Title: Guide to the heart: Differentiation of human pluripotent stem cells towards multiple cardiac subtypes

The handle http://hdl.handle.net/1887/82699 holds various files of this Leiden University dissertation.

Author: Schwach, V.

Title: Guide to the heart: Differentiation of human pluripotent stem cells towards multiple cardiac subtypes

Issue Date: 2020-01-15

hPSCs Differ

entiation

Magnetic Beads Non-cardiac cells

Heterogeneous

Population Cardiomyocytes

Purification

2

Chapter 2:

Generation and purification of human stem cell-derived cardiomyocytes

Verena Schwach ¹ and Robert Passier ^{1, 2}

1 Department of Anatomy and Embryology, Leiden University Medical Center, Einthovenweg 20, P.O. box 9600; ² Department of Applied Stem Cell Technologies, TechMed Centre, University of Twente, P.O. Box 217, 7500 AE Enschede, the Netherlands

Differentiation, 91: 126–138 (2016)

Abstract

Efficient and reproducible generation and purification of human stem cell-derived cardiomyocytes (CMs) is crucial for regenerative medicine, disease modelling, drug screening and study of developmental events during cardiac specification. Established methods to generate CMs from human pluripotent stem cells (hPSCs) include the Spin-embryoid body (Spin-EB) and monolayer-based differentiation protocol. In the presence of an optimized cocktail of growth factors under defined conditions, hPSCs differentiate efficiently into functional contracting CMs within 10 days.

Nevertheless, despite high efficiencies, cardiac-directed differentiations of hPSCs typically result in heterogeneous populations comprised of both CMs and uncharacterized non-cardiac cell-types. Therefore, generation of pure populations of stem cell-derived CMs is of fundamental importance for basic cardiac research and pre-clinical and possible clinical applications. For the purification of CMs from heterogeneous populations, fluorescent activated cell sorting (FACS) is a widely appreciated method. Nonetheless, FACS- based isolation of CMs comes along with several disadvantages, such as undesired contaminations and low viability of target cells. Here, we describe a convenient and rapid procedure for the purification of hPSCs-derived CMs under sterile culture conditions, resulting in high purity and viability of sorted CMs. Purification with VCAM1-coupled magnetic Dynabeads led to robust enrichment of CMs, which will especially be important for cardiac differentiations of cell lines with poor differentiation efficiencies. In addition, this will also be beneficial for the standardization and reproducibility of human stem cell–derived assays in the fields of cardiac disease modeling, drug discovery and disease modeling.

Abbreviations:

CMs – cardiomyocytes, CPCs - cardiac progenitor cells, EB – Embryoid body,

GFP - green fluorescent protein, hPSCs – human pluripotent stem cells,

SIRPα - Signal regulatory protein alpha, VCAM1 - Vascular cell adhesion

molecule 1

2

Introduction

As human primary cardiomyocytes (CMs) are difficult to obtain and do not proliferate in culture, stem cell-derived CMs provide a tremendous advantage in regenerative medicine, disease modelling, drug screening and studying early cardiomyogenesis. Despite the progress made in the efficiency of CM differentiation from human pluripotent stem cells (hPSCs), standardized comparisons between experiments from different stem cell lines or different laboratories are hampered by the variability of cardiomyocyte production.

Several human stem cell-based cardiomyocyte differentiation methods have been established and are continuously optimized to either increase the purity or yield of CMs(van den Berg et al., 2015; Birket et al., 2015a; Burridge et al., 2014; Dambrot et al., 2014; Elliott et al., 2011; Lian et al., 2013). However, differentiation protocols produce heterogeneous populations composed not only of CMs, but also non-cardiac cell-types, including, fibroblasts, smooth- muscle and endothelial cells (Birket et al., 2013). Both for assay development and in vivo applications it is crucial to obtain pure (or at least defined) and viable cell populations. Previously, antibodies recognizing the cell surface proteins VCAM1 (Vascular cell adhesion molecule 1) and SIRPα (Signal regulatory protein alpha) have been utilized for the purification of CMs from heterogeneous hPSCs-derived CM populations by fluorescent activated cell sorting (FACS) (Elliott et al., 2011; Skelton et al., 2014; Uosaki et al., 2011). FACS is a frequently used method for detection, quantification and isolation of fluorescent single cells, labelled via either genetic manipulation or cell surface-specific antibodies. However, genetic manipulation for the introduction of fluorescent reporter genes is time-consuming and is not feasible for every cell line. Furthermore, FACS-based purification of cell populations comes along with several disadvantages, such as undesired contaminations and low viability of target cells.

Here, we describe two established methods for the generation of CMs from

hPSCs, as well as an optimized monolayer-based protocol. Moreover, we

illustrate a robust procedure for the purification of hPSCs-derived CMs

under sterile culture conditions, utilizing VCAM1 antibody-coupled magnetic

Dynabeads. Bead-based purifications result in high purity and viability of

sorted CMs.

Results

Improved cardiomyocyte differentiation from hPSCs in monolayer cultures

In order to generate high quantities of CMs we followed the cardiac monolayer protocol (Dambrot et al., 2014). To efficiently monitor cardiac differentiation, the cardiac fluorescent reporter line hESC-NKX ^2.5eGFP/+ was chosen for these experiments (Elliott et al., 2011). In this reporter line green fluorescent protein (GFP) has been targeted to the genomic locus of the cardiac transcription factor NKX2.5. Upon differentiation towards the cardiac lineage, GFP becomes visible in cardiac progenitor cells (CPCs) and is further increased in functional CMs. For cardiac differentiation hPSCs were seeded at low density on matrigel-coated dishes and 24 h later mesoderm differentiation was induced, by supplementing BPEL with the cytokines BMP4 and Activin-A, as well as the WNT activator CHIR99021.

Within 3 days, differentiating cells organized into a confluent monolayer with mesenchymal-like cobblestone morphology. To further direct differentiation into the cardiac fate, WNT signaling was inhibited by addition of XAV939, a potent WNT antagonist. WNT was inhibited from day 3 till 7 for control monolayer differentiations. To increase cardiomyocyte differentiation efficiency, we shortened XAV39 treatment from day 3 to 4 with subsequent inhibition of TGF-ß signaling by SB431542 and induction of sonic hedgehog signaling using the small molecule SAG (Smoothened agonist) from day 4 till 10 (hereafter called SBS differentiations) (Figure 2.1A). In both control and SBS monolayer differentiations first contractions were observed around day 9 and from day 10 forward cells formed a contracting network-like structure.

First GFP expression was induced in CPCs around day 7 of differentiation and enhanced substantially until day 14 (Figure 2.1B). FACS analysis of day 14 cultures revealed that 60% (62% ± 3%, n=5) of cells expressed GFP in control monolayer differentiations, while in SBS monolayer differentiations 80% (80% ± 2%, n=5) of the cells robustly expressed GFP. Overall SBS treatment had a statistically significant benefit on the monolayer-based in- vitro cardiac differentiation when compared to control (Figure 2.1C and D).

Cell death was perceived between day 1 and 7 of differentiation in both

conditions.

2

Figure 2.1: A) Differentiation schedule for the efficient generation of CMs from hPSCs with the control monolayer method (indicated by an asterisk) or optimized monolayer protocol (SBS differentiation). B) Representative images of the morphological appearance of the differentiating cells during SBS differentiation at day 0, 3, 7, 10 and 14 in bright field (upper panel) and GFP fluorescence (lower panel) (10x, scale bar = 100 µm). C) Representative FACS plots displaying the percentage of GFP+ CMs.

D) Averaged GFP percentage calculated from five different differentiations (62% ± 3%)

with control strategy compared to an optimized monolayer protocol (SBS), including

a SAG and SB431542 treatment (80% ± 2). SBS treatment had a statistically

significant benefit on monolayer-based differentiations. Data are displayed as means

+SEM. Statistical significance was analyzed by Student’s paired t-test with p<0.05

considered significant.

Purification of heterogeneous stem cell-derived cardiomyocyte cultures

Cardiac differentiations of hPSCs routinely yield heterogeneous populations comprised of not only CMs, but also additional cell types, such as fibroblasts, endothelial, smooth muscle cells or uncharacterized differentiated cell types.

Previously it was shown that human stem cell-derived CMs robustly express the surface marker VCAM1. In order to determine the percentage of VCAM1 on CMs we performed monolayer differentiations followed by FACS analysis for both VCAM1- and GFP-positive cells. Control monolayer differentiations resulted in 55% of VCAM1 ⁺ /GFP ⁺ cells (55% ± 6%, n=3) and 9% VCAM1 ⁺ / GFP ^- cells (9% ± 4%, n=3) at day 13 or 14 of differentiation when compared to isotype-control stained cell suspensions. SBS monolayer differentiations yielded 70% of VCAM1 ⁺ /GFP ⁺ cells (70% ± 3%, n=3) and 14% VCAM1 ⁺ /GFP ^- cells (14% ± 5%, n=3) (Figure 2.2). Next, we used two different magnetic bead isolation methods, both based on VCAM1 expression, for purification of mixed cardiomyocyte cultures.

Figure 2.2: Representative FACS measurements of isotype- (left) or VCAM1 (middle) labeled cells together with GFP of A) control or B) SBS differentiations.

Overlay of the intensity values from isotype-labeled cells (red) and VCAM1-labeled

cells (blue). Averaged VCAM1 labeling together with GFP calculated from three

different differentiations. C) GFP ⁺ /VCAM1 ⁺ (55% ± 6%), GFP ^- /VCAM1 ⁺ (9% ± 4%),

GFP ⁺ /VCAM1 ^- (9% ± 4%) and GFP ^- /VCAM1 ^- (28% ± 1%) from control monolayer

differentiations (upper panel) compared to D) SBS, including SAG and SB431542

2

treatment, GFP ⁺ /VCAM1 ⁺ (70% ± 3%), GFP ^- /VCAM1 ⁺ (14% ± 5%), GFP ⁺ /VCAM1 ^- (7%

± 3%) and GFP ^- /VCAM1 ^- (9% ± 5%) (Lower panel). Data are displayed as means +SEM. Statistical significance was analyzed by Student’s paired t-test with p<0.05 considered significant.

Positive Magnetic Bead-based Isolation of hPSCs-derived CMs

In order to purify hPSC-CMs, VCAM1 ⁺ CMs were isolated from heterogeneous cultures by magnetic bead isolation in a 6-step process (Figure 2.3).

Figure 2.3: Schematic illustration of the 6-step process of CM purification based on magnetic bead isolation.

To evaluate CM purity after bead sorting, bright field and GFP pictures of all three populations (unsorted, negative and positive - hereafter called cardiopure - fractions) were acquired. To allow recovery of the CMs after dissociation and purification, pictures were captured 7 days after magnetic isolation. Fluorescent pictures of cardiopure populations explicitly displayed strong enrichment for GFP ⁺ CMs when compared to the negative population or the unsorted material (Figure 2.4). Moreover, the predominant part of CMs in the cardiopure fraction formed a syncytium-like monolayer as opposed to the cardiomyocyte clusters surrounded by non-CMs found in negative cell suspensions or unsorted cells. Bright field videos of contracting CMs in all three fractions were acquired and overlaid with GFP images to confirm functionality of bead-sorted CMs (Suppl. Videos 1-3). Majority of cells in the cardiopure fraction were characterized by an elongated cardiac- like morphology and detectable NKX2.5-GFP expression. A small percentage of contracting cells displayed a cardiac-like morphology bound to VCAM1- beads, but were clearly negative for GFP (Figure 2.5), which may represent NKX2.5 negative pacemaker cells (Birket et al., 2015a). To evaluate cardiac identity of the GFP-negative cells more closely, bright field videos

• 1X TrypLe

• Resuspension in Sort buffer

• 0.06 µg per 10

CMs in 150 µl Sort buffer

• 5 - 10 min at RT

• 1x with Sort buffer

• 20 min on rotator at 4°C

• 8 beads per CM in 0.9 ml Sort

buffer

• 1x with Sort buffer

• 1x with DMEM + 0.1% BSA

• In CM medium for culture

• Cell lysis

Dissociation VCAMI incubation Wash Bead incubation Wash Resuspension

of contracting CMs were overlapped with GFP pictures (Suppl. Video 4).

As judged by manual counting of living cells before and after purification, magnetic bead isolations typically recovered 70% ± 10% of total CMs in culture (data not shown).

Figure 2.4: Representative images of the morphological appearance of purified

fractions 7 days after magnetic bead isolation and re-plating in bright field (upper

panel) or GFP (lower) (10x, scale bar = 100 µm). QR codes to movies.

2

BF GFP

Figure 2.5: Bright field and GFP images of bead-sorted cardiomyocyte with VCAM1 coupled-beads 7 days after purification, demonstrating a GFP-negative, but VCAM1 ⁺ cardiomyocyte (40x, scale bar = 25 µm). QR code to movie.

In order to further assess not only purity of the cell fractions, but also

cardiac features, such as sarcomeric organization, immunostainings of the

resulting fractions were performed for typical cardiac markers. In contrast to

unsorted starting material or the negative population, almost every cell in the

cardiopure population robustly expressed cardiac α-ACTININ and NKX2.5 7

days after re-plating. In addition, sarcomeric α-ACTININ overlapped with

nuclear NKX2.5 expression in CMs of all three fractions (Figure 6). Also with

higher magnification, no major differences in sarcomeric organization were

observed between CMs before and after magnetic isolation (Figure 2.6).

Figure 2.6: Representative confocal single-stack images of the cardiopure population (top), negative fraction (middle panel) and the unsorted population without purification (lower panel) 7 days after purification. DAPI in blue, NKX2.5-GFP in green, α-ACTININ in red and NKX2.5 in grey with overlay on the left (40x, scale bar

= 50 µm) and right (63x, scale bar = 25 µm).

To assess purification on gene expression level, we quantified the expression pattern of typical sarcomeric cardiac markers such as ACTN2 or TNNT2, as well as the early cardiac marker NKX2.5 by quantitative PCR (qPCR) at day 14 (n=3). Gene expression profiling revealed 3 to 4- fold enrichment of ACTN2, TNNT2 and NKX2.5 in the cardiopure population compared to the negative or unsorted population (Figure 2.7).

Cardiopure

GFP

GFP GFP

GFP

GFP GFP

NKX2.5 NKX2.5

NKX2.5

NKX2.5 NKX2.5

NKX2.5 ACTININ

ACTININ ACTININ

ACTININ ACTININ DAPI

DAPI

DAPI

DAPI DAPI

DAPI

Cardiopure

Negative

Unsorted Unsorted

Negative

ACTININ

2

Figure 2.7: qPCR reveals significant upregulation of typical cardiac markers, such as ACTN2, TNNT2 or NKX2.5 in cardiopure to unsorted and negative populations at day 14 (n=3). Data are displayed as means + SEM. Statistical significance was analyzed by Student’s paired t-test with P<0.05 considered significant.

Positive Magnetic Bead Isolation with Bead Release

For continuation of cardiopure cultures after magnetic bead purification, it is recommendable to release beads from cells after isolation. For this, we used DSB-biotinylated VCAM1 antibodies. Specialized Dynabeads can easily be released from these antibodies in presence of a biotin-rich release buffer as illustrated in figure 2.8.

Figure 2.8: Experimental outline of the CM purification with VCAM1-coupled magnetic Dynabeads and subsequent bead release.

• 1X TrypLe

• Resuspension in Sort buffer

• 1.3 µg per 10⁶ CMs in 100 µl Sort buffer

• 5 - 10 min at RT

• 2x with Sort buffer

• 25 min on rotator at RT

• 40 beads per CM in 1 ml Sort buffer

• 1x with Sort buffer

• 1x with DMEM + 0.1% BSA

• 1 ml Release buffer

• 2-5 min at RT

Dissociation VCAMI incubation Wash Bead incubation Wash Bead Release

• In CM medium for culture

• In Sort buffer for flow cytometry

Resuspension

Positive magnetic isolation with bead release allows quantification of the percentage of CMs following this procedure. FACS analysis of cell suspensions before and immediately after isolation revealed a significant enrichment for CMs in the cardiopure compared to the negative fraction or unsorted starting population in control monolayer differentiations and even under optimized conditions (SBS monolayer differentiations). Flow measurement detected 91% ± 2% (n=3) GFP ⁺ cells in the cardiopure fraction compared to 55% ± 6%

(n=3) in the negative population and 65% ± 3% (n=3) in the initial unsorted population in control monolayer suspensions. Under optimized conditions, 97% ± 1% (n=3) of all cells in the cardiopure population robustly expressed GFP opposed to 75% ± 6% (n=3) in the negative population and 81% ± 3%

(n=3) in the unsorted material (Figure 2.9). Magnetic bead isolations with bead release typically recovered 50% ± 10% of total CMs in culture (data not shown).

Figure 2.9: A) Representative FACS plots displaying the percentage of GFP-

positive CMs purified by magnetic bead isolation with bead release in the unsorted

(left), cardiopure (middle) and negative population (right) in A) control monolayer

differentiations or B) SBS differentiations. Averaged GFP percentage calculated from

three different purifications in the unsorted (65% ± 3%), cardiopure (91% ± 2%)

and the negative population (55% ± 6%) in C) control monolayer differentiations

compared to D) optimized SBS conditions, unsorted population (81% ± 3%),

cardiopure (97% ± 1%) and negative population (75% ± 6%). Data are displayed as

means +SEM. Statistical significance was analyzed by Student’s paired t-test with

P<0.05 considered significant.

2

As identified from FACS analysis for GFP ⁺ cells after bead purification approximately 5% of cells in the cardiopure fractions are GFP ^- (Figure 2.9). Functional assessment of this small cell population confirmed their functional contracting phenotype (Suppl. Video 4). To further confirm cardiac identity of the GFP ^- cells, FACS staining for SIRPα, an important cardiac surface marker, was performed(Dubois et al., 2011; Elliott et al., 2011).

We found that, indeed, about 5% of the cells in the cardiopure populations purified from either monolayer or SBS-derived cells express SIRPα, while simultaneously being negative for GFP (Figure 2.10). Additionally, SIRPα staining disclosed that in total more than 99% of all cells in the cardiopure fraction express the cardiac marker SIRPα or are positive for NKX2.5-GFP (Figure 2.10).

Figure 2.10: Representative FACS displaying the percentage of GFP- or SIRPα- positive CMs in the unsorted (left), cardiopure (middle) and negative population (right) after purification at day 14 in A) control or B) SBS differentiations. Averaged percentage of SIRPα-or-GFP-positive CMs in the unsorted (81% ± 3%), cardiac (97%

± 1%) and negative population (75% ± 6%) in C) control or D: SBS, SIRPα-or-GFP-

positive cells (95% ± 2%), (99.5% ± 1 %) and (87% ± 2%) (n=3). Data are displayed as

means +SEM. Statistical significance was analyzed by Student’s paired t-test with

p<0.05 considered significant.

Bead-based purification of CMs generated with the Spin-EB differentiation method

Generation of hPSCs-derived CMs with the Spin-EB differentiation method The Spin-EB protocol is a frequently used method to generate CMs from hPSCs (Ng ES, Davis R, Stanley EG, 2008). In the presence of an optimized cocktail of growth factors, differentiation is directed towards mesoderm during the first 3 days. Upon commitment to CPCs around day 7 of differentiation, GFP expression was induced and increased until day 14, as expected. Obvious contraction became apparent around day 9. FACS measurement of day 14, dissociated EBs showed stable GFP expression in 65% (65% ± 2%, n=5) of the cells (Figure 2.11). Minor cell death was observed during differentiation.

Figure 2.11: A) Schematic representation of the differentiation schedule for the

generation of CMs from hPSCs following the Spin-EB protocol. B) Representative

images of the morphological appearance of the EBs during differentiation at day 0, 3,

7, 8, 10 and 14 in bright field (upper panel) or GFP fluorescence (lower panel) (10x,

scale bar = 100 µm). C) Representative FACS plot displaying the percentage of GFP-

2

positive CMs in a typical Spin-EB differentiation at day 14 of differentiation (left) and an averaged GFP percentage calculated from five different differentiations (65% ± 2).

Positive Magnetic Bead-based Isolation of Spin-EB-derived CMs

To establish if bead-based purifications allow separation of cardiac and non- cardiac cells generated with the Spin-EB method, we examined isolation of dissociated Spin-EB-derived CMs at day 14 of differentiation. Equivalent to monolayer-derived CMs, cardiopure populations from Spin-EBs showed strong enrichment of GFP-positive CMs when compared to the negative population (Figure 2.12).

Figure 2.12: Representative images of cardiopure (left) and negative (right)

populations separated by positive selection with VCAM1-coupled magnetic beads

without bead release. CMs were purified from Spin-EBs after dissociation at day

14 of differentiation. Images in bright field (upper panel) or GFP fluorescence (lower

panel) (20x, scale bar = 50 µm).

Discussion

An important feature of hPSCs is that they can differentiate to any cell- type of the human body, including CMs. Because adult human primary CMs are difficult to obtain and do not proliferate in culture, human stem cell-derived CMs may provide a valuable human-based platform for various research areas, including regenerative medicine, developmental biology, disease modeling, drug discovery and safety pharmacology. Since the first derivation of human pluripotent embryonic stem cells in 1998 and later the induced pluripotent stem cells (hiPSC), several procedures have been developed for the generation of CMs, which includes co-culture with visceral endodermal-like cells, as well as Spin-EB and monolayer-based protocols.

The scalable, serum-free, defined differentiation conditions of the Spin-EB and monolayer-based methods strongly contributed to the frequent and widely accepted use of these methods (Mummery et al., 2012).

Cardiac differentiations via the Spin-EB method have a closer resemblance to embryonic development regarding their spheroid, three-dimensional (3D) shape than monolayer-based directed cardiac differentiation protocols.

In contrast, monolayer differentiation cultures generally display a higher robustness and efficiency in cardiomyocyte differentiation and can therefore more easily be applied to different cell lines with minor optimization (Dambrot et al., 2014).

To further optimize cardiac differentiation in monolayer differentiations, we interfered with the TGFb and sonic hedgehog signaling pathways.

Inhibition of TGF-ß signaling by SB431542 and induction of sonic hedgehog signaling with hedgehog agonist SAG have previously been shown to induce proliferation of CPCs before the onset of NKX2.5 expression(Birket et al., 2015a). SAG and SB431542 were supplemented during different time windows between day 4 and 10 of differentiation. Subsequently, monolayer cultures were assessed for their efficiency to generate CMs from hPSCs.

Combined treatment of SB431542 and SAG positively affected cardiac

differentiation when supplemented from day 4 to 7 or 10, but did not markedly

improve efficiency or yield when only added after day 7. As judged by FACS

analysis, inclusion of SAG and SB431542 into the culture medium from day

4 – 10 most efficiently increased cardiomyocyte yield and percentage of GFP-

positive cells when compared to control monolayer differentiations (data not

shown).

2

Generating pure populations of CMs from hPSCs is not only of fundamental importance for studying cardiac specification, but also for pre-clinical and clinical applications. However, cardiac-directed differentiations of hPSCs and especially from patient-specific induced PSCs typically result in mixed populations comprised of both CMs and uncharacterized non-CMs (e.g.

endothelial, smooth muscle cells, fibroblasts etc.). Therefore, there is an urgent need to develop techniques that will allow purification of CMs.

Here, we describe the isolation of CMs from heterogeneous cultures with VCAM1-coupled magnetic beads. The separation of cardiac from non- cardiac cells with VCAM1-Sheep-Anti-Mouse IgG Dynabeads provides a fast and convenient method for the enrichment of CMs from stem cell differentiation cultures with high efficiency. As CMs do not proliferate in vitro, fast proliferating non-cardiac cells, such as fibroblast-like cells often increase in percentage after time in culture. As assessed by morphological observation, as well as immunostaining for cardiac markers, such as NKX2.5 and α-ACTININ 7 days after purification, 100% of the cells in the cardiopure fraction displayed cardiac characteristics. Moreover, analysis of gene expression patterns of cardiac-specific genes NKX2.5, TNNT2 and ACTN2 implies that purifications resulted in pure cardiac populations. This method is particularly suitable for gene expression studies, since Dynabeads do not interfere with RNA isolation and are compatible with common RNA isolation Kits. Although efficiency and purity of this method are convincing, bead coupling to CMs makes it incompatible with FACS analysis, prolonged culture and may interfere with biological assays. Nevertheless, sorted CMs showed a fast recovery and remarkable viability within two days after sort.

CMs contracted in culture for at least more than two weeks.

To purify CMs with subsequent bead release we utilized the Dynabeads®

FlowComp™ Flexi Kit for which we carefully optimized concentration of beads, incubation timing, as well as amount of antibody per cell. As evaluated by detection with a Streptavidin-PE conjugated antibody, biotinylated VCAM1 antibody molecules stained similar percentages of cells during FACS measurement when compared to unconjugated VCAM1 antibody stainings with anti-mouse IgG antibodies conjugated to APC, suggesting that biotinylation of the antibody does not interfere with isotype recognition.

Bead isolations yielded highly pure cardiac cultures as confirmed by FACS

measurement for GFP and SIRPα. SIRPα has been identified and validated

as important surface marker for human hPSCs-derived CMs (Dubois et

al., 2011; Elliott et al., 2011). Similar to isolations without bead release,

a fast though gentle separation of cardiac and non-cardiac cells, resulted

in vastly viable populations of cardiac cells. It is noteworthy to mention

that we detected a small percentage of GFP ^- cells in the cardiac fraction

after bead sort. In concordance, FACS analysis for VCAM1 showed a small

percentage of VCAM1 ⁺ /GFP ^- cells. Interestingly, VCAM1 ⁺ /GFP ^- displayed

spontaneous contractions and a cardiomyocyte morphology. These

findings suggest that these cells may represent a NKX2.5-negative cardiac

population. NKX2.5-negative CMs constitute a side population in the heart,

known as the sinus node, the primary pacemaker of the heart responsible

for the initiation of cardiac contraction. According to literature, stem cell-

derived NKX2.5-negative CMs express the cardiac surface marker SIRPα

(Birket et al., 2015a). FACS for SIRPα revealed that GFP-negative cells stain

positive for SIRPα, which indeed suggest that these GFP-negative cells from

the positive sorted cardiopure fraction belong to the cardiac lineage. One

disadvantage of this method is the higher loss of CMs when compared to

bead purifications without bead release. Unfortunately, increased amounts

of antibody or beads, as well as prolonged incubation times did not benefit

efficiency of bead sort. Approximately 10% of CMs appeared to be lost during

washing, depending on the purpose of purification it might be advisable

to wash once only and accept some minor contamination of non-cardiac

cells in the cardiopure fraction. Nevertheless, conventional FACS-based

sorting generally lead to a high degree of cell loss too (Tomlinson et al.,

2013). Furthermore, FACS occurs in an open system with consequently a

higher chance of contaminations. Also, based on fluidic parameters single

cell purifications in FACS machines cannot be scaled up limitless without

reduction in purity and recovery or increased time investment. In contrast,

bead isolations are not based on single cells and can be applied to purify

higher cell numbers in the same amount of time. Although bead-based

purification of stem cell-derived CMs has been attempted previously, high

purity isolations resulted in low viability of target CMs and very high cell

loss with a cardiomyocyte recovery of only 20% (Dubois et al., 2011). In

addition, MACS-based positive selection of CMs for VCAM1 or SIRPα has

been described (Fuerstenau-Sharp et al., 2015). However, in these settings,

recovery and viability of CMs was low, which may be related to the column-

based selection step during MACS purifications. Moreover, about 80% of

cardiac Troponin T-positive cells were lost during the procedure. Another

approach to purify CMs is metabolic selection that takes advantage from

the differences in metabolic requirements between cardiac and non-cardiac

2

cells. Prolonged maintenance of cells in presence of lactate-rich and glucose- depleted medium for more than 1 week very efficiently selects for CMs and eliminates non-cardiac cells (Fuerstenau-Sharp et al., 2015; Tohyama et al., 2013). In contrast to metabolic purifications, bead-based isolations of CMs are rapid and can be accomplished within a few hours. In addition, it still needs to be elucidated how metabolic selection during culture affects CMs.

In conclusion, we demonstrated that both purification methods with magnetic beads led to robust enrichment of CMs, which will especially be important for cardiac differentiations of cell lines with poor differentiation efficiencies.

In addition, this will also be advantageous for the standardization and

reproducibility of human stem cell–derived assays in the fields of cardiac

disease modeling, drug discovery and disease modeling. Consecutively,

pure populations of hPSCs-derived CMs will be beneficial for regenerative

medicine and development of advanced 3D tissue constructs to create so-

called organ-on-chip models (for disease modeling and drug screening),

more closely resembling the human organ in vivo.

Materials and Methods

Culture and maintenance of hPSCs

HESC-NKX2.5 ^eGFP/+ cells were maintained as undifferentiated colonies on irradiated mouse embryonic fibroblasts in hESC medium (DMEM-F12 (life technologies), Non-essential amino acids (life technologies), ß-mercaptoethanol (life technologies), KnockOut Serum Replacement (life technologies) and basic Fibroblast Growth Factor (bFGF) (Miltenyi Biotech)) (Elliott et al., 2011). Cells were passaged twice a week using TrypLE Select (Invitrogen) as described previously (Elliott et al., 2011). Differentiation of hPSCs was performed between passages 9 and 30.

Generation of stem cell-derived CMs with the Monolayer protocol

Efficient generation of hPSCs-derived CMs was performed using the

previously described monolayer protocol (Dambrot et al., 2014). Briefly,

thirty thousand hPSCs per cm ² were seeded on 8.3 µg/cm ² matrigel (growth

factor reduced; Corning) the day before differentiation. The following day,

differentiation was initiated by changing the hESC culture medium to BPEL

(Bovine Serum Albumin (BSA) Polyvinylalcohol Essential Lipids) (Ng et

al., 2008) containing growth factors BMP4 (R&D Systems) and Activin A

(Miltenyi Biotech) at a final concentration of 20 ng/ml and 1.5 μM CHIR

99021 (Axon Medchem). 3 days later, cells were refreshed with growth factor-

free BPEL containing 1 μM XAV (Tocris Bioscience). For typical monolayer

differentiations, the medium was changed to fresh BPEL at day 7 and 10

and 13. For optimized monolayer differentiations the medium was replaced

with BPEL supplemented with 5 μM SB431542 (Tocris Bioscience) and 1

μM SAG (Millipore) at day 4 and 7. At day 10, the medium was changed to

BPEL. Optionally and to prevent cell detachment between day 3 and 7, BPEL

containing XAV can additionally be supplemented with 50 μg/ml matrigel

(growth factor reduced; Corning).

2

Generation of stem cell-derived CMs with the Spin-EB protocol

The Spin-Embryoid body (Spin-EB) protocol is an established method to efficiently generate CMs from a wide range of hESC and hiPSC lines (Ng et al., 2008). Concisely, to allow feeder depletion, hPSCs were seeded on 8.3 µg/cm ² matrigel (growth factor reduced; Corning) the day before differentiation. The next day, cells were harvested using TrypLE select. 3000 cells were resuspended in 50 μl of BPEL supplemented with growth factors at a final concentration of 30 ng/ml BMP4 (R&D Systems), 30 ng/ml VEGF (Miltenyi Biotech), 40 ng/ml SCF (Miltenyi Biotech), 30 ng/ml Activin A (Miltenyi Biotech) and 1.5 μM CHIR 99021 (Axon Medchem). It is important to note that growth factor concentrations of Activin A and BMP4 need to be optimized for every new batch and cell line. Cells were transferred to 96-well plates in which outer wells were filled with 150 μl phosphate buffered saline (PBS). Subsequently, plates were centrifuged at 1500 rpm for 3 min to allow Spin-EB aggregation. On day 3 of differentiation, EBs were refreshed with 100 μl growth factor-free BPEL per well. 4 days later, EBs were plated on 0.1% gelatin coated wells to facilitate attachment. Optionally, EBs could also be refreshed with 100 μl BPEL without plating. Followed by another BPEL refreshment at day 10.

Purification of stem cell-derived CMs by Magnetic Bead Isolation Dissociation of hESC-derived CMs to single cells

Dissociation to single cell CMs was performed between day 13 or 14 of

differentiation. Monolayer cultures and Spin-EBs were washed with Ca ²⁺ and

Mg ²⁺ -free PBS ^- and 700 µl 1x TrypLE Select (Invitrogen) was added per 6-well

cell culture dish. Afterwards, the plates were incubated at 37°C not longer

than 10 min. To detach cells carefully, plates were tapped and cells were

dissociated by gently pipetting up and down. Dissociated cells were taken up

in at least 4.5 ml DMEM medium (life technologies) containing 10% fetal calf

serum (FCS). Samples were centrifuged at 250 g for 3 min and resuspended

in Sort buffer (PBS ^- , 0.1% BSA and 2 mM EDTA) for filtering with BD

Falcon™ 5 ml Polystyrene Tube (35 µm). Filtered cells were centrifuged and

used for RNA isolation or resuspended in Sort Buffer for bead purification or

FACS analysis, or in TID medium (triiodothyronine hormone (T3), insulin-

like growth factor 1 (IGF-1), glucocorticoid dexamethasone (Dex) previously

described by (Birket et al., 2015b) for re-plating and continued culture.

Positive Magnetic Bead Isolation

Positive isolation of CMs was carried out with Sheep-Anti-Mouse IgG Dynabeads® (life technologies) according to manufacturer’s instructions after optimization. Briefly, after dissociation and filtration of the initial cell suspension, 0.06 µg VCAM1 antibody (Mouse Anti-Human CD106, Clone 51-10C9, BD Biosciences) was used to label 10 ⁶ single cells at RT for 5-10 min in 150 µl Sort buffer (Ca ²⁺ and Mg ²⁺ -free PBS supplemented with 0.1%

BSA and 2 mM EDTA, not provided by the company). To wash the cells and remove unbound antibodies, cells were centrifuged in a microcentrifuge for approximately 30 s and resuspended in 1 ml Sort buffer, centrifuged again and resuspended in 900 µl Sort buffer and transferred to FACS tubes (BD Falcon™, 5ml round-bottom). For bead preparation, sufficient amount of beads were washed by resuspension in 1 ml Sort buffer, followed by buffer removal on the magnet and resuspension in the initial volume of Sort buffer.

For positive isolation of CMs, 20 μl pre-washed Dynabeads® were added per 106 VCAM1-labeled cells (8 beads per cardiomyocte). For bead coupling, samples were incubated on a rotator for 20 min at 4°C. Afterwards, tubes were placed on a magnet for 2 min and unbound cells (negative fraction) were separated from CMs by removing the supernatant. Bead-bound CMs were washed twice: once with 1 ml Sort buffer and once with 1 ml DMEM- 0.1%-BSA. Finally, cells were resuspended in an appropriate volume of TID medium for further culture or RNA lysis buffer. For up-or downscaling of cell numbers, buffer volumes, amount of antibody and beads were increased or decreased proportionally. This procedure, from start of dissociation until cell recovery, took approximately 75 min.

Positive Magnetic Bead Isolation with Bead Release

Positive isolation of CMs with bead release was accomplished with the

Dynabeads® FlowComp™ Flexi Kit including DSB-X™ biotinylation of

the antibody (life technologies) according to manufacturer’s instructions

with some changes. Briefly, after dissociation and filtration of the primary

cell suspension, 1.3 µg biotinylated VCAM1 antibody (Mouse Ant-Human

CD106, Clone 51-10C9, BD Biosciences, modified by biotinyation) was

used to label 10 ⁶ single cells at RT for 5-10 min in 100 µl Sort buffer. Cells

were washed twice with 1 ml Sort buffer, resuspended in 900 µl Sort buffer

2

and were transferred to FACS tubes. 40 μl pre-washed Dynabeads® were added per 10 ⁶ labeled cells (60 beads per cardiomyocyte). For bead coupling, samples were incubated on a rotator for 25 min at RT. Afterwards, tubes were placed on a magnet for 2 min to remove unbound cells (negative fraction).

Bead-bound CMs were washed twice as described before. To release the beads from the cells, bead-bound cells were resuspended in Release buffer (provided by the company) and placed on a rotator for 2 min at RT. For bead removal, the tube was then placed on the magnet for at least 2 min and the supernatant, containing purified CMs, was transferred into another FACS tube to repeat bead-removal. Finally, cells were centrifuged at 250 g for 3 min and resuspended in an appropriate volume of Sort buffer for FACS analysis or TID medium for further culture. For up-or downscaling of sample size, buffer volumes, amount of antibody and beads was changed proportionally. This procedure, from start of dissociation until cell recovery, took approximately 90 min.

Characterization of stem cell-derived CMs Microscopy

Bright field and fluorescent images at either 10x, 20x or 40x magnification were captured with a Leica AF-6000LX microscope (Leica Microsystems) with controlled temperature and CO ₂ .

FACS analysis

Dissociated and filtered samples were immediately analyzed for determining the percentage of GFP-positive CMs by FACS or stained for VCAM1 (1:500) (CD106, Clone 51-10C9, BD Biosciences) at RT for 10 min. Antibody detection was achieved with a donkey-anti mouse IgG antibody conjugated to APC (1:100) (Jackson Immunoresearch). As a negative control, an isotype-matched mouse IgG1 was used (Mouse IgG1Κ, Clone MOPC-21, BD Biosciences).

Samples were analyzed with a MACSQuant VYB Flow Cytometer (Miltenyi

Biotech). To eliminate cell debris or aggregated cells, events with very low or

high side and forward scatter were excluded. Subsequent data analysis was

performed with MACSQuantify™ Software and results were expressed as the percentage of positive or negative cells or as mean fluorescence intensity.

Immunostaining and Confocal Imaging

Dissociated samples, negative fractions or purified CMs were re-plated on matrigel-coated glass coverslips. After 1 week of recovery, cells were fixed in 2% paraformaldehyde for 30 min at room temperature, followed by permeabilization with 0.1% Triton X 100 (Sigma Aldrich) in PBS for 8 min at room temperature. Blocking was performed with 4% swine serum (Dako) in PBS for 1 h at RT and cells were incubated overnight at 4°C with primary antibodies rabbit polyclonal anti-NKX2.5 (clone H-114: sc-14033, Santa Cruz Biotechnology Inc) and mouse monoclonal anti-α-ACTININ (clone EA-53, Sigma, Saint Louis, Missouri). Detection of primary antibodies was achieved by incubation with corresponding secondary antibodies conjugated to either Alexa Fluor 647 (anti-rabbit Alexa 647 (Invitrogen)) or Cy3 (anti- mouse Cy3 (Jackson Immunoresearch)) at RT for 1 h. Nuclei were stained with 4’, 6-Diamidino-2-Phenylindole (DAPI) (Invitrogen) at RT for 5 min and coverslips were embedded with mowiol (Millipore).

Confocal images were captured with an inverted Leica TCS SP5 microscope (Leica Microsystems). Optical z-stacks stacks were acquired with a 40x or 63x oil immersion objective and image acquisition was performed with LAS AF software (Leica Microsystems).

Quantitative Real-time PCR

Total RNA was isolated from samples before and after purification with

the NucleoSpin RNA isolation Kit (Macherey-Nagel) according to the

manufacturer’s protocol. Approximately 500 ng cDNA were synthesized

with the iScript cDNA Synthesis Kit (BIO-RAD). Real-time quantitative PCR

(qPCR) was carried out in triplicate using the SybrGreen master mix (Applied

Biosystems) and the CFX384 Real-time PCR detection system. Non-template

reactions (replacing cDNA with RNase-free H ₂ O) were used as negative

controls. qPCR reactions were run with a 3-Step protocol: 30 s at 95°C,

2

followed by 40 cycles of 95°C for 5 s and 60°C for 30 s. Primer sequence information is provided in supplemental table 1. Data was analyzed with Bio-Rad CFX Manager software. HARP was used as housekeeping gene and fold changes in gene expression were calculated relative to negative fractions.

Targetgene Forward Reverse

TNNT2 TTCGACCTGCAGGAGAAGTT GCGGGTCTTGGAGACTTTCT NKX2.5 TTCCCGCCGCCCCCGCCTTCTAT CGCTCCGCGTTGTCCGCCTCTGT

ACTN2 CTGCTGCTTTGGTGTCAGAG TTCCTATGGGGTCATCCTTG

Statistical analysis

Results were described as means ± standard error of the means (SEM).

Differences between experimental groups were analyzed by paired Student’s

t-test. p < 0.05 was considered as statistically significant.

Acknowledgements

We would like to thank Richard P. Davis and Stefan Braam for the Spin-EB protocol and C. Dambrot for the monolayer protocol, as well as Harsha D.

Devalla, Matthew J. Birket, Marcelo C. Ribeiro and Dorien Ward for technical

advice on the Spin-EB and monolayer protocol. We also thank Valeria

Orlova and Matthew J. Birket for assistance with the experimental set-up of

the cardiomyocyte purification, Joop C. Wiegant for technical advice during

microscopy and Richard P. Davis for technical advice during FACS.

2

References

van den Berg, C.W., Elliott, D.A., Braam, S.R., Mummery, C.L., and Davis, R.P. (2015).

Differentiation of Human Pluripotent Stem Cells to Cardiomyocytes Under Defined Conditions.

In Methods in Molecular Biology, pp. 1–18.

Birket, M.J., Casini, S., Kosmidis, G., Elliott, D. a., Gerencser, A. a., Baartscheer, A., Schumacher, C., Mastroberardino, P.G., Elefanty, A.G., Stanley, E.G., et al. (2013). PGC- 1α and reactive oxygen species regulate human embryonic stem cell-derived cardiomyocyte function. Stem Cell Reports 1, 560–574.

Birket, M.J., Ribeiro, M.C., Verkerk, A.O., Ward, D., Leitoguinho, A.R., den Hartogh, S.C., Orlova, V. V, Devalla, H.D., Schwach, V., Bellin, M., et al. (2015a). Expansion and patterning of cardiovascular progenitors derived from human pluripotent stem cells. Nat. Biotechnol. 1–12.

Birket, M.J., Ribeiro, M.C., Kosmidis, G., Ward, D., Leitoguinho, A.R., Pol, V. Van De, Dambrot, C., Devalla, H.D., Davis, R.P., Mastroberardino, P.G., et al. (2015b). Contractile Defect Caused by Mutation in MYBPC3 Revealed under Conditions Optimized for Human PSC-Cardiomyocyte Function. CellReports 13, 1–13.

Burridge, P.W., Matsa, E., Shukla, P., Lin, Z.C., Churko, J.M., Ebert, A.D., Lan, F., Diecke, S., Huber, B., Mordwinkin, N.M., et al. (2014). Chemically defined generation of human cardiomyocytes. Nat. Methods 11, 855–860.

Dambrot, C., Buermans, H.P.J., Varga, E., Kosmidis, G., Langenberg, K., Casini, S., Elliott, D. a, Dinnyes, A., Atsma, D.E., Mummery, C.L., et al. (2014). Strategies for rapidly mapping proviral integration sites and assessing cardiogenic potential of nascent human induced pluripotent stem cell clones. Exp. Cell Res. 327, 297–306.

Dubois, N.C., Craft, A.M., Sharma, P., Elliott, D. a, Stanley, E.G., Elefanty, A.G., Gramolini, A., and Keller, G. (2011). SIRPA is a specific cell-surface marker for isolating cardiomyocytes derived from human pluripotent stem cells. Nat. Biotechnol. 29, 1011–1018.

Elliott, D. a, Braam, S.R., Koutsis, K., Ng, E.S., Jenny, R., Lagerqvist, E.L., Biben, C., Hatzistavrou, T., Hirst, C.E., Yu, Q.C., et al. (2011). NKX2-5eGFP/w hESCs for isolation of human cardiac progenitors and cardiomyocytes. Nat. Methods 8, 1037–1040.

Fuerstenau-Sharp, M., Zimmermann, M.E., Stark, K., Jentsch, N., Klingenstein, M., Drzymalski, M., Wagner, S., Maier, L.S., Hehr, U., Baessler, A., et al. (2015). Generation of Highly Purified Human Cardiomyocytes from Peripheral Blood Mononuclear Cell-Derived Induced Pluripotent Stem Cells. PLoS One 10, e0126596.

Lian, X., Zhang, J., Azarin, S.M., Zhu, K., Hazeltine, L.B., Bao, X., Hsiao, C., Kamp, T.J., and Palecek, S.P. (2013). Directed cardiomyocyte differentiation from human pluripotent stem cells by modulating WNT/β-catenin signaling under fully defined conditions. Nat. Protoc. 8, 162–175.

Mummery, C., Zhang, J., Ng, E., Elliott, D., Elefanty, A.G., and Kamp, T.J. (2012). Differentiation of Human ES and iPS Cells to Cardiomyocytes: A Methods Overview. Circ Res 111, 344–358.

Ng ES, Davis R, Stanley EG, E.A. (2008). A protocol describing the use of a recombinant protein- based, animal product-free medium (APEL) for human embryonic stem cell differentiation as spin embryoid bodies. Nat. Protoc. 3, 768–776.

Skelton, R.J.P., Costa, M., Anderson, D.J., Bruveris, F., Finnin, B.W., Koutsis, K., Arasaratnam, D., White, A.J., Rafii, A., Ng, E.S., et al. (2014). SIRPA, VCAM1 and CD34 identify discrete lineages during early human cardiovascular development. Stem Cell Res. 13, 172–179.

Tohyama, S., Hattori, F., Sano, M., Hishiki, T., Nagahata, Y., Matsuura, T., Hashimoto, H., Suzuki, T., Yamashita, H., Satoh, Y., et al. (2013). Distinct metabolic flow enables large-scale purification of mouse and human pluripotent stem cell-derived cardiomyocytes. Cell Stem Cell 12, 127–137.

Tomlinson, M.J., Tomlinson, S., Yang, X.B., and Kirkham, J. (2013). Cell separation:

Terminology and practical considerations. J. Tissue Eng. 4.

Uosaki, H., Fukushima, H., Takeuchi, A., Matsuoka, S., Nakatsuji, N., Yamanaka, S., and

Yamashita, J.K. (2011). Efficient and scalable purification of cardiomyocytes from human

embryonic and induced pluripotent stem cells by VCAM1 surface expression. PLoS One 6.