Peptidomic Analysis of Amniotic Fluid for Identification of Putative Bioactive Peptides in Ventricular Septal Defect Open Access

Ventricular septal defect (VSD) is one of the commonest congenital malformations of the heart, accounting for up to 40% of all cardiac anomalies [1]. The incidence of VSD is approximately 3.94 per 1000 live births [2,3], and congenital heart disease is causing society a heavy psychological and economic burden. Due to the progress of new molecular biology techniques, the past decade has witnessed spectacular development in elucidation of the molecular mechanisms of heart formation. In particular, several transcription factors (TFs) such as NKX2.5 [4], GATA4 [5], TBX5 [5] and HOMEZ [6] have been identified as being essential for heart. However, little is known about the proteomic and peptidomic studies related to the pathology of congenital heart disease (CHD). Besides, changes in genomic and transcriptomic approaches may not quantitatively correlate well with the expression of proteins [7,8] and cannot characterize post-translational modifications (PTMs) which are prominently involved in modulating many biological processes. Cardiovascular proteomics dramatically broadening our knowledge on the complex myocardial physiology and pathophysiological states with an enormous potential to rapidly advance the identification of disease mechanisms.

As we know, echocardiography is the mainstay of modern diagnosis of ventricular septal defect, but the diagnosis of congenital heart disease in fetuses by echocardiography depends highly on the examiner's skills and experiences, which may cause misdiagnosis. Reliable objective diagnostic biomarkers for VSD in early pregnancy is still needed for early detection and for predicting prognosis after birth, which may provide useful information to aid prenatal counselling and choice of delivery method. A greater understanding of the molecular basis of cardiac function will help guide the development of novel diagnostic and therapeutic strategies. A recent proteomic study of plasma protein in CHDs [9] was composed of multiple kinds of CHD. VSD is not only a common isolated cardiac malformation, but also an intrinsic component of several complex malformations. In the present study we focused on patients for whom VSD is the predominant form of congenital heart disease.

Recent technical advances in proteomics and associated methodologies will allow us to investigate VSD to an unprecedented depth. This is the first molecular biology and bioinformatics representation of peptidomic analysis on VSD. Peptidomics is an emerging branch of proteomics that investigates endogenously produced protein fragments [10]. Peptidomics can present a dynamic view of the health status of a tissue or sample, and they may provide insight into the complex and fluid interaction of activators, inhibitors, proteases, and protein substrates. In the past, intact proteins were considered as the functional units in vivo. In reality, most proteins experience proteolytic processing such as activation or degradation by enzymes [11]. Therefore, in most cases, endogenous peptides can interact directly with the cellular targets, producing a functional effect.

Amniotic fluid (AF) fills the amniotic cavity that is lined by amnion epithelial cells that are of fetal origin and hence reflect the pathological changes of the fetus [12,13]. Vesce et al. [12] found that ampicillin administration may directly decrease amniotic IL-6 and PGE2 release through management of bacterial and nonbacterial inflammatory complications of pregnancy by prostanoid and cytokine interaction. In addition, Morabito et al. [14] utilize mesenchymal stem cells from human amniotic fluid as a osteogenic model to study in-vitro cell responses to calcitonin, which provide new insight in cell therapies and pharmacological use of these molecules. The composition of AF is modified during pregnancy, and its protein profile reflects the physiological and pathological changes that affect both the fetus and the mother. In the present study, we conducted a comprehensive peptidomic analysis of human AF in healthy and VSD fetuses using tandem mass tag (TMT) labelling coupled with nano LC-MS/MS in hope to explore bioactive peptides during heart development and in the pathogenisis of VSD.

Novex NuPAGE 10% Bis-Tris Gels were obtained from Life Technologies (Rockford, USA); Coomassie Brilliant Blue-G250 and PageRuler prestained protein ladder were purchased from Thermo Fisher Scientific (Rockford, USA); HPLC-grade acetonitrile (ACN) was obtained from Tedia (Fairfield, OH, USA); HPLC-grade formic acid (FA) was purchased from Sigma/Fluka (Switzerland).

Amniotic fluid (AF) were obtained from pregnant women at 23-25 weeks of gestation who underwent amniocentesis and routine blood test at the second trimester of pregnancy in Nanjing Maternity and Child Health Care Hospital Affiliated to Nanjing Medical University from April 10^th to August 15^th, 2015. Samples were collected with written consent and ethics board approval from Nanjing Maternity and Child Health Care Hospital Affiliated to Nanjing Medical University. VSD fetuses were identified by a characteristic echo dropout in the ventricular septum in two-dimensional echocardiography (Fig. 1A), a typical systolic coloured flame crossing the septum (Fig. 1B). Fetuses with associated complex extra-cardiac anomalies (such as fetal edema and renal dysplasia) and chromosome abnormalities (such as Trisomy 21 and Trisomy 18) were excluded from the study.

The control group consisted of women with natural, singleton pregnancies who underwent termination, and whose screening ultrasound reports were normal during pregnancy. Samples were thus normal controls (n = 3) or VSD group members (n = 3). Samples were matched for gestational age and maternal age, and the detailed characteristics of patient and control groups are listed in Table 1. Samples were collected, stored on ice during transport to the laboratory (up to 1 h), centrifuged (1000 g, 10 min, 4°C) to remove cellular debris, protease inhibitor mixture (Complete Mini EDTA-free, Roche, Basel, Switzerland) was added and supernatants were stored at -80°C until needed.

Samples were centrifuged at 12,000 g at 4°C for 15 min after thawing on ice, ACN (20% (v/v)) was added to supernatants, and samples were briefly vortexed and incubated for 20 min at room temperature. Centrifugal concentrators (Amicon Ultra-15, Millipore) with a molecular weight cut-off (MWCO) of 10 kDa were washed with 0.5 ml H₂O and samples were passed through the filters according to the manufacturer's recommendations. The flow-through containing peptides was recovered and lyophilized, and the concentration of peptides/proteins in the supernatant was measured using the BCA protein assay (Pierce, Rockford, IL, USA).

Extracted peptides (35 µg) were dissolved in NuPAGE Bis-Tris LDS sample buffer, heated at 70 C for 10 min, and each sample was loaded onto a single lane. Separation was performed in NuPAGE MES running buffer and proteins were visualized using Coomassie Brilliant Blue staining as previously described [15]. The staining revealed that almost all high molecular weight proteins have been removed (Fig. 2).

Samples containing 100 µg peptides were reduced with 10 mM DTT at 60°C for 1 h, alkylated with 55 mM iodoacetamide (IAA) for 45 min at room temperature, desalted and dried in vacuo (Speed Vac, Eppendorf). AF peptides were labelled with isobaric tags based on total peptide amount at room temperature for 1 h. TMT labelling was performed according to manufacturer's instructions (TMT 6-plex Label Reagent, Thermo Scientific). TMT results were generated from analysis of isobaric tag combinations. Three normal control AF samples were labelled with reagents 126, 127 and 128, and three VSD AF samples were labelled with reagents 129, 130 and 131.

Peptides were analyzed by nanoflow LC (Easy-nLC, ThermoFisher Scientific Inc., San Jose, CA) coupled with an LTQ-Orbitrap Velos mass spectrometer (ThermoFisher Scientific Inc.). Reverse-phase separation of peptides was performed using buffer A (2% ACN, 0.5% acetic acid) and buffer B (80% ACN, 0.5% acetic acid) with a gradient. Eluted peptides were electrosprayed at a voltage of 1.8 kV into the mass spectrometer which was configured to collect high resolution (R = 60,000 at m/z 400) broadband mass spectra (m/z 375-1800) using the lock mass feature for the polydimethylcyclosiloxane ion generated in the electrospray process (m/z 445.12002). The 20 most abundant peptide molecular ions dynamically determined from the MS scan were selected for tandem MS using a relative collision-induced dissociation (CID) energy of 35%. The most intense product ion from the MS2 step was selected for higher energy collision-induced dissociation (HCD) MS3 fragmentation.

Extracted MS/MS spectra were searched against the database containing 20194 human protein sequences (released May 2015) using in-house PEAKS software (version 7.0, Bioinformatics Solutions, Waterloo, Canada). The fusion target-decoy approach was used for estimation of the false discovery rate (FDR) and was set at ≤1% (-10 log P ≥20.0) at both protein and peptide levels. Peptides were considered positively identified only if peptide peaks were present in at least two spectra per sample.

Relative quantification of the AF peptidome was performed using the TMT labelling approach in the PEAKS Q module. Feature detection was performed separately on each sample using the expectation-maximization algorithm. Features of identical peptides from different samples were aligned using high-performance retention time alignment algorithm [16]. Peptides and proteins were considered to be significantly altered between AF samples when the p-value was <0.05 and the fold change was >1.5. Identified results were included during the last step of TMT labelling quantification.

The isoelectric point (pI) of each peptide was calculated using the pI/Mw online tool (http://web.expasy.org/compute_pi/). Pathway and gene ontology (GO) analyses (http://geneontology.org) were carried to determine potential physiological functions. GO results included three categories; cellular component, biological process and molecular function. The threshold of significance was defined by the P-value and FDR. To further explore the significance of differentially expressed peptides and their precursor proteins, they were imported into the Ingenuity Pathway Analysis (IPA) Software v7.1 (Ingenuity Systems, Mountain View, CA) for pathway analysis as described previously [17]. This software can also be used for biochemical, biologic, and molecular functions. The identified proteins were mapped to associated network functions that were generated from existing literature from the Ingenuity Systems Knowledge Base.The MEROPS database (http://merops.sanger.ac.uk) was used to seek out candidate enzymes for given cleavage sites [18].

SPSS 15.0 software (SPSS Inc, Chicago, IL) were used for statistical analysis. T-test was used to compare values between control and VSD groups, and statistical significance was defined as p of less than 0.05.

A total of 63 peptides significantly differentially expressed (fold change >2.0) were listed in Table 2 and they were visualized using heat maps (Fig. 3A and 3B). Each of the six columns represents healthy control or VSD groups, and the magnitude of peptide expression is coloured according to the heat map scale. We identified and quantified 692 non-redundant peptides, 183 of which were differentially expressed in the amniotic fluid of healthy and VSD fetuses. Raw data have been uploaded to the public database: http://58.193.178.2/datacenter/.

Following analysis of the differentially expressed peptides identified in human amniotic fluid, we found that both molecular weight (MW) and isoelectric point (pI) were spread over a wide range, but most peptides were distributed below 1000 Da in molecular weight and 5.0-7.0 in isoelectric point (Fig. 4A and 4B). The relative distribution of pI versus MW was also investigated (Fig. 4C), and the MW of all identified peptides was below 3 kDa, confirming the isolated peptides were of adequate purity.

We combined peptidome analysis of human amniotic fluid by LC-MS/MS with bioinformatics analysis of cleavage sites at the N- and C-termini. Four cleavage sites were investigated for each identified up-regulated and down-regulated peptide (Fig. 4D and 4E), respectively. In up-regulated group the four dominant amino acids of the four cleavage sites (C-terminal amino acid of the preceding peptide, N-terminal amino acid of the identified peptide, C-terminal amino acid of the identified peptide, N-terminal amino acid of the subsequent peptide) were glycine (G), glycine (G), leucine (L), glycine (G), while in down-regulated group the four most amino acids were glycine (G), glycine (G), proline (P) ‚ proline (P). There were more than one peptides derived from the same precursor protein. The precusor proteins were listed in Fig. 4F with titin having largest number of related peptides.

Pathway and gene ontology (GO) analyses were carried out to determine potential roles for the identified peptides and their precursor proteins. The most common biological process, cellular component, and molecular function categories were listed in Fig. 5A-C, respectively. As amniotic fluid is composed of metabolite of various organs, and results also involved in multiple organs. Biological processes particularly involved in caidiogenesis were cell morphogenesis involved in differentiation, cardiovascular system development and circulatory system development. Precursor proteins significantly differentially expressed in the AF of the VSD group were mapped against canonical pathways, and those with the highest statistical significance were plotted as a bar chart (Fig. 5D).

Differentially expressed proteins were investigated using IPA software (Qiagen, Redwood City, CA, USA) that performs a comprehensive analysis of precursor proteins and related biological processes based on the proteomic data.

Primary data analysis indicated that some of the proteins that were differentially expressed in AF were related to cardiovascular system development and function. Possible downstream effects in the morphogenesis of cardiovascular system (p-value 4.00E-04), cardiogenesis (p-value 8.06E-03), hypoplasia of myocardium (p-value 4.24E-03), and abnormal morphology of cardiovascular system (p-value, 7.60E-03) were predicted for the VSD group. The differentially expressed precursor proteins related to cardiogenesis were shown in Fig. 6.

Analysis of potential molecular interaction networks of the 183 differentially expressed proteins identified 11 high scoring networks. Both 'Cardiovascular System Development and Function, Skeletal and Muscular System Development and Function, Cancer' (Fig. 7A) and 'Hereditary Disorder, Neurological Disease, Connective Tissue Disorders' (Fig. 7B) networks were significantly perturbed.

All identified peptides in cardiovascular field located within the domains of their respective precursor proteins are listed in Table 3, including the amino acid sequences, amino acids with PTM, peptide location and dormain/region description. In addition, we also the listed upstream entrez gene name, location, family and gene ID for human, mouse and rat in Table 4, which also indicated that most peptides were derived from evolutionarily conserved proteins or protein domains. Titin (TTN) was represented by the highest number of peptides, many of which are part of Ig-like or Fibronectin type-III domains. Predicted changes in upstream regulation revealed that most changes associated with TTN were related to Pkc(s) and NF-κB pathways.

We chose “VAAVNVKG” who has the highest fold change and is also in functional domain of Titin in Table 3 as our target. A search of the peptidase database (MEROPS) (30) identified a number of candidate peptidases that would likely cleave the starting and ending bond, also refering to Arg-Val and Gly-Met sites of VAAVNVKG (Table 5). This list included peptidases and proteases from different classes (serine, cysteine, and metallo), as well as soluble and membrane-bound forms of the enzyme. According to this list, the cathepsin K and matrix metallopeptidase-2 can cleave VAAVNVKG at both cleavage sites, suggesting that a single enzyme may regulate VAAVNVKG.

Technological advances facilitating the rapid analysis of whole genomes has yielded large amounts of gene sequence and expression data that can be utilized by proteomics and peptidomics approaches. Peptidomics is an emerging branch of proteomics that targets endogenously produced protein fragments [10]. Rather than examining a sample for intact proteins, peptidomics identifies endogenous protein fragments.

Recent technological advances in proteomics and peptidomics have been actively utilized to investigate AF proteins in order to better understand this complex biological fluid and to discover disease-specific biomarkers. Putative markers including intra-amniotic infection, premature rupture of amnion, and Down syndrome have been recently discovered or reanalysed [19,20,21]. We studied AF from normal healthy and VSD fetuses because AF is fetal in origin and so reflects the genotypic constitution and physiological status of the developing fetus. Although the collection of the biofluid through amniocentesis may be invasive, aliquots of it may be taken from routine analysis and the detailed characterization of AF dynamics in relation to pathological occurrences may lead to the discovery of new biomarkers for better prevention approaches and early detection strategies [22]. However, AF is dynamic in nature and the composition is known to differ depending on gestational age [23]. Our data was therefore restricted to fetuses of a gestational age of 23-25 weeks, the period during which most amniocenteses are performed and hence samples are most easily obtained.

To identify naturally occurring peptides in AF in an efficient manner, a novel and comprehensive approach was developed that involved centrifugal ultrafiltration to separate peptides based on MW. We found a 10 kDa MWCO filter capable of removing most abundant proteins from human AF without compromising peptide recovery. We used ACN to resolve peptides that become adsorbed on the surface of high MW proteins, and MS/MS assays identified nearly 700 non-redundant peptides across both groups, of which 183 were differentially expressed in the VSD group.

Proteomics is shaped by proteolysis through protein degradation and stable cleavage products were generated by restricted proteolysis [24]. It has been reported that peptidomic analyses yielded disease biomarkers representing cleavage fragments from bodily fluids [25]. As protease's specificity can be reflected by the cleavage events catalyzed by protease, different cleavage sites may represent different protease activities. We analyzed the cleavage sites of all the differentially expressed peptides (Fig. 4D and 4E) . We can see the cleavage sites among the 20 amino acids were distinctly different, which is worth our further exploration on changes in protease function. By analyzing cleavage site of “VAAVNVKG” we found protease cathepsin K and matrix metallopeptidase-2 can cleave VAAVNVKG at both cleavage sites, and previous studies have reported expression of metallopeptidase-2 during early cardiac and neural crest morphogenesis [26], which needs our further exploration. Because the MEROPS database cannot account for uncharacterized peptidases or proteases, we intend to either chemically inhibit peptidase activity in vitro experiments or genetically knock out the given peptidase in model animal as previously reported [27] to assess the global changes in the peptidome.

There exists much literature discussing the association between these proteins and cardiac function. For example, ataxia-telangiectasia mutated (ATM) protein and Prkdc collaborate to maintain genomic stability during gastrulation and early organogenesis [28]. COL1A1, COL2A1, and COL3A1 have been associated with Ehlers-Danlos syndrome, and type IV variant of this disease with symptomatic coronary artery dissection revealed impaired secretion and abnormally slow electrophoretic migration of type III collagen (COL3A) [29]. Overexpression of pro-α3(V) (COL5A) mRNA was observed in fetal heart and lung, and moderately high levels were detected in certain structures of the adult human heart [30]. COL18A1 inhibits the migration of endothelial cells in vitro but has no effect on the proliferation of these cells [31]. PKD1 knock-out mice embryos developed hydrops, cardiac conotruncal defects and renal cystogenesis [32]. Mice lacking ErbB4 exhibit defects in cranial neural crest cell migration and die because of the resultant defective heart development [33]. WT1 null mice suffer oedema and die at midgestation from malformed hearts [34]. MINT knock-out mice embryos had defects in cardiac septum and muscle tissue, consistent with Notch signalling-associated regulation of heart development [35]. McMahon et al. [36] proposed that COL4A1 and COL4A2 could be candidate genes for congenital heart disease (CHD) with a deletion in 13q within the 6Mb critical region for cardiac development. All this perspective has given us a reason to consider them as bioactive peptides.

Titin (TTN), also known as connectin, is a very large (3.4 MDa) protein and plays an important role movement generated by skeletal and cardiac muscle contraction [37]. Most previous studies on TTN have focussed on cardiomyopathy, and over 70 TTN mutations have been found to cause familial dilated cardiomyopathy [38,39,40,41,42,43]. During heart development, changes in Titin expression are likely to impact functional transitions and diastolic filling behaviour by regulating sarcomere compliance and sarcomere homeostasis [44,45]. Titin participates in the formation of cardiac muscle fibers and muscle tissue morphogenesis [46]. In the present study, we found 15 differentially expressed endogenous peptides derived from TTN mapped to functional domains of this protein. Whether these peptides were derived from TTN degradation or from newly synthesized material, their differential abundance in VSD AF may reflect changes in the expression and function of TTN in VSD fetuses, and they could therefore be putative bioactive peptides for VSD.

Our results indicated that changes in TTN were mainly related to Pkc(s) and NF-κB signalling pathways, suggesting it may be through these pathways that the differentially expressed proteins mediate malformation of the ventricular septum. Ahmad [47] observed mRNA and protein levels of titin in endomyocardial biopsy of DCM patients, and TNF-α may modulate the expression of these proteins via NF-κB pathway. Hidalgo [48] reported that PKC phosphorylation of titin is a novel and conserved pathway that links myocardial signaling and myocardial stiffness. Network analysis indicated that most of the differentially expressed proteins could also participate in other signalling pathways related to cardiovascular development such as the MAPK, ERK and GRB2 pathways. Biological interaction networks serve as useful tools for useful for postulation of new hypotheses, and can provide important insights into diseases such as VSD.

In summary, our study provides a profile of dysregulated peptidomic analysis in human amniotic fluid with VSD for the first time. It suggests a large amount of peptides participate in heart development and provides a background resource for future functional investigations of peptides related to VSD. Larger population study is required to improve statistical meaning. And further studies are necessary to reveal possible biological functions and molecular mechanisms exerted by these peptides in the process of VSD formation.

ACN (Acetonitrile); CHD (Congenital heart disease); DCM (Dilated cardial myopathy); GO (Gene Ontology project); IPA (Ingenuity Pathway Analysis); KDa (Kilo Dalton); LC-MS/MS (Liquid chromatogram tandem mass spectrometry); PTM (Post translational modification); VSD (Ventricular septal defect); WT1 (Wilms tumor protein); TMT (Tendam mass tag).

This work was funded by a grant from the National Natural Science Foundation of China (Grant No. 81370278 and 81570209). Dr. Fu-Qiang Wang is acknowledged for assistance with mass spectrometry.

Authors have nothing to disclose.

X. Li and L.-J. Wu contributed equally to this article.