Recycling of internalized integrins is a crucial step in adhesion remodeling and cell movement. Recently, we determined that the ADP-ribosylation factor–guanine nucleotide exchange factors (ARF-GEFs) cytohesin 2/ARNO and cytohesin 3/GRP1 have opposing effects on adhesion and stimulated β1 integrin recycling even though they are very closely related proteins (80% sequence identity). We have now determined the sequence differences underlying the differential actions of cytohesin 2/ARNO and cytohesin 3/GRP1. We found that the ability of cytohesins to promote β1 integrin recycling and adhesion depends upon the presence or absence of a key glycine residue in their pleckstrin homology (PH) domains. This glycine residue determines the phosphoinositide specificity and affinity of cytohesin PH domains. Switching the number of glycines in the PH domains of cytohesin 2 and cytohesin 3 is sufficient to reverse their effects on adhesion and spreading and to reverse their subcellular locations. Importantly, we also find that a mutant form of cytohesin 3/GRP1 that has three rather than two glycines in its PH domain rescues β1 integrin recycling in cytohesin 2/ARNO knockdown cells. Conversely, a mutant form of cytohesin 2/ARNO with two glycines in its PH domain fails to rescue β1 integrin recycling. Therefore, we conclude that phosphoinositide specificity is the sole functional difference that determines which cytohesin can promote integrin recycling.
Integrin-based cell to ECM adhesion formation, turnover and remodeling are required to convert cytoskeletal contractile forces into productive movement (Caswell and Norman, 2006). During migration of epithelial cells, the balance between E-cadherin based cell–cell adhesion and integrin based cell–ECM adhesion influences both the mode and extent of epithelial movement (Hay, 2005; Perl et al., 1998). Increased integrin based cell–ECM adhesion strength and contractility promotes epithelial migration and scattering (de Rooij et al., 2005; Sander et al., 1998). Conversely, enhanced E-cadherin based adhesion reduces lamellipodia formation, scattering and invasive movements (Borghi et al., 2010; Hordijk et al., 1997). These data suggest that upregulation of integrin exocytosis and cell surface integrin levels can promote epithelial migration.
Integrins can be internalized from the plasma membrane via a number of pathways (Altankov and Grinnell, 1995; Ylänne et al., 1995). Some internalized integrin is immediately recycled via the ‘short-loop’ pathway. The remainder of the internalized integrin recycles through the ‘long-loop’ pathway via the recycling endosome (Caswell and Norman, 2008). This pathway is subject to regulation by serum, growth factors and PKC activity (Gao et al., 2000; Ng et al., 1999; Powelka et al., 2004). These factors can promote both integrin recycling and migration.
The formation of trafficking carriers is initiated by the actions of small GTPases of the ADP-ribosylation factor (ARF) family. ARF6 is the GTPase that regulates the stimulated recycling of β1 integrin (Powelka et al., 2004). Activation of ARFs requires the action of guanine nucleotide exchange factors (GEFs). We have recently shown that the ARF-GEF cytohesin 2/ARNO and its closely related homologue, cytohesin 3/GRP1, have distinct functions in cell adhesion, spreading and migration. Furthermore, only cytohesin 2/ARNO regulates β1 integrin recycling. Finally, these proteins have distinct subcellular locations (Oh and Santy, 2010).
We have now identified the key differences underlying differential actions of cytohesin 2/ARNO and cytohesin 3/GRP1 in β1 integrin recycling. We find that the identity of the pleckstrin homology (PH) domain determines the effect of cytohesins on cell adhesion and spreading. The PH domain is highly conserved in cytohesin 2 and cytohesin 3. However, the most abundant splice variants of these proteins differ significantly in the phosphoinositide binding specificity of their PH domains (Klarlund et al., 2000; Ogasawara et al., 2000). Alternative splicing of a single three-nucleotide exon produces variants of cytohesin 1, 2 and 3 PH domains that contain either two or three glycine residues (Klarlund et al., 2000; Ogasawara et al., 2000). The presence or absence of this single glycine has a profound effect on their phosphoinositide-binding specificity (Klarlund et al., 2000; Ogasawara et al., 2000). The diglycine form has a strong selectivity for phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3 (PIP3)], while the triglycine form has an equal affinity for PtdIns(3,4,5)P3 and PtdIns(4,5)P2 (PIP2) (Klarlund et al., 2000). Cytohesin 2/ARNO primarily exists in the triglycine form and cytohesin 3/GRP1 is predominately in the diglycine form (Ogasawara et al., 2000). We have found that this difference in glycine number is the sole sequence difference that underlies their different effects on adhesion and β1 integrin recycling. These results implicate the phosphoinositide composition of the β1 integrin recycling compartment as a key player in stimulated integrin recycling.
The PH domains of cytohesin 2/ARNO and cytohesin 3/GRP1 are responsible for the differential effects of these proteins on cell adhesion to fibronectin
We constructed hybrid constructs where a single domain of cytohesin 2 was replaced with the equivalent domain of cytohesin 3. The coiled-coil (cc) domain, the polybasic (pb) domain or the PH domain of cytohesin 2 was replaced with the equivalent cytohesin 3 domain (Fig. 1A). Cytohesin 2/ARNO expression enhances cell adhesion to fibronectin, while cytohesin 3/GRP1 expression reduces cell adhesion to fibronectin (Fig. 1B) (Oh and Santy, 2010). If the relevant domain of cytohesin 3 has been swapped into cytohesin 2, then the cells expressing that construct should show impaired rather than enhanced binding to fibronectin. Cytohesin 2 with the cytohesin 3 cc domain (2 w/3cc) or cytohesin 2 with the cytohesin 3 pb domain (2 w/3pb) expression still significantly enhanced cell adhesion suggesting that these domains do not underlie the differences between cytohesin 2 and 3 (Fig. 1A,B). On the other hand, cells expressing cytohesin 2 with the cytohesin 3 PH domain (2 w/3PH) had significantly reduced cell adhesion when compared to the either control cells or to cells expressing wild-type cytohesin 2/ARNO (Fig. 1A,B). Importantly, cells expressing 2 w/3PH adhered to fibronectin similarly to cells expressing wild-type cytohesin 3/GRP1 (Fig. 1B). Therefore, we conclude that the key differences that determine if a cytohesin regulates integrin recycling lie in the PH domain.
The number of glycine residues in the PH domain of cytohesin 2/ARNO (3G) and cytohesin 3/GRP1 (2G) determines their effects on cell adhesion and spreading
Since we found that the identity of the PH domain determines the effect of cytohesin expression on cell adhesion to fibronectin, we tested the hypothesis that the phosphoinositide affinity of the PH domain is the critical difference. We made cytohesin constructs with reversed phosphoinositide binding affinity by altering the number of glycine residues in the PH domains. Specifically, we constructed cytohesin 2 with a diglycine PH (GG cytohesin 2) and cytohesin 3 with a triglycine PH (GGG cytohesin 3), which only differ in glycine numbers when compared to wild-type cytohesin 2/ARNO (3G) and wild-type cytohesin 3/GRP1 (2G). We then tested whether these mutant constructs alter cell adhesion to fibronectin. As shown in Fig. 2A, HeLa cells expressing GGG cytohesin 3 had significantly enhanced cell adhesion to fibronectin to a level similar to the cells expressing wild-type cytohesin 2/ARNO (3G). Conversely, GG cytohesin 2 expression reduced cell adhesion. This reduced adhesion was significantly different from the adhesion of control cells or cells expressing wild-type cytohesin 2/ARNO (3G), but was similar to the adhesion of cells expressing wild-type cytohesin 3/GRP1 (2G; Fig. 2A). We have previously shown that expression of cytohesin 2 increases cell surface β1 integrin levels while expression of cytohesin 3 reduces surface β1 integrin (Oh and Santy, 2010). We measured the levels of surface integrin in HeLa cells expressing wild-type cytohesin 2 or 3 or the cytohesins with altered glycine numbers. Cells expressing cytohesins with three glycines had enhanced levels of cell surface β1 integrin while those expressing cytohesins with two glycines had reduced surface levels of β1 integrin (supplementary material Table S1). Therefore, we concluded that the number of glycine residues of the PH domain is the critical determinant of the effect of cytohesin expression of cell adhesion.
As cells bind more tightly to fibronectin, they also spread more rapidly on the matrix. Consistent with the adhesion data, GG cytohesin 2 expressing HeLa or MCF-7 cells spread less than control cells and similar to cells expressing wild-type cytohesin 3/GRP1 (2G; Fig. 2B–D). Furthermore, GGG cytohesin 3 expressing HeLa or MCF-7 cells spread more than control cells and similar to cells expressing wild-type cytohesin 2/ARNO (3G; Fig. 2B–D). Spreading of MCF-7 cells were more effected by cytohesin expression than the spreading of HeLa cells, suggesting that HeLa cells may have alternative non-cytohesin dependent mechanisms for recycling integrins. We conclude that the different phophoinositide affinity of cytohesin 2 and 3 is responsible for their opposing effects on cell adhesion and spreading.
Differential localization of GG cytohesin 2 and GGG cytohesin 3 in spreading cells
We previously demonstrated that in spreading cells wild-type cytohesin 2/ARNO (3G) is localized at both the cell periphery and interior regions, while wild-type cytohesin 3/GRP1 (2G) is exclusively located at the cell periphery (Oh and Santy, 2010). Therefore, we determined where the GG cytohesin 2 and GGG cytohesin 3 mutants localize in spreading cells. GG cytohesin 2 was exclusively located at the cell periphery region and is restricted to the most basal level of the cell, which is the same localization as wild-type cytohesin 3/GRP1 (Fig. 3A). Conversely, GGG cytohesin 3 was localized at both peripheral and interior regions of spreading cells (Fig. 3B). GGG cytohesin 3 was extensively co-localized with wild-type cytohesin 2/ARNO (3G) and was less well co-localized with cytohesin 3 (Fig. 3B,C,E). Similarly GG cytohesin 2 was better co-localized with cytohesin 3 than with cytohesin 2 (Fig. 3A,D,E). These data all suggest that the number of glycine residues (3Gs vs. 2Gs) in the PH domain is also the important determinant of the differential localization of cytohesin 2 and cytohesin 3.
GGG cytohesin 3 expression rescues impaired β1 integrin recycling caused by knockdown of cytohesin 2/ARNO
Cytohesin 2/ARNO knockdown inhibits β1 integrin recycling (Oh and Santy, 2010). If phosphoinositide affinity is the only functional difference between cytohesin 2 and 3, then GGG cytohesin 3 should rescue β1 integrin recycling in cells where cytohesin 2/ARNO has been knocked-down while GG cytohesin 2 should not rescue β1 integrin recycling. MCF-7 cells were transfected with siRNA targeting cytohesin 2/ARNO or control siRNA. A day later the cells were transfected again with either empty plasmid or plasmid encoding myc–GGG-cytohesin-3 or myc-GG-cytohesin-2. After 24 hours of recovery, the recycling of β1 integrin was assayed. As shown in Fig. 4, the integrin is recycled back to the plasma membrane in 5 minutes in control cells. However, the cells treated with siRNA targeting cytohesin 2/ARNO have significantly impaired β1 integrin recycling (Fig. 4A,B) (Oh and Santy, 2010). Importantly, expression of GGG cytohesin 3 completely rescued β1 integrin recycling in the cells treated with siRNA against wild-type cytohesin 2/ARNO (Fig. 4A,B). Furthermore expression of GG cytohesin 2 failed to rescue β1 integrin recycling in these cells (Fig. 4A,B). This data strongly support the conclusion that the different phosphoinositide specificity of the cytohesin PH domains determines their differential localization and promotes distinct functions of the cytohesins in cell adhesion and migration.
Although cytohesin 2/ARNO and cytohesin 3/GRP1 share 80% sequence identity, cytohesin 2 and cytohesin 3 have distinct functions (Casanova, 2007; Oh and Santy, 2010). Cytohesin 2 and 3 have opposing effects on cell adhesion, spreading and migration (Oh and Santy, 2010). Additionally, only cytohesin 2 is required in β1 integrin recycling (Oh and Santy, 2010). In this study, we demonstrated that the number of glycine residues in the PH domain is the key difference underlying the differential actions of cytohesin 2/ARNO and cytohesin 3/GRP1 in cell adhesion, spreading and β1 integrin recycling. Previous studies have demonstrated that the number of glycine residues in cytohesin PH domains is the key determinant of the phosphoinositide binding affinity of these PH domains (Klarlund et al., 2000). Cytohesin PH domains with 2 glycines bind PtdIns(3,4,5)P3 with high affinity and specificity. The addition of a third glycine residue reduces the affinity for PtdIns(3,4,5)P3 and greatly increases the affinity of the PH domain for PtdIns(4,5)P2 (Klarlund et al., 2000). The structures of PH domains of cytohesin 3/Grp1 (2G) bound to the headgroup of PtdIns(3,4,5)P3 and cytohesin 2/ARNO (3G) bound to the headgroups of PtdIns(4,5)P2 and PtdIns(3,4,5)P3 have been solved and demonstrate the structural basis of the differential binding affinities (Cronin et al., 2004; Lietzke et al., 2000). The addition of a third glycine disrupts interactions of PtdIns(3,4,5)P3 with the peptide backbone thereby reducing the PtdIns(3,4,5)P3 binding affinity (Cronin et al., 2004). Furthermore the addition of this glycine causes PtdIns(4,5)P2 to bind in a different orientation into a high affinity pocket (Cronin et al., 2004). Our data support the conclusion that phosphoinositide specificity is the key functional difference determining which cytohesin functions in cell adhesion, spreading and integrin recycling.
We also found that localization of cytohesin 2 and cytohesin 3 depends on number of glycines in the PH domain. This observation is somewhat surprising since these proteins are recruited to membranes both by their PH domains and by binding to scaffold proteins via their coiled-coil domains, which is the least conserved region (Donaldson and Jackson, 2000; Lim et al., 2010; White et al., 2010). However, we found that a change in the number of glycine residues in the PH domain is sufficient to alter localization of cytohesin 2 and cytohesin 3. Protein–protein interactions mediated by the coiled-coil domain may therefore serve to stabilize cytohesins at membranes once they have been recruited there by PH domain–phosphoinositide interactions. Alternatively, we have shown that coiled-coiled domain mediated interactions are required for assembly of protein complexes that have signal transduction roles (White et al., 2010). The coiled-coil domain may therefore promote cytohesin-dependent signal transduction rather than cytohesin localization.
Our data implicate binding to PtdIns(4,5)P2 as the key function that allows a cytohesin to promote stimulated β1 integrin recycling. Other studies have reported the presence of PtdIns(4,5)P2 in early endosomal compartments but not in the later recycling compartments (Brown et al., 2001; Donaldson et al., 2009; Jovanovic et al., 2006). Therefore, it is not clear how PtdIns(4,5)P2 would be acting to recruit cytohesin 2 and induce integrin recycling. It is worth noting that our studies have investigated stimulated, not basal, integrin recycling. PtdIns(4,5)P2 might be produced transiently at recycling compartments in response to serum or other signals, but play no role in unstimulated recycling of proteins to the plasma membrane. Alternatively cytohesin 2 could be recruited onto early endosomal membranes by binding to PtdIns(4,5)P2 and then be maintained on these membranes as they traffic through the endosomal system by interactions with other proteins. In this scenario cytohesin 2 would be present on the endosomal membranes in an inactive state until it is turned on by additional signals. One final possibility is that as ARF6 is activated during recycling, it can activate phosphatidylinositol-4-phosphate 5 kinase to produce PtdIns(4,5)P2 (Brown et al., 2001; Donaldson et al., 2009; Grant and Donaldson, 2009; Jovanovic et al., 2006). Thus, PtdIns(4,5)P2 and cytohesin 2 may set up a positive feedback loop that promotes the recycling of endocytic cargo. Clearly further work will be necessary to determine where PtdIns(4,5)P2 and cytohesin 2 intersect with internalized β1 integrin.
The inhibitory action of cytohesin 3/GRP1
The increased adhesion and spreading in cells expressing cytohesin 2 is explained by the ability of this protein to promote β1 integrin recycling and thereby elevate surface integrin levels. Cytohesin 3, on the other hand, inhibits adhesion and migration. In contrast to cytohesin 2/ARNO, cytohesin 3/GRP1 is not involved in β1 integrin recycling due to its exclusive localization at the plasma membrane (Oh and Santy, 2010). However, cytohesin 2 and cytohesin 3 share many up- and downstream signaling molecules. Therefore, cytohesin 3/GRP1 expression may reduce cell adhesion and spreading by sequestering these common up- or downstream components at the plasma membrane, thereby perturbing formation of a functional complex at the recycling endosome. Scaffolding proteins such as GRASP, Cybr, CNK1 and CNK3/IPCEF regulate ARF6 activity by binding to cytohesins (Attar et al., 2012; Lim et al., 2010; White et al., 2010). A number of molecules including Rac, phosphatidylinositol-4-phosphate 5 kinase and PLD act downstream of ARF6 in signaling pathways (Logan and Mandato, 2006). Cytohesin 3/GRP1 expression in cells may affect all of these signaling pathways. Further investigation of the inhibitory action of cytohesin 3/GRP1 on β1 integrin recycling is required to determine which of these pathways is impaired by cytohesin 3 overexpression. Identifying these pathways will identify additional pathways that are limiting for β1 integrin recycling.
Cytohesin 2 localizes at both cell edges and interior regions, while cytohesin 3 exclusively locates at the plasma membrane (Oh and Santy, 2010). Therefore, we hypothesize that only cytohesin 2 functions at internal compartments, while both cytohesin 2 and cytohesin 3 should be involved with events at the plasma membrane (Oh and Santy, 2010). Consistent with this idea, recent studies have shown that both cytohesin 2 and cytohesin 3 are involved in insulin signal transduction. Knockdown of either cytohesin 2 or cytohesin 3 produces insulin resistance in mice (Hafner et al., 2006). Additionally, cytohesin function is required to generate PtdIns(4,5)P2-rich environment at the plasma membrane for recruitment of insulin receptor substrate 1 (IRS1) (Lim et al., 2010). Although both cytohesins should act at the plasma membrane, cytohesin 3 might still have unique functions in PtdIns(3,4,5)P3 dependent events due to its high affinity for this lipid. Even though both insulin and PDGF increase PtdIns(3,4,5)P3 levels at the plasma membrane, a rise is less and slower in response to PDGF (Oatey et al., 1999). Therefore, total cytohesin activity at the plasma membrane, which will be a sum of cytohesin 2 binding to either lipid and cytohesin 3 binding to PtdIns(3,4,5)P3, may be different in response to these different signals. Alternatively, cytohesin 3 recruitment to the plasma membrane in response to PtdIns(3,4,5)P3 may act as a coincidence detector to increase cytohesin activity when PtdIns(3,4,5)P3 production occurs in conjunction with other signals.
In this study, we demonstrated that the presence of a single additional glycine in the PH domain of cytohesin 2/ARNO differentiates its functions and localization from the functions and localization of cytohesin 3/GRP1. This result implicates the phophoinositide binding specificity of cytohesin 2/ARNO and cytohesin 3/GRP1 as a key determinant of their localization and functions. Furthermore these results implicate phosphoinositide production as a key signal in the stimulated recycling of β1 integrin, and therefore as a key player in the initiation of cell migration.
Materials and Methods
Cells and reagents
HeLa cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with antibiotics (Penicillin, streptomycin, Fungizone), 10% FBS and glutamine. MCF-7 cells were cultured in DMEM/F-12 with antibiotics (Penicillin, streptomycin, Fungizone), 10% FBS and nonessential amino acids. The TS2/16 anti-β1 integrin antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). M2 anti-flag antibodies (labeled and unlabeled) were purchased from Sigma-Aldrich (St. Louis, MO, USA). 9E10 anti-myc antibodies (labeled and unlabeled), anti-actin and anti-GapDH antibodies were purchased from Covance (Princeton, NJ, USA). Alexa-Fluor-647–TS2/16 was obtained from BioLegend (San Diego, CA, USA).
siRNAs and expression constructs
siRNA duplexes against the sequence 5′-GCAAUGGGCAGGAAGAAGU-3′ targeting human cytohesin 2/ARNO and control siRNA against firefly luciferase were purchased from Dharmacon (Layfayette, CO, USA). The siRNAs were transfected into MCF-7 cells by using Neon Transfection System (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's recommended protocol. Plasmids encoding myc-tagged cytohesin 2/ARNO or Myc-tagged cytohesin 3/GRP1 in pcDNA3 were constructed by PCR from myc-tagged cytohesin 2/ARNO or FLAG-tagged cytohesin 3/GRP1 in pCB7 provided by James Casanova (University of Virginia). Mutations were introduced by using QuikChange site-directed mutagenesis (Stratagene, Wilmington, DE, USA). Hybrid constructs were constructed by creating unique restriction sites in equivalent locations of cytohesin 2 and 3. The construct for expressing siRNA-resistant GG cytohesin 2 was created by introducing two silent point mutations into the siRNA site using site-directed mutagenesis. Transient transfections were performed according to the manufacturer's instructions using Lipofectamine LTX (Forward transfection (Invitrogen, Carlsbad, CA, USA) for MCF-7 cells and Lipofectamine 2000 (Forward transfection) for HeLa.
HeLa cells were transfected with various DNA constructs as described above. After 24 hours of expression, adhesion of the cells to fibronectin was tested as previously described (Oh and Santy, 2010). Briefly, the cells were harvested by incubation in PBS, 4 mM EDTA, 1 mM EGTA. The cells were resuspended in Optimem medium with 0.2% BSA. Equal numbers of cells were plated in wells coated with varying concentrations of fibronectin. After adhesion for 1 hour the cells were washed, fixed and stained with crystal violet.
Spreading of cells transfected with cytohesin constructs was performed as previously described (Oh and Santy, 2010). Briefly cells were transfected and harvested as described for the adhesion assay. They were plated on coverslips coated with 10 µg/ml of fibronectin and allowed to spread for 15 minutes (HeLa cells) or 30 minutes (MCF-7 cells). The cells were then fixed and stained with 9E10 anti-myc followed by Alexa-Fluor-488-conjugated anti-mouse secondary antibody and Rhodamine-conjugated phalloidin.
Co-localization of cytohesins
MCF-7 cells were transfected with the indicated cytohesin constructs and the next day allowed to spread on fibronectin as described above for the spreading assay. The cells were fixed with 4% paraformaldehyde and then stained with FITC-conjugated M2 anti-FLAG and biotinylated 9E10 anti-myc followed by Alexa-Flour-546-conjugated streptavidin or with Cy3-conjugated M2 anti-FLAG and Alexa-Fluor-488-conjugated 9E10 anti-myc antibodies. Co-localization of the cytohesins was assayed using deconvolution microscopy as previously described (Oh and Santy, 2010). The extent of co-localization was quantified by calculating the Pearson's correlation coefficient in multiple cells using Slidebook 5.0.
Analyzing rescue of β1 integrin recycling by microscopy
Two days before the assay, siRNA targeting cytohesin 2/ARNO or control siRNA was transfected into MCF-7 cells. The next day empty plasmid, plasmid encoding myc–GGG-cytohesin-3 or plasmid encoding siRNA-resistant myc–GG-cytohesin-2 were transfected into these cells using Lipofectamine LTX. After 8 hours of recovery, the cells were serum starved overnight. The next day, β1 integrin recycling was assayed as previously described (Oh and Santy, 2010). Briefly, TS2/16 anti-β1 integrin was bound to the surface of the cells in the cold for 2 hours. The cells were allowed to internalized the bound antibody for 1 hour at 37°C, and then remaining surface antibody was removed with by washing the cells twice with 0.5 M NaCl, 0.5% acetic acid, pH 3.0. Recycling of internalized integrin was initiated by the addition of warm DMEM with 20% FBS. At designated times the cells were chilled and fixed with 4% paraformaldehyde. The cells were left unpermeabilized and antibody that had been recycled to the cell surface was detected by incubation with Alexa-Fluor-594-conjugated anti-mouse secondary antibody.
MCF-7 cells were transfected with siRNAs and plasmid as described above for the rescue assay. The day after plasmid transfection total RNA was isolated using the RNeasy kit according to the manufacturer's instructions (Qiagen, Inc., Valencia, CA). RT-PCR to amplify cytohesin 2 and GapDH was performed as previously described (Oh and Santy, 2010), using 0.2 µg of the RNA as template.
Detection of surface β1 integrin levels
HeLa cells were transfected with the indicated cytohesin constructs and a GFP vector as described above. After 24 hours of expression, the cells were harvested as described above for the adhesion assay, fixed and stained with Alexa-Fluor-647-conjugated TS2/16 anti-β1 integrin. Levels of GFP and Alexa Fluor 647 fluorescence were measured using an FC500 benchtop flow cytometer. Alexa Fluor 647 levels in GFP positive cells were analyzed using the FlowJo (Tree Star, Inc., Ashland, OR, USA) PB histogram comparison algorithm.
This work was supported by grants from the American Heart Association [grant number SDG 0730229N to L.C.S.]; and the American Cancer Society [grant number RSG-09-168-01-CSM to L.C.S.].