Human pluripotent stem cell-derived cardiomyocytes: Genome-wide expression profiling of long-term in vitro maturation in comparison to human heart tissue

Abstract

Cardiomyocyte-like cells (CMs) derived from human pluripotent stem cells (hPSCs) present a valuable model for human disease modeling, studying early human development and, potentially, developing cell therapeutic approaches. However, the specification of early hPSC-derived CMs into defined cardiac subtypes such as atrial and ventricular cells is not well understood and, thus, poorly controlled. Moreover, the maturation status of hPSC-CMs is not well defined, yet it is known that these cells undergo at least some degree of maturation upon longer term in vitro culture. To gain insight into this process, and to assess their developmental status, we have recently generated a data set of hPSC-CMs monitoring global changes in gene expression upon long term maintenance in vitro, in comparison to human atrial and ventricular heart samples (GEO accession number GEO: GSE64189). These data present a rich resource for evaluating the maturation status of hPSC-CMs, for identifying suitable markers for subtype-specific gene expression, as well as for the generation of functional hypotheses. Here, we provide additional details and quality checks of this data set, and exemplify how it can be used to identify maturation-associated as well as cardiac subtype-specific markers.

Specifications
Organism/cell line/tissue	Homo sapiens/hPSC line FS3F.2/Atrial and ventricular heart biopsies
Sex	Male
Sequencer or array type	HumanHT-12 v4
Data format	Raw and processed
Experimental factors	hPSC-derived CMs: untreated cells harvested at weekly intervals
Experimental features	hPSCs were differentiated into cardiomyocytes and maintained under adherent culture conditions in low fetal calf serum-containing media. Samples were taken at weekly intervals up to 8 weeks. Human atrial and ventricular biopsy samples served as in vivo reference.
Consent	Written informed consent was obtained from patients where applicable.
Sample source location	Münster, Germany

Direct link to deposited data

Deposited data can be found here: http://www.dtd.nlm.nih.gov/geo/query/acc.cgi?acc=GSE64189.

Experimental Design, Materials and Methods

Generation and processing of samples

Cell culture. hPSCs, cell line FS3F.2 [1], were maintained in FTDA medium [2], on Matrigel™-coated dishes. Cardiac differentiation was induced as described [3]. In brief, fully confluent hPSC cultures were harvested using Accutase™ digestion, and replated onto Matrigel-coated 24-well plates (500,000 cells per well in 2 ml of day 0 differentiation medium). An aliquot of cells was used for RNA isolation (0 week time-point). Day 0 differentiation medium contained Knockout™ DMEM, insulin/transferrin/selenium, 10 μM Y27632, penicillin/streptomycin/l-Glutamine, 10 ng/ml FGF2, 1 ng/ml BMP4, and 1 μM CHIR99021 [3]. Day 1 medium contained Knockout™ DMEM transferrin/selenium penicillin/streptomycin/l-Glutamine, and 250 μM phospho-ascorbate (TS medium). On days 2 and 3, cells were fed with TS medium supplemented with 2 μM of WNT inhibitor IWP-2. Hence after, cells were maintained in basal TS medium. Spontaneous beating was observed from day 6 onwards. CM differentiation efficiency was above 85% as judged by FACS counting [3]. On day 7, the cells were harvested using 1 × TrypLE Select (Life Technologies), and pooled from independent samples. An aliquot of cells was used for RNA isolation (1 week time-point), and the remaining cells replated at ~ 250,000 cells per well of a Matrigel-coated 24-well plate, in 2% fetal calf serum/Knockout DMEM/penicillin/streptomycin/l-Glutamine. Thereafter, medium was replaced every 3–4 days. Samples of maturating hPSC-CMs were taken at weekly intervals up to 8 weeks, to be subjected to microarray analysis. Total RNA was isolated using Qiagen RNeasy columns with on-column DNA digestion.

Human heart samples. RNA samples from left and right atrial appendages have been previously described [4]. RNA samples were pooled from six independent patients per tissue type. The human left and right ventricular RNA samples were from a commercial supplier (Biocat #R1234138‐50‐BC and #R1234139‐50‐BC, respectively).

500 ng of total RNA from each biological sample was used as input for the generation of biotin-labeled cRNA using an Illumina® TotalPrep™ RNA amplification kit (Life Technologies). Following the manufacturer's instructions, in vitro transcription of double-stranded cDNA was performed for 14 h, in a PCR cycler. Purified biotin-labeled cRNA was eluted in a volume of 100 μl and quality-checked on a 2100 Bioanalyzer device (Agilent Technologies). cRNA samples were adjusted to 150 ng/μl in water, and hybridized onto Illumina HumanHT-12 v4 bead arrays following the manufacturer's instructions throughout. Hybridization was carried out at 58 °C for 18 h. Staining with streptavidin-Cy3 (GE Healthcare #PA43001) was carried out as recommended, at a concentration 1 μg/ml in blocking buffer. Dried bead arrays were scanned on a HiScan SQ device (Illumina) using default settings.

Technical and biological data quality assessment

Scanned images were confirmed to show an overall clean fluorescence spot morphology with high signal-to-noise ratio, and array data were confirmed to display an average P₉₅ intensity of > 800 (a.u.). Inspection of raw data in GenomeStudio suggested an overall high hybridization stringency, according to internal mismatch control probes, and no major hybridization artifacts. Following these routine checks, all separately hybridized samples were background-subtracted and normalized using the Cubic Spline algorithm in GenomeStudio. This revealed a high degree of similarity between the left/right human heart samples, suggesting that they could be combined in silico. To assess overall human heart-specific gene expression regardless of chamber-specific differences, data was additionally analyzed by combining all human atrial and ventricular samples using GenomeStudio software.

As a biological quality control step, known markers were used to assess differential gene expression between the distinct types of samples. In line with the expectations, hPSC-specific genes OCT4, NANOG and SOX2 were only expressed in the undifferentiated (0 week) cells. Conversely, structural cardiac markers (MYH6, MYL4, MYL7) were indeed only expressed in the differentiated samples and not in the undifferentiated cells (Fig. 1A, left). Focusing on gene expression changes upon long-term in vitro culture, maturation markers such as MYL2 and MYH7 were upregulated in the late (8 weeks) samples, whereas markers of immature hPSC-CMs were indeed overrepresented in the early (1 week) samples (Fig. 1A, middle). As supported by functional assays [3], however, there were only marginal differences between 4 weeks and 8 week-old hPSC-CMs, suggesting that the cells reach a rather stable transcriptomic state from approximately 4 weeks onwards (Fig. 1A, right). Furthermore, the expression pattern of the pan-cardiac marker ACTC1 (cardiac muscle alpha actin) served to indicate an overall stable cardiomyocyte signature in all differentiated samples (1 to 8 weeks, Fig. 1B).

Fig. 1.

Biological quality assessment. (A) Scatter plot analysis (power scale) of early hPSC-CMs versus undifferentiated hPSCs (left), late vs. early hPSC-CMs (middle), and 8 week vs. 4 week-old hPSC-CMs (right). Linear correlation coefficients are provided as a measure for global transcriptome similarity. Blue colored dots indicate data points of known marker genes. See text for discussion. (B) ACTC1 as a pan-cardiac marker is expressed at similar levels in all differentiated in vitro samples (from 1 week onwards). Error bars indicate bead standard deviation extracted from GenomeStudio.

Basic data analysis

A comparison of human atrial and ventricular samples allowed for the identification of marker genes. Using stringent filtering criteria (> 10-fold differences in gene expression), these included known structural genes, ion channels, as well as transcriptional regulators (Fig. 2, Table 1). For instance, myosin light chain 2 (MYL2, also known as MLC2v) presents a rather stringent ventricular marker. By contrast, the frequently used MYL7 (MLC2a) was only about 2-fold enriched in atrial tissue, suggesting that it does not well discriminate between human cardiac subtypes. Instead, the natriuretic peptide-encoding genes NPPA and NPPB are excellent atrial markers according to this analysis.

Fig. 2.

Comparison of adult human atrial and ventricular tissue (scatter plot of combined left/right samples). Selected marker genes are highlighted by colored dots. Note that MYL7 (MLC2a) is also highly expressed in ventricular tissue.

Table 1.

Marker genes discriminating between human atrial and ventricular tissue and corresponding expression levels in late (4–8 weeks) hPSC-CMs. P values for differential gene expression are below 0.01 in all cases.

Symbol	Atria signal	Ventricles signal	Fold change (V/A or A/V)	hPSC-CMs 4–8 weeks signal	Definition
Human ventricular markers
DLK1	5	341	68	924	Delta-like 1 homologue (Drosophila)
IRX4	5	204	41	252	Iroquois homeobox 4
MYL2	444	18,009	41	6769	Myosin, light polypeptide 2, regulatory, cardiac, slow
XDH	5	188	38	33	Xanthine dehydrogenase
TMEM190	7	255	37	46	Transmembrane protein 190
HYAL2	16	532	34	112	Hyaluronoglucosaminidase 2
CPNE4	20	650	32	Below detection	Copine IV
CYP1A1	5	158	32	Below detection	cytochrome P450, family 1, subfamily A, Polypeptide 1
IRX5	14	387	27	42	Iroquois homeobox protein 5
C3orf23	5	112	22	24	Chromosome 3 open reading frame 23
Human atrial markers
NPPB	15,780	19	824	2043	Natriuretic peptide precursor B
HAMP	1548	5	310	29	Hepcidin antimicrobial peptide
MYBPHL	922	5	184	Below detection	Myosin binding protein H-like
NPPA	21,892	143	153	1615	Natriuretic peptide precursor A
PLA2G2A	478	5	101	Below detection	Phospholipase A2, group IIA (platelets, synovial fluid)
COMP	396	5	79	Below detection	Cartilage oligomeric matrix protein
TCEAL2	761	12	63	19	Transcription elongation factor A (SII)-like 2
SLPI	274	5	55	Below detection	Secretory leukocyte peptidase inhibitor
DHRS9	1441	28	52	2251	Dehydrogenase/reductase (SDR family) member 9
HP	1316	27	49	Below detection	Haptoglobin

Analyzing the hPSC-based together with the in vivo data allows determination of shared and divergent gene expression between hPSC-CMs and human heart. In line with the expectations, scatter plot and clustering analyses between human heart versus (i) undifferentiated hPSCs, (ii) early (1 week) hPSC-CMs, and (iii) maturated hPSC-CMs (combined 4–8 weeks samples) suggest that over time, hPSC-CMs tend to become more similar to the human adult heart reference (Fig. 3). This tendency is also supported by the fact that a set of known immature CM markers (NKX2.5, IRX4, and others) declines upon long-term in vitro culture, whereas a set of maturation marker genes (MYH7, MYL2, and others) reached human heart-like expression levels over time (blue and orange colored genes, respectively, in Fig. 3). However, even in late hPSC-CMs a number of human heart genes were not expressed, and the global similarity to the in vivo reference appears to be rather limited (red colored genes in Fig. 3, Table 2).

Fig. 3.

Comparison of hPSC-CMs and human heart tissue. (A) Global correlation-based dendrogram showing early (1 week-old) and late (combined 4–8 weeks-old) hPSC-CMs clustering closer to human heart (combined atrial and ventricular samples) than to undifferentiated hPSCs. (B) Scatter plots comparing the indicated samples. Examples of genes enriched in the four types of samples are highlighted by different colors (black: hPSCs, blue: early hPSC-CMs, orange: late hPSC-CMs and human heart, red: human heart).

Table 2.

Selected genes enriched in early hPSC-CMs, late hPSC-CMs, and human heart. P values for differential gene expression are below 0.01 in all cases.

Symbol	hPSCs 0 week signal	hPSC-CMs 1 week signal	hPSC-CMs 4–8 weeks signal	Human heart atr. & ventr. signal	Definition
Genes upregulated in early hPSC-CMs
NKX2-5	Below detection	3201	1549	583	NK2 transcription factor related, locus 5
IRX4	67	1036	276	132	Iroquois homeobox 4
TBX2	Below detection	1536	331	152	T-box 2
COL2A1	20	694	39	Below detection	Collagen, type II, alpha 1
ISL1	23	269	17	Below detection	ISL1 transcription factor, LIM/homeodomain
HAND1	Below detection	4480	1231	632	heart and neural crest derivatives expressed 1
ID2	110	1339	83	228	Inhibitor of DNA binding 2
LEF1	15	501	47	Below detection	Lymphoid enhancer-binding factor 1
IRS1	53	546	241	117	Insulin receptor substrate 1
MDK	23	532	201	Below detection	Midkine (neurite growth-promoting factor 2)
Genes upregulated in late hPSC-CMs and human heart
MYH7	Below detection	938	11,781	13101	Myosin, heavy chain 7, cardiac muscle, beta
MYL2	Below detection	41	6838	7737	Myosin, light polypeptide 2, regulatory, cardiac, slow
TNNI3K	below detection	46	757	659	TNNI3 interacting kinase, transcript variant 2
HSPB7	Below detection	3339	5428	5614	Heat shock 27 kDa protein family, member 7
PLN	Below detection	1534	4645	4638	Phospholamban
CSRP3	below detection	540	1297	2317	Cysteine and glycine-rich protein 3
ACTN2	Below detection	813	1130	1739	Actinin, alpha 2
RBM20	below detection	440	1121	1426	RNA binding motif protein 20
TRIM63	Below detection	367	1116	1544	Tripartite motif-containing 63
CORIN	Below detection	444	934	976	Corin, serine peptidase
Genes upregulated in human heart
CASQ2	Below detection	10	23	6699	Calsequestrin 2 (cardiac muscle)
MB	Below detection	727	633	12450	Myoglobin, transcript variant 1
MYOM2	14	41	180	8272	Myomesin (M-protein) 2
TCAP	Below detection	264	922	8830	Titin-cap (telethonin)
MYH11	below detection	147	201	1835	Myosin, heavy chain 11
TNNI3	34	331	401	9157	Troponin I type 3 (cardiac)
S100A1	Below detection	10	10	1482	S100 calcium binding protein A1
DES	Below detection	14	56	4555	Desmin
HRC	24	1294	1105	4976	Histidine rich calcium binding protein
MYOM1	Below detection	4787	4668	10215	Myomesin 1, transcript variant

Discussion

Despite the fact that neither the in vitro-derived nor the in vivo samples consisted of pure populations of cardiomyocytes, this data set suggests that meaningful biological information can be extracted from it. The combined hPSC-CM/adult human heart data hence presents a useful resource for evaluating the maturation status of hPSC-CMs at the transcriptional level as well as for assessing cardiac subtype-specific gene expression. Notably, according to our analysis, the frequently used MYL7 (MLC2a) gene appears to be unsuited for discriminating between atrial and ventricular subtypes. Our data instead suggests alternative genes, such as NPPA and NPPB, as being well-suited markers for evaluating atrial subtype specification in hPSC-CMs.