lable at ScienceDirect

Biomaterials 85 (2016) 180e194

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

Transdifferentiation of human endothelial progenitors into smoothmuscle cells

HaYeun Ji a, 1, Leigh Atchison b, 1, Zaozao Chen a, Syandan Chakraborty a, Youngmee Jung d,George A. Truskey b, Nicolas Christoforou b, c, **, Kam W. Leong a, *

a Department of Biomedical Engineering, Columbia University, Mail Code 8904, 1210 Amsterdam Avenue, New York, NY 10027, USAb Department of Biomedical Engineering, Duke University, 136 Hudson Hall, P. O. Box 90281, Durham, NC 27708, USAc Department of Biomedical Engineering, Khalifa University, P. O. Box 127788, Abu Dhabi, United Arab Emiratesd Center for Biomaterials, Biomedical Research Institute, Korea Institute of Science and Technology, Seoul, South Korea

a r t i c l e i n f o

Article history:Received 3 November 2015Received in revised form23 January 2016Accepted 28 January 2016Available online 3 February 2016

Keywords:Direct transdifferentiationDirect reprogrammingSmooth muscle cell differentiationMyocardinTissue-engineered blood vessel

* Corresponding author.** Corresponding author. Department of Biomedicalsity, P. O. Box 127788, Abu Dhabi, United Arab Emira

E-mail addresses: nicolas.christoforou@kus(N. Christoforou), kwl2121@columbia.edu (K.W. Leon

1 Equally contributed to this work.

http://dx.doi.org/10.1016/j.biomaterials.2016.01.0660142-9612/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

Access to smooth muscle cells (SMC) would create opportunities for tissue engineering, drug testing, anddisease modeling. Herein we report the direct conversion of human endothelial progenitor cells (EPC) toinduced smooth muscle cells (iSMC) by induced expression of MYOCD. The EPC undergo a cytoskeletalrearrangement resembling that of mesenchymal cells within 3 days post initiation of MYOCD expression.By day 7, the reprogrammed cells show upregulation of smooth muscle markers ACTA2, MYH11, andTAGLN by qRT-PCR and ACTA2 and MYH11 expression by immunofluorescence. By two weeks, theyresemble umbilical artery SMC in microarray gene expression analysis. The iSMC, in contrast to EPCcontrol, show calcium transients in response to phenylephrine stimulation and a contractility an order ofmagnitude higher than that of EPC as determined by traction force microscopy. Tissue-engineered bloodvessels constructed using iSMC show functionality with respect to flow- and drug-mediated vasodilationand vasoconstriction.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Smooth muscle cells (SMC) are non-striated muscle cells foundin the walls of blood vessels and hollow organs including bladder,lung, and intestine [1]. They differ from other terminally differen-tiated cells such as cardiac and skeletal myocytes in that theymaintain plasticity in modulating from a contractile to a syntheticphenotype in response to various stimuli [2,3]. In the normalvasculature environment, SMC maintain their contractile pheno-type and show low proliferation and migration [4,5]. However,during development, post-vascular injury, or progression ofvascular disease, vascular SMC switch to a synthetic phenotype byincreasing the rate of cell proliferation, migration and syntheticproperties, while downregulating genes controlling the contractile

Engineering, Khalifa Univer-tes.tar.ac.ae, nc28@duke.edug).

process [6,7]. This phenotypic switching allows SMC to play a majorrole in vascular contraction, regulation of vessel tone and bloodpressure, as well as maintenance of a functional endothelium[8e10]. Therefore, SMC are an essential cell source for under-standing vascular disease mechanisms and for regenerative medi-cine applications.

Unfortunately, it is difficult to acquire a large number of SMCthrough current techniques involving biopsy and in vitro expansiondue to their limited proliferative capacity and quick senescenceacquisition [11]. To explore an alternative SMC cell source, previousstudies have used iPSC to differentiate to the SMC lineage [11,12].Although successful, this approach requiresmultiple differentiationsteps as well as purification from residual iPSC and alternate iPSC-derived cell lineages. Transdifferentiation offers an attractivealternative [13,14]. This process eliminates the need of goingthrough an intermediate pluripotent state, and therefore it is afaster and potentially more efficient method for acquiring a specificcell type. Direct transdifferentiation has already generated manydifferent types of cells including neuronal cells that would havedesirable implications for use with SMC [13,15].

For direct transdifferentiation to be effective, it is critical to

H. Ji et al. / Biomaterials 85 (2016) 180e194 181

select an easily obtainable cell source that can be quickly expandedin large quantities. In this sense, EPC derived from various bloodsources, including umbilical cord blood or peripheral blood, is anexcellent cell source for transdifferentiating into induced SMC [16].Blood can be easily extracted from patients, and the procedure forisolating EPC from whole blood sample is already well establishedin previous studies [17]. In addition, EPC proliferate quickly and canmaintain their endothelial phenotype even after multiple rounds ofreplication. SMC and EPC also make up the basic structure of bloodvessels, therefore their use in conjunction with each other is idealfor creating an autologous tissue-engineered blood vessel forpatient-specific application.

The direct transdifferentiation into SMC can be induced usingmyocardin (MYOCD), a master regulator of smooth muscle geneexpression [9]. Myocardin is a strong transcriptional co-activatorinvolved in activation of cardiac and smooth muscle related genesthrough interaction with serum response factor (SRF). There havebeen multiple studies that indicate myocardin plays an importantrole in regulating SMC development and differentiation [18e20].Wang et al. reported that myocardin expression in non-muscle cellsactivates smooth muscle related genes but not cardiac muscle-related genes [9]. Li et al. demonstrated that while myocardindeficient mutant embryos form proper cardiac development, theydie by day 10.5 from complete absence of vascular SMC [21]. Longet al. reported that myocardin induces structural and functionalattributes of a mature SMC with contractile phenotype in BC3H1myogenic cells [22]. In addition, Du et al. have demonstrated thatmyocardin expression is developmentally regulated in visceral andvascular SMC during embryonic development, and the forcedexpression of myocardin in undifferentiated mouse embryonicstem cells induces expression of SMC-restricted genes such asSM22a [23].

In this study, we demonstrate that a single transcriptional co-activator is sufficient to induce efficiently and quickly the directtransdifferentiation of human EPC into induced smooth musclecells (iSMC). We generated iSMC using a lentiviral gene deliverysystem allowing the inducible expression of MYOCD in EPC(Fig. 1A). Using immunofluorescence, flow cytometry, and micro-array gene expression analysis we confirmed the phenotypic con-version of iSMC from the EPC controls. We also measured thecalcium signaling activity and cell traction force to evaluate theirfunctional phenotypic characteristics. Finally we demonstrated theutility of these iSMC in the assembly of tissue-engineered bloodvessels (TEBV).

2. Results

2.1. MYOCD expression induces significant rapid phenotypicchanges in endothelial progenitor cells

Primary human endothelial progenitor cells (EPC) were stablytransduced using second-generation lentivirus with a constructallowing the inducible expression of the human transcriptional co-activator MYOCD. Following induction of MYOCD expressionthrough the exposure of transduced cells to doxycycline (DOX) wereadily detected nuclear localization of MYOCD within 2 days posttransduction as determined by immunofluorescence (Fig. 1B).MYOCD expression levels within the entire cell population wereinhomogeneous with some nuclei having high expression levels(white arrowheads) and some nuclei having no obvious signs ofMYOCD expression (yellow arrowheads). No MYOCD-positivenuclei were detected in the control cell population (M2rtTA only).Induction of MYOCD expression was associated with significantphenotypic changes in the transduced cells within 4 days (Fig. 1C).MYOCD-expressing EPC seemed to undergo endothelial-to-

mesenchymal transition as evident by a loss of their cobblestone-like shape (typical in endothelial cells) and conversion to elon-gated and spindle-like shape often seen in smooth muscle cells.Interestingly transduction of EPC with a higher virus titer to ensurethorough transduction of the target cell population was toxic withthe majority of cells undergoing apoptosis within 7 days post in-duction of MYOCD expression. Due to the potential cytotoxic effectof high MYOCD expression levels for the purpose of this work weonly used low virus titers when transducing human EPC. Further-more, temporal immunofluorescent staining against F-actin andvimentin (2 days prior to induction of MYOCD expression, 3 days,and 7 days post-induction of MYOCD expression) indicated agradual but significant cytoskeletal rearrangement of the cells un-dergoing transdifferentiation, which also suggested a transitionfrom an endothelial to a mesenchymal cytoskeletal phenotype(Fig. 1D).

2.2. Cells undergoing transdifferentiation downregulate endothelialprogenitor cell markers and express smooth muscle markers

We first used flow cytometry (FACS) to measure the expressionlevels of endothelial cell surface markers CD31 and CD105 on cellsundergoing transdifferentiation into induced smooth muscle cells(iSMC) via the induction of MYOCD expression (Fig. 2A). While EPCwere CD31 and CD105 double positive, 7 days of MYOCD expressioninduced a large and highly significant downregulation of both cellsurface markers. Although approximately 30% of the cellscontinued to express CD31, no significant expression of CD105 wasdetected. We hypothesize that the CD31 expression detected was aresult of EPC not transduced with the inducible lentiviral construct.

We next examined the relative expression of smooth musclegenes and endothelial progenitor cell genes via quantitative RT.PCRin the non-transduced EPC, EPC transduced only with M2rtTA, EPCtransduced with a low virus titer of MYOCD, and EPC transducedwith a high virus titer of MYOCD at day 7. MYOCD RNA expressionwas negligible and non-significant in the two control cell groups,confirming the previous immunofluorescence observations(Fig. 2B). MYOCD RNA levels were significantly upregulated (~150�and ~400�) relatively to the control EPC cell group in transducedcells. Moreover, induction of MYOCD expression was associatedwith a significant downregulation of RNAs encoding for endothelialmarkers CD31 and CDH5. Importantly, MYOCD expression for 7 daysinduced a significant upregulation of RNAs encoding for smoothmuscle genes ACTA2, MYH11, and TAGLN, further supporting ourhypothesis that the transcriptional co-factor induced EPC trans-differentiation into iSMC. Although high virus titers induced higherexpression levels of the three smooth muscle genes, their long-term cytotoxic effect prohibited their further use.

We then examined the protein expression and cellular organi-zation of endothelial markers CD31 and VWF, as well as the smoothmuscle markers TAGLN, MYH11, and ACTA2 in both iSMC andcontrol EPC (negative control and EPC transduced only withM2rtTA) (Fig. 2C). No significant MYH11 or ACTA2 expression wasdetected in control EPC whereas both proteins were readilydetected in transdifferentiated iSMC encompassing the entire cellarea. Interestingly we detected TAGLN expression in both EPC andiSMC although the protein was expressed at higher levels in iSMC(as determined by fluorescent intensity) and localized uniformlywithin the entire cell area rather than primarily close to the cellborders (EPC). As expected EPC stained positive for both CD31 andVWF whereas no VWF expressionwas detected on ACTA2þ iSMC orCD31 expression on TAGLNþ iSMC. Small colonies of EPC stainedpositive for either VWF, or CD31 was detected within the iSMCcultures, indicating there were cells that did not undergo trans-differentiation. The cell shape and area of the iSMC was

Fig. 1. Transdifferentiation of endothelial progenitor cells into induced smooth muscle cells through the transient overexpression of MYOCD. (A) Illustration depicting theexperimental design of our study. Human endothelial progenitor cells (EPC, CD31high/CD105high) were transduced with lentiviruses allowing the constitutive overexpression of areverse tetracycline-controlled transactivator (M2rtTA) and inducible overexpression of MYOCD. Induction of MYOCD overexpression via the exposure to doxycycline (DOX)stimulated morphological changes resembling that of the endothelial-to-mesenchymal transition. The reprogrammed cells (iSMC) resembled smooth muscle cells and expressedsmooth muscle actin (ACTA2) and smooth muscle myosin heavy chain 11 (MYH11). iSMC were characterized both morphologically and functionally. (B) EPC transduced with the twogenes (M2rtTA and MYOCD) expressed and localized MYOCD in their nucleus, following induction of expression via doxycycline exposure as determined by immunofluorescence(right images). MYOCD expression levels were determined to be inhomogeneous in the cell population as nuclear protein expression was either high (white arrows), low (redarrows), or absent (yellow arrows). No MYOCD expression was detected in control EPC transduced only with M2rtTA (left images). (C) EPC were first transduced with an M2rtTAlentivirus and subsequently with either low or high viral titers of MYOCD lentivirus. Induction of MYOCD overexpression (1 day) provoked a significant amount of cell death whichwas more evident in cells transduced with high viral titers of MYOCD. Following 7 days of MYOCD overexpression transduced EPC acquired a mesenchymal phenotype andcontinued to proliferate in the cell population transduced with a low titer of MYOCD lentivirus. Cells transduced with a high lentiviral titer did not recover. Control cells (M2rtTAonly) retained an endothelial morphology. (D) Epigenetic reprogramming of EPC into iSMC was associated with significant changes in the cytoskeletal organization of the cells (actinfilaments and intermediate filaments) as determined by immunofluorescence. EPC organized actin filaments on their periphery whereas iSMC contained actin filaments which transversed the entire length of the cell. EPC contained intermediate filaments in the entire cell area apart from the nucleus whereas in iSMC the intermediate filaments covered theentire cell. These changes were detectable as early as three days post induction of MYOCD overexpression and became more pronounced by day seven. (For interpretation of thereferences to color in this figure legend, the reader is referred to the web version of this article.)

H. Ji et al. / Biomaterials 85 (2016) 180e194182

Fig. 2. Phenotypic characterization of transdifferentiated iSMC. (A) Epigenetic reprogramming of EPC into iSMC was associated with a significant downregulation of endothelialcell surface markers CD31 and CD105 as determined by flow cytometry. (B) Exposure of transduced EPC to doxycycline induced a robust overexpression of MYOCD as determined bygene expression analysis. Transdifferentiation and iSMC derivation was associated with a significant downregulation of endothelial markers CD31 and CDH5, and a significantupregulation of smooth muscle markers ACTA2, MYH11, and TAGLN. Control EPC (i), M2rtTA (ii), M2rtTA þ MYOCDLow (iii), M2rtTA þ MYOCDHigh (iv). (C) EPC-to-iSMC trans-differentiation was associated with expression of smooth muscle proteins MYH11, ACTA2, and an upregulation/reorganization of TAGLN as determined by immunofluorescence.Endothelial proteins CD31 and VWF were absent in the transdifferentiated cells. Small foci of VWFþ/CD31þ EPC remained in cultures of iSMC which can be attributed to EPC nothaving be transduced with the MYOCD expressing lentivirus. Expression of CD31 or VWF by MYH11 or ACTA2 expressing iSMC was not detected.

H. Ji et al. / Biomaterials 85 (2016) 180e194 183

significantly different from that observed for the control EPC.

2.3. Global gene expression and molecular pathway analysissuggests that MYOCD expression stably induces the smooth musclemolecular phenotype

To further ascertain the overall effect that the induction of

MYOCD expression had on cells undergoing transdifferentiationweperformed microarray gene expression analysis on iSMC following2 weeks of continuous MYOCD expression (iSMC-2w) or 2 weeks ofcontinuous MYOCD expression followed by 2 weeks of cultureunder standard conditions (iSMC-4w). Control cells included non-transduced EPC and primary human smooth muscle cells isolatedfrom the umbilical artery (UASMC). To establish the purity of the

H. Ji et al. / Biomaterials 85 (2016) 180e194184

cell populations used for this set of experiments, prior to cell lysisand RNA isolation we performed flow cytometry on a cell sub-population and measured the endothelial cell surface markersCD31 and CD105 (Supplementary Fig. 1E). As expected approxi-mately 98% of EPC were double positive for CD31 and CD105 ascompared to only 0.03% of UASMC (UASMC were ~29% positive forCD105). iSMC at the 2 week time point were approximately 13%double positive for CD31 and CD105 whereas that percentagedecreased to 0.35% at the 4 week time point.

We first determined that iSMC at both time points and UASMCexhibited a variation in signal intensity for a number of probes(Fig. 3A). We selected probe sets exhibiting a significant level ofgene upregulation or downregulation (p-value: <0.05, fold change< or >1.5) in each of the three groups as compared to the control(Fig. 3B). By comparing the sets of either upregulated genes ordownregulated genes we identified 1239 commonly upregulatedgenes and 1823 commonly downregulated genes (Fig. 3C,Supplementary Table 1). We then performed hierarchical clusteringanalysis on all probes exhibiting significant upregulation ordownregulation in at least one of the cell groups that differed intheir level of gene expression patterns (Fig. 3D). We also performedpathway analysis on groups of genes that are significantly upre-gulated or downregulated either specifically in iSMC as comparedto EPC or iSMC and UASMC as compared to EPC (Fig. 3E,Supplementary Table 2). This analysis indicated that the upregu-lated genes were associated with pathways controlling smoothmuscle differentiation and function as well as pathways controllingspecific cell attachment to the extracellular matrix. On the otherhand the downregulated genes were associated with pathwayscontrolling various aspects of the cell cycle, suggesting that the cellsexited the cell cycle, a finding that is consistent with what weobserved in our cell cultures of iSMC. Importantly, a large array ofgenes associated with the smooth muscle contraction pathwaywere significantly upregulated in iSMC both at the 2 and 4 weektime point (Fig. 3F).

Finally, we performed a principal component analysis on theentire gene expression signature of our control and experimentalsamples (EPC, UASMC, iSMC) as well as control samples from pre-viously published studies: Skeletal Muscle [24,25], Brain [26], Liver[26], Umbilical Vein and Coronary Artery Endothelial Cells [27],Aortic Smooth Muscle Cells [28,29], Dermal Fibroblasts [30]. Thisanalysis confirmed that the transdifferentiated iSMC closelyresemble the control cells including adult aorta SMC and umbilicalartery SMC while at the same time being significantly differentfrom the starting EPC population (Fig. 3G).

To identify transcription factors associated with genes that aresignificantly upregulated or downregulated in the iSMC cell pop-ulation when compared to the EPC we used a publicly availablesource code PASTAA [31]. A set of unique transcription factorsassociated with either upregulated or downregulated genes wasdetermined (Supplementary Table 3). We also identified tran-scription factor-binding sites associated with the promoter andenhancer regions of upregulated and downregulated genes usingtheWEB-based GEne SeTAnaLysis Toolkit (Supplementary Table 4).

In addition, we examined the differences between iSMC at thetwo- and four-week time points (iSMC-2w Vs iSMC-4w). Using thesame selection criteria to identify genes with significant differencesbetween the two groups we determined that 486 genes wereupregulated and 507 genes were downregulated at the 4 weekstime point (Supplementary Fig. 1AeC, Supplementary Table 1).Pathway analysis suggested that downregulated genes were asso-ciated with pathways controlling cell attachment to the substrateand extracellular matrix, whereas upregulated genes were associ-ated with pathways controlling various aspects of the transcriptionprocess and particular mRNA processing (Supplementary Fig. 1D,

Supplementary Table 2).

2.4. Functional in vitro characterization of iSMC demonstrates theirsmooth muscle-like functional phenotype

To assess the functional phenotype of iSMC, we first evaluatedthe regulation of smooth muscle cell contractility by changes inintracellular calcium concentration. The calcium influx in cells inresponse tomolecular stimulators wasmonitored in real time usinga lentivirally-delivered genetically encoded calcium indicator (R-GECO) [32]. We determined the levels of calcium influx bymeasuring the fluorescence intensity change over time and con-verting that to RGB images corresponding to the time at peak in-tensity for selected iSMC at 2 weeks and 4 weeks post induction ofMYOCD expression as well as control EPC (Fig. 4AeC).

When incubated in Tyrode's solution no change in fluorescenceintensity was detected in all three cell groups. Potassium chloride(KCl), a chemical known to activate voltage-gated Ca2þ channels insmooth muscle cells [33], when administered locally via a smallvolume injection (100 ml, 50 mM) induced a small but significantincrease in fluorescence intensity in a small number of iSMC at the2 week time point (Fig. 4A). Although we did not detect a similareffect at the 4 week time point (Fig. 4B) when we increased theconcentration of KCl delivered (100 ml, 100 mM) we readily detec-ted a large and significant increase in fluorescence intensity(~300%, Supplementary Fig. 2). We did not detect any change influorescence in control EPC following exposure to KCl (Fig. 4C).Similarly, addition of 100 mMphenylephrine, a vasoconstrictor drug[34], induced a steep transient increase of 75.6% and 117% in thefluorescence intensity in iSMC at both the 2 week and 4 week timepoint, but not in control EPC.

The functionality of iSMC was also evaluated by traction forcemicroscopy [35,36]. We compared the traction forces produced incontrol EPC and iSMC (4 weeks). We used color-coding to displaythe magnitude of bead displacement in iSMC (Fig. 4E) and controlEPC (Fig. 4D). By measuring the displacement of fluorescent beadswithin the hydrogel substrate we determined that trans-differentiated iSMC induced a significantly higher substrate defor-mation than did EPC. The calculated cell traction force of singleiSMC was approximately 10 times higher than that generated byEPC, and comparable to that of native UASMC (Fig. 4F).

2.5. Functional characterization of iSMC in a collagen gel-basedtissue-engineered blood vessel (TEBV) configuration

The functionality of iSMC in a more physiologically relevant 3Denvironment was evaluated based on their vasoactive responses toknown stimuli in a tubular collagen construct that mimics the basicstructure of a small-diameter TEBV. This was achieved by assem-bling a collagen construct composed of iSMC as the vessel wall andEPC as the lumen (Fig. 5A). A construct consisting of EPC in both thewall and lumen was used as a negative control. We initially testedthe vessel response to changes in flow rate by measuring thechange in vessel diameter at the lowest and highest rates applied.Results showed that TEBV containing iSMC in the vessel walldilated 6.37 ± 0.64% in response to increasing flow rate from0.5 mL/min to 4 mL/min. The control TEBV containing EPC alsoshowed dilation, however, at significantly reduced levels (Fig. 5C).We then tested the ability of the TEBV to respond to the addition of1 mMphenylephrine, an a-adrenergic receptor agonist against SMC,in the perfusion by measuring the change in vessel diameter fromthe baseline flow rate without drug. The iSMC-TEBV constricted by2.0 ± 0.25% in response to phenylephrine, while EPC-TEBVresponded minimally. We also tested the ability of the TEBV torespond to the addition of 1 mM acetylcholine, a muscarinic

Fig. 3. Microarray gene expression analysis performed on four cell populations. (A) Plot of signal intensity ratios for individual chip probe when comparing iSMC (2 weeks postinduction of transdifferentiation, red), iSMC (4 weeks post induction of transdifferentiation, green), or control umbilical aorta smooth muscle cells (UASMC, blue) to control EPC. (B)Volcano plot displaying the relationship between the calculated fold change for individual chip probes versus the P-value as calculated using ANOVA statistical analysis (whencomparing iSMC 2 weeks, iSMC 4 weeks, or UASMC to EPC). Plot includes probes that are significantly upregulated or downregulated (Fold Change < or >1.5, P-value <0.05). (C)Venn diagrams displaying the numbers of common or unique genes that are either significantly upregulated or significantly downregulated when comparing each of the threegroups (iSMC 2 weeks, iSMC 4 weeks, UASMC) to control EPC. (D) Graphical representation of hierarchical clustering analysis performed on the union of all of the significantlyupregulated (Red, þ2.88) or significantly downregulated genes (Blue, �2.88). (E) Molecular pathways associated with significantly upregulated or downregulated genes whencomparing only iSMC to control EPC as determined by the WEB-based GEne SeT AnaLysis Toolkit (WebGestalt) [58]. The “Gene #” column refers to the number of identified genesthat belong to a particular pathway and the “P-value” column refers to the P-value of each of the pathways and based on the number of identified genes. (F) A calculated smoothmuscle contraction gene network based on known and predicted protein interactions (Co-expression, co-localization, pathway, physical interactions, predicted interactions, shareprotein domains) as determined by the GeneMANIA prediction server [59]. Black circles mark genes that are significantly upregulated in iSMC (2 and 4 weeks) as compared to thecontrol EPC. The table contains the level of gene upregulation and the P-value for each of the genes based on the ANOVA performed on all the data. Gray circles mark genespredicted to belong in the particular genetic network but found not to be significantly upregulated in iSMC as compared to EPC control cells. (G) Principal component analysisperformed on the normalized signal values for each of the chip probes as well as probes from previously published studies: Skeletal Muscle [24,25], Brain [26], Liver [26], UmbilicalVein and Coronary Artery Endothelial Cells [27], Aortic Smooth Muscle Cells [28,29], Dermal Fibroblasts [30]. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

H. Ji et al. / Biomaterials 85 (2016) 180e194 185

Fig. 4. Functional characterization of iSMC using calcium transient mapping and cell traction force mapping. (AeC) Qualitative and quantitative output of calcium transientwave mapping in iSMC at 2 weeks post initiation of transdifferentiation (A), iSMC at 4 weeks after doxycycline removal (B), and control EPC at 4 weeks (C). RGB images (left)highlight cells of interest (numbered circles) at the peak change in fluorescent intensity (RGECO-1) outlined on the graphs (dotted box). The graphs (right) show the change incalcium influx in the highlighted cells over time in response to the indicated substances as measured by the change in fluorescent intensity (RGECO-1) of the highlighted cells. Thebackground fluorescence is labeled as (�). (DeE) Elastic substrate traction mapping of an iSMC (D) and a control EPC (E). Images on the left are phase contrast microscopy images,middle images are bead displacement maps, and images on the right are the calculated constrained traction maps, where color bars indicate relative values. (F) Total cell tractionforce generated by EPC (n ¼ 12), iSMC (n ¼ 10), and UASMC (n ¼ 10). Error bars represent the standard deviation. There is a significant difference in total cell traction force betweenEPC and iSMC as well as EPC and UASMC, but there is no significant difference between UASMC and iSMC (#p < 0.001, Student's t-test). (For interpretation of the references to colorin this figure legend, the reader is referred to the web version of this article.)

H. Ji et al. / Biomaterials 85 (2016) 180e194186

receptor agonist against SMC and endothelial cells (EC), in theperfusion by measuring the change in the vessel diameter from thephenylephrine-mediated constricted state. Results showed thatiSMC-TEBV dilated by 1.9 ± 0.21% in response to acetylcholinewhile

the EPC-TEBV responded by an insignificant amount.Immunostaining showed a strong presence of smooth muscle-

specific contractile proteins such as ACTA2 and calponin (CNN1)[22] in iSMC-TEBV (Fig. 5B). While EPC-TEBV did show presence of

Fig. 5. Functional assessment of iSMC-TEBV. (A) Illustration depicting the overall process for assembling a TEBV. A dense collagen gel with iSMC (7 days post-transduction) ismolded into a hollow tubular construct using a syringe and a 1 mm diameter mandrel. The construct is then loaded/secured onto a perfusion chamber, and EPC are injected into thelumen of the construct to create an endothelium. The resulting TEBV is then matured under a continuous flow loop for up to 7 days before further assessment is performed. (B) Theexpression of contractile smooth muscle related genes, CNN1 and ACTA2, was detected by immunofluorescence in iSMC-TEBV (upper row), while the expression was minimal in thecontrol EPC-TEBV (lower row). VWF, a major endothelial marker, was detected in both iSMC-TEBV and control EPC-TEBV, implying the presence of endothelial cells. (CeE) The setupof a perfusion system containing the TEBV. (C) A TEBV of approximately 2.5 cm length is loaded within a perfusion chamber. (D) A continuous flow from the media reservoir into thechamber is delivered using a peristaltic pump. (E) An AMScope connected to video capture software is used to record the vessel diameter change in TEBV for functional assessment.(F) Functional vasoactivity of the TEBV was assessed by measuring the vessel diameter change in response to the addition of vasoactive drugs and increased flow rate. iSMC-TEBVshowed predicted vasoconstriction and vasodilation response to phenylephrine and acetylcholine, respectively, while the response was minimal in control EPC-TEBV. The vaso-dilation response to increased flow rate was higher in iSMC-TEBV than that of EPC-TEBV. N ¼ 3. Error bars are the standard error mean (One * for p-value <0.05, Two * for p-value<0.01, Student's t-test).

H. Ji et al. / Biomaterials 85 (2016) 180e194 187

ACTA2, there was no presence of calponin, which is a definitivemarker for contractile smooth muscle cells. Previous studies haveindicated that ACTA2 is present in both EC and SMC. Both TEBVconstructs stained positive for VWF, further indicating the presenceof an endothelium.

3. Discussion

We have developed a robust in vitro transdifferentiation systemto convert human endothelial progenitor cells into smooth musclecells via induced expression of the human transcriptional co-activator MYOCD using a lentiviral gene delivery system. Activa-tion of MYOCD expression induced quick phenotypic changes in the

H. Ji et al. / Biomaterials 85 (2016) 180e194188

EPC cell population similar to those observed during theendothelial-to-mesenchymal transition. Concomitantly with thephenotypic changes we observed a downregulation of endothelialmarkers (CD31, CD105, CDH5) and significant upregulation ofsmooth muscle cell markers (ACTA2, MYH11, TAGLN). Usingmicroarray gene expression analysis we determined that thederived iSMC activated pathways controlling smooth muscle celldifferentiation, smooth muscle contractility, and specific attach-ment to the extracellular matrix, and at the same time deactivatedpathways associated with cell cycling. Using calcium imaging wedemonstrated that iSMC responded appropriately when exposed toeither KCl or a vasoconstrictor drug. Using traction force micro-scopy we demonstrated a significant increase in the level ofcontraction force exerted by iSMC as compared to the control EPC.Finally, we applied iSMC to generate a TEBV with functional vaso-active responses.

The assembly of functional human TEBV would benefit from theinclusion of human SMC as they are a major player in native bloodvessels. Unfortunately human SMC are a difficult cell source toaccess and have limited proliferation capacity. Several groups haveattempted to bypass this issue by deriving SMC-like cells fromhuman embryonic stem cells (ESC) or human iPSC which can bereadily expanded in culture prior to their directed differentiation.Such attempts require the development of efficient smooth muscledifferentiation protocols and the derivation of SMC that closelyresemble their in vivo counterparts in order to closely replicatevascular function using the TEBV.

In a first such report Lee et al. described the generation of hu-man iPSCs from human aortic vascular smooth muscle cells(HAVSMC) and their subsequent differentiation into smoothmusclecells via the formation of embryoid bodies [37]. The differentiatedSMC have a similar morphology to that of the primary HAVSMC,expressed ACTA2, and showed a relative change in intracellularcalcium concentrations in response to membrane depolarizationexposure to a vasoconstrictor drug. Alternatively, Bajpai et al. re-ported on the iPSC derivation of functional human SMC via thedifferentiation through a mesenchymal stem cell intermediate. Thederived SMC both expressed and stained positive for ACTA2, CNN1,and MYH11. The authors also tested the contractile function ofderived SMC through the fabrication of small-diameter fibrinhydrogel-based cylindrical tissue constructs and determined thatthe cells exhibited high contractility levels in response to receptor-and non-receptor-mediated agonists [12]. In an elegant studyCheung et al. were able to generate distinct vascular SMC subtypesby first differentiating human ESCs or iPSCs into either neural crest,somite mesoderm, or lateral plate mesoderm and then inducingthem to differentiate into SMC via exposure toTGF-b1 and PDGF-BB[38]. Importantly, their protocols allowed for a high differentiationefficiency and SMC purity, and the derived SMC were highly func-tional. In a more recent study, Wang et al. described the differen-tiation of human iPSCs into proliferative SMC and their subsequentexpansion. They were able to induce the cells to assume a con-tractile phenotype, and following SMC seeding into a macroporousand nanofibrous PLLA scaffold they showed that the SMC retainedtheir differentiated phenotype and formed vascular tissuefollowing subcutaneous implantation in nude mice [11]. However,the differentiation protocol was at least 50 days long and the cellpurity was 76.5%.

A common denominator of the iPSC-directed differentiation intoSMC is first the necessary derivation, expansion, and detailedcharacterization of the iPS cells, which can take between 30 and 60days. Moreover, there is a requirement for a prolonged differenti-ation protocol which in the best case scenario can last up to 20 days[38]. Finally, there is the requirement of ensuring purity of the SMCpopulation as well as complete elimination of remaining

undifferentiated cells (ESC or iPS), which can form teratomas ifapplied in vivo. On the other hand, the protocol established in thisstudy has the advantages that 1) EPC can easily be obtained fromvarious blood sources including umbilical cord blood or peripheralblood [16,17]; 2) EPC can proliferate rapidly while maintaining theirendothelial phenotype; and 3) the transdifferentiation of EPC intoiSMC is efficient and quick, taking as few as 14 days to acquire iSMCwith phenotypic characteristics comparable to those of iPSC-SMCreported in the literature.

An important finding of this work is the capacity of MYOCD toactivate smooth muscle-specific genes and ultimately induce thetransdifferentiation of EPC into iSMC. The transcriptional co-activator MYOCD was first described by Wang et al. who used abioinformatics-based in silico screen to identify cardiac-specificgenes. They determined that MYOCD belongs to the SAP domainfamily of nuclear proteins and acts by associating with the serumresponse factor (SRF) in order to activate cardiac muscle specificgenes [39]. In subsequent work the same group determined thatMYOCD is a master regulator of smooth muscle gene expressionand is capable of activating smooth muscle expression in variousnon-muscle cell types by associating with SRF [9]. In a more recentstudy it was reported that forced MYOCD expression in human ESCderived-embryoid bodies only selectively regulated CArG-dependent smooth muscle genes and allowed an increase in thenumber of SMC-like cells derived following one month of differ-entiation [40]. Altogether, these reports alignwith our findings thattransient over-expression of MYOCD can convert EPC to SMC.

A hallmark of smooth muscle cell function is its ability to con-tract in response to various vasoreactive drugs such as phenyl-ephrine. This contractile response is regulated by transient receptorpotential cation channels that control the intracellular levels ofcalcium and thereby regulate cell contractility [41]. Our data indi-cate that iSMC responded to phenylephrine in a similar manner toprimary smooth muscle cells. Previous reports show that renalarterial smooth muscle cells initially respond with a sharp peak ofcalcium influx after addition of 10 mM phenylephrine and then ashort plateau before returning to baseline [42]. This peak responseis prevalent in the majority of our measured iSMC at similar levels.However, we did not see a sustained plateau after the initial cal-cium influx. Furthermore, iSMC at week-4 time point responded toKCl only at a higher concentration (100 mM) compared to that ofweek 2 (50 mM). Although this concentration is still within therange tested by previous studies [33], higher stimulation mighthave been necessary due to the reduced level of contractile proteinexpression, which is also confirmed by our microarray analysis,caused by the removal of doxycycline induction after week 2. Theseresults imply incomplete maturation or variation in the type ofsmooth muscle cell derived from our transdifferentiation method.The decrease in plateau presence in iSMC response betweenweek 2andweek 4may also allude to this maturation inconsistency as wellas the need to apply higher concentrations of phenylephrine(100 mM) in order to obtain such a response. Furthermore, studieswith pulmonary aortic smooth muscle cells show a sharp peak inresponse to phenylephrine that oscillates over longer time periods[43]. Interestingly enough, we did see oscillation of this calciumtransients throughout our cells as well as between cells, indicatingpotential cellecell communication, which is important for theregulation of vascular tone [44].

Adherent cells constantly generate traction forces via actino-myosin interactions with the actin cytoskeleton. These forces aretransmitted through stress fibers to focal adhesions that are linkedto the extracellular matrix [45]. Importantly, different cell typesgenerate varying levels of contractile force and the force amplitudeclosely correlates with particular cellular physiological functionsrequiring force generation, including contraction, adhesion,

H. Ji et al. / Biomaterials 85 (2016) 180e194 189

migration, and ECM reorganization [45e47]. We determined in thisstudy that the cell traction forces of iSMC is significantly higherthan that generated by control EPC, and comparable to that ofnative UASMC. As one of themajor types of force-generating cells inhuman body, SMC generate much larger contraction forces thanEPC. Thus, the significant increase in cell contraction force of iSMCdemonstrates a functional change compared to EPC, which could beattributed, but not limited, to an increase in ACTA2 level. Chen et al.previously reported that overexpression of ACTA2 in fibroblastssignificantly increased the cell contraction force. Their findingssuggest that the amount of ACTA2 correlates with cell traction force[46]. Here, we determined that iSMC have an elevated ACTA2expression (8 � 104 fold), accompanied by a significant increase incell traction force (10 fold). This increase in contractility also sug-gests that the iSMC have reached a level of maturity comparable tothat of iPSC-SMC. Bajpai et al. previously reported that smoothmuscle bundles madewith Stage-3-hiPSC (differentiated functionalSMC) have a 12-fold increase in cell contractile force over thosemade with Stage-2-hiPSC (immature stage), which shows a similarlevel of increase as we observed here [12].

Our pilot study of characterizing the iSMC in a 3D dense collagengel construct further demonstrates the functionality of the trans-differentiated cells. The iSMC-TEBV displayed contractile SMCprotein expression, and responded to the vasoactive drugs phen-ylephrine and acetylcholine, in contrast to the control EPC-TEBV.These results are consistent with the literature that shows SMCcontract in response to phenylephrine, and the addition of acetyl-choline can recover this constricted state by stimulating the pro-duction of nitric oxide from the healthy endothelium to induce SMCrelaxation [48]. In addition, our results suggest that iSMC-TEBVdilated in response to increased flow rate at much higher levelthan EPC-TEBV.While EPC-TEBV also dilated, this may be due to theexpansion of the collagen construct in which the cells areembedded, and the greater increase in dilation seen from the iSMC-TEBV can be attributed to the iSMC response to increasing flowrates, which also triggers endothelial production of nitric oxide.Altogether, these results highlight the functionality of the iSMC in aTEBV construct as well as the presence of a healthy endothelium.

We have only compared the iSMC-TEBV with the control EPC-TEBV in this study. Future studies should include the comparisonof iSMC-TEBV with TEBV generated from other cell types, such asfibroblasts [49] or mesenchymal stem cells (MSC) [50], for a betterassessment of iSMC as a suitable cell source for TEBV generation. Inaddition, further optimization of TEBV assembly involving the us-age of iSMC, such as optimal cell density, perfusion flow pattern,and maturation time, would be needed. Such optimization willallow for the use of these TEBV in a variety of applications related tocardiovascular disease and drug screening.

While drug-induced vascular injury can greatly influence thetoxicological evaluation of drug candidates under development, theexact mechanisms that cause smooth muscle and endothelial cellinjury have not been studied in detail. TEBV that mimic the mediallayer of smooth muscle cells and a luminal layer of endothelial cellsin small diameter blood vessels can be of great use to model suchdrug-induced vascular injury [51].

Previous generation of TEBV constructs have relied on the use offibroblasts or MSC to mimic the medial layer, as SMC are difficult toisolate and expand in the quantity necessary for generating a tissueconstruct. Nevertheless these TEBV contain a functional endothe-lium, and have effectively simulated the cytokine triggered endo-thelial activation and the subsequent adhesion of inflammatorycells onto the endothelium [52]. A healthy functioning endotheliumis critical for regulating the vascular tone and growth, and itsdysfunction can lead to cardiovascular disease such as hypertensionor coronary artery disease [53]. Thus, an endothelialized TEBV

would be important for the accurate study of the mechanisminvolved in endothelial activation and related inflammatory pro-cess during vascular injury [54].

However, the medial layer in MSC and fibroblast TEBV con-structs cannot accurately mimic the increased proliferation of SMCthat occurs after vascular injury. This phenotype switching isunique to SMC and is closely associated with vascular diseases suchas atherosclerosis [8]. Furthermore, SMC-EC interaction plays amajor role in maintaining such proliferative capacity of SMC andultimately regulating the vascular tone. For instance, the endo-thelium produces vasoconstrictor substances that modulate theproliferation of SMC [48]. Therefore, SMC-EC interaction is impor-tant for establishing a functional endothelium. In this regard, pre-vious TEBV designs have been limited by their cell source toaccurately model injury to the vasculature in terms of smoothmuscle and endothelial cells. Therefore, the usage of iSMC withEPCs for TEBV generation described in this study can serve as anattractive vascular injury model that overcomes previous limita-tions. iSMC-TEBV have the potential to more accurately depict thedisease phenotype and progression by incorporating the native celltype. Moreover, our iSMC express the contractile phenotype afterderivation, which is evident in their expression of MYH11 andfunctional behavior in response to vasoconstrictors. This is bene-ficial for baseline modeling of the healthy arteriole phenotype seenin vivo and could possibly be manipulated for disease modeling.

In addition, our transdifferentiation method of quickly derivinga large source of smooth muscle cells from an easily acquired cellsource has opened up many doors for the use of these cells in avariety of other applications, such as tissue engineering, geneediting, and disease modeling, that were not previously practicalwith primary smooth muscle cell sources.

In terms of tissue engineering, iSMC have the potential to helpfabricate multiple organ structures due to the broad presence ofsmooth muscle cells in the body, including, but not limited to gut,lung, esophagus, bladder, and blood vessels. Many of these struc-tures are primarily composed of SMC and EPC. Therefore, both cellsources used in our method can be effectively applied for patient-specific studies of these engineered organs using a reduced num-ber of primary cell sources, which tend to be much more difficult toacquire.

In conclusion, we have demonstrated a fast method of gener-ating SMC from human EPC through transdifferentiation using asingle transcriptional activator: MYOCD. The derived iSMC displaycharacteristics of SMC in terms of their phenotype and geneexpression profile, and show functionality in terms of contractileresponse when exposed to appropriate vasoactive stimuli both in2D and 3D cultures. Finally, incorporation of iSMC in a TEBVconstruct suggests that iSMC can be an excellent cell source formodeling drug-induced vascular injury and studying patient-specific disease mechanisms.

4. Materials and methods

4.1. Cell culture

Umbilical cord blood was obtained from the Carolina Cord BloodBank and all patient identifiers were removed prior to receipt. Theprotocol for the collection and the use of human blood for thisstudy was previously approved by the Duke University InstitutionalReview Board. Human umbilical cord blood derived endothelialprogenitor cells (EPC) were derived as previously described [17,55].Briefly, blood was diluted 1:1 with Hanks Balanced Salt Solution(HBSS, Invitrogen) and placed into Histopaque 1077 (Sigma). Themixture was then centrifuged at 740g for 30 min. Blood mono-nuclear cells (BMCs, buffy coat) were collected and washed three

H. Ji et al. / Biomaterials 85 (2016) 180e194190

times with complete endothelial cell growth medium containingEBM-2 (Lonza), EGM-2 supplement (Lonza), 8% fetal bovine serum(FBS, Atlanta Biologicals) and 1% antibiotic/antimycotic solution(Life Technologies). The BMCs were subsequently plated on cellculture plastic pre-coated with Collagen Type I (Rat tail, BD Bio-sciences) in complete endothelial cell (EC) growth medium. Me-dium was changed daily during the first week of culture. Coloniesappeared within 7e10 days following initial plating and weresubsequently passaged onto additional Collagen Type I pre-coatedcell culture plastic for expansion.

The derived human EPC were maintained in T-75 cell cultureflasks in EBM-2 supplemented with 1% penicillin/streptomycin,EGM-2 supplement, and 10% FBS. Media was changed every otherday. Cells were passaged with a ratio of 1:10 upon confluence andpassage 1 was defined as the first passage following confluence ofthe initially derived EPC. Cells were passaged up to passage 4 andfrozen down for later use.

Unless stated otherwise, EPC and induced smooth muscle cells(iSMC) were cultured in EBM-2 (Lonza), supplemented with EGM-2(Lonza), 10% fetal bovine serum (FBS, Atlanta Biologicals), and100U/ml Penicillin/Streptomycin (Life Technologies). Human um-bilical artery smoothmuscle cells (UASMC, Lonza) weremaintainedin DMEM-LG (Sigma), supplemented with 3% FBS (Atlanta Bi-ologicals) and 100U/ml Penicillin/Streptomycin (Life Technologies).2 mg/ml doxycycline (Sigma) was added to the culture medium forthe inducing transgenic MYOCD expression. The culture mediumwas changed every two days for all culture conditions except dur-ing lentiviral transduction when the medium was changed after24 h.

4.2. Plasmid construction and lentivirus preparation

DNA plasmid vectors were prepared as previously described[56]. Briefly the MYOCD DNA vector (Homo sapiens myocardin(MYOCD), transcript variant 1, mRNA, NCBI Reference Sequence:NM_001146312.2) was designed for the production of lentiviralparticles that allows the inducible expression of the proteinmolecule. We used a previously described target lentiviral vectorwhich controlled the expression of OCT4 [57] (Addgene plasmid19778, FU.tet.on.OCT4) and proceeded to remove the OCT4 geneand replace it withMYOCD. The fully sequencedMYOCD cDNA clone(BC126307, MHS4426-99239581, Open Biosystems) was clonedinto the FU.tet.on plasmid using SpeI/ blunt (50) and NotI/ blunt(30).

To produce viral particles we used a previously described secondgeneration lentivirus production system utilizing the psPAX2(packaging vector, Addgene plasmid 12660) and pMD2.G (envelopevector, Addgene plasmid 12559) vectors (Dr. Didier Trono). Wefollowed a procedure as previously described [56]. Briefly HEK293Tcells were maintained in a standard growth medium and expandedin T-75 tissue culture flasks pre-coated with a 0.1% solution ofporcine gelatin (Sigma). Cells were allowed to reach 90% confluenceat which point theywere transfected in the presence of Opti-MEM®

(Life Technologies) with a total of 24 mg of the three lentiviralvectors (12 mg expression vector, 7.7 mg of psPAX2 and 4.3 mg ofpMD2.G) using Lipofectamine 2000 (Life Technologies). The su-pernatant containing the viral particles was collected at 48 and 96 hfollowing initial transfection with a final volume of 20 ml. The su-pernatant was subsequently concentrated to a final volume ofapproximately 300 ml using Amicon Ultra-15 centrifugal filterunites (Millipore) and stored at 4 �C for immediate use or in smallaliquots at �80 �C for long term use. To transduce the human EPC,the cells were plated at a density of approximately 10,000 cells/cm2

is 6-well plates. The next day 2 ml of growth medium containing5 ml of each viral concentrate (FUW.M2rtTA, FU.tet.on.MYOCD) and

8 mg/ml sequabrene (Sigma) was used to transduce the cells.

4.3. Immunofluorescence and flow cytometry

Fluorescent cell imaging was performed on either a NikonEclipse TE2000-U using a Roper Scientific CoolSnap HQ camera andthe NIS Elements software suite, a Zeiss 510 inverted confocal mi-croscope, or an Olympus IX51 inverted fluorescent microscope(Olympus America Inc., Melville, NY, USA) at 20� magnificationusing HCImage software. Primary antibodies used for immunoflu-orescence included: anti-MYOCD (Santa Cruz, sc-21561), anti-Vimentin (Sigma, C9080), anti-TAGLN (Abcam, ab14106), anti-MYH11 (Abcam, ab683), anti-ACTA2 (Abcam, ab7817), anti-VWF(Abcam, ab6994), anti-CD31 (Life Technologies, 37-0700). Second-ary antibodies used were raised in either chicken or goat againstmouse or rabbit antibodies. They were purchased from Life Tech-nologies and conjugated to Alexa Fluor® 568 (red), Alexa Fluor® 488(green), or Alexa Fluor® 350 (blue). Cell nuclei were detected usingDAPI (Life Technologies, D1306). The actin cytoskeletonwas stainedand imaged using phalloidin conjugated to an Alexa Fluor® 488(Life Technologies). Cells were fixed using paraformaldehyde (2% inPBS, Electron Microscopy Sciences) and permeabilized using TritonX-100 (0.2% in PBS, Sigma). Blocking solution included FBS (10%).Flow cytometry analysis was performed on a FACSCanto flow cy-tometer (Duke Core Facility, BD Biosciences). Antibodies used were:CD105-APC (Biolegend, 323207), CD31-APC (Biolegend, 303115),Mouse IgG1,k Isotype control-APC (Biolegend, 400119). Meta-analysis of acquired data was performed using Cyflogic (www.cyflogic.com) and Microsoft Excel.

4.4. Gene expression analysis

Primers were designed using NCBI primer-BLAST. In order toavoid polymerization of non-specific DNA amplimers, whenapplicable primers were required to span an exoneexon junctionand the primer pair was required to be separated by at least oneintron on the corresponding genomic DNA. Total RNAwas collectedusing the RNeasy Mini Kit (Qiagen). Quantitative RT.PCR analysiswas performed on an 7900HT real time thermocycler using theQuantiTect SYBR Green one-step RT.PCR kit (Qiagen, 204243). Weincluded the option of getting a dissociation curve and also run thefinal reaction product on agarose gels in order to ensure that theonly amplimers detected and measured were the expected ones.We used the SDS software (ABI, version 1.4 or 2.4) to analyze theraw data and then additional analysis was performed on MicrosoftExcel. Relative quantification was performed using the DDCtmethod and statistical significance was determined using the T-Test. The primers used for this experiment are the following:MYOCD_Fw: CAAGCCAAAGGTGAAGAAGC, MYOCD_Rv: TAGCT-GAATCGGTGTTGCTG, CD31_Fw: CCAAGCCCGAACTGGAATCT,CD31_Rv: CACTGTCCGACTTTGAGGCT, CDH5_Fw: ATGCGGCTAGG-CATAGCATT, CDH5_Rv: TTTCCTGTGGGGGTTCCAGT, ACTA2_Fw:GCCAAGCACTGTCAGGAATC, ACTA2_Rv: GTCACCCACGTAGCTGTCTT,MYH11_Fw: AGTATCACGGGAGAGCTGGA, MYH11_Rv:ATGTCCTCTCGTCTCTGGCT, TAGLN_Fw: GTCTTCACTCCTTCCTGCGA,TAGLN_Rv: CTTGCTCAGAATCACGCCAT, ACTB_Fw: ACA-GAGCCTCGCCTTTGCCGAT, ACTB_Rv: CATGCCCACCATCACGCCCTG.

4.5. Microarray gene expression analysis

EPC were seeded at a density of 20,000 cells/cm2 and weresubsequently induced to transdifferentiate via MYOCD over-expression 24 h post initial seeding as previously described. Controlcell groups (wild-type EPC or UASMC) were similarly prepared(20,000 cells/cm2). 2 weeks post initiation of EPC

H. Ji et al. / Biomaterials 85 (2016) 180e194 191

transdifferentiation, cells were enzymatically dissociated and1 � 106 cells from each sample were used for flow cytometry. TotalRNA was collected from the remaining of dissociated cells (RNeasyMini, Qiagen) for iSMC, EPC, UASMC (n ¼ 3). This was repeated foriSMC at the 4-week time-point (n ¼ 3). One week post initiation oftransdifferentiation the culture medium used to maintain thederived iSMC was switched to SmBM (supplemented with SmGM-2, Lonza) and kept the same for the remainder of the experiment.Doxycycline was added to the iSMC culture medium only for theinitial 2 weeks.

Cells collected for flow cytometry analysis were incubated withanti-CD31 (PE conjugated, BD Pharmingen™) and anti-CD105 (APCconjugated, BD Pharmingen™) antibodies according to the manu-facturer's protocol. The control baseline for the analysis wasestablished using cells which were pre-incubated with mouseIgG1,k isotype control (conjugated to either PE or APC). Analysiswas performed using a FACSCanto II flow cytometer and theFACSDiva software package (BD Biosciences).

Total RNA was assessed for quality with Agilent 2100 Bio-analyzer G2939A (Agilent Technologies, Santa Clara, CA) andNanodrop 8000 spectrophotometer (Thermo Scientific/Nanodrop,Wilmington, DE). Hybridization targets were prepared with Mes-sageAmp™ Premier RNA Amplification Kit (Applied Biosystems/Ambion, Austin, TX) from total RNA, hybridized to GeneChip® Hu-man Genome U133A 2.0 arrays in Affymetrix GeneChip® hybridi-zation oven 645, washed in Affymetrix GeneChip® Fluidics Station450 and scanned with Affymetrix GeneChip® Scanner 7G accordingto standard Affymetrix GeneChip® Hybridization, Wash, and Stainprotocols. (Affymetrix, Santa Clara, CA). This work was performedat the Duke University microarray core facility.

Data analysis was performed as previously described [56].Briefly using the Partek Genomics Suite raw data (CEL files) wasimported and normalized using the RMA algorithm. SubsequentlyANOVA statistical analysis was performed on the entire data setsearching for significant differences between any one of the threegroups (iSMC 2 weeks, iSMC 4 weeks, UASMC) and the negativecontrol (EPC). Subsequently significantly upregulated or down-regulated genes were identified based on the fact that p-value<0.05 and Fold Change < or >1.5. Hierarchical clustering analysisand principal component analysis were performed within thesoftware suite itself. Identification of molecular pathways associ-ated with significantly upregulated or downregulated genes wasperformed using the WEB-based GEne SeT AnaLysis Toolkit (Web-Gestalt) [58]. Derivation of the molecular pathway associated withsmooth muscle contraction and based on the set of upregulatedgenes in the iSMC cell population was done using the GeneMANIAprediction server [59]. All the raw data can be accessed in the NCBIGene Expression Omnibus (GSE73469). Principal componentanalysis was performed with additional control data files frompreviously published studies that were uploaded on NCBI GeneExpression Omnibus or EMBL-EBI Array Express: Skeletal Muscle(E-GEOD-31243, E-GEOD-36297), Brain (E-TABM-1091), Liver (E-TABM-1091), Umbilical Vein and Coronary Artery Endothelial Cells(E-GEOD-10804), Aortic Smooth Muscle Cells (E-GEOD-29955, E-GEOD-59671), Dermal Fibroblasts (E-GEOD-34309). Data analysisfor the identification of transcription factor binding sites for genesthat are either significantly upregulated or downregulated in theiSMC cell population as compared to EPC was performed using theWEB-based GEne SeT AnaLysis Toolkit (WebGestalt) [58]. Dataanalysis for the identification of transcription factors associatedwith the same lists of genes was performed using PASTAA [31].

4.6. Calcium signal detection

Human EPC were seeded into four 10-cm cell culture dishes at a

density of 20,000 cells/cm2 per plate. 24 h post seeding, three of thefour plates were transduced with a lentivirus allowing the consti-tutive overexpression of M2rtTA. 24 h post-transduction, all plateswere passaged into fifty 35 mm plates at a density of 1 � 105 cellsper plate: wild type/not transduced EPC (10 plates) and M2rtTAtransduced EPC (40 plates). 24 h post passaging, twenty of theM2rtTA transduced plates were also transduced with a lentivirusallowing the inducible overexpression of MYOCD. All transductionswere performed in complete EGM medium with 8 mg/ml ofsequebrene (Sigma) to enhance transduction efficiencies. Inductionof MYOCD expression was initiated 24 h post transduction using2 mg/ml of doxycycline (Sigma). Five days post initiation of MYOCDoverexpression all plates were transduced with a lentivirus allow-ing the overexpression of a genetically encoded calcium indicator(RGECO-1) [32]. Calcium imaging was first performed on day 10post initiation of MYOCD overexpression. Moreover, the culturemedium was switched to SmBM (Lonza) supplemented withSmGM-2 (Lonza) for half the plates of each condition and doxycy-cline was removed. Calcium imaging was also performed on day 17day post initiation of MYOCD overexpression.

For calcium transient detection, the 35 mm plates were moun-ted onto a Nikon Eclipse TE2000 Inverted Microscope and main-tained in an environmental chamber set to 37 �C. Cells wereincubated in 37 �C Tyrode's solution (135 mM NaCl, 5.4 mM KCl,5 mM HEPES, 5 mM D-Glucose, 0.33 mM NaH2PO4, 1.8 mM CaCl2and 1 mM MgCl2; pH 7.4 and 290 mOsmol) during all imaging andstimulation steps. Cells in each plate were stimulated by directaddition of a drop of Tyrode's solution (mechanical control), 50 mMNaCl (negative control), 50e100mMKCl,10 mMphenylephrine, and100 mM phenylephrine, with an average of 50 s interval betweeneach stimulus. For each plate, the intensity of the RGECO-1 fluo-rescence signal was recorded following exposure to each stimulusand it was subsequently compiled into a single video file using theNikon NIS Elements software.

To determine the relative change in calcium concentrationwithin the stimulated cells we analyzed the recorded video filesusing ImageJ. The area of selected responsive cells and the back-ground was individually highlighted within each video file and thefluorescence intensity of the highlighted areas at each frame, rep-resented by the ‘mean gray value’ parameter in ImageJ, weremeasured throughout the time course of the video. Initial back-ground of all highlighted areas was averaged to determine theinitial background fluorescence (F), and the fold change in fluo-rescence intensity was calculated for each cell by subtracting theintensity at each time point from the initial fluorescence (DF/F).RGB images were created using ImageJ to highlight peak fluores-cence in concordance with fluorescent intensity graphs.

4.7. Traction force microscopy

Polyacrylamide substrate was prepared with previously pub-lished protocols [60,61]. Briefly, the fabrication of polyacrylamidesubstrate involves the following steps: (1) Glass surface modifica-tion: 0.5 ml of 0.1 M NaOH was added to the center of the 35 mmglass bottom dish (uncoated, no.0, MatTek) and kept for 20 min.NaOH solution was aspirated and dishes were air-dried for an hour.Glass coverslips (18 mm, VWR) were immersed in repel saline (GE)for 10 min and washed with 70% EtOH and later with distilledwater. (2) Silanization: the 35 mm glass bottom dish was silanizedwith 1 ml of 0.5% silane (97% 3-aminopropyltrimethoxylsilane,Sigma) for 20 min and dishes were washed gently. (3) Hydro-formylation: 1 ml of 0.5% glutaraldehyde in PBS was added to thecentral region in each dish for 30 min. The solution was aspiratedand dish was air-dried for an hour. (4) Polymerization: To make a4 kPa gel, 35 ml of gel mixture containing acrylamide (5%), bis-

H. Ji et al. / Biomaterials 85 (2016) 180e194192

acrylamide (0.1%), 0.01% of 0.22-mm-diameter red fluorescentcarboxylate-modified beads (Fluospheres, Invitrogen), 0.5% ofammonium persulphate (Sigma), and 0.05% TEMED (Bio-Rad) wasadded to the center of each dish. A coverslip prepared in step 1 wasgently put onto this gel mixture to form a glass-gel-glass sandwich.(5) Formation of the polyacrylamide gel: The gel was allowed topolymerize for 1 h. The fully polymerized gel (with cover slip on itstop) was soaked in 1 ml of 10 mM HEPEs buffer for 10 min and thecover slip was removed. (6) Gel surface activation: 150 ml of a so-lution containing 4 mM sulphosuccinimidyl-6-(4-azido-2-nitrophenylamino) hexanoate (Sulfo-SANPAH, Pierce) dissolved in0.05 M HEPES buffer (pH 8.5) was added to the top of the gel. Thedishes were then exposed to ultraviolet light for 20 min (until thecolor of the Sulfo-SANPAH solution changed from bright orange tobrown), washed with 0.05 M HEPES buffer (pH 8.5) twice and withPBS (pH 7.4) once. (7) Surface coating: 50 ml of rat tail Collagen typeI solution (20 mg/ml; BD) was added onto the gel. An untreated18 mm cover slip was used to cover the gel surface, and dish wasstored at 4 �C overnight. (9) Cell seeding and cell culture: the gelswere exposed to ultraviolet light for 20 min for sterilization andthen washed, and incubated with cell culture medium overnight.The medium was aspirated, and the gel was air-dried right beforeseeding the cells. iSMC, EPC and UASMC suspended in 2 ml ofcultural mediumwere seeded on a 35mmdish containing the gel ata density of 104 cells per dish 24 h before the traction force testswere implemented.

Confocal images and differential interference contrast (DIC)images were taken with an inverted Olympus FV1200 equippedwith a live cell chamber and an UPLSAPO 20� (NA 0.75) objective.The DIC Images for EPC, UASMC and iSMC on elastic substrate, andthe fluorescent beads on the top layer of elastic substrate weretaken. The images for substrate without cell attachment (referenceimages) were taken after each experiment; in this case, cells weretreated with Trypsin/EDTA (Life Technologies) for 30 min andallowed to completely detached from substrate.

The displacement field was computed by comparing red fluo-rescent bead images obtained during the experiment with thereference image. The calculation was done by a Matlab programanalysis using the method developed by Butler et al. in whichparticle imaging velocimetry is employed to measure thedisplacement of small windows that contain a number of beads[60,61]. The traction field was calculated from the displacementfield. The total contraction force was calculated by addition ofmagnitude of traction force in each the filed all together. Color barsin displacement filed and traction field indicate relative values inbeads displacement (unit: Micron) andmagnitude of traction (unit:Pascal), respectively.

4.8. Assembly and functional characterization of tissue engineeredblood vessels

Dense collagen gels have previously been shown to producemechanically robust constructs that allow for the replication ofbasic small diameter blood vessels that can undergo perfusion forlong time periods [62,63]. 1.5 � 106 dissociated and re-suspendediSMC in 300 ml of SmBM supplemented with SmGM-2 (Lonza)were embedded in 2.05 mg/ml of rat tail collagen type 1 solution(Corning) dissolved in 0.6% acetic acid. The solution was allowed togel by raising the pH to 8.5 using NaOH, and incubated in a cellculture hood at room temperature for 30 min in a 3 mL syringemold containing a 0.8 mm mandrel inserted through the middle(Fig. 5A). After gelation, the construct was placed on a 0.2 mmWhatman nylon membrane filter and subject to compressionaround the mandrel for 7e8 min to remove over 90% of the water.The collagen vessel construct was then slide off the mandrel and

into a chamber where it was sutured at both ends to grips thatallow for perfusion through the lumen of the vessel (Fig. 5D).0.5 � 106 dissociated and re-suspended EPC in 500 ml of EGM wereseeded into the lumen of the vessel by placing the EPC suspensionin a 1 mL luer-lock syringe and injecting them into the inlet ports ofthe chamber. The chamber was rotated at 10 rph for 30min at 37 �Cto allow for an even attachment of the EPC throughout the vessellumen. The chamber was then integrated into a continuous, steady,laminar perfusion circuit at a flow rate of 2 mL/min to ensure theapplication of physiological shear stresses of 6.8 dyn/cm2 to thevessel wall (Fig. 5EeF). The vessel was matured within the flowloop at 37 �C for 7 days. The same procedure was followed to createthe control vessel, but 1.5� 106 EPCwere embedded in the collagenconstruct in place of the iSMC.

To determine the functionality of the vessels, the change invessel diameter in response to gradual increase of flow rate, 1 mMphenylephrine and 1 mM acetylcholine were evaluated. In order tomeasure the response to flow-mediated dilation, the vessels wererun in the same perfusion circuit and subjected to 0.5, 1, 2, and4mL/min flow rate for 2min each and their change in diameter wasrecorded using a stereoscope (AmScope) and ISCapture software(Fig. 5EeF). Screenshots of the recorded video were taken at 2 minand 8min in order to determine the change in vessel diameter from0.5 mL/min to 4 mL/min. The diameter was calculated by averagingthe measurements across the width of the vessel at 4 random spotsover the entire vessel length using ImageJ. A similar approach wastaken to measure the vessel response to phenylephrine andacetylcholine. The vessels were run in the continuous perfusioncircuit and 1 mM phenylephrine was added to the media. After5 min, 1 mM acetylcholine was added to the media and perfusionwas continued for another 5 min. The 10 min procedure was againrecorded using a stereoscope and ISCapture software. Screenshotsof the video were taken at 30 s (prior to phenylephrine addition),5 min (after phenylephrine addition) and 10 min (after acetylcho-line addition). The vessel diameter at each time point was calcu-lated as previously stated. The vasoconstriction response tophenylephrinewas measured as the percent change from the initialdiameter to the diameter after 5 min. The vasodilation response toacetylcholine was measured as the percent change from thediameter after 5 min to the diameter after 10 min.

After the functional tests, vessels were taken out from thechamber and cut longitudinally to expose the vessel lumen. Theflattened vessels were then fixed in 4% paraformaldehyde (ElectronMicroscopy Sciences) for subsequent immunofluorescencestaining.

Author contributions

H.J., L.A., Z.C., N.C., and K.W.L. designed the project, H.J., L.A., Z.C.,S.C., Y.J. and N.C. performed the experiments, H.J., L.A., Z.C., N.C.,G.A.T. and K.W.L. analyzed the data, H.J., L.A., Z.C., N.C., G.A.T. andK.W.L. wrote the paper.

Competing financial interests

The authors declare no competing financial interests.

Acknowledgments

This work was supported by NIH grants UH2TR000505,4UH3TR000505, and the NIH Common Fund for the Micro-physiological Systems Initiative.

H. Ji et al. / Biomaterials 85 (2016) 180e194 193

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.biomaterials.2016.01.066.

References

[1] A. de Mel, et al., A potential platform for developing 3D tubular scaffolds forpaediatric organ development, J. Mater. Sci. Mater. Med. 26 (2015) 141, http://dx.doi.org/10.1007/s10856-015-5477-4.

[2] J.M. Ha, et al., Platelet-derived growth factor regulates vascular smoothmuscle phenotype via mammalian target of rapamycin complex 1, Biochem.Biophys. Res. Commun. (2015), http://dx.doi.org/10.1016/j.bbrc.2015.05.097.

[3] G.K. Owens, M.S. Kumar, B.R. Wamhoff, Molecular regulation of vascularsmooth muscle cell differentiation in development and disease, Physiol. Rev.84 (2004) 767e801, http://dx.doi.org/10.1152/physrev.00041.2003.

[4] S. Li, S. Sims, Y. Jiao, L.H. Chow, J.G. Pickering, Evidence from a novel humancell clone that adult vascular smooth muscle cells can convert reversibly be-tween noncontractile and contractile phenotypes, Circ. Res. 85 (1999)338e348.

[5] A.P. Somlyo, A.V. Somlyo, Ca2þ sensitivity of smooth muscle and nonmusclemyosin II: modulated by G proteins, kinases, and myosin phosphatase,Physiol. Rev. 83 (2003) 1325e1358, http://dx.doi.org/10.1152/physrev.00023.2003.

[6] M.R. Alexander, G.K. Owens, Epigenetic control of smooth muscle cell differ-entiation and phenotypic switching in vascular development and disease,Annu. Rev. Physiol. 74 (2012) 13e40, http://dx.doi.org/10.1146/annurev-physiol-012110-142315.

[7] D.A. Chistiakov, A.N. Orekhov, Y.V. Bobryshev, Vascular smooth muscle cell inatherosclerosis, Acta Physiol. (Oxf) 214 (2015) 33e50, http://dx.doi.org/10.1111/apha.12466.

[8] G.K. Owens, Regulation of differentiation of vascular smooth muscle cells,Physiol. Rev. 75 (1995) 487e517.

[9] Z. Wang, D.Z. Wang, G.C. Pipes, E.N. Olson, Myocardin is a master regulator ofsmooth muscle gene expression, Proc. Natl. Acad. Sci. U. S. A. 100 (2003)7129e7134, http://dx.doi.org/10.1073/pnas.1232341100.

[10] U. Hedin, B.A. Bottger, J. Luthman, S. Johansson, J. Thyberg, A substrate of thecell-attachment sequence of fibronectin (Arg-Gly-Asp-Ser) is sufficient topromote transition of arterial smooth muscle cells from a contractile to asynthetic phenotype, Dev. Biol. 133 (1989) 489e501.

[11] Y. Wang, et al., Engineering vascular tissue with functional smooth musclecells derived from human iPS cells and nanofibrous scaffolds, Biomaterials 35(2014) 8960e8969, http://dx.doi.org/10.1016/j.biomaterials.2014.07.011.

[12] V.K. Bajpai, P. Mistriotis, Y.H. Loh, G.Q. Daley, S.T. Andreadis, Functionalvascular smooth muscle cells derived from human induced pluripotent stemcells via mesenchymal stem cell intermediates, Cardiovasc. Res. 96 (2012)391e400, http://dx.doi.org/10.1093/cvr/cvs253.

[13] T. Vierbuchen, et al., Direct conversion of fibroblasts to functional neurons bydefined factors, Nature 463 (2010) 1035e1041, http://dx.doi.org/10.1038/nature08797.

[14] A.F. Adler, et al., Nonviral direct conversion of primary mouse embryonic fi-broblasts to neuronal cells, Mol. Ther. Nucleic Acids 1 (2012) e32, http://dx.doi.org/10.1038/mtna.2012.25.

[15] S. Chanda, et al., Generation of induced neuronal cells by the single reprog-ramming factor ASCL1, Stem Cell Rep. 3 (2014) 282e296, http://dx.doi.org/10.1016/j.stemcr.2014.05.020.

[16] S. Rafii, D. Lyden, Therapeutic stem and progenitor cell transplantation fororgan vascularization and regeneration, Nat. Med. 9 (2003) 702e712, http://dx.doi.org/10.1038/nm0603-702.

[17] D.A. Ingram, et al., Identification of a novel hierarchy of endothelial progenitorcells using human peripheral and umbilical cord blood, Blood 104 (2004)2752e2760, http://dx.doi.org/10.1182/blood-2004-04-1396.

[18] G.C. Pipes, E.E. Creemers, E.N. Olson, The myocardin family of transcriptionalcoactivators: versatile regulators of cell growth, migration, and myogenesis,Genes Dev. 20 (2006) 1545e1556, http://dx.doi.org/10.1101/gad.1428006.

[19] J. Chen, C.M. Kitchen, J.W. Streb, J.M. Miano, Myocardin: a component of amolecular switch for smooth muscle differentiation, J. Mol. Cell Cardiol. 34(2002) 1345e1356.

[20] J.M. Miano, Myocardin in biology and disease, J. Biomed. Res. 29 (2015) 3e19,http://dx.doi.org/10.7555/JBR.29.20140151.

[21] S. Li, D.Z. Wang, Z. Wang, J.A. Richardson, E.N. Olson, The serum responsefactor coactivator myocardin is required for vascular smooth muscle devel-opment, Proc. Natl. Acad. Sci. U. S. A. 100 (2003) 9366e9370, http://dx.doi.org/10.1073/pnas.1233635100.

[22] X.C. Long, R.D. Bell, W.T. Gerthoffer, B.V. Zlokovic, J.M. Miano, Myocardin issufficient for a smooth muscle-like contractile phenotype, Arter. Thromb.Vasc. 28 (2008) 1505e1510, http://dx.doi.org/10.1161/ATVBAHA.108.166066.

[23] K.L. Du, et al., Myocardin is a critical serum response factor cofactor in thetranscriptional program regulating smooth muscle cell differentiation, Mol.Cell Biol. 23 (2003) 2425e2437.

[24] J.M. Kristensen, et al., A PGC-1alpha- and muscle fibre type-related decreasein markers of mitochondrial oxidative metabolism in skeletal muscle ofhumans with inherited insulin resistance, Diabetologia 57 (2014) 1006e1015,

http://dx.doi.org/10.1007/s00125-014-3187-y.[25] L.R. Smith, H.G. Chambers, S. Subramaniam, R.L. Lieber, Transcriptional ab-

normalities of hamstring muscle contractures in children with cerebral palsy,PLoS One 7 (2012) e40686, http://dx.doi.org/10.1371/journal.pone.0040686.

[26] P.S. Pine, B.A. Rosenzweig, K.L. Thompson, An adaptable method using humanmixed tissue ratiometric controls for benchmarking performance on geneexpression microarrays in clinical laboratories, BMC Biotechnol. 11 (2011) 38,http://dx.doi.org/10.1186/1472-6750-11-38.

[27] H. Wessells, et al., Transcriptional profiling of human cavernosal endothelialcells reveals distinctive cell adhesion phenotype and role for claudin 11 invascular barrier function, Physiol. Genom. 39 (2009) 100e108, http://dx.doi.org/10.1152/physiolgenomics.90354.2008.

[28] C.S. Greene, et al., Understanding multicellular function and disease withhuman tissue-specific networks, Nat. Genet. 47 (2015) 569e576, http://dx.doi.org/10.1038/ng.3259.

[29] M. Olesen, V. Skov, M. Mechta, B.H. Mumm, L.M. Rasmussen, No influence ofOPG and its ligands, RANKL and TRAIL, on proliferation and regulation of thecalcification process in primary human vascular smooth muscle cells, Mol.Cell. Endocrinol. 362 (2012) 149e156, http://dx.doi.org/10.1016/j.mce.2012.06.004.

[30] X.M. Wang, et al., The gene expression profiles of induced pluripotent stemcells from individuals with childhood cerebral adrenoleukodystrophy areconsistent with proposed mechanisms of pathogenesis, Stem Cell Res. Ther. 3(2012) 39, http://dx.doi.org/10.1186/scrt130.

[31] H.G. Roider, T. Manke, S. O'Keeffe, M. Vingron, S.A. Haas, PASTAA: identifyingtranscription factors associated with sets of co-regulated genes, Bioinfor-matics 25 (2009) 435e442, http://dx.doi.org/10.1093/bioinformatics/btn627.

[32] Y. Zhao, et al., An expanded palette of genetically encoded Ca(2)(þ) indicators,Science 333 (2011) 1888e1891, http://dx.doi.org/10.1126/science.1208592.

[33] P.H. Ratz, K.M. Berg, N.H. Urban, A.S. Miner, Regulation of smooth musclecalcium sensitivity: KCl as a calcium-sensitizing stimulus, Am. J. Physiol. CellPhysiol. 288 (2005) C769eC783, http://dx.doi.org/10.1152/ajpcell.00529.2004.

[34] E.M. Collins, M.P. Walsh, K.G. Morgan, Contraction of single vascular smoothmuscle cells by phenylephrine at constant [Ca2þ]i, Am. J. Physiol. 262 (1992)H754eH762.

[35] S.A. Maskarinec, C. Franck, D.A. Tirrell, G. Ravichandran, Quantifying cellulartraction forces in three dimensions, Proc. Natl. Acad. Sci. U. S. A. 106 (2009)22108e22113, http://dx.doi.org/10.1073/pnas.0904565106.

[36] G.J. Ye, et al., The contractile strength of vascular smooth muscle myocytes isshape dependent, Integr. Biol. (Camb) 6 (2014) 152e163, http://dx.doi.org/10.1039/c3ib40230d.

[37] T.H. Lee, et al., Functional recapitulation of smooth muscle cells via inducedpluripotent stem cells from human aortic smooth muscle cells, Circ. Res. 106(2010) 120e128, http://dx.doi.org/10.1161/CIRCRESAHA.109.207902.

[38] C. Cheung, A.S. Bernardo, M.W. Trotter, R.A. Pedersen, S. Sinha, Generation ofhuman vascular smooth muscle subtypes provides insight into embryologicalorigin-dependent disease susceptibility, Nat. Biotechnol. 30 (2012) 165e173,http://dx.doi.org/10.1038/nbt.2107.

[39] D. Wang, et al., Activation of cardiac gene expression by myocardin, a tran-scriptional cofactor for serum response factor, Cell 105 (2001) 851e862.

[40] L. Raphel, A. Talasila, C. Cheung, S. Sinha, Myocardin overexpression is suffi-cient for promoting the development of a mature smooth muscle cell-likephenotype from human embryonic stem cells, PLoS One 7 (2012) e44052,http://dx.doi.org/10.1371/journal.pone.0044052.

[41] S. Earley, J.E. Brayden, Transient receptor potential channels in the vascula-ture, Physiol. Rev. 95 (2015) 645e690, http://dx.doi.org/10.1152/physrev.00026.2014.

[42] J. Utz, R. Eckert, W. Trautwein, Changes of intracellular calcium concentrationsby phenylephrine in renal arterial smooth muscle cells, Pflugers Arch. 438(1999) 725e731.

[43] H. Hamada, D.S. Damron, S.J. Hong, D.R. Van Wagoner, P.A. Murray, Phenyl-ephrine-induced Ca2þ oscillations in canine pulmonary artery smooth musclecells, Circ. Res. 81 (1997) 812e823.

[44] K.A. Dora, M.P. Doyle, B.R. Duling, Elevation of intracellular calcium in smoothmuscle causes endothelial cell generation of NO in arterioles, Proc. Natl. Acad.Sci. U. S. A. 94 (1997) 6529e6534.

[45] J.H. Wang, J.S. Lin, Cell traction force and measurement methods, Biomech.Model Mechanobiol. 6 (2007) 361e371, http://dx.doi.org/10.1007/s10237-006-0068-4.

[46] J. Chen, H. Li, N. SundarRaj, J.H. Wang, Alpha-smooth muscle actin expressionenhances cell traction force, Cell Motil. Cytoskelet. 64 (2007) 248e257, http://dx.doi.org/10.1002/cm.20178.

[47] S. Sinha, et al., Assessment of contractility of purified smooth muscle cellsderived from embryonic stem cells, Stem Cells 24 (2006) 1678e1688, http://dx.doi.org/10.1634/stemcells.2006-0002.

[48] K.A. Dora, J.M. Hinton, S.D. Walker, C.J. Garland, An indirect influence ofphenylephrine on the release of endothelium-derived vasodilators in rat smallmesenteric artery, Br. J. Pharmacol. 129 (2000) 381e387, http://dx.doi.org/10.1038/sj.bjp.0703052.

[49] N. L'Heureux, et al., Human tissue-engineered blood vessels for adult arterialrevascularization, Nat. Med. 12 (2006) 361e365, http://dx.doi.org/10.1038/nm1364.

[50] Z. Gong, L.E. Niklason, Small-diameter human vessel wall engineered frombone marrow-derived mesenchymal stem cells (hMSCs), FASEB J. 22 (2008)

H. Ji et al. / Biomaterials 85 (2016) 180e194194

1635e1648, http://dx.doi.org/10.1096/fj.07-087924.[51] C.B. Weinberg, E. Bell, A blood vessel model constructed from collagen and

cultured vascular cells, Science 231 (1986) 397e400.[52] Y. Jung, et al., Scaffold-free, human mesenchymal stem cell-based tissue

engineered blood vessels, Sci. Rep. 5 (2015) 15116, http://dx.doi.org/10.1038/srep15116.

[53] G.A. Truskey, C.E. Fernandez, Tissue-engineered blood vessels as promisingtools for testing drug toxicity, Expert Opin. Drug Metab. Toxicol. (2015) 1e4,http://dx.doi.org/10.1517/17425255.2015.1047342.

[54] P. Rajendran, et al., The vascular endothelium and human diseases, Int. J. Biol.Sci. 9 (2013) 1057e1069, http://dx.doi.org/10.7150/ijbs.7502.

[55] M.A. Brown, C.S. Wallace, M. Angelos, G.A. Truskey, Characterization of um-bilical cord blood-derived late outgrowth endothelial progenitor cells exposedto laminar shear stress, Tissue Eng. Part A 15 (2009) 3575e3587, http://dx.doi.org/10.1089/ten.TEA.2008.0444.

[56] N. Christoforou, et al., Transcription factors MYOCD, SRF, Mesp1 andSMARCD3 enhance the cardio-inducing effect of GATA4, TBX5, and MEF2Cduring direct cellular reprogramming, PLoS One 8 (2013) e63577, http://dx.doi.org/10.1371/journal.pone.0063577.

[57] N. Maherali, et al., A high-efficiency system for the generation and study ofhuman induced pluripotent stem cells, Cell Stem Cell 3 (2008) 340e345,

http://dx.doi.org/10.1016/j.stem.2008.08.003.[58] B. Zhang, S. Kirov, J. Snoddy, WebGestalt: an integrated system for exploring

gene sets in various biological contexts, Nucleic Acids Res. 33 (2005)W741eW748, http://dx.doi.org/10.1093/nar/gki475.

[59] D. Warde-Farley, et al., The GeneMANIA prediction server: biological networkintegration for gene prioritization and predicting gene function, Nucleic AcidsRes. 38 (2010) W214eW220, http://dx.doi.org/10.1093/nar/gkq537.

[60] J.P. Butler, I.M. Tolic-Norrelykke, B. Fabry, J.J. Fredberg, Traction fields, mo-ments, and strain energy that cells exert on their surroundings, Am. J. Physiol.Cell Physiol. 282 (2002) C595eC605, http://dx.doi.org/10.1152/ajpcell.00270.2001.

[61] X. Trepat, et al., Physical forces during collective cell migration, Nat. Phys. 5(2009) 426e430 doi, http://www.nature.com/nphys/journal/v5/n6/suppinfo/nphys1269_S1.html.

[62] C.E. Ghezzi, et al., An airway smooth muscle cell niche under physiologicalpulsatile flow culture using a tubular dense collagen construct, Biomaterials34 (2013) 1954e1966, http://dx.doi.org/10.1016/j.biomaterials.2012.11.025.

[63] J.B. Phillips, R. Brown, Micro-structured materials and mechanical cues in 3Dcollagen gels, Methods Mol. Biol. 695 (2011) 183e196, http://dx.doi.org/10.1007/978-1-60761-984-0_12.