|Home | About | Journals | Submit | Contact Us | Français|
Differentiation of prevalvular mesenchyme into valve fibroblasts is an integral step towards the development of functionally mature cardiac valves. Although clinically relevant, little is known regarding the molecular and cellular mechanisms by which this process proceeds. Genes that are regulated in a spatio-temporal pattern during valve remodeling are candidates for affecting this differentiation process. Based on its expression pattern, we have focused our studies on the role of the matricellular gene, periostin in regulating the differentiation of cushion mesenchymal cells into valve fibroblasts. Herein, we demonstrate that periostin expression is coincident with, and regulates type I collagen protein production, a major component of mature valve tissue. Adenoviral mediated knock-down of periostin in atrioventricular mesenchyme resulted in a decrease in collagen I protein expression and aberrant induction of myocyte markers indicating an alteration in AV mesenchyme differentiation. In vitro analyses using a novel “cardiotube” assay further demonstrated that expression of periostin regulates lineage commitment of valve precursor cells. In these cells, expression of periostin and collagen I are regulated, in part, by TGFb-3. We further demonstrate that TGFb3, through a periostin/collagen pathway, enhances the viscoelastic properties of AV cushion tissue surface tension and plays a crucial role in regulating valve remodeling. Thus, data presented here demonstrates that periostin, a TGFb-3 responsive gene, functions as a crucial mediator of chick AV valve maturation via promoting mesenchymal to fibroblast differentiation while blocking differentiation of alternative cell types (myocytes).
The development of properly functioning cardiac valves is a complex, dynamic process involving a myriad of factors (Person et al., 2005). Valvulogenesis commences with an endothelial to mesenchymal transition (EMT) resulting in undifferentiated mesenchyme between the myocardial and endocardial layers of the heart, known as endocardial cushions (ECs). As blood circulates through the heart, these EC swellings function as primitive valves prohibiting retrograde blood flow between the primitive atria and ventricle and the ventricle and aortic sac. Over the past thirty years much effort has been put forth to determine the inductive signals essential for proper EMT progression. To date, more than 100 genes have been linked to this important morphogenetic EMT event (Mjaatvedt and Markwald, 1989; Yamamura et al., 1997; Lakkis and Epstein, 1998; Mjaatvedt et al., 1998; Savagner, 2001; Camenisch et al., 2002; Schroeder et al., 2003; Holifield et al., 2004; Ma et al., 2005; Person et al., 2005; Stevens et al., 2008). The initial EMT event leading to the formation of these primitive valves is just the first phase in valvuloseptal morphogenesis. The molecular and cellular events that occur after this EMT event are quite complex and poorly understood. This “post-EMT” process, which will yield a mature AV valve leaflet and septal apparatus, involves the migration and proliferation of mesenchymal cells to distend the forming cushions into the lumen (Akiyama et al., 2004; Shelton and Yutzey, 2007). This is followed by a gradual attenuation (thinning) and continued elongation of prevalvular tissue. Concomitant with this early remodeling phase is differentiation of the cushion mesenchyme into fibroblastic interstitial cells (“valve interstitial cells”) (Weber, 1989; Schoen, 1999; Goldsmith et al., 2004; Aikawa et al., 2006; Liu et al., 2007). These cells, then, have undergone a fundamental phenotypic change. The matrix surrounding these newly differentiated fibroblasts matures into a highly organized fibrous connective tissue (Peacock et al., 2008), making the valve structurally more rigid and competent for handling the increasing hemodynamic load of the beating heart.
Understanding the regulatory pathways that specify mesenchymal cell differentiation into valve interstitial fibroblasts (vs. other mesodermal “options” or lineages) is currently in its infancy (Lincoln et al., 2006; Firulli and Conway, 2008; Levay et al., 2008). Of significance, recent data demonstrate that cushion mesenchymal cells are multipotential mesodermal cells that normally differentiate into fibroblasts but also retain the ability to progress down alternative cell lineages (Galvin et al., 2000; Garg et al., 2005). Abnormal conditions or signals within the heart (genetic mutations, alterations in hemodynamic loading, teratogens, carcinogens, etc.) may shift the balance of these mesenchymal cells to become non-fibroblasts, resulting in altered valve formation, valve dysfunction and secondary cardiac defects. As demonstrated by mutations in either the fibrillin-1 (Dietz et al., 1991) or notch1 genes (Garg et al., 2005), it is hypothesized that even small defects in the post-EMT cushion differentiation program, initially not detected at birth can, over time, result in valvular pathologies realized in childhood or adult life.
Thus, we propose that a more thorough understanding of post-EMT processes will provide unique insight into mechanisms driving valve differentiation, and ultimately function. Based on its reported expression profile in mice and the phenotype resulting from its targeted deletion, the protein encoded by the periostin gene has emerged as an important regulator of post-EMT valve development (Snider et al., 2007; Norris et al., 2008). Periostin is a secreted protein capable of interacting with various extracellular matrix components in addition to cell surface integrins and, as such, has been defined as a “matricellular” protein. Amino acid similarity indicates that periostin is highly conserved being evolutionarily related to the ancestral Drosophila fasciclin-1 gene: (Takeshita et al., 1993; Horiuchi et al., 1999; Litvin et al., 2005). The mammalian fasciclin gene family comprises 4 members: periostin, βIG-H3, stabilin-1, and stabilin-2. These genes have been demonstrated as playing important roles in cellular processes such as: adhesion, migration, cell-to-cell interactions and differentiation (Elkins et al., 1990; Hortsch and Goodman, 1990; Horiuchi et al., 1999; Ferguson et al., 2003a; Ferguson et al., 2003b). Periostin is robustly expressed in the post-EMT cushion mesenchyme as it undergoes differentiation, remodeling and maturation (Kruzynska-Frejtag et al., 2001; Snider et al., 2007; Norris et al., 2008). This expression is maintained throughout embryonic AV valvulogenesis and into neonatal and adult life where it is intensely expressed in the fibrosa of mature AV leaflets, their tendinous cords and the annulus fibrosa. The function of periostin in cellular post-EMT cushion processes, such as proliferation, apoptosis, migration and fibrous tissue formation/differentiation is currently unknown. However, the phenotype of adult periostin null mice indicates a role in the differentiation of post-EMT cushion mesenchyme into fibroblasts (Snider et al., 2007; Norris et al., 2008).
Here, we demonstrate that periostin expression precedes and regulates type I collagen accumulation. Adenoviral mediated knock-down of periostin resulted in a decrease of collagen I protein and an increase of the myocardial markers: sarcomeric myosin, α-MHC, and desmin. Over expression of periostin increased collagen I and decreased the levels of the myocardial proteins indicating the potential for periostin to affect the fate of AV mesenchymal cells by promoting fibroblast differentiation in addition to blocking myocardial differentiation. Furthermore, we demonstrate that periostin, and collagen type I are regulated, in part, by TGFβ-3. As a result, the viscoelastic properties of AV cushion tissue are affected in a stage dependent manner. Taken together the results of these experiments indicate, for the first time, that periostin is a potent, and essential regulator of chick AV valve maturation.
Descriptive analyses of periostin mRNA and protein expression in the chick have previously been reported by us(Norris et al., 2004; Kern et al., 2005; Norris et al., 2005). However, a detailed quantitative analysis of periostin and collagen I expression in the developing AV valves has yet to be determined. Thus, quantitative RT-PCR was performed on isolated AV valves at HH17, HH22, HH25, HH38, HH44 of chick development to determine if the transcript levels of periostin and collagen I were temporally regulated. Figure 1 demonstrates that the transcript levels of both collagen I and periostin dramatically increased (80 and 70 fold, respectively) during AV valve development when compared to the HH17 levels. To determine if this increase in message resulted in a concomitant increase in levels of periostin and collagen protein expression, Western blot analysis was carried out on developing chick AV valve tissues (Figure 1C). At HH25 of development, low levels of periostin were observed in the AV cushion (arrow) with undetectable amounts of collagen I at this time point. To rule out antibody sensitivity issues, Western blots were carried out on titrating amounts of purified collagen and periostin protein which determined that the relative sensitivity of each of the antibodies (at the concentrations used) were similar, and thus comparable (data not shown). By HH38, robust expression of both periostin and collagen I were detected in the AV valves with expression increasing by HH44. At HH38, multiple putative isoforms of periostin are easily detected (arrow heads). These alternative forms of periostin significantly increase from HH38 to HH44 with the presence of as many as 4 periostin variants. Additionally, a faint ~40–45 kDa band is detected at HH44 (asterisk). Genomic sequencing and analyses of intron/exon borders has suggested that this is probably not the result of splicing mechanisms, but may be a proteolytically cleaved fragment of periostin. Low level expression of both periostin and collagen I are evident in the myocardial fractions (double asterisks). This is most likely due to these tissues containing an increasing amount of cardiac fibroblasts, which have been shown by us and others to express appreciable amounts of periostin and collagen I(Norris et al., 2007a; Oka et al., 2007; Snider et al., 2007; Shimazaki et al., 2008; Lorts et al., 2009). Actin was used as a protein loading control.
Indirect Immunofluorescence was used to determine the localizations of these proteins during valve development (Figure 2). In HH22 hearts periostin was weakly detected in invading mesenchymal cells in the AV cushions. At this time point, collagen I expression was not detected. At HH25 and HH27, periostin accumulates in the proximal component of the AV cushions and is highly localized at the myocardial-cushion interface. At HH25, collagen I can be detected in a pattern very similar to that of periostin. By HH35, periostin and collagen I are co-localized throughout the AV leaflets. However, a gradient of expression is still evident with the ventricular aspect of the leaflet showing more intense staining for both periostin and collagen I. By HH38 of chick AV valve development, expression of periostin and collagen I is nearly identical with the majority of the leaflet staining positive for both proteins.
Hanging drop aggregates were generated using HH25 cushion explants as previously described. The time point (HH25) was carefully evaluated and utilized as the starting point for our assays for these primary reasons: (i) HH25 marks the beginning of the post-EMT process, (II) after HH25 the AV cushions are fused and cellular mixing occurs with significant contributions from other structures such as the dorsal mesenchymal protrusion, (III) evaluating younger cultures would be more applicable to studying the EMT process, and (IV) later cultures express more abundant levels of periostin, which would make the effect of blocking periostin expression muted and difficult to interpret. Cushion mesenchymal aggregates were stimulated with either periostin expressing (OX), antisense (αs), or green fluorescent protein (GFP) adenoviruses. Following 7 days in culture, Western analyses were performed on these tissue explants. As demonstrated in Figure 3A, the periostin expressing adenovirus (detected by anti-HA tag) and antisense virus both work to increase or decrease periostin expression, respectively. These adenoviruses have been characterized previously as being effective in generating more periostin (OX) or blocking periostin translation (αs) (Butcher et al., 2006; Norris et al., 2007b). A significant decrease in collagen I protein was evident when periostin expression was abrogated.
We next sought to determine if the effects described above could be recapitulated by using a 3-D model system that more closely mimics post EMT valve development. This dynamic 3-D tube model of AV leaflet formation has been described previously(Goodwin et al., 2005). When AV cushions are placed on the luminal side of a collagen tube containing spontaneously contracting cardiomyocytes (termed a “cardiotube”), the cushions progressively elongate into attenuated, leaflet-like structures over a 7-day period(Evans et al., 2003; Goodwin et al., 2005). For this study, HH25 AV cushion tissues were infected with either the OX, LacZ, or αs adenoviral constructs prior to being inserted into the cardiotubes. After seven days of culture, AV explants were analyzed for the presence and localization of collagen I (Figure 3B). Robust tracts of collagen I staining were observed in the OX treated AV explants (green, Figure 3B). The collagen I staining in these cultures appeared to be organized in a parallel (lamellar-like) fashion along the proximal-distal axis of the leaflet. LacZ treated (control) explants also had bright bands of collagen I staining, albeit the staining was confined to the periphery of the growing cushion. In contrast, blocking periostin expression in these cultures (αs infected) resulted in a significant decrease in collagen I immunointensity.
Due to the decrease in collagen I (a key marker of fibroblast differentiation) associated with inhibiting periostin expression in AV cushion cultures, we sought to determine if mesenchyme differentiation into fibroblasts had been altered. Western blot analyses of chick AV cushion aggregates demonstrated that by blocking periostin expression, there is an increase in the myocardial markers: e.g. α-myosin heavy chain and desmin (Figure 4A). Densitometric assessment demonstrated that when periostin expression was inhibited by viral antisense vectors there was a significant increase (~8-fold) in the number of cells expressing the myocardial marker, MF20 (Figure 4B). It should be noted that the control cultures often contained a small number of MF20 positive cells (due to incomplete removal of myocardium from the cushion explants). However, this was to our advantage as we noticed a significant decrease in the number of these contaminating MF20 positive cells when hanging drop cultures were treated with the periostin expressing adenovirus. The “cardiotube” model was subsequently utilized to examine whether these changes in fibroblast differentiation also occurred in the 3-D “valve” cardiotube model. As seen in Figure 4C, no detectable MF20 staining was seen in either the periostin expressing (OX) or LacZ cultures. However, the majority of explanted cushion cells expressed MF20 when periostin expression was inhibited (αs).
TGFβ family members have been widely studied for their ability to regulate extracellular matrix genes during development and disease. In addition, we and others have found that periostin expression is stimulated by members of the TGFβ family(Horiuchi et al., 1999; Goetsch et al., 2003; Chen et al., 2006; Inai et al., 2008). Whereas these experiments have focused primarily on the role of BMP proteins and/or TGFβ1, we aimed to evaluate the ability for TGFβ3, a signaling protein expressed in the AV cushions, to function as an upstream regulator of periostin and collagen expression and assay whether its stimulation could indirectly affect the material properties of AV cushion tissue. Using the same hanging drop culture system as described above, AV cushion explants were cultured either with or without TGFβ3 (10ng/ml). Treatment with TGFβ3 resulted in the up-regulation of both periostin and collagen I as compared to cultures lacking TGFβ3 stimulation (fig 5A). It is interesting to note that TGFβ3 stimulation resulted in an increase in both 90kDa and ~42 kDa variants of periostin. The 90kDa form is a recognized variant of periostin whereas the 42 kDa form may be due to proteolytic cleavage. The precise reason for this lower molecular weight periostin immunoreactive band is currently unknown. TGFβ3 treatment also affected the material properties of the developing valve mesenchyme. Surface tension measurements of the AV cushion tissue were found to be significantly increased in cultures treated with TGFβ3 regardless of the developmental stage of the cushion selected for testing (fig 5B).
The development of mature cardiac valves involves two main processes. These two processes are grouped broadly into EMT and post-EMT events(Lincoln et al., 2004; Hinton et al., 2006; Lincoln et al., 2006; Hinton et al., 2008). The initial process of EMT depends on the transformation of endothelial cells into mesenchyme(Person et al., 2005). The genetic, molecular and cellular cues integral to this process have been extensively studied in both the chick and mouse, and as such, have shed significant light on how these initial stages of valvulogenesis proceed. In regards to the mouse studies, much of the work has revolved around studying various targeted gene knock-outs and their resultant phenotypes. In many of these mouse mutants, the formation of endocardial derived mesenchymal cushions was abnormal resulting in embryonic lethality. Although informative, the lethality of these mice precluded the ability to investigate later events in valvulogenesis. In the chick system, it is experimentally possible to side-step issues of embryonic lethality or genetic redundancies using viral vectors to deliver genes at specific time-points and specific regions of the heart. Also, it provides an opportunity to directly test specific mechanisms that might explain a particular phenotype observed in mice knock out studies. For our purposes, the chick model also afforded the opportunity to separate EMT from post-EMT processes and to focus on the function of periostin in regulating cushion differentiation and remodeling into valve leaflets.
The periostin gene encodes a secreted, extracellular matrix protein that has been shown to have affects on cell migration and differentiation in various cell types and model systems(Gillan et al., 2002; Yan and Shao, 2006; Inai et al., 2008). Periostin is expressed by fibroblasts in a variety of connective tissues, directly binds to collagen type I and promotes fibril cross-linking (Goetsch et al., 2003; Katsuragi et al., 2004; Suzuki et al., 2004; Butcher et al., 2006; Kii et al., 2006; Takayama et al., 2006; Norris et al., 2007b; Oka et al., 2007; Shimazaki et al., 2008). We and others have recently reported that periostin knock-out mice may have differentiation defects in post-EMT cushion mesenchyme during post-EMT valvulogenesis(Snider et al., 2007; Norris et al., 2008; Vincentz et al., 2008). This was based on observing heterogeneous foci of undifferentiated mesenchyme and myocardial tissue within the mural leaflets as well as muscularized and truncated tendious cords (chordae tendineae). However, whether the ectopic cardiomyocytes seen within the leaflets and cords represented altered differentiation or migration of muscular tissue into the cushion primordia of the valves has not been established. A connection with collagen fibrillogenesis was also demonstrated in the null mice due to collagen cross-linking defects and smaller more uniform collagen fibril size when compared to the wild-type mice. (Norris et al., 2007b). Thus, in the present study, we sought to establish if differentiation of cushion tissue could be modified through manipulation of periostin protein expression. Using complementary in vitro assays: hanging drop assays and cardiotube cultures, we were able to show that differentiation of cushion mesenchyme is altered when periostin expression is inhibited. In control cells or cells over-expressing periostin, cushion mesenchyme differentiated into fibroblasts and collagen I expression was increased. Conversely, by specifically blocking periostin in both of these culture assays, collagen I levels were significantly reduced whereas the myocardial markers α-MHC and desmin were up-regulated. These findings suggest that the muscularized and poorly differentiated AV valves seen in periostin null mice are the direct consequence of the absence of a blocking or inhibitory effect of periostin on the differentiation of cushion cells into non-fibroblastic lineages.
It is interesting to note that quantitative real-time PCR of these cultures (data not shown) have failed to reveal a significant change in collagen I mRNA levels suggesting that periostin regulates collagen post-transcriptionally. Previous work on the periostin null mouse has demonstrated that periostin specifically interacts with collagen I protein and is required for collagen cross-linking and fibril maturation(Norris et al., 2007b). Thus, over-expression of periostin in the cardiotube cultures would appear to promote a higher order collagen fibril formation and thereby increase its stability. Conversely, by blocking periostin expression in the cardiotube cultures, collagen protein expression is seen as sparse and diffuse indicating poor organization of collagen (or degradation). Based on the overlapping expression domains and their previously identified interactions with each other, we hypothesized that periostin and collagen I are coordinately regulated both spatially and temporally during post-EMT AV cushion valvulogenesis. Basically, except at the earliest post-EMT stages, “as periostin goes so goes collagen I”.
The expression of both periostin and collagen is most abundant at the interface (boundary) between AV cushion mesenchyme and AV junctional myocardium. Remodeling of the AV junctional myocardium is essential for breaking the electrical connectivity between the atrial and ventricular wall and for separation (delamination) of the mural AV cushions from their myocardial template to form a free valve leaflet(Lamers et al., 1995; Oosthoek et al., 1998a; Oosthoek et al., 1998b; de Lange et al., 2004). The mechanisms by which the cushion/valves delaminate from the myocardium are unknown. Two pieces of data presented here implicate periostin as an important contributor to this remodeling process: (i) periostin protein accumulates at the interface in situ and (ii) when myocytes were inadvertently carried into hanging drop cushion tissue cultures, over time the myocytes disappeared when periostin was over-expressed. Although, we cannot yet discern the mechanism (either differentiation and/or apoptosis) for this reduction in myocyte numbers, it does appear that periostin expression has an affect on myocyte phenotype.
Growth factors coordinately regulating ECM production (including collagen and periostin) in the AV cushions haven’t been thoroughly investigated. However, based on their well-defined role in fibrogenesis (accumulation of excessive matrix) and their expression during AV cushion morphogenesis, members of the TGFβ superfamily are excellent candidates for mediating ECM synthesis during AV cushion development. We sought to define the putative role that one of these TGFβ signaling molecules, TGFβ3, plays in matrix regulation, and the effect that this regulation may have on the biomechanical properties of AV cushion tissue. TGFβ3 is a 25-kDa homodimeric peptide growth factor with diverse effects on cellular differentiation and proliferation including the promotion of cellular phenotypic changes, the control of migration, cellular adhesiveness and invasiveness, and the regulation of ECM deposition (Sporn et al., 1986; Cui et al., 1994; Watts et al., 1994). It is well known that TGFβ3 is a growth factor that induces the accumulation and deposition of ECM(Huojia et al., 2005; Miyanishi et al., 2006a; Miyanishi et al., 2006b). At the onset of atrioventricular cushion EMT, TGFβ3 is expressed in transforming endothelium, migrating mesenchymal cells and AV junctional myocardium(Millan et al., 1991; Dickson et al., 1993; Nakajima et al., 2000a; Nakajima et al., 2000b). It has also been shown that receptors for TGFβ3 are expressed in cushion tissue during development (Brown et al., 1996; Boyer et al., 1999; Brown et al., 1999; Nakajima et al., 2000a; Nakajima et al., 2000b; Boyer and Runyan, 2001; Camenisch et al., 2002). Therefore, it was not surprising that TGFβ3 dramatically changed the material properties of AV cushions by increasing the tissue cohesivity (surface tension) over developmental time points. This change in material properties and cohesivity, induced by TGFβ3, results in the stimulation of a periostin pathway, which ultimately promotes collagen fibrillogenesis. This lays a putative morphogenetic regulatory foundation for how AV cushion mesenchyme can remodeled into attenuated, fibrous valve leaflets.
Based on these findings, we propose that: (i) chick AV cushion mesenchyme is multipotential; being able to differentiate into either a myocyte or fibroblast lineage depending on the molecular signal, (ii) periostin is an important regulatory molecule promoting the differentiation of chick AV cushion mesenchyme into valvular fibroblasts, (iii) periostin plays a role in regulating mesenchyme-myocardial interactions and remodeling processes, and (iv) coordinate regulation of periostin and collagen I by a TGFβ3 pathway is important for enhancing the material properties of AV cushion tissue. A summary of these findings is presented in Figure 6.
Total RNA of explants was extracted by using RNeasy Mini Kit (QIAGEN) and the RNA integrity and concentration were measured on Agilent 2100 bioanalyzer using RNA 6000 nano reagents (Agilent Technologies). The time points of each of the tissues evaluated were carefully chosen based on the different morphogenetic processes ongoing during AV valvulogenesis. For example: HH17-initiation of EMT, HH22-active EMT, HH25-transition from EMT to post-EMT, HH38-post-EMT remodeling, HH44-post-EMT maturation. RNA was aliquoted to 500ng for reverse transcription. RNA was converted to cDNA by using Iscript cDNA Synthesis Kit (Bio-Rad). Real-time PCR was then performed on a Step One Plus real-time PCR system using SYBR Green PCR MasterMix (Applied Biosystems). Real-time PCR primers were designed by using Beacon Designer 6 software: Chicken Arbp--control (Forward: 5'-CCTGTGATGTGACTGTGC-3',Reverse:5'-ACTTTGTCTCCGGTCTTAATC-3'), Periostin (Forward: 5'-GGATGGTATGAGAGGATGTC-3', Reverse: 5'-GCAAAGAAAGTGAATGAACC-3'), Chicken collagen I alpha I: (Forward: 5'-TGGTGTTGATAGCAGCGACT-3', Reverse:5'-GTGTCCTCGCAGATCACCTC-3'). The relative expression of target genes was calculated as previously described Pfaffl (Pfaffl, 2001).
White leghorn chicken embryos (Columbia Farms, Columbia, SC) are staged according to Hamburger and Hamilton (Hamburger and Hamilton, 1992) and their hearts excised at the appropriate time points (HH25, HH38, HH44). AV valve tissues and myocardium were isolated from HH25, HH38, and HH44 chick hearts. For the later stages of chick development (HH38 and HH44) the mural and septal mitral leaflets of the left AV valves were isolated. Tissues from each time point were pooled and lysed in a small volume of 1X RIPA buffer (50 mM Tris pH7.4, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 0.1%SDS, 0.5% sodium deoxycholate) using sonication. Quantitation of the proteins was performed using standard Commassie Protein Assay (Pierce) and 10 µg of total protein was loaded onto a 4–15% SDS-PAGE. Gels were blotted onto nitrocellulose and probed with either a previously reported periostin antibody(Kern et al., 2005), collagen I (Abcam-24133), or actin for loading control (Chemicon-1501) at concentrations of 1:5000, 1:1000, 1:10,000, respectively. Goat α-rabbit or goat α-mouse secondary antibodies (1:10,000 dilution) were used and detection was performed using Visualizer (Millipore).
Production of tubular scaffold cultures. The production of the 3-D collagen type I tube scaffold has been described in detail previously (Evans et al., 2003; Yost et al., 2004). Briefly, a 25 mg/ml solution of bovine type I collagen is extruded with a device that contains two counter-rotating cones. The liquid collagen is fed between the two cones and forced through a circular annulus in the presence of an NH3-air (50–50 vol/vol) chamber. This process results in a tube of aligned collagen (Evans et al., 2003; Yost et al., 2004). The tubes produced for these experiments have a length of 3cm with a lumen diameter of 4mm and an exterior diameter of 5mm, and a wall thickness of 0.5mm. The collagen fibers within the tubes have a defined fiber angle of 18° relative to the central axis of the tube and have an average pore size of 10 µm.
The isolation of embryonic day 15 (E15) rat CMs has been described previously (Evans et al., 2003). Briefly, timed pregnant rats (Harlan Sprague Dawley) are sacrificed at 15 days post coitus (E15), and embryos are removed from placental membranes. Hearts are dissected from the E15 embryos and atria removed from the ventricles under a dissecting microscope (Olympus SZ60). Isolated ventricles are first minced and then subjected to a series of collagenase digestions until a single-cell suspension is achieved. Cells are quantified with a hemocytometer and seeded on to the tube scaffold at a density of 200-cells/105µm2 (or 1.5 × 106 cells per tube). Cells are then added into the lumen and exterior of the tube with a pipette. The cardiotubes are seeded with these fetal myocytes to create myocardial contractions in a 3-D environment that more closely simulates the geometry of the pulsatile heart tube in vivo. For these experiments, static cultures (in the absence of fluid flow) were utilized to determine the effects of cyclical contractile forces on the developing cushions. For these experiments spontaneously contracting cardiotubes are cultured for 7 days, in a bioreactor as previously described(Evans et al., 2003), prior to the addition of the HH25 AV cushion explants.
AV canals are microdissected and placed in Tyrodes buffer and dissected to expose and isolate the endocardial cushions away from the myocardium. HH25 cushions were placed in adenoviral stocks (> 109 PFU/ml) for one hour and then inserted into the cardiotube cultures. Periostin expressing, periostin blocking (antisense), and control (LacZ) adenoviruses were used in the infections. These adenoviruses have previously been reported as functional in expressing periostin or blocking periostin translation(Butcher et al., 2006; Norris et al., 2007b). The AV cushions are then placed into the lumen of the collagen tube by manually inserting the tissue. To aid in the placement of the AV cushions into the tubes, most of media is removed from the culturing dish prior to insertion into the lumen. This prevents the cushion from floating away. Cushions were allowed to adhere to the inside of the lumen for 12–16 hr prior to the addition of new media. The tubes were visualized daily to determine the presence of beating myocytes and AV cushions. These AV/cardiotube cultures were grown for seven days under standard tissue culture conditions using 10% FBS in DMEM media supplemented with antibiotics.
After 7 days in culture, the cardiotubes were microdissected into smaller doughnut-shaped sections. These smaller sections were embedded in agarose (5%) and then sectioned crosswise using a vibratome (Oxford) into 200mm thick sections. Sections were then processed for immunofluorescence. Tubes are fixed in 2% paraformaldehyde at 4°C for 12–16 hr, permeabilized in 0.25% Triton X-100/PBS for 30 minutes and blocked in 2% BSA/PBS for one hour at room temperature or overnight at 4°C. The following primary antibodies were used for immunostaining: MF-20 for sarcomeric myocytes (Developmental Studies), Periostin and AB755P (type I collagen, Chemicon). These were used at concentrations ranging from 1:50 to 1:250 according to the specific antibody. Finally, nuclei were stained with DAPI (Molecular Probes, Inc). Imaging was carried out on a Zeiss LSM 510 laser confocal microscope.
In order to quantitatively assess the affect of periostin expression on myocardial and fibroblast markers, hanging drop cultures (HH25 AV cushion aggregates) followed by Western analyses were performed. Hanging drop assays were performed as previously described(Butcher et al., 2006; Norris et al., 2007b) with one exception: cultures were grown for seven days in the presence of each of the adenoviruses. For each virus, a total of 10 HH25 AV cushion aggregates were infected. Adenoviral infection was performed after the AV cushions had formed spheroid aggregates. After 7 days in culture, the aggregates were pooled, lysed in 1X RIPA buffer and run on a 4–15% SDS-PAGE. Gels were blotted and probed with antibodies against collagen I (1:1000), hemagluttinin (1:1000; Sigma HA-7), periostin (1:5000), desmin (1:1000; abcam), myosin heavy chain (1:2000; abcam), and actin (1:10,000). Secondary antibodies and detection were performed as described above. These experiments were repeated a minimum of 3 times. For TGFβ3 stimulation, 10ng/ml of purified protein was added to the HH25 AV cushion hanging drop cultures. After 48 hours AV cushion aggregates were pooled, lysed and analyzed by Western using periostin, collagen I, and actin antibodies as previously described.
Tissue surface tension was measured with an in-house built compression plate tensiometer. Surface tension measurements were performed and calculated as previously described(Foty et al., 1996; Norris et al., 2007b). Calculation of the equilibrium force from the tensiometry force data was performed using Matlab software by fitting the force decay curve to a double exponential function(Forgacs et al., 1998). The geometric parameters were determined using an in-house written fitting program, with a precision of 3 µm. The program evaluates the explant’s recorded contour on the basis of variation in gray scale values in its vicinity, and by fitting the left and right profiles of the compressed explants at its equator to circles. HH25 explants were compressed twice with varying force with at least 30 min recovery in the uncompressed state. Additional surface tension measurements were performed by incubating the explants in 10ng/ml of TGFβ3 for 24 hour prior to compression.
We thank members of the Norris, Markwald, and Goodwin lab for their close reading of the manuscript and funding agencies as mentioned previously.
Grant Information: This study was partially funded by research Grants NIH-NHLBI: HL33756 (R.R.M), 1KO2 HL086901 (J.D.P), HL086856 (R.L.G); NIH-NCRR: COBRE P20RR016434-07 (R.R.M. and R.A.N); National Science Foundation: FIBRE EF0526854 (R.R.M. and R.A.N.); Foundation Leducq: Mitral 07CVD04 (R.A.N and R.R.M); SC INBRE: 5MO1RR001070-28 (R.A.N); and American Heart Association: 0765280U (RAN).