CD99-mediated immunological synapse formation potentiates CAR-T cell function

Nam, Giri; Yeon, Hye Ran; Park, Hyung Bae; Chang, Hanna; Kim, Ji Hwan; Cho, Byoung-Kyu; Jung, Hyeryeon; Yi, Eugene C.; Kim, Seoyeon; An, Joon-Yong; Lee, Ji Eun; Lee, Youngjae; Lee, Seoho; Lim, Hyeonji; Shon, Woo-Jeong; Hwang, Eun Mi; Ryu, Hoon; Chang, Jun; Choi, Kyungho; Choi, Eun Young

doi:10.1038/s41467-025-63184-w

Download PDF

Article
Open access
Published: 27 August 2025

CD99-mediated immunological synapse formation potentiates CAR-T cell function

Nature Communications volume 16, Article number: 7987 (2025) Cite this article

5111 Accesses
17 Altmetric
Metrics details

Subjects

Abstract

Despite the efficacy of chimeric antigen receptor (CAR)-T cells in selected hematological malignancies, further improvement on CAR-T designs is still desirable. We hypothesize that modifying the CAR structure to enhance immunological synapse (IS) stabilization and CAR target-binding may be a feasible strategy. Here we show that the membrane protein, CD99, is critical for IS formation in T cells by mediating actin-microtubule interaction. CD99 deficiency abolishes IS formation and prevents effective in vivo T cell immunity. Mechanistically, CD99 interacts with microtubules and actins through the transmembrane and cytoplasmic domains, respectively, with which myosin and IQGAP1 interact. As such, incorporating the transmembrane and juxtamembrane domains of CD99 into the CAR structure enhances IS formation and improves the therapeutic efficacy of human CAR-T cells against lymphoma in immune-deficient mice. Our data thus suggest that CD99-mediated IS stabilization may help improve CAR design and efficacy.

CAR-T therapy dilemma and innovative design strategies for next generation

Article Open access 27 March 2025

Chimeric antigen receptor with novel intracellular modules improves antitumor performance of T cells

Article Open access 15 January 2025

3D live imaging and phenotyping of CAR-T cell mediated-cytotoxicity using high-throughput Bessel oblique plane microscopy

Article Open access 06 August 2024

Introduction

Chimeric antigen receptor (CAR) is a recombinant receptor in which an extracellular tumor-targeting antibody moiety is linked to the intracellular signaling domain of the T cell receptor¹. When a CAR gene is introduced into T cells (CAR-T cells) from cancer patients, these T cells can recognize the tumor antigen and be activated via the CAR to kill the tumor cells. Although CAR-T cell therapy has shown remarkable efficacy in some hematological malignancies, its efficacy in general needs to be significantly improved². For this purpose, attempts have been made to modify the CAR structure. Most efforts to date have focused on the addition of one or more signaling domains from costimulatory molecules, such as CD28, 41BB, and OX40, to the intracellular region of the CAR³. However, increasing the number of costimulatory domains incorporated into a single CAR molecule may lead to excessive activation of CAR-T cells, resulting in adverse side effects, such as cytokine release syndrome^4,5. Therefore, a different approach may be required to further improve the CAR structure, such as strengthening the immunological synapse (IS) between CAR-T cells and the target tumor cells. This could be achieved by incorporating the domains from a key surface receptor responsible for IS formation into CAR. However, our current understanding of the molecular mechanism of IS formation and the receptors responsible is very limited.

The IS is a specialized structure formed at the site of contact of T cells with antigen-presenting cells (APCs) or target cells, and is known to enhance T cell activation and effector function^6,7,8. The mature IS consists of three radially symmetric substructures, known as the central (c-), peripheral (p-), and distal (d-) supramolecular activation clusters (SMACs)⁹. In the cytosol, the actin cytoskeleton forms distinct networks: a hypodense network in the cSMAC, a lamella-like arc network in the pSMAC, and a lamellipodia-like branched network in the dSMAC¹⁰. Microtubules are arrayed with radial symmetry, while the microtubule-organizing center (MTOC) is located immediately beneath the plasma membrane in the cSMAC¹¹. The unique shape of the IS cytoskeleton network is formed by reorganization of the actin and microtubule networks after TCR signaling. TCR signaling triggers actin polymerization at the T-APC contact site, promoting lamellipodia extension toward the APC^12,13. Actin is then depolymerized at the center of the contact site, followed by its retrograde flow from the lamellipodia to the central area along with maturation of the IS structure. TCR signaling also induces microtubule polymerization/elongation, which occurs in parallel with lamellipodia extension¹⁴. Subsequently, the MTOC translocates to the synapse side from the distal region (uropod) of T cells^15,16. Importantly, these actin and microtubule reorganization processes are tightly intertwined, probably through the physical interaction of actin and microtubule filaments^16,17,18,19. However, the mechanisms underlying the actin-microtubule physical interaction, and the existence of a membrane protein that orchestrates actin and microtubule reorganization are unclear.

CD99 is a transmembrane protein that is expressed at basal levels on various types of cells, but is upregulated on activated T cells and some tumor cells²⁰. Although CD99 was initially identified as a cell adhesion molecule²¹, its involvement in transendothelial migration of leukocytes and tumor cells has been demonstrated^22,23,24,25. Roles for CD99 in the generation of costimulatory effects on T cells^26,27, intracellular molecular trafficking in T cells^28,29,30, and recruitment of CD99 into IS during T cell activation³¹ have also been reported. Mechanistically, all of these functions are known to be associated with actin polymerization/reorganization^28,29,32. We also reported previously that the absence of the intact cytoplasmic domain of CD99 impairs molecular trafficking³³, suggesting the molecular interaction of this domain with cytoskeletal components. However, the molecular mechanisms remain unclear.

Here, based on the possibility of CD99-cytoskeleton interaction and the importance of actin polymerization in IS formation, we investigate the hypothesis that CD99 may also play a role in cytoskeleton reorganization to achieve IS formation. Our results demonstrate that CD99 ensures the interaction between actin and microtubules, which is crucial for coordinated cytoskeletal reorganization. CAR-T cells containing CD99 functional domains in the CAR structure exhibit enhanced IS formation and functional efficacy. Thus, this study identifies CD99 as a key membrane protein for IS formation and proposes a novel IS-stabilizing CAR design that could be generally applicable to enhance CAR-T cell therapy.

Results

Requirement of CD99 for IS formation and actin reorganization

To test the aforementioned hypothesis, we first examined the impact of the absence of CD99 on IS formation by comparing T-APC conjugate formation by CD99-deficient (Cd99^−/−) and wild type (WT) murine T cells in vitro. When cells from a WT CD8 cytotoxic T lymphocyte line specific for a minor histocompatibility antigen (MiHA) H60^34,35,36 (WT H60-CTL) were incubated with H60-peptide-presenting dendritic cells (H60-DCs), T-APC conjugates were formed, with binding of about three T cells on average per DC and > 25% of the H60-DCs being occupied by three or more WT H60-CTLs per DC (Fig. 1a). In contrast, Cd99^−/− H60-CTLs did not form such conjugates with H60-DCs, suggesting that CD99 is required for T-APC conjugate formation. Similar results were obtained with naïve TCR-transgenic J15^37,38 or OT-1 T cells specific for H60 or ovalbumin (OVA) complexed with H-2K^b, respectively (Supplementary Fig. 1a, b), implying that the CD99 dependency of T-APC conjugate formation is independent of antigen-experience and the antigen-specificity of T cells. In subsequent experiments, Cd99^−/− and WT H60-CTLs (hereafter, Cd99^−/−-cells and WT-cells, respectively) were used representatively. Cd99^−/−- and WT-cells exhibited comparable cell sizes and expressed similar levels of TCR and LFA on their cell surfaces prior to TCR-stimulation (Supplementary Fig. 1c). After T cell stimulation with APCs, however, the polarization of polymerized actin, TCR, and LFA-1 to the contact sites with cognate H60-DCs, indicators of mature IS formation in WT-cells, were not observed in Cd99^−/−-cells, in fluorescence staining experiments (Fig. 1b, Supplementary Fig. 1d).

The effects of CD99 on IS-forming actin dynamics were then examined in a longitudinal real-time actin-tracing experiment using Lifeact-mCherry³⁹-expressing retrovirus-transduced Cd99^−/−-cells and WT-cells that were activated on anti-CD3 antibody-coated plates. Confocal live-cell imaging of WT-cells showed that the lamellipodial actin ring expanded in the early phase (3:40 image in Fig. 1c), followed by the establishment of a radially symmetric actin network, typical of IS, with centripetal actin retrograde flow (6:20 and 14:20 images in Fig. 1c) in 10–15 min post-stimulation (Supplementary Movie 1). In Cd99^−/−-cells, however, despite initial actin polymerization, the typical IS actin network was not established within the ~30-min observation period, with detection of actin-free cell cortex area (Fig. 1c, Supplementary Movie 1). Both the basal and maximal intensities of filamentous actin (F-actin; polymerized) were significantly lower in CD3-stimulated Cd99^−/−-cells, with significantly smaller synapse size and lamellipodia width, compared to the WT counterparts (Fig. 1d, e). In addition, retrograde actin flow in Cd99^−/−-cells was not centripetal (after about 10:00 min in Supplementary Movie 1), so that dense actin-microclusters were rarely found in the central area of the cells (10:15 and 18:30 images for Cd99^−/−-cells in Fig. 1c, e). In line with these observations, although the TCR signal could be delivered by anti-CD3 stimulation in Cd99^−/−-cells, as demonstrated by nuclear translocation of NFAT and NFκB, and Erk phosphorylation in confocal microscopy of fixed cells, the degree of activation was far less than that in WT cells (Fig. 1f, Supplementary Fig. 1e–g). Taken together, these findings indicated that CD99 is required for the progression and amplification of TCR signaling-initiated actin dynamics leading to mature IS formation in T cells.

Requirement of CD99 for the physical interaction between actin and microtubules, and for microtubule reorganization

As actin flows toward the centrally localized MTOC in IS formation¹⁰, we assumed that the atypical localization of actin-microculsters in Cd99^−/−-cells reflected inappropriate microtubule reorganization. Therefore, we stained Cd99^−/−-cells, stimulated with plate-bound anti-CD3 antibody or H60-DCs, with anti-α-tubulin antibody to visualize the microtubules. As expected, Cd99^−/−-cells had fewer and less intensely stained perpendicularly growing microtubules (pMTs) compared to WT-cells, and showed impaired MTOC-translocation (Fig. 2a, supplementary Fig. 2a).

**Fig. 2: Requirement of CD99 for microtubule reorganization, and the physical interaction between actin and microtubules.**

Then, actin-microtubule interactions were examined in detail in a real-time microtubule-tracing experiment using Lifeact-transduced Cd99^−/−- and WT-cells stained with a tubulin-visualizing agent SiR-tubulin⁴⁰. Upon plate-bound anti-CD3 stimulation, the microtubules of WT-cells were newly polymerized, elongated, and tethered to the cell cortex, concomitant with lamellipodial extension (Fig. 2b, Supplementary Fig. 2b, Supplementary Movie 2). In Cd99^−/−-cells, however, there was little new microtubule outgrowth, and the microtubules were also untethered. Additionally, in WT-cells, shortening (contraction) or curling (bending) of pMTs near the cell cortex was seen, as noted previously¹⁶, followed by the generation of numerous and intensely staining pMTs and horizontally growing microtubules (hMTs), with an IS-typical microtubule network formed within 10–15 min. In contrast, in Cd99^−/−-cells, the new microtubules quickly fragmented and/or depolymerized, so that their staining intensity was low and their numbers declined over time (Fig. 2b, Supplementary Fig. 2b, Supplementary Movie 2). Therefore, CD99 is also crucial for appropriate microtubule dynamics following TCR-stimulation.

Importantly, simultaneous visualization of actin (Lifeact⁺) and microtubule (SiR-tubulin⁺) dynamics revealed that the two cytoskeletal components rarely interacted with each other in Cd99^−/−-cells, resulting in significantly less colocalization (indicated by a lower Manders’ coefficient of microtubules with F-actin) in the cell body (cytoplasmic core plus lamella) and lamellipodia than seen in WT-cells (Fig. 2c, Supplementary Fig. 2c, Supplementary Movie 3). Staining of the fixed cells with phalloidin and anti-α-tubulin antibody, in the absence of Lifeact and SiR-tubulin, reproduced colocalization of actin and microtubules, particularly in the peri-MTOC, lamella, and cortex regions of WT-cells, but not Cd99^−/−-cells (Fig. 2d). Structured illumination microscopy (SIM) further confirmed the presence of reduced levels of actin-microtubule colocalization and intensity of F-actin and microtubules in Cd99^−/−-cells compared to WT-cells in both the synapse plane and z-projection images (Supplementary Fig. 2d, Supplementary Movie 4). Interestingly, treatment of WT-cells with SMIFH2 or CK666, inhibitors of formin- or Arp2/3-mediated actin polymerization, respectively¹⁰, did not fully recapitulate the low levels of actin-microtubule colocalization and intensity of F-actin and microtubules observed in Cd99^-/--cells, despite inducing an overall reduction in the aforementioned parameters (Supplementary Fig. 2e). Furthermore, even when treated with phorbol 12-myristate 13-acetate and the ionophore ionomycin (PMA/Iono), agents that deliver artificial TCR signaling without TCR engagement and promote actin polymerization^41,42, Cd99^−/−-cells still exhibited lower microtubule and actin-microtubule colocalization levels compared to the WT-counterparts (Supplementary Fig. 2f), confirming that CD99 not only affects actin polymerization but also mediates proper microtubule reorganization and actin-microtubule colocalization. Consistently, the coimmunoprecipitation of actin and tubulin observed in WT-cells was not seen in Cd99^−/−-cells (Fig. 2e). Collectively, these results demonstrated that CD99 is critical for the physical interaction of actin and microtubules during IS formation, suggesting the importance of CD99 function-mediated actin-microtubule interaction for appropriate and coordinated reorganization of the two cytoskeletal components.

Importance of the cytoplasmic and transmembrane domains for CD99 function

Next, we analyzed the intracellular localization of CD99 to investigate how it functions in actin-microtubule interaction. Cd99^−/−-cells were transduced such that they expressed myc-tagged recombinant CD99-YFP fusion protein (rWT-CD99) as a reporter protein (Fig. 3a). Flow cytometric analysis confirmed expression of rWT-CD99 on the surface of stably transduced cells (rWT-cells) and showed comparable cell size and surface levels of TCR and LFA-1 between rWT-cells and mock-transduced cells (Supplementary Fig. 3a). Fixed-cell analysis of the rWT-cells showed that IS formation was successfully restored (as determined by the actin/microtubule colocalization index, actin ring width, pMT number, and MTOC-translocation), in contrast to mock-transduced Cd99^−/−-cells (Fig. 3b, Supplementary Fig. 3b, c). Live-cell imaging confirmed the proper dynamics and interaction of actin and microtubules in rWT-cells (Fig. 3c, Supplementary Fig. 3d, Supplementary Movie 5). Analysis of rWT-CD99 revealed its molecular localization in the cell body (but not in the MTOC area) and lamellipodia (Fig. 3c, Supplementary Movie 6), and suggested pSMAC and dSMAC localization in the IS. The pSMAC localization of CD99 was confirmed by positive costaining of endogenous CD99 with LFA-1, but not with the TCR, on the surface of WT-cell ISs (Supplementary Fig. 3e). Furthermore, rWT-CD99 was markedly enriched in the peri-MTOC, lamella, and cortical regions (Supplementary Fig. 3f), i.e., major sites of colocalization between actin and microtubules (Fig. 2d), in rWT-cells. Therefore, three-way colocalization of rWT-CD99, actin, and microtubules in these regions was seen frequently (Fig. 3d), suggesting that CD99 can interact simultaneously with these cytoskeletal proteins.

Insight into the mechanism of action of CD99 was obtained in studies using two mutants, Cyt-Mut-CD99 and TM-Mut-CD99, in which either the cytoplasmic or transmembrane domain was replaced with the corresponding nonfunctional domain from human CD4 (Fig. 3a). When the two mutants were introduced into Cd99^−/−-cells (Cyt-Mut-cells or TM-Mut-cells), no substantial differences in cell size and surface levels of TCR and LFA-1 were observed between the two mutant-cells and rWT-cells in the unstimulated state (Supplementary Fig. 3a). However, upon stimulation, the cytoskeleton reorganization for IS formation, defined according to the aforementioned criteria, was largely unrestored in both mutant-cells, as determined in both fixed (Fig. 3b, Supplementary Fig. 3b, c) and live (Fig. 3c, e, f, Supplementary Movie 6) cells. In addition, both mutants showed significantly less colocalization with actin and microtubules compared to rWT-CD99 (Fig. 3g), indicating that both domains are required for the function of CD99. However, the two mutants exhibited distinct defects. While Cyt-Mut-CD99 did not localize at the cell cortex, TM-Mut-CD99 was rarely detected at the cell body (Fig. 3c, Supplementary Movie 6). This was confirmed by flow cytometric analysis which showed a defect in cell-surface expression of Cyt-Mut-CD99, consistent with the defect in molecular trafficking to the plasma membrane of the mutant³³, while TM-Mut-CD99 showed increased cell-surface expression (Supplementary Fig. 3a). Furthermore, Cyt-Mut-CD99 showed drastic reduction of colocalization with actin, whereas colocalization of TM-Mut-CD99 with microtubules was greatly reduced (Fig. 3c, g, Supplementary Movie 7). These differences in cellular localization and impaired colocalization with either actin or microtubules between the two mutants were not due to aberrant functioning of the substituted human CD4 domains, since expression in Cd99^−/−-cells of alternative mutants containing human CD19 domains, designated Cyt^hCD19-Mut and TM^hCD19-Mut, resulted in the same effects (Supplementary Fig. 3g). These observations indicated that the cytoplasmic and transmembrane domains make different contribution to the function of CD99.

A detailed analysis of cytoskeleton dynamics revealed significantly fewer pMTs in Cyt-Mut-cells, with defective growth of peri-MTOC microtubules and untethering of microtubules from actin (Fig. 3c, f, Supplementary Fig. 3h, Supplementary Movie 6). The lack of actin-microtubule interaction required for lamellipodial extension resulted in highly unstable actin dynamics and an abnormal predominance of actin-free cortex (Fig. 3c, Supplementary Fig. 3h). Full lamellipodia extension was transient and followed by rapid retraction (Supplementary Fig. 3i). These observations were consistent with a previous report that actin-microtubule linkage enables coordinated microtubule growth and actin filament elongation⁴³. Therefore, CD99 was suggested to use its cytoplasmic domain to support interdependent actin and microtubule growth probably by actin-guided microtubule growth and microtubule-induced actin polymerization, thus enabling lamellipodial extension.

In TM-Mut-cells, actin polymerization persisted near the cell cortex, where TM-Mut-CD99 resides, albeit expressed at low levels (Fig. 3c and Supplementary Movie 7). Additionally, there were a greater number of long pMTs than in Cyt-Mut-cells, although the overall number of pMTs was lower than in rWT-cells (Fig. 3b, c, e). A low level of actin-microtubule colocalization, as detected in the peri-MTOC area (Fig. 3b, c, f), may facilitate pMT generation. This supports a role for the cytoplasmic domain (which is intact in TM-Mut-CD99) in pMT generation and actin polymerization in lamellipodia. However, tethering of pMTs to actin at the cell cortex was absent, and pMTs in the cortical region were broken and decayed over time, so that intense staining of pMTs and hMTs was not observed (Fig. 3c, Supplementary Fig. 3j). Due to this lack of pMT-actin interaction, periodic retraction of the lamellipodia was reduced, and the actin accumulation necessary to form an actin-arc network near the lamella was hindered (Fig. 3c, Supplementary Fig. 3i). These results indicated that the transmembrane domain of CD99 is required for stable tethering between actin and preformed pMTs near the cell cortex and for the retraction of actin and extended lamellipodia.

In summary, our results indicated that in the cell body, the cytoplasmic domain of CD99 is required for pMT formation and F-actin extension, while the transmembrane domain is for microtubule retention (Fig. 3h). In lamellipodia, the cytoplasmic domain of CD99 is needed for actin polymerization and lamellipodial extension, while the transmembrane domain is for pMT-tethering to cortical actin, which allows lamellipodia retraction.

Proteomic identification of IQGAP1 and myosin IIA as domain-mediated interaction partners of CD99

The molecular mechanism of action of CD99 was further investigated by proteomics analysis using coprecipitates of rWT-CD99. As expected, cytoskeletal proteins, including various components of the actomyosin (Actbl2, Iqgap1, Coro1a, Actc1, Myh9, Myl6), microtubule (Tuba1, Tuba4a, Tubb4b, Tubb5), and molecular trafficking (Fmnl, Rac2, and Rab27) systems, were identified as a major subset of the CD99 interactome (Fig. 4a, Supplementary Fig. 4a, b, Supplementary Data 1–4).

**Fig. 4: Identification of IQGAP1 and Myosin IIA as domain-mediated interaction partners of CD99.**

Next, we assessed the domain-specific interactions of CD99 with cytoskeletal components by comparing coimmunoprecipitates of Cyt-Mut-CD99 and TM-Mut-CD99 with those of rWT-CD99. Principal component analysis (PCA) of the interactome data revealed clear clustering of replicates and distinct separation between the three recombinant protein samples (Supplementary Fig. 4c). In general, cytoskeletal interactors were present at lower levels in the coprecipitates of both mutants than in those of rWT-CD99, with less binding to Cyt-Mut-CD99 than to TM-Mut-CD99 (Fig. 4b, Supplementary Fig. 4d, Supplementary Data 1, 5–7). Exceptionally, however, myosin (MyoII) proteins (Myh9 heavy (MyoIIA) and Myl6 light chains) showed less binding to TM-Mut-CD99 than Cyt-Mut-CD99, indicating that myosin proteins preferentially bind to the transmembrane rather than the cytoplasmic domain of CD99 (Fig. 4b, c). In contrast, IQGAP1 showed less binding to Cyt-Mut-CD99 than TM-Mut-CD99, although binding to both mutants was relatively strong. Therefore, IQGAP1 was suggested to have greater affinity for the cytoplasmic than the transmembrane domain of CD99. This pattern was confirmed by immunoblotting of each coimmunoprecipitate (Fig. 4d). IQGAP1 is a candidate molecule that mediates actin-microtubule linkage with domains capable of binding to F-actin and microtubules⁴⁴. The actin motor protein MyoIIA has been reported to play a role in microtubule reorganization⁴⁵. Therefore, IQGAP1 and MyoIIA were inferred to be cytoplasmic and transmembrane domain-mediated principal interactors involved in CD99-associated actin-microtubule interaction. Indeed, both MyoIIA and IQGAP1, as well as actin and α-tubulin, coprecipitated with CD99 in WT-cells (Fig. 4e), suggesting that these five proteins form a multiprotein complex.

CD99 facilitates interactions between IQGAP1, myosin IIA, and the cytoskeleton

To gain further insight into the above biochemical interactions, rWT-cells and mutant-cells were stained for the localization of IQGAP1 and MyoIIA, and examined by confocal microscopy. In rWT-cells, IQGAP1 and MyoIIA were enriched in the lamella, as well as in peri-MTOC and cortical regions, of the IS (Supplementary Fig. 4d), as reported previously^46,47. Additionally, rWT-CD99 colocalized with these two molecules in these regions (Fig. 4f, g). In contrast, the colocalization of Cyt-Mut-CD99 with IQGAP1 was greatly reduced, as was that of TM-Mut-CD99 with MyoIIA (Fig. 4f, h, Supplementary Fig. 4e), consistent with the biochemical data. Notably, the colocalization of MyoIIA and IQGAP1, as well as the three-way colocalization of MyoIIA, IQGAP1, and rWT-CD99, was significantly reduced in Cyt-Mut- and TM-Mut-cells, compared to rWT-cells (Fig. 4f, h, Supplementary Fig. 4e). The requirement of CD99 in the MyoIIA-IQGAP1 interaction was further supported by the reduced colocalization and coprecipitation of these two molecules in Cd99^−/−-cells (Fig. 4i, j, Supplementary Fig. 4f).

As CD99 mediates both actin-microtubule and MyoIIA-IQGAP1 interactions, we sought to determine whether CD99 is necessary for the interaction of the actin modulators MyoIIA and IQGAP1 with microtubules. The colocalization of microtubules with IQGAP1 or MyoIIA was significantly reduced in Cd99^−/−-cells compared to WT-cells, such that neither MyoIIA nor IQGAP1 coprecipitated with α-tubulin (Fig. 4k, l). These results suggest a role for CD99 as a hub protein that coordinates actin-microtubule crosstalk by binding to IQGAP1 and MyoIIA and modulating their interaction (Fig. 4m).

Ineffective T cell activation and in vivo immunity in the absence of CD99

Given the significance of IS for T cell activation and effector function⁷, the effects of CD99 deficiency on the efficiency of T cell activation were investigated. After stimulation with anti-CD3 and CD28 antibodies, polyclonal T cells, both CD4 and CD8 T cells, from Cd99^−/− mice were capable of dividing, but showed delayed cell division (CFSE-dilution) and activation (upregulation of CD69, CD25, and CD44), compared to the WT counterparts (Fig. 5a, b, Supplementary Fig. 5a, b). This was also the case in antigen-specific T cells, i.e., Cd99^−/− OT1 T cells stimulated with OVA-DCs (Supplementary Fig. 5c). Consistently, TCR signaling events, including the phosphorylation of ZAP70 and Erk, nuclear translocation of NFAT and NFκB, and sustained Akt activation, were all reduced in Cd99^−/− T cells (Supplementary Fig. 5d–f), recapitulating the biochemical features observed in Cd99^−/− CTL cells (Fig. 1f, Supplementary Fig. 1e–g). Furthermore, Cd99^−/− CD8 T cells exhibited lower expression levels of FasL and granzyme B compared to the WT-counterparts (Supplementary Fig. 5g).

Whether the activation of Cd99^−/− T cells was also delayed in vivo was investigated by the adoptive transfer of WT or Cd99^−/− H60-specific J15 (Thy1.2⁺) T cells into B6 mice, which were then injected intraperitoneally with H60-expressing splenocytes from H60-congenic mice³⁸. In response to H60-challenge, the expansion of Cd99^−/−J15 T cells in the spleen and the migration of these cells to the antigen-bearing site (the peritoneum) in the host were slower, compared to the WT counterparts (Fig. 5c, Supplementary Fig. 5h). Based on this inefficient reactivity of Cd99^−/− T cells for MiHA H60, whether CD99 deficiency in T cells would delay the rejection of MiHA-mismatched allografts was investigated using a well-established B6 anti-BALB.B skin-allograft model^48,49. When BALB.B tail-skin was transplanted onto WT or Cd99^−/− B6 recipients, Cd99^−/− recipients showed a significant delay in allograft rejection, with the T cells specific for MiHAs (H60 and H4) showing lower peak levels in the blood compared to WT recipients (Fig. 5d).

The physiological relevance of CD99 for T cell activation was further investigated in an antitumor T cell therapy model. WT OVA-specific OT-1 (CD45.1⁺) T cells, when adoptively transferred, reduced tumor growth in OVA-expressing B16 melanoma-bearing mice, which was not the case with Cd99^−/− OT-1 T cells (Fig. 5e). Consistently, Cd99^−/− OT-1 T cells were detected at lower levels in host blood compared to WT OT-1 cells. Collectively, these results suggest that CD99 deficiency in T cells limits the efficiency of T cell immunity in vivo, probably through defective CD99-mediated actin-microtubule crosstalk. However, as CD99 has been known to be also important for transendothelial migration of leukocytes⁵⁰, it should be noted that defective migration of T cells in the absence of CD99 cannot be formally excluded.

Transmembrane and juxtamembrane domains are minimal essential domains of mouse and human CD99

Having confirmed a critical role of CD99 in IS formation and T cell function, we investigated whether incorporation of the CD99 transmembrane and cytoplasmic domains into a tumor-targeting CAR would enhance the antitumor efficacy of CAR-T cells by potentiating IS formation. To minimize the CD99 components to be incorporated into the CAR, additional domain studies were performed. First, two additional transmembrane domain mutants, TM-SR (small-portion restored) and TM-LR (large-portion restored), were generated, in which small and large portions of the human CD4 transmembrane domain of TM-Mut-CD99 were restored to mouse CD99 sequences, respectively (Fig. 6a). In transduced Cd99^−/− cells, greater inward localization of mutant-CD99 from the cell cortex was achieved with TM-LR than TM-SR (Fig. 6b, Supplementary Fig. 6a). Concomitantly, hMT generation and actin-microtubule colocalization were gradually restored in the order TM-Mut < TM-SR < TM-LR. However, complete restoration was achieved only in rWT-cells, indicating that the intact transmembrane domain is essential for CD99 function.

**Fig. 6: Identification of the transmembrane and juxtamembrane domains as minimal essential domains of mouse and human CD99.**

A comparison of CD99 amino acid sequences from different species showed that the transmembrane and cytoplasmic domains, especially the transmembrane-juxtaposing sequences (hereafter referred to as Juxta, i.e., the juxtamembrane region), are highly conserved (Fig. 6c). Therefore, we generated a mutant (TJ-CD99) that retained the transmembrane and Juxta (TJ) regions but with deletion of the rest of the cytoplasmic domain (Fig. 6d). When introduced into Cd99^−/− cells, TJ-CD99 fully restored the colocalization and reorganization of actin and microtubules, as well as molecular localization, similar to rWT-CD99 (Fig. 6d, Supplementary Fig. 6b). Thus, the TJ region of CD99 is essential for IS formation.

Considering application to human CAR-T-cell generation, we then investigated whether the mouse CD99 findings could be recapitulated in human T cells. Human CD99 exists as two isoforms resulting from alternative splicing, one with a short cytoplasmic domain and the other with a long cytoplasmic domain, but both isoforms share the transmembrane and juxtamembrane regions²¹, enabling the evaluation of a single human TJ mutant. First, the function of human CD99 was tested using human Jurkat T cells. After transfection of an anti-CD99 short hairpin RNA (shRNA) plasmid selected for the effective knockdown (KD) of human CD99 (Supplementary Fig. 6c–e), the CD99 knocked-down (KD-HuCD99) Jurkat T cells exhibited defective formation of the actin and microtubule networks of the IS, in contrast to control cells transfected with scramble-shRNA (Fig. 6e). Furthermore, when the human TJ-mutant of CD99 (HuTJ-CD99-YFP) was expressed in KD-HuCD99-Jurkat cells, both actin and microtubules were properly reorganized (Fig. 6f, Supplementary Fig. 6f). HuTJ-CD99-YFP molecules localized to the cell body and cell cortex, where they were colocalized with actin and microtubules. Next, we extended our analysis to human primary T cells, utilizing alloreactive T cells generated through a mixed leukocyte reaction (MLR) of peripheral blood mononuclear cells (PBMCs) from allogeneic individuals. These alloreactive T cells were transfected with CD99 single guide RNA (sgRNA) to induce CD99 gene deletion, generating sgCD99-T cells. Upon TCR-stimulation, sgCD99-T cells also exhibited defects in actin and microtubule network formation of IS, including the polarization of F-actin and MTOC toward the coincubated target allogeneic cells (Supplementary Fig. 6g, h). Furthermore, sgCD99-T cells showed reduced biochemical and phenotypic readouts of TCR-signaling events, including nuclear translocation of NFAT and NFκB, phosphorylation of ERK, sustained activation of AKT, and upregulation of activation markers, compared to sgControl-T cells (Supplementary Fig. 6i-k). Altogether, these results provide further evidence for the conserved function of CD99 and the essentiality of the TJ region in human T cells.

Potentiated IS formation and antitumor effects via CD99 domain-incorporated CAR design

Finally, we evaluated whether the function of human CAR-T cells bearing the CD99-TJ region would be enhanced compared to conventional CAR-T cells. In the conventional CAR structure, such as the BBz-CAR targeting human CD19-expressing cells⁵¹, the antibody moiety is linked to the signaling cytoplasmic domains of 41BB and CD3 zeta chain via the extracellular and transmembrane domains of CD8 (Fig. 7a). These CD8-derived domains serve only as a physical linker rather than a functional entity. Therefore, we designed a new BBz variant-CAR, named HuTJ-BBz-CAR, incorporating the functional extracellular and TJ domains of human CD99 instead of the CD8-derived linker domains (Fig. 7a). CAR-T cells expressing BBz-CAR or HuTJ-BBz-CAR, with comparable CAR expression levels, were utilized in subsequent experiments (Fig. 7b). IS formation and effector function were then compared between HuTJ-BBz- and conventional BBz-CAR-T cells at multiple time points after coincubation with target (CD19⁺) Raji human lymphoma cells. Consistent with normal IS formation between T cells and APCs, cell-cell conjugate formation with the target cells was seen more frequently, and actin and MTOC were more polarized to the target cells in BBz-CAR-T cells than in untransduced (UnTD) control T cells (Fig. 7b, c, Supplementary Fig. 7a). Notably, HuTJ-BBz-CAR-T cells exhibited more efficient cell-cell conjugate formation, and earlier completion of F-actin polarization and MTOC-translocation than BBz-CAR-T cells (Fig. 7b, c, Supplementary Fig. 7a). HuTJ-BBz-CAR-T cells, both the CD4 and CD8 fractions, upregulated CD69, CD25, and CD44 to greater extents than BBz-CAR-T cells after the coculture with Raji tumor cells (Fig. 7d, supplementary Fig. 7b, c). The enhanced activation of HuTJ-BBz-CAR-T cells relative to BBz-CAR-T cells was maintained even after the deletion of the T cell’s endogenous CD99 gene (Supplementary Fig. 7d). Notably, CD99 deletion did not significantly impact the activation levels of either HuTJ-BBz-CAR or BBz-CAR-T cells, suggesting that CAR-signaling does not strictly depend on the function of endogenous CD99. Then, when subjected to long-term tumor-killing assay, HuTJ-BBz-CAR-T cells expanded greatly to significantly higher levels and exerted superior antitumor effects in comparison with BBz-CAR-T cells (Fig. 7e, Supplementary Fig. 7e). Regarding safety, although there has been some controversy in the literature regarding the oncogenic and oncosuppressive roles of CD99²⁵, in the absence of tumor target cells and cytokine supplementation during long-term culture, both HuTJ-BBz- and BBz-CAR-T cells, as well as UnTD-T cells, exhibited a continuous decline in cell numbers, suggesting their minimal oncogenic potential (Supplementary Fig. 7f). Collectively, these results indicated that the incorporated CD99-derived domains were functional in CAR-T cells, mediating potentiated IS formation, activation, and tumor-killing efficiency of HuTJ-BBz-CAR-T cells.

CD99 domain-incorporated CAR enhances CAR-T cell efficacy in vivo

The in vivo antitumor efficacies of the two lines of human CAR-T cells were compared in Raji tumor-bearing immunodeficient NSG mice. Tumor burden was monitored by implantation of luciferase-expressing Raji cells. After infusion of BBz-CAR-T cells (5 × 10⁶ cells), tumor growth was significantly suppressed by day 7 post-infusion (Fig. 8a). However, tumors re-grew and eventually killed all the mice (Fig. 8a, b). Notably, the infusion of HuTJ-BBz-CAR-T cells resulted in earlier tumor eradication and a much better survival rate for the mice than was the case with BBz-CAR-T cells (Fig. 8a, b). At the lower dose of CAR-T cells (1 × 10⁶ cells), with which BBz-CAR-T cells could not control tumor growth, HuTJ-BBz-CAR-T cells still completely eliminated tumors, indicating more than five-fold enhanced efficacy (Fig. 8c, d). HuTJ-BBz-CAR-T cells expanded with much faster kinetics in the peripheral blood than BBz-CAR-T cells, for both CD4 and CD8 fractions, in that HuTJ-BBz-CAR-T cells peaked on day 7, whereas BBz-CAR-T cells began to show major expansion on day 14 (Fig. 8e). Therefore, efficient expansion and tumor-killing activity of HuTJ-BBz-CAR-T cells were observed in vivo as well as in vitro. Altogether, the incorporation of CD99 TJ domains into the CAR improved CAR-T functionality, suggesting that CD99 functional domain-mediated IS stabilization enhances CAR-T cell functionality.

Discussion

This study demonstrated that CD99 mediates the physical interaction between actin and microtubules and that this interaction is critical for mature IS formation and effective T cell immunity, leading to the novel design of an IS-stabilizing CAR structure.

CD99 contributes to several biological processes and inflammatory responses, such as T cell adhesion^21,52,53; activation and differentiation^26,27; intracellular membrane protein trafficking in T, B, and epithelial cells^{28,29,30,33,54}; and lymphocyte diapedesis into inflamed vascular endothelium^32,50. Nevertheless, the uncertainty surrounding its molecular function and mechanism of action has hindered the acknowledgement of CD99 in the immunological field. Actin reorganization has been considered a molecular target of CD99 action, as the aforementioned phenomena are related to the actin cytoskeleton, and stimulation of CD99 by an agonistic antibody has been shown to induce action polymerization^28,29. Therefore, the requirement of CD99 for proper actin reorganization during IS formation is unsurprising. However, the contribution of CD99 to microtubule dynamics is a novel and completely unexpected finding. Furthermore, the observed physical and functional dissociation of actin and microtubules in the absence of CD99 strongly suggests that CD99 is the long-sought master-controller membrane protein that orchestrates cytoskeleton reorganization during IS formation - a protein that has been proposed but remained unidentified^55,56. Regarding human CD99, the alternatively spliced isoforms can exhibit divergent functions, depending on the cell type²¹, with both isoform being recruited to the IS in activated T cells³¹. Given that both isoforms share functional domains (TJ regions), we propose that the core function of CD99—namely, its role in controlling the reorganization and cross-talk between actin and microtubules—would be maintained, irrespective of the cooperative or counteractive relationships these isoforms may exhibit within the IS.

We identified the domains of CD99 responsible for the actin-microtubule linkage and their interactors associated with the linkage: the transmembrane domain for microtubule interaction via MyoII and the cytoplasmic domain for actin interaction via IQGAP1. IQGAP1 is a candidate mediator of the actin-microtubule linkage^11,44. Our identification of IQGAP1 as a binding partner of the cytoplasmic domain of CD99 is consistent with the above consensus. However, despite its ability to interact with IQGAP1, TM-Mut-CD99 did not fully restore the actin-microtubule linkage, indicating the involvement of additional factors such as MyoII. MyoII is an actin motor protein known to be primarily involved in lamellar actin-arc contraction in IS⁴⁵. However, the failure of TM-Mut-CD99 to interact with MyoIIA and restore pMT-actin linkage particularly at the cell cortex suggests that MyoII also plays an important role in actin-microtubule linkage at the cell cortex. This is supported by the results of general cell biology studies indicating that MyoII forms a contractile ring complex with actin and microtubules at the cell cortex when a cell divides, and generates the contraction force needed to close the ring during cytokinesis^57,58. Additionally, recruitment of IQGAP1, formin, lipids, and a membrane protein to the ring complex has been reported^59,60,61. Therefore, CD99 is likely to be “the” membrane protein component of the ring complex. We have also reported defective cytokinesis in some human CD99-downregulated tumor cells⁶². Taken together, these observations suggest that CD99-MyoIIA-IQGAP1 constitutes the general machinery for the physical and functional coupling of actin and microtubules in many cellular responses.

During IS formation, the actin-microtubule linkage at the peri-MTOC and cell cortex has been proposed to provide the traction force driving translocation of the MTOC to the synapse site^11,55. However, the results of our cytoskeleton dynamics analyses suggest that the actin-microtubule physical interaction has important functional roles beyond simple mechanical traction. At the peri-MTOC, the CD99-mediated interaction enables concurrent pMT generation/growth and actin ring expansion, as inferred from the defects in Cyt-Mut-cells. At the cell cortex, the linkage mediates microtubule contraction, leading to lamellipodial retraction, and the associated retrograde actin flow and then lamellar actin-arc formation, as inferred from the defects in TM-Mut-CD99 cells, which to our knowledge has not been reported previously. Furthermore, at the lamella, we propose that depolymerization of microtubule tips may lead to actomyosin contraction via linked F-actin, enabling inward flow of actin-microclusters across the lamella^10,45,63. Similar control of actomyosin contractility by microtubule depolymerization has been demonstrated in cell migration⁶⁴. Collectively, our findings highlight the significance of CD99-mediated actin-microtubule physical interaction in IS structure formation, providing mechanistic insight into the process of IS formation.

Incorporation of the functional domains of CD99 into a conventional BBz-CAR structure significantly improved the efficacy of CAR-T cells, potentiating IS formation by CAR-T cells. This validated the functionality of CD99 domains in the context of CAR-IS, underscoring the significance of IS formation as a mechanistic component of CAR-T cell functionality. Previous studies have demonstrated a positive correlation between IS quality and CAR-T cell performance in vivo⁶⁵, and proposed the need for an IS-stabilization strategy to enhance CAR-T cell functionality^66,67. However, there have been few reports on such strategies. Our identification of the molecular function and mechanism of action of CD99 represents a key advance enabling the development of IS-potentiating CAR-T cells.

Despite recent advances in CAR design, most CARs use the transmembrane domain as the physical linker between the extracellular antibody moiety and the intracellular signaling domains, with only the cytoplasmic domains being modified. The improvement of the efficacy of CAR-T cells by our novel CAR design, involving replacement of the CD8-linker domains with CD99 transmembrane and juxtamembrane domains, provides a new concept that the transmembrane domain can be transformed to boost CAR-T-cell functionality. This also has the advantage of making more space in the cytoplasmic domain available for further modification. It has been shown that CAR-T cells form nonclassical IS with indistinguishable pSMAC structure, rapidly recruiting lytic granules to the IS and quickly detaching from target cells^68,69. Considering the pSMAC-localization of CD99, we propose that the improved functionality of HuTJ-BBz-CAR T cells is a result of the gain-of-pSMAC-localizing CD99 functional domains, which would promote the establishment of classical and stable IS by strengthening actin-microtubule crosstalk in CAR-T cells. This is supported by enhanced activation marker upregulation and proliferation of HuTJ-BBz-CAR T cells.

In summary, we propose CD99 as a novel master regulator of IS stabilization via the actin-microtubule linkage, which could be applied in T-cell-mediated therapeutics.

Methods

Mice

C57BL/6 J (000664; B6), B6.PL-Thy1^a/CyJ (000406; Thy1.1⁺), B6.SJL-Ptprc^aPepc^b/BoyJ (002014; CD45.1⁺), C.B10-H2^b/LiMcdJ (001952; BALB.B), C57BL/6-Tg(TcraTcrb)1100Mjb/J (003831; OT-1) and NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (005557; NSG) mice were purchased from the Jackson Laboratory. H60 congenic mouse strain (B6.CH60; a kind gift from Dr. Derry Roopenian at the Jackson Laboratory, Maine, USA). J15 TCR transgenic mouse strain was developed in our laboratory^37,38. Cd99^−/− mouse strain (B6.Cd99^{Gt(pU-21T)44Imeg}; CARD ID #718)³³ were obtained from the Center for Animal Resources and Development (CARD) of Kumamoto University. All mice were housed in a specific pathogen-free animal facility at the Seoul National University College of Medicine and maintained in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC). The experimental use of animals was approved by the IACUC (SNU-221207-9-2, SNU-221207-5-1, SNU-221207-4-3, SNU-220704-9). Experimental and control group mice at the age of 8-12-week-old were housed separately in individually ventilated cage racks in the same room. Euthanasia was performed by CO₂ inhalation.

Cell culture

Cells were cultured with DMEM (for mouse cells) or RPMI-1640 (for human cells) media (Gibco) supplemented with 10% FBS (Hyclone), HEPES, sodium pyruvate, L-arginine, L-glutamine, 2-mercaptoethanol, non-essential amino, gentamycin, streptomycin, and penicillin under an atmosphere of 5% CO₂ at 37 °C.

Cell line establishment and cell engineering

H60-specific WT or Cd99^−/− CD8 CTL lines were established from WT or Cd99^−/− B6 mice immunized intraperitoneally with splenocytes from B6.CH60 mice, and maintained via weekly re-stimulation with irradiated B6.CH60 splenocytes⁴⁸. For introduction of recombinant CD99 molecules, early (<10) passage Cd99^−/− CTL lines were incubated with filtered retroviral supernatants harvested from Plat-E cells (Cell Biolabs #RV-101) transiently transfected with pMSCV-puro vector (Addgene #68469) containing Lifeact-mCherry, rCD99WT-YFP, Cyt-Mut-CD99-YFP, TM-Mut-CD99-YFP, Cyt^hCD19-Mut-CD99-YFP, TM^hCD19-Mut-CD99-YFP, TM-SR-CD99-YFP, TM-LR-CD99-YFP, or TJ-CD99-YFP gene, or an empty vector (mock expression), using the calcium phosphate transfection or polyethylenimine (PEI) method. After 2 additional days of culture in fresh medium, the transduced CTL cells were re-stimulated for passage in culture medium containing puromycin (Sigma-Aldrich) and selected for mCherry⁺ or YFP⁺ cells using the FACS Aria II (BD Biosciences). DC2.4 cells⁷⁰ transduced to express H60 (LTFNYRNL) or OVA (SIINFEKL), yielding H60-DC and OVA-DC, respectively, were cultured in the presence of puromycin. The B16F10-OVA cell line was generated by transduction of B16F10 murine melanoma cells (ATCC) with the lentivirus harboring the OVA-expressing puromycin-resistance cassette (pLEF-puro-OVA). The transduced cells were selected with puromycin and isolated as clones. The luciferase-GFP-expressing Raji cell line (Raji-Luc) were generated by spin infection with lentivirus (pLEF-LucGFP) harboring the luciferase-P2A-EGFP expression cassette. The GFP-high populations were sorted using the FACS Aria II.

In Jurkat cells, for knockdown of human CD99, anti-CD99 shRNA- or scramble shRNA-expressing plasmid (pSicoR-mCh; Addgene #21907) was transfected into the cells by electroporation using a Nucleofector2b device (Lonza). To express HuTJ-CD99, YFP- or HuTJ-YFP-expressing plasmid (pcDNA3.1; Thermo Fisher Scientific #V79020) was co-transfected with anti-CD99 shRNA-expressing plasmid (see Supplementary Table 1) using the method described above. Live cells were isolated 48 h post-transfection by Ficoll-gradient centrifugation (GE Healthcare #17-1440-02), stimulated on anti-human-CD3 antibody (OKT3; BioXcell #BE0001-2; 10 μg/ml)-coated coverslips, and observed using confocal microscopy.

For human primary T cell experiments, human peripheral blood mononuclear cells (PBMCs) from healthy donor leukapheresis samples were subjected to mixed lymphocyte reaction (MLR) with irradiated (20 Gy) human PBMCs from allogeneic leukapheresis samples to induce alloreactive T cell activation. To delete the CD99 gene, T cells were electroporated on day 4 of MLR with Cas9 ribonucleoprotein containing sgRNA targeting CD99 (cggcgaccagaacacccagc, synthesized by IDT) using the Neon Transfection System (Thermo Fisher Scientific). A scrambled sgRNA (sgControl; #1072544, IDT) was used as a control. The transfected T cells were rested in complete medium for another 6 days and restimulated with irradiated allogeneic PBMCs for 10 days for further expansion. The CD99-deleted alloreactive T cells were then sorted using the SH800S cell sorter (Sony Biotechnology) and further expanded by 1-2 rounds of allogeneic PBMC stimulation at 10-day intervals before use.

Cell-cell conjugate formation assay

Live B6 WT and Cd99^−/− CTLs purified by Ficoll-gradient separation at day 3 after re-stimulation for passage were stained with Cell Tracker green CMFDA (5-chloromethylfluorescein diacetate; Thermo Fisher Scientific #C7025) according to the manufacturer’s directions. CMFDA-labeled T cells (1 × 10⁵) and H60-DC or OVA-DC cells (2 × 10⁴) labeled with Cell Tracker orange CMTMR (5-(and-6)-(((4-chloromethyl)benzoyl)amino) tetramethylrhodamine; Thermo Fisher Scientific #C2927) were co-incubated on poly-L-lysine (Sigma-Aldrich #P4832)-coated coverslips at 37°C and examined by fluorescence microscopy. T cells from J15 and OT-1 mice on a WT or Cd99^−/− B6 background were purified by negative magnetic-activated cell sorting (MACS; Miltenyi Biotec) and then co-incubated with H60-DCs or OVA-DCs, as described above. After incubation, the cells were fixed with 4% paraformaldehyde for 15 min, washed, and viewed under a microscope (IX71; Olympus).

Confocal and structured illumination microscopy and confocal time-lapse live imaging

To visualize the IS, WT and Cd99^−/− CTLs were pre-stained with fluorescein isothiocyanate (FITC) or phycoerythrin (PE)-conjugated anti-CD99 (EJ2; Dinona; 1:200) together with FITC-, or PE-conjugated anti-TCRβ (H57-597; Thermo Fisher Scientific #11-5961-81; 1:500) or anti-LFA1 (2D7; BD Biosciences #553120; 1:500) antibody for 1 h at 4 °C. H60-DCs were stained with Alexa 647-conjugated anti-ICAM1 antibody (YN1/1.7.4; Thermo Fisher Scientific #12-0541-81; 1:500). Pre-stained T and H60-DC cells were co-incubated on the same poly-L-lysine coated coverslips; after fixation and washing, they were viewed under a FluoView 1000 confocal microscope (Olympus). For the imaging of fixed cells, T cells (1 × 10⁵) loaded onto anti-CD3 antibody (10 μg/ml; 145-2C11, BD Biosciences #553058)-coated coverslips or plated together with H60-DCs on poly-L-lysine coated coverslips were incubated for 7, 15, 30 min and fixed with 4% paraformaldehyde. For chemical treatment, T cells pretreated with the actin polymerization inhibitors CK666 or SMIFH2 were stimulated on an anti-CD3-coated plate for 15 min before fixation. In another experiment, T cells were treated with PMA (20ng/ml; Sigma-Aldrich #P8139) and Ionomycin (10 μM; Sigma-Aldrich #I0634) for 15min and fixed. The fixed cells were permeabilized with 0.25% Triton in phosphate-buffered saline (PBS) for 10 min and blocked with 10% bovine serum albumin (BSA; GeorgiaChem #BS1005) for 1 h at room temperature. Then, cells were stained with anti-myosin IIA (Cell Signaling Technology #3403; 1:200), anti-IQGAP1 (C-9; Santa Cruz Biotechnology #sc-376021; 1:200), anti-NFAT1 (D43B1; Cell Signaling #5861; 1:200), anti-NFκB p65 (D14E12; Cell Signaling #8242; 1:200), phosphor-Erk1/2 (D13.14.4E; Cell Signaling #4370; 1:200), or anti-α-Tubulin (DM1A; Millipore #05-829; 1:500) antibody (see Supplementary Table 2), washed, and stained with Alexa 647-conjugated phalloidin (Thermo Fisher Scientific #A22287; 1:1000), fluorescence-conjugated secondary antibodies (Thermo Fisher Scientific), and 4’,6-diamidino-2-phenylindole (DAPI; 0.5 μg/ml; Sigma-Aldrich #D9542). For Jurkat cell imaging, cells were stimulated for 15 min on cover slip coated with anti-human CD3 antibody (10 μg/ml; OKT3, BioXcell), and for human primary T cell analysis, cells were either stimulated on the coverslips or co-incubated with irradiated allogeneic PBMCs for 30 min prior to imaging. To perform super-resolution 3D structured illumination microscopy (SIM), the N-SIM system (Nikon) was used. 3D-SIM images of the microtubule and actin cytoskeleton were acquired with a 100×/1.49 CFI Apochromat TIRF oil-immersion objective, using excitation wavelengths of 488 nm and 561 nm. Z-stacks were acquired at 120 nm intervals to scan the entire cell. For live-cell imaging, the sorted cells stably transduced with Lifeact-mCherry for F-actin visualization were isolated by Ficoll separation and incubated for 1 h with 1 μM of SiR-Tubulin dye (Spirochrome AG #CY-SC002) containing 10 μM of the efflux pump inhibitor verapamil (Spirochrome AG #CY-SC002). After the CTLs were loaded onto an anti-CD3 antibody-coated coverslip, time-lapse images at 100× were collected every 20 s for 30–35 min using FluoView 3000 confocal microscope (Olympus).

Image analysis

Single-plane or z-stacked images were generated from raw microscopy data by exporting with FIJI ImageJ (NIH) or rendering with IMARIS (Bitplane). To remove non-specific background fluorescence, unstained control images were acquired using the same imaging settings, and their fluorescence intensities were subtracted. For fluorescence intensity measurements, a single cell was selected as the region of interest (ROI) by automatically defining the cell boundary on the thresholded F-actin image, and cell size (area) was also measured from the same image. Specific parts of the cell were selected using the rectangular selection tool. Distance and thickness measurements were performed using the line tool. Intensity was quantified based on these defined areas. For quantification of NFAT or NFκB nuclear translocation, the nucleus region was defined as the ROI by thresholding the DAPI channel, and fluorescence intensity of NFAT or NFκB was measured within this ROI. Colocalization analysis was conducted by calculating the Manders’ coefficient M1 using IMARIS⁷¹ and Pearson’s coefficient using ImageJ and the JACoP plug-in. All time-lapse images were analyzed using IMARIS. To analyze F-actin retraction dynamics, time-lapse images were aligned and straightened using ImageJ software, and the Multi Kymograph plug-in was used to generate kymographs. In most cases, the bottom layer along with one or two (when necessary) upper layer images were presented as a synapse plane image and used for analysis, unless otherwise specified in the figures and figure legends.

In vitro T cell stimulation, co-immunoprecipitation, and immunoblotting

Lymph node or splenic T cells (2 × 10⁶/well) from WT or Cd99^−/− mice were resuspended in pre-warmed serum-free DMEM, and stimulated with anti-CD3 (10 μg/ml) and anti-CD28 (0.5 μg/ml; 37.51, BD Biosciences #556620) antibodies at 37 °C for 0, 2, 5, 15, or 30 min before adding cold PBS containing 1 mM sodium orthovanadate. After centrifugation, cell pellet was resuspended in lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM EDTA, 10% Glycerol, 1% NP-40, 1 mM Na₃VO₄, 2 mM NaF, Protease inhibitor cocktail) and incubated on ice for 20 min cell lysis. Cell lysates were subjected to immunoblotting using anti-Zap70, anti-pZap70, anti-ERK1/2, and anti-pERK1/2 (Cell Signaling Technology) antibodies, and horseradish peroxidase-conjugated anti-rabbit (Santa Cruz Biotechnology) or anti-mouse (Jackson ImmunoResearch Labs) secondary antibody. For the co-immunoprecipitation assays, lysates were precleared for 1 h with protein-G sepharose beads (BioVision #6511-5) and then incubated with anti-α-tubulin (DM1A), anti-CD99 (EJ2), anti-myc (9E10; Santa Cruz Biotechnology #sc-40), or anti-YFP (3E6; Thermo Fisher Scientific #A11120) antibody. The antibody-bound proteins were pulled down using protein-G sepharose beads and eluted after boiling in sample buffer. Immunoblotting was performed as described above.

Proteomics analysis for CD99-interactome identification

Samples co-immunoprecipitated with anti-YFP antibody, or isotype IgG were fractionated on a 4–12% Bis-Tris gel (Thermo Fisher Scientific) and stained with Coomassie Brilliant Blue (Sigma-Aldrich). The entire gel lane was cut into 7–10 pieces and subjected to in-gel tryptic digestion following the general protocol. The extracted peptides were resuspended in 0.1% formic acid (FA, solvent A) and analyzed on a liquid chromatography-tandem mass spectrometry (LC-MS/MS) system integrated with Dionex 3000 ultra-high-performance liquid chromatograph and Q-Exactive Plus mass spectrometer. Peptides were loaded onto a C18 trap column (Acclaim PepMap100, 75 μm × 2 cm, 100 Å; Thermo Fisher Scientific) and separated on a C18 analytical column (PepMap RSLC, 75 μm × 50 cm, 100 Å; Thermo Fisher Scientific) using a 125-min linear gradient containing 5–35% solvent B (0.1% FA in acetonitrile) at a flow rate of 300 nL/min. The spray voltage of the column was set to 2.2 kV, and the heated capillary to 250 °C. The normalized collision energy was 27 eV. Q-Exactive was run in data-dependent acquisition mode with an MS survey scan followed by 10 MS/MS scans of the most abundant ions. The full MS scan range was 400– 1400 m/z and dynamic exclusion was applied for 20 s.

To assess the interactomes of CD99 and its mutants, the amounts of precipitated recombinant proteins were compared by a test run on SDS-PAGE followed by immunoblotting. After normalization of the levels of precipitated recombinant proteins, the samples were subjected to interactome studies. For quantitative comparison of interactome components among the three recombinant proteins (rWT, Cyt-Mut, and TM-Mut), the levels of proteins in the rWT interactome were used as a standard. For data analysis, differential expression analysis was performed using the edgeR package (v3.36.0)⁷². Raw spectral count data were normalized using the voom transformation from the limma package, which involved log-transforming the counts and applying precision weights to account for the mean-variance relationship. A linear model was fitted to the data, with contrasts specified to identify differentially expressed proteins. Proteins differentially expressed between WT-YFP and WT-Control with a false discovery rate (FDR) < 0.05 and a log2 fold change > 0 were considered CD99 interactome proteins. These proteins were further classified into functional subsets based on Gene Ontology (GO) pathway terms obtained from the Molecular Signatures Database (MsigDB). Proteins associated with the actin cytoskeleton (GO:0015629), along with the proteins Myl6, Actbl2, Actc1, and Iqgap1, were classified into the actin system group. Proteins associated with microtubule (GO:0005874) and the microtubule cytoskeleton (GO:0015630) were classified into tubulin system group. Proteins associated with vesicles (GO:0031982), cytoplasmic vesicles (GO:0031410), or secretory granules (GO:0030141) were classified into the molecular trafficking system group. Proteins associated with the myosin complex (GO:0016459), along with Myl12a, were classified into the myosin system group. Proteins associated with the cytoskeleton (GO:0005856), along with Rac2, Capza2, Rab1b, Rab27a, and Fmnl1, were classified into other cytoskeleton group. Proteins associated with the ribosome (GO:0005840), along with Arf4, Arf5, Rpn1, and Rps27, were classified into the ribosome group. Proteins associated with TCR signaling pathway (GO:0050852), along with Ptprc, Zap70, Lck, CD247, and CD3e, were classified into the TCR signaling group. Proteins associated with ATP metabolic process (GO:0046034) were classified into the mitochondria group. Interaction pairs from the STRING database (v11.0)⁷³ were used to construct a protein-protein interaction (PPI) network of CD99-interactome proteins, by selecting high-quality pairs with a combined score > 700. The network was visualized using Cytoscape software (v3.9.1; Cytoscape Team)⁷⁴. PCA was performed on the normalized proteomics data using Partek Flow software (Partek Inc.). Data were normalized using log2-scale transformation, and the first three principal components were visualized in 3D plots to assess sample clustering and variation across experimental groups. To compare the interactomes of CD99 mutants, log2 fold changes of the cytoplasmic mutant (Cyt-Mut) and transmembrane mutant (TM-Mut) relative to wild-type (rWT) were calculated from their average transformed spectral counts. These values were then visualized as a heatmap.

Flow cytometry

For FACS staining, non-specific binding was blocked with an anti-Fc receptor antibody (2.4G2; Tonbo Bioscience #70-0161-M001) prior to staining with the directly labeled monoclonal antibodies indicated in the figures and text. Dead cells were excluded by DAPI staining. The following antibodies (see Supplementary Table 2) were used to stain cell surface molecules. PE/Cy7 anti‑mCD8α (53‑6.7; BD Biosciences #552877; 1:2000), FITC anti‑mTCRβ (H57‑594; Thermo Fisher Scientific #11‑5961‑81; 1:1000), FITC anti‑mCD11a (2D7; BD Biosciences #553120; 1:1000), APC anti‑mCD25 (PC61.5; Thermo Fisher Scientific #17‑0251‑81; 1:2000), APC/Cy7 anti‑m/hCD44 (IM7; BioLegend #103028; 1:2000), eFluor450 anti‑mCD45.1 (A20; Thermo Fisher Scientific #48‑0453‑82; 1:500), PE anti‑mCD69 (H1.2F3; BioLegend #104508; 1:2000), BV605 anti‑mCD90.2 (30‑H12; BioLegend #105343; 1:4000), PE-conjugated H60/K^b-tetramer, H4/K^b-tetramer or OVA/K^b-tetramer, PE anti‑mCD99 (R&D Systems #FAB3905P; 1:500), APC anti‑mCD178 (MFL3; BioLegend #106609; 1:500), BV421 anti‑hCD3 (UCHT1; BioLegend #300434; 1:500), AlexaFluor647 anti‑hCD99 (DN16; Novus Biologicals #NB100‑64930AF647; 1:1000), BV510 anti‑hCD8 (SK1; BioLegend #344732; 1:500), APC/Cy7 anti‑hCD4 (RPA‑T4; BD Biosciences #557871; 1:500), BV605 anti‑hCD69 (FN50; BD Biosciences #562989; 1:250), PE anti‑hCD25 (M‑A251; BioLegend #985802; 1:1000), biotinylated anti‑FMC63 (Y45; Acro Biosystems #FM3‑BY54; 1:50), PE/TexasRed streptavidin (BD Pharmingen #551487; 1:2000), PE streptavidin (BD Pharmingen #554061; 1:8000). Cell proliferation were detected by labeling T cells (5 × 10⁵) from WT and Cd99^−/− mice with 5 μM CFSE (carboxyfluorescein diacetate succinimidyl ester; Thermo Fisher Scientific #C34554) and stimulating them with anti-CD3 (1 µg/ml) and anti-CD28 (0.5 µg/ml) antibodies. On days 2 and 3 after the stimulation, cell proliferation was assessed by flow cytometric analysis of CFSE dye dilution. For intracellular staining, cells were treated with PMA (20 ng/ml), Ionomycin (10 μM), and Brefeldin A (1×; Thermo Fisher Scientific) 4 h before harvest, then fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton-containing buffer for granzyme B (NGZB; eBioscience #12-8898-82; 1:4000) staining. Additionally, cells were fixed with 1.5% paraformaldehyde and permeabilized with methanol to stain for phospho-Akt (J1-223.371; BD Biosciences #558275; 10 µl/sample). Fluorescence intensity was assessed with a BD LSRII (BD Biosciences) or Northern Lights (Cytek Biosciences) and analyzed using FlowJo software (TreeStar).

Adoptive T cell transfer into hosts for immunization or into tumor bearing mice

CD8⁺ T cells from female WT or Cd99^−/− J15 mice were purified from the spleen and lymph nodes by negative MACS selection. After labelling with CFSE, cells were adoptively transferred (1 × 10⁶) 1 day before immunization with a single-cell suspension of splenocytes (2 × 10⁷ in 300 μL of PBS) from male B6.CH60 mice. In tumor regression experiments, naïve T cells (5 × 10⁵) from female CD45.1⁺ WT or Cd99^−/− OT-I mice were transferred into a female host implanted with B16F10-OVA (3 × 10⁵) at the flank. Tumor volume was measured every third day using an electronic caliper. On days 7 and 10 after OT-I cell transfer, peripheral blood was collected by eye bleeding for flow cytometric analysis. Tumor growth was monitored every 2–3 days using a digital caliper, and volume was calculated using the formula (length × width²)/2. Mice were euthanized if tumors exceeded 1500 mm³, showed ulceration, or caused distress, in accordance with IACUC guidelines. Euthanasia was performed by CO₂ inhalation.

Skin transplantation

Allogeneic tail skin transplantation was performed⁴⁹. WT or Cd99^−/− recipient mice (female, 8–12 weeks) were grafted with three pieces of full-thickness tail skin (1 × 1 × 0.5 cm³) derived from male BALB.B mice (male, 8–12 weeks), and one piece of syngeneic WT or Cd99^−/− mice, respectively. Graft survival was evaluated by daily visual inspection. Rejection was defined as necrosis of 75% of the transplanted skin surface.

Generation of CAR-T cells

The ORF sequence of the conventional CAR harboring CD8 extracellular and transmembrane domain (BBz-CAR) was acquired from a patent (US_2013_0287748_A1). HuTJ-BBz-CAR ORF was generated by replacing CD8 parts of BBz-CAR with human CD99 extracellular, transmembrane and juxtamembrane domain (amino acid 88–158, UniProt #P14209-1). The CAR ORF cDNAs were synthesized by Integrated DNA Technologies (IDT) and cloned into MSGV retroviral vector (Addgene #64269). Retroviruses for human T cell transduction were produced by establishing amphitropic retrovirus-producing stable cell lines from PG13 retroviral packaging cells (ATCC). For this purpose, retroviral plasmids were transfected into Phoenix ECO cells (ATCC) using Lipofectamine 3000 (Thermo Fisher Scientific) to produce ecotropic retroviruses. The retroviral supernatant was harvested 48 h after transfection and spin-infected into PG13 cells (derived from a mouse fibroblast) for stable transduction of the retroviral cDNA. Retrovirus-containing supernatants from these PG13 cultures were harvested and concentrated 5–10-fold using a centrifugal filter device (Amicon Ultra,100 kDa cut-off; Millipore) and used for human T cell transduction. For the generation of CAR-T cells, human peripheral blood was obtained from healthy volunteers who provided written informed consent according to a protocol approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-2008-114-1150). PBMCs prepared by Ficoll-gradient centrifugation (841×g, 20 min, RT) were activated for 48 h with plate-bound anti-CD3 (10 μg/ml; OKT3, BioXcell) and soluble anti-CD28 (2 μg/ml; CD28.2, BD Biosciences) antibodies. The concentrated retroviral solutions were attached to a 24-well retronectin-coated plate (20 μg/ml, TaKaRa) by centrifugation for 2 h at 2000×g. After removal of the viral solution, the activated T cells were added to the plate, followed by 10-min centrifugation at 1000×g and further incubation for 48 h in medium containing human IL-2 (300 IU/ml Proleukin; Novartis). The T cells were then washed and expanded in fresh medium containing human IL-2 (200 IU/ml) for 3–6 days before being used as CAR-T cells. To generate endogenous CD99 gene-deleted HuTJ-BBz- or BBz-CAR-T cells, T cells stimulated with anti-CD3 and CD28 for 48 h were electroporated with Cas9 ribonucleoprotein containing sgRNA targeting CD99 (cggcgaccagaacacccagc, synthesized by IDT) or a scramble-sgRNA control (sgControl; #1072544, IDT) using the Neon Transfection System and transduced with virus containing either HuTJ-BBz-CAR or BBz-CAR.

Evaluation of CAR-T cell activity in vitro and in vivo

To visualize cell-conjugate formation, 2 × 10⁵ CAR-T cells were co-incubated with 1 × 10⁵ CMTMR-labeled Raji B cells on poly-L-lysine-coated coverslips. After 15, 30 or 60 min, the cells were washed, fixed, and permeabilized, and then stained with phalloidin and anti-α-tubulin (DM1A). For flow cytometric analysis of CAR-T cell activation, 1 × 10⁵ CAR-T cells were co-incubated with 2 × 10⁴ Raji B cells in round-bottom 96 well plates for three days. To evaluate tumor killing efficacy of CAR-T cells in vitro, untransduced T or CAR-T cells (4000 cells as CAR⁺/well) were co-cultured with Raji tumor cells (16,000 cells/well) at an E:T ratio of 1:4 in 96 well plates in the absence of exogenous cytokines. On the day of analysis, dead cells were excluded by staining with 7-aminoactinomycin D (7-AAD) (BioLegend #420404, 5 μl/sample) and counted number of live tumor cells (7-AAD^-CD3^-) and CAR-T cells (7-AAD⁻CD3⁺) using 123 count eBead (containing about 10,000 beads; Thermo Fisher Scientific #01-1234-42, 10 μl/sample). To exclude the tumorigenic potential of CAR-T cells, HuTJ-BBz- and BBz-CAR-T cells were suspended in 500 μl of 10% RPMI-1640 medium (5 × 10⁵ cells/sample) and cultured without additional CAR stimulation or cytokines. Untransduced-T cells and the Jurkat cell line were used as controls. Cell viability and the number of viable cells were determined at each time point. 7-AAD-negative cells were gated as viable cells, and cell numbers were counted using a bead-based flow cytometry method. For Jurkat cells, cells were split by 1/5 or 1/10 every 3–4 days to prevent excessive cell growth, so the dilution ratio was taken into account when calculating the number of viable cells.

CAR-T cell activity in vivo was evaluated by intravenously injecting NSG mice (female and male, 6–8 weeks) with Raji-Luc tumor cells (5 × 10⁵/mouse). After 7 days, BBz-CAR-T cells or HuTJ-BBz-CAR-T cells (5 × 10⁶/mouse) were infused intravenously. Tumor engraftment and growth were measured by bioluminescence imaging (IVIS Spectrum; PerkinElmer) after intraperitoneal injection of D-luciferin (2 mg; Promega #P1043). The tumor burden was quantified as total flux (photons per s) using Living Image software (PerkinElmer). To estimate CAR-T cell numbers in blood, 10 μl 123 count eBeads were added to each RBC-lysed and FACS-stained sample tube. After 2000 beads were acquired by flow cytometer, number of human cells in the blood was calculated by the following formula: blood cell count (cells/μl) = (blood cell count × eBead volume)/(eBead count × blood volume) × eBead concentration.

Statistical analysis

Statistical analysis was performed using GraphPad Prism software version 9 (GraphPad Software Inc.), with statistical tests detailed in the figure legends. Unless otherwise stated, quantitative data are expressed as means ± SEM from data obtained from at least two independent experiments. Statistical significance was determined by one-way ANOVA with the Dunnett’s post-hoc test, two-way repeated measures ANOVA with Sidak post-hoc test, two-tailed unpaired Student’s t-test, or a log-rank (Mantel-Cox) test. P values lower than 0.05 were considered to be significant.

Reporting summary

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

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD039955. All data are included in the Supplementary Information or available from the authors, as are unique reagents used in this Article. The raw numbers for charts and graphs are available in the Source Data file whenever possible. Source data are provided with this paper.

References

June, C. H., O’Connor, R. S., Kawalekar, O. U., Ghassemi, S. & Milone, M. C. CAR T cell immunotherapy for human cancer. Science 359, 1361–1365 (2018).
Article ADS CAS PubMed Google Scholar
Labanieh, L. & Mackall, C. L. CAR immune cells: design principles, resistance and the next generation. Nature 614, 635–648 (2023).
Article ADS CAS PubMed Google Scholar
Dotti, G., Gottschalk, S., Savoldo, B. & Brenner, M. K. Design and development of therapies using chimeric antigen receptor-expressing T cells. Immunol. Rev. 257, 107–126 (2014).
Article CAS PubMed Google Scholar
Morris, E. C., Neelapu, S. S., Giavridis, T. & Sadelain, M. Cytokine release syndrome and associated neurotoxicity in cancer immunotherapy. Nat. Rev. Immunol. 22, 85–96 (2022).
Article CAS PubMed Google Scholar
Morgan, R. A. et al. Case report of a serious adverse event following the administration of T cells transduced with a chimeric antigen receptor recognizing ERBB2. Mol. Ther. 18, 843–851 (2010).
Article CAS PubMed PubMed Central Google Scholar
Grakoui, A. et al. The immunological synapse: a molecular machine controlling T cell activation. Science 285, 221–227 (1999).
Article CAS PubMed Google Scholar
Dustin, M. L., Chakraborty, A. K. & Shaw, A. S. Understanding the structure and function of the immunological synapse. Cold Spring Harb. Perspect. Biol. 2, a002311 (2010).
Article CAS PubMed PubMed Central Google Scholar
Demetriou, P. et al. A dynamic CD2-rich compartment at the outer edge of the immunological synapse boosts and integrates signals. Nat. Immunol. 21, 1232–1243 (2020).
Article CAS PubMed PubMed Central Google Scholar
Monks, C. R., Freiberg, B. A., Kupfer, H., Sciaky, N. & Kupfer, A. Three-dimensional segregation of supramolecular activation clusters in T cells. Nature 395, 82–86 (1998).
Article ADS CAS PubMed Google Scholar
Hammer, J. A., Wang, J. C., Saeed, M. & Pedrosa, A. T. Origin, Organization, dynamics, and function of actin and actomyosin networks at the T cell immunological synapse. Annu. Rev. Immunol. 37, 201–224 (2019).
Article CAS PubMed Google Scholar
Kopf, A. & Kiermaier, E. Dynamic microtubule arrays in leukocytes and their role in cell migration and immune synapse formation. Front Cell Dev. Biol. 9, 635511 (2021).
Article PubMed PubMed Central Google Scholar
Ritter, A. T. et al. Actin depletion initiates events leading to granule secretion at the immunological synapse. Immunity 42, 864–876 (2015).
Article CAS PubMed PubMed Central Google Scholar
de la Roche, M., Asano, Y. & Griffiths, G. M. Origins of the cytolytic synapse. Nat. Rev. Immunol. 16, 421–432 (2016).
Article PubMed Google Scholar
Martin-Cofreces, N. B. et al. End-binding protein 1 controls signal propagation from the T cell receptor. EMBO J. 31, 4140–4152 (2012).
Article CAS PubMed PubMed Central Google Scholar
Kupfer, A. & Dennert, G. Reorientation of the microtubule-organizing center and the Golgi apparatus in cloned cytotoxic lymphocytes triggered by binding to lysable target cells. J. Immunol. 133, 2762–2766 (1984).
Article CAS PubMed Google Scholar
Kuhn, J. R. & Poenie, M. Dynamic polarization of the microtubule cytoskeleton during CTL-mediated killing. Immunity 16, 111–121 (2002).
Article CAS PubMed Google Scholar
Stinchcombe, J. C., Majorovits, E., Bossi, G., Fuller, S. & Griffiths, G. M. Centrosome polarization delivers secretory granules to the immunological synapse. Nature 443, 462–465 (2006).
Article ADS CAS PubMed Google Scholar
Laan, L. et al. Cortical dynein controls microtubule dynamics to generate pulling forces that position microtubule asters. Cell 148, 502–514 (2012).
Article CAS PubMed PubMed Central Google Scholar
Yi, J. et al. Centrosome repositioning in T cells is biphasic and driven by microtubule end-on capture-shrinkage. J. Cell Biol. 202, 779–792 (2013).
Article CAS PubMed PubMed Central Google Scholar
Pasello, M., Manara, M. C. & Scotlandi, K. CD99 at the crossroads of physiology and pathology. J. Cell Commun. Signal 12, 55–68 (2018).
Article PubMed PubMed Central Google Scholar
Hahn, J. H. et al. CD99 (MIC2) regulates the LFA-1/ICAM-1-mediated adhesion of lymphocytes, and its gene encodes both positive and negative regulators of cellular adhesion. J. Immunol. 159, 2250–2258 (1997).
Article CAS PubMed Google Scholar
Schenkel, A. R., Mamdouh, Z., Chen, X., Liebman, R. M. & Muller, W. A. CD99 plays a major role in the migration of monocytes through endothelial junctions. Nat. Immunol. 3, 143–150 (2002).
Article CAS PubMed Google Scholar
Lou, O., Alcaide, P., Luscinskas, F. W. & Muller, W. A. CD99 is a key mediator of the transendothelial migration of neutrophils. J. Immunol. 178, 1136–1143 (2007).
Article CAS PubMed Google Scholar
Winger, R. C. et al. Cutting Edge: CD99 Is a Novel Therapeutic Target for Control of T Cell-Mediated Central Nervous System Autoimmune Disease. J. Immunol. 196, 1443–1448 (2016).
Article CAS PubMed Google Scholar
Manara, M. C., Pasello, M. & Scotlandi, K. CD99: a cell surface protein with an oncojanus role in tumors. Genes (Basel) 9, 159 (2018).
Waclavicek, M. et al. CD99 engagement on human peripheral blood T cells results in TCR/CD3-dependent cellular activation and allows for Th1-restricted cytokine production. J. Immunol. 161, 4671–4678 (1998).
Article CAS PubMed Google Scholar
Oh, K. I. et al. CD99 activates T cells via a costimulatory function that promotes raft association of TCR complex and tyrosine phosphorylation of TCR zeta. Exp. Mol. Med. 39, 176–184 (2007).
Article CAS PubMed Google Scholar
Choi, E. Y. et al. Engagement of CD99 induces up-regulation of TCR and MHC class I and II molecules on the surface of human thymocytes. J. Immunol. 161, 749–754 (1998).
Article CAS PubMed Google Scholar
Yoon, S. S. et al. Engagement of CD99 triggers the exocytic transport of ganglioside GM1 and the reorganization of actin cytoskeleton. FEBS Lett. 540, 217–222 (2003).
Article CAS PubMed Google Scholar
Bremond, A. et al. Regulation of HLA class I surface expression requires CD99 and p230/golgin-245 interaction. Blood 113, 347–357 (2009).
Article CAS PubMed Google Scholar
Pata, S. et al. Association of CD99 short and long forms with MHC class I, MHC class II and tetraspanin CD81 and recruitment into immunological synapses. BMC Res. Notes 4, 293 (2011).
Article CAS PubMed PubMed Central Google Scholar
Watson, R. L. et al. Endothelial CD99 signals through soluble adenylyl cyclase and PKA to regulate leukocyte transendothelial migration. J. Exp. Med. 212, 1021–1041 (2015).
Article CAS PubMed PubMed Central Google Scholar
Nam, G. et al. Interaction of CD99 with its paralog CD99L2 positively regulates CD99L2 trafficking to cell surfaces. J. Immunol. 191, 5730–5742 (2013).
Article CAS PubMed PubMed Central Google Scholar
Roopenian, D., Choi, E. Y. & Brown, A. The immunogenomics of minor histocompatibility antigens. Immunol. Rev. 190, 86–94 (2002).
Article CAS PubMed Google Scholar
Choi, E. Y. et al. Immunodominance of H60 is caused by an abnormally high precursor T cell pool directed against its unique minor histocompatibility antigen peptide. Immunity 17, 593–603 (2002).
Article PubMed Google Scholar
Choi, E. Y., Choi, K., Nam, G., Kim, W. & Chung, M. H60: a unique murine hematopoietic cell-restricted minor histocompatibility antigen for graft-versus-leukemia effect. Front. Immunol. 11, 1163 (2020).
Article CAS PubMed PubMed Central Google Scholar
Ju, J. M. et al. Escape from thymic deletion and anti-leukemic effects of T cells specific for hematopoietic cell-restricted antigen. Nat. Commun. 9, 225 (2018).
Article ADS PubMed PubMed Central Google Scholar
Kim, J. et al. Memory programming in CD8(+) T-cell differentiation is intrinsic and is not determined by CD4 help. Nat. Commun. 6, 7994 (2015).
Article ADS CAS PubMed Google Scholar
Riedl, J. et al. Lifeact mice for studying F-actin dynamics. Nat. Methods 7, 168–169 (2010).
Article CAS PubMed Google Scholar
Lukinavicius, G. et al. Fluorogenic probes for live-cell imaging of the cytoskeleton. Nat. Methods 11, 731–733 (2014).
Article CAS PubMed Google Scholar
Macian, F. et al. Transcriptional mechanisms underlying lymphocyte tolerance. Cell 109, 719–731 (2002).
Article CAS PubMed Google Scholar
Keller, H. U., Niggli, V. & Zimmermann, A. Diacylglycerols and PMA induce actin polymerization and distinct shape changes in lymphocytes: relation to fluid pinocytosis and locomotion. J. Cell Sci. 93, 457–465 (1989).
Article CAS PubMed Google Scholar
Preciado Lopez, M. et al. Actin-microtubule coordination at growing microtubule ends. Nat. Commun. 5, 4778 (2014).
Article PubMed Google Scholar
Fukata, M. et al. Rac1 and Cdc42 capture microtubules through IQGAP1 and CLIP-170. Cell 109, 873–885 (2002).
Article CAS PubMed Google Scholar
Liu, X., Kapoor, T. M., Chen, J. K. & Huse, M. Diacylglycerol promotes centrosome polarization in T cells via reciprocal localization of dynein and myosin II. Proc. Natl Acad. Sci. USA 110, 11976–11981 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Wang, J. C. et al. The Rap1-cofilin-1 pathway coordinates actin reorganization and MTOC polarization at the B cell immune synapse. J. Cell Sci. 130, 1094–1109 (2017).
Article CAS PubMed Google Scholar
Kumari, S. et al. Cytoskeletal tension actively sustains the migratory T-cell synaptic contact. EMBO J. 39, e102783 (2020).
Article CAS PubMed PubMed Central Google Scholar
Choi, E. Y. et al. Quantitative analysis of the immune response to mouse non-MHC transplantation antigens in vivo: the H60 histocompatibility antigen dominates over all others. J. Immunol. 166, 4370–4379 (2001).
Article CAS PubMed Google Scholar
Yoo, K. I. et al. Subdominant H60 antigen-specific CD8 T-cell response precedes dominant H4 antigen-specific response during the initial phase of allogenic skin graft rejection. Exp. Mol. Med 47, e140 (2015).
Article CAS PubMed PubMed Central Google Scholar
Dufour, E. M., Deroche, A., Bae, Y. & Muller, W. A. CD99 is essential for leukocyte diapedesis in vivo. Cell Commun. Adhes. 15, 351–363 (2008).
Article CAS PubMed PubMed Central Google Scholar
Milone, M. C. et al. Chimeric receptors containing CD137 signal transduction domains mediate enhanced survival of T cells and increased antileukemic efficacy in vivo. Mol. Ther. 17, 1453–1464 (2009).
Article CAS PubMed PubMed Central Google Scholar
Bernard, G. et al. The E2 molecule (CD99) specifically triggers homotypic aggregation of CD4+ CD8+ thymocytes. J. Immunol. 154, 26–32 (1995).
Article CAS PubMed Google Scholar
Bernard, G. et al. CD99 (E2) up-regulates alpha4beta1-dependent T cell adhesion to inflamed vascular endothelium under flow conditions. Eur. J. Immunol. 30, 3061–3065 (2000).
Article CAS PubMed Google Scholar
Sohn, H. W. et al. CD99 regulates the transport of MHC class I molecules from the Golgi complex to the cell surface. J. Immunol. 166, 787–794 (2001).
Article CAS PubMed Google Scholar
Martin-Cofreces, N. B. & Sanchez-Madrid, F. Sailing to and docking at the immune synapse: role of tubulin dynamics and molecular motors. Front Immunol. 9, 1174 (2018).
Article PubMed PubMed Central Google Scholar
Mastrogiovanni, M., Juzans, M., Alcover, A. & Di Bartolo, V. Coordinating cytoskeleton and molecular traffic in T cell migration, activation, and effector functions. Front. Cell Dev. Biol. 8, 591348 (2020).
Article PubMed PubMed Central Google Scholar
Chanet, S., Sharan, R., Khan, Z. & Martin, A. C. Myosin 2-induced mitotic rounding enables columnar epithelial cells to interpret cortical spindle positioning cues. Curr. Biol. 27, 3350–3358 e3353 (2017).
Article CAS PubMed Google Scholar
Verma, V., Mogilner, A. & Maresca, T. J. Classical and emerging regulatory mechanisms of cytokinesis in animal cells. Biology 8, 55 (2019).
Hedman, A. C., Smith, J. M. & Sacks, D. B. The biology of IQGAP proteins: beyond the cytoskeleton. EMBO Rep. 16, 427–446 (2015).
Article CAS PubMed PubMed Central Google Scholar
Laplante, C., Huang, F., Tebbs, I. R., Bewersdorf, J. & Pollard, T. D. Molecular organization of cytokinesis nodes and contractile rings by super-resolution fluorescence microscopy of live fission yeast. Proc. Natl Acad. Sci. USA 113, E5876–E5885 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Sundvold, H. et al. Arv1 promotes cell division by recruiting IQGAP1 and myosin to the cleavage furrow. Cell Cycle 15, 628–643 (2016).
Article CAS PubMed PubMed Central Google Scholar
Kim, S. H. et al. Generation of cells with Hodgkin’s and Reed-Sternberg phenotype through downregulation of CD99 (Mic2). Blood 92, 4287–4295 (1998).
Article CAS PubMed Google Scholar
Hammer, J. A. 3rd & Burkhardt, J. K. Controversy and consensus regarding myosin II function at the immunological synapse. Curr. Opin. Immunol. 25, 300–306 (2013).
Article CAS PubMed PubMed Central Google Scholar
Kopf, A. et al. Microtubules control cellular shape and coherence in amoeboid migrating cells. J. Cell Biol. 219, e201907154 (2020).
Xiong, W. et al. Immunological synapse predicts effectiveness of chimeric antigen receptor cells. Mol. Ther. 26, 963–975 (2018).
Article CAS PubMed PubMed Central Google Scholar
Liu, D. et al. The role of immunological synapse in predicting the efficacy of chimeric antigen receptor (CAR) immunotherapy. Cell Commun. Signal 18, 134 (2020).
Article CAS PubMed PubMed Central Google Scholar
Gao, Y., Wang, Y., Luo, F. & Chu, Y. Optimization of T cell redirecting strategies: obtaining inspirations from natural process of T cell activation. Front Immunol. 12, 664329 (2021).
Article CAS PubMed PubMed Central Google Scholar
Davenport, A. J. et al. Chimeric antigen receptor T cells form nonclassical and potent immune synapses driving rapid cytotoxicity. Proc. Natl Acad. Sci. USA 115, E2068–E2076 (2018).
Article CAS PubMed PubMed Central Google Scholar
Mukherjee, M., Mace, E. M., Carisey, A. F., Ahmed, N. & Orange, J. S. Quantitative imaging approaches to study the CAR immunological synapse. Mol. Ther. 25, 1757–1768 (2017).
Article CAS PubMed PubMed Central Google Scholar
Shen, Z., Reznikoff, G., Dranoff, G. & Rock, K. L. Cloned dendritic cells can present exogenous antigens on both MHC class I and class II molecules. J. Immunol. 158, 2723–2730 (1997).
Article CAS PubMed Google Scholar
Manders, E. M., Stap, J., Brakenhoff, G. J., van Driel, R. & Aten, J. A. Dynamics of three-dimensional replication patterns during the S-phase, analysed by double labelling of DNA and confocal microscopy. J. Cell Sci. 103, 857–862 (1992).
Article CAS PubMed Google Scholar
Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).
Article CAS PubMed Google Scholar
Szklarczyk, D. et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 47, D607–D613 (2019).
Article CAS PubMed Google Scholar
Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003).
Article CAS PubMed PubMed Central Google Scholar
Sievers, F. et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank professor Sang Hee Shim (Korea University, Seoul, Korea) for technical advice on the visualization of tubulin dynamics and Min Jae Kim (Ticaros, Seoul, Korea) for technical assistance. This work was supported by the National Research Foundation (NRF) grants (RS-2024-00405619, EYC; RS-2023-00260454, KC; RS-2024-00400897, KC; 2021R1A4A1029097, EYC and KC) funded by Ministry of Science and ICT, the Korea Drug Development Fund (KDDF) grant (RS-2024-00337759, KC) funded by the Ministry of Science and ICT, Ministry of Trade, Industry, and Energy, and Ministry of Health and Welfare, and KIST intramural grant (2E33701, HR), Republic of Korea.

Author information

These authors jointly supervised this work: Kyungho Choi, Eun Young Choi.

Authors and Affiliations

Department of Biomedical Sciences, Seoul National University, Seoul, 03080, Republic of Korea
Giri Nam, Hyung Bae Park, Hanna Chang, Ji Hwan Kim, Youngjae Lee, Seoho Lee, Hyeonji Lim, Kyungho Choi & Eun Young Choi
Institute of Human Environment Interface Biology, Seoul National University College of Medicine, Seoul, 03080, Republic of Korea
Giri Nam, Hanna Chang, Woo-Jeong Shon & Eun Young Choi
Department of Biochemistry and Molecular Biology, Seoul National University College of Medicine, Seoul, 03080, Republic of Korea
Hye Ran Yeon, Hyung Bae Park, Ji Hwan Kim, Ji Eun Lee, Youngjae Lee, Seoho Lee, Hyeonji Lim & Kyungho Choi
Cancer Research Institute, Seoul National University, College of Medicine, Seoul, 03080, Republic of Korea
Hye Ran Yeon, Hyung Bae Park, Ji Eun Lee & Kyungho Choi
Ticaros Inc, Seoul, 04784, Republic of Korea
Hye Ran Yeon, Hyung Bae Park & Ji Eun Lee
Department of Molecular Medicine and Biopharmaceutical Sciences, Graduate School of Convergence Science and Technology, Seoul National University, Seoul, 03080, Republic of Korea
Byoung-Kyu Cho, Hyeryeon Jung & Eugene C. Yi
Department of Integrated Biomedical and Life Science, Korea University, Seoul, 02841, Republic of Korea
Seoyeon Kim & Joon-Yong An
BK21FOUR R&E Center for Learning Health Systems, Korea University, Seoul, 02841, Republic of Korea
Seoyeon Kim & Joon-Yong An
School of Biosystem and Biomedical Science, College of Health Science, Korea University, Seoul, 02841, Republic of Korea
Joon-Yong An
Brain Science Institute, Korea Institute of Science and Technology, Seoul, 02792, Republic of Korea
Eun Mi Hwang & Hoon Ryu
Graduate School of Pharmaceutical Sciences, Ewha Womans University, Seoul, 03760, Republic of Korea
Jun Chang

Authors

Giri Nam
View author publications
Search author on:PubMed Google Scholar
Hye Ran Yeon
View author publications
Search author on:PubMed Google Scholar
Hyung Bae Park
View author publications
Search author on:PubMed Google Scholar
Hanna Chang
View author publications
Search author on:PubMed Google Scholar
Ji Hwan Kim
View author publications
Search author on:PubMed Google Scholar
Byoung-Kyu Cho
View author publications
Search author on:PubMed Google Scholar
Hyeryeon Jung
View author publications
Search author on:PubMed Google Scholar
Eugene C. Yi
View author publications
Search author on:PubMed Google Scholar
Seoyeon Kim
View author publications
Search author on:PubMed Google Scholar
Joon-Yong An
View author publications
Search author on:PubMed Google Scholar
Ji Eun Lee
View author publications
Search author on:PubMed Google Scholar
Youngjae Lee
View author publications
Search author on:PubMed Google Scholar
Seoho Lee
View author publications
Search author on:PubMed Google Scholar
Hyeonji Lim
View author publications
Search author on:PubMed Google Scholar
Woo-Jeong Shon
View author publications
Search author on:PubMed Google Scholar
Eun Mi Hwang
View author publications
Search author on:PubMed Google Scholar
Hoon Ryu
View author publications
Search author on:PubMed Google Scholar
Jun Chang
View author publications
Search author on:PubMed Google Scholar
Kyungho Choi
View author publications
Search author on:PubMed Google Scholar
Eun Young Choi
View author publications
Search author on:PubMed Google Scholar

Contributions

Conceptualization: G.N., K.C., and E.Y.C. Investigation: G.N., H.R.Y., H.B.P., H.C., J.H.K., J.E.L., B.K.C., Y.L., S.L., H.L., and E.M.H. Data Curation: H.J., B.K.C., and E.C.Y.; Formal Analysis: S.K., J.Y.A. and W.J.S. Resources: H.R. and J.C. Funding Acquisition: K.C. and E.Y.C. Project Administration: E.Y.C. Supervision: K.C. and E.Y.C. Writing-original draft: G.N., K.C., and E.Y.C. Writing-review & editing: G.N., K.C., and E.Y.C.

Corresponding authors

Correspondence to Kyungho Choi or Eun Young Choi.

Ethics declarations

Competing interests

G.N., H.R.Y., H.B.P., J.E.L., K.C., and E.Y.C. are inventors on a filed patent related to the CAR-T cells. K.C. and E.Y.C. are scientific co-founders and shareholders of Ticaros. H.R.Y., H.B.P., and J.E.L. are employees of Ticaros currently. The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Claire Hivroz, Michael Kalos, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

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

Supplementary information

Supplementary Information

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Supplementary Data 4

Supplementary Data 5

Supplementary Data 6

Supplementary Data 7

Supplementary Movie 1

Supplementary Movie 2

Supplementary Movie 3

Supplementary Movie 4

Supplementary Movie 5

Supplementary Movie 6

Supplementary Movie 7

Reporting Summary

Transparent Peer Review file

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Nam, G., Yeon, H.R., Park, H.B. et al. CD99-mediated immunological synapse formation potentiates CAR-T cell function. Nat Commun 16, 7987 (2025). https://doi.org/10.1038/s41467-025-63184-w

Download citation

Received: 02 June 2024
Accepted: 13 August 2025
Published: 27 August 2025
DOI: https://doi.org/10.1038/s41467-025-63184-w