Therapeutic microparticles functionalized with biomimetic cardiac stem cell membranes and secretome

Stem cell therapy represents a promising strategy in regenerative medicine. However, cells need to be carefully preserved and processed before usage. In addition, cell transplantation carries immunogenicity and/or tumourigenicity risks. Mounting lines of evidence indicate that stem cells exert their beneficial effects mainly through secretion (of regenerative factors) and membrane-based cell–cell interaction with the injured cells. Here, we fabricate a synthetic cell-mimicking microparticle (CMMP) that recapitulates stem cell functions in tissue repair. CMMPs carry similar secreted proteins and membranes as genuine cardiac stem cells do. In a mouse model of myocardial infarction, injection of CMMPs leads to the preservation of viable myocardium and augmentation of cardiac functions similar to cardiac stem cell therapy. CMMPs (derived from human cells) do not stimulate T-cell infiltration in immuno-competent mice. In conclusion, CMMPs act as ‘synthetic stem cells’ which mimic the paracrine and biointerfacing activities of natural stem cells in therapeutic cardiac regeneration.

M ultiple types of adult stem cells, such as mesenchymal stem cells, cardiac stem cells (CSCs), and endothelial progenitor cells have entered clinical investigations worldwide [1][2][3][4][5][6] . Differentiation of injected cells into the host tissues has been reported. However, these sporadic events could not explain the therapeutic benefits seen in animal models and human trials 7,8 . Later on, the field realized that one important mode of therapeutic action is the secretion of paracrine factors by injected stem cells that act like 'mini-drug pumps' to promote endogenous repair 9,10 . Moreover, stem cell membranes are not null in the repair process: contact with the injected stem cells triggers intracellular protective/regenerative pathways in the host cells 11,12 . On the basis of these two aspects, we proposed a 'coreshell' design of a therapeutic microparticle (MP) which mimicked  2 have similar sizes to those of CSCs. n ¼ 3 for each group. (i) CMMPs and Control MP 2 carried similar surface antigens as CSCs did. n ¼ 3 for each group. (j-l) Similar release profile of CSC factors (namely vascular endothelial growth factor (VEGF), insulin-like growth factor (IGF)-1 and hepatocyte growth factor (HGF)) was observed in CMMPs and Control MP 1 , indicating membrane cloaking did not affect the release of CSC factors from CMMPs and Control MP 1 . n ¼ 3 for each time point. All data are mean ± s.d. Comparisons between any two groups were performed using two-tailed unpaired Student's t-test. Comparisons among more than two groups were performed using one-way ANOVA followed by post hoc Bonferroni test.
stem cell biointerfacing during regeneration. This particle, named cell-mimicking MP (CMMP), contained control-released stem cell factors in its polymeric core and was cloaked with stem cell membrane fragments on the surface. Our hypothesis is that CMMP can exert similar regenerative outcomes as real CSCs but are superior to the later since they are more stable during storage and do not stimulate T-cell immune reaction since they are not real cells.
In the present study, we report for the first time a poylmer MP which emulates CSC functions during tissue repair. In a mouse model of myocardial infarction, injection of CMMPs led to preservation of viable myocardium and augmentation of cardiac functions similar to CSC therapy. CMMPs (derived from human cells) did not stimulate T-cell infiltration in immuno-competent mice, suggesting their excellent safety profile. Although our first application targeted the heart, the CMMP strategy represents a platform technology that can be applied to multiple stem cell types and the repair of various organ systems.

Results
Physiochemical and biological properties of CMMPs. The biochemical design and work model of CMMPs were outlined in Fig. 1a. Briefly, Control MP 1 were fabricated from poly(lactic-coglycolic acid) (PLGA) and conditioned media of human CSCs which were isolated from human hearts using the cardiosphere method as previously described 13,14 (Supplementary Fig. 1). The conditioned media contains various growth factors secreted by CSCs 10 . CSCs have been tested and proven safe and effective in Phase I/II clinical trials [1][2][3] . After that, MPs (Texas red succinimidyl ester-labelled; Fig. 1b, red) were cloaked with the membrane fragments of CSCs (green fluorescent DiO-labelled; Fig. 1b, green) to become the final product CMMP (Fig. 1c, red particle with green coat). Fluorescent imaging revealed there is no specific DiO outer layer fluorescence on Texas red succinimidyl ester-labelled MPs (Control MP 1 ) after 30 min co-culture ( Supplementary Fig. 2). Scanning electron microscopy (SEM) revealed the effective CSC membrane cloaking on CMMPs ( Fig. 1e) but not on non-cloaked MPs (Control MP 1 ; Fig. 1d). As another control particle, Control MP 2 was fabricated by cloaking empty PLGA particles with CSC membranes. We fabricated CMMPs, Control MP 1 and Control MP 2 with sizes similar to those of real CSCs (Fig. 1h). As an indicator of successful membrane cloaking, flow cytometry analysis confirmed the expression of major human CSC markers (for example, CD105, CD90) on CMMPs and Control MP 2 but not on Control MP 1 (Figs 1f,g and 2). Overall, both CMMPs and Control MP 2 carried similar surface antigens as CSCs did (Fig. 1i). Membrane cloaking did not affect the release of CSC factors (namely vascular endothelial growth factor, insulin-like growth factor-1 and hepatocyte growth factor) from CMMPs and Control MP 1 (   Fig. 3b) increased the numbers of NRCMs as compared with those from Control MP 2 (pink bar, Fig. 3b) or solitary NRCM culture (white bar, Fig. 3b), the greatest NRCM numbers were seen in those cocultured with CMMPs (green bar, Fig. 3b) and genuine CSCs (blue bar, Fig. 3b). Furthermore, CMMPs and Control MP 1 robustly promoted NRCM contractility (Fig. 3c) and proliferation (as indicated by Ki67-positive nuclei, Fig. 3d). Both CMMPs and Control MP 2 could firmly bind to cardiomyocytes, as cells did, while most non-cloaked Control MP 1 floated in the media (Fig. 3e). Such binding was confirmed by CMMPs' synchronized movement with adjacent beating cardiomyocytes (   increased the numbers of NRCMs as compared with those from Control MP 2 (pink bar) or solitary NRCM culture (white bar), but the greatest NRCM numbers were seen in those co-cultured with CMMPs (green bar) and genuine CSCs (blue bar). n ¼ 5 for each group. (c) Higher NRCM contractility was seen in those cultured with CMMPs (green bar) and CSCs (blue bar) compared with those cultured with Control MP 1 (red bar). n ¼ 5 for each group. (d) Representative images and quantitative analysis of NRCMs stained with alpha sarcomeric actin (green) and proliferation marker Ki67 (red), treated with Control MP 1 , Control MP 2 , CMMPs or CSC. n ¼ 5 for each group. Scale bar, 50 mm. (e) Representative images and quantitative analysis of CMMP (red) or Control MP 1 (red), Control MP 2 (red) binding to NRCMs (green). n ¼ 3 for each group. Scale bar, 50 mm. (f) Representative movie screenshots and quantitation of Control MP 1 's and CMMP's synchronized movement with adjacent beating cardiomyocytes. n ¼ 3 for each group. Scale bar, 50 mm. (g,h) Time-lapse imaging revealed the rolling (g) and travelling (h) of CMMPs on attached cardiomyocytes. Yellow arrows indicated the rolling or moving directions. Scale bar, 20 mm. All data are mean±s.d. * indicates Po0.05 when compared with Control group; # indicates Po0.05 when compared with Control MP 1 group; & indicated Po0.05 when compared with Control MP 2 . Comparisons between any two groups were performed using two-tailed unpaired Student's t-test. Comparisons among more than two groups were performed using one-way ANOVA followed by post hoc Bonferroni test.
CMMP therapy in immunodeficient mice with heart attack. To test the therapeutic potential of CMMPs, we employed a mouse model of myocardial infarction (heart attack) by left anterior descending artery (LAD) ligation (Fig. 4a). CMMPs or Control MP 1 were intramyocardially injected immediately after LAD ligation. Negative or positive control animals received injection of vehicle (PBS) or CSCs, respectively. Ex vivo fluorescent imaging at Day 3 revealed that more CMMPs were retained in the heart after injection than Control MP 1 (Fig. 4b) were. This was further confirmed by histology (Fig. 4c). This was consistent with CMMP's superior binding to cardiomyocytes in vitro (as seen in Fig. 3). In addition, ex vivo fluorescent imaging indicated that the majority of CMMPs remained in the heart after injection, while 'washed away' CMMP signal could be found in the lung and the liver ( Supplementary Fig. 5), consistent with the notion that the needle injection can cause vessel damage and the venous drainage brings the particles to the lungs 15 . The off-target expression in the liver may represent the leakage of CMMPs into the LV cavity during injection. Nevertheless, the majority of CMMPs remain in the heart after injection.
In vivo degradation of CMMPs was evident as only a negligible amount of particles remained in the heart at Day 28 ( Supplementary Fig. 6). A cohort of animals was killed at Day 7 for assessment of myocardial tissue apoptosis and infiltration of macrophages in CMMP-treated animals. TdT-mediated dUTP nick end labelling (TUNEL) staining revealed the anti-apoptosis effects of CMMP: less apoptotic nuclei were detected in areas with the presence of CMMPs (green nuclei, Fig. 4d). CMMP treatment did not cause the exacerbation of inflammation: the tissue densities of CD45-positive cells were indistinguishable in areas with or without CMMPs (Fig. 4e). Masson's trichrome staining 4 weeks after treatment ( Fig. 4f; red ¼ healthy myocardium and blue ¼ scar tissue) revealed Control MP 1 treatment (red bars, Fig. 4g-i) exhibited a certain degree of heart morphology protection compared with Control PBS injections (white bars, Fig. 4g-i). However, the greatest protective effects were seen in the animals treated with CMMPs (green bars, Fig. 4g-i). Such protective effects were similar to those injected with CSCs (blue bars, Fig. 4g-i). The bona fide efficacy indicator for stem cell therapy is the ability to ameliorate ventricular dysfunction or even boost cardiac function over time, gauged by echocardiography. Left ventricular ejection fractions (LVEFs) were measured at baseline (4 h post infarct) and 4 weeks afterwards. LVEFs were indistinguishable at baseline for all groups (Fig. 4j), indicating a similar degree of initial heart injury. Over the 4 week period, the LVEFs in control (PBS or saline)-treated animals continued deteriorating (white bar, Fig. 4k) while the Control MP 1 -treated animals exhibited a trend of LVEF augmentation (red bar, Fig. 4k) but did not reach statistical significance. CMMP treatment robustly boosted cardiac function with the highest LVEFs at 4 weeks (green bar, Fig. 4k). Such treatment effects were indistinguishable from those of CSC treatment with real CSCs (blue bars, Fig. 4k). Histological analysis indicated that such functional benefits by CMMP treatment were accompanied by remuscularization (Fig. 5a), proliferation of endogenous cardiomyocytes (Fig. 5b), augmentation of blood flow (Fig. 5c), and increase of vessel density (Fig. 5d) in the post-MI heart.
CMMP injection does not promote T-cell infiltration in normal mice. To evaluate the local T-cell immune response to CMMPs, immune-competent CD1 mice were intramyocardially injected with human CSCs or CMMPs. Animals were killed 7 days after injection for assessment of immune rejection in the heart, as gauged by CD3 þ and CD8 þ T cell infiltration (Fig. 6a). CMMP (red) injection elicits negligible T-cell rejection as very few CD3 þ (green) or CD8 þ (green) T cells were detected in the heart (Fig. 6c,e). In contrast, severe rejection was detected in mouse hearts treated with human CSCs: injected CSCs (red) were surrounded by clusters of CD3 þ (green) or CD8 þ (green) T cells (Fig. 6b,d). Quantitative analysis also confirmed that CMMP stimulated negligible local T-cell infiltration as compared with the severe T-cell stimulation by human CSCs (Fig. 6f,g).

Discussion
The last one and a half decades witnessed the booming of stem cell therapies for multiple diseases [16][17][18] . Deviating from the initial perspective that stem cells exert their therapeutic effects through direct cell differentiation and tissue replacement, the paradigm has shifted as emerging evidence suggests that most adult stem cell types exert their beneficial effects through paracrine mechanisms (soluble factors) [19][20][21] . In addition, studies further suggest that cell-cell contact between the injected cells and the host cells plays an important role in tissue regeneration 11 . PLGA, as a biocompatible and biodegradable polymer, has provided a safe and non-toxic building block for various control-release systems 22 . Previous studies have demonstrated the success of coating PLGA nanoparticles with cell membranes from red blood cells 23 , platelets 24 and cancer cells 25,26 . Inspired by these findings, we designed CMMPs and demonstrated the therapeutic effects of CMMPs in an experimental myocardial infarction model. The comparison between CMMP and CSCs is outlined in Supplementary Table 1. CMMP represents a synthetic MP functionalized with both stem cell membranes and secretome, harnessing the power of these two major components of stem cell-induced regeneration. Moreover, CMMP overcomes several major limitations of live stem cells as therapy products. First, stem cells need to be carefully cryo-preserved and thawed before they can be sent to the clinic. As living organisms, how the cells are prepared and processed can greatly affect the therapeutic outcomes. Second, stem cell transplantation carries certain risks (for example, tumourogenecity and immunogenicity if allogeneic or xenogeneic cells were used). CMMPs will most likely be delivered intramyocardially via direct muscle injection. Such injection normally requires open-chest surgery. However, percutaneous options are becoming available with the implementation of the NOGA mapping systems 27 . Moreover, our future studies will explore the potential of vascular delivery of CMMPs (for example, intracoronary, intravenous) with the focus on targeting CMMPs to the injury and promoting extravasation through the mechanism of angiopellosis 28,29 . One caveat of our study is that with the existing assay it is difficult to conclude whether cardiomyocytes (or their progenitors) really are proliferating and leading to remuscularization after CMMP injection. Although this proof-of-concept study targets the heart, CMMP represents a platform technology that is generalizable to other stem cell types and the repair of various other organ systems.

Methods
Derivation and culture of human CSCs. Institutional review board approval was obtained for all procedures, and informed consent was achieved from all patients. Human CSCs were derived from donor human hearts as previously described 5,13 . All data are mean ± s.d. Comparisons between any two groups were performed using two-tailed unpaired Student's t-test. Comparisons among more than two groups were performed using one-way ANOVA followed by post hoc Bonferroni test.
Fabrication of Control MPs and CMMPs. CSC factor-loaded PLGA microparticles (Control MP 1 ) were fabricated by a water/oil/water (w/o/w) emulsion technique. Briefly, human CSC conditioned media as the internal aqueous phase with polyvinyal alcohol (0.1% w/v) was mixed in methylene chloride (DCM) containing PLGA as the oil phase. The mixture was then sonicated on ice for 30 s using a sonicator with a Microtip probe (Misonix, XL2020, Farmingdale, NY, USA). After that, the primary emulsion was immediately introduced into water with polyvinyal alcohol (0.7% w/v) to produce a w/o/w emulsion. The secondary emulsion was emulsified for 5 min on a high-speed homogenizer. The w/o/w emulsion was continuously stirred overnight at room temperature to promote solvent evaporation. The solidified MPs, namely Control MP 1 , were then centrifuged, washed three times with water, lyophilized and stored at À 80°C.
To prepare CMMPs, DiO (Invitrogen)-labelled CSCs went through three freeze/thaw cycles. After which, the disrupted CSCs were sonicated for B5 min at room temperature along with the Control MP 1 . After that, the particles were washed three times in PBS by centrifugation. Control MP 2 was fabricated by cloaking empty PLGA particles with CSC membranes. Successful membrane coating was confirmed using fluorescent microscopy.
Protein release studies. Total protein contents in MPs were determined using the following method. Approximately 10 mg freeze-dried microparticles were dissolved in 1 ml DCM for 60 min. Then, 1 ml PBS was added into solution followed by agitation for 10 min to extract protein from DCM into PBS. After centrifugation,  Scale Bar, 50 mm. * indicates Po0.05 when compared with CMMP group. All data are mean ± s.d. Comparisons among more than two groups were performed using one-way ANOVA followed by post hoc Bonferroni test. The method to induce myocardial infarction in mice was based on previous studies 30 . Briefly, male SCID Beige mice were anaesthetized with 3% isofluorane combined with 2% oxygen inhalation. Under sterile conditions, the heart was exposed by a minimally invasive left thoracotomy and acute myocardial infarction (AMI) was produced by permanent ligation of the LAD coronary artery. Immediately after AMI induction, the heart was randomized to receive one of the following four treatment arms: (1) 'Control (PBS)' group: intramyocardial injection of 50 ml PBS into the heart immediately after AMI; (2) 'Control MP 1 ' group: intramyocardial injection of 1 Â 10 5 Control MP 1 in 50 ml PBS into the heart immediately after AMI; (3) 'CMMP' group: intramyocardial injection of 1 Â 10 5 CMMPs in 50 ml PBS into the heart immediately after AMI; (4) 'CSC' group: intramyocardial injection of 1 Â 10 5 CSCs in 50 ml PBS into the heart immediately after AMI. To enable visualization of Control MP 1 or CMMP in a cohort of animals, we pre-labelled the Control MP 1 or CMMP with Texas Red-X succinimidyl ester (1 mg ml À 1 (Invitrogen, Carlsbad, California)).
Ex vivo fluorescent imaging for biodistribution of CMMPs. Seven days after injection, a cohort of mice receiving CMMPs were killed; their heart, lung, spleen, liver, and kidneys were removed for biodistribution studies. Ex vivo fluorescent imaging was performed with an IVIS Xenogen In Vivo Imager (Caliper Lifesciences, Waltham, MA).
Heart morphometry. After the echocardiography study at 4 weeks, all animals were killed and hearts were collected and frozen in optimum cutting temperature (OCT) compound. Specimens were sectioned at 10 mm thickness from the apex to the ligation level with 100 mm intervals. Masson's trichrome staining was performed as described by the manufacturer's instructions (HT15 Trichrome Staining (Masson) Kit; Sigma-Aldrich). Images were acquired with a PathScan Enabler IV slide scanner (Advanced Imaging Concepts, Princeton, NJ). From the Masson's trichrome stained images, morphometric parameters including viable myocardium, infarct thickness and scar size were measured in each section with NIH ImageJ software. The percentage of viable myocardium as a fraction of the scar area (infarcted size) was quantified as described [31][32][33] . Three selected sections were quantified for each animal.
Cardiac function assessment. All animals underwent transthoracic echocardiography under 1.5% isofluorane-oxygen mixture anaesthesia in supine position at 4 h and 4 weeks. The procedure was performed by an animal cardiologist blind to the experimental design using a Philips CX30 ultrasound system coupled with an L15 high-frequency probe. Hearts were imaged in 2D in long-axis views at the level of the greatest LV diameter. EF was determined by using the formula (LVEDV-LVESV/LVEDV) Â 100%.
Immunogenicity studies for human CSCs and CMMPs. Immuno-competent male CD1 mice were anaesthetized with 3% isofluorane combined with 2% oxygen inhalation. Under sterile conditions, the heart was exposed by a minimally invasive left thoracotomy, and the heart was randomized to receive one of the two treatments: (1) 'CMMP' group: intramyocardial injection of 1 Â 10 5 CMMPs in 50 ml PBS into the heart; (2) 'CSC' group: intramyocardial injection of 1 Â 10 5 human CSCs in 50 ml PBS into the heart. To enable visualization of CMMPs or CSCs, they were pre-labelled with red fluorophore.
Statistical analysis. All results are expressed as mean ± s.d. Comparison between two groups was performed with two-tailed Student's t-test. Comparisons among more than two groups were performed using one-way ANOVA followed by post hoc Bonferroni test. Differences were considered statistically significant when the P valueo0.05.
Data availability. The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information files or from the corresponding author on reasonable request.