Homo sapiens: SKBR3 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_0033
Homo sapiens: MDA-BM-231 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_0062
Homo sapiens: NK-92MI Shanghai Biofeng Biotech. RRID: CVCL_3755
Mus musculus: B16F10 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_0159
Mus musculus: MC38 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_B288
Mus musculus: LLC National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_4358
Mus musculus: 4T1 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_0125
Mus musculus: EMT6 National Collection of
Authenticated Cell Cultures,
Shanghai, China
RRID: CVCL_1923
Homo sapiens: FreeStyle
TM
293-F Liu et al.
32
N/A
Mus musculus: B16F10 EGFR Liu et al.
32
N/A
Mus musculus: MC38 EGFR Liu et al.
32
N/A
Experimental models: Organisms/strains
Mouse: C57BL/6J Beijing Vital River Laboratory RRID: IMSR_JAX:000664
Mouse: BALB/c Beijing Vital River Laboratory RRID: IMSR_JAX:000651
Mouse: NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ
(NSG)
The Jackson Laboratory RRID: IMSR_JAX:005557
Mouse: C57BL/6-Tg(TcraTcrb)1100Mjb/J
(OT-I)
The Jackson Laboratory RRID: IMSR_JAX:003831
Mouse: B6.Cg-Thy1a/Cy
Tg(TcraTcrb)8Rest/J (Pmel-1)
The Jackson Laboratory RRID: IMSR_JAX:005023
Mouse: Rosa26-LSL-Cas9 knockin The Jackson Laboratory RRID: IMSR_JAX:024857
Oligonucleotides
sgRNA sequences to generate KO
cell lines, see Table S3
This paper N/A
Primers for TIDE analysis, see Table S4 This paper N/A
ICAM1 BSP_Fwd TTAAGTTTAGTTTGGTT
GGGAAAT
This paper http://www.urogene.org/methprimer/
ICAM1 BSP_Rev AACTCTAAATAACAAA
AAAACTCAAC
This paper http://www.urogene.org/methprimer/
Recombinant DNA
lentiCas9-Blast Addgene RRID: Addgene_52962
lentiCRISPRv2-puro Addgene RRID: Addgene_98290
lentiCRISPRv2-hygro Addgene RRID: Addgene_98291
lentiCRISPRv2-eGFP This paper N/A
lentiCRISPRv2-tdTomato This paper N/A
pMYs-GFP-U6-BbsI This paper N/A
pHAGE-EF1aL-eGFP Addgene RRID: Addgene_126686
pHAGE-miniCMV-eGFP This paper N/A
(Continued on next page)
e4 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Cell lines
A549, MDA-BM-231, B16F10, MC38 and LLC cells were maintained in DMEM medium supplemented with 10% fetal bovine
serum,100 mg/mL penicillin and 100U/mL streptomycin. A498, SW480, and 4T1 cells were maintained in RPMI 1640 medium sup-
plemented with 10% fetal bovine serum and penicillin-streptomycin. SKBR3 cells were maintained in McCoy’s 5A media supple-
mented with 10% fetal bovine serum, 100 mg/mL penicillin and 100U/mL streptomycin. NK-92MI cells were maintained in Alpha
MEM without ribonucleosides supplemented with 1.5 g/L sodium bicarbonate, 0.2 mM Myo-inositol, 0.1 mM 2-mercaptoethanol,
0.02 mM folic acid and 12.5% fetal bovine serum and horse serum. All the cells were incubated at 37C and 5% CO
2
.
Primary cultures
Human peripheral blood mononuclear cells (PBMCs) were obtained from healthy donors (SailyBio Technologies) under the Shanghai
Liquan Hospital institutional review board–approved protocol. All human participants provided written informed consent prior to their
participation in the study. On day 1, CD8
+
T cells were isolated using the EasySep Human CD8
+
T cell Enrichment Isolation Kit and
stimulated with precoated anti-human-CD3 (5 mg/mL), anti-human-CD28 (1 mg/mL), and human fibronectin (5 mg/mL). On day 2, lenti-
virus expressing the NY-ESO-1 (HA+) TCR construct was added to the 24-well plates, followed by centrifugation at 1100gfor 2.5 h at
32C. After infection, cells were incubated at 37C with 5% CO
2
. On day 5, NY-ESO-1 expressing T cells were sorted by FACS using
anti-HA antibodies and Alexa Fluor 488 Conjugate anti-rabbit IgG (H + L), F(ab’)2 Fragment. T cells were then expanded in human
T cell media (RPMI 1640 media with 10% heat-inactivated fetal bovine serum, 1mM sodium pyruvate, 2mM L-glutamine, 13
NEAA, 100 mg/mL penicillin and 100U/mL streptomycin) with recombinant human IL-2 (200IU/mL).
For OT-I or Pmel-1 CD8
+
T cells, CD8
+
T cells were isolated from mouse spleens using the MojoSort Mouse CD8
+
T cell Isolation kit
and activated using anti-mouse-CD3 (5 mg/mL) and anti-mouse-CD28 (1 mg/mL). For CD11a KO OT-1 cells, follow the same protocol
for infection as human CD8
+
T cells. On day 5, CD11a KO T cells were sorted by FACS using anti-mouse-CD54. T cells were cultured
and expanded in murine T cell media (RPMI 1640 media with 10% heat-inactivated fetal bovine serum, 20mM HEPES, 1mM sodium
pyruvate, 0.05mM 2-mercaptoethanol, 2mM L-glutamine, 100 mg/mL penicillin and 100U/mL streptomycin.) with recombinant
mouse IL-2 (20 ng/mL).
Mice
Six to eight-week-old C57BL/6J, BALB/c mice were purchased from Beijing Vital River Laboratory. NOD.Cg-Prkdcscid Il2rgtm1Wjl/
SzJ (NSG), C57BL/6-Tg (TcraTcrb)1100Mjb/J (OT-I), B6.Cg-Thy1a/Cy Tg(TcraTcrb)8Rest/J (Pmel-1) and Rosa26-LSL-Cas9 knockin
mice were purchased from The Jackson Laboratory and bred at the Laboratory Animal ResourcesCenter of Tsinghua University. The
mice were maintained in pathogen-free conditions with a 12/12-h light/dark cycle, 22C–26C, 30–70% relative humidity with sterile
pellet food and water ad libitum. All mice experiments were conducted according to Institutional Animal Care and Use Committee
(IACUC)–approved protocols of Tsinghua University, Beijing, China.
Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
psPAX2 Addgene RRID: Addgene_12260
pMD2.G Addgene RRID: Addgene_12259
Human sgRNA library Brunello in
lentiGuide-Puro
Addgene RRID: Addgene_73178
Software and algorithms
GraphPad Prism 9 GraphPad RRID: SCR_002798
FlowJo 10 BD RRID: SCR_008520
STAR v2.6.1day Dobin et al.
49
RRID: SCR_004463
DESeq2 v1.34.0 Bioconductor RRID: SCR_000154
methylKit Akalin et al.
50
RRID: SCR_005177
Seurat v5 Hao et al.
51
RRID: SCR_016341
MaGeCK Li et al.
52
https://sourceforge.net/p/mageck/
wiki/Home/
TIDE Brinkman et al.
53
https://tide.nki.nl/
QUMA Kumaki et al.
54
http://quma.cdb.riken.jp/top/
quma_main_j.html
AlphaFold 3 Abramson et al.
55
https://alphafold.com/
Cell Reports Medicine 6, 101975, March 18, 2025 e5
Article
ll
OPEN ACCESS
METHOD DETAILS
Generation of cell lines
Generation of NY-ESO-1 expressing cell lines
A498, SW480, and SKBR3 cells, all of which express the endogenous HLA-A2 allele, were infected with lentivirus expressing NY-
ESO-1 antigen linked with eGFP. In the case of constructing A549-NY-ESO-1 cells, the lentiviral construct also expresses the
HLA-A*02:01 heavy chain, which is necessary for presenting the NY-ESO-1 antigen to the NY-ESO-1 specific TCR used in this study.
Generation of Icam-1 overexpressing and vector cell lines
B16F10 and 4T1 Icam1 overexpressing (Icam1
OE
) cells were generated by lentiviral transduction of murine Icam1-T2A-GFP driven by
a miniCMV promoter. GFP+ Icam1 expressing cells were stained anti-mouse-CD54 then FACS sorted and used for subsequent ex-
periments. Vector control cells were generated by lentiviral transduction of GFP vector driven by a miniCMV promoter and GFP+ cells
were FACS sorted for use.
Generation of KO cell lines
The sgRNA sequences and corresponding backbone were used to generate KO cell lines were listed in Table S3. Cells were infected
with lentiviruses expressing sgRNA. Two days after infection, puromycin (2 mg/mL) or hygromycin B (50 mg/mL) was added to the
culture for selection of KO cell lines. For ICAM1 KO and B2M KO tumor cells were FACS sorted with following antibodies: anti-hu-
man-CD54 and anti-mouse-H2K
d
.
FACS-based genome wide CRISPR screen
A549-Cas9 cells were generated by transfection with lentivirus encoding Cas9-Blast and selected with blasticidin (12 mg/mL). A549-
Cas9 cells were then transfected with the Brunello lentivirus library at an infection rate of 10%. After 48 h of transfection, transduced
cells were selected using puromycin (2 mg/mL) for 2 days. Twelve days after viral transfection, cells were stained with APC anti-hu-
man-CD54 antibody and sorted into ICAM-1
high
and ICAM-1
low
subsets. Genomic DNA of the sorted cells was extracted using the
NucleoSpin Blood L kit following the manufacturer’s protocol. Amplification of the sgRNA cassettes by PCR was performed accord-
ing to the broad GPP protocol https://portals.broadinstitute.org/gpp/public/resources/protocols. MaGeCK RRA module was used to
process and analyze the CRISPR screen data.
In vitro tumor with NY-ESO-1 specific CTLs or NK-92MI cells co-culture
In vitro competition assay
The control cells were mixed with tdTomato
+
control or ICAM-1 KO cells at an approximate 1:1 ratio. The cell mixtures were seeded at
80% confluence in a 12-well plate. Six hours after seeding, NY-ESO-1 specific CTLs or NK-92MI cells were added at E:T ratio of 1:3 or
1:1, respectively. After 24 h of co-culture, the remaining tumor cells were collected and the ratio of mixed tumor cells was analyzed by
FACS (CytoFLEX S, Beckman) and FlowJo. The following antibodies were used for gating tumor cells: (1) Human tumor-CTL co-cul-
ture assay: Live & Dead Zombie NIR Fixable Viability Kit, anti-human-CD8a; (2) Human tumor-NK-92MI co-culture assay: Live & Dead
Zombie NIR Fixable Viability Kit, anti-human-CD56; (3) Murine tumor-CTL co-culture assay: DAPI, anti-mouse-CD8a, anti-
mouse-CD54.
CTL and NK-92MI-mediated killing assay with anti-ICAM-1 blockade
Tumor cells were seeded at 80% confluence in a 12-well plate. Isotype control mouse IgG1 antibodies (5 mg/mL) or Anti-ICAM1
monoclonal antibodies (5 mg/mL) were added to the wells. After 30 min of incubation with the antibodies, human NY-ESO-1 specific
CTLs or NK-92MI cells were added at E:T ratio of 1:3 or 1:1, respectively. After 24 h of co-culture, the remaining tumor cells were
collected, and viable cells were analyzed by flow cytometry.
Human CTLs and NK-92MI cell cytotoxicity assay
Tumor cells were seeded at 80% confluence in a 12-well plate. Six hours after seeding, NY-ESO-1 specific CTLs were added at E:T
ratio of 1:1 in the presence of 5 mg/mL Brefeldin A. NK-92MI cells were added at E:T ratio of 1:1. After 6 h of co-culture, NY-ESO-1
specific CTLs or NK-92MI cells were harvested then analyzed by FACS and FlowJo. The following antibodies were used: (1) NY-ESO-
1 specific CTLs cytotoxicity assay: Live & Dead Zombie NIR Fixable Viability Kit, anti-human-CD8a, anti-human-IFN-g; (2) NK-92MI
cells cytotoxicity assay: Live & Dead Zombie NIR Fixable Viability Kit, anti-human-CD56, anti-human-CD107a. The cells were stained
with Live & Dead Zombie NIR Fixable Viability Kit and surface markers in PBS for 15 min at room temperature. For intracellular cyto-
kine IFN-gstaining, cells were processed using the eBioscience Foxp3/Transcription Factor Staining Buffer followed by manufac-
turer’s instructions.
Cell proliferation assay
For cell proliferation and viability experiments, number of 100 A498 or 200 SW480 indicated tumor cells were plated in 96-well plates
and alive cells were quantified using CellTiter-Glo Luminescent Cell Viability Assay Kit every 2 days. The k-value is calculated using
exponential growth equation in GraphPad Prism 9 to represent the cell growth rate.
For cell competition proliferation assay, ICAM-1 KO and control tumor cells were mixture at approximately 1:1 ratio. Mixture tumor
cells were cultured and ratio of alive cells were analyzed using FACS every 2 days.
e6 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
Western blot
Total cell lysates or nuclear extractions were extracted by Nuclear and Cytoplasmic Protein Extraction Kit according to the manu-
facturer’s protocol. Protein concentrations were assayed by BCA Protein assay kit and 20mg total protein or nuclear protein was
loaded per lane onto 4–12% gradient, SurePAGE, Bis-Tris gels. Gels were transferred to Immobilon PVDF membranes. Membranes
were blocked in TBST containing 5% skim milk for 2 h at room temperature and incubated primary antibodies overnight in 4C and
followed by IgG (H&L)-HRP conjugated secondary antibodies in room temperature for 1 h. Enhanced chemiluminescence substrates
were used to visualize the specific bands on the membrane. Chemiluminescence was captured using Amersham Imager 600 (GE).
Tracking of Indels by DEcomposition (TIDE) analysis for gene editing efficiency
Genomic DNA of the target gene knockout tumor cells was extracted using the NucleoSpin Tissue, Mini kit for DNA from cells and
tissue following the manufacturer’s protocol. PCR amplification of a stretch of DNA 700bp enclosing the designed editing site
(sgRNA) using PrimerSTAR Max by following thermal cycling protocol, 98C for 2min and 30 cycles each of 98C for 10 s, 60C
for 5 s, and 72C for 7s. Pair of PCR products from control and target gene knockout tumor cells were sequenced and upload the
sequencing files on the TIDE https://tide.nki.nl/ to get the gene editing efficiency. The primers used were listed in Table S4.
Whole genome bisulfite sequencing (WGBS) and focused bisulfite sequencing of ICAM1 promoter region
The genomic DNAs of A549 control and UHRF1 KO tumor cells were isolated using the NucleoSpin Blood Mini kit for DNA from cells
and tissue. The EZ DNA Methylation-Gold Kit were used for DNA bisulfite conversion of genomic DNA (400ng) according to the man-
ufacturers’ instructions. The DNA bisulfite converted DNA were then used for WGBS and focused bisulfite sequencing. For WGBS,
clean reads were mapped to human reference genome by Bismark and only uniquely mapped reads were retained. CpG methylation
levels were detected with the R package ‘‘methylKit’’. Differential methylated sites were defined by a cutoff for 20 percentage of ab-
solute change of methylation. The promoter and CpG region of indicated genes were obtained from UCSC genome table browser.
For BSP, the PCR amplification of ICAM1 promoter region was using EpiTaq HS by following thermal cycling protocol, 98C for 2min
and 40 cycles each of 98C for 10 s, 55C for 30 s, and 72C for 45s. PCR products were isolated by Mini kit for Gel and PCR Clean-up
and subcloned into the TA/Blunt-Zero Cloning Kit then individual clones were sequenced. The methylation status of the region was
determined and analyzed with QUMA http://quma.cdb.riken.jp/top/quma_main_j.html.
In vivo animal studies xenograft mouse models
For tumor challenge, 1x10
6
4T1, 1x10
6
B16F10 and 1x10
6
MC38 cells were resuspended in PBS and injected subcutaneously (s.c.)
into the flanks of mice. 0.7x10
6
B16F10 EGFR cells were prepared and injected into NSG mice following the same protocol. For fusion
protein treatment, mice bearing tumors were intratumorally/intravenously (i.t./i.v.) injected with 5 mg/kg Cetuximab or Cet3ICAM1-
D1. For anti-PD1 treatment, mice bearing tumors were intraperitoneally (i.p.) injected with 5 mg/kg of anti-PD1 antibodies every three
days starting from day 7 post tumor inoculation. The length and width were measured every 2–3 days when the tumors became
palpable and tumor volume was calculated using the following formula: (length 3width
2
)/2. The endpoint was recorded when the
tumor volume reached 2000mm
3
or mice died. Randomization was performed on age and sex-matched mice when possible.
When measuring the tumor size, investigators were blinded for sample allocations when feasible.
Flow cytometry and analysis of tumor infiltrating lymphocytes
Tumors were dissociated in gentleMACS dissociator with collagenase type IV (1 mg/mL), DNAse type IV (20 units/mL) and hyaluron-
idase type V (0.1 mg/mL) for 30 min at 37C. Cells were passed through a 70-mm filter and a small fraction was used for FACS. Cells
were stained with fluorophore-conjugated antibodies in PBS containing 1% fetal bovine serum. The following antibodies were used:
anti-mouse-CD45, anti-mouse-CD3, anti-mouse-CD8a, anti-mouse-CD4, anti-mouse-NKp46, anti-human/mouse-Granzyme B,
anti-mouse-CD62L, anti-mouse-Ly108, anti-mouse-TIM3. Cells were first stained with Live & Dead Zombie NIR Fixable Viability
Kit and anti-mouse-CD16/32 in PBS for 15 min at room temperature to block the IgG Fc receptor. The cells were stained with surface
markers, fixed, and permeabilized for intracellular staining with eBioscience Foxp3/Transcription Factor Staining Buffer followed by
manufacturer’s instructions. Beckman Coulter CytoFLEX S was used for data collection and FlowJo was used for data analysis.
Bulk RNA-seq
A549 wild-type cells were cultured in replicates, stained with anti-human-CD54 antibody, and sorted ICAM1
high
and ICAM1
low
sub-
sets by FACS AriaIII (BD). The cells were washed with PBS and lysed using TRIzol. The reads were aligned to the GRCh38 using
STAR. Feature count was used to map aligned reads to genes and generate a gene count matrix. Statistical analysis of the differen-
tially expressed genes was performed using the DESeq2 R package.
Single cell RNA-seq of immune cells isolated from Cetuximab and Cet3ICAM1-D1 treated tumors
Single-cell suspensions were obtained from mice bearing MC38-EGFR treated with Cetuximab and Cet3ICAM1-D1 on day 14. Cells
were stained with Live & DEAD Zombie NIR Fixable Viability Kit and anti-mouse-CD45. Equally numbers of alive CD45
+
cells were
sorted by FACS AriaIII (BD) from each sample of Cetuximab and Cet3ICAM1-D1 treated tumors then pooled together, respectively.
The cells were washed with PBS for library construction of scRNA-seq.
Cell Reports Medicine 6, 101975, March 18, 2025 e7
Article
ll
OPEN ACCESS
Single cell RNA-seq analysis
SingleCellExperiment (v1.16.0) was used for quality control purposes. Cells with low log-transformed library size, low log-trans-
formed number of expressed genes, or high mitochondrial proportions that were more than 3 MADs (median absolute deviation)
from the median were identified as low-quality cells and discarded. Seurat (v5) was used for integration, normalization, dimensionality
reduction, clustering, UMAP visualization, and marker gene detection, based on which manual annotation was performed for each
cluster. Other downstream analyses were performed using custom R (v4.1.3) scripts.
Analysis of ICB cohorts
Human ICB RNA-seq datasets
20–22
(see Table S5) are obtained from public databases. Samples were stratified by response status,
and tumor ICAM-1 expression levels were compared between the responders and non-responders.
Experiments related to fusion protein
Design and production of fusion protein
Cet3ICAM1-D1 is composed of Fab fragment of Cetuximab and natural D1 domain of ICAM-1 (murine or human) fused with ‘‘LALA-
PG’’ mutations Fc fragment. AlphaFold 3 was used to predict the potential binding between human ICAM1’s natural D1 domain of
Cet3hICAM1-D1 and human CD11a. These fusion proteins were generated by transient co-transfection of indicated plasmids into
FreeStyle 293-F cells. The supernatant containing the fusion protein was purified using Protein A affinity chromatography according
to the manufacturer’s protocol. The heterogeneity was confirmed by SDS-PAGE.
Binding affinity of antibody to EGFR
All tumor cells were first stained with anti-mouse-CD16/32or Human TruStain FcX to block the IgG Fc receptor. Then, MC38 and
B16F10 EGFR tumor cells tumor cells were incubated with serial dilutions of Cetuximab and Cet3ICAM1-D1 in PBS for 30 min at
room temperature. Then, cells were stained followed by a fluorophore-conjugated anti-human IgG secondary antibody. Using
FACS to measure the mean fluorescence intensity.
In vitro tumor-T cells co-culture and cytotoxicity assay
MC38 and B16F10 EGFR tumor cells pulsed with 100 ng/mL SIINFEKL peptide were co-incubated with OT-1 cells in the presence of
serial dilutions of Cetuximab or Cet3ICAM1-D1 at E:T ratio of 1:1. ICAM-1 knockout and NY-ESO-1 positive A498 and SW480 tumor
cells with endogenous EGFR were co-incubated with NY-ESO-1 specific T cells in the presence of serial dilutions of Cetuximab or
Cet3hICAM1-D1 at E:T ratio of 1:1. Then the effector cells mediated killing assay and effector cells cytotoxicity assay were pro-
cessed as described above.
QUANTIFICATION AND STATISTICAL ANALYSIS
Statistical analyses were performed by using GraphPad Prism 9 software and presented as means ±SEM. Statistical comparisons
between multiple groups were determined by one-way ANOVA or two-way ANOVA with multiple comparisons. Statistical compar-
isons between two groups were determined by unpaired two-tailed Student’s T test. p< 0.05 was considered statistically significant
(*p< 0.05, **p< 0.01, ***p< 0.001 and ****p< 0.0001; ns, not significant). Statistical details for each experiment can be found in the
figures and the legends. For ICB analysis (Figure 1H), pvalues are derived from linear regression model which uses tumor ICAM-1
expression level to predict responding status (responder vs. non-responder).
e8 Cell Reports Medicine 6, 101975, March 18, 2025
Article
ll
OPEN ACCESS
