Methods 84 (2015) 3–16

Contents lists available at ScienceDirect

Methods

journal homepage: www.elsevier .com/locate /ymeth

Engineering mesenchymal stem cells for regenerative medicineand drug delivery

http://dx.doi.org/10.1016/j.ymeth.2015.03.0021046-2023/� 2015 Elsevier Inc. All rights reserved.

Abbreviations: MSC, mesenchymal stem cells; BM, bone marrow; UC, umbilical cord; iPSC, induced pluripotent stem cells; UC-MSC, UC-derived MSC; BM-Mderived MSC; NP, nanoparticle; AAV, adeno-associated virus; ASC, adipose-derived stem cells; ZFN, zinc finger nuclease; Epo, erythropoietin gene; TALENS, tranactivator-like effector nucleases; CRISPR, clustered regularly interspaced short palindromic repeats; MBs, microbubbles; PEI, poly(ethylenimine); Akt1, protein kinas2, B-cell lymphoma-2; HO-1, heme oxygenase-1; CXCR4, chemokine (C-X-C motif) receptor 4; CCR-1, C-C chemokine receptor type-1; BMP2, bone morphogenetic pTGF-b, transforming growth factor-b; IGF-1, insulin like growth factor-1; VEGF, vascular endothelial growth factor; Runx-2, runt-related transcription factor-2; PLLlysine; PCL, poly-e-caprolactone; ILs, interleukins; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand; scFvCD20-TRAIL, CD20-specific single chain FvTRAIL fusion; FIX, Factor IX; PLA, poly(D,L-lactide); PLMA, poly(D,L-lactic acid-co-a,b-malic acid); SPIO, superparamagnetic iron oxide; PTX, paclitaxel; HSC, hemastem cells; HCELL, hematopoietic cell E-/L-selectin ligand; PSGL-1, P-selectin glycoprotein ligand-1; CB, cord blood; FTVI, fucosyltransferase VI; SLeX, Sialyl-Lewishematopoietic stem/progenitor cells; NHS, N-hydroxy-succinimide; SDF-1, stromal cell-derived factor-1; VCAM-1, vascular cell adhesion molecule-1; Ale, alenICAM-1, intercellular adhesion molecule-1; MAdCAM, mucosal vascular addressin cell adhesion molecule.⇑ Corresponding author at: 500 W. 120th St., 351 Engineering Terrace, New York, NY 10027, United States.

E-mail address: kam.leong@columbia.edu (K.W. Leong).1 Contributed equally.

Ji Sun Park 1, Smruthi Suryaprakash 1, Yeh-Hsing Lao 1, Kam W. Leong ⇑Department of Biomedical Engineering, Columbia University, New York, NY 10027, United States

a r t i c l e i n f o

Article history:Received 28 November 2014Received in revised form 19 February 2015Accepted 2 March 2015Available online 11 March 2015

Keywords:Mesenchymal stem cells (MSC)Cell-based therapyTissue engineeringGenetic modificationNanoparticle (NP)Surface modification

a b s t r a c t

Researchers have applied mesenchymal stem cells (MSC) to a variety of therapeutic scenarios by harness-ing their multipotent, regenerative, and immunosuppressive properties with tropisms toward inflamed,hypoxic, and cancerous sites. Although MSC-based therapies have been shown to be safe and effective toa certain degree, the efficacy remains low in most cases when MSC are applied alone. To enhance theirtherapeutic efficacy, researchers have equipped MSC with targeted delivery functions using genetic engi-neering, therapeutic agent incorporation, and cell surface modification. MSC can be genetically modifiedvirally or non-virally to overexpress therapeutic proteins that complement their innate properties. MSCcan also be primed with non-peptidic drugs or magnetic nanoparticles for enhanced efficacy and exter-nally regulated targeting, respectively. Furthermore, MSC can be functionalized with targeting moietiesto augment their homing toward therapeutic sites using enzymatic modification, chemical conjugation,or non-covalent interactions. These engineering techniques are still works in progress, requiringoptimization to improve the therapeutic efficacy and targeting effectiveness while minimizing any lossof MSC function. In this review, we will highlight the advanced techniques of engineering MSC, describetheir promise and the challenges of translation into clinical settings, and suggest future perspectives onrealizing their full potential for MSC-based therapy.

� 2015 Elsevier Inc. All rights reserved.

1. Introduction

Mesenchymal stem cells (MSC) are adult stem cells capable ofself-renewal and differentiation into multiple lineages includingcartilage, adipose, and bone. MSC are characterized by their abilityto adhere to plastics under standard cell culture conditions,expressing CD44, CD73, CD90, CD105, but not CD45, CD34, andCD14 [1].

Friedenstein first reported MSC merely as proliferating fibrob-lastic cells from bone marrow capable of differentiating into

osteoblasts, chondrocytes and adipocytes. Along with their self-re-newal property, MSC secrete factors, such as growth factors, bothin an autocrine and paracrine fashion, which affect the surroundingmicroenvironment to promote angiogenesis, decrease inflamma-tion, and enhance tissue repair. Moreover, MSC exert strongimmunosuppressive properties, allowing them to be transplantedwithout any pre- or post-treatment. Additionally, they are easyto expand in culture and have multi-lineage differentiation poten-tial and tropism toward neo-angiogenic, tumor, and inflammatorysites. MSC also pose no risk of teratoma formation, nor are there

SC, BM-scriptione B; Bcl-

rotein-2;, poly-L-

antibodytopoieticX; HSPC,dronate;

4 J.S. Park et al. / Methods 84 (2015) 3–16

any ethical issues associated with the cell source. All of these prop-erties collectively make MSC an attractive candidate for cell-basedtherapies.

MSC have been isolated from a wide range of sources includingbone marrow (BM) [2], umbilical cord (UC), adipose tissue [3], liver[4], multiple dental tissues [5], and induced pluripotent stem cells(iPSC) [6]. Each of these sources has its own advantages anddisadvantages. BM is the most characterized and documentedsource of MSC. However, the collection of MSC from BM is painful,invasive, and characterized by a low yield [2]. MSC in the UC can beobtained from Wharton jelly, veins, arteries, the UC lining, and thesubamnion and perivascular regions. UC-derived MSC (UC-MSC)can be obtained through a painless collection method and havefewer associated ethical issues. They also renew faster than BM-derived MSC (BM-MSC) [7]. Adipose tissue is another popularsource, mainly because a large number of MSC can be obtainedthrough minimally invasive methods [3]. In all cases, these cellsneed to be monitored regularly to ascertain their quality. Whilethere are many sources for MSC, the quality of the MSC is highlyvariable from donor to donor and is significantly affected by ageand aging disorders. Furthermore, extended handling of MSCin vitro reduces their differentiation potential. To circumvent theseissues, MSC were recently derived from iPSC [6]. These cells havethe same in vitro and in vivo characteristics of BM-MSC, such asthe potential for adipogenesis, osteogenesis, and chondrogenesis.MSC derived from iPSC also display higher capacity for prolifera-tion and stronger telomerase activity, leading to better engraft-ment and survival after transplantation. Additionally, theydisplay superior capabilities in repairing tissue ischemia comparedto BM-MSC [6]. In addition to tissue regeneration, MSC have beenused to treat type-1 diabetes [8], myocardial infarction [9], graft-versus-host disease [10], inflammatory bowel disease [11], andcancer [12]. Currently, there are 395 ongoing or completed clinicaltrials worldwide using MSC or mesenchymal stromal cells [13],indicating the popularity of MSC for cell-based therapies.

In this review, we will highlight the advanced techniques usedto engineer MSC for tissue engineering and drug delivery applica-tions. The challenges and advantages of each technique will also beanalyzed and discussed. Numerous clinical trials have establishedthe safety of using MSC for cell-based therapies. However, the effi-cacy of MSC in vivo is still low due to poor survival, retention, andengraftment of the cells. The first section of this paper focuses ongenetic modification to enhance the survival, migration, and secre-tion of growth factors for their application in the field of regenera-tive medicine. This is followed by a discussion of MSC applicationsin cancer therapy and gene therapy. Although genetic modificationis a powerful tool, only protein-based drugs can be delivered usingthis approach. Additionally, the genetic modification could poten-tially affect the innate properties of MSC. Hence, over the lastfew years, nanoparticle (NP)-based MSC delivery systems havegained increasing attention. While numerous synthetic NP plat-forms have been designed and some have even shown promisingclinical outcomes, obstacles (including toxicity, specificity, anddelivery efficiency) remain to be overcome before translation. Incontrast, MSC offer intrinsic homing properties, low toxicity, andlow immunogenicity, which could lead to higher delivery effi-ciency compared to conventional nanomedicine platforms. Thesecond section of the paper focuses on combining conventionalNP platforms with MSC-based therapies. The various methods usedto load the therapeutic agents onto MSC, release the therapeuticagents from MSC, and the applications of such MSC-NP combina-tion are analyzed in detail. However, NP-based MSC therapy mustensure that the NP does not compromise the cell’s native proper-ties and it can deliver a suitable release profile. To deal with theseissues, researchers have used surface modification of MSC as analternative. Using various engineering approaches (enzymatic

modification, chemical modification, and non-covalent interac-tions), researchers immobilize targeting moieties onto the cell sur-face to direct MSC to the therapeutic site. As surface modificationconfers only transient expression of targeting molecules on MSC,it does not significantly affect the cells’ phenotype. The last sectionwill suggest future perspectives for translating MSC-basedtherapies.

2. Techniques for engineering MSC

2.1. Genetic modifications

The clinical application of MSC is often hampered by inadequatein vivo performance with respect to survival, retention, andengraftment. Genetic engineering is one approach to improve thein vivo performance of MSC. MSC are genetically engineered tosecrete factors that can protect MSC from apoptosis, increase theirsurvivability in hypoxic conditions, and enhance other innateproperties, such as migration, cardiac protection, and differentia-tion to a particular lineage. Moreover, genetic modifications havealso been used to engineer MSC to produce therapeutic proteinsfor treating diseases like hemophilia and diabetes, and for repair-ing musculoskeletal disorders. Genetic modification of MSC is usu-ally achieved via viral vectors although the use of non-viral vectorsis on the rise.

2.1.1. Viral transductionMSC are readily amenable to viral modification. Standard proto-

cols can lead to 90% transduced cells with no effect on lineage dif-ferentiation or the quality of the progeny [14,15]. Viraltransduction can also offer a long-term and stable production ofthe protein of interest. The most common vectors include retro-virus, lentivirus, baculovirus, and adeno-associated virus (AAV)[16]. Retrovirus leads to integration of the transgene into the hostgenome. While this results in a stable expression, it could also leadto insertional mutagenesis and activation of oncogenes [17].Retrovirus is used when long-term protein production is desirable,such as treatment of genetic diseases. Lentivirus also enablesstable transgene expression through integration into the genome.Non-integrating lentiviral vectors have also been designed thatcan circumvent the problems associated with integration [18].Baculovirus, on the other hand, is non-toxic; it neither replicatesnor integrates into the host genome and is capable of transducingwith high efficiency. Baculovirus can transduce adipose-derivedstem cells (ASC) with 95% efficiency and minimal toxicity [19].Finally, AAV is one of the most promising vectors as it is non-pathogenic to humans and results in long-term gene expression.However, a large fraction of the human population have neu-tralizing antibodies against AAV, which drastically reduces theirin vivo efficacy [20]. To circumvent the issue of activating onco-genes and to achieve targeted integration, Benabdallah et al. usedzinc finger nuclease (ZFN) to add erythropoietin gene (Epo) intothe chemokine (C-C motif) receptor-5 gene locus of MSC. ZFNwas delivered to MSC using adenovirus while Epo was deliveredusing integrase-defective lentiviral vectors. The MSC derived fromhuman BM, adipose tissue, and UCB was transduced with Epo bythe ZFN-driven targeted gene addition. When these cells wereinjected into the peritoneum of non-obese diabetic severe com-bined immunodeficient interleukin-2Rc null mice, the hematocritlevels rose from an average of 49% to more than 60% at day 10[21]. This study clearly demonstrates the potential of site directedinsertion (Fig. 1A) compared to the conventional random integra-tion using viral vectors. Site-directed integration can also beachieved using transcription activator-like effector nucleases(TALENS), and clustered regularly interspaced short palindromicrepeats (CRISPR).

++ ++

MSC

Nucleus

1. Complexation of plasmid

2. Endocytosis3. Nuclear transport

4. Release of the plasmid

5. Transcription

6. Translation

MSC

Nucleus

1. Endocytosis

2. Nuclear transport 3. Integration

into the genome

4. Transcription

5. Translation

Plasmid

Carrier

Nanoparticle or Microparticle complex

Protein of interest

Viral particle

DNA

Genome of MSC

Protein of interest

Random insertion Site-directed insertion

1. Normal insertion

2. Zinc finger nuclease assisted insertion

3. CRISPR cas9 assisted insertion

Site specific break in the genome

Insertion of the foreign DNA at the cleaved site

Fok I DNA cleaving domain

DNA binding domain

Cas9 protein

Guide RNA

Foreign DNA

+ + +

A

B

Fig. 1. Schematic diagram of the mechanisms of viral and non-viral gene delivery. (A) Viral gene delivery. In this case, the insertion of the target gene into the host genome isusually random. The insertion process can be ameliorated by cutting/inserting specific sequence using ZFN or CRISPR/Cas9 system. (B) Non-viral gene delivery. In this case,the plasmid does not integrate into the genome. Figure not drawn to scale.

J.S. Park et al. / Methods 84 (2015) 3–16 5

2.1.2. Non-viral transfectionWhile viral transductions have high efficiency, translation to

the clinics has been hampered by high production cost and adverseimmune reactions. Some viral vectors have limitations in thepackaging capacity of exogenous DNA and the possibility of onco-gene activation. Non-viral vectors, by contrast, are amenable toscale-up manufacturing, low in immune stimulation, and versatilewith a wide array of design choices (Fig. 1B). Use of non-viral genedelivery is also favorable for regenerative medicine, which requiresonly a transient expression. Unfortunately, MSC are difficult totransfect without affecting their viability, resulting in very lowefficiencies. Current methods used to transfect MSC can be dividedinto physical methods and chemical methods. Physical methodsinclude electroporation and nucleofection, both of which lead tosignificant cell death (�40%) if not handled properly [22], andsonoporation [23]. In the case of electroporation an electrical pulseis used to transiently open the pores of the cells. This opening isused to drive the nucleic acid directly into the cytoplasm.Nucleofection also uses electrical pulse but the nucleic acid isdirectly driven into the nucleus of cells. Conversely, sonoporation

uses mechanical vibration, such as ultrasound to increase thetransport of nucleic acid into the cells by enhancing the permeabil-ity of the cell membrane. siRNA was successfully delivered intoMSC using a combination of ultrasound and microbubbles (MBs)[23]. However, the acoustic intensity significantly affected the via-bility of the cells. Chemical methods include the use of lipid agents,polymeric carriers, dendrimers, and inorganic nanoparticles. Lipidand polymeric agents typically can transfect 2–35% of MSC [24].Dendrimers have shown great success in transfecting a wide vari-ety of cell types but not MSC, typically achieving 10–17% transfec-tion efficiency only [25]. Inorganic nanoparticles mostly incombination with polycations have also been use to transfectMSC. Gold NP decorated with Jet-poly(ethylenimine) (PEI) reagentcan condense DNA on the surface to achieve a 2.5-fold increase intransfection efficiency over conventional Jet-PEI polyplexes [26].Similarly, PEI attached to silica performed better than PEI alone,achieving 75% transfection efficiency in human MSC [27]. Theimprovement is attributed to electrostatic interactions within thecomplex that helps stabilize the complex and reduce the size.More recently, Muroski and colleagues developed gold NP

6 J.S. Park et al. / Methods 84 (2015) 3–16

modified with Ku70 peptide using N-cysteine, which was able totransfect rat MSC with a transfection efficiency approaching 80%[28]. This was achieved without affecting the viability and theproperties of the MSC.

2.1.3. Applications of genetically modified MSCApplications of genetic engineering of MSC can be divided into

tissue engineering and drug delivery. In the case of tissue engineer-ing and regenerative medicine, MSC are engineered mainly toincrease their survival, retention, migration, and growth factor pro-duction. While, for drug delivery purposes, MSC are used to exoge-nously produce protein-based therapeutics to treat cancer orgenetic diseases.

2.1.3.1. Tissue engineering and regenerative medicine. A majority ofMSC often dies within the first few hours of in vivo delivery. Tocircumvent this issue, genetic engineering has been used toincrease the survival of MSC in vivo. Overexpressing pro-survival,pro-angiogenic, or anti-apoptotic genes, such as protein kinase B(Akt1) [29], adrenomedullin, B-cell lymphoma-2 (Bcl-2) [30], andheme oxygenase-1 (HO-1) [31], has significantly increased thesurvival of MSC in vivo. Viral modification has also improved theirmigratory behavior towards sites of injury, inflammation, and can-cer through the overexpression of homing receptors, such as che-mokine (C-X-C motif) receptor 4 (CXCR4) [32], and C-C chemokinereceptor type-1 (CCR-1) [33]. MSC overexpressing CXCR4 home ingreater numbers than unmodified MSC in the damaged infarctregion of the myocardium. However, only a transient increase inCXCR4 expression is required for cells to reach the target site.With this in mind, Rejman and colleagues used cationic lipidsand linear PEI to deliver CXCR4 mRNA. Since this bypasses theneed to cross the nuclear envelope, transfection efficiencies of80% and 40% were obtained using cationic lipids and linear PEI,respectively [34]. The mRNA was also detected within 30 minand persisted in the cytoplasm for about 9 days, which was morethan enough time for the cells to reach the target site.

MSC have been most extensively used for treating bone-relateddiseases. Bone morphogenetic protein-2 (BMP2), transforminggrowth factor-b (TGF-b), latent membrane protein-1, insulin likegrowth factor-1 (IGF-1), and growth differentiation factor-5 arefew of the proteins used to engineer MSC to enhance the genera-tion of cartilage, bone, and tendon [35–37]. AAV-6 encodingBMP-2 and vascular endothelial growth factor (VEGF) have beenused to induce bone formation and vasculogenesis [38].However, most of the cells usually accumulate in the lung or thespleen. Therefore, to enhance the homing, MSC have been co-trans-duced with CXCR4 to increase migration; for example, MSC havebeen transduced with AAV to express runt-related transcriptionfactor-2 (Runx-2) and CXCR4 to increase homing and mechanicalstrength of the bone in a murine osteoporotic mouse model [39].MSC have also been engineered with retrovirus to express receptoractivator of nuclear factor-jB and CXCR4 to prevent bone loss [40].The three important corner stones of tissue engineering are cells,scaffold, and bioactive factors. Recently, Brunger et al. attemptedto combine all three factors by using poly-L-lysine (PLL) to immobi-lize lentivirus that encodes for TGF-b3 in a woven poly-e-caprolac-tone (PCL) scaffold seeded with MSC [41]. Using a bioactivescaffold with properties similar to the native tissue to transduceMSC with TGF-b3 led to a sustained and robust cartilaginous extra-cellular matrix formation.

Non-viral approaches to deliver the required growth factors arethe most rational approach to repair or induce bone formationsince constitutive expression of genes is not imperative here. Theabovementioned non-viral vectors are viable in delivering thera-peutic genes, such as osterix [42], core-binding factor alpha 1[43], and BMP2 [44], for bone regeneration [45]. Recently, cell

sheets were engineered using Lipofectamine� 2000 to produceanti-miR-138, a microRNA precursor family [46]. These cell sheetswith high cell number and good cell–cell contact uses the scaffold-free tissue engineering approach and can be easily layered andattached to tissue beds. Moreover, the cell sheets enable theregeneration of massive bone with good vascularization.

MSC are also known to enhance cardiac repair through neo-vascularization, protecting myocardium from ischemic cell death.However, the bottleneck in effective translation of MSC for cardiactherapy lies in their poor survival, retention, and engraftmentin vivo. Therefore, MSC have been engineered to enhance their sur-vival using Akt1 [29], connexin43 [47], heat-shock protein 20 [48],Bcl-2 [30], and HO-1 [31], homing properties using CXCR4 [32] andCCR-1 [33], and differentiation into cardiomyocytes using murinehyperpolarization-activated cyclic nucleotide-gated 2 [9]. Theyhave also been engineered to produce angiogenic factors usingVEGF [49] and hepatocyte growth factor [50], and anti-in-flammatory agents using interleukin-18 (IL-18) binding protein[51] to increase vascularization and reduce inflammation, respec-tively. For example, Yeh and colleagues used baculovirus to trans-duce ASC sheets to express VEGF [52]. The cell–cell junction in thecell sheets increased the survival and engraftment of the cell sheet.In addition, secretion of VEGF effectively arrested the loss of heartfunction, prevented myocardium fibrosis, reduced the infarct size,and increased the blood vessel formation.

2.1.3.2. Drug delivery for cancer and gene therapy. The inherent abil-ity of MSC to migrate towards cancer cells makes them attractiveas a cellular delivery vehicle for cancer therapy. MSC have beenengineered to produce a wide variety of cancer therapeutic pro-teins, such as interleukins (IL-2 [12], IL-12 [53], IL-18 [54]), inter-feron-b [55], and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) [56]. MSC have also been engineered toproduce prodrug-activating enzymes [57]. In this case, the geneti-cally modified cells are followed by administration of the inactiveprodrug locally, which is activated by the enzyme. This, in princi-ple, would provoke the cell-killing action only at the site populatedby MSC, minimizing the systemic toxicity. Some of the combina-tion therapies that have shown success are: herpes simplex virusthymidine kinase with Ganciclovir, Escherichia coli cytosine deami-nase with 5-fluorocytosine, carboxylesterase with irinotecan, andcytochrome P450 with cyclophosphamide or ifosfamide [57].Incorporating a tetracyclin-induced promoter to turn off the geneexpression confers safety to the transplanted MSC. Deliveringoncolytic virus is another popular approach taken for cancer treat-ment. These viruses are designed to specifically replicate only intumor cells and amplify only at the target site [58]. While thereare quite a few oncolytic viruses in clinical trials, their translationis greatly hindered by the host immune response and inefficientviral distribution. MSC have been shown to be an effective carrierof oncolytic virus in treating ovarian tumors [59] and human hep-atocellular carcinoma [60].

While engineered MSC have shown promise in treating cancer,monotherapy has not proven effective in treating highly heteroge-neous cancer. Additionally, cancerous cells gain resistance tochemotherapy during their evolution, prompting the need forcombination therapy. For example, TRAIL-engineered MSC deliv-ered alongside temozolomide was more effective in treatingglioblastomas than either of the individual therapies [61].However, to make sure that both the drugs reach the tumor sitesimultaneously, MSC have been engineered to secrete CD20-speci-fic single chain Fv antibody fused to TRAIL (scFvCD20-TRAIL). Thecombination was more efficient than delivering TRAIL-transducedMSC alone for treating non-Hodgkin’s lymphoma because of thesimultaneous inhibition of proliferation and induction of apoptosisin cancer cells [60].

J.S. Park et al. / Methods 84 (2015) 3–16 7

MSC have been also used to treat monogenetic diseases, such aslysosomal storage disease or hemophilia, both of which require ahigh level of protein production afforded by viral transduction.Lentiviral transduction of MSC to produce high levels of biologi-cally active b-glucuronidase led to successful cross-correction oflysosomal storage device in vivo [62–64]. Coutu et al. deliveredengineered MSC to produce Factor IX (FIX) for treating hemophiliausing a poly(lactic-co-glycolic acid) composite coated withbiomineralized collagen-1 [65]. The long-term self-renewal anddifferentiation of MSC was promoted by the calcium phosphateceramics, leading to therapeutic level production of FIX in hemo-philic mice for over 12 weeks. For diabetes therapy, MSC have beentransduced with AAV to increase expression of islet genes [8].Pancreatic duodenal homeobox-1 has also been used to induce dif-ferentiation of MSC towards the b cell phenotype [66]. These stud-ies demonstrate the potential of engineered MSC as a clinicallyrelevant platform for treating various other genetic diseases.

Although genetically engineered MSC may be appealing for amyriad of diseases, the long-term safety of viral gene therapyremains a concern. As of now, only in diseases with a favorablebenefit-to-risk ratio would such cell therapy be practical. Non-viralgene engineering may alleviate the safety concerns but its lowtransfection efficiency must be improved before it can rival viralmethods.

2.2. Therapeutic agent incorporation

MSC have emerged as targeted drug delivery vehicles becauseof their (1) tissue homing ability; (2) low immunogenicity as theydo not express histocompatibility complex class II molecule; and(3) anti-inflammatory property with the secreted growth factorsand cytokines. As genetic engineering can only deliver proteindrugs, NP delivery strategies can be applied to deliver non-peptidicdrugs and therapeutic nucleic acids. Drug-loaded NP can be cou-pled with MSC through either cellular internalization or cell sur-face anchorage. As MSC migrate to the target tissue, the drug canbe released in a local and sustained manner.

2.2.1. Loading via cellular internalizationCellular internalization is one of the drug loading approaches.

Drug-encapsulated NPs taken up via endocytosis have to escapethe endolysosomal compartment to endow MSC with drug deliveryproperty (Fig. 2A). The endocytic and cytoplasmic localization pro-cesses are influenced by the composition, size, and surface chargeof the NP [67,68]. Various types of NPs, such as polymeric or liposo-mal NPs, with diameters ranging from tens to several hundrednanometers have been applied, and also with both negativelyand positively charged NP. Poly(D,L-lactide) (PLA) and its analoguesare FDA-approved, biodegradable drug delivery polymers that beara weak anionic backbone under physiological conditions. Cellularinternalization of PLA NPs is mainly through pinocytosis or cla-thrin-mediated endocytosis [69]. After endocytosis, the polymerundergoes a charge transition from negative to positive due tothe acidic environment in the endosomal compartment and subse-quently escapes from the endosome. Roger et al. loaded MSC withcoumarin-6 encapsulated PLA NPs and showed that they did notaffect the MSC viability even when the cells were treated with adosage of 200 lg/mL NPs. The endocytosis of PLA NPs by MSCwas time and concentration dependent, and very efficient [70].Another PLA analogue, poly(D,L-lactic acid-co-a,b-malic acid)(PLMA), has also been used for MSC loading. PLMA NPs composedof iron oxide and the dye, FITC, was designed and used as a bifunc-tional contrast agent for MSC tracking under magnetic resonanceand fluorescent visualizations. These NPs were taken up by MSCwithout affecting their viability and differentiation capabilities

into adipocytes and osteocytes [71]. Similar results were reportedfor the negatively charged NP, mesoporous silica NP [72].

NPs with cationic surfaces have also been used to improve thecellular uptake by MSC [73–75]. Cationic polymers that have beenwidely used for non-viral gene delivery for MSC [76–78], such asPLL and PEI, are also efficient for drug loading applications.Kostura and colleagues coated superparamagnetic iron oxide(SPIO) NPs with PLL and showed that PLL-coating could signifi-cantly enhance the cell uptake by MSC. The percentage of cellsloaded with NPs reached nearly 100%, whereas no significantuptake was observed when MSC were treated with unmodifiedSPIO NPs [79]. The size of the coated PLL was further optimized:when SPIO NPs were coated with a moderate-sized PLL (MW388 kDa), more than 92% of cell loading efficiency was achieved,whereas only 80% was observed when lysine monomer was used[80]. In addition to PLL, a series of low molecular PEI (MW2 kDa)-SPIO NPs for MSC loading was designed and optimized.These PEI-SPIO NPs with sizes ranging from 20 to 80 nm alsoshowed high loading efficiency (93 ± 2%) and did not have signifi-cant toxicity on MSC [81].

In addition to polymeric NPs, liposomal NPs have been used todeliver therapeutic cargo to MSC [70,82,83]. They can also be takenup by MSC through endocytosis and release their cargos intra-cellularly by fusing to the endosomal membrane. In contrast topolymeric NPs, high doses of liposomal NPs are usually requiredto achieve effective cell loading efficiency, which leads to signifi-cant cytotoxicity [70,82]. In addition, small therapeutic cargoscan be directly loaded onto MSC without any carrier encapsulation.Pascucci et al. used murine MSC as a vehicle to deliver achemotherapy drug, paclitaxel (PTX). They demonstrated thatMSC were able to take up and release PTX although the drug inhib-ited the cells’ proliferation ability. In spite of this, in vitro resultsshowed that the released products from the PTX-primed MSCcould kill cancer cells [84].

The success of non-viral transfection has inspired researchers inthe field to apply the same concepts for loading MSC with thera-peutic agents and use them as drug vehicles. The abovementionedstudies show that MSC are capable of internalizing the therapeuticagent-loaded NPs or the drug directly. Generally, polymeric NPsare more efficient for drug loading compared to liposomal NPs[82] and a positive NP surface charge enhances the cell loading effi-ciency [75]. Beside these strategies, receptor-mediated endocytosisis also a possible way to enhance the cell loading efficiency.Decorating NPs with antibody against MSC surface antigen CD90could improve the kinetics of NP internalization [85].

2.2.2. Anchoring via lipid fusionAs explained in Section 2.2.1., liposomal NP has been used to

load MSC with therapeutic cargos. Fig. 2B shows the basic mecha-nisms of a non-covalent strategy to modify the MSC surface withtherapeutic agents. When NPs interact with the MSC surface, theyundergo membrane fusion and the cargos are transferred andanchored on the cell. As a result, a synthetic ‘‘receptor’’ could becreated on the cell surface. Dutta and colleagues demonstrated thisstrategy on fibroblasts. Various functional groups or fluorescentdyes can be transferred from the liposomes onto cells [86]. A simi-lar approach has been applied to MSC for surface modification.Biotin-modified liposome was used to present biotin on the cellsurface, which serves as the bridge for subsequent ligand immobi-lization via the biotin–streptavidin interaction [87].

2.2.3. Mechanisms and strategies to control the releaseFor MSC that are primed with therapeutic cargos, exocytosis

and simple diffusion are the two major routes of cargo release.Exocytosis is a removal process that varies with NP type and sizeas well as cell type. Cells can release half of the internalized cargos

MSC - MSC +

++

+

++

++

+

+

+

++

+

+

+

NP formation

Endocytosis

NP formation

Membrane fusion

A B

C

D

E

MSC MSC

Carrier Theapeutic agent

Nanoparticle

Cargo Carrier

Nanoparticle

Fig. 2. Schematic diagram of the selective mechanisms of MSC to uptake multifunctional NPs. (A) Endocytosis. Polymeric, liposomal NPs and small drugs can be loaded toMSC through this route. (B) Lipid fusion. MSC surface can be decorated and loaded with drugs through this route, but this is not the only mechanism for liposomal NPs. A andB are not drawn to scale. (C) Comparison on target tissue accumulation between SPIO-loaded luciferase MSC and unloaded luciferase MSC. MSC were injected into the muscleinjury site and localized by 3T external magnetic force. (D) Comparison on restenosis efficacy between magnet-guided and non-guided SPIO-loaded MSC in rabbit femoralartery injury model. Tissue sections were stained with H&E for visualization. (E) Enhancement of therapeutic agent accumulation by MSC in a brain glioblastoma model. NPaccumulation of NP-loaded MSC was 5-fold higher than NP only in the tumor site (indicated with an arrow). The figure is adapted with permission from (C), [82], � 2012Elsevier; (D), [95], � 2013 Elsevier; and (E), [97], � 2013 Elsevier.

8 J.S. Park et al. / Methods 84 (2015) 3–16

between 30 min to few hours [88,89]. In a fluorescent dye transferassay, more than 85% of the internalized dye was transferred from

MSC to the co-cultured cancer cells through exocytosis [90]. Incontrast, if the cargos were released from the endosomal

J.S. Park et al. / Methods 84 (2015) 3–16 9

compartments, which are typically governed by a combination ofmatrix degradation- and diffusion-controlled release mechanisms,the drug release profile from the MSC would be slower and moreprolonged. By combining the exocytotic and endosomal releasemechanisms, a biphasic profile may fit the release kinetics of thera-peutic agent-primed MSC. SPIO-loaded MSC studies have revealedthis phenomenon. Here the cells’ iron content dropped signifi-cantly in the beginning stage and cells released half of the ironwithin the first two days, which may be dominated by exocytosisroute. After that, relatively slow diffusion kinetics could beobserved as the iron was totally removed by two weeks [82,91].To control the diffusion kinetics, a multidrug resistance-1 genewas used to generate P-glycoprotein 1, which could actively pumpout the loaded drug from MSC cytosol [92]. While this modificationcan tune the release profile, it also raises safety concerns, whichwill be discussed in Section 3.

For the cargos anchored on the cell surface, the release profile isregulated by the dissociation constant between the ‘‘synthetic’’receptor and its ligand (the therapeutic cargos). For example, ifthe MSC surface is modified with biotin and then the cargo isloaded through biotin–avidin interaction, one may expect a veryslow release profile since the affinity between biotin and avidinis very strong (the dissociation constant is �10�16 M) [93].

2.2.4. Applications of therapeutic agent loaded MSCThe MSC loaded with therapeutic agents have been used for tis-

sue regeneration and cancer therapeutics. For tissue regeneration,it is essential that the multi-lineage differentiation potential ofMSC is not affected by the uptake or fusion of NPs. Most of theNPs have been reported not to influence the differentiationcapability of MSC with the exception of titanium and cobalt/chro-mium/molybdenum NPs [94]. As a result, NPs, especially, SPIO-containing NPs, have been incorporated with MSC for repairing abroad spectrum of tissue types. SPIO-containing NPs are generallyused for improving the homing ability of MSC. With the assistanceof external magnetism, SPIO-primed MSC localize on the injuredtissue site. Fig. 2C shows a typical example of the magneticforce-assisted tissue localization of SPIO-primed MSC. Luciferase-expressing MSC either with or without SPIO-loading was trans-planted into a rat with muscle injury. Subsequently, a 3T externalmagnetic force was applied on the site of injury, and a significantincrease of MSC localization in the case of SPIO-primed MSC wasobserved [95]. Similar approaches have been applied to other typesof tissue injuries including vascular [82,83] and cartilage repairs[96]. A preclinical arterial injury model has been used for mag-netic-assisted MSC delivery. The femoral artery injury on rabbitwas created and treated with SPIO-primed MSC and an externalmagnet guide. After treatment for three weeks, a significant reduc-tion of restenosis was seen without losing the level of re-endothe-lialization (Fig. 2D) [82].

Additionally, NP-loaded MSC are also used for cancer therapy.By using their homing properties, cancer drugs can be preciselydelivered to the tumor site, and drug-loaded MSC can release thedrug via exocytosis and diffusion. Exosomes from drug-loadedMSC have also shown to effectively kill cancer cells; for example,exosomes collected from PTX-loaded MSC were successful ininhibiting the growth of pancreatic adenocarcinoma cells in vitro[84]. Furthermore, Huang et al. have conducted an in vitro tran-swell assay to demonstrate that the tumor tropism of MSC is notaffected when MSC are loaded with NPs [97]. They have alsoshowed in a murine glioblastoma model that contrast agent-primed MSC can cross the blood–brain barrier to the tumor site.The result demonstrated that the MSC-treated group showed 5.2-fold increase of contrast agent accumulation on tumor comparedwith the group treated with only NPs (Fig. 2E) [97]. This indicates

that drug-loaded MSC may be more advantageous and efficientcompared with the conventional cancer therapeutics.

2.3. Transient cell surface modifications

Although genetic modification and therapeutic agent incorpora-tion techniques have been used to improve the native properties ofMSC to increase the therapeutic efficacy, the major limitation ofusing MSC as cellular therapeutic carriers is the inability to delivera therapeutically effective dosage to the desired site. Despite thefact that the homing property of MSC to inflammatory and tumorsites has been well documented [98], the exact mechanism of thehoming property of endogenous and exogenous MSC has still notbeen elucidated. However, it is accepted that several mediatorsincluding chemokines [30,99], cell adhesion molecules [100,101],matrix metalloproteinase [102,103], and cytokines [104,105] playa vital role. Nevertheless, relying solely on the homing property ofMSC is insufficient to achieve significant therapeutic cell dosageat the disease site as MSC often lack the homing receptorsexpressed on hematopoietic stem cells (HSC) or leukocytes, andhence are not as efficient. Moreover, ex vivo handling (passage fre-quency, confluency, and the microenvironment of the culture con-ditions) of the cells can cause them to lose their homing ability[103,106]. Furthermore, upon systemic administration, most MSCbecome filtered and accumulate in the lung, liver, and spleen beforethey are eliminated from the body [107,108]. Consequently,approximately less than 1% of the injected MSC reach the target site[107,109], which would not be enough to drive the therapeuticeffect via paracrine signaling. Therefore, several approaches havebeen used to enhance the therapeutic dosage of MSC. One approachmight be to inject more of them to ensure that an effective numberof cells reach the target site. However, direct injection (intravascu-lar or intra-arterial injection) of a large number of cells results incomplications attributed to microembolism, which may lead to vas-cular obstruction, stroke, and potentially, death [110,111].Furthermore, compared with the commonly used animal models,humans would require larger quantities of MSC, the production ofwhich may require excessive expansion that might diminish theirpluripotency [106]. Therefore, there is a limit to dose escalationfor improved therapeutic outcomes and a need to establish safeadministration protocol for high cell numbers.

A better approach would be to augment the homing efficiency ofMSC by using surface modification techniques to introduce target-ing moieties: (1) by exploiting the HSC’ and leukocyte’s transmigra-tion process to target activated endothelium of the therapeutic site;(2) by using chemical or non-covalent interactions; or (3) byexploiting biospecific recognition (antigen–antibody or ligand–re-ceptor). These techniques are not only versatile and economic,but also potentially safer than that of genetic modification andtherapeutic agent incorporation because they avoid the long-termsafety and regulatory issues, and toxicity issues, respectively.Because transmigration of transplanted MSC to the activatedendothelium occurs between 60 and 120 min [112], the expressionof transient homing molecules would enhance the targeting effi-ciency of MSC to the target site. Additionally, surface modificationtechniques can enhance the adhesion and engraftment of MSCwithout compromising their viability, adhesion, multipotency, pro-liferation, and differentiation potentials [87,113–115]. In the nextsection, we will discuss three different surface modification strate-gies that improve the cell homing property (Fig. 3), and therebypotentially avoid the risk and need to inject more cells.

2.3.1. Enzymatic modificationThe most popular method for enhancing the migratory behavior

of MSC is to functionalize MSC with adhesion ligands, such ashematopoietic cell E-/L-selectin ligand (HCELL) and P-selectin

HN

O

Endothelial Cells 4 pt

MSC

NHO

NHO

Enzymatic Modification

Chemical Modification Peptide

Conjugation

Antibody Conjugation

Surface Glycoprotein (CD44)

Hematopoietic cell E-/L-selectin ligand (HCELL)

E-selectin

P-selectin

Sialic acidGalactose

Fucose

StreptavidinBiotin

Polyacrylamide linkerSialyl-Lewis X (sLe ) x

Sulfonated-N-hydroxysuccinimide

Targeting Peptide

Palmitated Derivatives

Palmitated Protein G or A

Antibody

Bispecific Antibodies

N-Acetylglucosamine

Fig. 3. Schematic diagram of both covalent and non-covalent surface modification techniques used to enhance the homing property of MSC toward the endothelium oftherapeutic sites. Four main strategies depicted are: enzymatic modification, chemical modification, peptide conjugation, and antibody conjugation. The sugar chemistrylabeling for HCELL (in the left rectangle) is adapted from [151]. Figure not drawn to scale.

10 J.S. Park et al. / Methods 84 (2015) 3–16

glycoprotein ligand-1 (PSGL-1), which may be absent, insuffi-ciently expressed naturally [116], and/or lost during the expansionof cells ex vivo [106]. These ligands mediate adhesive interactions(e.g., tethering and rolling) due to the constitutive expression ofselectins on bone marrow and inflamed tissues [117]. For this pur-pose, the most characterized selectin ligands have been widelyused to engineer transmembrane glycoproteins on HSC and MSC.This concept originates from the endogenous lymphocyte infiltra-tion from blood to inflamed tissues, where the overexpression ofcell adhesion receptors mediate the cell rolling process via theselectin-dependent pathways [118]. Cell rolling enables the leuko-cyte to slow down and interact with the locally secreted chemoki-nes, which subsequently leads to the activation of integrins,followed by firm adhesion and invasion of leukocytes within theinflamed vasculature [118]. Within this cascade, the rolling stepinitiates the leukocyte extravasation and plays a crucial role inthe subsequent adhesion and transmigration processes [87].Several immunoprecipitation and mutation studies have demon-strated that the absence, modification, or blocking of the effectormolecules associated with the cell rolling response prevents firmadhesion under shear stress in vivo [100,104].

Xia and colleagues first tested this method by enzymaticallymodifying the surface of human umbilical cord blood (CB) CD34+cells with selectin-binding motifs (a-1,3-fucose) to improve theirhoming to the bone marrow under shear stress. For selectin to bindto its cell surface ligand, it first needs to be fucosylated to form gly-can determinants (epitopes). Therefore, to transiently displayfucose residues on the surface glycans, fucosyltransferase VI(FTVI) and the fucose donor substrate, guanosine diphosphatefucose, were used [119]. This modification generated a-1,3-fucose-functionalized CB CD34+ cells, which led to an increasedrolling response on the E- or P-selectin-immobilized plates at0.5 dyn/cm2 by 1.9- and 2.9-fold compared to the unmodifiedMSC, respectively [119]. In vivo results also demonstrated anincrease in bone marrow engraftment of systemically administeredCB cells in immunocompromised mice.

Extending this concept one step further, Sackstein et al. engi-neered MSC, which naturally lack the expression of E-selectin

ligands, to display selectin-binding motifs (a-2,3-sialic acid and a-1,3-fucose) on the endogenous surface glycans, which are carbohy-drate polymers on a glycoprotein. The resulting sialofucosylated gly-can epitope on CD44 is known as Sialyl-Lewis X (SLeX), the selectin-binding carbohydrate motif, which is an active site for HCELL andPSGL-1 expressed on human hematopoietic stem/progenitor cells(HSPC) and leukocytes, respectively. Enzymatic modification ofendogenous CD44 into its sialofucosylated glycoform (HCELL, asrepresented in Fig. 3) enables any cells expressing CD44 to bindfirmly to E- and L-selectin expressing cells without going throughconventional multistep chemical synthesis [113].

2.3.2. Chemical modificationEnzymatic modification of the cell surface glycan is limited to

the glycoproteins that already exist on the cell surface and maypotentially affect other critical cell surface molecules by non-speci-fic interactions. In contrast, chemical modifications enable MSC topresent several targeting moieties, including ligands [116], pep-tides [114], and polymers [120], via covalent bioconjugationmethod. Sarkar et al. covalently modified MSC to transiently deco-rate SLeX by first using N-hydroxy-succinimide (NHS) chemistry,and then subsequently using the biotin-streptavidin interaction.This targeting moiety, SLeX, can be immobilized on the surface ofMSC by first modifying the amine groups on the cell membraneto biotin-NHS and subsequently adding streptavidin to introducethe biotin binding sites. Then, P-selectin-binding MSC is obtainedby mixing and incubating them with either biotinylated lipid vesi-cles [87] or a polymer construct (�30 nm) [116], containing SLeX

determinants. Biotinylated lipid vesicles intercalate and fuse tothe cell membrane of MSC and display biotin on its surface, offer-ing a docking site for subsequent conjugation with biotinylatedSLeX after treating MSC with streptavidin. MSC modified withSLeX using this method resulted in promoted cell rolling, adhesion,and migration [121].

This versatile and simple technique can be used to display anytargeting moiety without affecting the microenvironment of themodified MSC, which is controlled by their paracrine factors (suchas stromal cell-derived factor-1 (SDF-1) and IGF-1) [116]. This

J.S. Park et al. / Methods 84 (2015) 3–16 11

conventional chemical conjugation technique has also beenrecently adapted to present bone-targeting protein conjugatedpolymer on MSC [120]. Moreover, chemical modification offers arelatively stable covalent binding of the targeting molecules onMSC – at least 6 h without the functionalized peptide getting inter-nalized – without affecting the cells’ viability, proliferation, andmultipotency [114].

2.3.3. Non-covalent modification2.3.3.1. Antibody conjugation. Another clinically attractive approachfor improving the specific homing of MSC is antibody conjugation,due to antibodies’ high affinity towards the overexpressed antigenson the target site. This strategy has been especially successful incancer immunotherapies by driving the immune system againstcancer cells using a bispecific antibody, which simultaneouslybinds to two different antigens, or an engineered receptor knownas chimeric antigen receptor, where the antibody-binding domainis connected to the T-cell activating domains [122]. As exploited inT-cells, antibody conjugation to the stem cells has also beenachieved to localize MSC at the therapeutic target. This is realizedby using palmitated proteins G [123–125] or A [126], or a bispecificantibody.

Engineering the cell surface with palmitated proteins, which actas hydrophobic anchors, allows the subsequent conjugation ofpotentially any antibody with an accessible Fc cassette or a proteinfused to this domain [127]. Chen and colleagues optimized the pro-tocol [128] from Kim and Peacock [126] and first established a con-cept called ‘‘protein painting.’’ This entails the intercalation ofpalmitated proteins on the cell membrane as a pre-coating stepto bind antibodies to cell surfaces while neither compromisingtheir binding affinity to specific antigens [126,128] nor the viabilityand differentiation potentials of the cellular carriers [123,129]. Thissimple and versatile method was first used to home chondrocytesfor cartilage repair in the rabbit explant model by binding antibodytargeting cartilage extracellular matrix [123].

Bispecific antibodies offer an interesting targeting strategy asthey can bind to two different antigens to guide stem cells to thedesired tissues and promote stem cell retention as well. This strat-egy was successful in homing murine stem cells expressing c-kit tothe murine injured myocardial cells overexpressing vascular celladhesion molecule-1 (VCAM-1) using anti-c-kit and anti-VCAM-1bispecific antibodies [130]. Similarly, this method was applied torepair the function of the infarcted heart by arming human HSCswith bispecific antibodies consisted of anti-human CD45 andanti-rat myosin light chain antibodies [131]. This technique wasadapted by Deng et al. to modify MSC with bispecific antibodies[132]. However, bispecific antibodies have not been widely usedwith MSC, likely due to the lack of a specific cell surface marker[133]. Moreover, the instability and the high production cost ofgenerating clinical-grade bispecific antibodies via heteroconjuga-tion of two antibodies (using chemical, genetic, or hybridomamethods) are the limiting factors for translating this modality toclinics.

2.3.3.2. Peptide conjugation. Although an antibody provides a speci-fic affinity towards its antigen, it is expensive to produce on a largescale with high purity. Peptide conjugation, which has also beenwidely used in preclinical studies, may be a more attractive strat-egy. Peptide production is inexpensive, relatively simple, and scal-able with high purity. Furthermore, using the phage displaytechnique, the peptide sequence with the highest binding affinityto the target tissue can be tested and further optimized. Along withthe chemical coupling method, peptide conjugation can be realizedby employing hydrophobic interactions where the homing peptideconjugated to lipid groups (palmitate derivatives) [134,135] areintegrated on the cell membrane composed of a lipid bilayer.

With coating of palmitated proteins or their derivatives, any typeof cell (MSC, red blood cells, chondrocytes, etc. [144]) can be con-jugated with peptide to home to the specific target.

2.3.4. Applications of surface-modified MSCSurface modification techniques introduce targeting moieties,

which can bind to the proteins overexpressed in the diseased ordamaged tissue via covalent or non-covalent interactions. Hence,the therapeutic dosage of MSC trafficking to the target is increasedand the subsequent therapeutic efficacy may be enhanced.Collectively, these methodologies that modify exogenous MSCdemonstrate the potential for engineered MSC to be used inregenerative medicine and drug delivery applications without sig-nificantly affecting the cells’ viability, multipotency, and dif-ferential potentials. Surface-modified MSC have been used forthree major applications: bone regeneration, repair of inflamed tis-sues, and treatment of myocardial infarction.

2.3.4.1. Bone regeneration. Although MSC differentiates into osteo-blasts, naturally they do not migrate to bone. Therefore, only byengineering MSC can they be targeted to bone for bone regenera-tion. Because of the constitutive expression of selectins on bonemarrow [117], modified MSC can target the bone marrow usingthe enzymatic modification technique. The forced glycosylationof the membrane surface glycans confers HCELL on MSC. As aresult, firm adhesion of glycosylated MSC to the microvascular ves-sel in bone marrow was achieved under hemodynamic shear stres-ses of up to 30 dyn/cm2 [113]. Intravenous injection of HCELL-expressing human MSC into immunocompromised mice also sig-nificantly enhanced the homing response to the calvaria of themice, as compared to controls. Likewise, HCELL facilitates themigration of other cells; for instance, HCELL-expressing murineHSPC resulted in more than 3-fold higher homing to the bone mar-row compared to the unmodified control [136]. HCELL also plays arole in mediating the metastasis of colon cancer and circulatingtumor cells [137,138]. Taken together, exploiting the tropisms tobone, MSC can be decorated with HCELL by enhancing thetransmigration of MSC to the bone via the selectin-mediated celladhesion pathway.

A recent paper published by D’Souza et al. combined a covalentconjugation method with polymer-based engineering to enhancethe homing property of MSC for treating bone injury sites [120].Atom transfer radical polymerization was used to synthesize abone-targeting moiety that contains alendronate (Ale). This drughas high affinity to bone and is known to decelerate bone resorp-tion by inducing the apoptosis of osteoclasts, thereby shifting thebone remodeling balance toward osteogenesis by osteoblasts[139]. MSC armed with bone-targeting polymer can home tohydroxyapatite crystals and rodent bone fragments, suggestingtheir potential for bone regeneration therapy. By contrast, otherstudies have used the biospecific recognition of the a4b1 integrinreceptor on the MSC surface and a synthetic peptide (LLP2A) mim-icking the ligand against the activated integrin to augmentosteogenesis [140,141]. A hybrid compound of Ale and LLP2Awas covalently coupled (LLP2A-Ale) and conjugated to MSC, whichallowed them to navigate and tether to bone and promote osteo-blast differentiation in vitro and in vivo. These studies demonstratethat functionalizing these dual targeting (MSC and bone) polymersor peptide-Ale hybrids on MSC can be used to treat bone fracturehealing and bone degenerative diseases, such as osteoporosis.

2.3.4.2. Inflammatory disease treatment. Modified MSC are not onlyexceptional in targeting bone marrow, but also in navigating toinflamed tissues expressing selectin [117] using the chemical mod-ification [87] and antibody conjugation [124] approaches.Functionalized MSC with SLeX using biotinylated lipid vesicles

12 J.S. Park et al. / Methods 84 (2015) 3–16

exhibited the rolling activity on the P-selectin surface under ashear stress of 0.5 dyn/cm2 without affecting the multilineage dif-ferentiation potentials, viability, proliferation, and adhesion kinet-ics of MSC [87]. Similarly, using an optimized biotinylated polymerconstruct resulted in a sturdier cell rolling response under a shearstress of up to 2 dyn/cm2 on the P-selectin substrate. An in vivostudy also indicated that the imaging agent-labeled, SLex-modifiedMSC navigate towards the inflamed ear 56% better than the nega-tive control [116]. Exploiting the same NHS chemistry, anotherstudy used a flexible linker to present E-selectin-binding peptideson MSC for promoting firm adhesion and the following rolling pro-cesses on E-selectin substrates under a shear stress of up to 10 dyn/cm2 [114]. Therefore, by exploiting the leukocyte extravasation viathe selectin-mediated rolling response, this simple, robust, andversatile strategy promotes targeting of MSC to the sites of inflam-mation upon systemic injection.

Ko and colleagues tailored MSC to the inflammatory sites byanchoring antibodies of the upregulated cell adhesion molecules,such as intercellular adhesion molecule-1 (ICAM-1) [124], theVCAM-1, and the mucosal vascular addressin cell adhesion mole-cule (MAdCAM) [125]. MSC coated with the ICAM-1 antibodiessubstantially promoted their adhesion to the activated endothelialcells in vitro [124]. Moreover, the homing abilities of the MAdCAMor VCAM-1 directing MSC were demonstrated by the prolongedsurvival (approximately 2.3-fold compared to unmodified MSC)of the treatment group injected with the MAdCAM antibody-coated MSC tested in the in vivo model of inflammatory bowel dis-ease. Efficient localization to colon and suppression of regulatory Tcells was achieved with the VCAM-1 antibody-coated MSC in vivo.Interestingly, this study demonstrates that better localization ofMSC does not always result in increased therapeutic efficacy,which agrees with the latest study that the therapeutic outcomedoes not correlate with the engraftment efficiency [142]. This willbe further discussed in Section 3.

2.3.4.3. Myocardial infarction repair. Employing molecular recogni-tion techniques, such as surface-modified MSC with bispecific anti-body [132] or peptide conjugation [134,135], Deng et al. improvedthe homing and therapeutic efficacy of MSC in a murine myocardialinfarct model by combining the bispecific antibodies (anti-mouseCD29 and anti-myosin light chain antibody) with ultrasound-mediated MBs. The combination approach significantly increasedthe homing of MSC at the target site and reduced the fibrosis densitycompared with the unmodified MSC or MSC coated with bispecificantibodies only [132]. This synergistic effect was due to theenhanced migration of the MSC, which changed their microenviron-ment by activating the STAT-mediated signaling pathway. Theincreased homing also improved the therapeutic efficiency.

MSC have also been presented with various palmitated-is-chemia-homing peptides in the rodent myocardial infarction modelusing the transient ‘‘protein painting’’ method, as explained inSection 2.3.3.1. By doing so, increased localization of peptide-coated MSC was achieved at the diseased site up to 10 times morethan the unmodified cells [134,135]. These results demonstratedthe enhanced localization of modified MSC with targeting peptidesin vivo; however, the stability of the peptide coated on MSCdecreased to 70% after 2 h of incubation at 37 �C [135]. This maybe due to internalization of the peptides anchored on the cell sur-face via endocytosis or proteolysis. Moreover, none of these studiestested whether the weak hydrophobic interaction between the cellmembrane and the homing peptide is maintained under physio-logically relevant shear stress. Hence, Lo et al. adapted this ‘‘paint-ing’’ method to present a fusion protein of PSGL-1 N-terminalpeptide and human IgG1 on MSC to assess whether they can attachto the selectin-expressing endothelial cell layers under flow.Without affecting the vital phenotypes of MSC, the tethering,

rolling, and retention under flow (up to 10 dyn/cm2) were signifi-cantly improved compared to the unmodified MSC [115]. The sta-bility of the peptide could not be compared with the previousstudies due to different incubation temperatures, but 66% of theMSC complexed with the fusion proteins were maintained after24 h of incubation at the room temperature, demonstrating a tran-sient expression of the fusion protein.

Potentially, these methods could generate engineered MSC totarget specific tissues for regeneration or drug delivery applica-tions as multiple peptides and antibodies can be coated onto thecell membrane.

3. Challenges and future directions

As discussed in the previous sections and summarized inTable 1 with selected examples of different engineering strategies,they offer new modalities and opportunities for MSC-based ther-apy and tissue regeneration. Each technique has its own benefitsand drawbacks, as summarized in Table 2, and the challenges ofthese strategies for translation are still being defined; they are nei-ther insignificant nor insurmountable. However, they would beworth addressing because of the amazing potential that MSC offer.

The International Society for Cellular Therapy has proposed theminimal criteria for defining human MSC [143], yet one of themajor barriers limiting the clinical translation of allogeneic MSCis the lack of consensus on MSC definitions to evaluate the qualityof MSC from different donors with varying ages, sources, andgenetically diverse backgrounds. Moreover, the lack of standard-ized protocols to extract, culture, and prepare MSC has led to largevariability in their properties. Therefore, engineering these hetero-geneous populations of MSC to begin with may affect the safetyprofile and lead to unpredictable clinical outcomes. Hence, a morestringent approach, such as automating the ex vivo expansion pro-cess, may be crucial for the standardized use of MSC in the future.Also, it is imperative to identify the relevant biomarkers and estab-lish a screening/monitoring method to evaluate, quantify, and pre-dict the clinical effectiveness of MSC for the further development ofMSC-based therapies.

Although viral modification is the most efficient and prominentengineering modality for enhancing the therapeutic functions ofMSC, its long-term safety concerns, including eliciting an immuneresponse and activating oncogenes, have hampered the clinicaltranslatability of this approach. It is also limited in the size andnumber of genes that can be simultaneously delivered. Yet, withadvancement in the field of genetic engineering, molecular-editingtools, such as TALENS and CRISPR, may be exploited to improvesafety and efficiency of inserting or expressing large or multiplegenes. Non-viral gene therapy, on the other hand, is severely lim-ited by the low transfection efficiency. A concerted effort to opti-mize all aspects of non-viral gene transfer ranging fromconstruction of efficient plasmids to design of stimulus-responsivegene carriers and to improved transfection protocol is required. Forinstance, microfluidics may be used as a rapid, highly reproducible,and scalable platform to transfect MSC with uniform propertiesand high transfection efficiency. Furthermore, combinatorial andhigh throughput approaches can be used to screen carrier designand formulation to improve the therapeutic performance of MSC.

NP incorporation strategies may have detrimental effects onMSC. For instance, proliferation of MSC is inhibited by PTX [84]and its viability is affected by other chemotherapeutic drugs, suchas vincristine and 5-fluorouracil [144,145]. Hence, MSC is not auniversal platform for delivering any therapeutic agents.Moreover, for loading therapeutic agent on NPs, it would be essen-tial to identify whether the properties of MSC are adverselyaffected because some NP carriers may affect the multipotency ofMSC [94]. Furthermore, as a delivery platform, cargo release

Table 1Summary of the select examples of the applications of MSC engineered with various modalities.

Engineering method Technique Target cargo Application Therapeutic outcome References

2.1 Geneticmodification

2.1.1 Viral transduction Akt1 Repair heart Reduced intramyocardial inflammation,regenerated 80–90% lost myocardial volume,and recovered systolic and diastolic cardiacfunction

[29]

TGF-b3 Repaircartilage

Significant differentiation towardschondrogenic lineage with deposition of ECM,rich in type II collagen andglycosaminoglycans

[41]

scFvCD20-TRAIL Treat non-Hodgkin’slymphoma

Achieved 65% of tumor regression comparedto control through intrinsic and extrinsicapoptotic pathway

[60]

2.1.2 Non-viral transfection Brain-derivedneurotropic factor

Treatneurologicaldisease

Displayed that more than 80% of cellsexpressed BDNF and lasted for 15 days in vitro

[26]

CXCR4 mRNA Increasehoming

Detected mRNA within 30 min of transfectionand persisted in the cytoplasm for about9 days

[34]

Anti-mir-138 Regeneratebone

Resulted in the regeneration of bone withgood vascularization

[46]

2.2 Therapeutic agentincorporation

2.2.1 Cellular internalization PTX Treat cancer Internalized PTX into the golgi-derivedvesicles; exerted cytotoxic effect on humanpancreatic cancer cell line (CFPAC-1) with anIC50 of 10.24 ng/mL

[84]

SPIO Regenerateskeletalmuscle

Significantly increased the localization of MSCupon exposure to external magnetic force andshowed superior muscle regeneration in vivo

[95]

Mesoporous silica NPwith Imaging agents(FITC, NIR dyeZW800, Gd3+ and64Cu)

Visualizetumor

Stably integrated to MSC without affectingtheir intrinsic properties; improved tumorhoming by 5.2-fold in the murineglioblastoma model compared with that of NPonly

[97]

2.2.2 Lipid fusion SLeX Improvehoming toinflamedtissue

Promoted cell rolling and P-selectininteraction by 52% compared to unmodifiedMSC at a shear stress of 0.5 dyn/cm2

[87]

2.3 Cell surfacemodification

2.3.1 Enzymatic modification (byengineering cell surfaceglycoproteins using FTVI)

HCELL (active site ofHCELL = SLeX ligand)

Increasehoming tobonemarrow

Achieved firm adhesion and transmigration ofMSC to the microvascular vessel in bonemarrow under hemodynamic shear stresses ofup to 30 dyn/cm2

[113]

2.3.2 Chemical modification (bycovalently coupling targetingligands)

SLeX Repairinflamedtissue

Promoted cell rolling, adhesion, and migrationin vitro and in vivo; increased 56% of migrationefficiency towards inflamed murine earcompared to control

[116],[121]

Polymer-Ale hybrid Regeneratebone

Enhanced homing to both human and rat bonefragments without affecting the proliferativeand differentiative potential of MSC

[120]

2.3.3 Non-covalentmodification (byemploying molecularrecognitiontechniques)

2.3.3.1Antibodyconjugation

Bispecific antibody Repairmyocardialinfarction

Significantly increased the homing of MSC atthe target site and reduced the fibrosis densitycompared with the unmodified MSC

[132]

2.3.3.2Peptideconjugation

Palmitated-ischemia-homing peptide

Repairmyocardialinfarction

Enhanced localization (10-fold increase) ofpeptide-coated MSC compared to their controlin vivo

[134],[135]

J.S. Park et al. / Methods 84 (2015) 3–16 13

kinetics needs to be optimized. Since MSC release the cargothrough exocytosis and diffusion, the kinetics of drug release fromthe MSC platform could be predicted if we can estimate how fastand how many NPs can cross the intracellular barriers and unloadthe cargo into the cytoplasm of MSC. A fluorescence resonanceenergy transfer-based approach using quantum dots and theirfluorescent pairs might be suitable to examine the release kineticsof NPs before engineering MSC [146,147]. For loading therapeuticagents through membrane anchoring, the release kinetics is con-trolled by the dissociation between the drug and its receptor, soit may be easier to predict the kinetics using techniques estab-lished for studying ligand–receptor interactions. However,researchers using this approach would face two other issues, drugdegradation and drug immunogenicity. Because the drug in thisapproach is not protected, enzymatic or hydrolytic degradationmay lead to inactivation of the therapeutic agent, which may alsoelicit an immune response.

Although investigations on the surface modification of MSChave not revealed serious compromise of their functions so far,potential danger remains until long-term and detailed in vivo stud-ies are conducted. Other than biological concerns, surface-modifiedMSC will be directly exposed to a shear stress environment, condi-tions that may lead to aggregation, internalization, and shedding ofthe active moieties. Variable expression of the targeting moleculesmay confound the homing and therapeutic efficacy of the engi-neered MSC, which would significantly hamper the clinical transla-tion of this approach. Therefore, surface-functionalized MSC needto be rigorously tested under clinically relevant conditionsin vitro and tested with large sample size to assess their impactin vivo. Furthermore, as homing and healing are both orchestratedby multiple effector molecules, improving the localization of modi-fied MSC to the target neither guarantees enhanced MSCtransmigration nor promises higher therapeutic outcome [125].For example, if the cells lack adhesion molecules, chemokine

Table 2The advantages and disadvantages of each MSC engineering method.

Engineering method Technique Advantages Disadvantages

2.1 Genetic modification 2.1.1 Viral transduction High transduction efficiency Long-term safety and regulatory issues; sizelimitation for the insert (<9 kb); labor-intensive; expensive; may cause insertionmutagenesis

2.1.2 Non-viral transfection Potentially safer than viral transduction;flexible insert size; cost-effective; simple andfast; high reproducibility

Low transfection efficiency; transient geneexpression

2.2 Therapeutic agentincorporation

2.2.1 Cellular internalization Cargo diversity; simple and fast; high loadingefficiency

Cargo may affect MSC viability; unable todecorate the cell surface

2.2.2 Lipid fusion Enables cell surface decoration Complicated; cargo release profile needs to beoptimized

2.3 Cell surface modification 2.3.1 Enzymatic modification Improve cell homing to affected endothelialbeds without compromising MSC’s nativeproperties; induce transmigration mediatedby selectin; simple

Limited to modifications on the existingendogenous glycoproteins; may loosetargeting moiety due to internalization orshedding of the cell over time (transientexpression)

2.3.2 Chemical modification Improve cell homing property withoutcompromising MSC’s native properties;versatile; simple

Transient expression of the targeting moiety

2.3.3.1 Antibody conjugation Improve cell homing property withoutcompromising MSC’s native properties;employs high specific affinity towards aspecific antigen; enables multi-targeting;versatile

High production cost on a large scale withhigh purity; lack of MSC-specific cell markers;Transient expression of the targeting moiety

2.3.3.2 Peptide conjugation Improve cell homing property withoutcompromising MSC’s native properties;versatile; exploits ligand-specific interactions;applicable to a wide range of target tissuesand cell types; low production cost on a largescale with high purity; enable to optimize thebinding affinity using phage display

Transient expression of the targeting moiety

14 J.S. Park et al. / Methods 84 (2015) 3–16

receptors, or other key associated molecules for transmigrationand invasion, presenting MSC with targeting moieties may not besufficient to enhance the homing or engraftment processes[119,124]. Nevertheless, surface modification approach bypassesmany risks associated with multiple MSC administrations, suchas activating the adaptive immunity. The variability in clinicalresponses may also be related to the presence or absence of appro-priate cues in the host’s microenvironment to maintain the bioac-tivity of MSC. Therefore, providing a suitable niche for implantedMSC at the affected tissue has become a paramount factor toimprove engraftment, which leads to extending the therapeuticoutcome and reducing the effective therapeutic dosage. There isample evidence that delivering pre-seeded cells in a polymericscaffold may promote their engraftment, survival, and tissueregeneration [148]. These biodegradable scaffolds, which aredesigned to mimic the native ECM, also provide structural andmechanical support for better cell engraftment at the target site.Although details of the scaffold-based strategies are not coveredherein, they have been discussed in other excellent reviews[148], [149], and [150]. Therefore, future studies need to also focuson inducing multitudes of biochemical and physical cues to affectthe microenvironment of the MSC, which may enhance the desiredtherapeutic efficacy. For instance, combining both the targetingand therapeutic agents (i.e. NPs, peptide drugs, exogenous genesetc.) on MSC and deliver them via an optimally designed implanta-ble scaffold (with optimized porosity, biocompatibility, degradabil-ity, and mechanical and biochemical properties) may be a fruitfuldirection.

4. Concluding remarks

Owing to their remarkable self-renewal, multipotent, immuno-suppressive, and homing properties, MSC are attractive candidates

for cell-based therapy in their own right. They can regenerate dam-aged tissues, deliver therapeutic gene(s)/agent(s), and exert para-crine effects, which lead to recruiting effector cells necessary forregenerative therapeutics and disease treatments. Recently,researchers have sought to further enhance these therapeuticproperties of MSC. Through various approaches, MSC can be engi-neered to express exogenous genes, incorporate therapeutic car-gos, or present targeting moieties to enhance their survival,homing, and therapeutic efficacy. These strategies offer therapeuticdosages of MSC and active agents at the target site, circumventingthe need for repetitive systemic administration. However, to main-tain or augment the tropism of engineered MSC to inflamed andcancerous sites for regeneration and treatment purposes, respec-tively, more detailed studies need to be conducted to better under-stand the behavior of native MSC in homing, healing, anddifferentiation. In addition, to be clinically translatable, optimalex vivo handling and preservation protocols and standards for qual-ity and safety profile of MSC need to be established. The usage ofMSC for personalized medicine and tissue repair has been promis-ing. As new and unexpected properties of MSC are discovered, ashas been the case in the past decade, MSC-based therapy is likelyto become even more prominent. The engineering strategiesdescribed in this review to modify MSC will further enhance theirfunctionality and potency. It is by no means an easy task to over-come the scientific hurdles of MSC modification, but it is a chal-lenge worthy of scientific undertaking given the potential reward.

Acknowledgments

We thank Dr. Chris Grigsby, Ms. Hayeun Ji and Ms. Chun-WeiChi for their kind input for the manuscript preparation. We wouldlike to acknowledge the fellowship support from the Korean-American Scientists and Engineers Association, United States (JP),Schlumberger Foundation, United States (SS), and Taiwanese

J.S. Park et al. / Methods 84 (2015) 3–16 15

Ministry of Education, Taiwan (YHL). Support from the NIH, UnitedStates (UH3TR000505, HL109442, AI096305) is also acknowledged.

References

[1] S. Natunen, M. Lampinen, H. Suila, I. Ritamo, V. Pitkanen, A.V. Nairn, J. Rabina,S. Laitinen, K.W. Moremen, W. Reutter, L. Valmu, Glycobiology 23 (2013)1004–1012.

[2] M. Gnecchi, L.G. Melo, Stem Cells in Regenerative Medicine, Humana Press,Totowa, NJ, 2009. pp. 281–294.

[3] K. Bieback, H. Klüter, Curr. Stem Cell Res. Ther. 2 (2007) 310–323.[4] T. Lu, P. Hu, X. Su, C. Li, Y. Ma, W. Guan, Cell Tissue Banking 15 (2014) 439–

450.[5] G.T.J. Huang, S. Gronthos, S. Shi, J. Dent. Res. 88 (2009) 792–806.[6] Q. Lian, Y. Zhang, J. Zhang, H.K. Zhang, X. Wu, Y. Zhang, F.F.-Y. Lam, S. Kang, J.C.

Xia, W.-H. Lai, K.-W. Au, Y.Y. Chow, C.-W. Siu, C.-N. Lee, H.F. Tse, Circulation121 (2010) 1113–1123.

[7] K.L. Seeberger, J.M. Dufour, A.M.J. Shapiro, J.R.T. Lakey, R.V. Rajotte, G.S.Korbutt, J. Tech. Methods Pathol. 86 (2006) 141–153.

[8] C. Moriscot, F. de Fraipont, M.-J. Richard, M. Marchand, P. Savatier, D. Bosco,M. Favrot, P.-Y. Benhamou, Stem Cells 23 (2005) 594–603.

[9] I. Potapova, A. Plotnikov, Z. Lu, P. Danilo, V. Valiunas, J. Qu, S. Doronin, J.Zuckerman, I.N. Shlapakova, J. Gao, Z. Pan, A.J. Herron, R.B. Robinson, P.R.Brink, M.R. Rosen, I.S. Cohen, Circ. Res. 94 (2004) 952–959.

[10] K. Zhao, R. Lou, F. Huang, Y. Peng, Z. Jiang, K. Huang, X. Wu, Y. Zhang, Z. Fan, H.Zhou, C. Liu, Y. Xiao, J. Sun, Y. Li, P. Xiang, Q. Liu, J. Am. Soc. Blood MarrowTransplant. (2014).

[11] H.M.A. Ahmed, G. Abdel Salam, Tarek M. Salman, Laila A. Rashed, Dina Sabry,Mona F. Schaalan, 22 (2014) 515–521.

[12] K. Nakamura, Y. Ito, Y. Kawano, K. Kurozumi, M. Kobune, H. Tsuda, A. Bizen, O.Honmou, Y. Niitsu, H. Hamada, Gene Ther. 11 (2004) 1155–1164.

[13] The United States National Institutes of Health Clinical Trial Database,www.clinicaltrials.gov; Accessed on February 18, 2015.

[14] J.D. Mosca, J.K. Hendricks, D. Buyaner, J. Davis-Sproul, L.C. Chuang, M.K.Majumdar, R. Chopra, F. Barry, M. Murphy, M.A. Thiede, U. Junker, R.J. Rigg,S.P. Forestell, E. Böhnlein, R. Storb, B.M. Sandmaier, Clin. Orthop. Relat. Res.(2000) S71–S90.

[15] J.A. Nolta, M.B. Hanley, D.B. Kohn, Blood 83 (1994) 3041–3051.[16] K.J. Airenne, Y.-C. Hu, T.A. Kost, R.H. Smith, R.M. Kotin, C. Ono, Y. Matsuura, S.

Wang, S. Ylä-Herttuala, Mol. Ther. 21 (2013) 739–749.[17] S. Hacein-Bey-Abina, A. Garrigue, G.P. Wang, J. Soulier, A. Lim, E. Morillon, E.

Clappier, L. Caccavelli, E. Delabesse, K. Beldjord, V. Asnafi, E. MacIntyre, L.D.Cortivo, I. Radford, N. Brousse, F. Sigaux, D. Moshous, J. Hauer, A. Borkhardt,B.H. Belohradsky, U. Wintergerst, M.C. Velez, L. Leiva, R. Sorensen, N.Wulffraat, S. Blanche, F.D. Bushman, A. Fischer, M. Cavazzana-Calvo, J. Clin.Invest. 118 (2008) 3132–3142.

[18] M.B. Banasik, P.B. McCray, Gene Ther. 17 (2010) 150–157.[19] C.-H. Lu, K.-J. Lin, H.-Y. Chiu, C.-Y. Chen, T.-C. Yen, S.-M. Hwang, Y.-H. Chang,

Y.-C. Hu, Tissue Eng. Part A 18 (2012) 2114–2124.[20] S. Nayak, R.W. Herzog, Gene Ther. 17 (2010) 295–304.[21] B.F. Benabdallah, E. Allard, S. Yao, G. Friedman, P.D. Gregory, N. Eliopoulos, J.

Fradette, J.L. Spees, E. Haddad, M.C. Holmes, C.M. Beauséjour, Cytotherapy 12(2010) 394–399.

[22] H. Aslan, Y. Zilberman, V. Arbeli, D. Sheyn, Y. Matan, M. Liebergall, J.Z. Li, G.A.Helm, D. Gazit, Z. Gazit, Tissue Eng. 12 (2006) 877–889.

[23] K. Otani, K. Yamahara, S. Ohnishi, H. Obata, S. Kitamura, N. Nagaya, J. Control.Release 133 (2009) 146–153.

[24] J.L. Santos, D. Pandita, J. Rodrigues, A.P. Pêgo, P.L. Granja, H. Tomás, Curr. GeneTher. 11 (2011) 46–57.

[25] Y. Gheisari, M. Soleimani, K. Azadmanesh, S. Zeinali, Cytotherapy 10 (2008)815–823.

[26] E. Uchimura, S. Yamada, L. Uebersax, S. Fujita, M. Miyake, J. Miyake, J. Biosci.Bioeng. 103 (2007) 101–103.

[27] J.S. Park, K. Na, D.G. Woo, H.N. Yang, J.M. Kim, J.H. Kim, H.-M. Chung, K.-H.Park, Biomaterials 31 (2010) 124–132.

[28] M.E. Muroski, T.J. Morgan, C.W. Levenson, G.F. Strouse, J. Am. Chem. Soc.(2014).

[29] Y.-S. Yu, Z.-Y. Shen, W.-X. Ye, H.-Y. Huang, F. Hua, Y.-H. Chen, K. Chen, W.-J.Lao, L. Tao, Chin. Med. J. 123 (2010) 1702–1708.

[30] Y. Li, X. Yu, S. Lin, X. Li, S. Zhang, Y.H. Song, Biochem. Biophys. Res. Commun.356 (2007) 780–784.

[31] T. Tsubokawa, K. Yagi, C. Nakanishi, M. Zuka, A. Nohara, H. Ino, N. Fujino, T.Konno, M.-A. Kawashiri, H. Ishibashi-Ueda, N. Nagaya, M. Yamagishi, Am. J.Physiol. Heart Circ. Physiol. 298 (2010) H1320–H1329.

[32] Z. Cheng, L. Ou, X. Zhou, F. Li, X. Jia, Y. Zhang, X. Liu, Y. Li, C.A. Ward, L.G. Melo,D. Kong, Mol. Ther. 16 (2008) 571–579.

[33] J. Huang, Z. Zhang, J. Guo, A. Ni, A. Deb, L. Zhang, M. Mirotsou, R.E. Pratt, V.J.Dzau, Circ. Res. 106 (2010) 1753–1762.

[34] J. Rejman, G. Tavernier, N. Bavarsad, J. Demeester, S.C. De Smedt, J. Control.Release 147 (2010) 385–391.

[35] C.P. Hodgkinson, J.A. Gomez, M. Mirotsou, V.J. Dzau, Hum. Gene Ther. 21(2010) 1513–1526.

[36] C.-H. Lu, Y.-H. Chang, S.-Y. Lin, K.-C. Li, Y.-C. Hu, Biotechnol. Adv. 31 (2013)1695–1706.

[37] N.K. Bleich, I. Kallai, J.R. Lieberman, E.M. Schwarz, G. Pelled, D. Gazit, Adv.Drug Deliv. Rev. 64 (2012) 1320–1330.

[38] S. Kumar, C. Wan, G. Ramaswamy, T.L. Clemens, S. Ponnazhagan, Mol. Ther.18 (2010) 1026–1034.

[39] C.-Y. Lien, K. Chih-Yuan, J. Bone Miner. Res. 24 (2009) 837–848.[40] S.W. Cho, H.J. Sun, J.-Y. Yang, J.Y. Jung, J.H. An, H.Y. Cho, H.J. Choi, S.W. Kim,

S.Y. Kim, D. Kim, C.S. Shin, Mol. Ther. 17 (2009) 1979–1987.[41] J.M. Brunger, N.P.T. Huynh, C.M. Guenther, P. Perez-Pinera, F.T. Moutos, J.

Sanchez-Adams, C.A. Gersbach, F. Guilak, Proc. Natl. Acad. Sci. U.S.A. 111(2014) E798–E806.

[42] Q.-G. Lai, K.-F. Yuan, X. Xu, D.-R. Li, G.-J. Li, F.-L. Wei, Z.-J. Yang, S.-L. Luo, X.-P.Tang, S. Li, Oral Surg. Oral Med. Oral Pathol. Oral Radiol. Endod. 111 (2011)412–419.

[43] S.-J. Park, K. Na, Biomaterials 33 (2012) 6485–6494.[44] F. Wegman, R.E. Geuze, Y.J. van der Helm, F. Cumhur, J. Alblas, J. Tissue Eng.

Regen. Med. 8 (2014) 763–770.[45] F. Wegman, F.C. Oner, W.J.A. Dhert, J. Alblas, Biotechnol. Genet. Eng. Rev. 29

(2013) 206–220.[46] J. Yan, C. Zhang, Y. Zhao, C. Cao, K. Wu, L. Zhao, Y. Zhang, Biomaterials 35

(2014) 7734–7749.[47] D. Wang, W. Shen, F. Zhang, M. Chen, H. Chen, K. Cao, Cell Biol. Int. 34 (2010)

415–423.[48] X. Wang, T. Zhao, W. Huang, T. Wang, J. Qian, M. Xu, E.G. Kranias, Y. Wang, G.-

C. Fan, Stem Cells 27 (2009) 3021–3031.[49] F. Gao, T. He, H. Wang, S. Yu, D. Yi, W. Liu, Z. Cai, Can. J. Cardiol. 23 (2007)

891–898.[50] H.-F. Duan, C.-T. Wu, D.-L. Wu, Y. Lu, H.-J. Liu, X.-Q. Ha, Q.-W. Zhang, H. Wang,

X.-X. Jia, L.-S. Wang, Mol. Ther. 8 (2003) 467–474.[51] M. Wang, J. Tan, Y. Wang, K.K. Meldrum, C.A. Dinarello, D.R. Meldrum, Proc.

Natl. Acad. Sci. U.S.A. 106 (2009) 17499–17504.[52] T.-S. Yeh, Y.-H.D. Fang, C.-H. Lu, S.-C. Chiu, C.-L. Yeh, T.-C. Yen, Y. Parfyonova,

Y.-C. Hu, Biomaterials 35 (2014) 174–184.[53] C.H. Ryu, S.-H. Park, S.A. Park, S.M. Kim, J.Y. Lim, C.H. Jeong, W.-S. Yoon, W.-I.

Oh, Y.-C. Sung, S.-S. Jeun, Hum. Gene Ther. 22 (2011) 733–743.[54] G. Xu, X.-D. Jiang, Y. Xu, J. Zhang, F.-H. Huang, Z.-Z. Chen, D.-X. Zhou, J.-H.

Shang, Y.-X. Zou, Y.-Q. Cai, S.-B. Kou, Y.-Z. Chen, R.-X. Xu, Y.-J. Zeng, Cell Biol.Int. 33 (2009) 466–474.

[55] X. Ling, F. Marini, M. Konopleva, W. Schober, Y. Shi, J. Burks, K. Clise-Dwyer,R.-Y. Wang, W. Zhang, X. Yuan, H. Lu, L. Caldwell, M. Andreeff, CancerMicroenviron. 3 (2010) 83–95.

[56] L.P. Mueller, J. Luetzkendorf, M. Widder, K. Nerger, H. Caysa, T. Mueller,Cancer Gene Ther. 18 (2011) 229–239.

[57] I. Amara, W. Touati, P. Beaune, I. de Waziers, Biochimie 105C (2014) 4–11.[58] O.G. Donnelly, F. Errington-Mais, R. Prestwich, K. Harrington, H. Pandha, R.

Vile, A.A. Melcher, Curr. Pharm. Biotechnol. 13 (2012) 1834–1841.[59] E.K. Mader, G. Butler, S.C. Dowdy, A. Mariani, K.L. Knutson, M.J. Federspiel, S.J.

Russell, E. Galanis, A.B. Dietz, K.-W. Peng, J. Transl. Med. 11 (2013). 20–20.[60] C. Yan, S. Li, Z. Li, H. Peng, X. Yuan, L. Jiang, Y. Zhang, D. Fan, X. Hu, M. Yang, D.

Xiong, Mol. Pharm. 10 (2013) 142–151.[61] S.M. Kim, J.S. Woo, C.H. Jeong, C.H. Ryu, J.-D. Jang, S.-S. Jeun, Stem Cells Transl.

Med. 3 (2014) 172–182.[62] A. Donsante, B. Levy, C. Vogler, M.S. Sands, J. Inherit. Metab. Dis. 30 (2007)

227–238.[63] J.W. Kyle, E.H. Birkenmeier, B. Gwynn, C. Vogler, P.C. Hoppe, J.W. Hoffmann,

W.S. Sly, Proc. Natl. Acad. Sci. 87 (1990) 3914–3918.[64] C.M. Simonaro, M.E. Haskins, J.L. Abkowitz, D.A. Brooks, J.J. Hopwood, J.

Zhang, E.H. Schuchman, Gene Ther. 6 (1999) 107–113.[65] D.L. Coutu, J. Cuerquis, R. El Ayoubi, K.-A. Forner, R. Roy, M. François, M.

Griffith, D. Lillicrap, A.-M. Yousefi, M.D. Blostein, J. Galipeau, Biomaterials 32(2011) 295–305.

[66] O. Karnieli, Y. Izhar-Prato, S. Bulvik, S. Efrat, Stem Cells 25 (2007)2837–2844.

[67] S.E.A. Gratton, P.A. Ropp, P.D. Pohlhaus, J.C. Luft, V.J. Madden, M.E. Napier, J.M.DeSimone, Proc. Natl. Acad. Sci. U.S.A. 105 (2008) 11613–11618.

[68] V. Holzapfel, A. Musyanovych, K. Landfester, M.R. Lorenz, V. Mailander,Macromol. Chem. Phys. 206 (2005) 2440–2449.

[69] J.K. Vasir, V. Labhasetwar, Adv. Drug Deliv. Rev. 59 (2007) 718–728.[70] M. Roger, A. Clavreul, M.C. Venier-Julienne, C. Passirani, L. Sindji, P. Schiller, C.

Montero-Menei, P. Menei, Biomaterials 31 (2010) 8393–8401.[71] L. Wang, K.G. Neoh, E.T. Kang, B. Shuter, S.C. Wang, Biomaterials 31 (2010)

3502–3511.[72] D.M. Huang, Y. Hung, B.S. Ko, S.C. Hsu, W.H. Chen, C.L. Chien, C.P. Tsai, C.T.

Kuo, J.C. Kang, C.S. Yang, C.Y. Mou, Y.C. Chen, FASEB J. 19 (2005). 2014-+.[73] M.R. Lorenz, V. Holzapfel, A. Musyanovych, K. Nothelfer, P. Walther, H. Frank,

K. Landfester, H. Schrezenmeier, V. Mailander, Biomaterials 27 (2006) 2820–2828.

[74] A.S. Arbab, L.A. Bashaw, B.R. Miller, E.K. Jordan, B.K. Lewis, H. Kalish, J.A.Frank, Radiology 229 (2003) 838–846.

[75] T.H. Chung, S.H. Wu, M. Yao, C.W. Lu, Y.S. Lin, Y. Hung, C.Y. Mou, Y.C. Chen,D.M. Huang, Biomaterials 28 (2007) 2959–2966.

[76] L.L. Farrell, J. Pepin, C. Kucharski, X.Y. Lin, Z.H. Xu, H. Uludag, Eur. J. Pharm.Biopharm. 65 (2007) 388–397.

[77] X.A. Chen, L.J. Zhang, Z.J. He, W.W. Wang, B. Xu, Q. Zhong, X.T. Shuai, L.Q.Yang, Y.B. Deng, Int. J. Nanomed. 6 (2011).

[78] N. Pimpha, P. Sunintaboon, S. Inphonlek, Y. Tabata, J. Biomater. Sci.-Polym. E21 (2010) 205–223.

16 J.S. Park et al. / Methods 84 (2015) 3–16

[79] L. Kostura, D.L. Kraitchman, A.M. Mackay, M.F. Pittenger, J.W.M. Bulte, NMRBiomed. 17 (2004) 513–517.

[80] M. Babic, D. Horak, M. Trchova, P. Jendelova, K. Glogarova, P. Lesny, V.Herynek, M. Hajek, E. Sykova, Bioconjug. Chem. 19 (2008) 740–750.

[81] G. Liu, Z.Y. Wang, J. Lu, C.C. Xia, F.B. Gao, Q.Y. Gong, B. Song, X.N. Zhao, X.T.Shuai, X.Y. Chen, H. Ai, Z.W. Gu, Biomaterials 32 (2011) 528–537.

[82] J. Riegler, A. Liew, S.O. Hynes, D. Ortega, T. O’Brien, R.M. Day, T. Richards, F.Sharif, Q.A. Pankhurst, M.F. Lythgoe, Biomaterials 34 (2013) 1987–1994.

[83] N. Landazuri, S. Tong, J. Suo, G. Joseph, D. Weiss, D.J. Sutcliffe, D.P. Giddens, G.Bao, W.R. Taylor, Small 9 (2013) 4017–4026.

[84] L. Pascucci, V. Cocce, A. Bonomi, D. Ami, P. Ceccarelli, E. Ciusani, L. Vigano, A.Locatelli, F. Sisto, S.M. Doglia, E. Parati, M.E. Bernardo, M. Muraca, G.Alessandri, G. Bondiolotti, A. Pessina, J. Control. Release 192 (2014)262–270.

[85] D. Sarkar, J.A. Ankrum, G.S.L. Teo, C.V. Carman, J.M. Karp, Biomaterials 32(2011) 3053–3061.

[86] D. Dutta, A. Pulsipher, W. Luo, H. Mak, M.N. Yousaf, Bioconjug. Chem. 22(2011) 2423–2433.

[87] D. Sarkar, P.K. Vemula, W.A. Zhao, A. Gupta, R. Karnik, J.M. Karp, Biomaterials31 (2010) 5266–5274.

[88] J. Panyam, V. Labhasetwar, Pharm Res 20 (2003) 212–220.[89] B.D. Chithrani, W.C.W. Chan, Nano Lett. 7 (2007) 1542–1550.[90] J.L. Munoz, S.A. Bliss, S.J. Greco, S.H. Ramkissoon, K.L. Ligon, P. Rameshwar,

Mol Ther Nucl Acids 2 (2013).[91] Z.Y. Huang, Y.L. Shen, N. Pei, A.J. Sun, J.F. Xu, Y.A. Song, G.Y. Huang, X.N. Sun,

S.N. Zhang, Q. Qin, H.M. Zhu, S. Yang, X.D. Yang, Y.Z. Zou, J.Y. Qian, J.B. Ge,Biomaterials 34 (2013) 9905–9916.

[92] Z.B. Gao, L.N. Zhang, J. Hu, Y.J. Sun, Nanomed. Nanotechnol. 9 (2013) 174–184.[93] O. Livnah, E.A. Bayer, M. Wilchek, J.L. Sussman, Proc. Natl. Acad. Sci. U.S.A. 90

(1993) 5076–5080.[94] C.C. Okafor, H. Haleem-Smith, P. Laqueriere, P.A. Manner, R.S. Tuan, J. Orthop.

Res. 24 (2006) 461–473.[95] A. Nakabayashi, N. Kamei, T. Sunagawa, O. Suzuki, S. Ohkawa, A. Kodama, G.

Kamei, M. Ochi, J. Orthop. Res. 31 (2013) 754–759.[96] T. Kobayashi, M. Ochi, S. Yanada, M. Ishikawa, N. Adachi, M. Deie, K. Arihiro,

Arthroscopy 24 (2008) 69–76.[97] X.L. Huang, F. Zhang, H. Wang, G. Niu, K.Y. Choi, M. Swierczewska, G.F. Zhang,

H.K. Gao, Z. Wang, L. Zhu, H.S. Choi, S. Lee, X.Y. Chen, Biomaterials 34 (2013)1772–1780.

[98] S. Kidd, E. Spaeth, J.L. Dembinski, M. Dietrich, K. Watson, A. Klopp, V.L.Battula, M. Weil, M. Andreeff, F.C. Marini, Stem Cells 27 (2009) 2614–2623.

[99] A.L. Ponte, E. Marais, N. Gallay, A. Langonne, B. Delorme, O. Herault, P.Charbord, J. Domenech, Stem Cells 25 (2007) 1737–1745.

[100] B. Ruster, S. Gottig, R.J. Ludwig, R. Bistrian, S. Muller, E. Seifried, J. Gille, R.Henschler, Blood 108 (2006) 3938–3944.

[101] C. Steingen, F. Brenig, L. Baumgartner, J. Schmidt, A. Schmidt, W. Bloch, J. Mol.Cell. Cardiol. 44 (2008) 1072–1084.

[102] C. Ries, V. Egea, M. Karow, H. Kolb, M. Jochum, P. Neth, Blood 109 (2007)4055–4063.

[103] A. De Becker, P. Van Hummelen, M. Bakkus, I.V. Broek, J. De Wever, M. DeWaele, I. Van Riet, Haematol. Hematol. J. 92 (2007) 440–449.

[104] V.F. Segers, I. Van Riet, L.J. Andries, K. Lemmens, M.J. Demolder, A.J. De Becker,M.M. Kockx, G.W. De Keulenaer, Am. J. Physiol. Heart Circ. Physiol. 290 (2006)H1370–H1377.

[105] Y. Ozaki, M. Nishimura, K. Sekiya, F. Suehiro, M. Kanawa, H. Nikawa, T.Hamada, Y. Kato, Stem Cells Dev. 16 (2007) 119–129.

[106] W.J.C. Rombouts, R.E. Ploemacher, Leukemia 17 (2003) 160–170.[107] I.M. Barbash, P. Chouraqui, J. Baron, M.S. Feinberg, S. Etzion, A. Tessone, L.

Miller, E. Guetta, D. Zipori, L.H. Kedes, R.A. Kloner, J. Leor, Circulation 108(2003) 863–868.

[108] S. Schrepfer, T. Deuse, H. Reichenspurner, M.P. Fischbein, R.C. Robbins, M.P.Pelletier, Transpl. Proc. 39 (2007) 573–576.

[109] A.I. Caplan, S.P. Bruder, Trends Mol. Med. 7 (2001) 259–264.[110] M. Janowski, A. Lyczek, C. Engels, J.D. Xu, B. Lukomska, J.W.M. Bulte, P.

Walczak, J. Cereb. Blood Flow Metab. 33 (2013) 921–927.[111] J.F. Ge, L. Guo, S. Wang, Y.L. Zhang, T. Cai, R.C.H. Zhao, Y.J. Wu, Stem Cell Rev.

Rep. 10 (2014) 295–303.[112] G.S.L. Teo, J.A. Ankrum, R. Martinelli, S.E. Boetto, K. Simms, T.E. Sciuto, A.M.

Dvorak, J.M. Karp, C.V. Carman, Stem Cells 30 (2012) 2472–2486.

[113] R. Sackstein, J.S. Merzaban, D.W. Cain, N.M. Dagia, J.A. Spencer, C.P. Lin, R.Wohlgemuth, Nat. Med. 14 (2008) 181–187.

[114] H. Cheng, M. Byrska-Bishop, C.T. Zhang, C.J. Kastrup, N.S. Hwang, A.K. Tai,W.W. Lee, X.Y. Xu, M. Nahrendorf, R. Langer, D.G. Anderson, Biomaterials 33(2012) 5004–5012.

[115] C.Y. Lo, A. Antonopoulos, A. Dell, S.M. Haslam, T. Lee, S. Neelamegham,Biomaterials 34 (2013) 8213–8222.

[116] D. Sarkar, J.A. Spencer, J.A. Phillips, W. Zhao, S. Schafer, D.P. Spelke, L.J.Mortensen, J.P. Ruiz, P.K. Vemula, R. Sridharan, S. Kumar, R. Karnik, C.P. Lin,J.M. Karp, Blood 118 (2011) e184–e191.

[117] R. Sackstein, J. Invest. Dermatol. Symp. Proc. 9 (2004) 215–223.[118] C. Weber, L. Fraemohs, E. Dejana, Nat. Rev. Immunol. 7 (2007) 467–477.[119] L. Xia, J.M. McDaniel, T. Yago, A. Doeden, R.P. McEver, Blood 104 (2004) 3091–

3096.[120] S. D’Souza, H. Murata, M.V. Jose, S. Askarova, Y. Yantsen, J.D. Andersen, C.D.

Edington, W.P. Clafshenkel, R.R. Koepsel, A.J. Russell, Biomaterials 35 (2014)9447–9458.

[121] D. Sarkar, W. Zhao, A. Gupta, W.L. Loh, R. Karnik, J.M. Karp, Methods Mol. Biol.698 (2011) 505–523.

[122] M.D. Lacher, M. Provenzano, Recent Pat. Anti-Cancer Drug Discovery 8 (2013)239–254.

[123] J.E. Dennis, N. Cohen, V.M. Goldberg, A.I. Caplan, J. Orthop. Res. 22 (2004)735–741.

[124] I.K. Ko, T.J. Kean, J.E. Dennis, Biomaterials 30 (2009) 3702–3710.[125] I.K. Ko, B.G. Kim, A. Awadallah, J. Mikulan, P. Lin, J.J. Letterio, J.E. Dennis, J. Am.

Soc. Gene Ther. 18 (2010) 1365–1372.[126] S.A. Kim, J.S. Peacock, J. Immunol. Methods 158 (1993) 57–65.[127] G. Zheng, A. Chen, R.E. Sterner, P.J. Zhang, T. Pan, N. Kiyatkin, M.L. Tykocinski,

Cancer Res. 61 (2001) 8127–8134.[128] A. Chen, G. Zheng, M.L. Tykocinski, J. Immunol. 164 (2000) 705–711.[129] A.I. Caplan, Tissue Eng. 11 (2005) 1198–1211.[130] L.G. Lum, H. Fok, R. Sievers, M. Abedi, P.J. Quesenberry, R.J. Lee, Mol. Dis. 32

(2004) 82–87.[131] R.J. Lee, Q. Fang, P.A. Davol, Y. Gu, R.E. Sievers, R.C. Grabert, J.M. Gall, E. Tsang,

M.S. Yee, H. Fok, N.F. Huang, J.F. Padbury, J.W. Larrick, L.G. Lum, Stem Cells 25(2007) 712–717.

[132] W. Deng, Q.W. Chen, X.S. Li, H. Liu, S.Q. Niu, Y. Zhou, G.Q. Li, D.Z. Ke, X.G. Mo,Cytotherapy 13 (2011) 431–440.

[133] T.J. Kean, P. Lin, A.I. Caplan, J.E. Dennis, Stem Cells Int. 2013 (2013) 732742.[134] J. Wagner, T. Kean, R. Young, J.E. Dennis, A.I. Caplan, Curr. Opin. Biotechnol. 20

(2009) 531–536.[135] T.J. Kean, L. Duesler, R.G. Young, A. Dadabayev, A. Olenyik, M. Penn, J. Wagner,

D.J. Fink, A.I. Caplan, J.E. Dennis, J. Drug Target. 20 (2012) 23–32.[136] J.S. Merzaban, M.M. Burdick, S.Z. Gadhoum, N.M. Dagia, J.T. Chu, R.C.

Fuhlbrigge, R. Sackstein, Blood 118 (2011) 1774–1783.[137] M.M. Burdick, J.T. Chu, S. Godar, R. Sackstein, J. Biol. Chem. 281 (2006)

13899–13905.[138] P.P. Jacobs, R. Sackstein, FEBS Lett. 585 (2011) 3148–3158.[139] D.E. Hughes, K.R. Wright, H.L. Uy, A. Sasaki, T. Yoneda, G.D. Roodman, G.R.

Mundy, B.F. Boyce, J. Bone Miner. Res. 10 (1995) 1478–1487.[140] M. Guan, W. Yao, R. Liu, K.S. Lam, J. Nolta, J. Jia, B. Panganiban, L. Meng, P.

Zhou, M. Shahnazari, R.O. Ritchie, N.E. Lane, Nat. Med. 18 (2012) 456–462.[141] W. Yao, N.E. Lane, Bone 70 (2015) 62–65.[142] L. von Bahr, I. Batsis, G. Moll, M. Hagg, A. Szakos, B. Sundberg, M. Uzunel, O.

Ringden, K. Le Blanc, Stem Cells 30 (2012) 1575–1578.[143] M. Dominici, K. Le Blanc, I. Mueller, I. Slaper-Cortenbach, F. Marini, D. Krause,

R. Deans, A. Keating, D. Prockop, E. Horwitz, Cytotherapy 8 (2006) 315–317.[144] J. Li, H.K.W. Law, Y.L. Lau, G.C.F. Chan, Br. J. Haematol. 127 (2004) 326–334.[145] J. Cao, M.H. Tan, P. Yang, W.L. Li, J. Xia, H. Du, W.B. Tang, H. Wang, X.W. Chen,

H.Q. Xiao, Cancer Lett. 263 (2008) 197–203.[146] C.L. Grigsby, Y.P. Ho, K.W. Leong, Nanomedicine-UK 7 (2012) 565–577.[147] Y. Wu, Y.P. Ho, Y.C. Mao, X.M. Wang, B. Yu, K.W. Leong, L.J. Lee, Mol. Pharm. 8

(2011) 1662–1668.[148] P.M. Kharkar, K.L. Kiick, A.M. Kloxin, Chem. Soc. Rev. 42 (2013) 7335–7372.[149] S. Sundelacruz, D.L. Kaplan, Semin. Cell Dev. Biol. 20 (2009) 646–655.[150] M. Ahearne, Interface Focus 4 (2014) 20130038.[151] R.D. Cummings, M.E. Etzler, in: A. Varki, R.D. Cummings, J.D. Esko, H.H.

Freeze, P. Stanley, C.R. Bertozzi, G.W. Hart, M.E. Etzler (Eds.), Essentials ofGlycobiology, Cold Spring Harbor (NY), 2009.