Capturing the stem cell paracrine effect using heparin-presenting nanofibres to treat cardiovascular diseases


  • Matthew J. Webber,

    1. Department of Biomedical Engineering, Northwestern University, Evanston, IL, USA
    Search for more papers by this author
  • Xiaoqiang Han,

    1. Department of Pathology, Northwestern University, Chicago, IL, USA
    2. Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
    Search for more papers by this author
  • S. N. Prasanna Murthy,

    1. Department of Pathology, Northwestern University, Chicago, IL, USA
    2. Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
    Search for more papers by this author
  • Kanya Rajangam,

    1. Department of Biomedical Engineering, Northwestern University, Evanston, IL, USA
    Search for more papers by this author
  • Samuel I. Stupp,

    1. Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
    2. Department of Materials Science and Engineering, Northwestern University, Evanston, IL, USA
    3. Department of Chemistry, Northwestern University, Evanston, IL, USA
    Search for more papers by this author
  • Jon W. Lomasney

    Corresponding author
    1. Department of Pathology, Northwestern University, Chicago, IL, USA
    2. Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
    • Northwestern University Feinberg School of Medicine, Ward 3-210, 303 East Chicago Avenue, Chicago, IL 60611, USA.
    Search for more papers by this author


The mechanism for stem cell-mediated improvement following acute myocardial infarction has been actively debated. We support hypotheses that the stem cell effect is primarily paracrine factor-linked. We used a heparin-presenting injectable nanofibre network to bind and deliver paracrine factors derived from hypoxic conditioned stem cell media to mimic this stem cell paracrine effect. Our self-assembling peptide nanofibres presenting heparin were capable of binding paracrine factors from a medium phase. When these factor-loaded materials were injected into the heart following coronary artery ligation in a mouse ischaemia-reperfusion model of acute myocardial infarction, we found significant preservation of haemodynamic function. Through media manipulation, we were able to determine that crucial factors are primarily < 30 kDa and primarily heparin-binding. Using recombinant VEGF- and bFGF-loaded nanofibre networks, the effect observed with conditioned media was recapitulated. When evaluated in another disease model, a chronic rat ischaemic hind limb, our factor-loaded materials contributed to extensive limb revascularization. These experiments demonstrate the potency of the paracrine effect associated with stem cell therapies and the potential of a biomaterial to bind and deliver these factors, pointing to a potential therapy based on synthetic materials and recombinant factors as an acellular therapy. Copyright © 2010 John Wiley & Sons, Ltd.

1. Introduction

Several clinical trials are under way to assess the intramyocardial transplantation of adult bone marrow-derived stem cells as a potential therapy to repair damaged myocardium following an acute infarct (Burt et al., 2008). The mechanism by which these transplanted stem cells elicit disease-ameliorating effects remains controversial (Field, 2006; Jaquet et al., 2005; Mazhari and Hare, 2007; Wang and Li, 2007). Some studies demonstrate cardiomyogenic transdifferentiation of transplanted stem cells, leading to the generation of de novo myocardium, restoring ventricular structure and enhancing functional performance (Orlic et al., 2001a, 2001b; Rota et al., 2007; Shake et al., 2002; Toma et al., 2002). However, poor retention within the myocardium (6% or less) of transplanted bone marrow-derived cells (Freyman et al., 2006), combined with little or no functional integration of residual cells (Scherschel et al., 2008), leads to questions of whether transplanted cells are capable of restoring significant function simply through de novo myocardium formation. More recently, this has led others to postulate that the mechanism for stem cell transplant efficacy, particularly in the heart, is paracrine factor-linked (Gnecchi et al., 2005; Pittenger and Martin, 2004; Tang et al., 2004, 2005; Zhang et al., 2007). The paracrine, or cell-help-cell, effect suggests that soluble proteins secreted by these transplanted cells are the major contributors to cardiac protection, repair and regeneration following infarct (Gnecchi et al., 2008). Mesenchymal stem cells in particular are known to secrete a broad variety of cytokines, chemokines and growth factors that may be involved in cardiovascular signalling (Caplan and Dennis, 2006). For example, factors such as vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) are found in significantly increased concentrations in injured myocardium following treatment with stem cells (Nagaya et al., 2005; Yoon et al., 2005).

Previously, we developed a class of peptide amphiphile (PA) molecules comprised of short peptide sequences covalently linked to a fatty acid tail and designed to spontaneously self-assemble into high aspect ratio nanofibres under physiological conditions (Beniash et al., 2005; Hartgerink et al., 2001, 2002). This technology aims to translate advancements in the field of nanoscience into a synthetic therapy, especially because of its natural components and biodegradable nature. One such molecule, a heparin-binding peptide amphiphile (HBPA), incorporates a consensus Cardin–Weintraub heparin-binding sequence and was designed to mimic natural heparin-binding proteins with a short synthetic peptide (Rajangam et al., 2006, 2008). Upon mixing, the negatively charged glycosaminoglycan interacts with the positively charged HBPA molecule, inducing the HBPA molecules to self-assemble into nanofibres. Bundling of these nanofibres produces a gel matrix consisting of fibrillar nanostructures of similar dimensions to natural extracellular matrix components. These self-assembled nanofibres immobilize and present heparin in a biomimetic fashion. The presentation of heparin makes possible the binding of a variety of proteins via heparin-binding domains, increasing cellular recognition of these factors. For example, heparin binds angiogenic factors, including VEGF and bFGF, leading to enhanced receptor dimerization and protection of these factors against enzymatic degradation (Ferrara et al., 2003; Herr et al., 1997; Klagsbrun, 1992; Schlessinger et al., 2000). When combined with nanogram quantities of these two angiogenic growth factors, our heparin-presenting nanofibre gel matrix led to extensive vascularization in a rat corneal assay (Rajangam et al., 2006). Further studies have found that this heparin-binding peptide amphiphile system demonstrates excellent biocompatibility and leads to the generation of a de novo vascularized tissue when transplanted in vivo (Ghanaati et al., 2009).

Given the therapeutic efficacy of the paracrine effect observed with stem cells in the myocardium, we set forth to investigate the effectiveness of our synthetic heparin-presenting nanofibre networks to capture and deliver paracrine factors therapeutically. Using factors derived from conditioned stem cell media, we aimed to evaluate our paracrine factor-loaded materials for efficacy in an ischaemia–reperfusion mouse myocardial infarction model. Our hope was to demonstrate that our engineered nanofibre delivery system could elicit a paracrine effect, frequently cited as the means of efficacy for stem cell therapies, within the myocardium. Moving forward, we aimed to then replicate this effect using a defined combination of recombinant factors to produce an acellular, synthetic therapy harnessing this paracrine effect.

2. Materials and methods

2.1. Synthesis and purification of heparin-binding peptide amphiphile

Heparin-binding peptide amphiphile (HBPA: Figure 1A) was synthesized as previously described (Ghanaati et al., 2009; Rajangam et al., 2006). HBPA was purified by standard reverse-phase HPLC and trifluoroacetate counter-ions were exchanged using 0.1 M hydrochloric acid, yielding the chloride salt of the peptide upon lyophilization.

Figure 1.

Chemical structure of heparin-binding peptide amphiphile (A), which self-assembles through interaction with heparin, allowing for biomimetic capture of heparin-binding paracrine factors (B). Using assembled HBPA gels, we found that the materials were able to capture fluorescently tagged bFGF from solution (C). A non-linear least squares equation, assuming a single binding site, was fitted to the data, with an R2 value of 0.97

2.2. Conditioned media preparation and manipulation

Primary mouse mesenchymal stem cells were graciously provided by Dr Christoff Westenfelder (University of Utah), having been isolated according to previously reported methods (Togel et al., 2005a, 2005b, 2008). Briefly, multipotent mesenchymal stromal cells (MSCs) were obtained from the bone marrow of anaesthetized C57/Bl6 mice (Charles River) by flushing both femurs with normal saline. The obtained cell suspensions were washed and plated in T25 flasks (Corning) containing α-MEM and preselected 10% fetal bovine serum (GIBCO), pH 7.40. Non-adherent cells were removed and adherent cells were detached (trypsin–EDTA) after 3 days, replated under identical conditions and passaged at subconfluence. MSCs were characterized by demonstrating positive FACS staining for CD44, CD90 and CD105 and negative staining for CD34 and CD45 and, in addition, their ability to differentiate into adipocytes, osteocytes and chondrocytes.

Conditioned medium (CM) was prepared as follows: 90% confluent primary mouse MSCs were incubated in serum-free DMEM for 18 h under hypoxic conditions—5% O2, 5% CO2 and 90% N2. After the hypoxic stress, the conditioned medium was removed from the cells. CM depleted of heparin-binding factors was prepared using heparin–sepharose chromatography (HS dCM), using a heparin–sepharose solid support (CL-6B, Amersham) and soaking the beads in medium for 24 h at 4 °C. Subsequently, the medium was further depleted using heparin–agarose chromatography (HS/HA dCM), following the same protocol but using heparin immobilized on 4% beaded agarose type I (Sigma). Separately, the complete CM was partitioned, based on size, using centrifugal filtration with 30 kDa and 10 kDa MW cut-off membranes (Centriprep YM), and concentration changes were adjusted using serum-free DMEM. This produced three size-based fractions of the original CM: factors > 30 kDa (pool 1 CM); factors of 10–30 kDa (pool 2 CM); and factors < 10 kDa (pool 3 CM). These size ranges were verified using SDS–PAGE.

2.3. Factor loading of HBPA networks

A solution of HBPA (3% w/v) was gelled by mixing with an equal volume of a solution of heparin sodium salt derived from porcine intestinal mucosa (2% w/v; Sigma-Aldrich). The resulting gel was soaked in 1 ml of either hypoxic conditioned medium or one of the various manipulations of this medium previously described, at 4 °C for 24 h. In the case where heparin was not used, disodium hydrogen phosphate (1.1% w/v) was mixed with the same concentration of HBPA to form a gel. Following 24 h incubation, the HBPA nanofibre gels were used for in vivo studies. Additionally, as a proof of concept, recombinant human bFGF (Peprotech) was tagged with rhodamine, suspended in PBS to a concentration of 40 nM and used to soak a preformed HBPA/heparin gel by the same methodology as that used to prepare materials for in vivo studies. The concentration of the medium phase was monitored through fluorescence (Ex/Em, 544/576). For in vivo studies using recombinant VEGF165 and bFGF (Peprotech), 1 ng each factor/µl gel was mixed with the HBPA/heparin material and injected immediately.

2.4. Mouse ischaemia-reperfusion infarction model

Female ICR (Harlan) mice, aged 3 months and weighing approximately 25 g, were anaesthetized with ketamine (100 mg/kg, administered via intraperitoneal injection). Following tracheal intubation, the heart was exposed and the left anterior descending (LAD) artery was temporarily ligated with a suture for 30 min to induce infarction. This model was predetermined to have an infarct size that was 44.6 ± 4.6% of the area at risk, with an area at risk consisting of 43.4 ± 3.2% of the left ventricle (n = 5). After 30 min of ligation, 10 µl preformed HBPA gel, having been soaked in various conditioned medium samples, was injected into the left ventricle (LV) wall and the tissue was reperfused. The consistency of the gel was such that it could be delivered via a small-volume Hamilton syringe with a 26 gauge needle. Additionally, control surgeries were performed, with animals receiving a saline injection following infarct. As a control for HBPA retention in the tissue, a PA conjugated to rhodamine was injected into the LV wall and found to persist for 30 days, evidenced by fluorescent signal in the sectioned tissue at this time. Animals were anaesthetized and phenotyped for functional performance at 30 days, using a pressure–volume catheter (Millar) inserted into the left ventricle through the right carotid artery, providing functional approximations of systolic blood pressure and left ventricular contractility (+dP/dt) and relaxation (−dP/dt). In total, eight animals were used for each of the 14 conditions, amounting to 112 animals in total. For each group, the number of animals to survive the surgery and subsequent follow-up was at least five. These studies were approved by the Northwestern University Animal Care and Use Committee.

2.5. Ischaemic hind limb rat model

Female Sprague–Dawley rats (Harlan) were used, aged 3 months and weighing approximately 280 g. The animals were anaesthetized with ketamine (100 mg/kg administered via intraperiotoneal injection). The left femoral artery was ligated and excised. The factor-loaded nanofibre gel was administered via intramuscular injection of 15 µl gel to the adductor magnus and gracilis anticus and 10 µl gel injected into the pectineus and adductor longus, for a total injection volume of 50 µl gel/animal. The animals were phenotyped at 30 days through microcatheter-facilitated barium-enhanced angiography (General Electric OEC 9800), through surgical preparation and administration via the distal abdominal aorta. The resulting images were quantified using a standard angioscoring method by a blinded evaluator, whereby a 9 × 5 grid was transposed onto the angiogram (Adobe Photoshop) in the femoral region and the number of times a blood vessel intersected the grid was counted. This value was divided by the total number of grid lines to obtain the final angioscore. In total, six or seven rats were used for each of the treatment groups, amounting to a total of 38 animals. All animals survived surgery and remained alive until the end of the study. These studies were approved by the Northwestern University Animal Care and Use Committee.

2.6. Statistics and data analysis

All error bars for graphs and reported values are standard deviations (SD). Significance between all groups was determined simultaneously, using a one-way analysis of variance (ANOVA) with a Bonferroni multiple comparisons post hoc test.

3. Results

3.1. Factor-binding capacity of HBPA

To test the ability of this material to extract and bind factors from a medium phase, we examined the concentration of dye-labelled bFGF in the medium used to soak the HBPA gel. As shown in Figure 1C, 58.5% of the labelled bFGF was adsorbed to the assembled HBPA nanofibre gel. Considering that the HBPA gel volume accounted for only 2% of the total volume of medium in this experiment, this is substantially greater than could be attributed to passive diffusion into the gel volume. The capacity of the gel, indicated by the saturation effect observed, is approximately 1 pmol bFGF/µl gel. We previously found that, once bound to the assembled HBPA nanofibre gel, a prolonged bFGF release occurs over a period of days (Rajangam et al., 2006).

3.2. CM-loaded HBPA preserves function post-MI

Our HBPA nanofibre gel was soaked in various media to bind factors and then injected into the LV following acute myocardial injury, using an ischaemia–reperfusion mouse model. Preliminary studies found that fluorescent conjugates of this material remained in the LV tissue for at least 30 days. Haemodynamic analysis, the most accurate measure of cardiac performance, reveals that the group treated with CM-loaded HBPA nanofibres exhibited significantly (p < 0.01) improved LV contractility (7804 ± 995 mmHg/s, Figure 2A) compared to that of an untreated group (4948 ± 953 mmHg/s). For this CM-loaded nanofibre group, LV relaxation (5800 ± 438 mmHg/s, Figure 2B) was also significantly (p < 0.05) greater than that of the untreated group (4630 ± 744 mmHg/s). On the basis of average systolic blood pressure (Figure 2C), once again there was a significant difference (p < 0.01) between the group receiving treatment with CM-loaded HBPA (82.2 ± 4.8 mmHg) and the untreated group (70.6 ± 5.2 mmHg). Strikingly, when compared to a healthy animal subjected to no procedure, giving a healthy baseline, the LV contractility of the CM-loaded HBPA group was not significantly different from that of this healthy group (9422 ± 1420 mmHg/s). However, when comparing our CM-loaded group to the healthy baseline on the basis of LV relaxation (7108 ± 745 mmHg/s, p < 0.01) and systolic blood pressure (93.5 ± 3.9 mmHg, p < 0.001), the difference was significant. While the CM-loaded HBPA does not completely preserve function equivalent to the healthy baseline level, it offers significant functional improvement when compared to an animal receiving no treatment following infarction. Nanofibres soaked in unconditioned medium (unCM) showed no functional improvement relative to untreated animals on the basis of any of the functional performance measurements. The same is true for animals receiving an injection of hypoxic CM alone without the HBPA nanofibre gel. Thus, the effect seen through the combination of hypoxic CM with our HBPA material is not obtained with either component alone.

Figure 2.

Functional analysis of mice 30 days following acute MI, showing left ventricular contractility (A), left ventricular relaxation (B) and systolic blood pressure (C). Mice receiving treatment with (left to right): HBPA loaded with hypoxic CM (n = 5); HBPA loaded with unconditioned medium (unCM, n = 6); and conditioned medium only (CM, n = 5). This is compared to a control group receiving no treatment following MI (n = 5) and to a healthy animal serving as a baseline control (n = 8). Significance is shown relative to the HBPA + CM group, and error bars indicate SD

3.3. Crucial factors are heparin-binding

Using heparin chromatography, the hypoxic CM was depleted of heparin-binding components in order to assess the importance of heparin to the observed therapeutic effect of the nanofibre gel and, moreover, to determine whether this effect could be depleted. As shown in Figure 3, soaking HBPA nanofibre gels in heparin–sepharose-depleted CM (HS dCM) resulted in LV contractility (6389 ± 1103 mmHg/s) less than that observed using non-depleted CM in combination with HBPA, although the decrease was not significant. When this heparin–sepharose medium was further depleted using a heparin–agarose (HS/HA dCM) solid support, the resulting LV contractility (4988 ± 576 mmHg/s) was significantly less (p < 0.001) than the treatment using complete hypoxic CM combined with HBPA, but not significantly different from an untreated control. When the HBPA molecule was assembled into nanofibre networks through the addition of phosphate ions (in place of the negatively-charged heparin) and soaked in complete hypoxic CM, the resulting LV contractility (5577 ± 537 mmHg/s) was significantly less (p < 0.01) than the treatment utilizing complete hypoxic CM with HBPA, but not significantly different from an animal receiving an infarction with no additional treatment.

Figure 3.

Left ventricular contractility at 30 days following acute MI, evaluating the importance of heparin. Treatments with material loaded with (left to right): complete hypoxic CM (n = 5); hypoxic CM depleted by heparin–sepharose chromatography (n = 5); hypoxic CM depleted by heparin–sepharose and heparin–agarose chromatography (n = 5); and HBPA assembled with phosphate instead of heparin (n = 5) and loaded with complete hypoxic CM. Dashed lines represent contractility values for healthy animals as a baseline comparison (small dashes) and animals induced with infarction but receiving no further treatment (large dashes). Significance is shown relative to the HBPA + CM group, and error bars indicate SD. The data for HBPA + CM, healthy and no-treatment groups are the same as those in Figure 2, and are displayed to assist with comparisons between groups

3.4. Crucial factors are primarily < 30 kDa

Through molecular weight-based fractionation, the hypoxic CM was divided into pools of different molecular weight ranges (Figure 4). When HBPA was loaded with medium containing only proteins > 30 kDa (pool 1 CM), the resulting LV contractility (5163 ± 605 mmHg/s) was significantly (p < 0.001) less than when the complete medium was used. In contrast, the pool containing proteins in the range 10–30 kDa (pool 2 CM, 7186 ± 748 mmHg/s) and the pool containing proteins < 10 kDa (pool 3 CM, 6592 ± 705 mmHg/s) exhibited contractility that was not significantly less than that for the complete hypoxic CM. When compared to the untreated control, both pool 2 (p < 0.01) and pool 3 (p < 0.05) showed significantly enhanced function.

Figure 4.

Left ventricular contractility at 30 days following acute MI, evaluating different molecular weight fractions of hypoxic conditioned media. Treatment with HBPA loaded with (left to right): complete hypoxic CM; fractioned hypoxic CM > 30 kDa (pool 1 CM, n = 5); fractioned hypoxic CM in the range 10–30 kDa (pool 2 CM, n = 5); and fractioned hypoxic CM < 10 kDa (pool 3 CM, n = 6). Dashed lines represent contractility values for healthy animals as a baseline control (small dashes) and untreated animals (large dashes). Displayed p values are relative to the HBPA + CM group, and error bars indicate SD. *, significant enhancement compared to the untreated controls (large dashes); *p < 0.05; **p < 0.01. Neither group differed significantly from the HBPA + CM group. The data for HBPA + CM, healthy and no-treatment groups are the same as those found in Figure 2, and are displayed to assist with comparisons between groups

3.5. VEGF and bFGF combined with HBPA preserves function

To begin to mimic the paracrine effect observed, recombinant VEGF and bFGF were added to our HBPA nanofibre gels (Figure 5). These factors were found to be present in the hypoxic CM using Western blotting (see Supporting information, Data supplement). When the HBPA was mixed with 10 ng of each of these factors, the resulting LV contractility (7399 ± 1233 mmHg/s) was significantly greater (p < 0.01) than that for the untreated control. Additionally, this value was significantly greater than that for the HBPA alone (4713 ± 1293 mmHg/s, p < 0.01) and also for growth factors alone (5382 ± 456 mmHg/s, p < 0.05), showing that function is preserved using HBPA in combination with growth factors. This was not observed when either HBPA or growth factors were used separately. All groups examined differed significantly from the function observed for our baseline group of healthy animals.

Figure 5.

Functional analysis of mice 30 days following acute MI, showing left ventricular contractility (A), left ventricular relaxation (B) and systolic blood pressure (C) for therapies using recombinant VEGF and bFGF. Mice receiving treatment with (left to right): HBPA loaded with VEGF and bFGF (n = 8); HBPA only (n = 5); and VEGF and bFGF (GF, n = 5). This is compared to a control group receiving no treatment following MI (n = 5) and to a healthy animal (n = 8). Significance is shown relative to the HBPA + GF group, and error bars indicate SD. The data for healthy and no treatment control groups are the same as those found in Figure 2, and are displayed to assist with comparisons between groups

3.6. Nanofibre gel applied to ischaemic hind limb

To investigate the therapeutic potential of our factor-loaded nanofibre gels in a different tissue site, we implemented a rat hind-limb ischaemia model (Figure 6). Treatment with hypoxic CM-loaded HBPA results in an angioscore (1.38 ± 0.13) significantly higher than that for an untreated group receiving a vehicle injection (0.81 ± 0.12, p < 0.001) or unloaded HBPA alone (1.12 ± 0.05, p < 0.05). Our material loaded with recombinant VEGF and bFGF (1.20 ± 0.15) was not significantly less than the hypoxic CM-loaded material, although some decrease was evident. The recombinant factor-loaded material produced an angioscore significantly greater than that for treatment with recombinant factors alone (0.92 ± 0.08, p < 0.01) and was also significantly greater (p < 0.001) than the untreated control. Interestingly, there was no significant difference between the recombinant factor-loaded HBPA and the unloaded HBPA, and the angioscore for HBPA alone was significantly greater (p < 0.01) than the group subjected to ischaemia with no additional treatment. When compared to the baseline level of a healthy control (1.19 ± 0.22), the material loaded with hypoxic CM did not represent a significant increase, while the recombinant factor-loaded HBPA was essentially the same as that obtained for healthy animals.

Figure 6.

Loaded materials applied to an ischaemic hind-limb model, showing angiography 30 days following ischaemic injury and material administration. Treatment groups evaluated: HBPA loaded with hypoxic CM (A, n = 6); HBPA loaded with VEGF and bFGF (B, n = 7); HBPA (C, n = 6); and recombinant VEGF and bFGF alone (D, n = 6). These were compared to an untreated control (E, n = 7) and a healthy animal (F, n = 6). Angiograms were quantified using an angioscoring method (G). Significance is shown relative to the HBPA + CM group, and error bars indicate SD

4. Discussion

Several clinical trials have demonstrated the beneficial effect of bone marrow-derived stem cells in ischaemic myocardium, although there remains speculation over the therapeutic mechanism. As described, this has led to several proposed mechanisms, including transdifferentiation into cardiomyocytes, lineage specific differentiation into endovascular cells, or paracrine effects. Although there is debate surrounding transdifferentiaton (Murry et al., 2004; Sussman and Murry, 2008), the preponderance of evidence suggests that transdifferenation and cell fusion are at most rare events and unlikely to be functionally significant (Nygren et al., 2004, 2008). Likewise, neovascularization of ischaemic myocardium and regeneration of vascular endothelium by stem cells is debated, and may not occur to an extent that can lead to the large increases in myocardial performance observed after stem cell therapy following MI (Jackson et al., 2001; Kocher et al., 2001). Given the growing evidence for paracrine effects being the primary mode of stem cell efficacy (Gnecchi et al., 2008), we set forth to develop a material that could capture and deliver paracrine factors produced by stem cells. In this way, we would be able to harness the paracrine effect using a synthetic delivery method, eventually translating this into a system that could mirror the effectiveness of cell therapies using a ‘cell-free’ synthetic material coupled with recombinant factors.

Hypoxic conditioned medium was derived from mesenchymal stem cells, which are among several types of stem cells currently being explored clinically for the treatment of ischaemic heart disease (Burt et al., 2008; Chen et al., 2004). Using a mouse ischaemia–reperfusion model of acute myocardial infarction, we measured the actual haemodynamic performance of the left ventricle, the truest measure of myocardial function. Our hypoxic CM-loaded HBPA material led to significantly better haemodynamic performance at 30 days following infarct when compared to an untreated group. Additionally, animals receiving CM-HBPA treatment did not vary significantly from healthy animals on the basis of LV contractility, the most crucial of our measures of performance, although some loss of function was evident in the other haemodynamic parameters evaluated. Moreover, we found that treatment with the CM-loaded HBPA nanofibre networks resulted in significantly greater myocardial functional performance than did treatment with the material alone or with medium alone, neither of which differed from a group receiving no treatment following infarction. This indicates a synergistic effect when the HBPA is soaked in the hypoxic CM and subsequently injected into the myocardium. This effect was not captured through either the material or paracrine factors in isolation and is likely due to the heparin-containing nanofibre gel concentrating heparin-binding factors, stabilizing factor-receptor complexes, prolonging factor activity by suppressing proteolysis, and increasing retention within the tissue through the introduction of supramolecular aggregates, rather than isolated proteins.

The importance of the heparin component to the activity of the nanofibre gel system in vivo is not surprising, given the potency of heparin-binding components in the hypoxic CM. When the hypoxic CM is depleted of heparin-binding components through heparin solid support chromatography, the overall effectiveness of the medium-loaded HBPA is diminished. With heparin–sepharose chromatography, the medium was partially depleted, as some effect was still observed. However, when this medium was further depleted using a heparin–agarose solid support, the effect of the CM-loaded HBPA was completely lost, with the resulting phenotype no different from an animal receiving no treatment following infarction. The phenotype of the treated animals indicates different levels of depletion, depending on the methods of chromatography, possibly due to heparin content or varying orientation of the covalently linked heparin oligomers to the solid supports. This depletion was verified by Western blotting (see Supporting information, Data supplement). This demonstrates the ability to scale the observed phenotypic response depending on the extent of depletion, indicating a dose–response relationship. Furthermore, using a gelled peptide nanofibre construct that did not present heparin, it was found that while proteins still adsorbed to the gel, no significant therapeutic benefit was observed in vivo. This highlights the importance of heparin in the HBPA system to bind and present these paracrine factors in a biologically relevant way, as the crucial factors are heparin binding and simply delivering proteins from a fibrous gel is not sufficient to elicit any response. A control of heparin alone without HBPA was not included in this study, as this treatment proved fatal due to resultant myocardial bleeding. This effect is consistent with the known activity of heparin as a potent anticoagulant and vasodilator. It is interesting to note that this same effect was not observed for heparin immobilized on the surface of HBPA nanofibres.

To begin to identify some of the factors contained in the hypoxic medium, we used Western blotting (see Supporting information, Data supplement) to probe for the presence of VEGF165 and bFGF, potent regulators of angiogenesis and cellular behaviour that are known to be heparin-binding and may also have cardioprotective attributes (Caplan and Dennis, 2006; Gnecchi et al., 2008). VEGF and bFGF have been found in increased concentrations in the post-infarct myocardium following stem cell transplant (Nagaya et al., 2005; Yoon et al., 2005). Also, VEGF gene therapy has been applied to treat myocardial ischaemia (Losordo et al., 1998) and bFGF has been found to enhance myocardial function when administered following ischaemia in an angiogenesis-independent role (Jiang et al., 2004). Further, we have previous experience using these two factors with our HBPA system (Rajangam et al., 2006). We initiated our recombinant studies by loading VEGF and bFGF directly onto the HBPA material, because of the known potency of both of these factors, their demonstrated presence in our CM samples and our past success using these factors with HBPA. Although VEGF165 is 38.2 kDa, there exist other isoforms of VEGF that are < 30 kDa, a molecular weight criterion suggested by our size fractionation studies. Additionally, the size fractionation technique used to establish this criterion is more dependent on protein structure and hydrodynamic radius than the corresponding molecular weight, so the suggestion of a 30 kDa limit for effective factors is likely somewhat flexible. Western blotting data for the three molecular weight fractions was inconclusive, so while we can verify the presence of VEGF in the whole conditioned medium, which molecular weight-based pool it is found in is not known. Excitingly, we were able to recapitulate the observed conditioned medium effect, preserving significant function though the combination of our HBPA nanofibre networks with nanogram quantities of these two growth factors—an effect that was not obtained from the growth factors alone. Perhaps the most surprising synergy between the nanofibre gel and growth factors was the very small amount of growth factors needed, 10 ng VEGF and bFGF/animal. Our material potentiates the effect of these factors, likely through heparin-dependent activity, as previously discussed, since the factors alone have no effect.

The mechanism by which paracrine factors, presented using our HBPA nanofibre networks, augment cardiac function is not known. Histology failed to reveal differences in infarct size or capillary density in treated vs. untreated animals, suggesting that the mechanism is not directly linked to enhanced angiogenesis or a reduction in fibrosis. Additionally, there was no difference in LV remodelling. While we were initially surprised that capillary density was not augmented by the HBPA combined with angiogenic factors, it was in keeping with studies demonstrating that the failing heart is not O2-limited (Gong et al., 2003; Murakami et al., 1999). Given our observations, we propose that the paracrine factors delivered with our HBPA nanofibre elicit trophic effects on the myocardium after infarct and can augment function by modulation of cardiac metabolism. Although heart failure is multifactorial (Mann and Bristow, 2005), there is substantial evidence suggesting that the failing heart is devoid of fuel and that modulation of cardiac metabolism can lead to marked functional improvement (Ingwall and Weiss, 2004; Neubauer, 2007). There is also clinical evidence of delayed recovery of regional function after reperfusion in stunned myocardium (Dilsizian et al., 2005). We hypothesize that the delivery of paracrine factors helps erase this ischaemic memory and helps tap the cardiometabolic reserve (Taegtmeyer, 2004). Glucagon-like peptide 1 has been shown to increase glucose uptake and improve LV performance in dogs with pacing-induced dilated cardiomyopathy, so the paradigm for paracrine factors to augment cardiac metabolism, and subsequently cardiac function, has been previously established (Nikolaidis et al., 2004). In order to explore this mechanism, further studies using 1H- and 31P-NMR and other biochemical methods will be performed to examine the absolute levels and ratio of phosphocreatine (PCr) to creatine (Cr) in treated vs. untreated groups.

HBPA loaded with paracrine factors from hypoxic CM also demonstrated efficacy in an ischaemic model within a different tissue site, increasing arteriogenesis in a critical hind limb ischaemia model. Angiography revealed a large number of collateral blood vessels formed in response to the hypoxic CM-loaded HBPA group relative to groups receiving no treatment following induction of ischaemia. This group also exhibited higher blood vessel density via angiography than did a healthy control, which should not be surprising, given that the healthy animals had no injury and thus less impetus for collateral formation. In the myocardial infarction model, neither the HBPA alone nor recombinant growth factors alone were sufficient to improve function. However, in the hind limb ischaemia model, the HBPA alone did significantly enhance vasculature relative to the untreated group, which was not observed using growth factors alone. It is possible that, in skeletal muscle, the material is able to bind and potentiate endogenous factors that are released in response to this ischaemic injury, supported by the insignificant increase when small quantities of factors are added to the HBPA compared to using only the HBPA alone.

Our studies to date using HBPA nanofibre networks indicate a marked ability to augment function and healing in ischaemic tissues. The mechanisms, while closely linked to paracrine effects, differ in the two tissue sites examined here and are congruent with the known regenerative capacity of each tissue. In the ischaemic hind limb, HBPA alone has some effect on arteriogenesis, likely by accelerating a naturally occurring collateral formation mechanism, but maximal effect is seen with the paracrine factor-loaded HBPA. In the heart, which has one of the most limited regenerative capacities in the body, only the paracrine factor-loaded HBPA has an effect on function, which we postulate to be via trophic modulation of cardiac metabolism. In this capacity, our HBPA nanofibre system serves as a synthetic extracellular matrix and biomimetic delivery vehicle for paracrine factors. Its use in these studies demonstrates the marked biological activity of the paracrine factors released by stem cells, coupled with our material to augment function after an MI or enhance vasculature following critical ischaemia. Since the paracrine factor-loaded HBPA can induce a paracrine effect resembling that previously demonstrated for stem cells on the heart, we propose that this synthetic recombinant system may be used to treat patients with acute MI as well as chronic heart failure, eliminating the real complications of rejection and tumourigenesis seen with stem cell therapies.


The authors gratefully acknowledge funding support from the US Army Telemedicine and Advanced Technology Research Center (TATRC; Grant No. W81XWH-05-1-0381), awarded to S.I.S. Additionally, J.W.L. and S.I.S. acknowledge assistance provided through the Northwestern University Senyei Translational Research Award. M.J.W. is supported by the Northwestern Regenerative Medicine Training Program (RMTP), NIH Award No. 5T90-DA022881. We thank Mark Seniw for his assistance with molecular graphics. Antibodies for Western blotting were provided by Dr Susan Crawford of the Department of Pathology, Northwestern University. Mesenchymal stem cells were provided by Dr Christoff Westenfelder, University of Utah.

Supporting information

The following supporting information is available in the online version of this article:

  • Western blotting

  • Hypoxic CM contains VEGF and bFGF

  • Figure S1. Western blotting for (A) VEGF and (B) bFGF for the various media manipulations used in this study