L-selectin controls trafficking of chronic lymphocytic leukemia cells in lymph node high endothelial venules in vivo

B-cell chronic lymphocytic leukemia (CLL) is the most common leukemia in adults with highly variable clinical progression, characterized by accumulation of malignant B cells in blood, bone marrow, lymph nodes (LNs), spleen, and liver.^1,2  Accessory cells such as T cells, mesenchymal stromal cells, and nurse-like cells constitute a protective microenvironment for CLL cells in lymphoid organs.^3-5  They prevent the apoptosis of CLL cells and support their clonal proliferation within specific structures known as proliferation centers. The tissue microenvironment is thus believed to play a critical role in CLL disease progression.^3-5  Indeed, a bulky lymphadenopathy is a feature of advanced disease correlating with poor clinical outcome. Importantly, CLL cell migration to LNs provides protection from chemotherapy and immunotherapy.⁶  Interfering with CLL cell migration or retention in LNs could thus result in a greater efficacy of conventional therapy for advanced CLL patients.

Several chemokine receptors have been proposed to play important roles in CLL cell trafficking (ie, CXC chemokine receptors [CXCRs] CXCR4 and CXCR5, and CC chemokine receptor 7 [CCR7]).^5,7  CXCR4, the receptor for CXC chemokine ligand 12 (CXCL12), was the first chemokine receptor shown to be expressed and functional on CLL cells.⁸  CLL cells were later found to express high levels of CCR7^9,10  and CXCR5,^10-12  the receptors for chemokines CC chemokine ligands (CCLs) CCL19/CCL21 and CXCL13, respectively. In vitro chemotaxis assays have revealed that CLL cells efficiently migrate in response to all ligands of CCR7, CXCR4, and CXCR5.^8-12  Together, these studies suggested that lymphoid chemokines and their receptors, which play critical roles in B-cell homing to LNs and other secondary lymphoid organs, are likely to be also important for CLL cell migration in vivo.^8-12 

Despite these important advances on the role of chemokine receptors in CLL cell trafficking, the mechanisms regulating CLL cell migration in vivo remain incompletely characterized. For instance, it is currently unknown which molecules control the initial capture and trafficking of CLL cells in high endothelial venules (HEVs), specialized blood vessels for lymphocyte entry in LNs.^13,14  B lymphocytes use a multistep adhesion cascade (rolling, sticking, crawling) to attach to HEV walls in vivo.¹³  Whether CLL cells exhibit a similar behavior has not yet been determined. Here, we used an in vivo imaging approach to address these unresolved and important issues. Intravital microscopy analysis of the mouse LN microcirculation^14,15  allowed us to visualize for the first time the rolling, sticking, and crawling of human CLL cells on HEV endothelium in vivo, and to demonstrate the critical role of the lymphocyte homing receptor L-selectin (CD62L)^16-18  in these processes. Interestingly, we observed downregulation of L-selectin on CLL cells from patients under treatment with the phosphoinositide 3-kinase δ inhibitor idelalisib,^19,20  suggesting an inhibitory effect of this drug on CLL cell migration to LNs.

Patient’s tissues and blood were obtained following standard ethical procedures (Helsinki principles), after informed written consent, and stored at the HIMIP collection. According to the French law, the HIMIP collection was declared to the Ministry of Higher Education and Research (DC 2008-307 collection 1); a transfer agreement (AC 2008-129) was obtained after approbation by the “Comité de Protection des Personnes Sud-Ouest et Outremer II” (ethical committee). Clinical and biological annotations of the samples have been declared to the Commission Nationale de l’Informatique et des Libertés (CNIL). Animal experiments were conducted according to institutional guidelines for animal handling and using protocols approved by the Institut de Pharmacologie et de Biologie Structurale and Région Midi-Pyrénées animal care committees.

Peripheral blood samples from 30 patients with CLL/small lymphocytic lymphoma (SLL) who underwent routine follow-up visits at Service d’Hématologie (Institut Universitaire du Cancer Toulouse [IUCT]–Oncopôle, Toulouse, France) were collected. Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation over Ficoll-Plaque (GE Healthcare) and used immediately. Purity of CD19⁺CD5⁺ B cells was assessed to be >90% by flow cytometry. CLL/SLL patient characteristics are described in Table 1. Patients were untreated at the time of blood collection, with the exception of patients treated with idelalisib (Table 1). Bulky disease was determined by a clinical examination and defined as presence of nodal mass ≥5 cm and/or spleen size ≥6 cm below left costal margin at physical examination. Peripheral blood samples from 6 second- to third-line patients (Table 1) who received idelalisib 150 mg twice a day (national patient drug access program) in combination with anti-CD20 immunotherapy were collected before treatment and at the indicated time under treatment.

Wide-field intravital microscopy was performed as previously described.^14,15  In brief, calcein-labeled PBMCs were injected into the right femoral artery of C57BL/6 mice (Charles River Laboratories), and fluorescent events in the left inguinal LN microcirculation were visualized and recorded. Rolling fractions (the percentage of rolling cells in the total flux of cells in each venule), sticking fractions (the percentage of rolling cells that subsequently arrested for >30 seconds), and rolling speed (rolling velocity [V_roll]) were determined as previously described.^14,15  When indicated, cells were pretreated for 30 minutes with a function blocking anti-L-selectin (CD62L) antibody (Ab) (6 μg/mL, DREG56; eBioscience) or with an isotype control. Multiphoton imaging was performed on a 7MP upright microscope (Carl Zeiss). Excitation was provided by a Ti-Sapphire femtosecond laser, Chameleon Ultra 2 (Coherent Inc) tuned at 800 nm; emitted fluorescence was collected with 2 non-descanned detectors: Channel 2 (500-550 nm) for carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled PBMCs and Channel 3 (565-610 nm) for Qtracker-565 (Molecular Probes)-labeled vessels. Mice were anesthetized and placed on a customized stage for securing mice and immobilizing the inguinal LN on a heating pad with temperature feedback to an mTCII micro-Temperature Controller (Cell MicroControls). The crawling speed (crawling velocity [V_crawl]) was determined with Imaris (Bitplane) on image sequences. See supplemental Methods (available on the Blood Web site) for additional information.

LNs were either fixed in 10% formalin and embedded in paraffin or embedded in optimal cutting-temperature compounds and snap-frozen in liquid nitrogen. For histology, 3-µm-thick sections were used; for confocal microscopy, 5-µm-thick sections were used. HEV density and HEV pockets were analyzed using the HEV-specific antibody MECA-79 (ATCC). The detailed procedures and antibodies used are described in supplemental Methods. Panoramic Viewer and HistoQuant software were used for, respectively, viewing and analyzing the digitalized slides (RTM 1.5.053; 3D Histech).

Freshly isolated PBMCs were incubated with the following fluorochrome-conjugated antibodies (Abs), CD19-phycoerythrin (PE; HIB19), CD5-fluorescein isothiocyanate (4CHT2), CD62L-PECy5 (DREG-56), CXCR4-allophycocyanin (12G5), or CD11a-PE (HI-111) (all from BD Biosciences), and analyzed with a FACSCalibur flow cytometer (BD Biosciences) and FlowJo software (Tree Star). The mean fluorescence intensity (MFI) ratio (MFIR) was determined by dividing the MFI of the analyzed antigens by the MFI of the isotype control. In some experiments, freshly isolated PBMCs were incubated ex vivo for the indicated time at 37°C with idelalisib (5 µM; Selleckchem) or dimethylsulfoxide (untreated conditions) before analysis by flow cytometry.

Data are shown as mean ± standard deviation (SD) or mean ± standard error of the mean (as indicated) and were analyzed using a 2-tailed unpaired Student t test. When indicated, a 2-tailed paired Student t test or Mann-Whitney test were instead applied. Differences were considered statistically significant when P < .05.

Table 1 presents the characteristics of the different groups of patients analyzed in this study. Blood samples (BLs) were obtained from untreated CLL/SLL patients (with or without bulky disease) and idelalisib-treated CLL patients. For the idelalisib-treated group, blood was collected both before and under treatment of each patient. LN biopsies were obtained from a different group of CLL/SLL patients with various clinical characteristics.

To analyze the HEV network, we stained human CLL and control LNs with the HEV-specific Ab MECA-79 which recognizes sulfated ligands for lymphocytes (Figure 1A). HEV density was significantly increased in CLL LNs (n = 13) and SLL LNs (n = 5) in comparison with control LNs (n = 8) (Figure 1B). HEVs were located both outside and inside proliferation centers (Figure 1C-D), and expressed the 6-sulfosialyl Lewis X oligosaccharides ligands for the lymphocyte homing receptor L-selectin, recognized by the HEV-specific Abs G72 and G152²¹  (Figure 1E). L-selectin⁺ cells were located inside, around (<10 μm), or at a distance (>10 μm) from MECA-79⁺ HEVs (Figure 1F). However, there was no specific enrichment of L-selectin⁺ cells around MECA-79⁺ HEVs (Figure 1G), probably because L-selectin shedding occurs during extravasation through HEVs.²² 

Confocal microscopy analysis revealed that CD20⁺ CLL cells are frequently found within HEV pockets in human CLL LNs (Figure 2). HEV pockets are highly dynamic structures that are closely associated with lymphocyte extravasation through HEVs.^13,23,24  Indeed, they have been shown to disappear when lymphocyte migration to LNs is inhibited.^23,24  CD20⁺ cells nested in HEV pockets were separated from the HEV lumen by the membrane of the HEV endothelial cells stained with MECA-79 and CD31 antibodies (Figure 2). As previously described in murine LNs,²³  HEV pockets in human CLL LNs were able to house up to 4 to 5 CD20⁺ cells. Together, these observations suggested intense trafficking of CLL cells through HEVs in CLL LNs.

To visualize the interaction of human CLL cells with HEV endothelium in vivo, we next performed intravital microscopy analyses in LNs of wild-type mice, making use of the evolutionary conservation of the vasculature and the unique capacity of human leukemic cells to migrate to murine bone marrow and lymphoid organs.²⁵  The venular tree of the inguinal LN consists of up to 5 branching orders (I-V), and higher order venules (III-V) correspond to HEVs^13-15  (Figure 3A). Human CLL cells, obtained from freshly isolated samples to avoid any phenotypic alteration after culture or freezing, were fluorescently labeled ex vivo, and injected in recipient mice for in vivo imaging of their behavior in the LN microcirculation. Human CLL cells were observed to roll and stick (arrest) within LN venules with numerous interactions in order III to V HEVs (Figure 3B). More interactions were observed for CLL cells from patients with bulky disease and high circulating lymphocyte counts (≥40 000 cells/mm³; Table 1; supplemental Videos 1-2) than for CLL cells from patients without bulky disease (Table 1; supplemental Video 3). Indeed, 60% of the CLL cells from patients with bulky disease rolled in order V HEVs and >10% of these cells arrested, compared with <30% and 3%, respectively, for CLL cells from patients with no bulky disease (Figure 3B). In addition, CLL cells from patients with bulky disease rolled more slowly in HEVs than cells from patients without bulky disease (Figure 3C). Intravital 2-photon laser-scanning microscopy revealed that after arrest, CLL cells were able to crawl on the HEV endothelium, some of them in the opposite direction of blood flow (Figure 3D; supplemental Video 4). The average intraluminal velocity of the CLL cells was ∼6 μm per minute (Figure 3E). This velocity is similar to that of murine B cells in LN HEVs.²⁶  We concluded that human CLL cells are able to roll, stick, and crawl on HEV endothelium in vivo.

We then investigated the molecular mechanisms controlling the behavior of CLL cells within LN HEVs. We first analyzed the expression of the lymphocyte homing receptor L-selectin,^17,18  which mediates the initial capture and rolling interactions of lymphocytes on HEV endothelium,^13,16,27,28  and is known to be expressed on CLL cells.^29,30  We found that L-selectin is expressed at significantly higher levels on CLL cells from patients with bulky disease compared with cells from patients without bulky disease (Figure 4A). In contrast, similar expression levels between the 2 subgroups of patients were observed for chemokine receptor CXCR4 and integrin LFA-1, which mediates the sticking of lymphocytes to HEV endothelial cells.¹³  In the course of these studies, we identified a third subgroup of patients with bulky disease but lower circulating lymphocyte counts (<40 000 cells/mm³; CLL/SLL patients BL01-03, Table 1). Interestingly, CLL/SLL cells from this subgroup expressed low levels of L-selectin and had a reduced capacity to roll and stick to HEVs (supplemental Figure 1), compared with CLL cells from patients with bulky disease and high circulating lymphocyte counts (≥40 000 cells/mm³; CLL patients BL15-BL24, Table 1). Together, these results suggested that differences in L-selectin expression could explain the different behavior of CLL cells in LNs HEVs in vivo. Functional analyses revealed that a blocking antibody to L-selectin had a strong inhibitory effect on CLL trafficking in LN HEVs; it reduced the rolling and completely abrogated the sticking of CLL cells to HEV walls (Figure 4B). We obtained similar results with circulating CLL cells from 3 patients with bulky disease (Figure 4B-D). Interestingly, reduction in rolling and sticking upon blockade of L-selectin also occurred with CLL cells from patients without bulky disease (3 patients shown in Figure 4C-D). We concluded that L-selectin is the critical determinant controlling the initial rolling and sticking interactions of CLL cells with HEV endothelium in vivo.

Idelalisib (formerly known as GS-1101 and CAL-101) is a drug, recently approved for CLL patients, which induces LN shrinkage and lymphocytosis as a consequence of CLL cell redistribution into the peripheral blood.^5,19,20,31  To study the effect of idelalisib on CLL cell trafficking to LNs in vivo, we first measured the expression level of L-selectin at the surface of CLL cells, before and during idelalisib treatment. Interestingly, we observed that L-selectin expression is decreased on circulating CLL cells from patients treated with idelalisib (Figure 5A; supplemental Figure 2). Intravital microscopy analysis of CLL cells from a patient with bulky disease before treatment and after 3 or 7 weeks of treatment revealed that rolling and sticking interactions of CLL cells with HEV endothelium in vivo are reduced after treatment with idelalisib (Figure 5B). The inhibition of CLL cell sticking to HEV walls was associated with a fourfold increase in the rolling velocity of CLL cells after idelalisib treatment (Figure 5C), suggesting that cells rolled too quickly for stable arrest. To determine whether the effect of idelalisib on L-selectin expression is direct, we performed additional in vitro experiments with circulating CLL cells from 5 individual patients (Figure 5D). However, we did not observe any change in the expression of L-selectin on CLL cells after up to 4 hours of incubation with idelalisib, suggesting an indirect effect of the drug in vivo (Figure 5E). We concluded that idelalisib treatment induces a downregulation of L-selectin on CLL cells in vivo, that may impair HEV-mediated recruitment of CLL cells in LNs.

In this study, we investigated the first step of CLL cell migration to LNs: their interaction with HEV blood vessels. We made use of the evolutionary conservation of the vasculature to analyze the behavior of human CLL cells in the mouse LN microcirculation using intravital microscopy and multiphoton in vivo imaging. This strategy allowed us to visualize for the first time the rolling, sticking, and crawling of CLL cells on HEV endothelium in vivo. Functional analyses revealed that the lymphocyte homing receptor L-selectin controls the interaction of CLL cells with HEVs in vivo. Interestingly, L-selectin was downregulated on CLL cells from patients treated with idelalisib, a phosphoinositide 3-kinase δ inhibitor recently approved in clinics for CLL treatment.

Our results show that, similar to B cells, CLL cells use a multistep adhesion cascade to bind to HEV walls in vivo. CLL cells circulating in the blood tether and roll on HEV walls through the binding of L-selectin to HEV endothelial cells. Rolling of CLL cells on the HEV endothelium is essential for subsequent arrest, as illustrated by the fact that a function-blocking antibody to L-selectin, which reduced rolling, completely blocked the arrest of CLL cells (Figure 4B,D). We observed a significantly increased rolling fraction and decreased rolling velocity for CLL cells from patients with bulky disease and high circulating lymphocyte counts (≥40 000 cells/mm³) (Figure 3C) which exhibit high expression levels of L-selectin (Figure 4A). The decreased rolling velocity is likely to facilitate chemokine-induced integrin-mediated arrest of CLL cells.¹³  In contrast, we observed a decreased sticking fraction and an increased rolling velocity (Figure 3B-C) for cells from CLL patients without bulky disease (including 3 patients with very high leukocytosis, ie, lymphocyte counts >100 000 cells/mm³), which express lower levels of L-selectin (Figure 4A). This suggested a possible defect for these cells in trafficking through LN HEVs in vivo. However, a certain level of rolling and sticking was still observed and it was dependent on L-selectin (Figure 4C-D). In addition, we observed low L-selectin expression and reduced rolling and sticking fractions for cells from patients with bulky disease but lower circulating lymphocytes counts (<40 000 cells/mm³, including 1 SLL patient; supplemental Figure 1). These patients may thus have a defect in CLL/SLL cell entry in LNs through HEVs. L-selectin expression was also decreased on CLL cells from patients after treatment with idelalisib in vivo (Figure 5A) and this was associated with increased rolling velocity (Figure 5C) and reduced sticking of CLL cells to HEV walls (Figure 5B). These observations are in agreement with previous findings showing that the number of L-selectin molecules at the cell surface dictates the rolling behavior of T and B lymphocytes within LN HEVs.³²  Together, our findings indicated that L-selectin is a critical determinant of CLL cell trafficking in HEVs in vivo.

A major strength of our study is providing the first direct visualization and functional analysis of human CLL cell trafficking in the LN microcirculation in vivo. It is important to point out that the HEV ligands for L-selectin (6-sulfo-sialyl Lewis X) are sulfated sugars that are identical between human and mouse.¹³  These L-selectin ligands are expressed to high levels in both human and mouse HEVs, and they are recognized by the HEV-specific monoclonal Ab (mAb) MECA-79, both in human and mouse. L-selectin binding to HEV endothelial cells is thus evolutionarily conserved between human and mouse. Indeed, our intravital microscopy studies with function-blocking antibodies against human L-selectin (Figure 4B-D) convincingly demonstrate that L-selectin controls the rolling and firm arrest (sticking) of human CLL cells in LN HEVs in vivo. In the future, it would be interesting to further characterize the role of L-selectin in CLL trafficking through HEVs in mouse models of CLL, for instance by using the well-characterized Eμ-TCL1 model.³³  Finally, it would have been nice to complement our in vivo imaging studies in mice with in vitro studies using human HEV endothelial cells. Unfortunately, it is not possible to culture human HEV endothelial cells in vitro because we showed previously that they rapidly de-differentiate outside of the lymphoid tissue microenvironment.³⁴  In addition, no human HEV cell line or cultured human endothelial cells that exhibit the bona fide HEV phenotype (ie, expressing the sulfated ligands recognized by L-selectin and HEV-specific mAb MECA-79) have been described yet.

We found that the density of HEVs was increased in human CLL LNs and that CD20⁺ CLL cells frequently accumulated within HEV pockets (Figure 2). To the best of our knowledge, this is the first description of HEV pockets in human LNs. In mouse LNs, HEV pockets have been shown to function as “waiting areas” in which lymphocytes were held until space was made available to them for entry into the LN tissue.²³  They were found to be highly dynamic structures that are continuously altered in size and location by lymphocyte migration. Interestingly, an antibody that blocks L-selectin–mediated lymphocyte entry in LNs through HEVs induced the disappearance of HEV pockets in vivo.²³  Therefore, the presence of many HEV pockets occupied by CLL cells in human CLL LNs strongly suggests that HEVs mediate high levels of CLL cell entry in LNs.

In this study, we have focused on CLL cell entry (or re-entry) in LNs. However, another key aspect in CLL cell trafficking is the egress from LNs. In normal B cells, this process depends on the sphingosine-1 phosphate receptor S1PR1,^13,26  a receptor which has been shown to be reduced on CLL cells from patients with unfavorable prognosis.³⁵  Defective egress caused by impaired S1PR1 expression was proposed to contribute to the extended survival of CLL cells by prolonging their residency in the prosurvival niche of LNs.³⁵  Egress occurs only minimally in SLL, and differential expression of S1PR1 may also explain some of the differences between CLL and SLL. Interestingly, idelalisib has recently been shown to increase S1PR1 expression on CLL cells,³⁶  an effect that could contribute to the rapid resolution of lymphadenopathy (LN shrinkage) and redistribution of CLL cells into the blood (lymphocytosis) that have been observed after idelalisib treatment.^19,31  In this study, we found that treatment with idelalisib results in decreased levels of L-selectin on CLL cells in vivo and reduced binding of CLL cells to HEV walls in LNs. We do not know the mechanism for this downregulation of L-selectin in vivo but it may be related to the profound CLL cell redistribution induced by idelalisib^5,19,20,32  because our in vitro studies indicate that it is not a direct effect. Idelalisib has been shown to interfere with integrin-mediated adhesion of CLL cells to endothelial cells in vitro³⁷  and to inhibit chemotaxis of CLL cells toward lymphoid chemokines.²⁰  Idelalisib could thus not only induce exit of CLL cells from LNs,^5,20  but also inhibit their re-entry through HEVs via L-selectin downregulation and interference with integrin-mediated adhesion and chemotaxis. Both effects, induction of CLL cell exit and inhibition of re-entry, could contribute to LN shrinkage and redistribution of CLL cells into the blood.

Together, our findings suggest that interfering with L-selectin–mediated CLL cell trafficking in HEVs could represent a novel strategy to block dissemination of CLL cells to LNs. This strategy could be particularly useful to increasing the efficacy of conventional therapy for advanced CLL patients with bulky lymphadenopathy and high levels of L-selectin on their leukemic cells. Targeting L-selectin might prove to be better than targeting chemokine receptors because multiple chemokine receptors (ie, CXCR4, CXCR5, CCR7) contribute to CLL cell migration,^8-12  whereas a single molecule (ie, L-selectin) controls the initial capture and rolling interaction of CLL cells with HEV endothelium in vivo.

The authors thank the Institut de Pharmacologie et de Biologie Structurale (IPBS)_TRI facility, the IPBS_ANEXPLO facility, and the INSERM Unité Mixte de Recherche 1043 cellular imaging facility. The authors are grateful to R. Kannagi for gift of Abs G72 and G152, V. Thuries, L. Jalabert, and G. Perez for help with immunohistochemistry, and F. X. Frenois for whole-slide imaging studies (Service d’Anatomie et de Cytologie Pathologique, IUCT-Oncopôle); they are also grateful to all patients and their families and to the SILLC Association.

This work was supported by Laboratoire d’Excellence Toulouse Cancer (LABEX TOUCAN, “Integrative Analysis of Resistance in Hematological Cancers”), Fondation Recherche Innovation Thérapeutique Cancérologie (Fondation RITC), Région Midi-Pyrénées, and Fondation ARC Pour la Recherche sur le Cancer (ARC Equipment No. 8505).

Contribution: F.L. performed the experiments, analyzed the results, and was involved in study design and manuscript preparation; E.B., C.L., and C.M. performed the experiments and analyzed the results; J.-J.F. was involved in study design; L.Y. was involved in study design and analyzed the results; and J.-P.G. designed the study, analyzed the results, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Jean-Philippe Girard, IPBS-CNRS, 205 Route de Narbonne, 31077 Toulouse cedex 4, France; e-mail: jean-philippe.girard@ipbs.fr.