The integrated stress response promotes immune evasion th…

The transcription factor ATF4 and its effector lipocalin 2 (LCN2) have a key role in immune evasion and tumour progression, and targeting the ATF4–LCN2 axis might provide a way to treat several types of solid tumour by increasing anti-cancer immunity.

Cancer cells activate the integrated stress response (ISR) to adapt to stress and resist therapy1. ISR signals converge on activating transcription factor 4 (ATF4), which controls cell-intrinsic transcriptional programs that are involved in metabolic adaptation, survival and growth2,3. However, whether the ISR–ATF4 axis influences anti-tumour immune responses remains mostly unknown. Here we show that loss of ATF4 decreases tumour progression considerably in immunocompetent mice, but not in immunocompromised ones, by enhancing T cell-dependent anti-cancer immune responses.

An unbiased genetic screen of ATF4-regulated genes identifies lipocalin 2 (LCN2) as the principal ATF4-dependent effector that impairs anti-tumour immunity by favouring infiltration with immunosuppressive interstitial macrophages. Furthermore, we find that LCN2 promotes T cell exclusion and immune evasion in preclinical mouse models, and correlates with decreased T cell infiltration in patients with lung and pancreatic adenocarcinomas.

Anti-LCN2 antibodies promote robust anti-tumour T cell responses in mouse models of aggressive solid tumours. Our study shows that the ATF4–LCN2 axis has a cell-extrinsic role in suppressing anti-cancer immunity, and could pave the way for an immunotherapy approach that targets LCN2. Immune checkpoint inhibitors (ICIs) have transformed anti-cancer therapy and become a first-line treatment for several cancers4,5.

However, many solid tumours do not respond to standard-of-care ICIs6,7,8. Gaining a better understanding of the mechanisms that underlie immune evasion of ICI-refractory cancers is therefore essential to improve anti-cancer treatments8,9. Solid tumours progressively reshape their immediate tumour microenvironment (TME), producing features that affect the fitness and characteristics of cells within the TME, such as hypoxia, nutrient scarcity and waste accumulation1,10,11,12,13.

In response, cancer cells activate the integrated stress response (ISR), an evolutionarily conserved cellular defence system that is triggered by diverse stressors, including misfolded proteins, amino acid starvation and mitochondrial dysfunction1,3,11,14,15,16,17,18,19. The central driver of ISR activation is the phosphorylation of eukaryotic translation initiation factor 2α (eIF2α), which reduces global translation while selectively promoting the translation of activating transcription factor 4 (ATF4), a central regulator of transcriptional programs that support adaptation and survival1.

The ISR and its master transcriptional effector, ATF4, are emerging as key players that respond to several intrinsic stressors during tumorigenesis and contribute to therapy resistance1,2,3,15,17,18,20. Although these TME features also profoundly influence anti-tumour immunity, the cancer-cell-extrinsic contribution of the ISR and the ATF4 axis to this process remains mainly unknown10,11,16,21. So far, most studies investigating the role of ATF4 in cancer cells have been performed in vitro or in immunodeficient animals, and consequently its potential role in adaptation and resistance to host immune responses during tumorigenesis has not been defined.

To evaluate the importance of ATF4 in tumorigenesis, we used a Kras and p53 (also known as Trp53) mutant (KrasLSL-G12D/+;p53fl/fl; hereafter, KP) genetically engineered mouse model (GEMM) of lung adenocarcinoma (LUAD), with conditional CRISPR–Cas9-based loss of Atf4 (sgAtf4) or control (sgTom) (Fig. 1a,b). Loss of Atf4 decreased the tumour burden, as compared with Atf4 wild-type (WT) tumours (Fig.

1c). To dissect the importance of ATF4 in tumour progression, we generated isogenic Atf4 knockout (Atf4KO) and WT (Atf4WT) lung cancer KP primary cell lines (Extended Data Fig. 1a) and transplanted them subcutaneously (s.c.) into syngeneic immunocompetent C57BL/6J mice (Fig. 1d and Extended Data Fig. 1b,c). In agreement with the GEMM experiment (Fig. 1c) and previous studies2,3,15, ATF4-deficient tumours grew significantly more slowly than control tumours did, despite showing no differences in markers of cellular proliferation or apoptosis (Extended Data Fig.

1d) and no differences in proliferation in vitro (Extended Data Fig. 1a), excluding the possibility that the intrinsic properties of the ATF4-deficient cells caused their growth delay in immunocompetent mice. Notably, Atf4KO tumours grew as well as Atf4WT tumours did in immunodeficient NOD scid IL2Rγnull (NSG) mice (Fig. 1e and Extended Data Fig. 1c). Consistent with these findings, ATF4 was required for tumour growth in two additional syngeneic models—Lewis lung carcinoma (LLC) (Fig.

1f and Extended Data Fig. 1e) and B16F10 melanoma (Fig. 1g and Extended Data Fig. 1f). By contrast, loss of ATF4 in B16F10 tumours had no effect on tumour growth in immunodeficient NU/J (nude) mice (Fig. 1h and Extended Data Fig. 1g). Furthermore, pharmacological inhibition of ATF4 with integrated stress response inhibitor (ISRIB), which allosterically antagonizes phosphorylated eIF2α22 (Fig. 1i), resulted in a significant reduction in tumour growth and prolonged survival in immunocompetent mice (Fig.

1j,k and Extended Data Fig. 1h,i), but not in immunodeficient NSG mice (Fig. 1l and Extended Data Fig. 1j,k). Collectively, these results show that ATF4 is required for tumour progression by curtailing anti-tumour immunity in several cancers. a, Scheme of the GEMM KrasLSL-G12D/+;p53fl/fl;Rosa26LSL-Cas-P2A-GFP model, in which we infected mice with bifunctional lentiviruses (pUSEC), which express Cre recombinase and sgRNAs for conditional CRISPR–Cas9-based loss of Atf4 (sgAtf4) or control (sgTom).

b, Representative IHC of lung tumours stained for ATF4. Arrowheads indicate ATF4-positive cells. Scale bars, 50 μm. c, Tumour burden 14 weeks after infection with lentiviruses carrying guides against control sgTom (n = 7) and sgAtf4 (n = 8). Two-tailed t-test. d,e, Tumour growth in C57BL/6J (Atf4WT n = 4, Atf4KO n = 6) (d) and NSG (Atf4WT n = 10, Atf4KO n = 11) (e) mice injected s.c.

with KP primary LUAD cells (Atf4WT or Atf4KO). Data were analysed by a repeated-measures two-way analysis of variance (ANOVA) with Fisher’s least significant difference (LSD) test. NS, not significant. f, Tumour growth of LLC s.c. injected cell lines with Atf4WT (n = 13) or Atf4KO (n = 15) status in C57BL/6J mice. Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test.

g,h, Tumour growth of B16F10 s.c. transplanted tumours with Atf4WT (n = 17) or Atf4KO (n = 8) status in C57BL/6J mice (g) and nude mice (h) (Atf4WT n = 8, Atf4KO n = 8). Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. i, Mechanism of ISR pharmacological inhibition by ISRIB, which disrupts the dimerization of phospho (p)-eIF2α and the subsequent accumulation of ATF4.

j, Luminescence of orthotopically transplanted KP tumours in the lungs of C57BL/6J mice treated daily with 2.5 mg per kg ISRIB (n = 7) or vehicle (n = 5). Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. k, Survival of C57BL/6J mice with orthotopic KP tumours (vehicle n = 11, ISRIB n = 12). Data were analysed by a log-rank (Mantel–Cox) test.

l, Luminescence of orthotopically transplanted KP tumours in the lungs of NSG mice treated with ISRIB (n = 4) or vehicle (n = 3). Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. Data are mean ± s.e.m. The illustrations in a,i were created in BioRender. Bossowski, J. (2025) https://BioRender.com/wryu8nk. To identify ATF4-driven immunosuppressive factors, we performed a pooled negative-selection CRISPR–Cas9 screen in a syngeneic C57BL/6J mouse using a custom library of 477 putative ATF4 targets23 (Fig.

2a and Supplementary Table 1). We identified candidate genes whose single guide RNAs (sgRNAs) were preferentially depleted in the immunocompetent hosts—including Atf4, consistent with our previous data (Fig. 2b, Extended Data Fig. 2a and Supplementary Table 2). Moreover, we identified lipocalin 2 (Lcn2) as one of the most significantly depleted genes (Fig. 2b and Extended Data Fig. 2a). a, Schematic of the CRISPR library experimental set-up to screen ATF4-regulated genes in vivo and in vitro.

b, Differential score of gene dropout frequency between KP tumours implanted into immunodeficient NSG versus immunocompetent C57BL/6J mice and isolated 12 days after s.c. transplantation. c,d, Growth of Lcn2WT and Lcn2KO KP tumours transplanted into C57BL/6J mice (n = 8 in Lcn2WT, n = 10 in Lcn2KO group) (c) and NSG immunodeficient mice (n = 24 in Lcn2WT, n = 13 in Lcn2KO group) (d).

Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. e, Normalized luminescence signal from orthotopic KP tumours expressing Lcn2WT (n = 6), Lcn2KO (n = 5), or Lcn2KO KP cells with ectopic overexpression of mouse Lcn2 (mLcn2, n = 6) in C57BL/6J mice. Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test. f, Relative tumour growth in C57BL/6J mice injected s.c.

with KP cells with Atf4 KO or WT status, with ectopic overexpression of control (Ctr) vector or Lcn2 (Lcn2OE) (Atf4WTCtr n = 7, Atf4WTLcn2OE n = 7, Atf4KOCtr n = 7, Atf4KOLcn2OE n = 6). Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. g, Growth, in C57BL/6J mice, of s.c. implanted Lcn2KO KP tumours ectopically expressing empty vector (n = 9), Lcn2WT (n = 9), Lcn2 lacking the secretion signal peptide sequence (sec_del, n = 14) and two independent triple mutant Lcn2 constructs expressing Lcn2 with substituted amino acids responsible for binding to iron (Lcn2Y127A/K147A/K156A, n = 6; Lcn2R103G/K147A/K156A, n = 7).

Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test. h, Scheme of the KPC GEMM KrasLSL-G12D/+;p53fl/fl;Rosa26LSL-Cas-P2A-GFP model, in which we infected mice with bifunctional lentiviruses (pUSEC), which express Cre recombinase and double-guides for conditional CRISPR–Cas9-based loss of Lcn2 (sgLcn2) or control (sgNeo). i, Representative magnetic resonance imaging (MRI) images 14 and 21 weeks after infection of KPC mice with sgLcn2 or sgNeo lentivirus.

j, Histological tumour burden (LUAD) 24 weeks after tumour initiation (sgNeo n = 7, sgLcn2 n = 6). Data were analysed by a two-tailed t-test. k, Scheme of the orthotopic PDAC transplant mouse model. l, Tumour weight of PDAC (KPC7) tumours five weeks after orthotopic transplantation with Lcn2WT (n = 6) or Lcn2KO (n = 7) cells. Data were analysed by a two-tailed t-test. Data are mean ± s.e.m.

The illustrations in a,h,k were created in BioRender. Bossowski, J. (2025) https://BioRender.com/wryu8nk. LCN2 is a secreted glycoprotein that has been implicated in inflammatory responses24,25. We performed a series of in vivo experiments using Lcn2 knockout (Lcn2KO) and wild-type (Lcn2WT) KP cells. Similar to our findings for ATF4 (Fig. 1d–h), loss of LCN2 resulted in a reduction in tumour growth in C57BL/6J mice (Fig.

2c and Extended Data Fig. 2c) but not in NSG mice (Fig. 2d and Extended Data Fig. 2d). Furthermore, loss of LCN2 suppressed orthotopic KP tumour growth in the lungs of immunocompetent mice (Fig. 2e), whereas reintroducing LCN2 restored tumour growth to levels comparable with those of Lcn2WT tumours (Fig. 2e and Extended Data Fig. 2e,f). We observed no differences in proliferation between the groups (Extended Data Fig.

2b,g), similar to what was seen for ATF4 loss (Extended Data Fig. 1d). Next, we overexpressed LCN2 in ATF4-deficient cells and found that exogenous LCN2 rescued the growth of Atf4KO tumours to levels comparable to those of Atf4WT tumours (Fig. 2f and Extended Data Fig. 2h), demonstrating that LCN2 has a crucial role in ATF4-driven immune suppression. Secreted LCN2 acts as a siderophore-bound iron sequestrator26.

On the basis of previous reports27,28, we tested two Lcn2 mutants that abrogate iron–catechol binding. Tumours expressing either mutant grew at the same rate as that of tumours expressing Lcn2WT (Fig. 2g and Extended Data Fig. 2i–m). Moreover, ectopic expression of a secretion-deficient Lcn2 mutant did not rescue the growth of Lcn2KO tumours (Fig. 2g and Extended Data Fig. 2l–o). These findings show that LCN2 secretion, but not its binding to iron, is required for its immunosuppressive effects.

Using the KP GEMM, we observed that LCN2 loss (sgLcn2) suppressed tumorigenesis, compared with control tumours (sgNeo) (Fig. 2h–j and Extended Data Fig. 2p). The level of LCN2 protein in bronchoalveolar lavage fluid (BALF) was more than twofold lower in mice with Lcn2-deficient tumours than it was in control mice (Extended Data Fig. 2q), indicating that tumour-derived LCN2 makes a considerable contribution to the total lung TME LCN2 levels in the autochthonous LUAD model.

Next, we evaluated whether LCN2 is required for the growth of other solid tumour lineages using syngeneic orthotopic models of melanoma (B16F10) and pancreatic ductal adenocarcinoma (PDAC) (KPC7). We observed that Lcn2KO tumours grew significantly more slowly than Lcn2WT tumours did (Fig. 2k,l and Extended Data Fig. 2r–u), revealing that LCN2 regulates tumorigenesis in several solid tumour types.

To determine whether LCN2 contributes to tumour initiation or to the growth of established tumours, as suggested by the ISRIB experiments (Fig. 1j–l), we used a doxycycline-inducible short hairpin RNA (shRNA) system to silence LCN2 (shLcn2) in an orthotopic KP LUAD model. Silencing LCN2 in established tumours (14 days after implantation) significantly reduced tumour growth, prolonged survival and decreased tumour burden (Extended Data Fig.

2v–ab), indicating that LCN2 represents a therapeutic vulnerability in advanced disease. We next assessed the ISR–ATF4-dependent mechanisms that regulate the expression of Lcn2 in cancer cells. Lcn2 transcriptional regulation has been reported to be driven mainly by pro-inflammatory nuclear factor-κB (NF-κB) signalling29. However, our work and previous studies suggest that ATF4 activates Lcn2 expression directly30,31.

Both glutamine deprivation and tunicamycin—classical inducers of the ISR—increased the levels of ATF4 protein and the expression of Lcn2; this induction was abolished in ATF4-deficient cells (Extended Data Fig. 3a–c). Similarly, treatment with ISRIB, which was found to curb tumour progression (Fig. 1j,k), abolished the induction of Lcn2 by glutamine deprivation and tunicamycin, without affecting the interleukin-1β (IL-1β)- and NF-κB-dependent induction of Lcn2 (Extended Data Fig.

3d). Moreover, the combination of IL-1β stimulation and acute glutamine deprivation resulted in higher levels of Lcn2 than did either stimulus alone—as opposed to Asns, a canonical ATF4 target, which was not affected by IL-1β stimulation (Extended Data Fig. 3e,f). We observed similar ISR-mediated regulation of Lcn2 in the human lung carcinoma cell lines A549 and H1299 (Extended Data Fig. 3g,h).

We next tested whether Lcn2 expression is induced by clinically relevant metabolic inhibitors, including the glutaminase inhibitor CB-839 and phenformin, a mitochondrial respiration inhibitor and analogue of metformin32,33. Both inhibitors induced the expression of Lcn2 in vitro, which was abolished in Atf4KO KP cells (Extended Data Fig. 3i). To corroborate our in vitro findings, we tested the effect of phenformin on Lcn2 levels in tumour cells in vivo.

Mice bearing established lung tumours were administered oral phenformin daily for three days, after which the expression of Lcn2 was analysed in sorted tumour cells (Extended Data Fig. 3j). Lcn2 expression was increased in cancer cells from phenformin-treated mice, and correlated with a higher frequency of ATF4-positive cells (Extended Data Fig. 3k–l), consistent with ATF4-mediated induction of Lcn2.

Conversely, treatment with ISRIB in vivo reduced the proportion of ATF4-positive tumour cells (Extended Data Fig. 3l). An in silico analysis of the vicinity of the Lcn2 locus identified an ATF4-binding motif (Supplementary Table 3). ATF4 chromatin immunoprecipitation (ChIP)–quantitative PCR (qPCR) showed that ATF4 occupancy was enriched at the promoter regions of Lcn2, as well as the canonical targets Chac1 and Asns (Extended Data Fig.

3m). We analysed datasets from The Cancer Genome Atlas (TCGA) for lung, pancreatic and melanoma tumours to assess co-expression between LCN2 and an ATF4 transcriptional signature, and stratified patients into low, medium and high LCN2 groups for each cancer type (Extended Data Fig. 3n–p). Across all datasets, LCN2 expression showed strong concordance with the ATF4 signature, reinforcing the link between ATF4 activity and LCN2 expression.

In addition, in non-small-cell lung cancer (NSCLC), we observed that the NF-κB signature was positively associated with LCN2 levels, and, to a lesser extent, with the ATF4 signature (Extended Data Fig. 3q,r). Altogether, these data suggest that in addition to the well-known regulation of LCN2 expression by NF-κB, ATF4 also controls LCN2 expression in tumours. The dependence of tumour growth on LCN2 in immunocompetent hosts indicates that tumour-derived LCN2 establishes an immunosuppressive TME.

Consistent with this model, Lcn2KO KP LUAD tumours showed increased levels of CD8+ and CD4+ tumour-infiltrating leukocytes (TILs) and a concomitant reduction in CD4+FOXP3+ regulatory T (Treg) cells, compared with Lcn2WT tumours (Fig. 3a,b and Extended Data Fig. 4a). Treatment with ISRIB phenocopied LCN2 loss by increasing the levels of CD8+ TILs (Extended Data Fig. 4b,c), and Lcn2KO PDAC tumours also showed enhanced immune infiltration (Extended Data Fig.

4d). a,b, Immunofluorescence images stained for CD4 and CD8 cells of Lcn2KO (n = 6) and Lcn2WT (n = 6) KP tumours from orthotopically transplanted mice (a) and their quantification (the number of lung tumours quantified per mouse ranged from 13 to 57) (b). Data were analysed using a two-tailed t-test. c, Growth of subcutaneous Atf4WT and Atf4KO tumours in C57BL/6J mice with CD4+ and CD8+ depletion (Atf4WT isotype n = 10, anti-CD4 n = 6, anti-CD8 n = 10; Atf4KO isotype n = 8, anti-CD4 n = 7, anti-CD8 n = 10).

Data were analysed by a repeated-measures two-way ANOVA with Tukey test. Violin plots show the distribution of values; central lines indicate the median and dashed lines denote the first and third quartiles. d, Orthotopic growth of Lcn2KO tumours treated with isotype (n = 9), CD4-depleting (n = 8) or CD8-depleting (n = 8) antibodies. Data were analysed by Kruskal–Wallis tests followed by a Dunn’s multiple-comparisons test of the last measurement point.

e, ExCITE-seq immune analysis of myeloid cell clusters in orthotopically transplanted KP LUAD with doxycycline-inducible shLcn2 (n = 2) and shCtr (n = 3) KP tumours eight days after adding doxycycline in the food. UMAP, uniform manifold approximation and projection. f, LUMICKS AFS measurements of mLCN2-coated beads binding to BMDMs. BMDMs were transduced with siRNA against LCN2 receptor (siSlc22a17) or a non-targeting siRNA (Ctr).

Beads coated with an antibody against F4/80 were used as a positive binding control. Data were analysed using one-way ANOVA with Tukey’s multiple-comparisons test. g, Gene set enrichment analysis (GSEA) of hallmark pathways in subpopulations of myeloid cells between shCtr and shLcn2 tumours from e. FDR, false discovery rate; NES, normalized enrichment score. h,i, Representative results of relative gene expression of Cxcl9 (h) and Il6 (i) after 6 h of stimulation of BMDMs with LPS (0.5 ng ml−1) with preconditioning for 24 h with 500 ng ml−1 recombinant lipocalin 2 (mLCN2).

One-way ANOVA test with Tukey’s multiple comparison test. j, Quantification of CD8 T cell recruitment in 3D collagen matrix embedded with BMDM–KP cancer-cell co-cultures. Treatment with anti-CXCL9 antibody curtailed T cell infiltration as compared with isotype control (Welch’s t-test, biological replicates: n  =  3). Data are mean ± s.e.m (c,d,j) or mean ± s.d. (f,h,i). We reasoned that ATF4-deficient tumours might be effectively controlled by adaptive immunity20.

Indeed, the impaired growth of Atf4KO lung tumours was entirely reversed by the depletion of either CD4+ or CD8+ T cells (Fig. 3c). We observed similar T cell-dependent growth suppression in orthotopically implanted Lcn2KO KP lung tumours (Fig. 3d). These findings indicate that both CD4+ and CD8+ T cells are necessary to mount an effective anti-tumour response, which is absent in ATF4–LCN2-proficient tumours.

To obtain a more granular and unbiased view of the immune TME in the context of LCN2 suppression, we performed expanded cellular indexing of transcriptomes and epitopes by sequencing (ExCITE-seq) of live extravascular CD45+ cells from whole-lung digests of mice bearing control or LCN2-deficient tumours. ExCITE-seq enables simultaneous evaluation of the transcriptome and surface epitopes of single cells in a high-throughput manner, and allows analysis of the antigen receptor repertoire in T lymphocytes.

Using a doxycycline-inducible Lcn2 shRNA system (Extended Data Fig. 2v–ab), we silenced Lcn2 for eight days and analysed CD45+ immune populations from collected lung tumours (Extended Data Fig. 4e). In agreement with immunohistochemistry (IHC) data (Fig. 3a,b), the shLcn2 tumours showed an increase in the proportion of CD4+ (9.1% versus 5.8%) and CD8+ (10.6% versus 4.4%) T cells, as compared with the control (shCtr) condition (Extended Data Fig.

4e,f). We observed a relative decrease in the populations of macrophages (from 34.6% in shCtr to 25.6% in shLcn2) and neutrophils (from 26% to 13.7%) (Extended Data Fig. 4f). To elucidate the effects of LCN2 on T cells, natural killer (NK) cells and innate lymphoid cells, we divided them into nine subclusters. In agreement with our immunofluorescence data (Extended Data Fig. 4a), we found lower proportions of Treg cells in LCN2-deficient tumours (15.9% shCtr versus 5.1% in shLcn2; Extended Data Fig.

4g) and robust clonal expansion of CD8-effector and -exhausted cells in the shLcn2 condition, with little evidence of clonal expansion in other T cell subclusters or in the WT KP tumours (Extended Data Fig. 4h). To further validate the effects of tumour-derived LCN2 on T cells, we performed a flow cytometry analysis of Lcn2KO tumours and compared their immune landscape with those of Lcn2WT tumours and Lcn2-repleted tumours (Lcn2KO with ectopic Lcn2 re-expression).

We observed a marked decrease in the intratumoural abundance of Treg cells in Lcn2KO tumours, relative to both Lcn2WT and Lcn2-repleted tumours (Extended Data Fig. 4i). To determine whether these changes in Treg levels were related to the Lcn2 genetic status of the tumours, and not driven by overall tumour burden, we performed immune profiling after short-term suppression of Lcn2 (7 and 14 days). Again, we observed that acute suppression of Lcn2 led to decreased levels of CD4+FOXP3+ Treg cells in tumours (Extended Data Fig.

4j). Moreover, we found a lower abundance of Treg cells in tumours from ISRIB-treated mice (Extended Data Fig. 4k). Collectively, these data indicate that tumour-derived LCN2 is associated with an increased accumulation of intratumoural CD4+FOXP3+ Treg cells7,21. Tumour-infiltrating innate immune cells, particularly macrophages and neutrophils, actively contribute to immune evasion and tumour progression within the TME34,35.

Previous studies have shown that myeloid cells have a key role in regulating lung cancer immunity and shaping the therapeutic outcomes of immune checkpoint blockade36,37,38,39,40. Notably, LCN2 can modulate the inflammatory phenotype of these myeloid cells, particularly macrophages24,41,42,43. ExCITE-seq data revealed decreased levels of macrophages within Lcn2-silenced tumours (Extended Data Fig.

4e,f). We identified five macrophage subclusters: alveolar macrophages (AMs); monocytes; and three interstitial macrophages (IMs; Cx3cr1, Spp1 and Arg1) (Fig. 3e). We observed a higher fraction of AMs (39% shLcn2 versus 20% shCtr), with concomitant decreases in the Spp1 and Cx3cr1 IM subclusters (35.9% and 28.7% in shCtr versus 16.4% and 18.2% in shLcn2, respectively), in tumours with Lcn2 suppression (Fig.

3e). Acute knockdown of Lcn2, analysed by flow cytometry, also led to a reduction in IMs (SiglecFlowCD11bhigh) in tumours (Extended Data Fig. 4l–n), corroborating the ExCITE-seq results. As a secreted factor, LCN2 signals through a cell-surface receptor, with SLC22A17 mediating its endocytosis and downstream receptor-dependent functions41,43,44,45. ExCITE-seq gene expression allowed us to examine the transcriptional levels of Slc22a17 among all identified cell clusters.

Slc22a17 expression was restricted mainly to macrophages, with negligible levels in other immune cells, as corroborated by independent datasets (Extended Data Fig. 4o,p). To determine whether LCN2 influences macrophages, we analysed its direct binding to SLC22A17 and examined the downstream functional effects. Using the LUMICKS acoustic force spectroscopy (AFS) system, we detected binding between recombinant-mouse-LCN2-coated beads and bone-marrow-derived macrophages (BMDMs) (Fig.

3f). This interaction was mediated by SLC22A17, because small interfering RNA (siRNA)-mediated knockdown of Slc22a17 abolished LCN2 binding, whereas the binding of F4/80-antibody-coated beads, targeting an independent macrophage marker, was unaffected. The functional effects of macrophages on tumour progression are determined mostly by their phenotype and polarization state40,46. After suppression of Lcn2, IMs showed an increase in the activity of anti-tumorigenic pathways (interferon-γ (IFNγ) and IFNα response, allograft rejection) and a decrease in the activity of pro-tumorigenic pathways, including oxidative phosphorylation (Fig.

3g and Extended Data Fig. 4q). We examined CXCL9 and IL-6, well-recognized macrophage-associated mediators of anti-tumour immunity that have been reported40,47,48,49 to be regulated by LCN2. Whereas CXCL9, a T cell chemokine, supports a stronger immune response against cancer, IL-6 is generally associated with weakened anti-cancer immunity. An ExCITE-seq analysis of macrophage clusters indicated that knockdown of Lcn2 (shLcn2) in tumour cells upregulated Cxcl9 expression specifically in Spp1 and Cx3cr1 IMs, whereas Il6 expression was diminished in the Arg1 macrophage population (Extended Data Fig.

4r). Treating BMDMs in vitro with recombinant LCN2 (rLCN2) in the context of low-dose lipopolysaccharide (LPS) challenge (0.5 ng ml−1) led to the suppression of Cxcl9 expression and a concomitant increase in the expression of Il6, mimicking the effects of tumour-derived LCN2 in IMs that were seen in vivo (Fig. 3h,i). Furthermore, the LCN2-mediated regulation of Cxcl9 and Il6 was, to a large extent, abolished in macrophages after knockdown of Slc22a17 (Extended Data Fig.

4s,t). These results confirm that tumour-derived LCN2 can modulate the inflammatory state of macrophages at least partly through binding to SLC22A17 on macrophages. Overall, our results suggest that tumour-cell-derived LCN2 promotes an immunosuppressive transcriptional state of IMs, reminiscent of the inflammatory macrophage states that are associated with immune evasion in preclinical models and patients34,40,46,50,51.

Impaired growth of Lcn2KO tumours requires CD8+ T cells (Fig. 3d) and is accompanied by increased levels of CD8+ TILs (Fig. 3a,b). Consistently, in a three-dimensional (3D) co-culture system, Lcn2KO KP cells promoted greater CD8+ T cell infiltration than did Lcn2WT cells (Fig. 3j and Extended Data Fig. 4u). This effect was significantly reduced by CXCL9 neutralization (Fig. 3j). Collectively, these findings support a model in which tumour-derived LCN2 limits the anti-tumour effector function of T cells by fostering immunosuppressive IM accumulation in the TME, and highlight the functional importance of CXCL9 suppression in LCN2-associated immune evasion.

To examine the clinical relevance of our findings in preclinical mouse models, LCN2 protein levels were examined in tumour microarrays from 105 patients with LUAD with known clinical and histological characteristics. Samples were scored according to the intensity of LCN2 IHC staining, and patients were stratified on the basis of histological assessments of cancer differentiation patterns52. A higher tumour grade is indicative of increased clinical aggressiveness and shortened overall survival52.

Notably, the levels of LCN2 staining strongly correlated with tumour grade (Fig. 4a). Regions with a high LCN2 staining intensity often corresponded to regions with low levels of CD3+ IHC staining (Fig. 4b), and we observed a statistically significant relationship between high levels of LCN2 and the absence of TILs (Fig. 4c). a, LCN2 score and tumour grade in biospecimens from patients with LUAD (n = 105).

One-way ANOVA test with Tukey’s multiple comparison test. b, Representative LCN2 and CD3 IHC staining of the human LUAD biopsy quantified in a, with a similar pattern among n = 105 biopsies. Scale bar, 500 μm. c, Violin plot of LCN2 score in samples from a subgrouped on the basis of the presence of TILs in the sample. Data were analysed using a two-tailed t-test. Violin plots show the distribution of values; central lines indicate the median and dashed lines denote the first and third quartiles.

d, LCN2 and CD3 IHC staining of a representative biopsy from a patient with PDAC (quantified in e), with a similar pattern among n = 33 PDAC biopsies. Scale bar, 500 μm. e, Correlation between CD3 staining and LCN2 score among 33 PDAC biopsies. f, Nearest neighbour analysis of PDAC samples on the basis of immunofluorescence Opal staining and single-cell annotation of LCN2-positive (LCN2+) cells, LCN2-negative (LCN2−) cells and T cells (either CD4+ or CD8+).

The statistical analyses were computed using a Wilcoxon rank-sum test, and P values were adjusted using Benjamini–Hochberg. g,h, iTIL/sTIL ratio (g) and macrophage density (macrophage count in cancer and stroma/combined area (mm2)) (h) across LCN2 expression groups (low, moderate and high) in the TCGA pan-cancer analysis. Mann–Whitney U test (low versus high) P = 7.54 × 10−99 (g) and P = 7.15 × 10−58 (h).

i, Overall survival of patients with NSCLC in the microarray database of 2,166 samples, divided by bulk LCN2 expression levels into low- and high-expression groups. A total of 1,484 patients were assigned to the LCN2-low category, and 682 were assigned to the LCN2-high category. LCN2 arbitrary expression intensity ranged between 2 and 48,717, and the established cut-off was set to 1,844. HR, hazard ratio; numbers in brackets represent 95% confidence intervals.

j, Overall survival of 976 patients with cancer treated with immunotherapy (anti-PD1, anti-PD-L1 or anti-CTLA4) among patients registered in the Gene Expression Omnibus (GEO), European Genome-phenome Archive (EGA) and TCGA databases. A total of 375 patients were assigned to the LCN2-low category, and 601 were assigned to the LCN2-high category. LCN2 expression ranged between 2 and 1,258,740, and the established cut-off was set to 16.

Box plots show the median (centre line), the interquartile range (25th–75th percentiles; box) and whiskers extending to the minimum and maximum values. Data are mean ± s.e.m. We performed a similar investigation of tumour microarrays collected from 33 patients with PDAC, another ICI-refractory cancer type (Fig. 4d,e), and observed a similar negative correlation between LCN2 and CD3 staining levels.

Consequently, we reasoned that tumour-derived LCN2 might act in a spatially localized manner. To test the close-proximity model of LCN2’s action, we performed nearest-neighbour single-cell spatial analysis on tumour tissue from a patient with PDAC, annotating cells by LCN2 status (positive versus negative cytoplasmic staining). Spatial neighbour analysis revealed that LCN2 expression was associated with increased spatial exclusion of T cells (Fig.

4f and Extended Data Fig. 5a,b), supporting a mechanism in which LCN2 acts in a locally confined manner. Together, the results from LUAD and PDAC are consistent with our observations in preclinical models (Fig. 3a,b and Extended Data Fig. 4a–g), showing that there is a spatial exclusivity of LCN2 expression and TILs within tumour tissue. Next, we used transcriptomic data and whole-slide images (WSIs) from TCGA, including 23 epithelial solid tumours, and analysed them with Lunit SCOPE IO53 (Extended Data Fig.

5c). First, we stratified the cases according to the previously established immune phenotypes of inflamed, immune-excluded and immune-desert, and evaluated the levels of LCN2 and ATF4 expression across these groups (Extended Data Fig. 5d,e). Consistent with our laboratory findings, both LCN2 and ATF4 showed higher expression in the immune-excluded than in the inflamed category. Next, we stratified cases into three groups (low, moderate and high) according to LCN2 expression levels, and looked at several microenvironmental variables, including intratumoural-to-stromal T cell ratio (iTIL/sTIL; Fig.

4g) and macrophage density (Fig. 4h). In line with our preclinical mouse model results (Fig. 3), we observed a lower iTIL/sTIL ratio in the group with high LCN2 expression, concomitant with higher macrophage density (Fig. 4g,h). Next, we assessed the survival of patients with lung cancer on the basis of LCN2 expression in a publicly available microarray database of 2,166 samples54. High expression of LCN2 was associated with significantly shorter overall survival, with a median survival of 79 months in the LCN2-low cohort and and 52 months in the LCN2-high cohort (Fig.

4i). A similar correlation with survival, although to a lesser extent, was observed in 1,195 samples from patients with PDAC, with a median survival of 17.9 months in the LCN2-high group versus 20.2 months in the LCN2-low group (Extended Data Fig. 5f). We validated our findings in an independent NSCLC cohort (Lunit) and observed similar shortened survival in the group of patients with high LCN2 expression, as well as increased expression of LCN2 in patients characterized by the immune-excluded phenotype (Extended Data Fig.

5g,h). An analysis of overall survival in patients with cancer (n = 976) who were treated with immunotherapy revealed that the LCN2-low cohort had a better overall survival (Fig. 4j). In summary, LCN2 expression in human lung and pancreatic cancers positively correlates with tumour grade, decreased T cell infiltration, shorter overall survival and poor response to immunotherapy. Because LCN2 is a secreted protein, it should be readily targetable with antibodies.

To test the therapeutic efficacy of targeting LCN2, we developed synthetic antibodies against mouse LCN2 (mLCN2), and chose a clone that bound tightly to mLCN2 but not to human LCN2 (hLCN2) for further analyses (referred to as anti-mLCN2 hereafter) (Fig. 5a). To assess the efficacy of targeting LCN2 against immunogenic tumours, we used a platform in which male-derived primary tumour cells are transplanted into a female host55.

In this model, treatment with anti-mLCN2 significantly reduced the progression of orthotopically transplanted tumours and extended survival (Fig. 5b and Extended Data Fig. 6a,b). a, Biolayer interferometry sensorgrams of anti-mLCN2 and anti-hLCN2 in the monovalent Fab format binding to biotinylated hLCN2 and mLCN2 immobilized on streptavidin sensor tips. Derived dissociation constant (Kd) values with s.d.

from a 1:1 binding model are shown. b, Bioluminescence signal of tumour growth of C57BL/6J female mice orthotopically transplanted with KP LUAD tumours and treated with anti-mLCN2 antibody (10 mg per kg, biweekly) starting at ten days after tumour transplantation (Ctr n = 11, anti-mLCN2 n = 8). Data were analysed using a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test.

c, Final tumour weight of KP PDAC tumours from mice treated with anti-mLCN2 at 10 mg per kg, biweekly for three weeks (Ctr n = 11, anti-mLCN2 n = 7). Data were analysed using a one-tailed t-test. d, Bioluminescence signal of tumour growth from C57BL/6J mice transplanted with Lcn2KO KP LUAD tumours expressing human LCN2 (+hLCN2) by tail-vein injection and treated with anti-human LCN2 antibody (anti-hLCN2, 10 mg per kg, biweekly; Ctr n = 5, anti-hLCN2 n = 5).

Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. e,f, KP LUAD (e) and PDAC (f) IHC images of representative sections of tumours treated with anti-hLCN2 and anti-mLCN2, respectively. Quantification of LUAD is presented in Extended Data Fig. 6o,p. Scale bars, 50 μm (e) and 100 μm (f). g, Quantification of TILs in KP PDAC tumours orthotopically transplanted into C57BL/6J female mice and treated with anti-mLCN2 antibody (10 mg per kg biweekly; Ctr n = 8, anti-LCN2 n = 7).

Tumour quantification of CD4 and CD8 was assessed using three to five fields of view. Data were analysed using a two-tailed t-test. Violin plots show the distribution of values; central lines indicate the median and dashed lines denote the first and third quartiles. h, Survival of male mice bearing KP LUAD tumours and treated with anti-mLCN2 antibody (10 mg per kg, biweekly), anti-PD1 antibody (10 mg per kg, three times per week) and the combination of both (Ctr n = 7, anti-hLCN2 n = 8, anti-PD1 n = 6, combination n = 7).

Data were analysed using the log-rank (Mantel–Cox) test. Data are mean ± s.e.m. Treating male mice with sex-matched KP LUAD tumours with anti-mLCN2 resulted in a significant inhibition of tumour growth in this immunologically ‘cold’ syngeneic model that is refractory to ICIs. This was accompanied by a delayed onset of weight loss in the mice, indicative of decreased tumour burden and extended survival (Extended Data Fig.

6c–e). Furthermore, we observed no significant changes in the cell counts of the lymphocyte, neutrophil, monocyte, eosinophil or basophil populations in the peripheral blood (Extended Data Fig. 6f). Post mortem visual examination revealed no major changes in vital organs (liver, heart, spleen and kidney). We observed significant suppression of tumour growth by anti-mLCN2 in an orthotopically transplanted model of PDAC, another aggressive solid cancer (Fig.

5c and Extended Data Fig. 6g). These results indicate that anti-mLCN2 can suppress tumour growth without causing acute toxicity, and suggest an effective therapeutic strategy for lung and pancreatic cancer. To unambiguously assess the importance of tumour-derived LCN2 and accelerate the translation of anti-LCN2 antibodies, we developed an antibody that binds tightly to human LCN2 (anti-hLCN2) but only marginally to mLCN2 (Fig.

5a), and used Lcn2KO KP cells expressing hLCN2, termed hLCN2 chimeric KP cells hereafter (Extended Data Fig. 6h). Ectopically expressed hLCN2 rescued the growth of Lcn2KO cells in immunocompetent mice (Extended Data Fig. 6i,j), indicating that hLCN2 can functionally substitute for mLCN2, and confirming that cancer-cell-derived LCN2 is essential for tumour growth. The combination of hLCN2 chimeric KP cells and anti-hLCN2 enabled us to specifically target tumour-derived hLCN2 without affecting host-derived mLCN2.

Similar to anti-mLCN2 against Lcn2WT KP cells (Fig. 5b,c), anti-hLCN2 suppressed the tumour progression of orthotopically transplanted hLCN2 chimeric KP cells (Fig. 5d and Extended Data Fig. 6k). Similar efficacy was observed for s.c. implanted tumours (Extended Data Fig. 6l,m). Moreover, BMDM binding to mLCN2-coated beads was effectively disrupted using anti-mLCN2 antibody, but not anti-hLCN2 antibody (Extended Data Fig.

6n), suggesting that anti-mLCN2 functions by inhibiting the interaction between LCN2 and macrophages, and further validating the specificity of these antibodies. Tumours in the KP LUAD and PDAC mice were overwhelmingly devoid of TILs, consistent with the literature describing those tumours as immunologically cold56,57 (Fig. 5e–g). By contrast, both anti-mLCN2 treatment of PDAC mice and anti-hLCN2 treatment of humanized KP LUAD mice resulted in increased levels of CD4+ and CD8+ TILs (Fig.

5e–g and Extended Data Fig. 6o,p). We observed a similar increase in T cell infiltration with anti-mLCN2 treatment using a 3D co-culture system of Lcn2WT KP and BMDMs (Extended Data Fig. 6q). Furthermore, in agreement with the macrophage inflammatory state observed in the orthotopic mouse model (Fig. 3g and Extended Data Fig. 4q,r), anti-hLCN2 treatment resulted in lower IL-6 concentrations and increased IFNγ concentrations in BALF, compared with untreated controls (Extended Data Fig.

6r–t). These results show that cancer-cell-derived LCN2 has a key role in suppressing T cell infiltration in tumours, and that this can be inhibited by anti-LCN2 antibodies. Finally, because anti-hLCN2 treatment increased CD4+ and CD8+ T cell infiltration (Fig. 5e), we reasoned that it might render KP LUAD tumours responsive to ICIs. Treatment with anti-hLCN2 alone significantly extended survival (Fig.

5h), whereas treatment with anti-PD1 did not significantly alter mouse survival, as expected from previous studies55. Notably, the combination of anti-hLCN2 and anti-PD1 extended the median survival of mice to 26 days, as compared with 19 days in the Ctr group, 20 days in the anti-PD1 group and 24.5 days in the anti-LCN2-treated group (Fig. 5h). Altogether, our data with LUAD mouse models suggest that antibody-directed targeting of tumour-derived LCN2 is a promising anti-cancer therapeutic strategy with a reasonable safety profile.

This study shows that, in the context of a fully immunocompetent setting, loss of ATF4 in several solid tumours significantly hinders tumorigenesis, and that this immunomodulatory role of ATF4 is mediated through the induction of LCN2. Specifically, we show that LCN2 suppresses the infiltration of both CD4+ and CD8+ T cells, ultimately impairing anti-cancer immune responses. We found that LCN2 has a profound effect on the transcriptional state of macrophages, which can have a crucial role in modulating the infiltration and function of both CD4+ and CD8+ T cells34,35,36,37,39,40.

Whereas previous studies have mainly addressed immune-cell-intrinsic functions of LCN225,43,45,58,59, our work specifically dissects the role of tumour-derived LCN2, providing insight into its paracrine and non-cell-autonomous effects in shaping the tumour–immune microenvironment. Binding of LCN2 to SLC22A17 on macrophages drives polarization toward a Cxcl9-low immunosuppressive phenotype. Correspondingly, knockdown of Slc22a17 mostly abolishes the LCN2-mediated regulation of Cxcl9 and Il6 (Extended Data Fig.

4s,t). In rLCN2-naive conditions, Slc22a17 knockdown also affects the levels of Cxcl9 expression after LPS challenge, indicating a potential LCN2-independent function or serum-related cross-reactivity. Further studies are needed to determine how the LCN2–SLC22A17 interaction modulates macrophage transcriptional states and impairs adaptive immunity. It is also important to determine whether this mechanism translates to other disorders in which immunosuppressive macrophages have a role in pathogenesis and disease progression42,43,60.

LCN2 is known for its anti-bacterial role25,41,43,45,59,61, mainly through its ability to sequester bacterial siderophore–iron complexes from the environment, thereby inhibiting bacterial growth26. We found that the immunomodulatory function of LCN2 does not rely on iron sequestration (Fig. 2g). This view is consistent with previous findings that LCN2 regulates the immune function and phenotype of Treg cells62 and macrophages43 independently of iron, and with our finding that anti-mLCN2 inhibits the interaction between LCN2 and macrophages (Extended Data Fig.

6n). Elucidating the interaction between anti-LCN2 antibodies and LCN2 will be crucial to understanding LCN2–SLC22A17 signalling. Although LCN2 is a soluble protein capable of diffusing away from its source cells, our work suggests that it operates locally. LUAD and PDAC clinical samples revealed distinct local gradients of LCN2 (Fig. 4a–h and Extended Data Fig. 5a–e). We speculate that LCN2 is anchored to the extracellular matrix in the vicinity of cancer cells and/or binds to its receptors on macrophages immediately after being secreted from cancer cells.

Future structure–function studies of LCN2 will define the molecular mechanisms that underlie its immunosuppressive functions at the molecular and cellular levels. These studies could also illuminate the spatially restricted localization of LCN2 in affected tissues. We found that anti-LCN2 antibodies have robust efficacy in aggressive tumour models. Our experiments using the hLCN2 chimeric KP cells and anti-hLCN2 specifically showed that targeting cancer-cell-derived LCN2 is sufficient to achieve this efficacy and promote T cell infiltration (Fig.

5d,e and Extended Data Fig. 6k–t). Although systemic administration of anti-mLCN2 in mice did not result in detectable toxicity (Extended Data Fig. 6d,f), selective neutralization of tumour-derived LCN2 might offer a treatment approach with a greater therapeutic window. Such tumour-selective targeting could be facilitated by advanced antibody-engineering technologies63,64,65. Together, this study establishes LCN2 neutralization as a promising immune-modulating therapy against treatment-refractory solid tumours.

The KP cells used here were established previously66. Atf4 and Lcn2 knockout cell lines were generated by transient transfection of PX458 (Addgene, 48138) expressing guides targeting either Atf4 or Lcn2. The list of sequences and primers is included in Supplementary Table 4. Single GFP-positive clones were selected, and gene of interest (GOI) loss was validated by western blot or ELISA. B16F10 cells were a gift from the I. Aifantis laboratory.

Cells were cultured in RPMI with 10% fetal bovine serum (FBS) and gentamicin. mLCN2 and hLCN2 KP cells were selected and maintained in hygromycin-supplemented medium (800 μg ml−1). Commonly available cell lines have been authenticated using short tandem repeat (STR) profiling. All cells were mycoplasma-negative. The mouse KPC7 pancreatic cancer cell line, derived from a tumour in LSL-KrasG12D/+;LSL-Trp53R172H/+;Pdx1-Cre mice, was used.

This cell line is maintained on a C57BL/6 background. Doxycycline-induced knockdown of Lcn2 was achieved by cloning hairpin targeting Lcn2 into the pLKO.1-TETON-Puro vector (Addgene plasmid 21915). In brief, hairpins were designed according to the Genetic Perturbation Platform (Broad Institute), and the annealed hairpin is ligated into the EcoRI+AgeI–digested backbone with Quick Ligase (NEB) at a 3:1 insert:vector molar ratio.

Sequences and primers are listed in Supplementary Table 4. Vectors were transduced into cells through lentivirus and selected with puromycin (8 μg ml−1). Oligos were obtained from Integrated DNA Technologies. All experiments were approved by the New York University (NYU) Institutional Animal Care and Use Committee (IA16-01627). KrasLSL-G12D/+;Trp53fl/fl;Rosa26LSL-Cas9-P2A-GFP/LSL-Cas9-P2A-GFP (KPC) mice have already been described55,66,67,68,69,70,71.

Mice aged six to eight weeks of both sexes with the appropriate genotype were selected to begin tumour initiation studies with the bifunctional lentiviruses (pUSEC), which express Cre recombinase and sgRNAs for conditional CRISPR–Cas9-based loss of Atf4 (sgAtf4) or control (sgTom). The sgRNAs used are listed in Supplementary Table 4. For the experiments with the Lcn2-loss GEMM, we used a double-guide lentiviral system with a control vector carrying sgNeo-1 and sgNeo-2, and KPC mice with Lcn2 loss infected with lentivirus carrying sgLcn2-1 and sgLcn2-2 guides.

The total lung area occupied by each tumour was measured on haematoxylin and eosin (H&E)-stained slides using NIS-Elements software (Nikon). Transplant experiments were performed using nude (JAX strain 002019), NOD scid gamma (NSG; JAX strain 005557) or C57BL/6J (JAX strain 000664) mice. Cells (100,000 cells in 100 μl PBS) were injected s.c. into each flank of the mouse. B16F10 melanoma cells (2 × 105) were injected s.c.

into female mice aged four to six weeks. LLC cells (5 × 105) were injected s.c. into female mice aged four to six weeks. Tumours were measured by caliper, and tumour volume was calculated according to the formula 0.5 × length × width2. Tumours were not allowed to grow more than 2 cm in any linear measurement, and did not exceed that limit in any of the experiments performed. To generate orthotopic lung tumours, KP cells expressing luciferase–GFP were injected intravenously (100,000 cells in 100 μl PBS) into male C57BL/6J (JAX strain 000664) mice, and tumour burden was measured by bioluminescence (Perkin Elmer IVIS Spectrum in vivo imaging system, d-luciferin, Perkin Elmer 122799).

Data were analysed using Living Image software. Sample sizes are provided in the figure legends. The minimum sample size was predetermined on the basis of the stochastic variability expected in the individual model, and taking into consideration the difference in the expected effect magnitude. When possible, a higher number of mice was used to reduce waste and increase confidence. Precise numbers are provided in the figure legends.

For the PDAC orthotopic model experiments, 8- to 12-week-old female C57BL/6 mice were used. Syngeneic KPC7 cells, with or without Lcn2 knockout (Lcn2KO), were orthotopically implanted (200,000 cells per mouse) into the mice, and tumour growth was monitored weekly using ultrasound. Five weeks after implantation, the mice were euthanized, primary tumours were collected and the tissues were formalin-fixed for further IHC analysis.

In addition, KPC7 cells were orthotopically implanted (200,000 cells per mouse) into C57BL/6 mice. When the tumour size reached 100 mm3, the mice were randomized into two groups to receive either a control vehicle (Ctr) (n = 10) or anti-mLCN2 (n = 10) treatment by intraperitoneal (i.p.) injection. Pancreatic tumour burden was quantified by ultrasound imaging before and after the drug treatment.

After the treatment, the mice were euthanized, and tumour tissues were collected for IHC studies. All experiments were approved by the NYU Institutional Animal Care and Use Committee (IA16-01627). More information on library cloning is provided in a previous report23. In brief, the oligo pool from a reconstituted oligonucleotide library synthesis (OLS) library was single-amplified with barcoded primers and inserted into an Esp3I-digested pUSEPB (U6-sgRNA-EFS-Puro-P2A-TagBFP) vector72.

The screen included a total of 3,240 sgRNAs targeting 470 ATF4-controlled genes, 25 essential genes and 279 unique non-targeting Ctr sgRNAs (Supplementary Table 1). KP cells with stable expression of Cas9–blasticidin were generated by introducing the lentivector pLX_311-Cas9, which expresses blasticidin resistance from the SV40 promoter and Cas9 from the EF1a promoter, as described73. For the in vitro screen, at least 3.3 × 106 cells were maintained throughout the experiment to ensure a representation of at least 1,000×.

A total of 6 × 107 cells were transduced with the sgRNA library in 10-cm dishes at a multiplicity of infection of 0.3, then selected with 10 μg ml−1 puromycin for 30 h. After this, the medium was refreshed for 24 h for medium without puromycin, after which cells were collected, counted and used for screening in vitro and in vivo. A fraction of cells representing 1,000× coverage were pelleted and stored at −80 °C (t0 population).

Cells were passaged every 2–3 days, and 3.3 × 106 cells were plated after each passage. After 10 population doublings (t10), 3.3 × 106 cells were pelleted and stored at −80 °C. For the in vivo screen, the ATF4 CRISPR library was divided into two sub-libraries, containing 1,626 and 1,614 sgRNAs, respectively. KP cells were infected and selected in a similar manner as for the in vitro screen, and 1.6 × 106 were injected s.c.

into the flanks of C57BL/6J (n = 4) and NSG (n = 4) mice. Twelve days after s.c. transplantation, tumours were resected and genomic DNA was isolated using lysis buffer (50 mM Tris, 50 mM EDTA, 1% SDS, pH 8.0) supplemented with 30 μl of 20 mg ml−1 proteinase K (QIAGEN, 19131) that was added to tissue (up to 200 mg) or cells (up to 30 million cells) and incubated at 60 °C overnight.

This was followed by phenol:chloroform:isoamyl alcohol (25:24:1) phase separation, isopropanol precipitation and ethanol wash. Five micrograms of in vitro and 10 μg of in vivo samples’ genomic DNA were used for screen deconvolution with NEB Q5 High-Fidelity 2X Master Mix (M0492L). Integrated sgRNAs were amplified by PCR to attach sequencing adapters and barcodes, as described previously23. PCR products of around 248 bp in length were size-selected on 2% agarose and purified using the QIAquick Gel Extraction Kit (QIAGEN, 28704).

Samples were sequenced on an Illumina NextSeq 500 (75-nt single-end reads) at NYU Genome Technology Center. sgRNA dropout results were identified using the MAGeCK algorithm74. Cloning of CRISPR sgRNAs was performed as previously described into the USEC vector71. The list of sequences and primers is included in Supplementary Table 4. Lentivirus was generated by co-transfection of HEK293 cells with viral vector and packaging plasmids psPAX2 (Addgene, 12260) and pMD2.G (Addgene, 12259) using JetPrime transfection reagent (101000046).

Medium containing the virus was collected 72 h after transfection and filtered through a 0.45-μm filter. For in vivo experiments, the virus was concentrated by ultracentrifugation at 25,000 rpm for two hours at 4 °C. The virus pellet was resuspended in PBS and stored at −80 °C until use. Virus was titred using the GreenGo reporter cell line71. Sections were immunostained on a Leica BondRX automated stainer according to the manufacturer’s instructions.

In brief, tissues underwent deparaffinization online, followed by epitope retrieval for 60 min at 100 °C with Leica Biosystems ER2 solution (pH 9, AR9640) and endogenous peroxidase activity blocking with H2O2 (provided in the Leica BOND Polymer Refine Detection System, DS9800). Sections were then incubated with primary antibodies against ATF4 (Cell Signaling Technology, 11815S, RRID: AB_2616025) at 1:50 for 60 min at room temperature.

Primary antibodies were detected with anti-rabbit horseradish peroxidase (HRP)-conjugated polymer and 3,3′-diaminobenzidine (DAB) substrate that are provided in the Leica BOND Polymer Refine Detection System. After counter-staining with haematoxylin, slides were scanned at 40× on a Hamamatsu NanoZoomer (2.0-HT) and the image files were uploaded to the NYU Grossman School of Medicine’s OMERO Plus image data management system (Glencoe Software).

Chromogenic IHC for human LUAD biospecimens was performed on a Ventana Medical Systems Discovery Ultra platform using Ventana reagents and detection kits unless otherwise specified. Unconjugated polyclonal goat anti-human lipocalin 2 (LCN2, R&D Systems, AF1757, JBH0919031, RRID: AB_354974) antibody was optimized on a formalin-fixed, paraffin-embedded 14-core tissue microarray containing normal brain, liver and kidney.

All samples were sectioned at 4 μm, collected onto plus microscope slides (Thermo Fisher Scientific, 22-042-924) and stored at room temperature before use. Initial optimization testing determined antigen retrieval requirements at a fixed concentration. Subsequent optimization manipulated the concentration and/or incubation to establish the final protocol parameters. In addition, unconjugated rabbit anti-human ready-to-use CD3, clone 2GV6 (IVD CD3, Ventana Medical Systems, 790-4341, J27879, RRID: AB_2335978) was assayed according to the manufacturer’s instructions.

In brief, slides were heated at 60 °C for one hour and deparaffinized on-instrument using Discovery Wash (Ventana Medical Systems, 950-510) at 69 °C for 24 min. Antigen retrieval was performed in Cell Conditioner 1 (Ventana Medical Systems, 950-500) for 24 min for LCN2 and 64 min for CD3, both at 91 °C. Slides were treated with 3% H2O2 for 8 min to quench endogenous peroxidase. Anti-LCN2 was diluted 1:100 in Dulbecco’s PBS (Thermo Fisher Scientific, J67670.AP) and incubated for three hours at room temperature.

Primary antibody was detected with rabbit anti-goat HRP-conjugated multimer (Ventana Medical Systems, 760-4311) and incubated for 8 min, followed by ChromoMap (760-159) DAB detection. CD3 was applied neat and incubated for 32 min at 37 °C, followed by ultraView DAB detection (Ventana Medical Systems, 760-500). Slides were washed in distilled water, counterstained with haematoxylin, dehydrated and mounted with permanent medium.

Negative controls consisted of the primary antibody substituted with antibody diluent. For multiplex immunofluorescence staining with Akoya Biosciences Opal reagents, slides were incubated with the first primary antibody and secondary polymer pair, and then underwent HRP-mediated tyramide signal amplification with a specific Opal fluorophore. The primary and secondary antibodies were subsequently removed with a heat retrieval step, leaving the Opal fluorophore covalently linked to the antigen.

This sequence was repeated with subsequent primary and secondary antibody pairs and a different Opal fluorophore at each step (see Supplementary Table 5 for reagent details). Sections were counterstained with spectral DAPI (Akoya Biosciences, FP1490) and mounted with ProLong Gold Antifade (Thermo Fisher Scientific, P36935). Semi-automated image acquisition was performed either on a Leica AT2 whole-slide bright-field scanner at 40× magnification or on an Akoya Vectra Polaris (PhenoImager HT) multispectral imaging system at 20× magnification using PhenoImager HT 2.0 software in conjunction with Phenochart 2.0 and inForm 3.0 to generate unmixed whole-slide qptiff scans.

Image files were uploaded to the NYU Grossman School of Medicine’s OMERO Plus image data management system (Glencoe Software). Tumour microarray samples from biopsies from patients with LUAD were assessed and graded as previously described52. Levels of LCN2 and TILs in individual tumour biopsies were graded in a blinded manner. PDAC human biospecimens from the tissue microarray panel were annotated and graded in a blinded manner, followed by quantification of CD3- and LCN2-positive cells within the tumour area.

Only biopsies with PDAC histology were included in the analysis. Treatment with ISRIB (0.25 mg per kg) or vehicle control (1:1 dimethyl sulfoxide (DMSO), Thermo Fisher Scientific, D128-500, and polyethylene glycol 400) was administered daily through i.p. injections. For T cell-depletion experiments, anti-CD8 (2.43, BioXcell BE0061), anti-CD4 (YTS 191, BioXcell BE0119) or isotype control (rat IgG2b, LTF-2, BioXcell BE0090) was administered at 150 μg i.p.

twice a week. For treatments with anti-PD1 (29F.1A12, BioXcell), antibodies were diluted in PBS and injected i.p. at 200 μg per mouse three times a week until the end point of the experiment. Mice were sedated with ketamine and xylazine, and were then injected with 2 μg of APC anti-CD45 (2 μg per mouse diluted in 100 μl PBS; BioLegend, 30-F11) retro-orbitally 3 min before lung collection.

The lung lobes were minced on a glass slide and then digested (collagenase IV (Sigma-Aldrich, C5138) and DNAse I (Sigma-Aldrich, DN25) in RPMI with 10% FBS) for 35 min at 37 °C. Digestion was stopped with EDTA (1 mM). Digested tissue was filtered into a single-cell suspension through a 100-μm filter, followed by red blood cell lysis (BD Pharm Lyse, 555899). Cells were then washed and suspended in a staining buffer.

Cells were then stained with live–dead staining (Zombie UV fixable viability dye; BioLegend, 423107) and PeCy7 anti-CD45 (see ‘Flow cytometry’ for staining protocol). Approximately 500,000 lung immune cells from each mouse were sorted as live+IV−CD45−CD45+, and 50,000 tumour cells were sorted as live+CD45−GFP+. Sorted samples were multiplexed using cell hashing antibodies (BioLegend) and stained with antibody-derived tags.

Cells from each sample were pooled and loaded into 10X Chromium75,76. Gene expression, together with hashtag oligo (HTO) libraries, was processed using Cell Ranger (v5.0.0) in multi-mode. Cell-containing droplets were selected using the HTODemux function available in the Seurat programme. Unique molecular identifier (UMI) count matrices from each modality were imported into the same Seurat object as separate assays.

Viable cells were filtered on the basis of having more than 200 genes detected and less than 10% of total UMIs stemming from mitochondrial transcripts. HTO counts were normalized using centred log ratio transformation before hashed samples were demultiplexed using the Seurat::HTODemux function. Protein counts were normalized using the centred log ratio transformation. RNA counts were normalized using the Seurat::SCTransform function with regressions of cell-cycle score and ribosomal and mitochondrial percentages.

Multimodal integration was performed using the weighted nearest neighbour method in Seurat. In brief, a weighted nearest neighbour network was constructed on the basis of modality weights estimated for each cell using the Seurat::FindMultiModalNeighbors function with the top 40 and top 30 principal components from normalized RNA and protein counts, respectively. A shared nearest neighbour graph was then built on the basis of the first 40 principal components, followed by identification of cell clusters using the Leiden algorithm and Seurat::FindClusters function at several resolutions to identify potential rare cell types.

Cell types were annotated on the basis of canonical cell-type markers and differentially expressed genes of each cluster identified using the Seurat::FindAllMarkers function with a logistic regression model. Clusters expressing markers of the same cell type were merged into a single cluster. Cells were then projected onto a uniform manifold77 using the top 40 principal components for visualization.

Data have been deposited in the Gene Expression Omnibus (GEO) under accession number GSE308689. Mice were euthanized, and lungs were digested into a single-cell suspension as described above. Single cells were transferred to a 96-well round-bottom plate and resuspended in fluorescence-activated cell sorting (FACS) buffer (0.5% bovine serum albumin (BSA), 0.1% sodium azide and 1 mM EDTA). Live–dead staining was initially performed per protocol (Zombie UV fixable viability dye; BioLegend, 423107).

Cells were then blocked with Fc block (2.4G2, BioXcell) for 10 min on ice. Antibody cocktail for surface staining was then added for 15 min on ice (see antibody list in Supplementary Table 5), and then the samples were washed with FACS buffer. Cells needing intracellular staining for FOXP3 were fixed and permeabilized using the FOXP3 staining buffer kit (eBioscience, 00552300). Intracellular Fc blocking was applied for 10 min on ice, followed by intracellularly staining with FOXP3 antibody for one hour on ice.

Cells were then washed and resuspended in FACS buffer. Next, cells were washed, transferred to a 96-well round-bottom plate and resuspended in FACS buffer. Surface staining was done as described above. The cells were then washed and resuspended in FACS buffer. The samples were filtered with a 100-μm filter and then run on a BD LSRFortessa. Data were analysed using FlowJo v10. Two hundred microlitres of 10.57-µm far-red melamine resin beads (Microparticles, MF-FluoRed-AR1286) were coated with LCN2 or F4/80 antibody.

Unless specified, all volumes were 200 µl, incubations were done at room temperature on a platform rocker and two washes were performed with 1× PBS after each incubation. Sequentially, beads were incubated with 0.01% poly-l-lysine (Sigma-Aldrich, P4707) for 30 min, 100 mM BS(PEG)9 (Thermo Fisher Scientific, 21582) for 20 min, 0.5 mg ml−1 avidin (Thermo Fisher Scientific, 21128) for 20 min and 5 µM of biotinylated LCN2 or F4/80 antibody for 20 min.

Beads were washed three times with 1× PBS, resuspended in 200 µl of 1× PBS and kept on ice until the avidity experiment. BMDMs were disassociated from culture flasks and resuspended at a concentration of 8 × 107 cells per ml, and seeded on z-Movi (LUMICKS) microfluidic chips that were coated with poly-l-lysine (Sigma, P4707). Z-Movi chips seeded with BMDMs were placed in a 37 °C incubator for at least two hours for attachment, then 20 µl of protein or antibody on beads was flowed onto the z-Movi chip and incubated with BMDMs for 2 min.

Anti-mouse or anti-human LCN2 Fabs (1 µM) were co-injected with 20 µl of protein or antibody on beads. After incubation, an acoustic force ramp from 0 to 1,000 pN over 2 min 30 s was applied within the z-Movi chip, and protein and antibody on-bead detachment was observed using real-time fluorescence imaging on the z-Movi system. Replicates were performed on different z-Movi chips with randomized run orders for protein and antibody conditions.

Avidity experiments were processed using proprietary Oceon software. Brightfield and fluorescence pre-flow images and time-lapse movies were loaded in Oceon. One hundred to two hundred events were recognized within the field of view for each run. Oceon software drew a region of interest at a single cell size. Detachment events were recognized as loss of fluorescent signal within the region of interest.

Using pre-flow images, beads that remained stuck from previous runs were automatically removed from analysis. Beads that were bound to exposed glass and not target cells were also automatically excluded. Files produced by Oceon were imported into Microsoft Excel for organization before plotting with GraphPad Prism 10.2.3. Avidity data were plotted as (1) a function of force and (2) a bar chart at the indicated force.

Macrophages and pan-monocytes were derived from the tibiae and femurs of C57BL/6 male mice (age 6–10 weeks). Pan-monocytes were isolated from the bone-marrow cell suspension using a pan-monocyte isolation kit (Miltenyi Biotec, 130-100-629) and cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated FBS (HIFBS), 1% penicillin–streptomycin (Gibco, 15-140-122), 1% sodium pyruvate (Gibco, 11-360-070) and 0.05 mM 2-mercaptoethanol (Gibco, 21-985-023). Lcn2KO and Lcn2KO+mLcn2 cancer cells were cultured in RMPI supplemented with 10% HIFBS, 800 μg ml−1 hygromycin (Gibco, 10-687-010) and 50 μg ml−1 gentamicin (Gibco, 15-750-060).

Mouse spleens were collected to extract CD8a+ T cells using the CD8a+ T Cell Isolation Kit (Miltenyi Biotec, 130-104-075). T cells were activated by stimulation with IL-2 (BioLegend, 575402) and activation beads from the T Cell Activation/Expansion Kit (Miltenyi Biotec, 130-093-627) for three days and cultured in RMPI supplemented with 10% HIFBS, 1% penicillin–streptomycin, 1% sodium pyruvate, 1% non-essential amino acids solution (Gibco, 11-140-050), 1 M HEPES (Gibco, 15-630-106) and 0.05 mM 2-mercaptoethanol.

KP cells were seeded at a density of 2.2 × 106 per 15-cm tissue culture plate and incubated for 30 h. Cells were treated with tunicamycin (200 ng ml−1) for 15 h without medium replacement; control samples received an equivalent volume of PBS. Cells were cross-linked in 1% formaldehyde in PBS for 8 min at 37 °C. Cross-linking was quenched by washing once with 10 ml of 5 mg ml−1 BSA in cold PBS (3 min, room temperature), followed by two washes with cold PBS.

Cells were scraped in 700 μl cold PBS, pelleted (2,000 rpm, 2 min) and lysed in 1 ml of 1% SDS ChIP lysis buffer supplemented with 1× protease inhibitor cocktail (Thermo Fisher Scientific, P178444). Lysates were incubated on ice for 20 min and stored at −80 °C. Chromatin fragmentation was performed with sonication (Bioruptor Plus, high intensity, 21 cycles; 30 s on, 30 s off).

DNA was purified using the Zymo-Spin ChIP kit following the manufacturer’s instructions: cross-link reversal at 65 °C for 90 min with proteinase K, followed by DNA binding, washes and elution in 8 μl DNA elution buffer. DNA integrity and fragment size distribution were confirmed by 1.5% agarose gel electrophoresis. Chromatin immunoprecipitation: protein G dynabeads (50 μl; Invitrogen, 10-003-D) were washed twice with PBS + 5 mg ml−1 BSA and incubated with 5 μg antibody (ATF4 or IgG control) for six hours at 4 °C.

Antibody-coated beads were washed and incubated with 500 μl sonicated chromatin overnight at 4 °C. Beads were washed six times with RIPA buffer (with 15-min rotations every two washes) and twice with cold TE buffer. Elution and reverse cross-linking: beads were resuspended in 250 μl TE buffer with 5 μl RNase A (Qiagen, 19101) and incubated for one hour at 37 °C, followed by proteinase K digestion (1 μl, 15 min, 65 °C).

Chromatin was eluted in 50 μl 2× elution buffer (Zymo) for 30 min at 65 °C with vortexing every 5 min, and reverse cross-linking was continued for one hour at 65 °C. DNA was purified with the Zymo-Spin ChIP kit and eluted in 20 μl elution buffer for downstream qPCR analysis. Pan-monocytes were seeded under suspension culture conditions in 24-well flat-bottom ultra-low-adhesion plates at day 0 and transfected with 10 μM of Slc22a17 siRNA (mm.Ri.Slc22a17.13.1) or scrambled negative control siRNA (Qiagen, 102728) using lipofectamine RNAiMAX reagent (Invitrogen, 13-778-075) in OptiMEM medium (Gibco, 31985062) following the manufacturer’s protocol, in the presence of 5 ng ml−1 macrophage colony-stimulating factor (M-CSF) to promote cell survival.

After 72 h, cells were collected and used for 3D monocyte infiltration. Lcn2KO and Lcn2KO+mLcn2 cancer cells were seeded in a 2 mg ml−1 collagen type I (Thermo Fisher Scientific, CB-40236) solution at a concentration of 2 × 106 cells per ml, and placed in each well of a 96-well plate. These collagen solutions were allowed to polymerize for 30 min, forming a 3D matrix. Next, 50,000 pan-monocytes stained with CellTracker Deep Red (Invitrogen, C34565) were seeded in 200 µl of medium per well.

For the blocking antibody experiment, the LCN2 blocking antibody was added at a concentration of 1 µg ml−1. For the siRNA knockdown experiment, pan-monocytes were seeded with penicillin–streptomycin-free DMEM. After 24 h of co-culture, the plate was imaged using the ImageXpress Micro Confocal system to assess the number of monocytes recruited inside the collagen matrix, using five z-stacks at a 20-µm interval.

The number of monocytes was quantified using the TrackMate plug-in in ImageJ. Bone-marrow cells were seeded in a 96-well plate (2 × 105 cells per well) and were treated with 20 ng ml−1 M-CSF for five days to induce macrophage differentiation. Macrophages were co-cultured with Lcn2KO or Lcn2KO+mLcn2(Lcn2WT) cancer cells at a 100:1 ratio for two days in 96-well plates. Next, these cancer-conditioned macrophages (2 × 107 cells per ml) and cancer cells (2 × 107 cells per ml) were seeded in a collagen 3D matrix, followed by the seeding of 5 × 104 CellTracker Deep Red-stained CD8 T cells within each well.

For the blocking antibody conditions, LCN2 and CXCL9 blocking antibodies were added at concentrations of 1 µg ml−1 and 40 µg ml−1, respectively. After 48 h of co-culture, the plate was imaged using a Zeiss LSM700 Confocal to assess the number of CD8 T cells recruited into the collagen matrix, using three z-stacks at a 15-µm interval. The number of CD8 T cells was quantified using the TrackMate plug-in in ImageJ.

Synthetic genes encoding human LCN2 (Uniprot ID: P80188) and mouse LCN2 (Uniprot ID: P11672), including their secretion signal sequences (residues 1–20), were cloned into the plasmid pBCAG in such a way that they were fused C-terminally with an Avi-tag and a polyhistidine tag. The plasmids were used to transfect Expi293F cells following the standard protocol from the vendor (Thermo Fisher Scientific).

Transfected cells were incubated at 37 °C with 8% CO2 and collected on day 7 after transfection. After removing cells by centrifugation, the supernatant was dialysed against 20 mM sodium phosphate buffer pH 7.4 containing 500 mM NaCl. The dialysed solution was filtered and loaded onto a HisTrap excel column (Cytiva) pre-equilibrated in 20 mM sodium phosphate buffer pH 7.4 containing 500 mM NaCl, and eluted with 20 mM sodium phosphate buffer pH 7.4, containing 500 mM NaCl and 0.5 M imidazole.

Eluted fractions from the HisTrap column containing the expressed protein were pooled and dialysed against 20 mM sodium phosphate buffer pH 7.4 containing 500 mM NaCl. For in vitro biotinylation, the protein was dialysed against 50 mM bicine buffer pH 8.0, and the biotinylation reaction was initiated by adding 10 mM magnesium acetate, 10 mM ATP, 0.5 mM biotin and in-house-prepared BirA enzyme at a 0.1 molar ratio to LCN2 (all final concentrations).

The reaction mixture was incubated at 30 °C for one hour, and the LCN2 protein was purified with a gravity-flow Ni-NTA column and dialysed against gel filtration buffer (50 mM Tris HCl buffer pH 7.5 containing 150 mM NaCl). Finally, the sample was loaded onto a Superdex S75 Increase 10/300 GL size-exclusion column (Cytiva). Fractions containing the LCN2 protein were pooled, concentrated, aliquoted and stored at −80 °C until needed.

The sorting of a synthetic human antibody library was performed as described previously78,79. In brief, the phage library was incubated with biotinylated antigens at concentrations of 100 nM in the first and second rounds, 50 nM in the third round and 20 nM in the fourth round. Phage clones from the sorted library were assessed by phage multiplex bead binding assay80. The genes for Fab clones were subcloned into a bacterial expression vector using Golden Gate assembly (NEB)81, and Fab samples were produced from the 55244 Escherichia coli strain (ATCC) using published procedures78,79.

Antibodies in the mouse IgG1 format were produced by Biointron. Biolayer interferometry measurements were performed in 50 mM Tris HCl buffer pH 7.5 containing 150 mM NaCl, 0.5% BSA and 0.005% Tween 20 at 30 °C on an Octet RED96e instrument (Sartorius). Streptavidin sensors were used to immobilize a biotinylated analyte, and binding to a nonbiotinylated ligand was measured. Data were analysed with Octet data analysis software v12.0.2.59 (Sartorius).

RNA sequencing (RNA-seq) data for five cancer cohorts from the publicly available TCGA programme—lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), non-small cell lung cancer (NSCLC; LUAD + LUSC), pancreatic adenocarcinoma (PDAC) and skin cutaneous melanoma (SKCM)—were downloaded via the UCSC Xena platform in HTSeq-FPKM format. Only tumour samples were retained. Fragments per kilobase of transcript per million mapped reads (FPKM) values were converted to transcripts per million (TPM) using the formula: TPM = (FPKM/sum(FPKM)) × 106.

Subsequently, log2(TPM + 1) transformation was applied to stabilize variance and facilitate visualization. All analyses were done using log2-transformed TPM values unless otherwise specified. An ATF4-downregulated gene signature was compiled from a publicly available dataset published by Igarashi et al.82. For each tumour sample, a signature score was computed as the mean log2(TPM + 1) expression of genes in the signature.

LCN2 expression was analysed independently and also used to stratify samples into tertiles. The low, medium and high LCN2 expression groups were defined using the 25th and 75th percentiles of log2 (TPM + 1) LCN2 expression within each cancer type. We used Luminex xMAP technology to quantitatively and simultaneously detect 32 mouse cytokines, chemokines and growth factors. The multiplexing analysis was performed by Eve Technologies using the Luminex 200 system (DiaSorin) with Bio-Plex Manager software (Bio-Rad).

Thirty-two markers were measured in the samples using the Eve Technologies Mouse Cytokine/Chemokine 32-Plex Discovery Assay Array (MD32) as per the manufacturer’s instructions for use (MILLIPLEX Mouse Cytokine/Chemokine Magnetic Bead Panel, MCYTOMAG-70K, MilliporeSigma). The 32-plex consisted of eotaxin/CCL11, G-CSF/CSF-3, GM-CSF, GROα/CXCL1/KC/CINC-1, GROβ/CXCL2/MIP-2/CINC-3, IFN-γ, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12(p40), IL-12(p70), IL-13, IL-15, IL-17, IP-10/CXCL10, LIF, LIX, MCP-1/CCL2, M-CSF, MIG/CXCL9, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, TNF and VEGF-A.

Assay sensitivities of these markers range from 0.3 to 30.6 pg ml−1. Individual analyte sensitivity values are available in the MilliporeSigma MILLIPLEX protocol. Pairwise Pearson correlation between LCN2 expression and the ATF4-downregulated gene signature was computed within each cancer cohort. For stratified analysis, samples were grouped by LCN2 expression tertiles. Groupwise differences were assessed using Wilcoxon rank-sum tests, and P ≤ 0.05 was considered statistically significant.

For spatial immune profiling and associated molecular marker analysis, we used transcriptomic data and WSIs from TCGA, encompassing 23 types of epithelial solid tumour: bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma (CESC), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), oesophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), testicular germ cell tumour (TGCT), thyroid carcinoma (THCA), uterine corpus endometrial carcinoma (UCEC) and uveal melanoma (UVM).

All materials were accessed through the Genomic Data Commons Portal (https://portal.gdc.cancer.gov/). For quantitative analysis of H&E WSIs, we used Lunit SCOPE IO, an artificial intelligence (AI) model developed using more than 26 tumour types. The cell detection model was constructed using 20,617 patches extracted from 5,609 WSIs (3,798 for training and 1,811 for optimization). The tissue segmentation component used 76,110 patches from 18,935 WSIs (15,936 for training and 2,999 for optimization).

Board-certified pathologists annotated 2,828,448 cells, 1.70 × 1010 μm2 of cancer and stromal regions, 706,320 lymphocytes and 56,972 macrophages to train the model. The spatial distribution of TILs was evaluated by partitioning each WSI into 0.25-mm2 grids regardless of image dimensions. The AI system quantified both intratumoural and stromal TIL densities and categorized immune phenotypes on the basis of established thresholds: the inflamed phenotype required intratumoural TIL density ≥130 per mm2; the immune-excluded phenotype was defined by intratumoural TIL density <130 per mm2 with stromal TIL density ≥260 per mm2; and the immune-desert phenotype required TIL densities below these thresholds in both compartments.

The predominant immune phenotype for each WSI was determined by grid distribution analysis: inflamed when ≥33.3% of grids displayed inflamed characteristics; immune-excluded when ≥33.3% of grids showed immune-excluded patterns while inflamed grids remained <33.3%; and immune-desert in all other cases. Intratumoural TIL density = lymphocyte count in cancer area/cancer area (mm2). Macrophage density = macrophage count in cancer and stroma/combined area (mm2).

RNA expression data from TCGA, quantified as RNA-seq by expectation maximization (RSEM) values, underwent log2(RSEM + 1) transformation for scaling. We examined the differential expression of LCN2 and ATF4 between inflamed and immune-excluded phenotypes using analysis done with the Wilcoxon rank-sum test (Mann–Whitney U test), implemented in Python’s scipy.stats library to identify significant differences between these specific immune contexts.

Statistical significance was determined at P < 0.05, and results were visualized using box plots with individual data points to represent the distribution of gene expression across these two phenotypes. Furthermore, we stratified cases into three groups (low, moderate and high) based on LCN2 expression levels, and compared several microenvironmental variables, including stromal-to-intratumoural TIL ratio and macrophage density, using the same statistical approach.

The validation cohort consisted of 380 patients with NSCLC who were treated at Samsung Medical Center (SMC) with ICIs, including anti-PD1, anti-PD-L1 and anti-CTLA-4 agents, administered as either monotherapy or combination therapy, between September 2014 and July 2022. Immune phenotypes were determined using Lunit SCOPE IO following the same methodology as that described for the TCGA cohort. RNA expression data, quantified as TPM values, underwent log2(TPM + 1) transformation for scaling.

To investigate the association between LCN2 expression and immune exclusion patterns, we compared LCN2 expression levels between inflamed and immune-excluded phenotypes using the Wilcoxon rank-sum test. To elucidate the relationship between LCN2 expression and immune exclusion patterns, survival analysis was conducted exclusively on patients with inflamed and excluded immune phenotypes. Overall survival was defined as the time from treatment initiation to death from any cause or last follow-up.

Patients were stratified into high and low LCN2 expression groups on the basis of the median expression value, and survival differences were evaluated using Kaplan–Meier analysis and Cox proportional hazards models. Statistical analysis was performed using GraphPad Prism v9. All data are expressed as mean and s.e.m., unless otherwise specified. Data were analysed by the statistical tests indicated in the figure legends.

All tests were two-tailed unless specified otherwise. Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Data have been deposited in the GEO under accession number GSE308689. Any additional data are available upon reasonable request. Source data are provided with this paper. Pakos-Zebrucka, K. et al. The integrated stress response.

EMBO Rep. 17, 1374–1395 (2016). Tameire, F. et al. ATF4 couples MYC-dependent translational activity to bioenergetic demands during tumour progression. Nat. Cell Biol. 21, 889–899 (2019). Gwinn, D. M. et al. Oncogenic KRAS regulates amino acid homeostasis and asparagine biosynthesis via ATF4 and alters sensitivity to l-asparaginase. Cancer Cell 33, 91–107 (2018). Hellmann, M. D. et al. Genomic features of response to combination immunotherapy in patients with advanced non-small-cell lung cancer.

Cancer Cell 33, 843–852 (2018). Seth, R. et al. Systemic therapy for melanoma: ASCO guideline update. J. Clin. Oncol. 41, 4794–4820 (2023). Benjamin, D. J., Haslam, A., Gill, J. & Prasad, V. Targeted therapy in lung cancer: are we closing the gap in years of life lost?. Cancer Med. 11, 3417–3424 (2022). Genova, C. et al. Therapeutic implications of tumor microenvironment in lung cancer: focus on immune checkpoint blockade.

Front. Immunol. 12, 799455 (2021). Global Burden of Disease 2019 Cancer Collaboration. Cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life years for 29 cancer groups from 2010 to 2019: a systematic analysis for the Global Burden of Disease Study 2019. JAMA Oncol. 8, 420–444 (2022). Rangachari, D. & Costa, D. B. From hope to reality: durable overall survival with immune checkpoint inhibitors for advanced lung cancer.

J. Clin. Oncol. 37, 2511–2513 (2019). Akman, M. et al. Hypoxia, endoplasmic reticulum stress and chemoresistance: dangerous liaisons. J. Exp. Clin. Cancer Res. 40, 28 (2021). Cubillos-Ruiz, J. R., Bettigole, S. E. & Glimcher, L. H. Tumorigenic and immunosuppressive effects of endoplasmic reticulum stress in cancer. Cell 168, 692–706 (2017). Li, F. & Simon, M. C. Cancer cells don’t live alone: metabolic communication within tumor microenvironments.

Dev. Cell 54, 183–195 (2020). Obacz, J. et al. Regulation of tumor–stroma interactions by the unfolded protein response. FEBS J. 286, 279–296 (2019). Verginadis, I. I. et al. A stromal integrated stress response activates perivascular cancer-associated fibroblasts to drive angiogenesis and tumour progression. Nat. Cell Biol. 24, 940–953 (2022). Ye, J. et al. The GCN2–ATF4 pathway is critical for tumour cell survival and proliferation in response to nutrient deprivation.

EMBO J. 29, 2082–2096 (2010). Bhattacharya, S. et al. Anti-tumorigenic effects of type 1 interferon are subdued by integrated stress responses. Oncogene 32, 4214–4221 (2013). Nguyen, H. G. et al. Development of a stress response therapy targeting aggressive prostate cancer. Sci. Transl. Med. 10, eaar2036 (2018). Ghaddar, N. et al. The integrated stress response is tumorigenic and constitutes a therapeutic liability in KRAS-driven lung cancer.

Nat. Commun. 12, 4651 (2021). Krall, A. S. et al. Asparagine couples mitochondrial respiration to ATF4 activity and tumor growth. Cell Metab. 33, 1013–1026 (2021). Li, J. et al. Non-cell-autonomous cancer progression from chromosomal instability. Nature 620, 1080–1088 (2023). De Ponte Conti, B. et al. mTOR-dependent translation drives tumor infiltrating CD8+ effector and CD4+ Treg cells expansion.

eLife 10, e69015 (2021). Zyryanova, A. F. et al. ISRIB blunts the integrated stress response by allosterically antagonising the inhibitory effect of phosphorylated eIF2 on eIF2B. Mol. Cell 81, 88–103 (2021). Doench, J. G. et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR–Cas9. Nat. Biotechnol. 34, 184–191 (2016). Moschen, A. R., Adolph, T. E., Gerner, R.

R., Wieser, V. & Tilg, H. Lipocalin-2: a master mediator of intestinal and metabolic inflammation. Trends Endocrinol. Metab. 28, 388–397 (2017). Adler, O. et al. Reciprocal interactions between innate immune cells and astrocytes facilitate neuroinflammation and brain metastasis via lipocalin-2. Nat. Cancer 4, 401–418 (2023). Flo, T. H. et al. Lipocalin 2 mediates an innate immune response to bacterial infection by sequestrating iron.

Nature 432, 917–921 (2004). Rehwald, C. et al. The iron load of lipocalin-2 (LCN-2) defines its pro-tumour function in clear-cell renal cell carcinoma. Br. J. Cancer 122, 421–433 (2020). Hoette, T. M. et al. Immune interference in Mycobacterium tuberculosis intracellular iron acquisition through siderocalin recognition of carboxymycobactins. ACS Chem. Biol. 6, 1327–1331 (2011). Iannetti, A. et al.

The neutrophil gelatinase-associated lipocalin (NGAL), a NF-κB-regulated gene, is a survival factor for thyroid neoplastic cells. Proc. Natl Acad. Sci. USA 105, 14058–14063 (2008). Tillmann, S. et al. The unfolded protein response is a major driver of LCN2 expression in BCR–ABL- and JAK2V617F-positive MPN. Cancers 13, 4210 (2021). El Karoui, K. et al. Endoplasmic reticulum stress drives proteinuria-induced kidney lesions via lipocalin 2.

Nat. Commun. 7, 10330 (2016). Wen, J. et al. Efficacy of metformin therapy in patients with cancer: a meta-analysis of 22 randomised controlled trials. BMC Med. 20, 402 (2022). Pillai, R. et al. Glutamine antagonist DRP-104 suppresses tumor growth and enhances response to checkpoint blockade in KEAP1 mutant lung cancer. Sci. Adv. 10, eadm9859 (2024). Lavin, Y., Mortha, A., Rahman, A. & Merad, M. Regulation of macrophage development and function in peripheral tissues.

Nat. Rev. Immunol. 15, 731–744 (2015). Engblom, C., Pfirschke, C. & Pittet, M. J. The role of myeloid cells in cancer therapies. Nat. Rev. Cancer 16, 447–462 (2016). Casanova-Acebes, M. et al. Tissue-resident macrophages provide a pro-tumorigenic niche to early NSCLC cells. Nature 595, 578–584 (2021). Noy, R. & Pollard, J. W. Tumor-associated macrophages: from mechanisms to therapy. Immunity 41, 49–61 (2014).

Miret, J. J. et al. Suppression of myeloid cell arginase activity leads to therapeutic response in a NSCLC mouse model by activating anti-tumor immunity. J. Immunother. Cancer 7, 32 (2019). Molgora, M. et al. TREM2 modulation remodels the tumor myeloid landscape enhancing anti-PD-1 immunotherapy. Cell 182, 886–900 (2020). Bill, R. et al. CXCL9:SPP1 macrophage polarity identifies a network of cellular programs that control human cancers.

Science 381, 515–524 (2023). Czech, M. et al. Lipocalin-2 expression identifies an intestinal regulatory neutrophil population during acute graft-versus-host disease. Sci. Transl. Med. 16, eadi1501 (2024). Hermann, A. et al. Lipocalin-2 levels in acute and chronic graft-versus-host disease following allogeneic hematopoietic stem cell transplantation. Exp. Hematol. 74, 25–32 (2019). Warszawska, J.

M. et al. Lipocalin 2 deactivates macrophages and worsens pneumococcal pneumonia outcomes. J. Clin. Invest. 123, 3363–3372 (2013). Yao, F. et al. A targetable LIFR-NF-κB–LCN2 axis controls liver tumorigenesis and vulnerability to ferroptosis. Nat. Commun. 12, 7333 (2021). Toyonaga, T. et al. Lipocalin 2 prevents intestinal inflammation by enhancing phagocytic bacterial clearance in macrophages.

Sci. Rep. 6, 35014 (2016). Wang, L. et al. PD-L1-expressing tumor-associated macrophages are immunostimulatory and associate with good clinical outcome in human breast cancer. Cell Rep. Med. 5, 101420 (2024). An, H. S. et al. Lipocalin-2 promotes acute lung inflammation and oxidative stress by enhancing macrophage iron accumulation. Int. J. Biol. Sci. 19, 1163–1177 (2023). House, I. G. et al. Macrophage-derived CXCL9 and CXCL10 are required for antitumor immune responses following immune checkpoint blockade.

Clin. Cancer Res. 26, 487–504 (2020). Bent, E. H. et al. Microenvironmental IL-6 inhibits anti-cancer immune responses generated by cytotoxic chemotherapy. Nat. Commun. 12, 6218 (2021). Byles, V. et al. The TSC–mTOR pathway regulates macrophage polarization. Nat. Commun. 4, 2834 (2013). Rashidfarrokhi, A. et al. Tumor-intrinsic LKB1–LIF signaling axis establishes a myeloid niche to promote immune evasion and tumor growth.

Preprint at bioRxiv https://doi.org/10.1101/2023.07.15.549147 (2023). Moreira, A. L. et al. A grading system for invasive pulmonary adenocarcinoma: a proposal from the International Association for the Study of Lung Cancer Pathology Committee. J. Thorac. Oncol. 15, 1599–1610 (2020). Hwang, J.-E. et al. Tumor immune microenvironment biomarkers for recurrence prediction in locally advanced rectal cancer patients after neoadjuvant chemoradiotherapy.

Cancers 16, 3353 (2024). Győrffy, B. Transcriptome-level discovery of survival-associated biomarkers and therapy targets in non-small-cell lung cancer. Br. J. Pharmacol. 181, 362–374 (2024). Zavitsanou, A. M. et al. KEAP1 mutation in lung adenocarcinoma promotes immune evasion and immunotherapy resistance. Cell Rep. 42, 113295 (2023). Wood, L. D., Canto, M. I., Jaffee, E. M. & Simeone, D. M. Pancreatic cancer: pathogenesis, screening, diagnosis, and treatment.

Gastroenterology 163, 386–402 (2022). Werba, G. et al. Single-cell RNA sequencing reveals the effects of chemotherapy on human pancreatic adenocarcinoma and its tumor microenvironment. Nat. Commun. 14, 797 (2023). Guo, H., Jin, D. & Chen, X. Lipocalin 2 is a regulator of macrophage polarization and NF-κB/STAT3 pathway activation. Mol. Endocrinol. 28, 1616–1628 (2014). Watzenboeck, M. L. et al.

Lipocalin 2 modulates dendritic cell activity and shapes immunity to influenza in a microbiome dependent manner. PLoS Pathog. 17, e1009487 (2021). Treekitkarnmongkol, W. et al. Augmented lipocalin-2 Is associated with chronic obstructive pulmonary disease and counteracts lung adenocarcinoma development. Am. J. Respir. Crit. Care Med. 203, 90–101 (2021). Zhuang, X. et al. Ageing limits stemness and tumorigenesis by reprogramming iron homeostasis.

Nature 637, 184–194 (2025). La Manna, G. et al. Neutrophil gelatinase-associated lipocalin increases HLA-G+/FoxP3+ T-regulatory cell population in an in vitro model of PBMC. PLoS ONE 9, e89497 (2014). Mimoto, F. et al. Exploitation of elevated extracellular ATP to specifically direct antibody to tumor microenvironment. Cell Rep. 33, 108542 (2020). Ishihara, J. et al. Targeted antibody and cytokine cancer immunotherapies through collagen affinity.

Sci. Transl. Med. 11, eaau3259 (2019). Desnoyers, L. R. et al. Tumor-specific activation of an EGFR-targeting probody enhances therapeutic index. Sci. Transl. Med. 5, 207ra144 (2013). Romero, R. et al. Keap1 loss promotes Kras-driven lung cancer and results in dependence on glutaminolysis. Nat. Med. 23, 1362–1368 (2017). Ding, H. et al. Activation of the NRF2 antioxidant program sensitizes tumors to G6PD inhibition.

Sci. Adv. 7, eabk1023 (2021). LeBoeuf, S. E. et al. Activation of oxidative stress response in cancer generates a druggable dependency on exogenous non-essential amino acids. Cell Metab. 31, 339–350 (2020). Lignitto, L. et al. Nrf2 activation promotes lung cancer metastasis by inhibiting the degradation of Bach1. Cell 178, 316–329 (2019). Sayin, V. I. et al. Activation of the NRF2 antioxidant program generates an imbalance in central carbon metabolism in cancer.

eLife 6, e28083 (2017). Sánchez-Rivera, F. J. et al. Rapid modelling of cooperating genetic events in cancer through somatic genome editing. Nature 516, 428–431 (2014). Soto-Feliciano, Y. M. et al. A molecular switch between mammalian MLL complexes dictates response to menin–MLL inhibition. Cancer Discov. 13, 146–169 (2023). Doench, J. G. et al. Rational design of highly active sgRNAs for CRISPR–Cas9-mediated gene inactivation.

Nat. Biotechnol. 32, 1262–1267 (2014). Li, W. et al. MAGeCK enables robust identification of essential genes from genome-scale CRISPR/Cas9 knockout screens. Genome Biol. 15, 554 (2014). Mimitou, E. P. et al. Multiplexed detection of proteins, transcriptomes, clonotypes and CRISPR perturbations in single cells. Nat. Methods 16, 409–412 (2019). Stoeckius, M. et al. Simultaneous epitope and transcriptome measurement in single cells.

Nat. Methods 14, 865–868 (2017). McInnes, L., Healy, J., Saul, N. & Großberger, L. UMAP: uniform manifold approximation and projection. J. Open Source Softw. 3, 861 (2018). Miller, K. R. et al. Tcell receptor-like recognition of tumor in vivo by synthetic antibody fragment. PLoS ONE 7, e43746 (2012). Oury, J. et al. Mechanism of disease and therapeutic rescue of Dok7 congenital myasthenia. Nature 595, 404–408 (2021).

Hattori, T. et al. Multiplex bead binding assays using off-the-shelf components and common flow cytometers. J. Immunol. Methods 490, 112952 (2020). Maso, L. et al. Molecular basis for antibody recognition of multiple drug–peptide/MHC complexes. Proc. Natl Acad. Sci. USA 121, e2319029121 (2024). Igarashi, T. et al. Clock and ATF4 transcription system regulates drug resistance in human cancer cell lines.

Oncogene 26, 4749–4760 (2007). J.P.B. received funding from the ARC Fondation Postdoctorat 2 retour de l'étranger 2024 scholarship. T.P. is supported by National Institutes of Health (NIH) grants (R37CA222504, R01CA227649, R01CA283049 and R01CA262562) and by an American Cancer Society Research Scholar grant (RSG-17-200-01–TBE). T.P. and S.K. were supported by NIH grant R01CA297605. S.B.K. is supported by NIH grant CA271245 and by the Sale Johnson Philanthropic Fund and the LEO Foundation.

The project was also supported by funds from NIH Melanoma SPORE grant NCI P50 CA225450 (J. S. Weber and I. Osman) and U54 CA2630001 (E. Hernando and I. Osman). I.K.Z. is supported by NSF CAREER 2337385 and A.W.L. by NIH T32EB034216. We thank members of the Experimental Pathology Research Laboratory (RRID: SCR_017928), which is partially supported by the Cancer Center support grant NIH/NCI P30CA016087 at NYU Langone’s Laura and Isaac Perlmutter Cancer Center.

The original multispectral imaging system was awarded through a shared instrumentation grant S10 OD021747. We thank P. Meyn for advice in DNA sequencing experiments; the Center for Biospecimen Research and Development at NYU Langone Health (NYULH) for support with tissue acquisition; the NYULH Center for Biospecimen Research and Development for access to tumour biospecimens; Histology and Immunohistochemistry Laboratory (RRID: SCR_018304) for IHC work, which is supported in part by P30CA016087; and J.

Teixeira for supporting and motivating this work. Jozef P. Bossowski, Ray Pillai, Mari Nakamura, Ali Rashidfarrokhi, Yuan Hao, Katherine Wu, Andre L. Moreira, Cristina Hajdu, Sahith Rajalingam, Sarah E. LeBoeuf, Hortense Le, Hao Wang, Aristotelis Tsirigos, Kwok-Kin Wong, Sergei B. Koralov & Thales Papagiannakopoulos John Kilian, Ali Rashidfarrokhi, Takamitsu Hattori, Eliezra Glasser, Akiko Koide, Sergei B.

Koralov, Shohei Koide & Thales Papagiannakopoulos David H. Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA, USA T.P. and S.K. funded, supervised and directed the project. J.P.B., T.P. and S.K. designed experiments. F.J.S.-R. designed a custom CRISPR library and provided scientific advice. S.K., J.K., A.K., T.H. and E.G. developed antibodies against LCN2.

J.P.B., J.K., R.P., A.R., A.W.L., M.N., R.L., K.W. and L.W. performed in vitro and in vivo experiments. A.W.L., A.L.M. and I.K.Z. performed in vitro co-culture studies. C.H. analysed patient biopsies. Y.H. and H.L. performed bioinformatic analyses of CITE-seq and TCGA data. S.L., J.W.O., C.J., H.K., C.-Y.O. and S.-H.L. provided AI-based WSI Lunit analysis. A.A.H.P. performed experiments. A.T., K.-K.W., S.B.K., M.P., F.J.S.-R., D.M.S., I.K.Z., R.P., A.A.H.P, K.W., H.W.

and V.I.S. provided conceptual advice. J.P.B. assembled figures. J.P.B., T.P. and S.K. wrote and edited the manuscript. All authors read and approved the final version of the manuscript. T.P. received funding from Pfizer Medical Education Group, Dracen Pharmaceuticals, Kymera Therapeutics, Bristol Myers Squibb and Agios Pharmaceuticals, not related to the submitted work. S.K. is a co-founder of, receives consulting fees from and holds equity in Aethon Therapeutics; is a co-founder of and holds equity in Revalia Bio; and has received research funding from Aethon Therapeutics, Argenx BVBA, Black Diamond Therapeutics and Puretech Health, all outside of the current work.

S.B.K. has previously received funding from Micreos, BMS and Kymera Therapeutics. T.P., S.K., J.P.B. and J.K. are listed as inventors on a patent application (PCT/US24/19489: ‘Anti-lipocalin-2 antibodies and uses thereof’). Nature thanks David Feldser, Constantinos Koumenis and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

(a) Phenotypic characterization of Atf4WT and Atf4KO KP LUAD cells. Western blot of protein lysates from KP cells with Atf4WT and Atf4KO status (similar results were obtained in n = 3 independent experiments). Representative images of crystal violet staining, clonogenic assay, and in vitro proliferation rates (n = 3 biological replicates). Data were analysed with a two-tailed t-test. (b,c) Individual tumour growth kinetics of experiments depicted and described in Fig.

1d,e. (d) Quantification of KI67-positive and Cleaved Caspase 3-positive cells from KP tumours collected from C57BL/6J mice with Atf4WT (n = 8) and Atf4KO (n = 5) status. Data were analysed with a two-tailed t-test. (e–g) Individual tumour growth kinetics of experiments depicted in Fig. 1f–h. (h,i) Averaged (h) and individual (i) tumour growth of KP LUAD s.c. transplanted tumours treated with ISRIB or vehicle from day 9 after tumour transplantation in C57BL/6J mice.

Vehicle n = 20, ISRIB n = 20. Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. (j) Tumour growth of KP LUAD orthotopically transplanted tumours in NSG mice treated with ISRIB (2.5 mg/kg) or vehicle. Vehicle n = 3, ISRIB n = 4. Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. (k) Tumour growth of KP LUAD s.c.

transplanted tumours in NSG mice treated with ISRIB (2.5 mg/kg) daily from day two after implantation. Vehicle n = 10, ISRIB n = 9. Arrows indicate ISRIB treatment regimen. Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. Data are mean values +/− SEM. (a) Waterfall plot of genes significantly depleted (p < 0.05) in C57BL/6J versus NSG mice.

Genes are ranked according to their negative depletion scores. (b) Representative crystal violet-stained images from clonogenic assays of KP LUAD cells: WT (sgTom) and Lcn2 knockout (sgLcn2). (c,d) Individual growth of Lcn2KO KP tumours transplanted into (c) C57BL/6J (n = 8 in Lcn2WT, n = 10 in Lcn2KO group) and (d) NSG immunodeficient mice (n = 24 in Lcn2WT, n = 13 in Lcn2KO group), as depicted in Fig.

2c,d. Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. (e) Normalized individual growth bioluminescence signal from orthotopic KP tumours expressing Lcn2 WT (n = 6), Lcn2 knockout (Lcn2KO, n = 5), and Lcn2KO KP cells with ectopic overexpression of mouse Lcn2 (mLcn2, n = 6) as depicted in Fig. 2e. Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test.

(f) Tumour burden quantification based on H&E histology. Lcn2WT n = 8, Lcn2KO n = 6, Lcn2KO+mLcn2 n = 10. Data were analysed by a two-way ANOVA with Fisher’s LSD test. (g) Levels of KI67-positive cells were quantified in KP tumours as depicted in Fig. 2e. Data were analysed by a one-way ANOVA with Šidák’s multiple-comparisons test. (h) Final weight of tumours as depicted in Fig. 2f.

Atf4WTCtr n = 7, Atf4WTLcn2OE n = 7 Atf4KOCtr n = 7, Atf4KOLcn2OE n = 5. Mean with SEM, data were analysed by a one-way ANOVA with Fisher’s LSD test. (i) Schematic graph of lentiviral construct V2-EFS_Hygro with mouse WT Lcn2 (WT), Lcn2 lacking secretion signal peptide sequence (sec_del), and two independent triple mutant Lcn2 constructs expressing Lcn2 with substituted amino acids responsible for iron-binding features (Lcn2Y127A K147A K156A; Lcn2R103G K147A K156A).

(j,k) Binding of iron–catechol to mLcn2 and its mutants assessed using absorption spectroscopy. (j) Size-exclusion chromatograms of mLcn2 proteins used in the absorption assay. (k) Iron–catechol mixture (25 μM FeCl3 and 75 µM catechol) in the presence of mLCN2 proteins (30 μM; red curves). The spectrum of the iron–catechol mixture in the absence of mLCN2 is shown in black in the top panel with λmax = 575 nm (the vertical line marked for reference).

Spectra of mLCN2 in the absence of the iron–catechol mixture are shown in blue. The binding of the iron–catechol mixture to mLCN2-WT results in the blue shift of the spectrum. (l) Individual kinetics of tumour growth of s.c. transplanted tumours with different LCN2 form status in C57BL/6J mice as depicted in Fig. 2g. KP cells were stably transfected with lentiviral constructs as depicted in (Extended Data Fig.

2i) and hygromycin selected. Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test. (m) Tumour weight resected 28 days after s.c. implantation from mice as depicted in Fig. 2g. Bars represent SEM values. Empty vector n = 9, WT Lcn2 (WT, n = 10), Lcn2 lacking secretion signal peptide sequence (sec_del, n = 10), and Lcn2Y127AK147AK156A, n = 7; Lcn2R103G K147A K156A, n = 7).

Data were analysed by a two-way ANOVA with Holm-Šídák’s multiple-comparisons test. (n,o) ELISA (n) and Western blot (o) of KP cells Lcn2KO with Lcn2 lentiviral constructs as depicted in (i). Cancer cells were stably transfected with lentiviral constructs and hygromycin selected to obtain stable cell lines. Representative results of n = 2 similar biological replicates. (p) Calculated tumour volume based on MRI imaging of KPC mice 14 weeks after sgNeo (n = 6) and sgLcn2 (n = 6) infection.

Data were analysed by a two-tailed t-test. (q) LCN2 protein levels measured by ELISA in BALF (sgNeo n = 7, sgLcn2 n = 7). Data were analysed by a two-tailed t-test. (r) average and (s) individual tumour growth of B16F10 s.c. transplanted tumours with Lcn2WT (n = 17) and Lcn2KO (n = 10) status in C57BL/6J mice. Data were analysed by a two-way ANOVA with Šidák’s multiple-comparisons test.

(t,u) Growth of orthotopic KPC PDAC tumours Lcn2WT (n = 8) and Lcn2KO (n = 8). Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. (v) Normalized levels of Lcn2 gene expression in KP LUAD cells stably infected with doxycycline-inducible shLcn2 or shCtr KP cells that were treated under the indicated conditions for 24 h. Mouse interleukin 1 beta (IL1b, 5 ng/mL) Doxycycline (DOX, 1 μg/mL).

Data were analysed by a two-way ANOVA with Dunnett’s multiple-comparisons test. (w,x) Tumour growth of doxycycline-inducible knockdown of Lcn2 in orthotopically engrafted KP tumours in mouse lungs (shCtr, n = 7; shLcn2#1, n = 7; shLcn2#2, n = 3), and their survival (y,z). Mice were given a doxycycline diet (200 mg/kg, Bio-Serv, S3888) or standard chow (Ctr). Data were analysed by a two-way ANOVA with Dunnett’s multiple-comparisons test and Log-rank (Mantel–Cox) test.

(aa) Tumour burden based on H&E histology of lungs from mice depicted in (w). Data were analysed by a two-way ANOVA with Dunnett’s multiple-comparisons test. (ab) Representative histology pictures of lung lobes from mice bearing shLcn2 KP tumours, as depicted in (w). Data are mean values +/−SEM unless specified otherwise. (a) Schematic of the ISR pathway upon various stressors (mitochondrial, cytoplasmic, and endoplasmic reticulum disruptors).

(b) Western blot of ATF4 levels in KP cells under glutamine deprivation (-Q), tunicamycin (Tun, 1 ug/mL), and IL-1β stimulation (5 ng/mL) over 6, 12, and 24 h after stimulation (similar results were obtained in n = 2 independent experiments). (c) Lcn2 gene expression in KP cells after 24 h of stimulation. Atf4KO cells were infected and selected with two independent CRISPR–Cas9 targeted guides.

Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test. (d) Normalized levels of Lcn2 gene expression in KP LUAD cells that were treated under the indicated conditions for 24 h. Tunicamycin (Tun, 1 ug/mL), mouse interleukin 1 beta (IL1b, 5 ng/mL), cultured in RPMI lacking glutamine (-Q) with or without addition of ISRIB (500 nM) or DMSO as a vehicle. Samples compared to Veh using two-way ANOVA with Šídák’s multiple-comparisons test.

(e-f) Relative gene expression of Lcn2 (e) and Asns (f) after 16 h stimulation of KP cells with IL-1β under glutamine-deprived conditions or in the presence of glutamine (Q). Data were analysed using two-way ANOVA with Tukey’s multiple-comparisons test. Bars represent s.d. (g,h) Relative gene expression after 24 h of stimulation, as in (d) co-treated with ISRIB 500 nM or vehicle in human LUAD A549 (g) and H1299 (h) cell lines.

Data were analysed by a repeated-measures two-way ANOVA with Fisher’s LSD test. (i) Gene expression of KP cells treated with CB-839 (500 nM) and Phenformin (1 mM) for 24 h. (j) Relative levels of Lcn2 gene expression in sorted tumour cells from mice treated with phenformin (O.P. daily, started 72 h prior to cell collection, 100 mg/kg, n = 3) and ISRIB (daily I.P., 2,5 mg/kg, n = 4) and control mice (n = 5).

(k) IHC levels of ATF4-positive cells in tumours isolated from mice treated with phenformin and ISRIB. Data analysed using two-way ANOVA with Dunnett’s multiple-comparisons test. (l) Representative scans of ATF4 IHC staining quantified in (k). (m) ATF4 ChIP–qPCR of KP cells treated with tunicamycin. KP cells were treated with 200 ng/mL of Tunicamycin for 4 h, and binding sites for Lcn2 and two positive controls for ATF4-binding regions (Chac1 and Asns) and a negative control (gene desert).

Data were analysed by a repeated-measures one-way ANOVA with Fisher’s LSD test. (n,o) Correlation between ATF4 signature, NF-κB signature, and LCN2 levels in the TCGA NSCLC dataset. (p–r) Correlation between ATF4 signature and LCN2 levels in TCGA datasets. Correlation plots of patients with lung (p), pancreatic (q), and skin melanoma (r) tumours, with respective LCN2 levels subgroup analysis.

Category box plots are created from LCN2 quartiles (low: ≤25%, high: ≥75%, medium: between 25% and 75%). Data are mean values +/− SEM unless specified otherwise. The illustration in a was created in BioRender. Bossowski, J. (2025) https://BioRender.com/wryu8nk. (a) Quantification of intratumoural Foxp3+/CD4+ levels from Lcn2KO and Lcn2WT KP orthotopic lung tumours. Violin plots show the distribution of values; central lines indicate the median, and dashed lines denote the first and third quartiles.

(b) Quantification of CD8+ TILs in IHC tumours from C57BL/6J mice treated with ISRIB for 72 h and (c) representative IHC image. Violin plots show the distribution of values; central lines indicate the median, and dashed lines denote the first and third quartiles. (d) IHC quantification of TILs in Lcn2KO and Lcn2WT KPC PDAC tumours. (e,f) ExCITE-Seq immune analysis of sorted CD45+ immune-cell clusters from orthotopically transplanted KP LUAD with doxycycline-inducible shLcn2 and shCtr KP tumours 8 days after adding doxycycline and (f) their relative levels.

(g,h) Subcluster analysis of T cell clusters (g) T cell populations with TCR clonality (h) (represented by coloured dots, clones of >3 T cells) (i–k) Flow-cytometry quantification of Foxp3+ CD25+ Treg levels in KP tumours WT, KO, and KO with ectopically expressed Lcn2 (i); doxycycline-inducible shLcn2 KP tumours under 7 and 14 days after adding doxycycline (j); treated with ISRIB or vehicle (k).

(l,m) Flow cytometry of interstitial and alveolar macrophage numbers in lung digests of mice with KP shLcn2 tumours 7 days after addition of doxycycline (Ctr, n = 5; Dox, n = 4). (n) Gating strategy used to quantify interstitial and alveolar macrophages in mouse lungs. (o) Single-cell expression levels of Slc22a17 in ExCITE-Seq of sorted CD45+ immune-cell clusters from orthotopically transplanted KP LUAD and tumour cells (“epithelial” cluster).

(p) Skyline presenting the expression profile of the Slc22a17 gene in a reference population of cell types, Ultra-Low Input (ULI) RNA-seq data. Gene-expression data for Slc22a17 were obtained from the ImmGen RNA-seq Skyline (https://www.immgen.org), accessed on 08/02/2025. (q) GSEA of hallmark IFNγ gene set response in myeloid subclusters from orthotopically transplanted KP LUAD with doxycycline-inducible shLcn2 and shCtr KP tumours 8 days after adding doxycycline.

The statistical analyses were computed using the Wilcoxon rank-sum test, and p-values were adjusted using Benjamini–Hochberg. (r) Cxcl9 and Il6 transcript levels in macrophage clusters in ExCITE-seq analysis of orthotopically transplanted KP LUAD with doxycycline-inducible shLcn2 (n = 2) and shCtr (n = 3) KP tumours 8 days after Lcn2 suppression. The statistical analyses were computed using the Wilcoxon rank-sum test, and p-values were adjusted using Benjamini–Hochberg.

(s,t) Relative levels of Il6 (s) and Cxcl9 (t) in response to LPS in BMDMs with LCN2 receptor Slc22a17 knockdown with and without LCN2 preconditioning. Bars represent SEM with a one-way ANOVA with Fisher’s LSD test. (u) Scheme of experimental set-up of CD8+T cell recruitment in BMDMs/KP cells co-cultures within a 3D matrix as presented in Fig. 3j. Nv – naïve, Tfh: T follicular helper, Treg – T regulatory, eff/ex- effector/exhausted, nv/cm- naïve/central memory, DC – Dendritic cells, cDC – conventional dendritic cells, pDC – plasmacytoid dendritic cells.

Data are mean values +/− SEM unless specified otherwise. The illustration in u was created in BioRender. Bossowski, J. (2025) https://BioRender.com/wryu8nk. (a) Representative image of nearest neighbour analysis in PDAC tissue sample. Each dot represents a single cell annotated as LCN2pos (green), CD8 cell (brown) or neither (grey). (b) Nearest neighbour analysis quantification of distances between cells annotated based on IF staining as LCN2-positive (LCN2pos), LCN2-negative (LCN2neg), CD4+ or CD8+ cells (All_Tcell) and Foxp3+ cells (Treg).

The Wilcoxon rank-sum test was used to analyse the data. (c) Representative images of H&E samples analysed by Lunit SCOPE IO from LCN2-high (top 10%, left) and LCN2-low (bottom 10%, right) samples of TCGA LUAD. Cancer area (CA) and Cancer stromal area (CS) are highlighted in blue and green areas. Tumour cells and lymphocytes are marked by red and yellow dots, respectively. (d) LCN2 and (e) ATF4 expression across immune phenotypes (Inflamed, Immune-excluded, and Immune-desert) in TCGA pan-cancer analysis.

Data analysed using the Wilcoxon rank-sum test. (f) Overall survival of 1195 PDAC cancer samples divided by bulk Lcn2 expression levels into low and high Lcn2 expression groups. (g) Overall survival analysis based on LCN2 expression levels in the SMC lung cancer cohort, excluding the immune-desert phenotype. Kaplan–Meier curves showing differences in overall survival between low and high LCN2 expression groups (HR = 0.73, p = 0.043).

(h) LCN2 expression across immune phenotypes (Inflamed, and Immune-excluded) in the SMC lung cancer analysis. Data analysed using the Mann–Whitney U test. Box plots show the median (centre line), the interquartile range (25th–75th percentiles; box), and whiskers extending to the minimum and maximum values. Data are mean values +/− SEM unless specified otherwise. (a) Individual bioluminescence of tumour progression as depicted in Fig.

5b. (b) Survival of C57BL/6J female mice orthotopically transplanted with KP tumours and treated with anti-mLCN2 antibody (10 mg/kg, biweekly) starting at 10 days after tumour transplantation. Data analysed using the Log-rank (Mantel–Cox) test. (c) Relative bioluminescence signal from C57BL/6J mice tail-vein injected with KP primary LUAD cells and treated with anti-mLcn2 (10 mg/kg, biweekly).

Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. (d–f) C57BL/6J mice treated from day 9 after KP tumour implantation with anti-mouse LCN2 antibody (10 mg/kg biweekly). (d) Mouse weight over the course of the experiment. Ctr n = 7, amLCN2 n = 5. Data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test.

(e) Survival of C57BL/6J mice orthotopically transplanted with KP tumours and treated with anti-mouse LCN2 antibody as in Extended Data Fig. 5c. Ctr n = 6, amLCN2 n = 5. Data analysed using the Log-rank (Mantel–Cox) test. (f) Blood cell count in anti-mouse LCN2 antibody-treated group over pre-treatment initiation (day 0), 3, and 8 days after treatment start. N = 5. (g) Tumour volume of KP PDAC orthotopically transplanted into C57BL/6J female mice and estimated by weekly ultrasound imaging.

Ctr, n = 6, amLCN2 n = 6. Bars represent SEM and data were analysed using a two-way ANOVA with Fisher’s LSD test. (h) Scheme of in vivo platform developed to assess the anti-tumour efficacy of synthetic antibodies against LCN2. (i) Growth of s.c. transplanted KP tumours Lcn2WT n = 8, Lcn2KO n = 10 and Lcn2KO ectopically expressing Human LCN2 (+hLCN2), n = 15. Data were analysed by a repeated-measures two-way ANOVA with Tukey test.

(j) Bioluminescence of KP tumours transplanted into C57BL/6J male mice. KP tumours transduced and selected with genetic constructs of Lcn2 as depicted in (h). Lcn2KO n = 12, hLCN2 n = 12, mLCN2 n = 12. Data were analysed by a two-way ANOVA with Dunnett’s multiple-comparisons test. (k) Individual bioluminescence of tumour progression in mice as depicted in Fig. 5d. (l,m) Growth (l) and final volume analysed with two-tailed t-test (m) of s.c.

transplanted Lcn2KO cells expressing hLCN2 n = 12 and treated with anti-human LCN2 antibody (n = 12, anti-hLCN2, 10 mg/kg, biweekly). Kinetics data were analysed by a repeated-measures two-way ANOVA with Šidák’s multiple-comparisons test. (n) LUMICKS AFS measurements of mLCN2-coated beads binding to BMDMs. Beads coated with an antibody against F4/80 were used as a positive binding control (F4/80).

BMDMs were co-treated with Fab’s specifically binding to either human (α-human) or mouse (α-mouse) LCN2. Data were analysed using one-way ANOVA with Tukey’s multiple-comparisons test. (o,p) Quantification of CD8+ (o) and CD4+ (p) TILs in IHC tumours from C57BL/6J mice from e (Ctr, n = 5; anti-hLCN2, n = 5). Data analysed using a two-tailed t-test. Violin plots show the distribution of values; central lines indicate the median, and dashed lines denote the first and third quartiles.

(q) Normalized levels of CD8 T cell recruitment in 3D matrix embedded with BMDMs/KP cancer-cell co-cultures. anti-mLCN2 antibody treatment abolished T cell infiltration as compared to the Isotype control. (q) includes a subset of normalized data previously shown in Fig. 4j (Welch’s t-test, biological replicates: n = 3). (r) Multiplex Analysis of Cytokines by Luminex xMAP technology in BALF of mice treated with anti-hLCN2 n = 4 and Ctr (n = 5) collected 5 days after treatment initiation.

Data analysed using the Mann–Whitney U test. (s,t) Excerpt of (s) IL-6 and (t) IFNγ BALF levels as presented in (r). Data are mean values +/− SEM unless specified otherwise. The illustration in h was created in BioRender. Bossowski, J. (2025) https://BioRender.com/wryu8nk. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material.

You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/. Bossowski, J.P., Pillai, R., Kilian, J. et al. The integrated stress response promotes immune evasion through lipocalin 2. Nature (2026). https://doi.org/10.1038/s41586-026-10143-0

Summary

This report covers the latest developments in samsung. The information presented highlights key changes and updates that are relevant to those following this topic.

---