Structural surfaceomics reveals an AML-specific conformation of integrin β2 as a CAR T cellular therapy target

Mandal, Kamal; Wicaksono, Gianina; Yu, Clinton; Adams, Jarrett J.; Hoopmann, Michael R.; Temple, William C.; Izgutdina, Adila; Escobar, Bonell Patiño; Gorelik, Maryna; Ihling, Christian H.; Nix, Matthew A.; Naik, Akul; Xie, William H.; Hübner, Juwita; Rollins, Lisa A.; Reid, Sandy M.; Ramos, Emilio; Kasap, Corynn; Steri, Veronica; Serrano, Juan Antonio Camara; Salangsang, Fernando; Phojanakong, Paul; McMillan, Melanie; Gavallos, Victor; Leavitt, Andrew D.; Logan, Aaron C.; Rooney, Cliona M.; Eyquem, Justin; Sinz, Andrea; Huang, Benjamin J.; Stieglitz, Elliot; Smith, Catherine C.; Moritz, Robert L.; Sidhu, Sachdev S.; Huang, Lan; Wiita, Arun P.

doi:10.1038/s43018-023-00652-6

Download PDF

Article
Open access
Published: 30 October 2023

Structural surfaceomics reveals an AML-specific conformation of integrin β₂ as a CAR T cellular therapy target

Nature Cancer volume 4, pages 1592–1609 (2023)Cite this article

Subjects

Abstract

Safely expanding indications for cellular therapies has been challenging given a lack of highly cancer-specific surface markers. Here we explore the hypothesis that tumor cells express cancer-specific surface protein conformations that are invisible to standard target discovery pipelines evaluating gene or protein expression, and these conformations can be identified and immunotherapeutically targeted. We term this strategy integrating cross-linking mass spectrometry with glycoprotein surface capture ‘structural surfaceomics’. As a proof of principle, we apply this technology to acute myeloid leukemia (AML), a hematologic malignancy with dismal outcomes and no known optimal immunotherapy target. We identify the activated conformation of integrin β₂ as a structurally defined, widely expressed AML-specific target. We develop and characterize recombinant antibodies to this protein conformation and show that chimeric antigen receptor T cells eliminate AML cells and patient-derived xenografts without notable toxicity toward normal hematopoietic cells. Our findings validate an AML conformation-specific target antigen and demonstrate a tool kit for applying these strategies more broadly.

The surfaceome of multiple myeloma cells suggests potential immunotherapeutic strategies and protein markers of drug resistance

Article Open access 15 July 2022

The target atlas for antibody-drug conjugates across solid cancers

Article 21 December 2023

Structural insights into Siglec-15 reveal glycosylation dependency for its interaction with T cells through integrin CD11b

Article Open access 13 June 2023

Main

Cellular therapies are one of the most exciting modalities in cancer care¹. However, safely applying these therapies to cancers beyond B cell malignancies has remained clinically challenging². A major hurdle remains in the identification of surface antigens that are specifically expressed on tumor cells but not on other essential tissues, with the goal of minimizing ‘on target, off tumor’ toxicity^3,4.

Recently, we were intrigued by the discovery of an activated conformation of integrin β₇ as a specific cellular therapy in multiple myeloma⁵. This change in protein state led to the opportunity to target the active conformation of integrin β₇ while sparing other normal blood cells, where this protein remained in the closed resting conformation. This finding raised the exciting hypothesis that given aberrancies in tumor signaling, metabolism or cell–microenvironment communication, cancer-specific surface protein conformations may in fact be widespread. However, this result in myeloma was the serendipitous outcome of a hybridoma screen. Thus, here, we aimed to develop a technology to systematically probe this possible untapped source of tumor-specific surface antigens. Specifically, we took advantage of cross-linking mass spectrometry (XL–MS)⁶. Although XL–MS is most often used to define protein–protein interactions or structural constraints⁶, this approach can also yield low-resolution structural information for hundreds or thousands of proteins in a sample^7,8.

However, a major hurdle in XL–MS is the low fraction of cross-linked peptides compared to total peptides in any given sample⁹. Therefore, to focus on cell surface antigens, we combined XL–MS with cell surface capture (CSC), a method to specifically enrich cell surface N-linked glycoproteins¹⁰. We and others have used CSC to successfully identify immunotherapy targets based on surface protein abundance^11,12. Here, by combining XL–MS and CSC in ‘structural surfaceomics’, we aim to move to the next level of protein-centric target discovery.

As an initial proof of principle, we apply structural surfaceomics to acute myeloid leukemia (AML), a frequently diagnosed hematologic malignancy with dismal prognosis¹³. Thus far, chimeric antigen receptor (CAR) T cells in AML have generally led to either notable toxicities or disappointing clinical efficacy^14,15. One major hurdle to CAR T therapy for AML is lack of optimal immunotherapy targets¹². Leading targets include CD33, CD123 and CLL-1/CLEC12A. However, these are all expressed not only on AML blasts but also on normal myeloid cells and/or hematopoietic stem and progenitor cells (HSPCs), driving the potential for major toxicity as well as issues of intratumoral heterogeneity, leading to suboptimal efficacy^12,14,16,17. Thus, there remains a great need to identify AML-specific cellular therapy targets that may eliminate tumor cells while sparing normal myeloid cells.

Here, we apply structural surfaceomics to an AML model and identify the activated conformation of integrin β₂ as a promising immunotherapeutic target. We develop and characterize humanized recombinant antibodies specific to the activated conformation of this protein. We further demonstrate that CAR T cells that target integrin β₂ are efficacious in AML models and, importantly, do not show any evidence of toxicity in normal hematopoietic cells in a humanized immune system (HIS) mouse model, unlike anti-CD33 CAR T cells. In addition, our findings suggest structural surfaceomics as a strategy to unlock a previously unexplored class of immunotherapy targets invisible to standard discovery strategies.

Results

Development of the structural surfaceomics technology

Our overall strategy for structural surfaceomics is to first use a bifunctional chemical cross-linker applied to live cells, followed by glycoprotein oxidation and biotinylation using the CSC strategy (Fig. 1a). Our goal is to ‘freeze’ the native protein conformation in situ, thereby preserving relevant structural information, and use streptavidin-based enrichment of surface proteins to increase MS coverage of our most relevant peptides.

**Fig. 1: XL–MS and surface glycoprotein capture strategy to identify conformation-specific cancer antigens.**

As an initial model system, we used the Nomo1 AML cell line, which is derived from an individual with monocytic leukemia¹⁸. Using Nomo1 cells, we explored two complementary XL–MS strategies. One strategy incorporates the MS-cleavable cross-linker disuccinimidyl sulfoxide (DSSO), which we and others have used frequently to study protein–protein interactions^19,20,21. We also used the recently described non-cleavable cross-linker disuccinimidyl phenyl phosphonic acid (PhoX), which incorporates a phosphonate-based handle for enrichment via immobilized metal affinity chromatography (IMAC)⁹. We applied these strategies in separate experiments to Nomo1 cells using cellular input of 0.4 × 10⁹–5 × 10⁹ cells (Fig. 1a,b).

XL–MS can identify interlinked (type 2; bridging two separate peptides), intralinked (‘loop linked’, type 1; two lysines cross-linked in the same peptide) and monolinked (‘dead end’, type 0; single modified lysine) peptides. Inter- and intralinked peptides could be informative for our strategy, whereas monolinked peptides are not. For DSSO, we used our previously published computational approach¹⁹ to analyze these data and also adapted this strategy to a publicly available version compatible with the Trans-Proteomic Pipeline (TPP)²², called Ving (Extended Data Figs. 1a and 2 and Methods). In our initial DSSO experiment, we enriched cross-linked peptides by size-exclusion chromatography (SEC) alone, whereas in our subsequent experiment, we followed SEC with tip-based, reversed-phase high-pH fractionation (HpHt) to optimize coverage²³. Between these two DSSO experiments, a total of 700 unique interlinked peptides from 236 proteins were identified (Fig. 1c). Of these cross-links, 42.4% mapped to UniProt-annotated membrane-spanning proteins, demonstrating a strong focus on this compartment. The PhoX sample, processed using IMAC and SEC, resulted in 85.3% of total peptides demonstrating a cross-linked lysine (669 unique interlinks, 1,257 loop links and 6,534 uninformative monolinks), derived from 782 proteins (Fig. 1d). Although enrichment for membrane-spanning proteins for PhoX was less than DSSO, at 27.9%, this value was still broadly consistent with our prior studies using CSC alone²⁴. Combining these data, our ‘structural surfaceomics’ approach identified 2,390 total interlinked and intralinked peptides on Nomo1 cells.

Active conformation of integrin β₂ as a potential AML target

We manually compared the cross-linked peptides obtained from structural surfaceomics to published structures in the Protein Data Bank (PDB). In DSSO data, we were particularly intrigued to find several cross-links mapping to the protein integrin β₂ and its heterodimer partner integrin α_L (PDB 5E6R)²⁵. We first noted several intraprotein cross-links within integrin β₂ itself that fit the Cα lysine–lysine distance constraints of the DSSO cross-linker, <20 Å. However, we found four cross-links that did not match the Cα–Cα distance constraint on the available crystal structure, extending to ~38.5 Å between Lys 194 and Lys 196 of the βI domain of integrin β₂ and Lys 305 and Lys 330 on the I domain of integrin α_L^25,26 (Fig. 2a). Notably, the crystal structure appears to represent the inactive, closed form of this integrin heterodimer^25,26. Our XL–MS data suggested that these domains are instead in closer proximity on Nomo1 cells, potentially consistent with the open, active conformation in these AML tumor cells (Extended Data Fig. 3a).

**Fig. 2: Activated integrin β₂ is a conformationally selective antigen in AML.**

This finding was notable as integrin β₂ has been identified on several immune cell types, including monocytes, neutrophils, natural killer cells and T cells^27,28. However, at the protein level, it is known to largely remain in the closed, resting conformation until cellular activation after exposure to appropriate cytokines, adhesion molecules or other proteins^29,30,31,32. Furthermore, a previous study suggested that constitutive signaling through integrin β₂ maintains proliferation in AML blasts³³. Taken together, these results suggest that aberrant AML biology may lead to constitutive activation of integrin β₂, thus creating a possible tumor-specific conformation that, when targeted, would largely spare normal resting hematopoietic cells.

To explore this hypothesis, we took advantage of the mouse monoclonal antibody M24, which is widely used to selectively recognize the activated form of integrin β₂ by flow cytometry³⁴. We confirmed that four AML cell lines of varying genotypes (Nomo1, THP-1, HL-60 and MV-4-11) all showed clear M24 staining, in addition to high levels of total integrin β₂ as detected by the TS1/18 clone (Fig. 2b). By contrast, the B cell malignancy cell lines BV-173 and Namalwa showed total integrin β₂ expression but no discernable expression of the activated conformation (Fig. 2b). To extend this result to normal hematopoietic progenitors, we further obtained granulocyte–macrophage colony-stimulating factor (GM-CSF)-mobilized peripheral blood samples from five hematopoietic stem cell transplant donors at our institution. We found that CD34⁺ HSPCs from these individuals showed no evidence of activated integrin β₂ by flow cytometry (Fig. 2c), although they did express total integrin β₂ (Extended Data Fig. 3b). This result provides an initial suggestion of a favorable therapeutic index for this target.

To further evaluate activated integrin β₂ in primary AML, we obtained deidentified bone marrow aspirate specimens from ten individuals at our institution (Fig. 2d) and two patient-derived xenograft (PDX) models of AML from the Public Repository for Xenografts (PRoXe) biobank³⁵ (Extended Data Fig. 3c). Gating on mature blasts, we found that activated integrin β₂ appeared highly expressed in 9 of 12 total samples analyzed. We further analyzed bulk RNA-sequencing (RNA-seq) data across three AML tumor datasets (The Cancer Genome Atlas (TCGA) and BEAT AML datasets (adult) and the TARGET dataset (pediatric))^36,37,38 and found high blast expression of ITGB2 transcript across AML genotypes as well as potential complementarity with CD33 and IL3RA (CD123) (Fig. 2e and Extended Data Fig. 4a). However, we note that transcript expression alone cannot report whether surface integrin β₂ is in the activated or resting conformation. Toward the safety profile of this target, we evaluated single-cell RNA-seq data in the Human Protein Atlas³⁹. ITGB2 transcript was only detectably expressed in hematopoietic cell types (Fig. 2f), with high expression across the myeloid lineage^31,32. Already, this transcript expression pattern compares favorably with that of other known AML immunotherapy targets (Extended Data Fig. 4b). However, we anticipate that conformation-selective targeting will lead to an additional layer of discrimination between tumor and normal cells not available to other targets.

Characterization of antibody binders for active integrin β₂

We next sought to develop CAR T cells targeted to active integrin β₂ as a proof-of-principle therapeutic for AML. We first explored two commercially available antibody clones to active integrin β₂, M24 (ref. ⁴⁰) and AL57 (ref. ⁴¹). Using the sequence of these antibodies, we designed single-chain variable fragment (scFv) binders and incorporated them into a CD28-based CAR backbone. Although we found no activity for AL57-based scFvs, we did find that both the designs (Vh–Vl and Vl–Vh) of the M24-derived scFv did indeed lead to some Nomo1 cytotoxicity (Extended Data Fig. 5). Here and throughout the study, we also used a previously described anti-CD33 CAR as a positive control¹⁶.

Although this result was promising that CAR T cells targeting integrin β₂ could be developed, these M24-derived CAR T cells showed relatively limited in vitro potency. Furthermore, the M24 framework sequences are fully murine⁴⁰, increasing the potential for immunogenicity in humans. Therefore, we sought to develop alternative CAR T cell designs.

As a first step, we used our previously described antigen-binding fragment (Fab) phage display platform⁴² based on a fully human framework sequence to perform selections versus recombinant integrin β₂ (Fig. 3a). From a library of ~10¹⁰ binders, we identified ten initial hits versus integrin β₂, five of which were validated by biolayer interferometry (BLI) and nonspecific enzyme-linked immunosorbent assays (ELISAs; Fig. 3b,c and Extended Data Fig. 6a–c) to have binding affinities to integrin β₂ in the low-nanomolar range and to lack binding of irrelevant proteins, respectively (Extended Data Fig. 6b,c and Supplementary Table 1). These five Fabs were cloned into a human IgG1 backbone and recombinantly expressed in mammalian cells (Extended Data Fig. 6a). As a validation system, we chose Jurkat T-ALL cells, which we found express high levels of integrin β₂ with a fraction appearing to show constitutive activation at baseline (Fig. 3d). Encouragingly, four of our five recombinant antibodies to integrin β₂ showed positive signal by flow cytometry (Fig. 3d).

**Fig. 3: Antibody 7065 binds preferentially to the active conformation of integrin β₂.**

We next took advantage of the fact that integrins can be biochemically converted from the inactive, closed conformation to the active, open conformation by treatment with the divalent cation Mn²⁺ (ref. ⁴³). Although two clones (7060 and 7062) did not show any responsiveness to 2 mM Mn²⁺ treatment, clones 7065 and 7341 showed increased signal in response to Mn²⁺ (Fig. 3d). Indeed, the profile of 7065 appeared similar to the well-validated antibody M24, with limited signal in the absence of Mn²⁺ but an approximately threefold increase in mean fluorescence intensity after cation exposure. The higher signal from 7341 at baseline suggests that it may also have some binding to the closed conformation of integrin β₂. Flow cytometry of primary AML samples and cell lines with clone 7065 (Extended Data Fig. 6d,e) showed a similar profile as those stained with M24 (Fig. 2d). These findings suggest that clone 7065 may be particularly selective for the activated conformation of integrin β₂.

Development of antiactive integrin β₂ CAR T cells

The sequences of 7065 and 7341 were next engineered into the scFv format and cloned into a CAR backbone with a CD28 co-stimulatory domain (Fig. 4a). For each antibody, we explored two different scFv orientations, either Vh–Vl or Vl–Vh, with a 3× Gly₄-Ser linker. Based on Nomo1 cytotoxicity in vitro, the 7065 Vl–Vh design appeared to be more efficacious (Extended Data Fig. 7a) than control ‘empty’ CAR T cells (CAR backbone but no antibody binder). This 7065 design also showed no discernible activity versus the negative-control AMO-1 cell line, a multiple myeloma cell line that does not express activated integrin β₂ (Extended Data Fig. 7a,b).

**Fig. 4: aITGB2 CAR T cells derived from the 7065 antibody are cytotoxic to AML cells.**

Although these initial in vitro experiments were promising, we did anecdotally observe decreased proliferation and final yield of these CAR T cells during manufacturing. Furthermore, even the best-performing CAR T cell design had moderate Nomo1 cytotoxicity compared to positive-control anti-CD33 CAR T cells (Extended Data Fig. 7a). We hypothesized that T cell stimulation was leading to integrin β₂ activation, and thus some degree of CAR T cell ‘fratricide’, during expansion. To test this hypothesis, we used an approach used for other CAR T cell targets present on activated T cells, such as CD70 (ref. ⁴⁴), where CRISPR–Cas9 ribonucleoprotein (RNP) is used to knock out ITGB2 before T cell stimulation (Extended Data Fig. 7c). Single guide RNA 1 (‘sgRNA-1’) showed no evidence of genomic DNA cleavage at predicted off-target sites (Extended Data Fig. 7e) and was thus used for downstream manufacturing. Using this protocol, we no longer observed any deficit in CAR T cell expansion (Extended Data Fig. 7d), and, furthermore, we observed in vitro cytotoxicity versus Nomo1 and THP-1 cells comparable to anti-CD33 (Fig. 4b). The CAR T cells were also found to have potent degranulation against Nomo1 cells (Extended Data Fig. 7f). As expected, antitumor cytotoxicity correlated with antigen density (Fig. 2b and Extended Data Fig. 7g).

We further varied the Vl–Vh linker length and found largely consistent cytotoxicity (Extended Data Fig. 8a). We thus chose either the ‘3×’ or ‘4×’ linker designs as lead candidates for evaluation. To assess proliferation kinetics of antiactive integrin β₂ (aITGB2) CAR T cells, we performed live-cell imaging assays of Nomo1 cocultures. We found that at a 1:1 effector-to-tumor (E:T) ratio, aITGB2 CAR T cells showed levels of proliferation and cytotoxicity that were similar to those observed with anti-CD33 CAR T cells (Fig. 4c). However, at a 1:10 E:T ratio, aITGB2 CAR T cells outperformed anti-CD33 (Fig. 4c). Both CAR T cells showed similar proliferation in this coculture assay (Fig. 4c). aITGB2 CAR T cells were also found to be efficacious against primary AML samples (Extended Data Fig. 8b) and led to increased release of several cytokines after tumor exposure compared to empty control CAR T cells (Extended Data Fig. 8c). Further characterization based on memory, exhaustion and activation signaling markers revealed favorable features of aITGB2 CAR T cells (Extended Data Fig. 8d–f). Taken together, these findings encourage preclinical investigation of aITGB2 CAR T cells as an AML therapy.

aITGB2 CAR T cells are specific to active integrin β₂

We next evaluated specificity of our CAR T cells for the active conformation of integrin β₂. First, we used our Cas9 RNP strategy to confirm that ITGB2 knockout in Nomo1 cells fully abrogated aITGB2 CAR T cell activity (Fig. 4d,e). To further evaluate conformation specificity, we demonstrated no aITGB2 CAR T cell cytotoxicity versus the B cell leukemia cell line Namalwa, which stains for total integrin β₂ but not M24 (Fig. 2b and Extended Data Fig. 9a). As a second test, in an overnight assay, we incubated green fluorescent protein (GFP)-labeled aITGB2 CAR T cells with normal donor peripheral blood mononuclear cells (PBMCs). At baseline, we found that aITGB2 showed no cytotoxicity in resting CD3⁺ T cells, which are positive for total integrin β₂ only (Fig. 5a,e and Extended Data Fig. 9b). However, with PBMC stimulation using ionomycin, lipopolysaccharide and interleukin-2 (IL-2), we found partial depletion of the GFP^– (that is non-CAR T cell, derived from PBMCs) T cell population (Fig. 5a–c). Indeed, this partial depletion was consistent with the fraction of T cells expressing active integrin β₂ after stimulation (Fig. 5b). These results suggest that aITGB2 CAR T cells specifically eliminate target cells displaying the activated conformation of this protein.

**Fig. 5: Toxicity assessment of aITGB2 CAR T cells demonstrates a promising safety profile.**

aITGB2 CAR T cells appear to have minimal toxicity in HSPCs

As ITGB2 only appears to be expressed in hematopoietic cells (Fig. 2f), we focused further toxicity analysis on these populations. By M24 flow cytometry on peripheral blood, we showed that resting T and B cells did not express active integrin β₂ (Extended Data Fig. 9b). Granulocytes and monocytes appeared strongly positive for active integrin β₂; however, it is well known that this finding is an artifact of ex vivo activation of these cells after blood collection⁴⁵. We reasoned that evaluating potential aITGB2 CAR T cell cytotoxicity in mature myeloid cells would thus require in vivo studies.

In parallel, we also performed clonogenic assays using GM-CSF-mobilized peripheral blood. Consistent with the lack of active integrin β₂ on CD34⁺ HSPCs by flow cytometry (Fig. 2c), we found no toxicity against HSPCs after coculture with aITGB2 CAR T cells (Fig. 5d and Extended Data Fig. 9c). Similarly, in PBMCs, we observed no depletion of T cells (Fig. 5e), consistent with our findings in Fig. 5a. Surprisingly, we saw a modest depletion of CD19⁺ B cells compared to that observed with empty control CAR T cells; however, a similar depletion was also seen with anti-CD33 CAR T cells, arguing against target specificity (Extended Data Fig. 9d). We also tested our CAR T cells versus pathogen-specific T cells, where we found modest depletion only after exogenous activation (Extended Data Fig. 9e,f), consistent with donor T cells (Fig. 5c). As expected, based on known artifactual integrin β₂ activation (Extended Data Fig. 9g), and confirming in vitro potency of aITGB2 CAR T cells versus primary cells, we found strong depletion of monocytes and neutrophils (Extended Data Fig. 9h).

We next moved into a HIS mouse model, where CD34⁺ HSPCs isolated from GM-CSF-mobilized peripheral blood were intravenously implanted into busulfan-treated NOD scid gamma (NSG)-SGM3 mice⁴⁶ (Fig. 5f). Mice were monitored by peripheral blood draw at 8 weeks after implantation to confirm hematopoietic engraftment, as assessed by ≥1.5% circulating human CD45⁺ cells. At this time, we treated all engrafted mice (16 of 25 total implanted) with aITGB2, anti-CD33 or empty CAR T cells, and 6 d later, we killed the mice and analyzed peripheral blood. Although rigorous quantification of CD14⁺ cells was not possible due to high variability in myeloid engraftment at the time of CAR T cell treatment, we found no discernible depletion after treatment with aITGB2 CAR T cells (Fig. 5g). Importantly, we found a significant depletion of total human CD45⁺ cells after treatment with CD33 CAR T cells (Fig. 5h). Long-term monitoring up to 4 weeks found similar results (Extended Data Fig. 9i). These results recapitulated the expected toxicity of targeting CD33 given its expression on HSPCs and myeloid cells. By contrast, human CD45⁺ cells continued to expand in mice treated with either aITGB2 or empty CAR T cells (Fig. 5h).

Furthermore, we probed the 7065 antibody clone and found that it was cross-reactive with mouse activated integrin β₂ (Extended Data Fig. 9j). To further assess toxicity, we thus performed a complete blood count analysis of mouse blood from our HIS mouse study. Five days after aITGB2 CAR T cell treatment, we found no depletion of any mouse PBMC types (Fig. 5i). We also performed an analogous study with a non-tumor-bearing immunocompetent mouse model (C57BL/6), challenged them with mouse aITGB2 CAR T cells and found a similar safety profile (Extended Data Fig. 9k–m). Taken together, these results suggest that treatment with aITGB2 CAR T cells may carry minimal toxicities to bystander immune cells, thus underscoring a promising safety profile.

aITGB2 CAR T cells are efficacious against PDX models of AML

Finally, we evaluated in vivo efficacy of aITGB2 CAR T cells. We established two separate monocytic leukemia PDXs obtained from PRoXe³⁵ via intravenous implantation in NSG mice. Both samples appeared to express active integrin β₂ as determined by flow cytometry with M24 (Extended Data Fig. 3c). Five days after implantation of 2 × 10⁶ PDX AML cells, we administered 5 × 10⁶ empty, aITGB2 or CD33 CAR T cells. Tumor burden was monitored by peripheral blood draw and evaluation of human CD45⁺ cells and/or ultrasonography for spleen size (Fig. 6b,c and Extended Data Fig. 10a,b). In both models, we saw marked elimination of human CD45⁺ cells as well as decreased spleen size in mice treated with aITGB2 or CD33 CAR T cells, with prominent outgrowth of tumor cells in empty control (Fig. 6b,c). In both models, survival was significantly improved in mice treated with aITGB2 CAR T cells compared to those treated with empty control CAR T cells, and survival was similar between mice treated with aITGB2 CAR T cells and those treated with anti-CD33 CAR T cells (Fig. 6a). aITGB2 CAR T cells also showed favorable expansion and persistence properties in the peripheral blood (Extended Data Fig. 10c). In a Nomo1 cell line xenograft mouse model implanted in NSG mice, we again noted improved tumor control over empty CAR T cells as well as similar efficacy of aITGB2 CAR T cells and anti-CD33 CAR T cells (Fig. 6d,e). However, in this aggressive model, neither tested CAR T cells could lead to complete tumor eradication. Toward initial investigation of a possible mechanism of relapse after aITGB2 CAR T cell treatment, we performed flow cytometry on mouse spleens after killing mice at day 42 after tumor implantation. Gating on human CD45⁺ AML blasts, we found no evidence of downregulation or loss of activated integrin β₂ (Extended Data Fig. 10d). This initial experiment suggests that loss of the activated conformation of integrin β₂ may not be an immediate mechanism of resistance to our structurally selective targeting.

**Fig. 6: Efficacy of aITGB2 CAR T cells against AML models in vivo.**

Discussion

Our structural surfaceomics approach presented here, integrating XL–MS with cell surface glycoprotein enrichment, is a technology designed to expand the targetable space of cell surface immunotherapy antigens. This strategy may also carry promise in applications in other basic or translational science fields.

However, we acknowledge that our current methodology carries limitations. The first limitation is sample input. XL–MS has traditionally required large sample inputs (10⁹ cell scale) and extensive mass spectrometer time for analysis. These limitations led us to focus our initial efforts on a single AML cell line with multiple XL–MS approaches. However, future optimization of enrichable cross-linkers, alternative cross-linker reactivities and/or further technological MS advances may enable broader-scale profiling of both tumor and normal cells or even primary samples. The second limitation is the analysis and validation of potential targets. In the current study, we manually compared identified cross-links to PDB structures to find targets of interest. Future work will aim to develop automated computational structural analysis to identify the most promising targets for workup. Additional investigation will be required to evaluate constraints of our method based on surface protein glycosylation or other biophysical parameters. Further work will also clarify the underlying reasons for limited overlap between cross-links found by DSSO and PhoX, which may relate to acquisition on different mass spectrometers, different analysis pipelines and/or different radius of reactivity. In terms of validation, we chose to first investigate integrin β₂ in depth because we had flow cytometry and biochemical (that is, Mn²⁺) tools to probe its conformation status. For other potential targets, these tools will not exist a priori. We thus anticipate future efforts to develop alternative strategies (for example, ‘disulfide locking’, as used in structural biology studies of membrane proteins⁴⁷) to generate putative tumor-selective conformations for recombinant antibody selection and subsequent validation.

The active conformation of integrin β₂ carries promise compared to other AML immunotherapy targets given a potentially improved safety profile. Although we anticipate some unwanted activity toward activated myeloid or T cells, we predict that this toxicity will be markedly lower than CD33, CD123 or CLL-1 that are expressed widely on mature myeloid cells⁴⁸. CD70 is also expressed on activated T cells, but this property has not hindered its clinical development as an AML CAR T cell target (NCT04662294). Our ex vivo results also suggest that depletion of activated T cells may be limited. Future studies in immunocompetent mouse models of inflammation or infection may be useful to evaluate the impact of aITGB2 CAR T cell depletion of activated hematopoietic cells.

In terms of efficacy, like many other AML targets^12,49, we observed heterogeneity of active integrin β₂ on primary human tumor samples. Stratification of individuals by flow cytometry for this target may thus be required for any future clinical studies. Furthermore, PDX-B, despite higher aITGB2 expression, relapsed more rapidly than PDX-A after either aITGB2 or CD33 CAR T cell treatment. Future PDX modeling will further delineate the relationship between antigen density and in vivo aITGB2 CAR T cell efficacy. The favorable safety profile of aITGB2 CAR T cells may also create future opportunities for multitargeting CARs targeting two or more antigens with complementary but heterogeneous tumor expression patterns. Future antibody engineering efforts, or incorporation of emerging engineered T cell antigen receptor designs⁴⁴, may be able to enhance the efficacy of aITGB2 CAR T cells targeting tumors with low antigen levels.

In conclusion, our studies demonstrate a potential systematic approach to identify and target conformation-specific antigens in cancer. Humanized aITGB2 CAR T cells, discovered via this approach, stand as a promising proof-of-principle therapeutic warranting further preclinical evaluation in AML and a pathway for other applications of structurally directed immunotherapeutic targets.

Methods

Ethics statement

All primary AML samples used were obtained under Institutional Review Board-approved protocols by the University of California San Francisco (UCSF), Committee on Human Research and following the Declaration of Helsinki. Informed consent was obtained from participants for research purposes, although for this study, all samples were fully deidentified and could not be linked back to relevant clinical information, including participant sex. All mouse experiments were conducted in accordance with an approved protocol by the UCSF Institutional Animal Care and Usage Committee. Mice were housed in the UCSF Animal Care Facility Laboratory Animal Resource Center at the Helen Diller Family Cancer Center at UCSF Mission Bay. Animals were housed in an individual specific pathogen-free suite with up to five mice per ventilated cage and ad libitum access to food and water. The suite was maintained on a 12-h light/12-h dark cycle with controlled temperature (~19–23 °C) and humidity (30–70%) conditions.

Cell lines, PDX and human samples

Nomo1 and BV-173 cell lines were obtained from DSMZ, and THP-1, HL-60, MV-4-11, Jurkat, U937, Namalwa and S49.1 cell lines were obtained from ATCC. All cell lines were grown in RPMI-1640 medium (Gibco, 11875093) supplemented with 20% fetal bovine serum (FBS; BenchMark, Gemini, 100-106) and 100 U ml^–1 penicillin–streptomycin (UCSF Cell Culture Facility). All cells were grown at 37 °C with 5% CO₂. All AML PDXs were procured from the PRoXe at Dana–Farber Cancer Institute under an appropriate materials transfer agreement. Primary AML samples were obtained from the UCSF Hematologic Malignancies Tissue Bank and the Pediatric Hematopoietic Tissue Cell Bank.

Cross-linking and cell surface labeling

DSSO-based (Sigma-Aldrich, 909602) XL–MS involving HpHt and PhoX-based (Thermo Fisher Scientific, A52286) XL–MS were each performed with 2.4 × 10⁹ cells (in batches of 6 × 10⁸). Cells were collected and washed (300g, 5 min) three times with PBS. Cross-linker DSSO or PhoX predissolved in DMSO (Sigma-Aldrich, 276855) was then added to the cells at a final concentration of 10 mM and incubated at room temperature for 45 min. Cross-linking was followed by biotinylation of cell surface proteins. Briefly, N-linked sugar residues of the cells were oxidized with 1.6 mM sodium metaperiodate (VWR, 13798-22) for 20 min at 4 °C, followed by installation of biotin on those residues using 10 mM aniline (Sigma-Aldrich, 242284) and 1 mM biocytin hydrazide (Biotium, 90060) for 90 min at 4 °C. The cells were then washed with PBS, snap-frozen and stored at −80 °C until further processing.

Cell surface proteomics sample preparation

Frozen cell pellets were thawed on ice and resuspended in 1 ml of RIPA lysis buffer (MilliporeSigma, 20-188) with Halt protease inhibitor (Thermo Scientific, 78430) and 1 mM EDTA (Invitrogen, 15575-038), followed by sonication for lysis. Lysates were centrifuged at 17,000g for 10 min at 4 °C, and clarified supernatant was mixed with 0.5 ml of Neutravidin beads (Thermo Scientific, PI29204), followed by incubation at 4 °C for 2 h in an end-to-end rotor. Beads were washed extensively by vacuum manifold (Promega) with 50 ml of RIPA lysis buffer + 1 mM EDTA, 50 ml of PBS + 1 M NaCl and 50 ml of 2 M urea (VWR, 97063-798) + 50 mM ammonium bicarbonate. Beads were resuspended in 50 mM Tris (pH 8.5) + 4 M urea + 10 mM TCEP (Gold Biotechnology, TCEP10) and 20 mM iodoacetamide (VWR, 97064-926). Ten micrograms of trypsin-LysC (Thermo Scientific, PRV5073) mix was added for on-bead digestion with simultaneous reduction and alkylation. After 2 h, the mixture was diluted to 1.5 M urea using 50 mM Tris (pH 8.5) and incubated overnight (16–20 h). Beads were eliminated by centrifugation, and the resulting supernatant was acidified with 0.5% trifluoroacetic acid (TFA). Peptides were desalted using a SOLA HRP Column (Thermo Scientific, 60109-001) and eluted with 50% acetonitrile (ACN) + 0.1% formic acid (FA).

IMAC purification for PhoX

Dry peptides were reconstituted in 80% ACN + 0.1% TFA. Superflow Ni-NTA beads were stripped off using EDTA and reloaded with FeCl₃ (Sigma-Aldrich, 451649) on a polyprep chromatography column (Bio-Rad, 7326008). Fe³⁺-loaded beads were transferred to C18 tips (Nest Group, SEM SS18V.25) and incubated for 4–6 min, followed by washing with 0.5% FA. The bound peptides were eluted from beads with 0.5 M potassium phosphate buffer (pH 7.4). Peptides eluted from beads were bound to C18 chromatographic material of the Nest tips, washed three times with 0.5% FA and eluted with 50% ACN + 0.1% FA.

SEC

Size-based fractionation of peptides was performed using a Superdex Peptide 3.2/300 (GE Healthcare) column and high-performance liquid chromatography (HPLC; Agilent 1260 Infinity II). Dried peptides were reconstituted in the mobile phase of 30% ACN + 0.1% TFA and loaded on the column. Run time was 90 min at a flow rate of 50 µl min^–1, and 45 fractions (2 min per fraction) were collected. Fractions associated with the desired molecular weight were dried down in a SpeedVac and stored at −80 °C for MS analysis.

LC–MS and data-dependent acquisition analysis

Peptide samples prepared for building the Nomo1 cell surfaceome custom database were loaded on to the EASY-Spray nanocolumn (Thermo Scientific, ES900) installed on a Dionex Ultimate 3000 NanoRSLC instrument coupled with a Q-Exactive Plus mass spectrometer (Thermo Scientific). Peptides were separated primarily over a 313-min gradient ranging from 2.4% to 32% ACN with a flow rate of 0.3 μl min^–1. MS scans were performed from m/z 299 to 1,799 at a resolution of 70,000 full-width at half-maximum (FWHM) at m/z 200 with an automatic gain control (AGC) target of 3 × 10⁶ and maximum injection time of 100 ms. The resolution for MS/MS scans was set to 17,500 FWHM at an m/z of 200 with an AGC target of 2 × 10⁵ and maximum collision-induced dissociation of 200 ms. Normalized collision energies of 27%, 30% and 33% in stepped higher collision-induced dissociation mode were used for fragmentation of the top 15 most intense precursor ions with an isolation window of 1.7 m/z. Peptides with a charge state of 2⁺ or higher were considered for MS/MS. Dynamic exclusion was set to 20 s.

MS-generated .raw files were processed using MSFragger⁵⁰ within FragPipe v14.0 with default settings unless stated otherwise. Briefly, the spectral data were searched against the human proteome database (UniProt, downloaded 11 May 2021; 20,395 entries). The contaminant and decoy protein sequences were added to the search database using the inbuilt feature of the FragPipe v14.0 pipeline downstream statistical analysis. The inbuilt tools PeptideProphet and ProteinProphet were used for statistical validation for 1% false discovery rate (FDR).

HpHt-based fractionation of DSSO cross-linked peptides

The SEC fractions 13 and 14, which are enriched with DSSO cross-linked peptides (Extended Data Fig. 1b), were further fractionated by HpHt as described previously²³. Briefly, the HpH tip was constructed in a 200-µl pipette tip by packing C8 membrane (Empore 3M) and 5 mg of C18 solid phase (3 μm; Durashell, Phenomenex). The column was sequentially washed with three different solvents/solutions: methanol, ACN and ammonia water (pH 10; 90 µl each). Samples were loaded on the column, followed by washing with 90 µl of ammonia water (pH 10) and a series of ammonia water containing increasing concentration of ACN (6, 9, 12, 15, 18, 21, 25, 30, 35 and 50%). The fractions with 25, 30, 35 and 50% ACN were combined with fractions containing 6, 9, 12 and 21% ACN, respectively, for MS analysis.

LC–MS³ analysis of DSSO cross-linked peptides

The SEC–HpHt fractions were subjected to LC–MS³ analysis using an UltiMate 3000 RSLC nano-HPLC system coupled to an Orbitrap Fusion Lumos mass spectrometer (Thermo Scientific), as described previously²³. Peptides were separated by reversed-phase LC (50 cm × 75 μm Acclaim PepMap C18 column, Thermo Scientific) with over an 87-min gradient of ACN (4% to 25%) at a flow rate of 300 nl min^–1. Initial survey (MS¹) scans were measured in the Orbitrap with a scan range from 375 to 1,800 m/z, a resolution of 60,000 FWHM and an AGC target of 4 × 10⁵ with a maximum injection time of 75 ms at top speed per 4 s of cycle time. Ions with a charge of 4⁺ or greater were selected for MS² and subjected to fragmentation using collision-induced dissociation with a normalized collision energy of 23%. For MS² scans, the scan range was set to auto mode, with a resolution of 30,000 FWHM, an AGC target of 5 × 10⁴, a precursor isolation width of 1.6 m/z and a maximum injection time of 100 ms. A targeted inclusion on ions with a mass difference corresponding to the difference in alkene and thiol DSSO fragments (31.9721 Da) was used to select precursors for MS³ analysis. For MS³ scans, higher collision-induced dissociation was used with a normalized collision energy of 28%, the AGC target was set to 2 × 10⁴, and the maximum injection time was set to 125 ms.

Identification of DSSO cross-linked peptides

Peak lists were extracted from the LC–MSⁿ raw files using the in-house software PAVA (UCSF), and the extracted MS³ spectra were searched against a SwissProt database (2021.10.02 version; 20,387 entries) concatenated with its randomized decoy sequences using Protein Prospector (v.6.3.5). The mass tolerances allowed were ±20 ppm for precursor ions and 0.6 Da for fragment ions. The database search was performed with trypsin as a protease with a maximum of three allowed missed cleavages. Cysteine carbamidomethylation was set as the fixed modification. Variable modifications included N-terminal protein acetylation, methionine oxidation and N-terminal conversion of glutamine to pyroglutamic acid. Additionally, three specific modifications resulting from DSSO were included in the search: thiol (C₃H₂SO, +86 Da), alkene (C₃H₂O, +54 Da) and sulfenic acid (C₃H₄O₂S, +104 Da)¹⁹. The in-house software XL-Tools was used to automatically identify, summarize and validate cross-linked peptides based on Protein Prospector database search results and MSⁿ data. No decoy hits were found after the integration of MS¹, MS² and MS³ data.

Development of the MS³-based XL–MS analysis tool

We developed Ving to assess the MS²/MS³-based cleavable cross-linking database search results to produce a set of cross-linked spectrum matches (CSMs; Extended Data Fig. 1a). Ving is open-source, publicly available software and conceptually based on our previously published search methodology¹⁹ but now adapted to the proteomics analysis suite, the TPP. Ving input consists of spectral data in mzML format⁵¹ and database search results in PepXML format⁵².

Ving functions by parsing data to create spectral groups (SGs) consisting of MS² and MS³ data originating from a single precursor ion. Next, database search results from the MS² and MS³ scan events, performed using the TPP²² as described previously, are added to each SG. A series of thresholds categorize each SG to determine probable CSMs after peptide sequence assignments to all MS² and MS³ spectra within all groups. First, MS² peptide sequence assignments with a probability of >0.8 are labeled as single, non-linked peptide spectrum matches (PSMs). Evidence of an internal lysine residue with a hydrolyzed cross-linker mass refines the classification to dead end or monolinked PSMs. For the remaining SGs, the MS³ peptide assignments and probabilities are evaluated. If the SG has multiple MS³-level peptide sequence identifications with a probability of >0.8 and a cross-linker-modified lysine residue, those sequences are further evaluated as candidate CSMs. If two peptide sequence masses plus the cross-linker mass sum together to match the mass of the original precursor ion, then the group is classified as a CSM. If none of the peptide sequences sum to the precursor mass, then the SG is classified as an incomplete CSM. If the SG has one or zero MS³-level peptide sequences with a probability of >0.8, the group is classified simply as unknown PSMs. After all SGs are evaluated, a human-readable text summary is reported to the user.

LC–MS analysis of PhoX cross-linked peptides

PhoX cross-linked peptide samples were analyzed on a timsTOF Pro mass spectrometer (Bruker Daltronics) as described previously⁵³. Briefly, peptides from each SEC fraction 9–24 (Extended Data Fig. 1c) were loaded on to the column operated using an UltiMate 3000 RSLC nano-HPLC system (Thermo Scientific), and eluted peptides were analyzed with the timsTOF Pro mass spectrometer using a CaptiveSpray source (Bruker Daltonics). Peptides were first trapped on a C18 precolumn (Acclaim PepMap 100, 300 μm × 5 mm, 5 μm, 100 Å; Thermo Scientific), and eluted peptides were subsequently separated on a μPAC 50 column (PharmaFluidics) over 180 min with an ACN gradient ramping up from 3% to 35%, during which the flow rate changed from 900 to 600 nl min^–1 for the first 15 min, followed by a constant flow rate of 600 nl min^–1.

For MS analysis with the timsTOF Pro mass spectrometer, the mobility-dependent collision energy ramping settings were 95 eV at an inversed reduced mobility (1/k₀) of 1.6 V s^–1 cm^–2 and 23 eV at 0.73 V s^–1 cm^–2. Collision energies were interpolated linearly between the two 1/k₀ values and were kept constant above or below. TIMS scans were not merged, and the target intensity per individual parallel accumulation serial fragmentation precursor ion was kept at 20,000 (with an intensity threshold of 1,000). The mobility range of each scan was kept between 0.6 and 1.6 V s^–1 cm^–2 with a ramp and accumulation time of 166 ms, and the mass range for MS and MS/MS was set to 100–1,700 m/z. The number of parallel accumulation serial fragmentation MS/MS scans triggered was 14 per cycle (2.57 s), with a maximum of seven allowed precursors per mobilogram. The precursor ion charge states selected for fragmentation ranged between 3⁺ and 8⁺, and the isolation width was 2 Th for precursor m/z up to 700 and 3 Th for precursor m/z > 800, in between which it was ramped linearly. The active exclusion was set to 0.4 min (mass width of 0.015 Th and 1/k₀ width of 0.015 V s^–1 cm^–2).

TimsTOF MS data analysis

TimsTOF MS data were converted to .mgf format using MSConvert⁵⁴. The .mgf files were then processed for identification of cross-linked peptides using pLink-2 (ref. ⁵⁵) with default settings unless stated otherwise. All files were searched against the Nomo1 cell surfaceome-specific custom database (5,280 entries) generated from regular data-dependent acquisition analysis based on prior CSC data (that is, non-cross-linked). For pLink-based cross-linked peptide analysis, trypsin was set as the protease, allowing three missed cleavages. Cysteine carbamidomethylation was set as the fixed modification with methionine oxidation and N-terminal acetylation as the variable modifications. The search was performed with a ±20 ppm mass tolerance window for precursor and fragment ions, and results were reported at 1% FDR.

Flow cytometry

Immunostaining of cells was performed as per the instructions from the antibody vendor unless stated otherwise. Briefly, 1 × 10⁶ cells were resuspended in 100 µl of FACS buffer (PBS + 2% FBS) with 1 µg of antibody added. Cells were incubated at 4 °C for 10–15 min and washed three times with FACS buffer. For staining the active form of integrin β₂, the antibody incubation step was performed at 37 °C for 1 h. For staining primary AML cells for activated integrin β₂, the FACS buffer was RPMI-1640 + 5% FBS + 2% bovine serum albumin (BSA) + 50 µg ml^–1 DNase I (Gold Biotechnology, D-301-500). For other primary antibody cell stainings, FACS buffer was D-PBS + 5% FBS + 2% BSA + 5 mM EDTA + 50 µg ml^–1 DNase I with Human Trustain (Biolegend, 422302). Compensation used UltraComp eBeads Compensation Beads (Invitrogen, 01-2222-42). All antibodies were diluted 1:20 unless stated otherwise. Flow cytometry analysis was performed on the CytoFLEX platform (Beckman Coulter), and data were analyzed using FlowJo_v10.8.1.

Phage display selections

A synthetic, phage-displayed Fab library⁴² was selected for binding to either integrin β₂/integrin α_M (R and D 4047-AM, antibodies 7062 and 7065) or integrin β₂/integrin α_L (R and D 3868-AV, antibodies 7060 and 7341) recombinant protein complexes. Briefly, integrin β₂ recombinant protein complexes were immobilized on Maxisorp Immuno plates (Thermo Fisher, 12-565-135) and used for positive binding selections with library phage pools that were first exposed to neutravidin-coated wells to deplete nonspecific binders. After four rounds of binding selections, clonal phage was prepared and evaluated by phage ELISA and sequencing as previously described⁴².

Antibody production

Antibodies were produced using the human Expi293 expression system (Thermo Fisher). Expi293 cells (in a volume of 2 ml) were transiently transfected with construct DNA using FectoPro transfection reagent (Polyplus Transfection, 101000014). Following a 5-d expression period, antibodies were purified using rProteinA Sepharose (GE Healthcare) and stored in phosphate buffer (50 mM NaH₂PO₄, 75 mM Na₂HPO₄, 100 mM H₃PO₄ and 154 mM NaCl).

BLI binding assays

Binding of human integrin β₂ antibodies was tested against three different integrin β₂ complexes, including integrin β₂/integrin α_M (R and D 4047-AM), integrin β₂/integrin α_X (R and D 5755-AX) and integrin β₂/integrin α_L (R and D 3868-AV). To determine binding kinetic parameters, BLI was performed on an Octet HTX instrument (Sartorius) at 1,000 r.p.m. and 25 °C. All proteins were diluted in assay buffer (PBS, 1% BSA and 0.05% Tween 20). Tested and negative-control antibodies at 2 µg ml^–1 were first captured on AHQ biosensors to achieve binding signals of 0.8–1.3 nm. Unoccupied Fc-binding sites on the antibody-coated sensors were subsequently quenched with 20 µg ml^–1 Fc protein. After equilibration with assay buffer, biosensors were dipped for 600 s into wells containing a fivefold serial dilution of integrin β₂ complexes (association phase), followed by transfer back into assay buffer for an additional 600 s (dissociation phase). Assay buffer alone served as a negative control. Binding response data were reference subtracted and globally fitted with a 1:1 binding model using ForteBio’s Octet Systems software v9.0.

Nonspecific ELISA panel

The ELISA protocol to assess interactions between the antibodies and unrelated macromolecules was performed as described previously⁵⁶. The tested antigens included cardiolipin (50 μg ml^–1; Sigma, C0563), keyhole limpet hemocyanin (5 μg ml^–1; Sigma, H8283), lipopolysaccharide (10 μg ml^–1; InvivoGen, tlrl-eblps), single-stranded DNA (1 μg ml^–1; Sigma, D8899), double-stranded DNA (1 μg ml^–1; Sigma, D4522) and insulin (5 μg ml^–1; Sigma, I9278). In addition, binding of each antibody was also tested against empty wells (BSA-only control) and wells containing goat anti-human Fc (positive control, 1 µg ml^–1; Jackson, 109-005-098). Antigens were coated at 30 µl per well in 384-well Maxisorp plates and incubated at 4 °C overnight. Plates were blocked with 0.5% BSA for 1 h at room temperature and washed with PBS + 0.05% Tween 20. Antibodies were added at 100 nM and allowed to bind for 60 min at room temperature. Plates were washed with PBS + 0.05% Tween 20, and binding was detected with anti-κ-horseradish peroxidase (1:5,000; Southern Biotech, 2060-05) and developed with TMB substrate (KPL (Mandel), KP-50-76-03).

Lentiviral construct generation and production

Second-generation lentivirus constructs were used for transducing CAR expression cassettes in human T cells. Packaging plasmids used were pCMV delta R8.2 and pMD2.G. The transfer vector carrying the CAR expression cassette was pHR lentiviral vector with the SFFV promoter.

All transfer lentiviral plasmid constructs were generated using NEBuilder HiFi DNA Assembly master mix (New England Biolabs, E2621L) as per the vendor’s instructions with minor modifications. The DNA fragments containing the binder (scFv) sequence along with the 40-base pair vector-compatible flanking region for Gibson assembly were procured from Twist Bioscience. Target CAR plasmid backbone was linearized with BamHI-HF (New England Biolabs, R3136T). Stbl3 competent Escherichia coli cells (QB3 MacroLab, University of California, Berkeley) were used for transformation, and colonies obtained were screened by Sanger sequencing (Genewiz).

Lenti-X cells cultured in DMEM and plated 1 d before in a six-well plate (1 million cells per well) were transfected with the lentiviral plasmid constructs using polyethylenimine (Polysciences, 40,000 Da molecular weight, 24765-100). The spent media in cultures containing the lentivirus were collected 72 h after transfection. Twelve microliters of the stock concentration of polyethylenimine (2.5 mg ml^–1) was used for transfecting 3 µg of plasmid per well. In total, 1.35 µg, 0.165 µg and 1.5 µg of lentiviral plasmids pCMV delta R8.2, pMD2.G and pHR (transfer vector), respectively, were used per well.

Human primary T cell isolation

Primary T cells were isolated from LeukoPaks (Stem Cell Technologies, 200-0092). CD8⁺ and CD4⁺ cells were isolated separately using an EasySep human CD8⁺/CD4⁺ T cell isolation kit (StemCell, 17952 (CD4); StemCell, 17953 (CD8)) based on magnetic bead separation. Isolated cells were stored frozen with 10% DMSO (MP Biomedicals, 196055). In total, primary T cells from five different donors were used for in vitro and in vivo studies.

Human CAR T cell generation

T cells were thawed and grown in T cell medium consisting of Optmizer CTS medium (Gibco, A10221-01) + CTS supplement (Gibco, A10484-02) + 5% human AB serum (Valley Biomedical, HP1022) + penicillin/streptomycin + GlutaMAX (Gibco, 35050-061). Recombinant human IL-7 (Peprotech, 200-07) and IL-15 (Peprotech, 200-15; a final concentration of 10 ng ml^–1 each) were freshly added to cells every 2–3 d. For manufacturing CAR T cells, primary T cells (CD4⁺ or CD8⁺) were thawed and cultured overnight. For aITGB2 CAR T cells, the cells were then additionally nucleofected with ribonuclease complex of ITGB2 sgRNA and Cas9 using a P3 Primary Cell 4D-Nucleofector X kit S (Lonza, V4XP-3032) and 4D-Nucleofector (Lonza) with its built-in program EO-115. Cells were then stimulated with 20 µl of CD3/CD28 Dynabeads (Thermo Fisher Scientific, 11131-D) per 1 million cells on day 0. The following day, lentivirus carrying the CAR expression cassette was added to the cells. The virus was withdrawn from the culture after 24 h, followed by two to three rounds of PBS washes by centrifugation. Dynabeads were magnetically withdrawn on day 4. On day 6 or 7, CAR⁺ cells were sorted by magnetic-activated cell sorting using the Myc tag of the CAR constructs with biotinylated c-Myc antibody (Milteni Biotec, 130-124-877). CAR T cells were used for studies within days 10–14 of culture.

CAR T cell cytokine analysis

CAR T cells were cocultured with target (tumor) cells at a 1:1 E:T ratio for 24 h, and supernatant was snap-frozen in liquid nitrogen. Samples were analyzed at Eve Technologies Corporation with a Luminex 200 system (Luminex) using a Human High Sensitivity 14-Plex Discovery Assay (MilliporeSigma) according to the manufacturer’s protocol.

Retrovirus production and transduction

HEK293T cells (3.5 × 10⁶) were seeded in a 10-cm dish. Medium was replaced with 5 ml of complete DMEM (DMEM supplemented with 10% fetal bovine serum, 2 mM L-glutamine), and cells were transfected with 7.5 μg of pCL-ECO plasmid and 7.5 μg of MSCV plasmid using Lipofectamine LTX with Plus reagent (Invitrogen, 15338030). The transfection mix was prepared in 3 ml of Opti-MEM medium (Gibco, 31985062) and incubated for at least 30 min at room temperature before being added dropwise onto the cell culture. Twenty-four hours after transfection, the medium was exchanged for 6 ml of complete DMEM collection medium. Retrovirus was collected, sterile filtered and frozen at −80 °C for storage at 24 and 48 h.

Mouse T cell isolation and culture

Spleens from mice were crushed and strained, and T cells were isolated using an EasySep mouse T cell isolation kit (StemCell Technologies, 19851). Cells were cultured in RPMI-1640 (Gibco, 11875093) supplemented with FBS (10%), penicillin–streptomycin (100 U ml^–1), sodium pyruvate (1 mM), HEPES (10 mM), β-mercaptoethanol (Gibco, 21985-023), MEM non-essential amino acids (1×; Gibco, 11140050) and human IL-2 (200 U ml^–1; Peprotech, 200-02).

Mouse CAR T cell generation

The scFv of the 7065 antibody or human CD19 scFv was cloned in the mouse CAR backbone (MSCV plasmid) with N-terminal mouse CD8a signal peptide and C-terminal mouse CD28–mouse CD3z domain, with P2A sequence followed by Thy1.1. Mouse T cells were cultured and activated for 24 h using Dynabeads Mouse T-Expander CD3/CD28 (Gibco, 11452D) and magnetically removed thereafter. T cells (2 × 10⁶) were nucleofected with Cas9 ribonuclease complex (RNP; Lonza, V4SP-3096) using a 4D-Nucleofector 96-well unit (Lonza, AAF-1003S) and Lonza program code DN-100 with 60 pmol of Cas9 protein (QB3 MacroLab) and 120 pmol of sgRNA (UCCCUCCUCUAGAACUUCAC; Synthego) preincubated at 37 °C for 10–15 min. The culture medium was then added to the cells and incubated (37 °C, 5% CO₂) for 1 h. Retronectin-coated (Takara, T100B) plates were used for culture, and 1 ml of retrovirus carrying the CAR expression cassette was added with 10 μg ml^–1 polybrene. Cells were spinfected (2,000g, 30 °C, 60 min) and incubated overnight in a CO₂ incubator at 37 °C. Medium was replenished regularly, and cell density was maintained at approximately 2 × 10⁶ cells per ml.

T cell activation assay

PBMCs were treated with 3 µM ionomycin (Sigma-Aldrich, 407950) + 25 ng ml^–1 lipopolysaccharide (Sigma-Aldrich, L4391) + 100 U ml^–1 IL-2 (Prospec, CYT-209) and cultured overnight. Cells were then co-stained with CD3 and CD69 and analyzed by flow cytometry.

In vitro cytotoxicity assay

AML cell lines were engineered to stably express luciferase using lentiviral transduction. The cell lines were cocultured overnight with CAR T cells. d-Luciferin (150 μg ml^–1; Gold Biotechnology, LUCK-1G) was added to each well and incubated for 3–5 min at room temperature, followed by luciferase detection using GloMax Explorer (Promega). For each ratio (CAR T cells:tumor cells), the bioluminescence readings from the tumor cells cocultured with untransduced T cells were considered 100% viable for normalization.

Degranulation assay

CAR T cells were cocultured with tumor cells at ratio of 2:1 for 6 h at 37 °C with anti-CD107a and GolgiStop (BD Biosciences, 51-2092KZ). Cells were washed twice by centrifugation at 500g for 5 min at room temperature. Levels of CD107a were then measured with a flow cytometer as a readout of degranulation on GFP⁺ CAR T cells.

Generation of ITGB2-knockout cells

Knockout cell lines or primary T cells were generated using in vitro nucleofection of Cas9 ribonuclease protein complex. Briefly, 2 μl of each sgRNA (100 µM; Synthego Corporation) and recombinant Cas9 protein (40 µM; QB3 MacroLab, University of California, Berkeley) was incubated at 37 °C for 15 min to generate ribonuclease complex, which was then nucleofected using a 4D-Nucleofector (Lonza) with the built-in program DS-137 for cell lines (using Lonza V4XC-2032) and EO-115 for primary T cells (using Lonza V4XP-3032). The sgRNAs used in this study were obtained from the Brunello library⁵⁷ (Supplementary Table 3).

Off-targeting analysis of sgRNA

Potential off-target cleavage analysis was determined using the online tool CRISPOR⁵⁸. The top five potential hits (genomic loci) were selected for PCR amplification (for primers, see Supplementary Table 2) from genomic DNA of ITGB2-knockout and wild-type primary human T cells extracted using a Monarch Genomic DNA Purification kit (New England Biolabs, T3010S). PCR amplicons were sequenced, and data were analyzed using the Synthego ICE Analysis Tool to determine the percentages of insertions and deletions.

Clonogenic assay

CD34⁺ cells (1 × 10³) from healthy donor GM-CSF-mobilized peripheral blood were co-incubated with aITGB2 CAR T cells, My96 CAR T cells, empty CAR T cells, untransduced T cells or medium only (IMDM, 2% FBS and penicillin/streptomycin) at an E:T ratio of 1:1 for 5 h in V-bottom 96-well plates in triplicate in methylcellulose-based medium supplemented with recombinant cytokines (MethoCult H4434 Classic, StemCell Technologies). After 13–14 d, colonies were classified and counted as granulocytes, erythrocytes, monocytes, megakaryocytes; granulocytes, monocytes; granulocytes; monocytes or erythrocytes. Images were acquired with a Keyence microscope using ×10, brightfield and color mode.

Generation of pathogen-specific T cells

CD45RA^– healthy donor PBMCs were isolated using MACS MicroBead Technology (Miltenyi, 130-045-901) and pulsed with a pool of overlapping peptide libraries (15 mers overlapping by 11 amino acids) spanning the entire protein sequences of EBNA1, LMP1 and LMP2; BZLF1 and BRLF1 of Epstein–Barr virus; IE and pp65 of cytomegalovirus; hexon and penton of adenovirus and large T and VP1 or BK virus (JPT technologies) and expanded with IL-7 (10 ng ml^–1) and IL-15 (5 ng ml^–1; R&D Systems, 207-IL-010 and BT-015-010). On day 9, the cells received a second stimulation with irradiated pepmix-pulsed autologous ATCs pulsed with the same peptide libraries and irradiated HLA^– lymphoblastoid cell lines (ULCLs) at a EBVST:ATC:K562cs/ULCL ratio (EBVST, Epstein-Barr virus-specific T cells; ATCs, activated T cells) of 1:1:5 with IL-7 and IL-15. Expanding cells were split as required and cryopreserved on day 16 of culture. Aliquots were thawed for analysis.

Generation of HIS mice

The NSG-SGM3 strain (NOD.Cg-Prkdc^scidIl2rg^tm1Wjl Tg(CMV-IL3,CSF2,KITLG)1Eav/MloySzJ) obtained from Jackson Laboratories was used to generate HIS mice. All mice used were 6–8 weeks old (either all male or all female for a particular study). Each mouse was treated with busulfan (12.5 mg per kg (body weight)) for 2 d consecutively, followed by 1 d of recovery and injection of 7 × 10⁵ CD34⁺ human hematopoietic cells intravenously through the tail vein. Fully deidentified human CD34⁺-cell enriched blood samples were obtained from the Bone Marrow and Transplantation Laboratory at UCSF and were sorted by magnetic-activated cell sorting using a CD34 MicroBead kit (Miltenyi Biotec, 130-046-702) and incubated with anti-CD3 (Biolegend, 317302, clone OKT3) for T cell depletion 10 min before injection. Flow cytometry for human CD45⁺ cells was performed on mouse blood samples drawn 8 weeks later to determine engraftment efficiency (>1.5% threshold).

In vivo CAR T cell efficacy in PDX models of AML

All mice used in the experiments were 6–8 weeks old (either all male or all female for a particular study) and were obtained from either Jackson Laboratories (NSG-SMG3) or were bred in-house (NSG) at the Preclinical Therapeutics Core of UCSF. In total, 1 × 10⁶ Nomo1 cells were injected in each mouse. For PDXs, 2 × 10⁶ cells were injected in each mouse irradiated with 250 cGy 4–6 h before injection. In total, 5 × 10⁶ CAR T cells at a 1:1 ratio of CD4⁺:CD8⁺ CAR T cells were injected 5 d after tumor injection. For Nomo1 cells, tumor burden was determined by bioluminescence imaging with a Xenogen In Vivo Imaging System (Caliper Life Sciences). For PDXs, flow cytometry analysis of blood draws and ultrasonography of spleen size were used as readouts for tumor burden. No animals were excluded from the analysis. For such disseminated tumor models, our institution does not specify an allowable maximal tumor burden.

Statistics and reproducibility

All statistical analyses were performed using GraphPad Prism v.9 unless stated otherwise. The data are represented as mean ± s.e.m., and P values of <0.05 were considered statistically significant. No statistical method was used to predetermine sample size, but our sample sizes are similar to those reported in previous publications^11,16,44. Data distribution was assumed to be normal, but this was not formally tested. No data were excluded from the analyses. Animals were randomized based on body weight before treatment. The Preclinical Core Facility staff was blinded to mouse treatment and relevant outcomes. The other investigators were not blinded to allocation during the experiments and outcome assessment.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Raw proteomic data generated here have been deposited at the ProteomeXchange/PRIDE repository under accession numbers PXD035404, PXD035589 and PXD035591.

Previously published bulk RNA-seq datasets (TCGA AML, BEAT AML and TARGET AML) that are reported here are available under accession codes phs000178, phs001657 and phs000218, respectively, through the NIH/NCI GDC Data Portal.

The following are the other public database repertories used in the study: https://www.proteinatlas.org/ (ITGB2, CD33, CD123 and CLEC12A), https://www.proteinatlas.org/humanproteome/single+cell+type, https://www.proteinatlas.org/humanproteome/single+cell+type/blood+&+immune+cells and https://www.rcsb.org/ (PDB 5E6R). Source data are provided with this paper.

Code availability

The Ving software package for analysis of DSSO XL–MS data is available at GitHub at https://github.com/mhoopmann/Ving.

References

Lim, W. A. & June, C. H. The principles of engineering immune cells to treat cancer. Cell 168, 724–740 (2017).
CAS PubMed PubMed Central Google Scholar
Brooks, J. D. Translational genomics: the challenge of developing cancer biomarkers. Genome Res. 22, 183–187 (2012).
CAS PubMed PubMed Central Google Scholar
Beck, A., Goetsch, L., Dumontet, C. & Corvaïa, N. Strategies and challenges for the next generation of antibody–drug conjugates. Nat. Rev. Drug Discov. 16, 315–337 (2017).
CAS PubMed Google Scholar
Brudno, J. N. & Kochenderfer, J. N. Toxicities of chimeric antigen receptor T cells: recognition and management. Blood 127, 3321–3330 (2016).
Hosen, N. et al. The activated conformation of integrin β₇ is a novel multiple myeloma-specific target for CAR T cell therapy. Nat. Med. 23, 1436–1443 (2017).
CAS PubMed Google Scholar
Leitner, A., Walzthoeni, T. & Aebersold, R. Lysine-specific chemical cross-linking of protein complexes and identification of cross-linking sites using LC–MS/MS and the xQuest/xProphet software pipeline. Nat. Protoc. 9, 120–137 (2014).
CAS PubMed Google Scholar
Yu, C. & Huang, L. Cross-linking mass spectrometry (XL–MS): an emerging technology for interactomics and structural biology. Anal. Chem. 90, 144–165 (2018).
CAS PubMed Google Scholar
Piersimoni, L., Kastritis, P. L., Arlt, C. & Sinz, A. Cross-linking mass spectrometry for investigating protein conformations and protein–protein interactions—a method for all seasons. Chem. Rev. 122, 7500–7531 (2022).
CAS PubMed Google Scholar
Steigenberger, B., Pieters, R. J., Heck, A. J. R. & Scheltema, R. A. PhoX: an IMAC-enrichable cross-linking reagent. ACS Cent. Sci. 5, 1514–1522 (2019).
CAS PubMed PubMed Central Google Scholar
Wollscheid, B. et al. Mass-spectrometric identification and relative quantification of N-linked cell surface glycoproteins. Nat. Biotechnol. 27, 378–386 (2009).
CAS PubMed PubMed Central Google Scholar
Nix, M. A. et al. Surface proteomics reveals CD72 as a target for in vitro-evolved nanobody-based CAR-T cells in KMT2A/MLL1-rearranged B-ALL. Cancer Discov. 11, 2032–2049 (2021).
CAS PubMed PubMed Central Google Scholar
Perna, F. et al. Integrating proteomics and transcriptomics for systematic combinatorial chimeric antigen receptor therapy of AML. Cancer Cell 32, 506–519 (2017).
CAS PubMed PubMed Central Google Scholar
Kantarjian, H. et al. Acute myeloid leukemia: current progress and future directions. Blood Cancer J. 11, 41 (2021).
PubMed PubMed Central Google Scholar
Mardiana, S. & Gill, S. CAR T cells for acute myeloid leukemia: state of the art and future directions. Front. Oncol. 10, 697 (2020).
PubMed PubMed Central Google Scholar
Maucher, M. et al. Current limitations and perspectives of chimeric antigen receptor-T-cells in acute myeloid leukemia. Cancers 13, 6157 (2021).
CAS PubMed PubMed Central Google Scholar
Kenderian, S. S. et al. CD33-specific chimeric antigen receptor T cells exhibit potent preclinical activity against human acute myeloid leukemia. Leukemia 29, 1637–1647 (2015).
CAS PubMed PubMed Central Google Scholar
Gill, S. et al. Anti-CD123 chimeric antigen receptor T cells (CART-123) provide a novel myeloablative conditioning regimen that eradicates human acute myeloid leukemia in preclinical models. Blood 122, 143 (2013).
Google Scholar
Quentmeier, H. et al. Expression of HOX genes in acute leukemia cell lines with and without MLL translocations. Leuk. Lymphoma 45, 567–574 (2004).
CAS PubMed Google Scholar
Kao, A. et al. Development of a novel cross-linking strategy for fast and accurate identification of cross-linked peptides of protein complexes. Mol. Cell. Proteomics 10, M110.002212 (2011).
PubMed Google Scholar
Liu, F., Rijkers, D. T. S., Post, H. & Heck, A. J. R. Proteome-wide profiling of protein assemblies by cross-linking mass spectrometry. Nat. Methods 12, 1179–1184 (2015).
CAS PubMed Google Scholar
Klykov, O. et al. Efficient and robust proteome-wide approaches for cross-linking mass spectrometry. Nat. Protoc. 13, 2964–2990 (2018).
CAS PubMed Google Scholar
Deutsch, E. W. et al. Trans-Proteomic Pipeline, a standardized data processing pipeline for large-scale reproducible proteomics informatics. Proteomics Clin. Appl. 9, 745–754 (2015).
CAS PubMed PubMed Central Google Scholar
Jiao, F. et al. Two-dimensional fractionation method for proteome-wide cross-linking mass spectrometry analysis. Anal. Chem. 94, 4236–4242 (2022).
CAS PubMed PubMed Central Google Scholar
Ferguson, I. D. et al. The surfaceome of multiple myeloma cells suggests potential immunotherapeutic strategies and protein markers of drug resistance. Nat. Commun. 13, 4121 (2022).
CAS PubMed PubMed Central Google Scholar
Sen, M. & Springer, T. A. Leukocyte integrin α_lβ₂ headpiece structures: the α_i domain, the pocket for the internal ligand, and concerted movements of its loops. Proc. Natl Acad. Sci. USA 113, 2940–2945 (2016).
PubMed PubMed Central Google Scholar
Chen, X. et al. Requirement of open headpiece conformation for activation of leukocyte integrin α_Xβ₂. Proc. Natl Acad. Sci. USA 107, 14727–14732 (2010).
CAS PubMed PubMed Central Google Scholar
Fagerholm, S. C., Guenther, C., Asens, M. L., Savinko, T. & Uotila, L. M. β₂-integrins and interacting proteins in leukocyte trafficking, immune suppression, and immunodeficiency disease. Front. Immunol. 10, 254 (2019).
CAS PubMed PubMed Central Google Scholar
Schittenhelm, L., Hilkens, C. M. & Morrison, V. L. β₂ integrins as regulators of dendritic cell, monocyte, and macrophage function. Front. Immunol. 8, 1866 (2017).
PubMed PubMed Central Google Scholar
Herter, J. & Zarbock, A. Integrin regulation during leukocyte recruitment. J. Immunol. 190, 4451–4457 (2020).
Google Scholar
Ley, K., Laudanna, C., Cybulsky, M. I. & Nourshargh, S. Getting to the site of inflammation: the leukocyte adhesion cascade updated. Nat. Rev. Immunol. 7, 678–689 (2007).
CAS PubMed Google Scholar
Petri, B., Phillipson, M. & Kubes, P. The physiology of leukocyte recruitment: an in vivo perspective. J. Immunol. 180, 6439–6446 (2008).
CAS PubMed Google Scholar
Phillipson, M. & Kubes, P. The neutrophil in vascular inflammation. Nat. Med. 17, 1381–1390 (2011).
Oellerich, T., Oellerich, M. F., Engelke, M. & Silvia, M. β₂ integrin-derived signals induce cell survival and proliferation of AML blasts by activating a Syk/STAT signaling axis. Blood 121, 3889–3900 (2017).
Google Scholar
Dransfield, I., Cabañas, C., Craig, A. & Hogg, N. Divalent cation regulation of the function of the leukocyte integrin LFA-1. J. Cell Biol. 116, 219–226 (1992).
CAS PubMed Google Scholar
Townsend, E. C. et al. The Public Repository of Xenografts enables discovery and randomized phase II-like trials in mice. Cancer Cell 29, 574–586 (2016).
CAS PubMed PubMed Central Google Scholar
Weinstein, J. N. et al. The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 45, 1113–1120 (2013).
PubMed PubMed Central Google Scholar
Tyner, J. W. et al. Functional genomic landscape of acute myeloid leukemia. Nature 562, 526–531 (2018).
CAS PubMed PubMed Central Google Scholar
Bolouri, H. et al. The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions. Nat. Med. 24, 103–112 (2017).
PubMed PubMed Central Google Scholar
Uhlén, M. et al. Tissue-based map of the human proteome. Science 347, 1260419 (2015).
PubMed Google Scholar
Hogg, N. & Selvendran, Y. An anti-human monocyte/macrophage monoclonal antibody, reacting most strongly with macrophages in lymphoid tissue. Cell. Immunol. 92, 247–253 (1985).
CAS PubMed Google Scholar
Shimaoka, M. et al. AL-57, a ligand-mimetic antibody to integrin LFA-1, reveals chemokine-induced affinity up-regulation in lymphocytes. Proc. Natl Acad. Sci. USA 103, 13991–13996 (2006).
CAS PubMed PubMed Central Google Scholar
Persson, H. et al. CDR-H3 diversity is not required for antigen recognition by synthetic antibodies. J. Mol. Biol. 425, 803–811 (2013).
CAS PubMed Google Scholar
Ye, F., Kim, C. & Ginsberg, M. H. Reconstruction of integrin activation. Blood 119, 26–33 (2012).
CAS PubMed PubMed Central Google Scholar
Mansilla-Soto, J. et al. HLA-independent T cell receptors for targeting tumors with low antigen density. Nat. Med. 28, 345–352 (2022).
CAS PubMed PubMed Central Google Scholar
Blanter, M. et al. Method matters: effect of purification technology on neutrophil phenotype and function. Front. Immunol. 13, 820058 (2022).
CAS PubMed PubMed Central Google Scholar
Allen, T. M. et al. Humanized immune system mouse models: progress, challenges and opportunities. Nat. Immunol. 20, 770–774 (2019).
DeCaen, P. G., Yarov-Yarovoy, V., Zhao, Y., Scheuer, T. & Catterall, W. A. Disulfide locking a sodium channel voltage sensor reveals ion pair formation during activation. Proc. Natl Acad. Sci. USA 105, 15142–15147 (2008).
CAS PubMed PubMed Central Google Scholar
Ma, H., Padmanabhan, I. S., Parmar, S. & Gong, Y. Targeting CLL-1 for acute myeloid leukemia therapy. J. Hematol. Oncol. 12, 41 (2019).
PubMed PubMed Central Google Scholar
Li, S., Mason, C. & Melnick, A. Genetic and epigenetic heterogeneity in acute myeloid leukemia. Curr. Opin. Genet. Dev. 36, 100–106 (2016).
PubMed PubMed Central Google Scholar
Kong, A. T., Leprevost, F. V., Avtonomov, D. M., Mellacheruvu, D. & Nesvizhskii, A. I. MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry-based proteomics. Nat. Methods 14, 513–520 (2017).
CAS PubMed PubMed Central Google Scholar
Martens, L. et al. mzML—a community standard for mass spectrometry data. Mol. Cell. Proteomics 10, R110.000133 (2011).
PubMed Google Scholar
Keller, A., Eng, J., Zhang, N., Li, X. J. & Aebersold, R. A uniform proteomics MS/MS analysis platform utilizing open XML file formats. Mol. Syst. Biol. 1, 2005.0017 (2005).
PubMed PubMed Central Google Scholar
Ihling, C. H., Piersimoni, L., Kipping, M. & Sinz, A. Cross-linking/mass spectrometry combined with ion mobility on a timsTOF Pro instrument for structural proteomics. Anal. Chem. 93, 11442–11450 (2021).
CAS PubMed Google Scholar
Chambers, M. C. et al. A cross-platform toolkit for mass spectrometry and proteomics. Nat. Biotechnol. 30, 918–920 (2012).
CAS PubMed PubMed Central Google Scholar
Chen, Z. L. et al. A high-speed search engine pLink 2 with systematic evaluation for proteome-scale identification of cross-linked peptides. Nat. Commun. 10, 3404 (2019).
PubMed PubMed Central Google Scholar
Jain, T. et al. Biophysical properties of the clinical-stage antibody landscape. Proc. Natl Acad. Sci. USA 114, 944–949 (2017).
CAS PubMed PubMed Central Google Scholar
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).
CAS PubMed PubMed Central Google Scholar
Concordet, J. P. & Haeussler, M. CRISPOR: intuitive guide selection for CRISPR/Cas9 genome editing experiments and screens. Nucleic Acids Res. 46, W242–W245 (2018).
CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank D. Sheppard (UCSF) for providing his expert opinion and consultation on integrin biology. We also thank S. K. Elledge (UCSF) and M. A. Burlingame (UCSF) for technical assistance with MS sample analysis. We thank N. Wiita for assistance in figure graphics. We also thank the participants and their families who contributed research specimens to the associated tissue banks. We acknowledge funding from the Michelson Prize-2019 (to K.M.) awarded by the Michelson Medical Research Foundation and Human Vaccine Project; NIH R21 CA263299 (to A.P.W.); NIH R01GM074830, NIH R35GM145249 and NIH R01GM130144 (to L.H.); Canadian Institutes of Health Research (MOPS-136944) and Bristol-Myers Squibb (to S.S.S.); National Cancer Institute Cancer Center Support Grant P30CA082103 (to UCSF to support the Pediatric Hematopoietic Tissue Cell Bank); NIH R01GM087221, NIH S10OD026936 and the National Science Foundation award 1920268 (to R.L.M. and M.R.H.) and American Society of Hematology Research Training Award for Fellows and Chan Zuckerberg Biohub Physician-Scientist Fellowship Program (to W.C.T.). Flow cytometry was performed at the UCSF Laboratory for Cell Analysis, and mouse studies were performed at the UCSF Preclinical Therapeutics Core, both part of the Helen Diller Family Comprehensive Cancer Center and supported by P30CA082103.

Author information

Authors and Affiliations

Department of Laboratory Medicine, University of California San Francisco, San Francisco, CA, USA
Kamal Mandal, Gianina Wicaksono, Adila Izgutdina, Bonell Patiño Escobar, Matthew A. Nix, Akul Naik, Emilio Ramos, Melanie McMillan, Victor Gavallos, Andrew D. Leavitt & Arun P. Wiita
Department of Physiology and Biophysics, University of California Irvine, Irvine, CA, USA
Clinton Yu & Lan Huang
The Donnelly Centre, University of Toronto, Toronto, Ontario, Canada
Jarrett J. Adams, Maryna Gorelik & Sachdev S. Sidhu
School of Pharmacy, University of Waterloo, Kitchener, Ontario, Canada
Jarrett J. Adams & Sachdev S. Sidhu
Institute for Systems Biology, Seattle, WA, USA
Michael R. Hoopmann & Robert L. Moritz
Department of Pediatrics, Division of Hematology/Oncology, University of California San Francisco, San Francisco, CA, USA
William C. Temple, Juwita Hübner, Benjamin J. Huang & Elliot Stieglitz
Department of Pediatrics, Division of Allergy, Immunology, and Bone Marrow Transplantation, University of California San Francisco, San Francisco, CA, USA
William C. Temple
Department of Pharmaceutical Chemistry and Bioanalytics, Institute of Pharmacy, Martin-Luther University Halle-Wittenberg, Halle, Germany
Christian H. Ihling & Andrea Sinz
UCSF/Gladstone Institute for Genomic Immunology, San Francisco, CA, USA
William H. Xie & Justin Eyquem
Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, San Francisco, CA, USA
Juwita Hübner, Veronica Steri, Juan Antonio Camara Serrano, Fernando Salangsang, Paul Phojanakong, Elliot Stieglitz & Arun P. Wiita
Center for Cell and Gene Therapy, Baylor College of Medicine, Houston Methodist Hospital–Texas Children’s Hospital, Houston, TX, USA
Lisa A. Rollins, Sandy M. Reid & Cliona M. Rooney
Department of Medicine, Division of Hematology/Oncology, University of California San Francisco, San Francisco, CA, USA
Corynn Kasap, Aaron C. Logan, Justin Eyquem & Catherine C. Smith
Department of Bioengineering and Therapeutic Sciences, University of California San Francisco, San Francisco, CA, USA
Arun P. Wiita
Chan Zuckerberg Biohub San Francisco, San Francisco, CA, USA
Arun P. Wiita

Authors

Kamal Mandal
View author publications
You can also search for this author in PubMed Google Scholar
Gianina Wicaksono
View author publications
You can also search for this author in PubMed Google Scholar
Clinton Yu
View author publications
You can also search for this author in PubMed Google Scholar
Jarrett J. Adams
View author publications
You can also search for this author in PubMed Google Scholar
Michael R. Hoopmann
View author publications
You can also search for this author in PubMed Google Scholar
William C. Temple
View author publications
You can also search for this author in PubMed Google Scholar
Adila Izgutdina
View author publications
You can also search for this author in PubMed Google Scholar
Bonell Patiño Escobar
View author publications
You can also search for this author in PubMed Google Scholar
Maryna Gorelik
View author publications
You can also search for this author in PubMed Google Scholar
Christian H. Ihling
View author publications
You can also search for this author in PubMed Google Scholar
Matthew A. Nix
View author publications
You can also search for this author in PubMed Google Scholar
Akul Naik
View author publications
You can also search for this author in PubMed Google Scholar
William H. Xie
View author publications
You can also search for this author in PubMed Google Scholar
Juwita Hübner
View author publications
You can also search for this author in PubMed Google Scholar
Lisa A. Rollins
View author publications
You can also search for this author in PubMed Google Scholar
Sandy M. Reid
View author publications
You can also search for this author in PubMed Google Scholar
Emilio Ramos
View author publications
You can also search for this author in PubMed Google Scholar
Corynn Kasap
View author publications
You can also search for this author in PubMed Google Scholar
Veronica Steri
View author publications
You can also search for this author in PubMed Google Scholar
Juan Antonio Camara Serrano
View author publications
You can also search for this author in PubMed Google Scholar
Fernando Salangsang
View author publications
You can also search for this author in PubMed Google Scholar
Paul Phojanakong
View author publications
You can also search for this author in PubMed Google Scholar
Melanie McMillan
View author publications
You can also search for this author in PubMed Google Scholar
Victor Gavallos
View author publications
You can also search for this author in PubMed Google Scholar
Andrew D. Leavitt
View author publications
You can also search for this author in PubMed Google Scholar
Aaron C. Logan
View author publications
You can also search for this author in PubMed Google Scholar
Cliona M. Rooney
View author publications
You can also search for this author in PubMed Google Scholar
Justin Eyquem
View author publications
You can also search for this author in PubMed Google Scholar
Andrea Sinz
View author publications
You can also search for this author in PubMed Google Scholar
Benjamin J. Huang
View author publications
You can also search for this author in PubMed Google Scholar
Elliot Stieglitz
View author publications
You can also search for this author in PubMed Google Scholar
Catherine C. Smith
View author publications
You can also search for this author in PubMed Google Scholar
Robert L. Moritz
View author publications
You can also search for this author in PubMed Google Scholar
Sachdev S. Sidhu
View author publications
You can also search for this author in PubMed Google Scholar
Lan Huang
View author publications
You can also search for this author in PubMed Google Scholar
Arun P. Wiita
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

K.M. and A.P.W. conceptualized the study, acquired the funding, performed data analysis/interpretation and wrote the paper. K.M., G.W., C.Y., J.J.A., W.C.T., A.I., B.P.E., M.G., M.R.H., C.H.I., A.N., W.H.X., J.H., L.A.R., S.M.R., J.A.C.S., F.S., P.P. and B.J.H. performed experiments and/or data analysis. E.R., C.K., M.M., V.G., E.S. and C.C.S. acquired primary samples and/or performed associated analyses. A.P.W., M.A.N., A.C.L., V.S., A.S., A.D.L, C.M.R., J.E., S.S.S., L.H. and R.L.M. provided resources and/or supervised the study.

Corresponding author

Correspondence to Arun P. Wiita.

Ethics declarations

Competing interests

K.M., J.J.A., S.S.S. and A.P.W. have filed a patent related to the antibody sequences described herein. A.P.W. has received research funding from Genentech. C.S. has received research funding from Revolution Medicines, Abbvie and Erasca and has served on advisory boards for Genentech, Abbvie and Astellas. C.M.R. has equity in Allovir and Marker Therapeutics, has received research support from Tessa Therapeutics and is on the Scientific Advisory Board of Tessa Therapeutics and Marker. The spouse of C.M.R. has served on advisory boards for Walking Fish Therapeutics, CellGenix, Marker Therapeutics, Tessa Therapeutics, Abintus, Allogene, Bellicum Pharmaceuticals, Bluebird Bio, Athenex, Memgen, Turnstone Biologics, Coya Therapeutics, TScan Therapeutics, Onkimmune, Poseida Therapeutics and Allovir. The other authors declare no competing interests.

Peer review

Peer review information

Nature Cancer thanks Stephen Gottschalk, Alexander Leitner and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Ving and XL-MS SEC.

a, Schematic workflow describing the working principle of Ving. b, Representative SEC trace of peptides obtained from DSSO cross-linked samples. c, Representative SEC trace of peptides obtained from PhoX cross-linked samples. For both strategies (DSSO and PhoX), samples were processed in 4 separate batches and each time the SEC trace pattern was similar. XL peptides refers to cross-linked peptides.

Extended Data Fig. 2 Representative MS spectra demonstrating the MS3 based strategy of XL-MS.

Example MS spectra demonstrating the MS³ based strategy of XL-MS. The cross-linked peptides of 811.57 m/z is selected for MS², at which step the cross-linker is cleaved in the collision cell generating two separate peptides of 606.26 m/z and 942.45 m/z, respectively. These two high-abundance peptides are then selected for MS³ where they undergoes full fragmentation for peptide identification. We also note the respective modification on Lysine residues resulting from the cross-linker.

Extended Data Fig. 3 Discerning active integrin β₂ expression.

a, Cartoon diagram showing proposed inactive and active conformations of ITGB2. b, Flow cytometry plot showing presence of total integrin β₂ on CD34+ HSPCs from GM-CSF mobilized peripheral blood. Cells were gated on singlet cells for analysis. Deidentified patient samples were used for this analysis (n = 5, independent donors). c, Flow cytometry analysis showing expression of active ITGB2 in PDX models of AML (PDX-A and PDX-B). The y-axis represents percent count normalized to mode. Cells were gated on human CD45+ population cells for analysis. Representative plot from n = 2 separate PDX models of AML.

Extended Data Fig. 4 ITGB2 transcript expression.

a, AML subtype specific expression analysis of ITGB2 and other notable AML targets of patient samples from TARGET database. b, Single cell sequencing data showing expression of notable AML target across normal human tissues and immune cells (adapted from Human Protein Atlas⁴²).

Extended Data Fig. 5 Initial anti-active integrin β₂ CAR-T designs.

Luciferase based cytotoxicity analysis of M24 and A57 antibody derived CAR-T cells against AML cell line Nomo-1 (Data from single experiment performed in triplicate). The luciferase signals of the cytotoxicity assays were normalized against untransduced CAR-T of their respective E:T ratios.

Extended Data Fig. 6 Characterizing recombinant antibodies to integrin β₂.

a, Representative SEC trace of the antibody 7065 showing distinct peak, for quality check. b. Non-specific ELISA panel showing specificity profiles of the antibodies obtained from phage display selection. (Data from single experiment performed in duplicates) c, Representative BLI plots showing binding affinities (K_D) of the antibodies against ITGB2 with their alpha partners. Each experiment was performed with n = 3 different concentrations of antibody. d, Flow cytometry analysis showing binding of antibody 7065 on the AML cells. The y-axis represents percent count normalized to mode. Gating strategy shown in Supplementary Information 1a. n = 3 independent experiments. e, Flow cytometry analysis showing binding of antibody 7065 on the primary AML cells, gated on CD34+ blasts. The y-axis represents percent count normalized to mode. Gating strategy shown in Supplementary Information 1c. n = 7 independent patient samples.

Extended Data Fig. 7 Evaluation of aITGB2 CAR-T designs incorporating recombinant antibodies.

a, Flow cytometry screen for cytotoxicity and activation status of aITGB2 CAR-T designs vs. AML cell line Nomo1 (n = 1 for each design). Similarly, as a demonstration of specificity, cytotoxicity and activation status was also checked against AMO-1 (multiple myeloma cell line that does not express integrin β₂). Cells were gated on single cells for analysis. Gating strategy shown in Supplementary Information 2a. b, Flow cytometry analysis showing absence of ITGB2 in AMO-1. Cells were gated on single cells for analysis. Flow cytometry gating strategy similar to shown in Supplementary Information 1a. c, Flow cytometry analysis showing knockout efficiency of the various sgRNA used for knocking out ITGB2 in primary T cells (n = 1 for each sgRNA). Cells were gated on single cells for analysis. Flow cytometry gating strategy similar to that shown in Supplementary Information 1a. d, Plot showing proliferation of aITGB2-CAR-T cells, ‘with ITGB2 knockout’ vs ‘Non-Targeting (NT) – control’. Scrambled sgRNA was used for NT-control. (Data from single experiment performed in triplicate). e, Plots showing off-target analysis of the sgRNA used for CAR-T manufacturing using CRISPOR tool. Top 5 potential hits were assessed. f, Degranulation assay of aITGB2 and anti-CD33 CAR-T against Nomo1 based on CD107a staining. CAR positivity denoted by GFP tag on CAR construct. E:T ratio was 1:1 and 6 hours incubation time (n = 1). Cells were gated on single cells for analysis. Gating strategy similar to shown in Supplementary Information 2a. g, Luciferase assay-based cytotoxicity analysis showing cytotoxicity of aITGB2-CAR-T cells against AML cell lines with varying antigen density (Representative of n = 2 independent experiments with similar results, each experiment performed in triplicate). The flow cytometry data showing antigen density has been repurposed from Fig. 2b. Only aITGB2 CAR-T manufacturing involved knocking out integrin β₂, and only in (f) and (g).

Extended Data Fig. 8 Additional aITGB2 CAR-T evaluation.

a, Luciferase assay-based cytotoxicity analysis showing efficacy of 7065 based aITGB2 CAR-T with 1x – 4x Gly₄Ser (L1-L4) linker between heavy and light chain (Data from single experiment performed in triplicate). The luciferase signals of the cytotoxicity assays were normalized against untransduced T-cells of their respective E:T ratios. b, Bar plots showing cytotoxicity of aITGB2 CAR-T against primary AML patient samples. n = 3 technical replicates and 3 independent patient samples. E:T ratio was 3:1 with overnight incubation. Cells were gated on CD34+ cells for analysis. Gating strategy similar to Supplementary Information 1c. c, Cytokine profiling of aITGB2 CAR-T comparing against Empty CAR-T control upon pre vs post tumor exposure at E:T ratio of 1:1 for overnight incubation (Data from single experiment performed in triplicate). d–f, Representative flow cytometry-based histogram analysis showing activation markers (d), exhaustion markers (e) and memory markers (f) of CAR-T in pre vs post tumor exposure at E:T ratio of 1:1 for overnight incubation. Gating strategy similar to Supplementary Information 2d. Antibody dilutions used in d–f was 1:50. Only aITGB2 CAR-T manufacturing involved knocking out integrin β₂. All the statistical data in this figure are represented as mean ± SEM.

Extended Data Fig. 9 Determining specificity of aITGB2 CAR-T.

a, Luciferase assay-based cytotoxicity analysis showing no activity of aITGB2 CAR-T vs. Namalwa (B-ALL) line which does not harbor active ITGB2 although it does have total form of ITGB2 (see Fig. 2b). Nomo-1 as the positive control (Data from single experiment performed in triplicate). The luciferase signals of the cytotoxicity assays were normalized against untransduced T-cells of their respective E:T ratios. b, Flow cytometry analysis showing absence of active ITGB2 and presence of total ITGB2 in T and B cells (n = 3 independent experiments). Cells were gated on single cells for analysis. Gating strategy similar to that shown in Supplementary information 1b. c, Flow cytometry analysis showing no discernible impact of aITGB2 CAR-T against CD34+ HSPCs from GM-CSF mobilized peripheral blood. The y-axis represents percent count normalized to mode. Gating strategy similar to that shown in Supplementary information 2b (n = 1 donor). d, Flow cytometry analysis showing non-specific depletion of B cells with aITGB2 and anti-CD33 CAR-T (Representative of n = 2 independent experiments with similar results, each experiment performed in triplicate). e, Flow cytometry plots showing activation of pathogen-specific T cells using CD69 as a marker. f, Quantitative analysis showing no depletion of pathogen-specific T cells with aITGB2 CAR-T when at resting state, but moderate depletion when activated exogenously. Gating strategy shown in Supplementary information 2f (Data from single experiment performed in triplicate). g, Flow cytometry analysis showing presence of active ITGB2 in myeloid cells (n = 3 independent experiments). Cells were gated on single cells for analysis. Gating strategy similar to that shown in Supplementary information 2c. h, Flow cytometry analysis showing cytotoxicity of aITGB2-CAR against neutrophils and monocytes in vitro. Cells were gated on single cells for analysis. i, Quantitative plot showing absence of cytotoxicity of aITGB2 CAR-T against human CD45+ cells in HIS mice at noted time points. Peripheral blood samples were used for this analysis. CD33 CAR-T serves as a positive control where significant cytotoxicity was observed (n = 5 mice in each arm). Two-tailed t-test were performed for statistical analysis. [p-value = 0.009 (2-week, ITGB2-CAR vs CD33-CAR); p-value = 0.047 (2-week, Empty-CAR vs CD33-CAR); p-value = 0.035 (4-week, ITGB2-CAR vs CD33-CAR)]; *p ≤ 0.05; **p ≤ 0.01 j, Flow cytometry analysis showing cross reactivity of 7065 antibody against the murine ITGB2 on S49.1 cell line. The y-axis represents percent count normalized to mode. Cells were gated on single cells for analysis. Gating strategy similar to shown in Supplementary information 1a. k, Quantitative bar plots showing validation of mice aITGB2 CAR-T for its toxicity against murine S49.1 cells. E:T ratio was 2:1 with overnight coculture. hCD19 CAR-T, being specific for human CD19 only, was used as a negative control in murine CAR-T format. Gating strategy shown in Supplementary information 2f. (Data from single experiment performed in triplicate). l, Flow cytometry analysis showing knockout efficiency of the mice sgRNA used for knocking out ITGB2 in primary mice T cells (n = 1). Cells were gated on single cells for analysis. Flow cytometry gating strategy similar to that shown in Supplementary Information 1a. m, Complete blood count profiling of HIS mice treated with aITGB2 CAR-T periodically over 6 weeks (data from n = 6 mice). Only aITGB2 CAR-T manufacturing involved knocking out integrin β₂. All the statistical data in this figure are represented as mean ± SEM. All the mice used in this figure were females.

Extended Data Fig. 10 aITGB2 CAR-T efficacy in PDX models.

a, Flow cytometry analysis and bar graph of peripheral blood draw showing tumor burden at 6 and 8 weeks post tumor injection of PDX-A. The y-axis represents percent count normalized to mode. Cells were gated on single cells for analysis. Representative of data from n = 6 mice per arm; plots of only mice alive at designated time point. b, Flow cytometry analysis and bar graph of peripheral blood draw showing tumor burden at 3.5 weeks post tumor injection of PDX-B. The y-axis represents percent count normalized to mode. Cells were gated on single cells for analysis. Representative of data from n = 6 mice per arm; plots of only mice alive at designated time point. c, Plots showing expansion and persistence of aITGB2 CAR-T in mice peripheral blood, analyzed periodically for 3 months (n = 6 mice per arm, subsequent data points were from mice alive at that time point). d, Flow cytometry analysis showing active ITGB2 density of tumor cells harvested from relapse Nomo-1 mice model (n = 4 mice per condition). The y-axis represents percent count normalized to mode. Cells were gated on human CD45+ cells for analysis. Gating strategy similar to that shown in Supplementary information 2e. Only aITGB2 CAR-T manufacturing involved knocking out integrin β₂. All the statistical data in this figure are represented as mean ± SEM. All the mice used in this figure were females.

Supplementary information

Supplementary Information

Supplementary Information 1 and 2.

Reporting Summary

Supplementary Table

Supplementary Tables 1–3.

Source data

Source Data

Statistical source data for all figures in separate tabs of a single Excel sheet.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, 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 license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license 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 license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Mandal, K., Wicaksono, G., Yu, C. et al. Structural surfaceomics reveals an AML-specific conformation of integrin β₂ as a CAR T cellular therapy target. Nat Cancer 4, 1592–1609 (2023). https://doi.org/10.1038/s43018-023-00652-6

Download citation

Received: 19 August 2022
Accepted: 12 September 2023
Published: 30 October 2023
Issue Date: November 2023
DOI: https://doi.org/10.1038/s43018-023-00652-6