Analysis of the genomic expression profile in trisomy 18: insight into possible genes involved in the associated phenotypes

Abstract

Trisomy 18, sometimes called Edwards syndrome, occurs in about 1 in 6000 live births and causes multiple birth defects in affected infants. The extra copy of chromosome 18 causes the altered expression of many genes and leads to severe skeletal, cardiovascular and neurological systems malformations as well as other medical problems. Due to the low rate of survival and the massive genetic imbalance, little research has been aimed at understanding the molecular consequences of trisomy 18 or considering potential therapeutic approaches. Our research is the first study to characterize whole-genome expression in fibroblast cells obtained from two patients with trisomy 18 and two matched controls, with follow-up expression confirmation studies on six independent controls. We show a detailed analysis of the most highly dysregulated genes on chromosome 18 and those genome-wide. The identified effector genes and the dysregulated downstream pathways provide hints of possible genotype–phenotype relationships to some of the most common symptoms observed in trisomy 18. We also provide a possible explanation for the sex-specific differences in survival, a unique characteristic of trisomy 18. Our analysis of genome-wide expression data moves us closer to understanding the molecular consequences of the second most common human autosomal trisomy of infants who survive to term. These insights might also translate to the understanding of the etiology of associated birth defects and medical conditions among those with trisomy 18.

Introduction

In 1960, Edwards et al. (1) and Smith et al. (2) described a constellation of clinical features resulting from the presence of an extra copy of chromosome 18, a condition now commonly called ‘Edwards syndrome.’ Although there is a high risk of fetal loss and stillbirth (3, 4), trisomy 18 remains the second most common human autosomal aneuploidy after trisomy 21, with a frequency of around 1 in 6000 live births (5, 6). There is a suspected increased prevalence of trisomy 18 cases due to higher maternal age at conception, the most well-established risk factor in the last 20 years (7); however, recent epidemiological studies have actually shown a decrease. This is related to the use of prenatal diagnosis and an associated high rate of pregnancy termination (7, 8). In 2008, Crider et al. (8) estimated the prevalence of trisomy 18 in the United States at 1 in 2500 pregnancies and 1 in 8600 live births.

One of the unique characteristics of trisomy 18 is that more females than males are born alive. In 1996, Huether et al. (9) compared sex ratios in fetuses and live-born infants that were diagnosed with autosomal aneuploidies of chromosomes 13, 18 and 21. For trisomy 13 and 21, they found no difference in estimates of the sex ratios at the time of prenatal diagnosis or at birth. For trisomy 18, they found a significant loss of males during the second half of the pregnancy, with a male-to-female ratio of 0.69 at birth. In 2006, Niedrist et al. (10) confirmed these results, finding that after 18 weeks of gestation, the risk of spontaneous loss of males was twice as high as the risk of loss of females. Crider et al. (8) showed that the male-to-female ratio among fetuses electively terminated was close to 1, whereas at birth it was 0.66. This difference in sex ratios extends to the survival rate after birth, where female survival is higher than male survival (5, 6, 11).

Trisomy 18 is associated with an array of anomalies of the cardiovascular, nervous and musculoskeletal systems. About 80–100% of infants with trisomy 18 are born with some form of congenital heart defect, among the most common being ventricular and atrial septal defects, patent ductus arteriosus and polyvalvular disease (12–16).

Individuals with trisomy 18 suffer from marked developmental and intellectual disability resulting from structural central nervous system abnormalities, including cerebellar hypoplasia, agenesis of the corpus callosum, microgyria, hydrocephalus and myelomeningocele, present in about 5% of infants with trisomy 18 (12, 17). Functional neurologic features include hypotonia in infancy, hypertonia in older children and central apnea and seizures, the latter occurring in 25–50% of children with trisomy 18, and usually controlled with pharmacological therapy (12).

Many musculoskeletal malformations are also present in most cases of trisomy 18 and include clubfeet, rocker bottom feet, clenched fists with overlapping fingers, polydactyly and syndactyly and arthrogryposis. Cereda and Carey (18) estimated that 5–10% of infants with trisomy 18 have major malformations of the limbs and 50% present with positional foot deformities.

To date, most studies have focused on the prenatal detection of trisomy 18 and possible treatment in the newborn period. Only a few studies have attempted to investigate the genomic changes in individuals with trisomy 18 that could lead to a better understanding of the phenotypes observed.

In 2009, Cody et al. (19) published the first draft of a gene dosage map of chromosome 18 that annotated each gene or region and its associated clinical phenotype. These annotations are continually updated based on clinical and model-based information (20). This tool was developed to provide clinicians with a guide for the prediction and treatment of chromosome 18 abnormalities. Koide et al. (21) used cell-free amniotic fluid supernatant from trisomy 18 and control fetuses to study the transcriptomic differences. They found 251 genes that were differentially expressed, of which only 2.8% were located on chromosome 18. Here we compared fibroblast cell cultures obtained from cases with trisomy 18 with age- and sex-matched controls to gain insight into the molecular mechanisms that underlie trisomy 18 and advance our knowledge about the second most common human aneuploidy.

Results

Chromosome 18 gene expression profile

First we performed a focused analysis of the genes on chromosome 18. We expected a ratio of expression of all genes on chromosome 18 to be increased by 1.5 in the trisomy 18 cases in contrast to controls. Figure 1 shows the expression levels of cases and controls as the log10 of the fragments per kilobase of transcript per million mapped reads (FPKM) values separately for males and females, respectively, for 120 annotated genes on chromosome 18. As expected, we saw that most genes showed higher expression levels in cases than controls for both males (76%) and females (92.5%). This is represented by the points in both the male and female graphs falling below the line that represents a 1:1 ratio of expression between cases and controls. This finding also implies that there are more genes with expression levels that varied from this pattern in the male sample (24%) than in the female sample (7.5%). We identified seven genes with a fold change (FC) of at least ±4 in the male sample and 3 in the female sample.

Figure 1

Log10 of the FPKM values for all the annotated genes in chromosome 18 (N = 120) for the male and female samples. The x-axis shows values for the male or female case (trisomy 18), and the y-axis shows values for male or female control (Euploid).

Among these more highly dysregulated genes, we focused our attention on PIEZO2 due to the possible relationship between its known function as the principal mechanotransduction channel for proprioception (30) and its association with phenotypes such as arthrogryposis and scoliosis (31, 32), which are commonly observed in trisomy 18. We found that the expression of PIEZO2 was significantly reduced, despite being located on the triplicated chromosome 18. For the male with trisomy 18, there was an almost 39-fold reduction of expression in contrast to the control. In our female case, PIEZO2 expression was also reduced by a magnitude of 5.5-fold in contrast to control (Table 2). To confirm these results, we measured the relative expression of PIEZO2 using qRT-PCR (Fig. 2) as an alternative method and included three additional male and female controls (Table 1).

Figure 2

qRT-PCR results showing normalized fold expression for the genes of interest. The left column shows results for the male case (trisomy 18) and four controls (Euploid). The right column shows results for the female case (Trisomy 18) and four controls (Euploid).

Table 1

List of samples used in the research

Sample ID	Sex	Age	Type	Experiment
GM2732	Male	4 days	Case	RNA-Seq + qRT-PCR
GM143	Female	90 days	Case	RNA-Seq + qRT-PCR
CRL2097	Male	Newborn	Control	RNA-Seq + qRT-PCR
UMB5653	Female	100 days	Control	RNA-Seq + qRT-PCR
5817	Male	213 days	Control	qPCR
5837	Male	71 days	Control	qPCR
5891	Male	1 year 27 days	Control	qPCR
5655	Female	91 days	Control	qPCR
5761	Female	32 days	Control	qPCR
5770	Female	145 days	Control	qPCR

Genome-wide dysregulation observed in trisomy 18

Next we examined the rest of the genome to determine to what extent the extra copy of chromosome 18 dysregulates genome-wide expression. Figure 3 shows the expression levels of cases and controls as the log10 of the FPKM values for males and females, respectively, for all the genes not on chromosome 18. Overall, we found that approximately 56% of non-chromosome 18 genes have expression levels that do not differ by more than 20% for cases and controls, both in males and females. Due to the fact that we are comparing one case to its control, we focused our attention on those genes with extreme differences in expression levels, equivalent to a fold change (FC) of ±25 in males (Table 3) and females (Table 4). Again, we found sex-specific differences. First, the number of genes with this extreme dysregulation in females is 30 in contrast with 81 genes in males. Second, the number of downregulated genes is similar to the number of upregulated genes in females (11 vs 19), whereas our male case shows 60 downregulated genes in contrast to 21 upregulated genes. This overall increased dysregulation suggests an explanation for the increased severity of outcomes in males in contrast to females.

Figure 3

Log10 of the FPKM values for the genome-wide annotated genes minus the genes in chromosome 18 for the male and female samples. The x-axis shows values for the male or female case (Trisomy 18), and the y-axis shows values for the male or female control (Euploid).

Among the highly dysregulated genes, we found several transcription factors and protein-coding genes that could be linked to the phenotypes observed in trisomy 18. Among them, the short stature homeobox 2 (SHOX2) gene was identified as an excellent candidate gene for many of the severe symptoms associated with trisomy 18. The expression of SHOX2 in the male case with trisomy 18 was close to 0, in contrast with strong expression in the male control. In females, we saw upregulation in our case equivalent to just 1.7-fold (Table 2). We were able to verify this through qRT-PCR with the addition of three more control males and three control females (Fig. 2). Again, this male-specific dysregulation could help explain the increased severity and lethality among males in contrast with females with trisomy 18.

Table 2

Candidate genes studied, fold changes observed in males and females and system affected

Chr Location	Gene	Alias	Fold Change Male	Fold Change Female	System
Chr 1	ALX3	Aristaless-like homeobox 3	−254.9	7.16	Skeletal
Chr 3	SHOX2	Short stature homeobox 2	<−1000	1.73	Skeletal
Chr 4	HAND2	Hand and neural crest derivatives expressed 2	5.75	1.69	Skeletal
Chr 7	TWIST1	Twist family BHLH transcription factor 1	−2.99	−1.51	Skeletal
Chr 11	ALX4	Aristaless-like homeobox 4	−4.9	21.2	Skeletal
Chr 12	TBX5	T-box 5	−4.72	1.66	Heart
Chr 14	BMP4	Bone morphogenetic protein 4	−19.3	−1.32	Skeletal, heart
Chr 15	ACAN	Aggrecan	−62.84	−351.25	Skeletal
Chr 15	FMN1	Formin 1	−25.93	−64.63	Skeletal
Chr 17	TBX4	T-box 4	<−1000	−4.88	Skeletal
Chr 17	NOG	Noggin	−28.27	−1.55	Skeletal
Chr18	PIEZO2	Piezo-type mechanosensitive ion channel component 2	−38.58	−5.53	Skeletal

The existing literature regarding the functions and possible implications of altered expression of SHOX2 led us to check the expression of several other potentially associated highly dysregulated genes, such as the transcription factor TBX4 (33) and the protein-coding gene ACAN (34). A more detailed description of the functions of these genes and their interactions is provided in the Discussion. Briefly, we found an almost absent expression of TBX4 in males and a 5-fold downregulation in females (Table 2). Similarly, for ACAN, we observed a 62.84 FC reduction in our male case, consistent with the loss of SHOX2 expression. In the female case, although SHOX2 was not downregulated, we found an almost absent expression of ACAN (Table 2). We were able to confirm this result through qRT-PCR with the addition of three control males and three control females (Fig. 2).

Another highly dysregulated gene in both males and females is FMN1 (formin 1), which is associated with skeletal development (35, 36) (Tables 3 and 4). We found that its expression was downregulated in males by 25.9-fold and 64.6-fold in females. We confirmed these fold changes using qRT-PCR (Fig. 2).

Among the most dysregulated genes in males, we also found NOG (noggin) (Table 3), a bone morphogenetic protein (BMP) signaling inhibitor known to be associated with brachydactyly and neural tube defects, conditions that are often found in trisomy 18. We found NOG to be significantly reduced: the male with trisomy 18 showed the equivalent to a negative fold change of −28.2; in the female, the change was less pronounced, but still reduced by −1.55-fold (Table 2). The qRT-PCR confirmed these results both in males and females (Fig. 2).

Finally, we found another transcription factor, ALX3 (aristaless-like homeobox 3), to be highly dysregulated. Defects in ALX3 expression have been linked to frontonasal dysplasia I and neural tube defects. ALX3 was downregulated in the male case, equivalent to a negative fold change of −255, and upregulated in the female case (7-fold upregulation) (Table 2).

Discussion

Trisomy 18 leads to malformations associated with three major developmental processes, including the skeletal, the cardiovascular and the neurological systems. Our results show dysregulation of genes involved in these three processes and offer some insight into possible dysregulated genes responsible for the clinical manifestations of trisomy 18. Here we focus on genes that may be involved in the associated skeletal and heart defects to show the value of these types of expression studies.

Development of the skeletal system

Our strongest candidate gene on chromosome 18 associated with skeletal development was PIEZO2; this gene was downregulated in the trisomy 18 cases despite being located on chromosome 18. PIEZO2 acts as the major mechanotransducer of mammal mechanosensory neurons or proprioceptors (30). Several studies have examined the effects of gain- and loss-of-function mutations in PIEZO2, but perhaps more relevant to the situation in trisomy 18 is the fact that the loss of PIEZO2 expression gives rise to some of the skeletal problems affecting patients with trisomy 18. In 2015, Woo et al. (30) show that the lack of expression of PIEZO2 in their mouse lines caused abnormal limb positions and uncoordinated body movements. In humans, mutations leading to a loss of PIEZO2 are associated with skeletal conditions, including arthrogryposis and scoliosis, as well as other problems like muscular atrophy of the distal lower limbs and mild distal sensory involvement (31, 32). The effects of a gain of PIEZO2 expression show some of the same skeletal problems as observed when there is a loss-of-function. In 2013, Coste et al. (37) described two distinct gain-of-function PIEZO2 mutations in patients with distal arthrogryposis type 5.

Both arthrogryposis multiplex congenita (38–40) and scoliosis (18, 41) are hallmarks of trisomy 18. The observation that both loss-of-function and gain-of-function mutations in PIEZO2 lead to skeletal problems suggests that the expression of this gene needs to be tightly controlled. The important question that remains is why trisomy 18 leads to a significant reduction in the expression of PIEZO2 despite being present in three copies.

We also focused our attention on one of the most highly dysregulated genes outside chromosome 18 in males. SHOX2, a transcription factor located in chromosome 3, has been the subject of several investigations to understand its function both in mouse and human. Developmental studies in mouse show that SHOX2 is expressed in the embryonic mesodermal tissues of the face and is involved in nose and palate formation, the developing heart mesoderm and the mesenchyme of the developing limbs (42, 43). These same studies determined that the human and mouse orthologs are highly homologous, showing 100% homology at the protein level. Blaschke et al. (43) mapped the human SHOX2 gene to chromosome 3q25–26, the candidate region for Cornelia de Lange syndrome (CDL). With respect to skeletal development, CDL is characterized by growth retardation, facial deformities including cleft palate and reductive limb development (44). Many of these symptoms overlap with those of trisomy 18, including growth retardation, limb deformities, scoliosis and orofacial clefts (18).

Using a mouse model lacking SHOX2 expression, Glaser et al. (33) established a feedback connection between SHOX2 and another transcription factor, TBX4. They showed that their interaction depended on developmental context: during forelimb and hindlimb development, TBX4 activates SHOX2; in the forelimb bud, SHOX2 regulates TBX4. We showed a significant loss of expression in TBX4 in our male case and a 5-fold reduction in the female.

Table 3

List of genes in our male case/control showing a dysregulation with a minimum fold change of plus/minus 25

Male case vs male control FC ≤ −25	Male case vs male control FC ≥ 25
Gene name	Gene name	Gene name
HSPB3	OLFML1	KRT14
SNORD104	NBEA	ALDH1A3
INPP4B	SFRP2	NCAM2
FAM162B	SYPL2	FZD8
GFRA1	TFAP2C	LRRC17
PM20D2	FAM19A5	NRN1
STK32B	ADGRD1	BMP6
SEPP1	DAPK2	CXCL6
STMN2	FAM65C	LYPD1
FAM86FP	NFASC	IL33
SHOX2	ISL2	RBP1
MAF	PRELP	EN1
GLIS1	TRPC6	PAX8-AS1
BRWD1	ACKR3	ANXA10
ALDH1A1	TTC12	SCRG1
ITIH5	ACAN	PLPP2
ADH1B	GRIA1	CDH18
HOTAIR	FAM174B	NMRAL1
TBX4	WFDC1	HOXB-AS3
PGM5	SPESP1	CLGN
OGN	C1orf115	HOXB7
HSD17B2	MYO1D
LINC00839	OSR2
HOXD11	IGFBP2
C8orf4	LMO7-AS1
PTH1R	ELN
GALNT16	NOG
IGF2	FXYD1
DAPK1	FMN1
ALX3	KISS1

Table 4

List of genes in our female case/control showing a dysregulation with a minimum fold change of plus/minus 25

Female case vs female control FC ≤ −25	Female case vs female control FC ≥ 25
Gene name	Gene name
PI4KAP1	HOXD11
ACAN	COL15A1
HOTAIR	C8orf4
MYO19	AGTR1
PSMG4	MMP1
FMN1	ANXA10
PGM5	CDC25A
THBS4	FAM65C
PITX1	NMRAL1
MAGI1	CD24
SFRP2	SH3YL1
	GLIS1
	TBX1
	CCNE2
	TAGLN3
	INPP4B
	BLM
	TCF19
	XRCC4

ACAN, or aggrecan, may be another player in this system, as it is a transcription target of SHOX2 and is involved in skeletal development (34). ACAN plays a key role in the cartilage extracellular matrix and is important in mediating chondrocyte–chondrocyte and chondrocyte–matrix interactions (45). Aza-Carmona et al. (34) showed that SHOX2 is a key part of the complex SOX5/SOX6/SOX9/SHOX2 that activates the ACAN enhancer. Our data suggest that ACAN is significantly downregulated both in the male and female cases (Table 2).

Limb malformations, including oligosyndactyly, brachydactyly and syndactyly, are also prominent features in trisomy 18 (46–48). The significantly reduced expression level of FMN1 (formin 1) may also be involved in malformations of the limbs, as suggested by a patient with a homozygous deletion of 263-kb at the FMN1 locus who presented with oligosyndactyly, among other symptoms (36). A knockout mouse model of FMN1 showed signs of oligodactyly, potentially as a result of increased activity of the BMP signaling (35).

Noting this association, we examined the expression level of NOG, a BMP signaling inhibitor, and found that it was significantly reduced in our cases (Table 2). In humans, several reports have established an association between mutations in the NOG gene and brachydactyly (50–53), a known phenotype present in trisomy 18. Also, NOG has been associated with other developmental bone disorders such as symphalangism (49–52), multiple synostosis syndrome (49) and tarsal–carpal coalition syndrome (52).

Finally, the presence of frontonasal dysplasia in trisomy 18 fetuses (54, 55) led us to examine ALX3 and ALX4, two genes associated with frontonasal dysplasia I and II, respectively, when the expression is downregulated (56–58). In our data, ALX3 and ALX4 expression levels were downregulated in the male case, but both were upregulated in the female case (Table 2).

Based on the literature, expression levels of TWIST1 and HAND2 affect an interaction network involved in limb development, including ALX3 and ALX4 (59–61). For example, mouse lines with reduced expression of TWIST1 showed polydactyly, but reduced expression of HAND2 could recover this phenotype. Thus, a balance of TWIST1 and HAND2 expression levels may be key to the correct development of the limb proximodistal axis, along with ALX3 and ALX4. Our results showed a downregulation of TWIST1 in our male case but also showed increased levels of expression in HAND2 (Table 2). Overall, the balance of expression levels of this interaction network may be required for normal limb development.

We think that the dysregulation of key transcription factors such as SHOX2, TBX4, ALX3, ALX4 and TWIST1, along with protein-coding genes like PIEZO2, ACAN, FMN1, NOG and HAND2, plays a key role in trisomy 18-associated skeletal development.

Development of the heart

Congenital heart malformations are a common feature in patients with trisomy 18. Examination of our expression data to identify candidate genes for abnormal heart development led us to three transcription factors known to interact: SHOX2, TBX5 and BPM4. In mouse, TBX5 regulates the expression of SHOX2, and SHOX2 controls BMP4 expression in the pacemaker region of the developing heart (62). Studies have shown that heart defects due to a lack of SHOX2 lead to embryonic lethality in mice (43, 63). Indeed, our data show a downregulation of SHOX2, TBX5 and BMP4 in our male case. In the female, expression of SHOX2 is slightly increased, potentially explaining the relatively normal expression levels of TBX5 and BMP4, at least in contrast to the male case (Table 2). We speculate that the observed dysregulation of SHOX2, along with the transcription factors BMP4 and TBX5, may explain, at least in part, abnormal heart development resulting from trisomy 18 and, once again, may explain the sex-specific survival rates.

Here we report the first study on the whole-genome expression of differentiated cells from patients with trisomy 18. A detailed analysis of the most highly dysregulated genes provides hints of possible genotype–phenotype relationships to some of the most common symptoms observed in trisomy 18. We focus on some of the hallmark features of this syndrome, namely, skeletal and heart malformations, but provide data on expression levels genome-wide for future studies.

We also provide a possible explanation for the sex-specific differences in survival in trisomy 18. A unique characteristic of trisomy 18 is the higher survival rate among females in contrast to males, both prenatally and postnatally (3, 5, 6, 8, 9, 10, 64). Based on our whole-genome expression data, we speculate that the moderate differences in global expression dysregulation observed in females in contrast to males leads to this increased survival rate.

Combined with previous research that developed tools to better understand human trisomy 18 and its phenotype (19–21, 65), our research on genome-wide expression moves us closer to understanding the second most common human autosomal trisomy of infants who survive to term. This syndrome leads to clinically significant abnormalities and considerable emotional distress among families. It is also important to note that while our work provides insights on possible relationships between specific gene expression levels and phenotypes observed in trisomy 18, the limitation of this study resides in our small sample size. More data and future studies containing a bigger sample size are needed to confirm our results and to add to new discoveries related to the global genomic changes that occur as a consequence of the extra copy of chromosome 18.

Materials and Methods

Sample description

Our approach was to identify the differences in gene expression between two cases of trisomy 18 and two age- and sex-matched controls (Table 1). Fibroblast cell samples were obtained from different biorepositories: the Coriell Institute for Medical Research (GM2732, GM143), ATCC Biorepository Services (ATCC® (CCD-1079Sk ATCC® CRL-2097™) and the NIH biorepository (UMB5653).

Additional samples for follow-up of qRT-PCR studies were obtained from Coriell Institute for Medical Research (Table 1).

Cell culture

Fibroblasts were cultured in 6-well plates using DMEM (Gibco cat. #10569-010) with 10% fetal bovine serum (Gibco cat. # 10439024), 1% MEM non-essential amino acid solution (Gibco cat. # 11140050) and 0.1% β-mercaptoethanol (Gibco cat. #21985-023) and then placed at 37°C and 5% CO_2. At 80% confluence, cells were collected and passed into new plates. Briefly, 0.05% trypsin was added to the cultures and incubated for 3 min at 37°C, after which the cells were collected and spun down at 300 g for 5 min, counted and plated in new plates and media. After six passages, RNA was extracted from the cultures. Extracted total RNA was submitted to the Emory Integrated Genomics Core for RNA quality control and RNA sequencing in triplicate for each of the four samples in our study. RNA quality was tested using the RNA integrity number (RIN) method (22). To proceed to RNA sequencing, we established a cutoff RIN of 8.5, which guarantees high-quality RNA suitable for RNA sequencing. Prior to sequencing, we karyotyped all four samples for each of the three RNA replicates to confirm the original chromosome constitution. To do so, cultures were plated in T25 flasks and sent to WiCell Research Institute (Madison, WI) for karyotyping. A minimum of 20 metaphases were analyzed to confirm the expected karyotype for controls and cases.

RNA sequencing

RNA sequencing was performed in triplicate for each sample. Once we obtained the results of the RNA sequencing, RNA-Seq reads for the samples (median number of reads ~ 83.4 million) were first checked for quality using FastQC (23). The RNA-Seq reads were then processed into transcript-level expression data using the ‘new Tuxedo’ protocol (24). Briefly, the RNA-Seq reads were first mapped to hg38 using HISAT (25). Next, the reads were sorted, assembled and FPKM counts obtained using SAMtools (26) and stringTie (27). Ballgown (28) was then used to run the differential expression analysis.

Quantitative PCR

To verify the expression levels of the genes of interest resulting from the initial comparison of control and trisomy 18 samples, we performed qRT-PCR using RNA from each sample and additional controls (Table 1). Starting with RNA obtained from fibroblast cultures, we used a BIORAD iScript cDNA synthesis kit to obtain cDNA following the manufacturer’s protocol. 100 ng of each reverse transcription reaction were then used as a template for qRT-PCR using TaqMan® Fast Advanced Master Mix with TaqMan probes supplied by the manufacturer. Cycling conditions were followed as recommended by the TaqMan® Fast Advanced Master Mix user guide. Relative expression levels were determined based on the delta–delta CT method (29), using GAPDH as an endogenous control.

Acknowledgements

We thank The Claude B. Cook Charitable Trust for a generous gift to supporting this project.

Conflict of interest statement

The authors certify that they have NO financial or non-financial conflict of interest in the subject matter or materials discussed in this manuscript.