SDF-1 fused to a fractalkine stalk and a GPI anchor enables functional neovascularization


  • Georg Stachel,

    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    Search for more papers by this author
  • Teresa Trenkwalder,

    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    Search for more papers by this author
  • Franziska Götz,

    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    Search for more papers by this author
  • Chiraz El Aouni,

    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    Search for more papers by this author
  • Niklas Muenchmeier,

    1. Medizinische Poliklinik Innenstadt and Ludwig-Maximilians-University of Munich, Germany
    Search for more papers by this author
  • Achim Pfosser,

    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    Search for more papers by this author
  • Claudia Nussbaum,

    1. Walter-Brendel Centre of Experimental Medicine, Ludwig-Maximilians-University of Munich, Germany
    Search for more papers by this author
  • Markus Sperandio,

    1. Walter-Brendel Centre of Experimental Medicine, Ludwig-Maximilians-University of Munich, Germany
    Search for more papers by this author
  • Antonis K. Hatzopoulos,

    1. Department of Medicine, Division of Cardiovascular Medicine, Vanderbilt University, Nashville, Tennessee, USA
    2. Department of Cell and Developmental Biology, Vanderbilt University, Nashville, Tennessee, USA
    Search for more papers by this author
  • Rabea Hinkel,

    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    Search for more papers by this author
  • Peter J. Nelson,

    1. Medizinische Poliklinik Innenstadt and Ludwig-Maximilians-University of Munich, Germany
    Search for more papers by this author
  • Christian Kupatt

    Corresponding author
    1. Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Ludwig-Maximilians-University and DZHK (German Centre for Cardiovascular Research), Partner Site Munich Heart Alliance, Munich, Germany
    • Correspondence: Christian Kupatt, M.D., Medizinische Klinik und Poliklinik I, Klinikum Großhadern, Marchioninistr. 15, 81377 Munich, Germany. Telephone: +49-89-7095-6092; Fax: +49-89-7095-8572; e-mail:

    Search for more papers by this author

  • Author contributions: G.S. and T.T.: collection and assembly of data, data analysis and interpretation, and manuscript writing; F.G., N.M., and C.N.: collection and assembly of data and data analysis and interpretation; C.E.A.: conception and design and data analysis and interpretation; A.P.: administrative support and collection and assembly of data; M.S.: administrative support, data analysis and interpretation, and final approval of manuscript; A.K.H.: administrative support, provision of study material, and final approval of manuscript; R.H.: final approval of manuscript, manuscript writing, collection and assembly of data, data analysis and interpretation, and administrative support; P.J.N.: administrative support, provision of study material, conception and design, and final approval of manuscript; C.K.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript. G.S., T.T., P.J.N., and C.K. contributed equally to this article.


The facilitated recruitment of vascular progenitor cells (VPCs) to ischemic areas might be a therapeutic target for neovascularization and repair. However, efficient and directed attraction of VPCs remains a major challenge in clinical application. To enhance VPC homing, we developed a fusion protein (S1FG), based on the biology of stroma-derived factor-1/CXCL12 and the mucin backbone taken from fractalkine/CXCL12. A GPI-anchor was included to link the fusion-protein to the cell surface. HUVECs transfected with S1FG were capable of increasing firm adhesion of CXCR4+-mononuclear cells (THP-1) under shear stress conditions in vitro. In an in vivo rabbit model of chronic hind limb ischemia, local S1FG application enhanced the recruitment of adoptively transferred embryonic EPCs (eEPCs) to the ischemic muscles 2.5-fold. S1FG combined with eEPClow (2 × 106) yielded similar capillary growth as eEPChigh (5 × 106) alone. Compared to controls, collateral formation was increased in the S1FG eEPClow group, but not the eEPChigh group without S1FG, whereas perfusion was found enhanced in both groups. In addition, S1FG also increased collateral formation and flow when combined with AMD3100 treatment, to increase circulating levels of endogenous VPC. These data demonstrate that the fusion protein S1FG is capable of enhancing the recruitment of exogenously applied or endogenously mobilized progenitor cells to sites of injury. Recombinant versions of S1FG applied via catheters in combination with progenitor cell mobilization may be useful in the treatment of chronic ischemic syndromes requiring improved perfusion. Stem Cells 2013;31:1795-1805


Therapeutic neovascularization by circulating proangiogenic cells has been suggested as a potential therapy in peripheral artery disease patients lacking surgical or interventional treatment options. Currently, several populations of endothelial progenitor cells (for review cf. [1]) and mesenchymal stem cells [2, 3] are under clinical investigation for induction of therapeutic neovascularization [4, 5]. Although evidence suggests efficacy of first generation cell therapeutic approaches (for review cf. [6]), it is conceivable that this treatment would benefit from enhanced recruitment of the therapeutic cell population in the ischemic region [1]. This holds particularly true, since physiologic progenitor cell function—usually present in preclinical in vivo studies—is generally impaired in the presence of cardiovascular risk factors [7]. For example, diabetes mellitus affects the nitric oxide-mediated mobilization of proangiogenic cells [8], whereas smoking is associated with an increase of oxidants blunting the proangiogenic efficacy [9]. Increased body weight also carries the risk of a p38-mediated loss of proangiogenic function during episodes of weight gain [10].

In addition, previous studies have shown that homing of circulating progenitor cells depends on active interaction with the vascular wall, which might be quiescent at the time of cell application, thus minimizing recruitment of circulating proangiogenic cells [11]. Indeed, strategies to stimulate vascular adhesion properties have documented increased recruitment of EPCs [12].

An enhancement of cell homing has been proposed as a potential treatment option to help compensate for the functional loss of vascular progenitor cells (VPCs) in these patients. Stroma-derived factor-1 (SDF-1)/CXCL12 [13], insulin-like growth factor-1 [14], and the proangiogenic peptide LL37 [15] have been successfully used to prime EPCs for increased recruitment to ischemic tissue. SDF-1 appears particularly well suited, as it is capable of enhancing the recruitment of endothelial [4] and smooth muscle [16] progenitors as well as proarteriogenic mononuclear cells [17]. This spectrum of VPC recruitment appears well suited to the complex task of neovascularization comprising endothelial proliferation and capillary growth (angiogenesis) [18], pericyte and smooth muscle recruitment (microvessel maturation) [19], and monocyte/macrophage-dependent collateral growth (arteriogenesis) [20]. Notably, disturbance of the balanced growth and maturation of the microvascular and macrovascular compartments apparently results in failure to increase perfusion [21, 22].

We have previously used clonal embryonic endothelial progenitor cells (eEPCs) of murine origin [23] as a model system to induce neovascularization [15, 24, 25]. These cells resemble adult human proangiogenic cells with respect to induction of neovascularization [24] and SDF-1-dependent recruitment [26, 27].

To test the hypothesis that enhanced VPC recruitment will improve neovascularization in a stable chronic ischemic hind limb model, we designed and tested a fusion protein coupling SDF-1/CXCL12 to a fractalkine/CX3CL1 mucin stalk and a glycosylphosphatidylinositol-membrane anchor (S1FG), which was overexpressed in the ischemic tissue to promote homing of CXCR4+ progenitor cells. As cell source for these experiments, either murine eEPCs were locally applied via retroinfusion or endogenous circulating bone-marrow-derived VPCs were mobilized via transient AMD3100 application.

Materials and Methods

Chemicals were purchased from Sigma-Aldrich (Taufkirchen, Germany), Merck KGaA (Darmstadt, Germany,, and Carl Roth GmbH & Co. KG (Karlsruhe, Germany, Contrast agent Solutrast 370 was obtained from Bracco Altana Pharma AG (Konstanz, Germany,

Construction of the S1FG-Plasmid

The human CXCL12 (SDF-1) coding sequence (bp 90–359 of GenBank entry NM_199168.2) was amplified by PCR from human cDNA using 5′CGACCCGAATTCGCGCTCGTCC3′ and 5′GGCTGTTGTGCTCTAGACTTGTTTAAAGCTTTCTC3′ as primers. The resulting 331 bp PCR product comprised the entire coding sequence of CXCL12 except the stop codon and an EcoRI restriction site at the 5′ end as well as an XbaI site at the 3′ end. These restriction sites were used to ligate the CXCL12 gene in-frame into the vector containing the GPI signal sequence of LFA-3, immediately 3′ of the signal sequence. The GPI anchor signal sequence of LFA-3 (amino acids 203–232 of the translation of GenBank entry NM_001799.2) had previously been amplified from a human inflamed kidney cDNA sample [28], whereby XbaI (5′) and SalI (3′) restriction enzyme recognition sequences had been introduced via the PCR primers.

The mucin domain of CX3CL1 (Fractalkine; amino acids 100–341 of the translation of GenBank entry BC016164.1) was then amplified from a human inflamed kidney cDNA sample using the primers 5′CGGCGGGCGTCTAGAGGCGGCACCTTCG3′ and 5′CGTGTCGGCGCTAGCCTGCCTCCGGG3′. These primers introduced XbaI and NheI restriction sites at the 5′ and 3′ ends, respectively, which were used to ligate the PCR product in-frame in between the SDF-1 (CXCL12) gene and the GPI signal sequence via the connecting XbaI site. The resulting construct (S1FG, Fig. 1A) thus consisted (in 5′ to 3′ or N- to C-terminal order) of the SDF-1 coding sequence lacking a stop codon, the mucin domain of CX3CL1, and the GPI signal sequence of LFA-3. In addition, a cleavage-resistant fusion plasmid (S1FG-R) featuring a mutated CD26 binding site as well as a deleted MMP-9 cleavage site in the mucin domain was designed and cloned.

Figure 1.

Design and functionality of the S1FG fusion protein. (A): Construction of the recombinant fusion protein (S1FG), combined of SDF-1, the mucin domain of CX3CL1, and the GPI-anchor. (B): Representative images of S1FG-eEPC binding in vitro. Overexpression of S1FG-GFP (green) in human microvascular endothelial cells leads to eEPCs (DiD = red fluorescence), adhesion at site of S1FG expression. (C): Fluorescence-activated cell sorting analyses of CXCR4 receptor expression on the eEPC cell surface. Overexpression of S1G or S1FG in cocultured CHO-cells leads to a decrease of CXCR4 on eEPCs cell surface, indicating internalization of the activated receptor. (D): Coincubation of Fluo-4 Dye loaded eEPCs with transfected CHO-cells showed an increase in calcium mobilization in the presence of S1G or S1FG, displayed as delta fluorescence compared to coincubation with control CHO-cells, indicating activation of the intracellular signaling cascade of the CXCR4 receptor. **, p < .01. Abbreviations: GFP, green fluorescent protein; SDF-1, stroma-derived factor-1.

For functional analysis of S1FG-mediated eEPC adhesion, a S1FG-green fluorescent protein (GFP)-tagged construct was cloned and expressed in human microvascular endothelial cells (HMEC) plated on 35 mm µ-Slides (Ibidi, Martinsried, Germany, Twenty minutes after coincubation with DiD-labeled eEPCs, plates were washed with phosphate-buffered saline (PBS) and confocal microscopy was performed to analyze colocalization (Fig. 1B).

eEPC Studies

The isolation of murine eEPCs has been previously described [23]. Briefly, thrombomodulin-expressing cells were isolated from thrombomodulinlacZ/+ transgenic mice embryos on day E7.5 and cultivated on feeder layers. Single colonies were then picked and expanded on gelatin-coated plates and propagated. All experiments were performed with clonal cells. The eEPCs were grown at 37°C in Dulbecco's modified Eagle's medium with L-Glutamine and 20% fetal bovine serum.

CXCR4 Internalization Assay

For the analysis of CXCR4 internalization, CHO-cells (Chinese Hamster Ovarian cells) overexpressing S1FG, S1G, or nontransfected were transferred to a 96-well plate in a concentration of 6 × 105 cells per well. eEPCs were cocultured in a concentration of 2 × 105 per well and a CXCR4-specific antibody (1.5 µg/ml; e-Bioscience 12–9991-81) was added at 4° for 45 minutes. After 30 minutes incubation, cells were washed and resuspended in ice cold fluorescence-activated cell sorting (FACS) buffer. After washing, cells were resuspended in FACS buffer and measured for CXCR4 expression on the surface of eEPCs. Results are displayed as Δ fluorescence from cocultured nontransfected CHO-cells.

Calcium Mobilization Assays

The intracellular calcium mobilization was measured in Fluo-4-loaded eEPCs (Fluo-4 NW calcium assay kit; Invitrogen) cocultured with transfected CHO-cells expressing S1G, S1FG, or untransfected controls. After calcium loading for 45 minutes and 37°C according to the manufacturer's instruction, 50 µl of 5 × 106 eEPCs/ml in assay buffer was transferred to 96-well plates. Then 50 µl of assay buffer resuspended CHO-cells were added in a concentration of 1 × 107 cells per milliliter. Immediately, fluorescence was measured using 494 nm excitation and 516 nm emission wavelength in a microplate reader (Safire 2, Tecan, Austria, Measurements were taken every 20 seconds for a total time period of 100 seconds. Results are given as relative fluorescence.

Flow Chamber Shear Stress Adhesion Experiments

For shear dependent adhesion, Ibidi slides (Martinsried, Germany) were seeded with human umbilical vein endothelial cells (HUVECs) previously transduced by an AAV6 encoding S1FG, S1G, SDF-1, or GFP (transfection efficacy 70%). For specific blockage, THP-1 cells were preincubated with the following antibodies: anti-SDF-1 (Abcam, ab49124, Abcam, Cambridge, UK, http:/ anti-L-selectin (Becton-Dickinson, Heidelberg, Germany,, DREG-56), or control-antibody. After 18 hours TNFα-stimulation (Agilent Technologies, Santa Clara, CA,, Germany, 15 ng/ml), activated HUVECs were superfused with human THP-1 cells at a constant shear stress of 1 dyn/cm2 (venular shear stress) and 10 dyn/cm2 (for arteriolar shear stress). All cells stationary for more than 10 seconds were counted as adherent cells.

Rabbit Model of Hind Limb Ischemia

New Zealand White rabbits were obtained from Charles River Deutschland GmbH (Sulzfeld, Germany, Animal care and all experimental procedures were carried out in accordance to the German and National Institutes of Health animal legislation and were approved by the local animal care and use committee (AZ 2531-140-07). Animal experiments were performed at the Walter-Brendel-Centre of Experimental Medicine, as previously described [15]. Briefly, at day 0 (d0), animals (n = 5 per group) were subjected to right femoral artery excision. At day 7, baseline angiography of each hind limb using automated contrast agent injection was performed, followed by treatment as indicated. On day 35, a second angiography was obtained to elucidate the therapeutic effects. Thereafter, the animals were sacrificed and muscle samples were acquired for further analysis (supporting information Fig. S1A).

S1FG Application In Vivo

Liposomes (Effectene, Qiagen, Hilden, Germany, encoding for S1FG, SDF-1, or GFP (250 µg cDNA each) were administered via retroinfusion into the anterior tibial vein as previously described [29]. To observe the therapeutic effects of exogenous cell application after S1FG-transfection of the hind limb, animals were treated with retroinfusion of S1FG, SDF-1, or saline on day 7 ± eEPClow (2 × 106) or eEPChigh (5 × 106 = eEPCs only group) on day 9. In order to examine the effect of endogenous cell mobilization, S1FG or S1FG-R plasmid was retroinfused on day 7 and/or 1 ml of saline or 1 mg of AMD3100 (Sigma-Aldrich, Deisenhofen, Germany, was injected intraperitoneally on days 9, 10, and 11. AMD3100 sufficed to increase the VPC population in vivo (supporting information Fig. S1B, S1C).

PCR analysis of a reporter gene revealed a maximum of liposomal expression at day 3 (supporting information Fig. S1D). To assess S1FG protein expression, calf muscle samples were obtained 4 days after transfection of the S1FG-GFP construct. Endothelial cells were stained with an anti-PECAM-1 antibody (sc-1506, Santa Cruz Biotechnology Inc., Santa Cruz, CA, and examined by fluorescence microscopy.

Adhesion of eEPCs In Vivo

In vivo, adhesion to postcapillary venules 5 × 106 DiD-colored eEPCs were applied via retroinfusion on day 9 (with or without S1FG pretransfection on day 7). By virtue of this application mode, eEPCs were distributed homogenously throughout the limb with gentle pressure (<20 mmHg) and outflow occlusion with a handcuff [30]. Hind limb musculature from day 11 was analyzed under fluorescence microscopy and 10 representative high power fields per muscle were counted.

Capillary Density Measurements

Vascular growth of the microcirculation (e.g., angiogenesis) was analyzed by PECAM-1-antibody staining (sc-1506, Santa Cruz Biotechnology Inc.) and a rhodamine-conjugated secondary antibody (sc-2094, Santa Cruz Biotechnology Inc.). Nuclear staining was performed by DAPI. Capillaries and muscle fibers were quantified in at least eight representative cross-sectional muscle areas per animal. Results are given as capillary per muscle fiber ratio (c/mf).

Collateral Growth Measurement

Collateral growth was examined by a standardized grid overlay on a representative angiographic picture. The crossing points of the template and contrast-agent filled visible vessels between the inguinal ligament and the knee joint were counted, weighing in length and branching of collaterals, as previously described [30]. Results are given as a ratio between day 35 and day 7.

Perfusion Measurement

Perfusion was assessed by frame count analysis, as previously described [30]. Briefly, perfusion velocity was assessed by measuring the time (number of frames, recording time = 25 frames/minute) the contrast agent needed for the passage from the proximal landmark (iliac bifurcation) to the distal landmark (knee joint) and normalized to the left (nonischemic) leg. Results are given as ratio of day 35 to day 7 values.

Statistical Analysis

All data are presented as mean ± SD. Differences between several groups were tested using ANOVA and Student Newman Keul's post hoc analysis. p < .05 was considered statistically significant. All data were assessed by SPSS software package (Chicago, IL, Version 19.0,


S1FG, a Bioactive Fusion Protein Induces CXCR4 Activation and Internalization

The chemokine ligand SDF-1 is widely used by stem and progenitor cells equipped with its receptor CXCR4. For efficient function, chemokines also require adhesion molecules that are activated in response to chemokine receptor stimulation. The chemokine fractalkine shows enhanced leukocyte capture properties as it is membrane expressed and includes a long mucin stalk that allows the chemokine to function with reduced requirements for additional adhesion molecules [31]. To provide SDF-1 with proadhesive properties for stem and progenitor cells, SDF-1 was fused to the fractalkine mucin stalk. Finally, for membrane association, a GPI-anchor taken from LFA-3 was included in the construct (S1FG, Fig. 1A). To assess the importance of the mucin stalk, a second construct was generated where the SDF-1 was directly linked to the GPI-anchor (S1G). In vitro overexpression of the fusion protein (S1FG) in HMECs enabled S1FG-dependent eEPC-adhesion as displayed in Figure 1B. Red-labeled eEPCs are shown to adhere to the HMECs at regions where S1FG-GFP (green) is expressed. To further validate the bioactivity of the S1FG fusion protein, induced internalization of CXCR4 was used to assess receptor activation. FACS analyses of surface CXCR4 expression on eEPCs were assessed after coculture with CHO-cells overexpressing either S1G or S1FG. We found a significant decrease of cell surface expression of CXCR4 (48% ± 7% and 46% ± 4%) on eEPCs as compared to coculture with nontransfected CHO-cells (76% ± 6%, Fig. 1C).

Activation of CXCR4 also results in intracellular calcium shift from the endoplasmatic reticulum into the cytoplasm. As a further assay for specific bioactivity of S1FG or S1G, intracellular calcium influx in response to stimulation by these constructs was assessed using Fluo-4 Dye loaded eEPCs cocultured with S1G, S1FG expressing, or control CHO-cells. Presence of either S1G or S1FG on CHO-cells triggered a higher intracellular calcium mobilization than control CHO-cells indicating bioactivity of the receptor (Fig. 1D).

S1FG Effects In Vitro and In Vivo Recruitment of VPCs

To determine the influence of the mucin stalk in the context of physiological flow conditions, a parallel flow chamber assay was used to assess adhesion under flow at venous shear stress (1.0 dyn/cm2). A significant increase of firm adhesion of CXCR4+ THP-1 cells was seen on S1FG-transfected HUVECs as compared to SDF-1, S1G, or GFP-transfected cells (Fig. 2A–2C, supporting information Fig. S2C). Preincubation of HUVECs with an SDF-1 antibody abolished this effect (Fig. 2D, supporting information Fig. S2C). Of note, no firm adhesion was observed under arterial shear stress (10 dyn/cm2) during an observation time of 8 minutes. Taken together, under venous shear stress, S1FG provides increased firm adhesion superior to the SDF-1 adhesion, but sensitive to SDF-1 blockade.

Figure 2.

S1FG enhances adhesion of mononuclear cells. (A, B): Adhesion of mononuclear THP-1 cells (native in A, marked in B) under shear stress (1 dyn/cm2, 10-fold magnification) to HUVECs transfected with an AAV encoding for either SDF-1, S1G, or S1FG. Firm adhesion (C) showed a significant increase after S1FG overexpression. (D): Blocking either SDF-1 or L-Selectin led to a decrease in THP-1 firm adhesion. *, p < .05; **, p < .01. Abbreviations: GFP, green fluorescent protein; SDF-1, stroma-derived factor-1.

Since in a clinical setting the recruitment of proinflammatory cells such as polymorphonuclear neutrophils (PMN) [32] via S1FG might confound the beneficial effect of vascular progenitor recruitment, we investigated the adhesion of PMN on an S1FG-transfected HUVEC monolayer in vitro. Of note, although S1FG and wild-type fractalkine increased PMN adhesion compared to controls (43 ± 8 cells/field and 68 ± 8 cells/field vs. 18 ± 2 cells/field in pcDNA-controls), wild-type fractalkine carried a higher risk of PMN recruitment than S1FG (supporting information Fig. S2A, S2B). Thus, S1FG appears significantly less proinflammatory than fractalkine, which is well-known for its capability to provide leukocyte capture at low shear rates [31]. To test the efficacy of S1FG in recruiting circulating VPCs, we injected S1FG in the muscle of rabbit hind limbs followed by retrograde injection of fluorescently labeled eEPCs in the tibial vein during outflow obstruction (cf. Materials and Methods). Histological analysis of hind limb muscle sections revealed expression of the GFP-labeled S1FG fusion molecule, which correlated with adherent CXCR4-positive eEPCs in the vicinity (Fig. 3A). The majority of cells was attracted to the lower limb (Fig. 3C) (40 ± 8 vs. 15 ± 3 cells/field in the gastrocnemic muscle, 48 ± 5 vs. 7 ± 3 cells/field in the much smaller fibularis muscle), whereas there was a 2.6-fold increase (24 ± 2 cells/field) in the thigh muscles (Fig. 3D). Notably, this effect did not occur in the untransfected left hind limb (left leg: 13 ± 4 cells/field (thigh) vs. 11 ± 2 cells/field, data not shown).

Figure 3.

S1FG increases eEPC adhesion to endothelial cells in ischemic muscles. (A): Compared to the adhesion of DiD-labeled eEPChigh (5 × 106) in calf muscle of control (pcDNA-pretransfection) animals (left picture), S1FG-pretransfection (green) recruited eEPClow (2 × 106) to the transfected ischemic tissue 2 days after cell retroinfusion (scale bar = 10 µm). (B): Low-power field of either control or S1FG-transfected rabbit hind limb tissue displaying DiD-labeled eEPC recruitment (scale bar = 50 µm). (C, D): Quantification of eEPC adhesion after S1FG-pretreatment and eEPClow application showed significantly higher values than eEPChigh alone in calf muscles (C) and in thigh muscles (D). *, p < .01.

In Vivo Effect of S1FG on Neovascularization Via Recruitment of Exogenously Applied Cells

In order to assess the functional effects of S1FG- and eEPC-treatment in vivo, we conducted long-term experiments and measured capillary density, collateral growth, and limb perfusion. As in vivo cell adhesion studies suggested an approximately 2.5-fold increase in cell recruitment after pretreatment of the ischemic limb with S1FG (see above, Fig. 3B–3D), we tested the prediction that using S1FG the number of exogenously applied eEPCs could be reduced by a factor of 2.5 (S1FG + 2 × 106 eEPClow compared to 5 × 106 eEPChigh) without affecting the overall number of recruited eEPCs. Capillary growth in ischemic lower limb increased similarly in either eEPC group (1.8 ± 0.1 capillaries/muscle fiber (S1FG+ eEPClow) and 1.9 ± 0.1 c/mf (eEPChigh) versus 1.3 ± 0.1 c/mf (control), Fig. 4A, 4B). Of note, transfection of SDF-1+eEPClow did not increase capillary density compared to controls (1.5 ± 0.1 c/mf vs. 1.5 ± 0.2 c/mf Fig. 4A, 4B), suggesting that enhanced recruitment was primarily due to adhesion of eEPCs instead of increased chemoattraction.

Figure 4.

S1FG improves eEPC-induced neovascularization. (A): Representative examples of PECAM-1 (Rhodamine) and 4',6-diamidino-2-phenylindole dilactate (DAPI) (blue) stained ischemic calf muscles in saline (control), eEPChigh (5 × 106), S1FG + eEPClow (2 × 106), and SDF-1 + eEPClow (2 × 106) treated animals (scale bar = 20 µm). (B): Quantification of capillary density in the calf of the ischemic leg shows an increase in capillary/muscle fiber ratio (c/mf) after eEPChigh or S1FG + eEPClow treatment. (C): Representative angiographic images of day 7 and day 35 after femoral artery excision. (D): Quantification of collateral formation reveals an increased collateral formation after eEPChigh retroinfusion, an effect mimicked by S1FG-pretransfection with eEPClow application. (E): Analysis of perfusion mirrors the change in collateral growth. *, p < .05. Abbreviation: SDF-1, stroma-derived factor-1.

Collateral growth assessment revealed that application of SDF-1+eEPClow evoked a modest but not significant arteriogenic response (130% ± 10% vs. 105% ± 4% in controls Fig. 4C, 4D), whereas pretransfection of S1FG clearly increased collateral growth, even with the eEPClow concentration (201% ± 11%). Similarly, a modest improvement of conductance vessel formation was achieved by eEPChigh application alone (140% ± 7%). Even after correcting for the anticipated increase in eEPC recruitment in the S1FG+eEPClow group (2 × 106 eEPClow compared to 5 × 106 in the eEPChigh group), collateral formation significantly exceeded the level of the eEPChigh group.

In agreement with the observed changes in collateralization, perfusion was modestly but not significantly increased by SDF-1+eEPClow (140% ± 7% vs. 102% ± 10% in controls, Fig. 4E). Perfusion was accelerated by retroinfusion of eEPChigh (169% ± 3%) and, consistent with the collateral growth, further increased by combining eEPClow with S1FG-pretreatment (217% ± 7%, Fig. 4E). Thus, S1FG increases the biological efficacy at least in part by increasing recruitment of exogenously applied eEPCs, despite the observation that the exogenously applied eEPC population is fading over time [24].

In Vivo Effect of S1FG on Neovascularization Via Recruitment of Endogenous Progenitor Cells

Since enhancing recruitment of endogenous progenitor cells is an attractive alternative approach to cell transplantation in the clinical setting, we tested whether S1FG pretreatment in combination with mobilization of endogenous VPCs from the bone marrow is capable of inducing neovascularization. In order to mobilize progenitor cells, we injected the short-acting CXCR4 antagonist AMD3100 intraperitoneally on 3 consecutive days, starting 48 hours after S1FG transfection. This agent leads to egress of vascular progenitors from the bone marrow, but is short acting so that it would have minimal effects on S1FG or S1G/CXCR4 interactions. To confirm release of CXCR4+ cells from the bone marrow, CXCR4—GFP transgenic mice were analyzed for CXCR4+ cell concentration in the peripheral blood. AMD3100 pretreatment increased CXCR4+ cells three- to fourfold, as shown by FACS analysis performed 3 days after pulsatile AMD3100 treatment (supporting information Fig. S1B, S1C). Moreover, after S1FG+AMD treatment CD45+ cells expressing CD31 were recruited at higher percentage to ischemic murine hind limbs than in control animals (supporting information Fig. S3A, S3B), similar to monocytic CD45+CD11b+ cells, whereas neutrophil recruitment was found less pronounced (supporting information Fig. S3A–S3F).

In rabbits, application of S1FG alone increased capillary density to a modest, but (1.8 ± 0.2 c/mf vs. 1.3 ± 0.1 c/mf) significant extent versus control animals, whereas AMD3100 alone (1.5 ± 0.1 c/mf) showed no significant changes in capillary sprouting (Fig. 5A, 5B). However, the combination of S1FG and AMD3100 increased capillary density (1.8 ± 0.1 c/mf), without significantly exceeding the level provided by S1FG alone (Fig. 5B). In order to test whether protease cleavage of S1FG was limiting its efficacy, we applied S1FG-R, a mutant of S1FG resistant to cleavage of the SDF-1 by matrix metalloproteinase 2 and dipeptidylpeptidase-4 (DPP4) [33] and the fractalkine stalk by TACE [34] and ADAM-10 [35]. Comparison of S1FG and S1FG-R in combination with AMD3100 did not display significant differences in capillary density (1.8 ± 0.1 c/mf vs. 1.7 ± 0.1 c/mf in S1FG-R + AMD3100), indicating that S1FG cleavage did not take place in vivo to a significant degree.

Figure 5.

Enhanced endogenous progenitor cell mobilization and recruitment mimics the effect of exogenous eEPC application. (A): Examples of PECAM-1 and DAPI stained ischemic calf muscles (scale bar = 20 µm). (B): Capillary growth did not differ between the groups. (C): Representative angiographic pictures of day 7 and day 35. (D, E): Quantification of collateral growth (D) and perfusion (E) reveals that mobilization of endogenous progenitor cells via AMD3100 after S1FG-pretransfection is enhanced compared to single factor therapy. Use of the protease-resistant S1FG (S1FG-R) did not further increase the effect of S1FG. *, p < .05. Abbreviation: DAPI, 4',6-diamidino-2-phenylindole dilactate.

Quantification of collaterals revealed that the application of AMD3100 alone elevated the collateral count above control level (154% ± 8% vs. 105% ± 4% in controls, Fig. 5C, 5D), which also translated into an increase of perfusion (130% ± 4% vs. 102% ± 10% in controls, Fig. 5E). Still, AMD3100 alone was significantly less effective in initiating collateral growth than the combination with S1FG treatment (195% ± 5%) or with the degradation-resistant mutant S1FG-R (178% ± 5%, Fig. 5D). Similarly, S1FG alone resulted in a modest increase of collateral formation and perfusion, but remained significantly less effective than the combination of S1FG or S1FG-R + AMD3100, which equally augmented perfusion (193% ± 9% vs. 180% ± 7%). Taken together, our findings show that S1FG is capable of recruiting circulating VPCs which initiate neovascularization on the microvascular and macrovascular level.


Our data suggest that the efficacy of cell therapy to stimulate neovascularization can be increased by enhanced cell recruitment to the target area by in vivo overexpression of an engineered S1FG fusion protein in the ischemic endothelium. Mechanistically, S1FG is capable of providing firm adhesion of VPCs under venular shear stress conditions in vitro (Fig. 2) and in vivo after regional venous retroinfusion (Fig. 3), and thus further increase eEPC-mediated neovascularization (Fig. 4). Of note, the S1FG-mediated gain of collateral growth and perfusion after exogenous progenitor cell application exceeded the predicted efficacy of the cell dose (cf. Materials and Methods, Fig. 4D, 4E), indicating simultaneous enhancement of endogenous progenitor cell recruitment. In order to explore this capability, we combined S1FG-transfection with mobilization of endogenous bone marrow-derived VPCs by AMD3100 on 3 consecutive days. The AMD3100 mobilization approach, which on its own was inefficient with respect to the neovascular growth response, boosted the S1FG-mediated collateral growth and perfusion to the level achieved by exogenous cell application. Thus, S1FG in combination with AMD3100 appears as an attractive principle for cell-mediated treatment of chronic ischemic muscle disease.

It has been observed a decade ago that SDF-1 is instrumental in providing stem and progenitor cell homing to the injured heart [36]. Although SDF-1 itself is capable of recruiting a variety of stem and progenitor cells via its chemoattractant function, however, an overlap to inflammatory cells may limit its use in settings which are sensitive to overwhelming inflammation, for example, myocardial ischemia and reperfusion [32].

Mobilization of endogenous progenitor cells from the bone marrow has been achieved by a variety of factors, among them granulocyte colony-stimulating factor (G-CSF). Although the amount of circulating progenitor cells is rising substantially upon G-CSF application, meta-analysis of clinical trials conducted with G-CSF after myocardial ischemia and reperfusion did not yield a significant improvement. More recently, it has been used in combination with SDF-1 in order to enhance homing, for example, by blocking of SDF-1 cleavage through DPP IV [37]. Of note, G-CSF itself appears to exert its function as mobilizer of bone marrow cells through proteolytic cleavage of the SDF-1 [38], which via binding to CXCR4 is a major regulator of bone marrow cell retention [39]. Temporarily disrupting this binding by the use of the short acting small molecule CXCR4 antagonist (AMD3100) [40] appeared as an attractive approach, since it rendered the mobilized cells responsive to another SDF-1 signal thereafter. Exploiting this virtue of SDF-1, Jujo et al. demonstrated a beneficial use of a single dose of AMD3100 in experimental myocardial infarction, whereas continuous release of AMD3100 by an osmotic minipump over 14 days adversely affected infarct size and remodeling [41].

Of note, although SDF-1 upregulation was revealed during the first 3 days after myocardial infarction already a decade ago [42], only recently this observation triggered a successful translation into antiremodeling therapy of mouse hearts subjected to permanent coronary occlusion by combining the SDF-1 axis with other agents such as G-CSF [37] or sonic hedgehog [43]. Similarly, novel formulations of SDF-1, for example, by nanofiber packaging of protease-resistant SDF-1 [33], or by combination with GPIV, a platelet collagen receptor [44], significantly increased the efficacy of SDF-1.

Although SDF-1 is viewed as an important stem cell recruiting molecule, the role of neovascularization in SDF-1 modulated remodeling after myocardial infarction is not fully elucidated yet. Thus, we turned to a hind limb neovascularization model in which ischemia is present without fibrosis and/or remodeling. In the rabbit hind limb ischemia model, both angiogenesis and arteriogenesis have to be enhanced in order to gain perfusion [22]. In this model, S1FG, a membrane bound version of SDF-1, acting as an adhesion-molecule (cf. Fig. 1), sufficed to induce a robust increase of functional neovascularization, providing a gain of perfusion, when combined with supply of either exogenously transfused or endogenously mobilized VPCs (Figs. 4, 5).

Of note, capillary density was not increased by the higher number of CXCR4+ cells recruited by S1FG (Figs. 4B, 5B), apparently not driving the subsequent gain of perfusion. This finding replicates previous results of a complementary strategy, enhancing eEPC recruitment by elevating the NFκB activity of this cell population [15]. A potential ceiling to angiogenesis maybe due to the limitations of the model, which may allow for only transient increases of circulating progenitor cells, which are abrogated at day 14 after exogenous application [24] and may return to normal levels within days after transient endogenous recruitment mobilization by AMD3100 application. Nonetheless, S1FG combined with exogenous eEPCs or AMD3100 raised the arteriogenic growth level. The interplay between early angiogenic microcirculatory growth and subsequent collateral growth requires a high proportion of mature microvessels, as illustrated by vascular endothelial growth factor-A (VEGF-A) overexpression. Recombinant adeno-associated viruses containing VEGF-A at modest doses induce a high number of capillaries, but are unable to provide arteriogenesis unless microvessel maturation is boosted by the pericyte and smooth muscle cell recruiting molecule PDGF-B [22]. In addition, a substantial recruitment of progenitor cells took place in the thigh musculature, potentially allowing for a direct activation of arteriogenesis (Fig. 3D).

Why does S1FG increase adhesion of progenitor cells to a larger extent than SDF-1 (cf. Figs. 2, 3)? From the early observation of Abbott and colleagues, it is known that the endogenous local SDF-1 signal in an acute ischemic heart muscle is peaking at 72 hours, and fading thereafter [42]. Therefore, extension of the presence of the chemokine signal pathway may be a valuable approach for neovascularization [45]. However, SDF-1 itself is not an efficient molecule to draw circulating progenitor cells to an endothelium lining under flow conditions (Fig. 2, cf. [44]), since it mediates cell activation, but not firm adhesion. The combination with the mucine stalk of fractalkine altered this chemokine feature toward an adhesion molecule of VPCs. Using the improved progenitor adhesion, S1FG triggered a larger vascular growth response of exogenous and endogenous VPCs.

This mechanism is in support of other approaches of progenitor cell enhancement [13, 29, 46]. However, universal promotion of circulating cell adhesion may also increase recruitment of proinflammatory cells to atherosclerotic plaques, which display an array of chemokines [16, 47] and adhesion molecules [48], exacerbating atherosclerosis [49, 50]. In contrast, a significant advantage of the described approach is that enhancing local expression of chemoattractants and/or adhesion molecules provides a more selective way of directing circulating cells to the target region.

In conclusion, we demonstrate that S1FG, an engineered chemokine-based adhesion molecule comprising an SDF-1 head, a fractalkine stalk, and a GPI anchor, elicited an improved homing of exogenously applied and endogenously mobilized VPCs. The finding of an improved neovascular response on both, the microvessel and the macrovessel compartment, indicates the initiation of a balanced vascular growth response, which correlates with an enhanced progenitor cell homing in the calf and thigh musculature. The inclusion of a GPI-anchor in these constructs allows the isolation of recombinant versions of S1FG. Recombinant proteins with GPI-anchors can be efficiently incorporated into cell surface [51]. This suggests that in future recombinant S1FG or related agents could be applied via a catheter directly to sites of ischemic damage. Finally, the structural growth response translated into a gain of perfusion, indicating a close correlation between progenitor cell homing and vascular growth response. Shifting this correlation to a significantly higher level by S1FG expression on the endothelium of ischemic hind limbs thus is an appealing strategy for enhancement of progenitor cell therapy in chronic peripheral artery disease.


This study was in part supported by the Deutsche Forschungsgemeinschaft (Grant KU 1019/11-2 to C.K.), the Bundesministerium für Bildung und Forschung (Grant 01GU 1105A to C.K.), and FöFoLe-grants of the Ludwig-Maximilians-University to G.S., F.G., and T.T. Wild-type fractalkine plasmid was kindly provided by S. Massberg, Munich, Germany. CXCR4-GFP transgenic mice were kindly provided by F. Krombach. We are indebted to Elisabeth Raatz, Susanne Helbig, Tien Cuong Kieu, and Susanne Bierschenk for excellent technical assistance.

Disclosure of Potential Conflicts of Interest

The authors declare no potential conflict of interest.