|Home | About | Journals | Submit | Contact Us | Français|
Understanding the mechanisms by which stem cells home precisely to regions of injury or degeneration is of importance to both basic and applied regenerative medicine. Optimizing regenerative processes may depend on identifying the range of molecules that subserve stem cell trafficking. The “rolling” of extravasating cells on endothelium under conditions of physiological flow is the first essential step in the homing cascade and determines cell adhesion and transmigration. Using a laminar flow chamber to simulate physiological shear stress, we explored an aspect of this process by using human neural stem cells (hNSCs). We observed that the interactions between hNSCs and TNF-α-stimulated human endothelium (simulating an inflamed milieu) are mediated by a subclass of integrins -- α2, α6 and β1, but not α4 and αv or the chemokine-mediated pathway CXCR4-SDF-1α, suggesting not only that the mechanisms mediating hNSC homing via the vasculature differ from the mechanisms mediating homing through parenchyma, but also that each step invokes a distinct pathway mediating a specialized function in the hNSC homing cascade. (TNF-α stimulation also up-regulates VCAM-1 expression on the hNSCs themselves and increases NSC-endothelial interactions.) The selective use of integrin subgroups to mediate homing of cells of neuroectodermal origin may also be used to insure that cells within the systemic circulation are delivered to the pathological region of a given organ to the exclusion of other, perhaps undesired, organs.
Understanding the mechanisms by which “solid organ” somatic stem cells home precisely to regions of injury or degeneration (a “niche”) is of importance to both basic and applied regenerative medicine. Repair by stem/progenitor cells – whether endogenous or transplanted – may be limited in part by an inability to ensure a sufficient number of reconstituting cells in the damaged area at the opportune time. Optimizing regenerative processes, therefore, may depend on first identifying the range of molecules that subserve trafficking. We explored an aspect of this process by using human neural stem cells (hNSCs) 1.
The migration of hNSCs is a tightly regulated process. Previously, we and others have shown that hNSCs rely on inflammatory chemokines for homing through parenchyma to focal areas of injury (emulated by stroke) 2. Although controversial, there is also a suggestion that stem cells from outside the central nervous system (CNS), – either endogenous hNSCs resident in non-neural zones (e.g., bone marrow) 3, 4 or administered exogenously 5, 6 – may constitutively home to intracranial pathology. Under these situations, it is unclear what mechanism might mediate such selective homing given the characteristics of cerebrovascular endothelium which, as a component of the blood-brain barrier (BBB), serves to separate CNS parenchyma from the vascular compartment, playing a crucial role in regulating the trafficking of cells. Such a process consists of two major phases: (1) extravasation and (2) seeding of the niche. Extravasation involves the interaction of the NSC with the vascular endothelium under conditions of physiological flow and includes (a) the “rolling” of cells, (b) adhesion to the luminal surface of endothelial cells, and (c) transmigration across the endothelium. Thus, proper adhesive interactions between endothelial cells and the NSC are required for the successful extravasation of hNSCs. For intraparenchymal trafficking, as well, interactions between hNSCs and endothelium play a role, such as when CNS vasculature is ruptured following trauma or when less stable vessels form during the neoangiogenesis accompanying tumor formation and progression.
We report here evidence for an aspect of NSC trafficking -- the critical initial step of the homing cascade, i.e., “rolling” -- that is regulated by their first low-affinity interaction with vascular endothelial cells, and is independent of chemokine-dependent homing mechanisms. There is growing evidence for the importance of wall shear stress in the regulation of endothelial cell function 7, 8. Shear forces induce rapid activation of signaling cascades, transcription factors, and differential gene expression in endothelial cells 9, 10. Therefore, we used a parallel laminar flow chamber to simulate physiological shear stress conditions in order to investigate the mechanisms underlying cellular interactions between hNSCs and TNF-α-stimulated endothelium, a condition that models many intracranial pathological processes such as traumatic injury, inflammation, and degeneration. We found that integrins α2, α6 and β1, but not integrins α4 and αv, mediate “rolling” of hNSC on TNF-α stimulated human vascular endothelial cells (HUVECs) in a CXCR4-SDF-1α-independent manner. Of interest as well, direct TNF-α stimulation of the hNSCs themselves results in up-regulation of VCAM-1 expression on hNSCs and also increases hNSC-endothelial interaction.
A stable line of hNSCs was isolated from the ventricular zone of a late first trimester (13 week) human fetal cadaver as previously described 1, 2, 11 and expanded with mitogens without genetic manipulation or augmentation. They were maintained in Neurobasal Medium containing B27 Supplement, Glutamax (1%; Gibco-Invitrogen, Carlsbad, CA), fibroblast growth factor-2 (FGF-2; 20ng/ml; Calbiochem, San Diego, CA), heparin (8µg/ml; Sigma, St Louis, MO), and leukemia inhibitory factor (LIF; 10ng/ml; Chemicon, Temecula). The cells were fed twice per week. Both adherent and floating clusters were passaged. When clusters reached 10 cell diameters, they were dissociated with accutase and then split with a seeding density of 50,000 cells/cm2. For FACS analysis and laminar parallel flow chamber experiments, a single cell suspension was prepared by incubating the cells in phosphate buffered saline (PBS)-based enzyme-free cell dissociation buffer (Gibco-Invitrogen, Carlsbad, CA). Where indicated, hNSCs were pre-incubated with SDF-1α (100ng/ml, Upstate biotechnology, Lake Placid, NY) or manganese (Mn, 2mM, Sigma, St Louis, MO) for 5 min at 37°C.
The rolling and adhesion of hNSCs was assessed in vitro using a parallel plate laminar flow chamber as previously described12 (Fig. 1A). Briefly, 22 mm2 glass cover slips were coated with L-Poly-Lysine (10µg/ml, Sigma) overnight at 4°C and washed twice with PBS. Human umbilical vein endothelial cells (HUVECs; Clonetics, Walkersville, MD) were grown on these glass cover slips until 100% confluent in cell-type specific media (Clonetics, Walkersville, MD). Defined levels of flow (wall shear stress) were applied to the cover slip in the flow chamber (100 µm thickness) by perfusing warm media (RPMI containing 0.75mM Ca2+ and Mg2+ and 0.2% HSA) through a constant infusion syringe pump (Harvard Apparatus, Holliston, MA). The flow chamber was then perfused with a cellular suspension of hNSCs (10 ml at 1×105 cells/ml) at various shear stresses (2×105 per each shear stress). At least three slides with human endothelial cells were run in each experimental group. The interactions of the injected cells with the endothelial layer were observed on random fields in the central sector of each slide using an inverted phase contrast microscope, and the images were video-recorded. “Rolling” hNSCs flowed slowly and demonstrated multiple discrete flow interruptions, while “adherent” cells remained stationary at a given point for extended periods of time (>30 sec). (See Supplementary Material for a more detailed description of the methodology employed). The scoring was performed by trained but blinded researchers whose assessments, when uncoded, were in concordance in >98% of cases. Where indicated, hNSCs or the monolayers of human endothelial cells were pre-incubated with TNF-α (10ng/ml, 4 hours, 37°C); and hNSCs were pre-incubated with function-blocking antibodies directed against integrins β1, α1, 2, 3, 4, 5, 6, ν, CXCR4 and VCAM-1 (25µg/ml, 45 min, room temperature, Chemicon, Temecula CA). SDF-1α (100ng/ml, Upstate biotechnology, Lake Placid, NY), RANTES, MCP-1 (100ng/ml Chemicon, Temecula, CA) and MIG (Sigma, St. Luis, MO) were added to the suspension of hNSCs 5min prior to experiment. All results are expressed as the number of “rolling” cells/field representing the average from 3 slides (4 fields/slide).
To detect the expression of CXCR4, and integrins β1, α1, 2, 3, 4, 5, 6, and ν (Chemicon, Temecula CA), hNSCs were incubated with the specific antibody (10µg/ml) for 30 min at 4°C and then washed with FACS buffer (2% FCS, 0.1% BSA, 0.01% NaN3 in PBS). Control cells were incubated with isotype-matched control IgG (Strategic Biosolutions, DE). Thereafter, the cells were incubated with a FITC-conjugated secondary antibody (Biosource International, Camarillo, CA). VCAM-1 expression was detected by VCAM-1 specific antibody (Pharmingen, San Diego, CA) on control or TNF-α (10ng/ml; 4 hours) pre-treated HUVECs or hNSCs. Fluorescence analysis was performed on a FACScan (Becton Dickinson, San Diego, CA) according to standard procedures.
Triplicate cultures of hNSCs were cultured as described above, harvested and total RNA was isolated using a Qiagen RNA isolation kit. Probe preparation and chip hybridization was performed according to the manufacturer’s recommendations (Illumina, San Diego, CA). Full microarray analysis is provided in the Supplementary On-line Material and can be downloaded as an Excel spreadsheet. Members of the RGS (Regulator of G-protein signaling) gene family were defined as detectable if their hybridization signal intensities in all three samples were detected with at least 99% confidence.
Expression of the mRNA for RGS3 and RGS16 was detected by RT-PCR. RNA was isolated from hNSCs using a kit from Qiagen Corp. (Valencia, CA). A Qiagen kit was also used to make cDNA initiated by oligo-dT. A segment of the cDNA encoding regions stretching from amino acid 360 to 462 was detected using RT-PCR for RGS 3 by employing the following primers: forward primer 5’-AGA CGG CGG AAT GAG TCC CCT GG-3’ and reverse primer 5’-GA GTC CAG GTT GAC CTC CTT GCA T-3’. For RGS16, the following primers were used: forward primer 5’-AAG ATC CGA TCA GCT ACC AAG C-3’ and reverse primer 5’-GGG CTC GTC CAG GCT GCA GCT-3’. The derived cDNA fragments were hard copied into pCR II (Invitrogen Corporation, Carlsbad, CA) and sequenced using automated DNA sequencing employing a capillary ABI 3730 sequencer.
A single cell suspension of hNSCs was seeded on fibronectin-coated glass-slides (10mg/ml, 1 hour at 37°C), fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton-X100 and used for the intracellular detection of RGS3 and RGS16 proteins. Rabbit anti-human RGS3-specific antibody (1:500, Novus Biologicals, Littleton, CO) and chicken anti-human RGS16-specific antibody (1:500, Chemicon, Temecula, CA) staining was visualized with goat anti-rabbit Alexa-Flour 594-conjugated (Invitrogen, Carlsbad, CA) and goat anti-chicken Alexsa-Fluor 488-conjugated (Aves Labs, Tigard, OR) secondary antibodies (1:1000). Images were captured with an Olympus FLUOVIEW FV 1000 (× 60 objective) and analyzed using Open Lab software.
hNSCs were isolated and expanded (with mitogens alone) from the primary CNS germinal zone (ventricular zone) of a fetal forebrain as previously described1, 2, and maintained as a stable line to ensure reproducible conditions across experiments. These hNSCs are known to home in vivo to intracranial pathologies such as stroke lesions and tumors 1, 2, 11, as well as to participate in normal cerebrogenesis (including primate) 2, 13. For these studies, we used a parallel laminar flow chamber assay to investigate the mechanisms mediating the “homing cascade”-initiating adhesive interactions (called “rolling”) between hNSCs and inflamed human endothelium (Fig. 1A). (See also Supplementary Material for a more detailed description of the methodology employed). hNSCs were allowed to interact with untreated or TNF-α-exposed human endothelial cell monolayers by infusing them into the laminar flow chamber. TNF-α stimulation of human endothelium emulates their condition in a range of intracranial pathological conditions, particularly those accompanied by an inflammatory signature. The number of hNSCs “rolling” on TNF-α-exposed-human endothelium was significantly higher compared to control (Fig. 1B). As noted above, “rolling” is the initial step in somatic cell engagement with endothelium. The term refers to those cells that are impeded from flowing briskly through the chamber by instead making intermittent low-affinity contacts with the endothelium along its length, creating the appearance of “rolling” along its surface (Fig. 1A, schematic). In the blood stream, circulating cells undergo high shear stress; rolling is the mechanism that selectively slows down subsets of cells (out of larger population of all fast-flowing cells) in order to permit their ultimate firm adhesion and transmigration. This process is essential preparation for cell adhesion; blocking rolling, blocks adhesion and ultimately homing 14. The molecular mechanisms that mediate rolling – a process that adds an essential dimension of selectivity – are distinct from these leading to adhesion 15, 16 (relative proportions for which are presented in Supplementary Fig. 1). Those involved in NSC rolling have yet to be investigated.
Since TNF-α up-regulates the expression of VCAM-1 on human endothelium 17–19 and because hNSCs express the chemokine receptor CXCR4 2 (Supplementary Fig. 2), we anticipated that the interaction between hNSCs and TNF-α-exposed-endothelium would be facilitated by SDF-1α (the cognate ligand for CXCR4). Interestingly, the number of rolling hNSCs was unchanged in the presence of SDF-1α (Fig. 1B). Furthermore, the number of hNSCs which rolled on TNF-α-treated human endothelium was not significantly inhibited by CXCR-4-specific antibodies, as compared to control isotype-matched IgG (Fig. 1B). In accordance with this observation, FACS analysis showed that activation of integrin β1, which is normally stimulated by SDF-1α stimulation 20, was not influenced by SDF-1α in these cells (Fig. 1C). Therefore, the SDF-1/CXCR4 pathway did not appear to predominate in these cells under these conditions. The function of CXCR4 can be negatively regulated by a family of proteins named “regulator of G-protein signaling” (RGS). Previously, CXCR4 expression without a response to SDF-1α has been observed in lymphocytes that express RGS3 and RGS16 21–24. We asked whether a similar mechanism might be operative in the nervous system as well, i.e., whether these molecules are expressed by hNSCs. Gene expression profiling of hNSCs demonstrated the expression of 12 members of the RGS family (RGS 3, 4, 5, 6, 7, 11, 12, 13, 14, 16, 17 and 20). RGS3 and RGS16 are of particular interest because they have been shown to inhibit CXCR4 signaling. Their expression at the mRNA and protein levels was confirmed in the hNSC (Fig. 1D), suggesting that RGS expression might determine whether the CXCR-4/SDF-1 pathway functions in these cells.
Since it has been shown that NSCs isolated from the adult mouse brain express functionally active CCR1, CCR2, CCR5 and CXCR3 25, and because chemokine-mediated activation of integrins has been demonstrated for mouse lymphocytes 26, we investigated the effect of MCP-1, MIG and RANTES on the rolling of hNSC on human endothelium. We found that, as with SDF-1α, MCP-1, MIG and RANTES do not affect the adhesive interactions of hNSCs with human endothelium (Supplementary Fig. 3). This finding is consistent with our gene microarray data which demonstrated low or no expression of CCR1, CCR2, CCR5 and CXCR3 in these hNSCs (Full microarray gene profile provided in On-Line Supplementary Material).
Given that chemokine-mediated pathways appeared not to be operative in the regulation of hNSC rolling on inflamed human endothelium, we sought to determine which molecules might be pivotal. FACS analysis demonstrated that, in addition to β1, hNSCs express high levels of integrin α2, α6 and αν on their surface. No or low expression of integrin α1, α3, α4 and α5 were detected (Fig. 2A). To address the question of whether these integrins mediate low-affinity adhesive interactions of hNSC with TNF-α-stimulated human endothelium under conditions of physiological shear stress, function-blocking antibodies directed against specific integrins were tested in the parallel laminar flow chamber assay. Antibodies directed against α2, α6 and β1 inhibited rolling of hNSCs on TNF-α-stimulated human endothelial cells (Fig. 2B) whereas those against α4 and αv had no significant effect (Fig. 2C). Taken together, these data suggest that integrins α2, α6 and β1 (but not α4 and αν) mediate the initial interaction -- i.e. rolling -- between hNSCs and injured human endothelium.
Up to this point, our study entailed examining the effect of an inflamed (i.e., TNF-α−stimulated) human endothelium on hNSCs. Because, in an actual inflammatory niche in vivo, the hNSCs are also exposed to inflammation, we next examined whether TNF-α stimulation of the hNSCs themselves might impair their ability to interact with the vasculature. Therefore, we first determined whether TNF-α changed the expression of α4 integrin and VCAM-1 on hNSCs. While α4 integrin expression remained unchanged (i.e., low), TNF-α actually increased the expression of VCAM-1 (Fig. 3A). Further, TNF-α treated hNSCs evinced increased rolling under shear stress conditions, a finding that could be blocked by a VCAM-1-specific antibody but not by an α 4–specific antibody (Fig. 3B).
Together, these data suggest that the migration of hNSC is a tightly regulated process controlled by multiple mechanisms mediated not only by soluble factors such as chemokines and components of the extracellular matrix, but also by a subset of adhesion molecules. The fact that these subsets may vary from the vascular bed of one organ compared to that of another may have not only physiological importance but also therapeutic relevance. For example, it has been suggested that stem cells, including hNSCs, can be used to address pathology via intravascular injection. However, there is presently no strategy for specifically directing cells, once in the systemic circulation, to a targeted organ to the exclusion of an undesired organ. A knowledge of the unique mechanisms that mediate adherence of hNSCs to the endothelium of one organ vs. another – particularly if based on different adhesion molecule systems – may allow a more precise targeting of cells. Here we establish not only that integrins play this role (as opposed to chemokine-receptor interactions) but that a particular subset of integrins (to the exclusion of others) may be pivotal for refining such homing.
This work was supported by National Institutes of Health (NIH), grants R21DK067084 and K18 HL081096 (SKK), Studienstiftung des Deutschen Volkes (FJM), TRDRP Postdoctoral Fellowship 14FT-0126 (NS) and Franklin Delano Roosevelt Scholars Award from the March of Dimes (EYS). The authors thank Sirak Simavoryan for his help with flow chamber experiments and Dustin Wakeman for his help with hNSC culture.