Enforced sialyl‐Lewis‐X (sLeX) display in E‐selectin ligands by exofucosylation is dispensable for CD19‐CAR T‐cell activity and bone marrow homing

Abstract CD19‐directed chimeric antigen receptors (CAR) T cells induce impressive rates of complete response in advanced B‐cell malignancies, specially in B‐cell acute lymphoblastic leukemia (B‐ALL). However, CAR T‐cell‐treated patients eventually progress due to poor CAR T‐cell persistence and/or disease relapse. The bone marrow (BM) is the primary location for acute leukemia. The rapid/efficient colonization of the BM by systemically infused CD19‐CAR T cells might enhance CAR T‐cell activity and persistence, thus, offering clinical benefits. Circulating cells traffic to BM upon binding of tetrasaccharide sialyl‐Lewis X (sLeX)‐decorated E‐selectin ligands (sialofucosylated) to the E‐selectin receptor expressed in the vascular endothelium. sLeX‐installation in E‐selectin ligands is achieved through an ex vivo fucosylation reaction. Here, we sought to characterize the basal and cell‐autonomous display of sLeX in CAR T‐cells activated using different cytokines, and to assess whether exofucosylation of E‐selectin ligands improves CD19‐CAR T‐cell activity and BM homing. We report that cell‐autonomous sialofucosylation (sLeX display) steadily increases in culture‐ and in vivo‐expanded CAR T cells, and that, the cytokines used during T‐cell activation influence both the degree of such endogenous sialofucosylation and the CD19‐CAR T‐cell efficacy and persistence in vivo. However, glycoengineered enforced sialofucosylation of E‐selectin ligands was dispensable for CD19‐CAR T‐cell activity and BM homing in multiple xenograft models regardless the cytokines employed for T‐cell expansion, thus, representing a dispensable strategy for CD19‐CAR T‐cell therapy.

enforced sialofucosylation of E-selectin ligands was dispensable for CD19-CAR T-cell activity and BM homing in multiple xenograft models regardless the cytokines employed for T-cell expansion, thus, representing a dispensable strategy for CD19-CAR T-cell therapy.

K E Y W O R D S INTRODUCTION
Immunotherapy has revolutionized cancer treatment. Adoptive cell immunotherapy using T cells genetically redirected to a tumor-specific antigen by chimeric antigen receptors (CAR) has induced impressive rates of complete response in advanced B-cell malignancies, especially in B-cell acute lymphoblastic leukemia (B-ALL). 1,2 Unfortunately, however, CAR T-cell-treated patients eventually progress due to either poor CAR T-cell persistence and/or disease relapse. Common limitations associated to CD19-CAR T-cell treatment, but extendable to other CARs are: (i) failure to achieve complete remission, (ii) relapse with potential antigen loss, (iii) cytokine release syndrome (CRS) and related toxicities, and (iv) existence of multitreated patients not eligible for CAR T-cell therapy due to low counts of T cells. [3][4][5] In the clinical practice, adoptive cell therapies are systemically infused via the bloodstream. However, the bone marrow (BM) is the primary location for acute leukemia initiation and maintenance. [6][7][8] Anatomically, the BM microenvironment confers cellular interactions and signals promoting leukemia initiation, maintenance and progression, as well as drug resistance of leukemic cells. 9 Current challenges associated to CAR T-cell treatment in acute leukemia patients might be partially overcome by a rapid and effective CAR T-cell redirection to the BM. In fact, efficient seeding of systematically infused CAR T cells in the leukemic BM might enhance CAR T-cell activity and persistence, eventually providing key clinical benefits associated to the potential reduction of the CAR T-cell dose infused, namely less procedure-related toxicities, lower production costs, and broader patient's inclusion criteria.
Although several mechanisms regulate the homing of circulating cells to the BM, 10-12 the ability of circulating cells to traffic to the BM initially relies on their robust adherence to the E-selectin receptor (CD62E) displayed in the vascular endothelium (VE). Adhesive interactions between the E-selectin receptor and its cognate ligand, tetrasaccharide sialyl-Lewis X (sLeX), displayed on circulating cells dictate adherence of circulating cells to the VE, the first step of such biological process. 8 Cell bind-ing activity to E-selectin receptor is specifically exerted by the sialofucosylated E-selectin ligands CLA, CD43E, and HCELL resulting from the sLeX instalment (α-2,3sialic acid and α-1,3-fucose binding determinants on Nglycans) in the native E-selectin ligands PSGL1, CD43, and CD44, respectively. 13,14 Of note, native E-selectin ligands can be converted into sLeX-displaying (sialofucosylated) E-selectin ligands through a straightforward glycan engineering approach involving minimal ex vivo cellular manipulation based on a α-1,3-fucosyltransferase enzymatic reaction and guanosine diphosphate-fucose (GDPfucose) substrate. [15][16][17] Such exofucosylation reaction was previously shown to endow BM homing abilities to hematopoietic stem/progenitors cells (HSPCs), mesenchymal stem/stromal cells (MSCs), and immune cells. 15,18,19 Here, using CD19-CAR T cells as a working model, we sought (i) to characterize the basal/cell-autonomous display of sLeX in CAR T-cells activated with either IL-2 or IL-7/IL-15, and (ii) to assess whether exofucosylation of E-selectin ligands to enforce sLeX display improves CD19-CAR T-cell activity and BM homing. Our results revealed that cell-autonomous sialofucosylation steadily augments in culture-and in vivo-expanded CAR T cells, and that the type of cytokines used during T-cell activation influences both the cell surface display of sLeX in CD19-CAR T cells and the CD19-CAR T-cell efficacy and persistence in vivo. However, enforced sLeX display in Eselectin ligands by exofucosylation was dispensable for both CD19-CAR T-cell activity and BM homing, regardless the cytokines employed for T-cell expansion. Collectively, glycoengineered sLeX display in CAR T-cells systemically administered is a dispensable strategy for improved CAR T-cell function.

2.1
Cell-autonomous and exofucosylation-enforced expression of sLeX in CD19-CAR T cells T cells from peripheral blood (PB) of healthy donors (n = 5) were activated using CD3/CD28 plus either IL-2 or IL-7/ IL-15 and transduced on day 2 with CD19-CAR-expressing lentivectors ( Figure 1A). The levels of sialofucosylation (sLeX display in cell surface) were analyzed by FACS on CAR T cells over the 9-day activation/expansion period using the HECA452 MoAb, which recognizes sLeX. 20 Basal expression of sLeX (HECA452+) was found in approximately 31 ± 5% of T cells at day 0 ( Figure 1B). Interestingly, in vitro T-cell activation/expansion led to a cell-autonomous gradual increase in sLeX-expressing CAR T cells ( Figure 1B). Of note, IL-7/IL-15 activation consistently rendered higher frequency of sialofucosylated (HECA452+) CAR T cells than IL-2-based activation (75 ± 7% vs 50 ± 5% at day 9; Figure 1B), in a T-cell proliferation/expansion independent manner ( Figure 1C). Using a well-established FTVII-based exofucosylation reaction 17 ( Figure 1D), 100% of culture-expanded CAR T cells became HECA452+ within 48 h, regardless the cytokines used during T-cell activation ( Figure 1E). Western Blot (WB) analysis using E-selectin-Ig immunoprecipitates clearly identified CD43 (CD43E), and partially PSGL1 (CLA), as the E-selectin ligands carrying sLeX in exofucosylated CAR T cells ( Figure 1F). We next analyzed the phenotype of the expanded T cells using a CCR7 and CD45RA staining, and found that neither the cytokines used nor the exofucosylation reaction affected the T-cell phenotype (TN, TCM, TEFF/EM, TEMRA) upon 9 days of expansion ( Figure 1G). Collectively, although the cytokines used for T-cell activation influence the level of sLeX display in CAR T cells, cell-autonomous sialofucosylation gradually increases in culture-expanded CAR T cells, with up to 80% of CAR T cells being endogenously fucosylated at the end of the in vitro expansion.

Exofucosylation enhances neither cytotoxic activity nor homing of CAR T cells in vitro
We first prompted to analyze in vitro the cytotoxic activity of exofucosylated CD19-CAR T cells. The cytotoxicity and specificity of both BT-(control) and FTVII-treated (sialofucosylated) CD19-CAR T cells were identical in in vitro assays against CD19+ (NALM6, SEM) and CD19-(Jurkat) cell lines at multiple effector:target ratios (Figure 2A). We next investigated the potential of exofucosylated CD19-CAR T cells to migrate through TNF-α-stimulated Human Umbilical Vein Endothelial Cells (HUVEC) monolayers using standard in vitro transwell migration assays (Figure 2B). As expected, the expression of both E-selectin receptor and VCAM-1, a key vascular cell adhesion molecule, was upregulated in HUVEC cells upon TNF-α stimulation, thus, mimicking an activated microvasculature environment 21,22 (Figure 2B, right panels).
Regardless the cytokines used for T-cell activation, exofucosylation rendered HECA452 expression in 100% of the FTVII-treated CD19-CAR T cells while not affecting the expression of the VLA-4, the putative VCAM-1 ligand, confirming the specificity of the FTVII treatment (Figure 2C). BT-and FTVII-treated CD19 CAR T cells (upper chamber) showed identical migratory capacity toward target cells (bottom chamber) through either nonstimulated or TNF-α-stimulated HUVEC monolayers ( Figure 2D), which translated into identical cytotoxicity of target cells by the migrating CAR T cells, in 24 h assays ( Figure 2E). Taken together, enforced exofucosylation of CD19-CAR T cells enhances neither cytotoxicity nor homing in vitro.

Exofucosylation enhances neither homing to BM/spleen nor activity/persistence of CAR T cells in vivo
We next assessed whether enforcing sLeX display by exofucosylation promotes rapid migration of CD19 CAR T cells to BM and spleen. A sum of 3 × 10 6 BT-or FTVII-treated CD19-CAR T cells were i.v. infused in NSG mice previously intra-BM transplanted with CD19+ target cells, and the ability of CD19-CAR T cells to colonize the BM and spleen was analyzed as early as 24 and 72 h after (Supporting information Figure S1A). In line with the in vitro data, similar numbers of BT-and FTVII-treated CAR T cells were found in PB and BM 24 and 72 h ( Figure S1B) after CAR T-cell infusion, regardless the cytokines used during T-cell stimulation. These data suggest that, at least as a "stand-alone" strategy, exofucosylation of CAR T cells does not speed-up CD19-CAR T-cell colonization of BM.
We next investigated whether exofucosylation endows CD19-CAR T cells with an enhanced cytotoxic activity or prolonged persistence. Three in vivo models using differentially aggressive targets cells were employed ( cells (in the SEM model) to more accurately assess whether FTVII-treatment may provide an improved in vivo cytotoxic activity of CAR T cells when administered in limited numbers, and found that exofucosylation did not endow CAR T cells with an improved antileukemia effects regardless the cell dose infused (Supporting information Figure  S2). We finally measured the frequency of sialofucosylated (HECA452+) CAR T cells in both BM and PB at sacrifice of the B-ALL PDX models, and found identical levels (∼80%) of sLeX display in both BT-and FTVII-treated CAR T cells ( Figure 4E), further indicating cell-autonomous sialofucosylation of in vivo-expanded CAR T cells.
Of note, the results from these three in vivo models were fully reproduced with IL-7/IL-15-activated/expanded CAR T cells (Supporting information Figure S3), further validating that glycoengineered sLeX display in CAR T cells is dispensable for CAR T-cell activity and persistence, regardless the cytokines used during T-cell stimulation. However, regardless enforced exofucosylation of CAR T cells, IL-2-expanded CD19-CAR T-cells displayed a better control of the disease coupled to a higher T-cell persistence than IL-7/IL-15-expanded CD19 CAR T cells in all the in-vivo leukemia models used (NALM6, SEM, and PDXs, Figures 3 and 4 versus Supporting information Figure S3).

DISCUSSION
CAR T-cell therapy has been acclaimed as a revolution in cancer treatment following the impressive results in hematological B-cell malignancies, especially in refractory/relapse B-ALL. However, despite the impressive response rates, CD19-directed adoptive cell immunotherapy is on its infancy, and unfortunately, a large proportion of CD19-CAR T-cell-treated patients eventually progress due to either poor CAR T-cell persistence and/or disease relapse. 23,24 Indeed, many studies in the coming years are expected to seed light into key molecular and cellular immunological mechanisms underlying CAR T-cell biology. 25 Furthermore, CAR T cells for solid tumors are lagging behind in part because the need to circumvent the physical barriers of the tumor architecture such as subverted tumor vasculature, impediments of CAR T-cell trafficking, and immunesuppressive microenvironment. 26 Similarly, the primary location for acute leukemogenesis is the BM, and the BM microenvironment provides leukemic cells with cellular interactions and signals promoting leukemia initiation, progression, and chemoresistance. [6][7][8][9] However, CAR T cells in patients suffering from acute leukemias are systemically infused via the bloodstream. Practically all cellular therapies systemically administered to treat hematological malignancies such as transplantation of unmodified or gene therapy-modified HSPCs, 27 or infusion of donor unmodified immune cells rely on efficient seeding in the leukemic BM. 28 Similarly, cell therapy based on MSCs for graft-versus-host disease or inflammatory conditions also rely on successful MSC trafficking/homing to the damaged tissue. 29,30 Here, we have hypothesized that CAR T-cell immunotherapies (CD19-CAR as a working model) in acute leukemia patients may also benefit from a rapid and effective redirection of effector cells to BM. Efficient seeding of infused CAR T cells in the leukemic BM might enhance their activity and persistence, eventually providing many clinical benefits associated to the potential reduction in the CAR T-cell dose to be infused, namely less CRS, lower production costs, and broader patientťs inclusion criteria. Previous studies from several laboratories have suggested that enforced expression ex vivo of E-selectin ligands (exofucosylation) leads to transendothelial migration of systemically administered HSPCs, MSCs, and T cells at E-selectin-expressing endothelial beds. 16,19,[31][32][33] Here, we have addressed the role of cell-autonomous and enforced sialofucosylation (sLeX display) in E-selectin ligands in the cytotoxic activity and homing ability of systemically administered CD19 CAR T cells. Taking advantage of state-of-the-art in vitro assays as well as short-and long-term in vivo xenograft models using several B-ALL cell lines and PDXs, our FACS and biochemical data revealed that cell-autonomous sialofucosylation steadily increases in culture-and in vivo-expanded CAR T cells. In contrast, a study by Mondal et al. has recently shown that in vitro-expanded CAR T cells do not exhibit sLeX expression/E-selectin binding capacity. 34 One may attribute such differences to the use of different scFvs, T-cell activation conditions, biological differences of the donor T-cells employed (HLA haplotypes, age, comorbilities), etc. We have systematically compared side-by-side IL-2-versus IL-7/IL-15-based T-cell activation conditions, concluding that the cytokines used for T-cell activation influence the degree of cell-autonomous sLeX expression in CAR T cells. However, regardless of the cytokines used for T-cell activation, cell-autonomous sLeX expression/Eselectin binding capacity gradually increased in cultureand in vivo-expanded CAR T cells. Moreover, identical levels of sLeX expression/sialofucosylation were observed in in vitro-expanded CAR+ and CAR-T cells, suggesting a CAR-independent cell-autonomous sialofucosylation of activated culture-expanded T cells (data not shown).
Our results using multiple in vitro and in vivo xenograft models revealed that further enforced sLeX-installation in E-selectin ligands improves neither the cytotoxic activity nor BM/spleen homing of vascularly administered CD19-CAR T cells, regardless of the cytokines used for T-cell activation. Of note, the T-cell phenotype was not altered by either the cytokines used for T-cell expansion or the exofucosylation reaction. This is in line with the reported cell-autonomous steady sialofucosylation of in vitro culture-expanded CAR T cells prior to in vivo infusion. Furthermore, the frequency of HECA452+ CAR T cells was found very similar in xenografts infused with either BT-or FTVII-treated CAR T cells, suggesting that cellautonomous sialofucosylation of T cells in vivo seems sufficient for proper in vivo effector function. A major difference between our experimental design and that by Mondal et al. is that, in our study, CAR T cells were infused in NSG mice previously transplanted with CD19+ target cells, thus, making our in vivo model more informative. It should be noted that CAR T cells are not expected to migrate or persist in BM or spleen in the absence of target antigen. Therefore, our xenograft models permit a physiologically more relevant in vivo assessment of (i) the trafficking ability of the infused exofucosylated CAR T cells in the presence of target antigen-expressing leukemic niches, and (ii) the cytotoxic activity of exofucosylated CAR T cells. In a different adoptive immunotherapy context, tumor infiltrated lymphocytes (TILs) have been recently reported to display an increased in vivo but also in vitro cytotoxic activity upon exofucosylation, suggesting that the enhanced sLeX expression seemed important for the target cell recognition. 35 Furthermore, the mechanisms for activation/expansion and target cell recognition of TILs clearly differ from those from CAR T cells, further explaining such experimental discrepancies between the distinct effector cells used for adoptive cell immunotherapies. Of note, regardless enforced exofucosylation of CAR T cells, IL-2-expanded CD19-CAR T cells showed a better control of the disease coupled to a higher T-cell persistence than IL7-/IL-15-expanded CD19-CAR T cells in all the in-vivo leukemia models employed in the present study (NALM6, SEM, and PDXs), suggesting that adequate T-cell expansion protocols may benefit the manufacturing and clinical outcome of CAR T cells. Collectively, our results support that the cytokines used during T-cell activation influence both the degree of cell-autonomous sialofucosylation and the CD19-CAR T-cell efficacy and persistence in vivo. However, at least as a "stand-alone" strategy, glycoengineered exosialofucosylation of E-selectin ligands seems dispensable for CD19-CAR T-cell activity and BM homing in multiple xenograft models regardless of cytokines employed for T-cell expansion. Which and how alternative cellular and molecular mechanisms regulate the migration of circulating CAR T-cells to BM needs to be explored in further studies.

CD19-CAR vector, lentiviral production, T-cell transduction, activation and expansion
Our clinically validated pCCL lentiviral second-generation CD19CAR backbone containing a human CD8 transmembrane (TM) domain, human 4-1BB and CD3z endodomains, and a T2A-GFP cassette has been reported elsewhere. 36,37 CAR-expressing viral particles pseudotyped with VSV-G were generated in 293T cells using standard polyethylenimine transfection protocols, and were concentrated by ultracentrifugation as described elsewhere. 38 Viral titers were consistently in the range of 10 8 TU/mL. Peripheral blood mononuclear cells (PBMCs) were isolated from buffy coats from healthy volunteers by using Ficoll-Hypaque gradient centrifugation. Buffy coats were obtained from the Barcelona Blood and Tissue Bank upon institutional review board approval (HCB/2018/0030). T cells were plate-bound activated with anti-CD3 and anti-CD28 antibodies for 2 days and then transduced with CAR-expressing lentivirus (multiplicity of infection = 10) in the presence of either interleukin-2 (IL-2, 50 UI/mL, Mitenyi Biotec) or IL-7 and IL-15 (10 ng/mL, Mitenyi Biotec). 38 Proper CAR expression, T-cell activation, and expansion was confirmed at the end of the activation period, as previously described. 38

Exofucosylation reaction
CD19-CAR T cells expanded for 9 days either with IL-2or IL-7/IL-15 were treated on Hanks' Balanced Salt solution (0.1% human serum albumin and 10 mM HEPES) with GDP-fucose (Biosynth Carbosynth, Compton, UK) and FTVII (RD Systems). One million cells were incubated in 20 μL of buffer containing 1 mM of GDP-fucose and 70 μg/mL of purified FTVII enzyme at 37 • C for 1 h as previously detailed. 17 Control cells were incubated in the same solution but without FTVII/GDP-fucose (buffertreated [BT] cells). After the enzymatic reaction, cells were always washed twice with PBS before downstream experiments.

E-selectin-Ig immunoprecipitation
CAR T-cell lysates were prepared using lysis buffer containing 150 mM NaCl, 50 mM Tris-HCI (pH 7.4), 2% Nonidet P-40, 2 mM CaCl2, and protease and phosphatase inhibitor cocktails (Roche). When indicated, 5 mM EDTA was added to the lysis buffer as negative control condition. A sum of 5 × 10 6 cells were pelleted per condition, washed with PBS, and lysed in 500 μL of lysis buffer. Cell lysates were incubated on ice for 15 min and centrifuged at 12 000 g for 10 min. Cell lysates were then precleared overnight using protein G-agarose beads (Roche) and incubated for 2 h at 4 • C with 3 μg of murine E-selectinhuman Fc chimera ("E-Ig," R&D Systems), as described. 17 Agarose beads were then washed twice with lysis buffer, and immunoprecipitated glycoproteins were collected by boiling the beads in the presence of 2-mercaptoethanol in Laemmli loading buffer. For western blot (WB) analysis, immunoprecipitates were resolved on a 7.5% SDS-PAGE gel (Bio-Rad Laboratories) and then transferred onto a polyvinylidene difluoride (PVDF) membrane (Bio-Rad). The membrane was then blocked with blocking reagent (Chemiluminescence Western Blotting Kit, Roche), and incubated with monoclonal antibodies (MoAb) against PSGL1 (clone KPL1, BD), CD43 (clone IG10, BD), and CD44 (clone 2C5, R&D). Protein bands were detected by chemiluminescence using Lumi-Light substrate (Chemiluminescence Western Blotting Kit).

In vitro cytotoxicity assays
Luciferase (Luc)/GFP-expressing-NALM6 cells were kindly provided by Prof. RJ Brentjens (MSKCC, NY). SEM were generated by retroviral transduction and GFP-based FACS-selection. 39 Jurkat were purchased from DSMZ. Target cells were labeled with 3 μM eFluor670 (eBioscience) and incubated with BT-and CD19-CAR T cells at different Effector:Target (E:T) ratios. CAR T-cell-mediated cytotoxicity was determined by analyzing the residual alive (7-AAD-) eFluor670+ target cells after 24 h CAR T-cell exposure.

HUVEC transwell assays
HUVEC were maintained in EGM-2 Endothelial Cell Growth Medium-2 BulletKit (Lonza, Cultek SLU), as previously described. 40 Early passage HUVECs were plated on 24-well Transwell plates (5 μm polycarbonate membrane, 6.5 mm insert), and stimulated with 40 ng/mL of TNF-α (R&D) for 4 h at 37 • C to activate cell surface expression of E-selectin and VCAM-1.17 A sum of 2 × 10 5 of each NALM6 cells and BT-or FTVII-treated CAR T cells were seeded in the bottom and upper chamber, respectively. Twenty-four hours later, the absolute number of alive (7AAD-) NALM6 and CAR T cells present in the bottom chamber were quantified using Trucount tubes (BD Biosciences). 38

FACS analysis
Cell staining and FACS analysis were performed as extensively described 38  CD22-APC, and CD19-BV421. IgG1-APC, IgG1-PE, and Rat IgM-BV421 were used as isotype controls. All MoAb were purchased from BD Biosciences. Supporting information Figures S1B and S4 show the gating strategies for T-cell analysis.

Statistical analysis
Data are shown always from at least three individual donors. At least five animals were used per condition. Pvalues were calculated by an unpaired two-tailed Studentťs t-test using Prism software (GraphPad).

A C K N O W L E D G M E N T S
We are indebted to Juan José Rodríguez-Sevilla and PM's lab members for their technical feedback and constructive discussions.
We thank CERCA Programme / Generalitat de Catalunya and Fundacio Josep Carreras-Obra Social la Caixa for their institutional support. The research leading to these results has received funding from the European Research Council (CoG-2014-646903, PoC-2018-811220), the Spanish Ministry of Economy and Competitiveness (MINECO, SAF2016-80481R and SAF2019-108160R), the Fundacion Uno entre Cienmil, "la Caixa" Foundation (ID 100010434) under the agreement LCF/PR/HR19/52160011, and the Spanish Association against cancer (AECC, Semilla 2019) to PM. DSM is partially supported by a Sara Borrell fellowship from the Instituto de Salud Carlos III. MV is partially supported by a Juan de la Cierva fellowship from the MINECO. SRZ was supported by Marie Sklodowska Curie Fellowship. PM is an investigator of the Spanish Cell Therapy cooperative network (TERCEL). Manuscript writing: DSM, PM.

C O N F L I C T O F I N T E R E S T
The authors have nothing to disclose.

A P P R O VA L A N D C O N S E N T T O PA R T I C I PAT E
This study was IRB-approved by the Barcelona Clinic Hospital Ethics Committee (HCB/2019/0450). All in vivo procedures were approved by the Animal Care Committee of The Barcelona Biomedical Research Park (HRH-17-0029-P1).

AVA I L A B I L I T Y O F D ATA A N D M AT E R I A L
The datasets and materials generated in this study are available from the corresponding author on reasonable request.