Joint analysis of left ventricular expression and circulating plasma levels of Omentin after myocardial ischemia

Patients presenting for elective aortic valve replacement surgery by a single surgeon were enrolled in this study. They provided written informed consent with Institutional Review Board approval. Clinical characteristics of the study population are shown in Table 1. Patients were deemed to have CAD if obstructive lesions were noted on left heart catheterization just prior to aortic valve surgery requiring concomitant revascularization of at least one vessel. Punch biopsies (~3–5 μg total RNA content) were obtained intra-operatively from the site of a routinely placed surgical vent in the anterolateral apical left ventricle at the initiation of cardiopulmonary bypass (CPB) and again after an average of 81.73 ± 27.89 min of aortic cross clamp time. Specifically, the biopsies were taken from within the myocardium and did not include infarcted tissue. Pre-ischemia and post-ischemia biopsies were taken from the same site in the track made for the LV vent. During aortic cross clamping, the heart was infused with intermittent cold blood cardioplegia (8:1 blood to crystalloid ratio) for myocardial protection.

Tissue samples were immediately placed in RNAlater^® (Ambion, Life Technologies, USA), and after 48 h at +4 °C were stored at −80 °C until RNA extraction. Total RNA was isolated with Trizol and RNA quality was assessed using the Agilent Bioanalyzer 2100 (Agilent) with no samples being excluded for poor quality. Library preparation and sequencing has been described previously [21], but briefly, ribosomal RNA was removed by performing 1–2 washings of RNA annealed to poly-T oligos beads (Invitrogen). RNAs were reverse transcribed using random hexamers (Invitrogen). Double-stranded DNA (dsDNA) synthesis was performed using Pol I and RNA-ase H. Short fragments were purified with QiaQuick PCR extraction kit (Qiagen) and resolved with EB buffer for end reparation and poly(A) addition then ligated with sequencing adaptors for cluster generation and sequencing on the Illumina HiSeq 2000 (Illumina, San Diego, CA). As samples were analyzed at different times, different read lengths were employed, initially using single-end reads and then transitioning to paired-end reads ranging from 36 to 100 base pairs.

Raw reads produced by the Illumina sequencer imaging files were filtered to remove reads containing adaptor sequences, containing >5% unknown nucleotides, or having >50% of reads with base quality scores <5. Using Tophat v2.0.5 and Bowtie2 [22‐24], reads were aligned to the Homo Sapiens reference genome (UCSC hg19). For paired-end reads, mate inner distance was set at 165 bp and mate standard deviation at 37 bp. For other parameters, we used the default settings in Tophat v2.0.5 and Bowtie2. Final read alignments having more than two mismatches or more than two gaps or more than two edit distances were discarded. The anchor length was 8 without any mismatches being tolerated. The minimum and maximum intron lengths were 70 and 500,000 respectively. The maximum insertion and deletion lengths were 3. Fragments per kilobase per million mapped reads (FPKM) levels were determined using Cufflinks [25]. Differential expression analysis was performed using Limma (v3.30.11) [26]. A linear model was constructed to test the effect of ischemia on gene expression, denoted y, while blocking on the patient from which each sample was collected, which adjusts for within and between-sample variability:where β denotes regression coefficients. Read counts were normalized while incorporating sample-level weights using voom [27], and genes were filtered for inclusion in differential expression analysis in accordance with practices established by the authors of edgeR [28]. For each gene, positive or negative log2fold change, which was interpreted as an increase or decrease, respectively, in gene expression due to ischemia, was determined. p-values were adjusted for multiple comparison testing using the Benjamini Hochberg method.

Human visceral (omental) and abdominal subcutaneous adipose tissue samples were collected from patients undergoing weight loss reduction surgery at the University of Massachusetts Medical School between 2005 and 2009 [29]. The microarray data from visceral (VAT) and subcutaneous adipose tissue (SAT) samples have been deposited in the GEO database under accession code GSE20950. Samples of epicardial adipose tissue (EAT) adjacent to the right coronary artery, and SAT from the sternum, were collected from patients without CAD who were having elective surgery for valvular heart disease. RNA was isolated using QIAGEN mini-lipid RNA extraction kits and analyzed on an Agilent 2100 Bioanalyzer. Complementary RNA (cRNA) synthesis and hybridization to Affymetrix 1.0 ST microarrays was performed by the UMASS Medical School Genomics Core Facility. For both studies, informed consent was given by the patients and the study was approved by the University of Massachusetts Medical School Institutional Review Board. Subcutaneous (inguinal) and visceral (gonadal) adipose tissue was harvested from 13 week old male C57BL/6 mice fed normal chow. RNA from paired samples (n = 6) was prepared with QIAGEN mini-lipid RNA extraction kits and analyzed on an Agilent 2100 Bioanalyzer. Complementary RNA (cRNA) synthesis and hybridization to Affymetrix Mouse Gene 1.0 ST microarrays was performed by the UMASS Medical School Genomics Core Facility. This data is publically available in the GEO database under accession code GSE28440.

Plasma samples were obtained from venous blood obtained before and after CPB in the same population of cardiac surgical patients from which the left ventricular biopsies were taken. Concentrations of ITLN1 were measured using ELISA (Millipore #EZH0MNTN1-29K) according to the manufacturer’s instructions.

Sections of EAT were fixed overnight in 10% formalin, sectioned, and stained for ITLN1 using a sheep anti-human antibody (R&D Biosystems #AF4254) at 10 μg/ml. The secondary antibody was a donkey Anti-Sheep IgG NorthernLights NL557-conjugated antibody. Sectioning and immunofixation were done by the UMASS Morphology core.

Primary human mesothelial cells were purchased from Zen-Bio. Cells were cultured according to the manufacturer’s instructions (Human Adult Mesothelial Cell Manual ZBM0025.01). Briefly, cells were thawed and plated in MSO-1 media at a density of 5000 cells/cm². After they were 85–90% confluent, they were split at a 1:6 ratio and seeded onto collagen coated plates. On day 4, cells were serum starved for 24 h with Media 199 (Gibco) containing 5% penicillin/streptomycin, normocin, and 0.5% FBS. After 24 h, the media was changed and vehicle or 5 ng/ml TGFβ1(Sigma T7039) was added. Cells were harvested for RNA extraction after 48 h of TGFβ1 treatment. RNA was isolated using TRIzol as previously described. cDNA was synthesized with the iScript cDNA systhesis kit (BioRad) with 1 μg RNA template. qRT-PCR was performed using the iTaq Universal Sybergreen supermix (BioRad) and a CFX96 Connect Real Time Detection System. Relative expression was normalized to ribosomal phosphoprotein P0 (RPLP0) using the method of Livak et al. [30]. Primer sequences were obtained from primer bank and were as follows: E-cadherin (CDH-1) (5′cgagagctacacgttcacgg 3′, 3′gggtgtcgagggaaaaatagg 5′); VMAC (5′gccctagacgaactgggtc 3′, 3′ggctgcaactgcctaatgag5′); fibronectin (5′cggtggctgtcagtcaaag3′, 3′aaacctcggcttcctccataa5′); SNAI1 (5′ tcggaagcctaactacagcga 3′, 3′agatgagcattggcagcgag 5′); SNAI2 (5′cgaactggacacacatacagtg 3′, 3′ctgaggatctctggttgtggt 5′).

Baseline continuous variable patient characteristics are presented as means and standard deviations except ITLN1 mRNA levels which are described with median and interquartile range. Categorical variables are described with total number and percentages. Differences between pre-ischemia and post-ischemia ITLN1 protein levels and ITLN1 RNA levels were determined using a paired t test. Univariate linear regression analysis was performed on pre-ischemia ITLN1 RNA and ITLN1 plasma levels against each other and clinical parameters including age, body mass index (BMI), hemoglobin A1c (HgA1c), sex, CAD, and post-operative day 1 creatine kinase MB fraction (CKMB). Similarly, univariate linear regression analysis was also conducted on post-ischemia ITLN1 RNA and ITLN1 plasma levels against the same clinical parameters in addition to aortic cross clamp time. These studies were conducted in R studio [31] using the general linear model function and results are reported as p-values and beta coefficients. p-value <0.05 was considered significant.

We used our prior CPB model in patients undergoing aortic valve replacement to obtain pre-ischemia (blood and left ventricular tissue harvested at the start of CPB) and post-ischemia samples (tissue harvested after a mean of 81.73 ± 27.89 min of ischemic cross clamp time) [21]. This model has previously been validated to mimic ischemia in human hearts [32, 33]. In a small cohort of patients, ITLN1 was the most significantly down-regulated gene in this model [21]. In the current larger study from 132 paired samples, median levels of pre-ischemia and post-ischemia left ventricular ITLN1 RNA expression were 59.96 FPKM (interquartile range 5.43–303.57) and 2.46 FPKM (interquartile range 0.84–11.00) respectively. In this population, post-ischemia RNA levels were significantly lower than pre-ischemia levels (mean difference 280.48, 95% CI 449.43–111.54, p = 0.001) (Fig. 1a). From the same group of patients, 106 paired samples had mean pre-ischemia and post-ischemia ITLN1 plasma levels of 32.23 ± 15.22 and 37.46 ± 16.84 ng/ml respectively. In this population post-ischemia protein levels were significantly higher than pre-ischemia levels (mean difference 5.24, 95% CI 3.52–6.95, p < 0.001) (Fig. 1b). In patients with CAD compared to those without, pre-ischemia left ventricular expression levels of ITLN1 were slightly higher (median 89.64 FPKM interquartile range 4.60–331.00 vs median 43.53 interquartile range 5.62–281.85, p = 0.839) while pre-ischemia plasma levels of ITLN1 were slightly lower (mean 30.94 ± 14.44 ng/ml vs mean 33.56 ± 16.03 ng/ml, p = 0.378) but these associations were not significant.

We next investigated the correlation between pre-ischemia and post-ischemia ITLN1 expression in the left ventricle and clinical parameters using univariate linear regression models. There were no significant correlations identified. We repeated these correlations with plasma protein levels. Age was positively correlated with plasma protein levels before and after ischemia (Table 2). There were no significant associations of plasma protein levels or left ventricular expression of ITLN1 with patients on or off statin medications (data not shown).

Given the proximity of our cardiac biopsy samples to epicardial adipose tissue (EAT), which has also been shown to express ITLN1 [8, 19], we further wanted to characterize the expression of ITLN1 in this compartment. Clinical studies have demonstrated that the amount of EAT correlates with visceral adipose tissue (VAT), and is increased in aging and obesity [34, 35]. Therefore we hypothesized that the transcriptome of EAT may be similar to that of VAT, including high expression of ITLN1. In order to compare the gene expression profile of VAT with EAT, we utilized our previously published work [29]. We first compared gene expression in paired SAT and VAT samples from eight non-diabetic patients who underwent weight reduction surgery. Next, we compared gene expression in paired EAT and SAT samples from nine patients without CAD having surgery for heart disease. We then evaluated the two sets of differentially regulated genes in order to indirectly compare the gene expression of EAT and VAT. In the paired VAT vs. SAT comparison, there was differential expression of 845 genes (Fold change >2.0, p < 0.05). In the paired SAT vs. EAT comparison, 188 genes were differentially expressed (FC > 2.0, p < 0.05). Of these 188 genes, 46% were commonly differentially regulated in VAT (Fig. 2). These included ITLN1 and other canonical VAT specific genes such as TCF21, SRFP2, and ALOX15 (Table 3). Remarkably, geneset enrichment analysis of the differentially expressed genes in both the SAT vs. VAT and SAT vs. EAT comparison yielded almost identical results (Table 4). The top three enriched pathways in VAT and EAT included complement, classical complement, and lectin induced complement. Therefore, even in the absence of CAD or diabetes, both VAT and EAT seem to play an important role in innate immunity.

Next, we aimed to more carefully characterize the expression of ITLN1 in EAT. We first confirmed that expression of ITLN1 is highly enriched in human VAT, and nearly absent in SAT (Fig. 3a). Similar to other studies [7, 8], we also were able to show that ITLN1 expression in human VAT is isolated to the stromal vascular fraction (SVF) (Fig. 3b). Mice, on the other hand, have equivalent low levels of ITLN1 expression in both VAT and SAT (Fig. 3c).

Since the SVF can contain multiple different cell types including epithelial cells, macrophages, fibroblasts, and precursory adipocytes [7], we sought to further characterize the location of ITLN1 in EAT. Immunohistochemistry demonstrated that ITLN1 is predominantly localized to the mesothelium, which constitutes the visceral pericardium, and not the adjacent adipocytes (Fig. 4a–d).

We then wanted to determine if ITLN1 expression requires an epithelial cell fate. Primary human mesothelial cells were treated with TGFβ1 (5 ng/ml) for 48 h in order to induce epithelial to mesenchymal transition. qRT-PCR demonstrated that ITLN1 RNA expression was dramatically reduced within 48 h (Fig. 5a). We also confirmed that expression of ITLN1 was absent in a human adipocyte cell line before and after adipogenic differentiation (SGBS D0 and D14); whereas the adipocyte specific gene Perilipin 1 (PLIN1) was increased dramatically in this cell line, but not in the mesothelial cells (Fig. 5b). Furthermore, using qRT-PCR, we confirmed that the decreased ITLN1 mRNA expression was paralleled by an early transition to a mesenchymal lineage, characterized by increased expression of the mesenchymal markers VMAC, SNAI1, and SNAI2 (Fig. 5c). Interestingly, many genes that have been shown to play a role in the epithelial to mesenchymal transition were among the complete list of RNA-seq differentially expressed genes from our pre- and post-ischemic human myocardial samples. In particular, there was upregulation of the mesenchymal markers MMP9, TCF4, BMP1, and SNAI2, and downregulation of the epithelial markers KRT19 and CDH-1 (Additional file 1: Table S1).