P-selectin glycoprotein ligand regulates the interaction of multiple myeloma cells with the bone marrow microenvironment

The progression and dissemination of multiple myeloma (MM) involves the continuous spread of MM cells in and out of the BM.^1,2  Interactions of MM cells with the BM microenvironment play a critical role in the pathogenesis of MM and in the development of drug resistance.^3-5  We have shown that blocking the CXCR4/SDF1 axis disrupts the interaction of MM cells with the BM microenvironment, which in turn enhances the efficacy of therapeutic agents against MM cells.⁶ 

Selectins are cell-surface adhesion molecules that contain a lectin-like domain with selectivity for binding to specific saccharide chains.⁷  Each of the 3 types of selectins (E, L, and P) has a unique tissue distribution (in the endothelium, leukocytes, and platelets, respectively),⁷  and distinct classes of leukocytes use specific combinations of selectins to interact with endothelium.^8,9  The binding affinity of selectins to their ligands is relatively low, but is nonetheless strong enough to serve as a biologic brake that rapidly decelerates leukocytes as they roll on endothelial cells (ECs). While rolling, leukocytes are activated by binding to selectin and by chemoattractants such as CXCR4/SDF-1; activation increases the affinity of the integrins on leukocytes for ligands found in the endothelium. The presence of a chemotactic signal outside of a venule induces leukocytes to squeeze between the ECs of the venule and migrate into the target organ (extravasation)^10,11 ; inhibition of rolling by blocking selectins decreases this extravasation.^12-14  Small-molecule inhibitors of selectins have clinical activity.^12,15-18  Synthetic inhibitors of selectin also have a demonstrated ability to improve the manifestation of psoriasis and allergen-induced asthma in humans¹⁹  and in mouse models of skin inflammation.²⁰  These agents are currently being tested in clinical trials for the treatment of inflammatory diseases and in preclinical studies of solid-tumor metastasis.^21-24 

P-selectin glycoprotein ligand-1 (PSGL-1) is a dimeric, mucin-type glycoprotein ligand that is expressed by all leukocytes and is involved in the homing of leukocytes to target tissues.²⁵  PSGL-1 plays an important role in organ targeting during inflammation, and inhibition of PSGL-1 represents an attractive basis for anti-inflammatory strategies.^25,26  Earlier studies have shown that PSGL-1 is highly expressed in MM,^22,27  is a novel therapeutic target for mAb-mediated MM immunotherapy, plays a role in humoral immunotherapy of MM, and combined treatment with PSGL-1 mAb and chemotherapy improves tumor cytotoxicity.^23,28 

In the present study, we targeted PSGL-1 by inhibiting its interaction with selectins in the microenvironment as a therapeutic prospect but with a focus on adhesion dynamics that involve PSGL-1 on MM cells and its interaction with selectins in the BM microenvironment. We show that PSGL-1 regulates the adhesion and homing of MM cells to cells in the BM microenvironment, including ECs and BM stromal cells (BMSCs). We also document that this interaction regulates the proliferation and development of drug resistance by MM cells, both in vitro and in vivo. Furthermore, inhibiting this interaction with GMI-1070, a selective novel therapeutic agent that targets selectins, enhances the sensitization of MM cells to bortezomib in vitro and in vivo. These data highlight the critical contribution of PSGL-1 to the regulation of growth, dissemination, and drug resistance in MM in the context of the BM microenvironment.

The pan-selectin inhibitor GMI-1070 was obtained from GlycoMimetics. Recombinant SDF-1, recombinant selectins, and Abs against E-, L-, and P-selectins were purchased from R&D Systems. mAbs for Western blotting were procured from Cell Signaling Technologies, anti–PSGL-1 Ab was from Millipore, and anti–β1-integrin Ab (clone 12G10) was from Abcam. Calcein-AM cell-labeling dye was obtained from Invitrogen. Scramble-siRNA and SmartPool siRNAs for PSGL-1, E-selectin, L-selectin, and P-selectin were from Dharmacon. Lipofectamin-2000 was from Invitrogen.

The MM1s cell line was purchased from the ATCC, whereas the OPM1, OPM2, H929, RPMI8226, U266, and U266LR7 lines were the kind gift of Prof Jesús F. San Miguel (Salamanca, Spain). Human umbilical vein ECs (HUVECs) were from Lonza. CD138⁺ cells were isolated from MM patients by bead selection, and BMSC cultures were established as described previously.⁶ 

Informed consent was obtained from all patients in accordance with the Declaration of Helsinki for primary MM patient samples. Approval of the animal study protocol was obtained from the Institutional Review Board of the Dana-Farber Cancer Institute.

To detect PSGL-1, BM aspirates from 17 MM patients and 3 healthy subjects were rinsed with PBS, fixed with 4% formaldehyde in PBS, dehydrated with ethanol, embedded in paraffin, and sectioned. Tissues were then immunostained with mouse anti–human PSGL-1.

To determine the gene expression of PSGL-1 in MM, we used datasets from the Gene Expression Omnibus (Mayo Clinic series number GSE 6477). Expression levels of PSGL-1 (probe ID 209879_at) were compared in plasma cells from healthy subjects (n = 15), monoclonal gammopathy of undetermined significance (MGUS) patients (n = 20), smoldering MM patients (n = 23), and newly diagnosed MM patients (n = 68).

MM cell lines (MM1s, OPM1, OPM2, H929, RPMI8226, U266, and U266LR7) or HUVECs were treated with mouse anti–human Abs for PSGL-1; with E-, L-, or P-selectin (5 μg/mL); or with an isotype control for 1 hour on ice. Cells were then immersed in FITC-secondary Ab. Expression was determined with flow cytometric analysis and quantified as the ratio of the mean-fluorescence intensity (MFI) of the detected target to the MFI of isotype control.

To detect selectin on primary ECs, mononuclear cells from BM aspirates of MM patients were treated with primary Abs against each selectin, followed by a FITC–anti-mouse secondary Ab, washed, and treated with APC-conjugated anti-CD31 Ab. ECs were gated as APC-CD31⁺ cells, and the expression of each selectin was quantified as the ratio of the FITC-MFI of each selectin to the FITC-MFI of isotype control.

MM1s cells were cultured overnight in 6-well plates in Opti-MEM medium and washed, and cells in each well were immersed overnight in a mixture of Lipofectamin-2000 (7 μL) with scramble-siRNA or SmartPool PSGL-1 siRNA (100 pmol) in a final volume of 2 mL of Opti-MEM. Similarly, HUVECs were transfected with scramble siRNA or with SmartPool siRNA for E-, L-, or P-selectin. Twenty-four hours later, the transfection solution was replaced with complete medium and cells were used after an additional 24-48 hours.

MM1s cells transfected with scramble siRNA or PSGL-1 siRNA or treated with increasing concentrations (0, 250, and 500μM) of GMI-1070, were incubated with free human Fc-chain (isotype control) or with chimeras of human-Fc chain and recombinant human E-, L-, or P-selectin (10 μg/mL), followed by FITC-conjugated mouse anti–human Fc. The interaction of the selectins with MM cells was analyzed by flow cytometry and quantified as the ratio of the MFI of each selectin to the MFI of the isotype control.

HUVECs or BMSCs were cultured overnight to confluence in 96-well plates (5 × 10³ cells/well) before initiating the adhesion assay. Selectins on HUVECs were inhibited by knockdown with siRNA for E-, L-, or P-selectin; by treatment with blocking Abs against E-, L-, or P-selectin (10 μg/mL for 1 hour); or by increasing concentrations of GMI-1070 (0, 250, and 500mM for 1 hour) before performing the adhesion assay. MM1s, OPM1, and H929 cells were serum-starved for 3 hours, prelabeled with Calcein-AM, added to the HUVECs or BMSCs (1 × 10⁵ cells/well), and allowed to adhere for 2 hours at 37°C. Nonadherent cells were aspirated away, the HUVECS or BMSCs were washed, and fluorescence intensity was measured using a fluorescent-plate reader (Ex/Em = 485/520 nm). In some experiments, MM cells were treated with scramble-siRNA or siRNA for PSGL-1, and in other experiments with isotype control Ab or with a blocking Ab of PSGL1 (clone KPL-1; Millipore).

HUVECs were grown to confluence in 96-well plates (5 × 10³ cells/well) before performing the adhesion assay. HUVECs were then activated with vehicle or TNF-α (30 U/mL) for 3 hours or with IL-4 (3 ng/mL) for 24 hours, followed by histamine (2.25mM) for 4 hours. The cells were then treated with mouse anti–human Ab against each of the selectins for 1 hour, followed by a secondary FITC Ab for 1 hour. Nonbound Abs were washed away and the expression of selectins was assessed by measuring the fluorescence intensity (Ex/Em = 485/520 nm).

Adhesion of MM cells to HUVECs under conditions of shear flow was measured with a parallel plate flow chamber. HUVEC monolayers were grown in 35-mm tissue-culture plates, and activated. A MM cell suspension (2 × 10⁶cells/mL) was perfused through the chamber over the HUVEC monolayers at a shear rate corresponding to a wall shear stress of 0.9 dynes/cm². For each experiment, the cell suspension was allowed to flow through the chamber for 3 minutes, after which time digital images were collected to quantify the results. The IC300 digital image system (Inovision) driven by a Silicon Graphics Indigo2 workstation was used to acquire and analyze the images.

HUVECs (5 × 10³ cells/basket) were incubated overnight in the upper chamber of 8-micron pore filters (Costar; Corning) before performing the adhesion assay. Selectins on HUVECs were inhibited by knockdown with siRNA or by 1 hour of treatment with blocking Abs (10 μg/mL) or by 1 hour of treatment with GMI-1070 (0, 250, and 500μM) before carrying out the transendothelial migration assay. MM1s cells were serum-starved for 3 hours and then added to the upper chamber of the basket (2 × 10⁵ cells/well), and left to migrate for 4 hours at 37°C toward the lower chamber, which contained 0 or 30nM of SDF1α. In some cases, PSGL-1 in the MM cells was knocked down by siRNA, and in other experiments, MM cells were treated with isotype control Ab or with a blocking Ab of PSGL1. In other cases, migration was assayed without precoating the filter; in these, MM1s were treated with scramble siRNA, PSGL-1 siRNA, or with increasing concentrations of GMI-1070. Those cells that made it to the lower chambers were counted via flow cytometry.

In vivo flow cytometry was used to examine extravasation.⁶  SCID mice were pretreated for 1 hour with vehicle, anti–P-selectin Ab (250 μg/kg IV), or GMI-1070 (25 mg/kg IP). Calcein-labeled MM1s cells (1 × 10⁶) were injected into anesthetized mice and immediately thereafter, a 488-nm laser was focused on an artery in the mouse's ear. Signals were detected through a 680/25-nm bandpass filter, and analyzed with MATLAB Version 7.2 software. Cell counts were obtained every 5 minutes for 40 minutes after the cell injection.

We used in vivo confocal microscopy to test the homing of MM1s cells to the BM in vivo. Calcein-labeled MM1s cells (1 × 10⁶) were injected into anesthetized SCID mice that had been pretreated with vehicle or GMI-1070 (25 mg/kg IP) for 1 hour before the cell injection. For some cases, PSGL-1 in the MM cells was knocked down by siRNA.

In vivo confocal microscopy was performed as described previously.^6,29  Briefly, a skin flap was made in the scalps of the mice to expose the underlying skull surface and Evans blue dye was injected intravenously immediately before imaging to visualize blood vessels. High-resolution images with cellular detail were captured 30 minutes after cell injection through the intact mouse skull at depths of up to 250 μm from the surface of the skull using a 30 × 0.9–numerical aperture water-immersion objective lens (Lomo). Multiple imaging depths were acquired and a maximum intensity z-projection was performed in ImageJ software to merge the images. Calcein-labeled MM cells and Evans blue dye were excited with 491-nm and 638-nm single-photon lasers and detected via 528/19-nm and 680/25-nm bandpass filters, respectively.

MM1s cells (5 × 10⁶) were serum-starved for 3 hours and stimulated for different amounts of time with recombinant P-selectin (10 μg/mL). MM1s cell were treated with recombinant P-selectin (10 μg/mL for 30 minutes) in the presence or absence of GMI-1070 (500μM). The effect of GMI-1070 on signaling induced in MM cells by adhesion to HUVECs was also tested. HUVECs were treated with 0, 250, and 500μM GMI-1070 for 1 hour, followed by washing with PBS. Nontreated MM cells were cocultured with the HUVECs for 1 hour and then harvested by placing the dishes on ice for 10 minutes and dislodging the MM cells from the HUVECs by gentle pipetting. The MM cells were lysed and immunoblotted for the detection of p-FAK, p-Src, cofilin, p-AKT, p-GSK3, and α-tubulin.

MM1s cells were treated for 30 minutes with recombinant P-selectin (10 μg/mL) in the presence or absence of GMI-1070 (500μM); nontreated MM1s cells served as a control. Cells were fixed in formaldehyde for 15 minutes at room temperature, spun onto slides, washed, blocked, stained with 5 μg/mL of anti–β1-integrin Ab at 4°C overnight, followed by FITC-labeled secondary Ab for 1 hour, washed, mounted, and analyzed with a Nikon inverted TE2000 microscope equipped with a 20× Plan-Fluor differential interference contrast 0.5–numerical aperture objective. The percentage of highly labeled cells was calculated for each condition and normalized to the percentage in the nontreated control.

HUVECs or BMSCs (5 × 10³ cells/well) were plated overnight in 96-well plates. MM1s cells (2 × 10⁵ cells/mL) that were transfected with scrambled siRNA or with siRNA for PSGL-1 were cultured for 24 hours with or without BMSCs or HUVECs. In some cases, HUVECs were pretreated with GMI-1070 (0 or 500μM) before the addition of the MM1s to the culture. In some cases, MM1s cells were treated with vehicle, bortezomib (5nM), or dexamethasone (25nM). Proliferation was quantified with the bromodeoxyuridine assay 24 hours after coculture according to the manufacturer's instructions (BD Biosciences).

MM1s cells transfected with scramble siRNA or PSGL-1 siRNA were injected IV into the tail vein (3 × 10⁶ cells in 100 μL of PBS per mouse) or into the BM of the tibia (2 × 10⁵ cells in 20 μL of PBS per mouse) of SCID/bg mice. Mice were killed after 1 week. BM was extracted from 2 femurs and the tibia of mice injected intravenously. RBCs were lysed and mononuclear cells were incubated for 1 hour with mouse anti–human-CD138 Ab conjugated to PerCP-Cy5.5 (5 μg/mL) and analyzed using a flow cytometer.

MM1s cells expressing luciferase were injected intravenously into 40 SCID mice; treatment was started after tumors were first detected by bioluminescence imaging. Mice were divided into 4 groups of 10 each: (1) the control group, which received weekly IP injection of vehicle and were implanted with Alzet-Pump-2002 (flow rate, 0.5 μL/h) loaded with vehicle every 2 weeks for 4 weeks; (2) the GMI-1070–treated group, which received weekly IP injection of vehicle and were implanted with a pump loaded with 200 μL of 150 mg/mL of GMI-1070 that was changed every 2 weeks for 4 weeks; (3) the bortezomib-treated group, which received weekly IP injections of bortezomib 1.5 mg/kg and were implanted with a pump loaded with vehicle that was changed every 2 weeks for 4 weeks; and (4) the combination group, which received weekly IP injection of bortezomib 1.5 mg/kg and were implanted with a pump loaded with 200 μL of 150 mg/mL GMI-1070 that was changed every 2 weeks for 4 weeks. Tumor progression was monitored once a week by bioluminescence imaging while the mice were on treatment (4 weeks) and after the treatment was stopped (4 weeks).

Results were reported as the means ± SD for 3 experiments. Samples were compared by the Student t test, 2-factor ANOVA, or the log-rank test. Results were considered significantly different at P < .05.

The expression of PSGL-1 on MM cells was examined using immunohistochemistry for BM biopsies from 17 MM patients and using flow cytometry for fresh MM samples. In agreement with previous findings, we confirmed that PSGL-1 was highly expressed on MM cells in all 17 MM patients tested (Figure 1A). The expression of PSGL-1 on MM cells was also confirmed by flow cytometry for 7 MM cells lines and, again, high levels of expression of PSGL-1 were detected on all 7 cell lines tested (Figure 1B). PSGL-1 was also differentially expressed on plasma cells isolated from patients at different stages of MM. PSGL-1 gene expression increases with disease progression from MGUS to smoldering MM and then to active disease, as shown by analysis of the published gene expression datasets by Chng et al³⁰  (Figure 1C). These observations indicate that PSGL-1 is highly expressed on all MM cells.

We also examined the expression of selectin on MM cells relative to plasma cells obtained from the BM of healthy donors. Unlike the high expression of the selectin ligand, we found no expression of E-, L-, or P- selectins on any of the MM cells tested (primary samples and cell lines; supplemental Figure 1, available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

When we tested the interaction of recombinant selectins with plasma cells derived from healthy subjects, from MM patients, or from MM cell lines, we observed that that E-selectin did not interact with MM cells and normal plasma cells, whereas L- and P-selectins were highly interactive with MM cell lines and MM primary samples but not with normal plasma cells (Figure 1D). This is in agreement with previous reports showing that PSGL-1 interacts with both L- and P-selectins.³¹  Moreover, the differences in the interaction between MM cells and normal plasma cells were apparently due to lower expression of PSGL-1 in normal plasma cells or potentially because of different glycosylation patterns of the PSGL-1, which is known to alter binding of selectins to the ligand.^32,33 

To confirm the role of PSGL-1 in the interaction of selectins with MM cells, we knocked down PSGL-1 in MM1s cells using siRNA (as verified by flow cytometry; supplemental Figure 2). Knock-down of PSGL-1 decreased the high level of interaction of L- and P-selectins with MM cells (the effect was stronger for interaction of P-selectin on MM cells). E-selectin did not bind with MM cells and the knockdown of PSGL-1 did not alter this effect (Figure 1E). We further confirmed that GMI-1070, a specific selectin inhibitor that is currently being used in clinical trials, inhibits the interaction of L- and P-selectins with MM cells (Figure 1F). Supplemental Figure 3 shows that GMI-1070 did not induce cytotoxicity in MM cells or HUVECs.

We first determined the expression of selectins on ECs in 5 BM aspirates collected from patients with MM and in HUVECs. E- and P-selectins (but not L-selectin) were highly expressed on ECs isolated from the BM of MM patients and on HUVECs (Figure 2A). To investigate the role of each of selectin in the interaction of ECs with MM cells, we knocked down each of the selectins in HUVECs (as confirmed by flow cytometry; supplemental Figure 4). As shown in Figure 2Bi, knockdown of P-selectin, but not of E- or L-selectins, decreased the adhesion of MM cells to HUVECs. Similarly, blocking Abs for P-selectin (but not for E- or L-selectins) inhibited the adhesion of MM cells to HUVECs (Figure 2Bii). Moreover, increasing the expression of P-selectin (by stimulation with IL-4 and histamine) enhanced the adhesion of MM cells to HUVECs, whereas increasing the expression of E-selectin (by stimulation with TNFα) did not increase this adhesion (Figure 2C). These results indicate that P-selectin regulates the adhesion of MM cells to ECs.

We next investigated role of PSGL-1 in adhesion of MM cells to ECs. As shown in Figure 2D, knockdown of PSGL-1 in MM cells decreased the adhesion of MM cells to HUVECs. Similarly, inhibition of the interaction between P-selectin in HUVECs and PSGL-1 in MM cells decreased the adhesion of MM cells to HUVECs in a dose-dependent manner (Figure 2E). These results were reconfirmed using a blocking Ab, and we found that blocking PSGL-1 decreased the adhesion of MM cells to HUVECs (Figure 2F). These findings indicate that interactions between selectins and selectin ligands also regulate the adhesion of MM cells to ECs.

To understand the physiologic process of extravasation of MM cells through ECs, we tested the role of the interaction of PSGL-1 with P-selectin during migration of MM cells through a monolayer of ECs in response to the chemokine SDF1α. Figure 3A shows that blocking P-selectin by siRNA (Figure 3Ai) or blocking Abs (Figure 3Aii) significantly reduced the transendothelial migration of MM cells by 70%, whereas blocking E- or L-selectin had no such effect.

Moreover, down-regulation of PSGL-1 on MM cells by siRNA decreased the transendothelial migration of MM cells through the EC monolayer in vitro (Figure 3B). Similar results were obtained when the interaction between selectin on ECs and PSGL-1 on MM cells was inhibited using the selectin inhibitor GMI-1070 (Figure 3C). To confirm that the effect on transendothelial migration was through the specific interaction between selectins in ECs and PSGL-1 in MM cells, we eliminated the EC component from the experiment and tested the effect of down-regulation of PSGL-1 in MM cells and of GMI-1070 on the trans-well (through the filter) migration of MM cells in response to SDF1 through. Neither down-regulation of PSGL-1 (Figure 3D) nor treatment with GMI-1070 (Figure 3E) had an effect on the migration of MM cells in response to SDF1α. These results were reconfirmed using a blocking Ab, and we showed that blocking PSGL-1 decreased the transendothelial migration of MM cells (Figure 3F). These observations indicate that the interactions of P-selectin in ECs with PSGL-1 in MM cells play a critical role in transendothelial migration, which is the first step of the extravasation process and homing of MM cells into the BM in vivo.

To examine the role of the interaction of PSGL-1/P-selectin on homing of MM cells to the BM in vivo, we used in vivo flow cytometry, which measures the number of circulating cells as the number of cells that pass in an appropriate artery in the mouse ear per minute.²⁹ Figure 4A shows that in control nontreated mice, more than 90% of the MM cells had extravasated within 25 minutes after injection, whereas in mice pretreated with blocking P-selectin Ab or with GMI-1070 (25 mg/kg), MM cells displayed delayed extravasation: more than 50% of the cells were still circulating at 35 minutes after injection.

We also used in vivo live confocal imaging to examine homing of MM cells to the BM, in which we detect the presence of MM cells in the BM niches of the mouse skull. PSGL-1 in MM cells was knocked down or mice were treated with GMI-1070 to interfere with the interaction between ECs and MM cells. Figure 4Bi shows that knockdown of PSGL-1 in MM cells decreased the number of MM cells that had homed to the BM in an average of 18 images taken from 3 different mice. Figure 4Bii shows representative images of the BM in the calvaria of mice, which shows less homing of Calcein-labeled MM cells in cells transfected with PSGL-1 siRNA compared with scramble siRNA. Blood vessels were labeled with Evans blue. Similarly, Figure 4Ci shows that inhibition of selectins in the host environment by GMI-1070 (25 mg/kg) decreased the number of MM cells that homed to the BM. Figure 4Cii shows representative images of homing MM cells in mice treated with GMI-1070 or vehicle: fewer MM cells were present in the BM vascular niches in the GMI-1070–treated mice compared with control mice.

To further understand the nature of the interaction between P-selectin and PSGL-1 in adhesion to ECs, we studied the molecular events and adhesion-related cell signaling in MM cells. Treatment of MM cells with recombinant P-selectin activated cell-adhesion signaling, as evidenced by increased phosphorylation of FAK, Src, cofilin, AKT, and GSK-3α/β. Maximal activation of most proteins was achieved at 30 minutes (Figure 5A). Inhibition of P-selectin by GMI-1070 (500μM for 1 hour) reversed the activation of these kinases induced by recombinant selectin (Figure 5B). Similarly, coculture of HUVECs with MM cells induced activation of the same adhesion-related signaling in MM cells, which was inhibited by GMI-1070 in a dose-dependent manner (Figure 5C). Given that selectin ligands activate adhesion-related kinases, we tested the effect of P-selectin on the activation of integrins in MM cells. Recombinant P-selectin increased the activation and clustering of β1-integrin on the MM-cell surface, which was inhibited by blocking P-selectin with GMI-1070 (Figure 5Di). Representative immunofluorescent images of β1-integrin clustering are shown in Figure 5Dii.

Previous studies have shown that adhesion of MM cells to cells in the BM microenvironment leads to increased proliferation of MM cells.³⁴  We hypothesized that the PSGL-1/P-selectin axis plays a role in the proliferation of MM cells induced by BMSCs and ECs. To examine this possibility, we first determined the expression of selectins on BMSCs derived from the BM of 5 MM patients. Figure 6A shows that, as for ECs derived from BM of MM patients, BMSCs express E- and P-selectins but not L-selectin. Inhibition of the interaction of PSGL-1 in MM and P-selectin with BMSCs and cells by knockdown of PSGL-1 in MM cells (Figure 6Bi) or GMI-1070 (Figure 6Bii) reduced the adhesion of MM cells to BMSCs.

We also studied the role of the PSGL-1/P-selectin axis in MM-cell proliferation induced by BMSCs and ECs. Knockdown of PSGL-1 on MM cells reduced the proliferation of MM induced by BMSCs (Figure 6Ci) and ECs (Figure 6Cii). Similarly, inhibition of P-selectin in BMSCs (Figure 6Di) or ECs using GMI-1070 had similar effects (Figure 6Dii). In addition, we found that the mitochondria of MM1s cells incubated with stroma were less primed for apoptosis compared with those cultured in the absence of stroma. This reduction in priming was reversed by selectin inhibition with GMI-1070 (supplemental Figure 5).

To determine the role of PSGL-1/P-selectin in tumor initiation in vivo, we examined the level of MM cells in the BM of mice 1 week after injection of MM cells either IV or directly into the BM of the tibia as models of tumor initiation. Figure 6E shows that, when injected intravenously, knockdown of PSGL-1 in MM cells decreased the tumor initiation in vivo to 40% compared with scramble siRNA–transfected cells. Figure 6F shows that, when injected directly to the tibia, knockdown of PSGL-1 in MM cells decreased tumor initiation in vivo to 65% compared with scramble siRNA–transfected cells.

We showed previously that the interaction of MM cells with the BM microenvironment induces drug resistance⁶ ; therefore, in the present study, we examined the role of PSGL-1/P-selectin in this effect in vitro and in vivo.

Coculture of MM cells with either BMSCs or ECs induced resistance to bortezomib and dexamethasone, as demonstrated by reduced inhibition of proliferation (compared with MM cells cultured alone). Inhibition of P-selectin by GMI-1070 restored the sensitivity of MM cells to bortezomib and dexamethasone, even when cocultured with BMSCs (Figure 7A) and ECs (Figure 7B).

We next confirmed the effect of GMI-1070 on the sensitivity of MM cells to bortezomib in vivo. We examined the effect of GMI-1070 and bortezomib used either alone or in combination on MM tumor progression in vivo. MM tumors were established in SCID mice and treatment was initiated after tumor detection by bioluminescence imaging at 4 weeks after the MM-cell injection. Unlike the effect on tumor initiation, inhibition of the PSGL-1/P-selectin axis by treatment of mice with GMI-1070 alone did not affect tumor growth compared with the vehicle-treated group (Figure 7C). As expected, treatment with bortezomib alone delayed MM tumor growth; however, tumor progression was detected even when the animals were still being treated with bortezomib. Interestingly, the combination of bortezomib and GMI-1070 inhibited tumor growth completely while animals were on treatment (first 28 days). Furthermore, tumor growth was delayed after stopping therapy in the mice treated with the combination of GMI-1070 and bortezomib (Figure 7C).

Similarly, GMI-1070 treatment did not affect the survival of mice compared with the vehicle-treated group (50% of the animals died at 28 days in both groups). Bortezomib alone increased the survival of the animals: 50% of the mice died at day 38. The combination of bortezomib and GMI-1070 significantly enhanced survival: up to day 42, 100% of the animals in this group were alive, whereas only 40%, 10%, and 0% of the bortezomib group, vehicle group, and the GMI-1070 group were alive at this time point, respectively. Moreover, the animals in the combination group reached 50% survival at day 52 (Figure 7D).

One of the crucial steps in the early phase of tumor dissemination appears to be the interaction of tumor cells with the endothelium for extravasation and homing, which leads to the formation of new metastatic lesions.^9,35,36  Selectins are molecules expressed on the cell surface of ECs that have been shown to promote the first interaction between an extravasating cell and the blood-vessel wall.⁷  Selectins were previously shown to play an important role in seeding of tumor cells in distant organs and in facilitating metastasis.³⁷  Metastasis was dramatically reduced in mice with double deficiency of P- and L-selectin, suggesting a synergistic action of the 2 selectins in this process.⁸ 

In the present study, we examined the role of PSGL-1 and its interaction with P-selectin in tumor progression in MM. We found that PSGL-1 was highly expressed in MM cells and that P-selectin was highly expressed on cells in the BM microenvironment, including BMSCs and ECs. Functionally, in vitro studies showed that the interaction of PSGL-1 with P-selectin was the most critical in regulating adhesion to ECs and in transendothelial migration. These results were confirmed by down-regulation of PSGL-1 in MM cells and by down regulation of selectins in ECs. Similar results were obtained using the pan-selectin inhibitor GMI-1070, which is currently in clinical trials for vasoocclusive crisis in sickle cell disease. These findings were confirmed in vivo in mice: down-regulation of PSGL-1 in MM cells or inhibition of selectin using blocking Abs and GMI-1070 decreased the extravasation and homing of MM cells to the BM. The interaction between P-selectin and PSGl-1 mediated adhesion through activation of adhesion-related cell signaling and activation of integrins in MM cells. Future studies to examine mechanisms of interaction of selectins and integrins in MM are warranted, but were beyond the scope of this study.

We have shown previously that the interaction of MM cells with the BM microenvironment, including BMSCs and ECs, induces MM proliferation and drug resistance.^6,34  Although selectins are thought to regulate the early steps of homing, in the present study, we show that they are also critical regulators of the interaction of MM with the BM microenvironment, which is critical for MM proliferation and drug resistance. We also showed that E- and P-selectins were expressed in ECs as well as BMSCs isolated from the BM of MM patients. Inhibition of the PSGL-1/P-selectin interaction by down-regulation of PSGL-1 in MM cells or by inhibition of selectin with GMI-1070 reversed the proliferative effects and drug resistance induced by BMSCs and ECs in vitro. These findings were confirmed in vivo in mice, in which inhibition of the PSGL-1/P-selectin interaction decreased tumor initiation, sensitized MM cells to bortezomib, and enhanced survival of the animals. Tumor initiation was decreased in vivo because of knockdown of PSGL-1 after both IV injection and direct injection to the BM in the tibia; however, the decrease after IV injection was more than that after tibial injection. This result indicates that the decrease in tumor initiation is not only due to prevention of homing of MM cells, but also to inhibition of proliferation induced by the interaction with the BM microenvironment. These results confirm our hypothesis that disruption of the interaction of MM cells with the BM microenvironment sensitizes MM cells to therapy,⁶  and suggests the interaction between PSGL-1 and P-selectin as a therapeutic target.

In summary, these studies delineated the important role of PSGL-1 in the interaction of MM cells with the BM microenvironment as a regulator of adhesion of MM cells to ECs and BMSCs. In addition, it highlights the interaction between PSGL-1 in MM cells and P-selectin in the microenvironment as a therapeutic target for the prevention of tumor progression and drug resistance in MM.

The authors thank Sonal Jhaveri for editing the manuscript.

This study was supported in part by the Multiple Myeloma Research Foundation and by the National Institutes of Health (grants R01CA125690 and 1R01CA152607).

National Institutes of Health

Contribution: A.K.A. designed and performed the research, analyzed the data, and wrote the manuscript; P.Q., F.A., C.P., B.T., T.C., J.T.P., V.M., L.M.F., and R.C. performed the research and analyzed the data; P.M., A.S., H.T.N., C.P.L., J.L.M., A.L.K., and A.L. analyzed the data; A.M.R. analyzed the data and wrote the manuscript; and I.M.G. designed the research and wrote the manuscript.

Conflict-of-interest disclosure: I.M.G. is on the advisory board for Millennium/Takeda, Genzyme, Celgene, Onyx, and Novartis. A.M.R. is on the advisory board of Roche. J.T.P. and J.L.M. are employees of GlycoMimetics. The remaining authors declare no competing financial interests.

Correspondence: Irene M. Ghobrial, MD, Medical Oncology, Dana-Farber Cancer Institute, 44 Binney St, Mayer 548A, Boston, MA, 02115; e-mail: irene_ghobrial@dfci.harvard.edu.