Heart and Circulatory Physiology

Structural and mechanical factors influencing infarct scar collagen organization

Scott D. Zimmerman, William J. Karlon, Jeffrey W. Holmes, Jeffrey H. Omens, James W. Covell


Although large collagen fibers in myocardial infarct scar are highly organized, little is known about mechanisms controlling this organization. The preexisting extracellular matrix may act as a scaffold along which fibroblasts migrate. Conversely, deformation within the ischemic area could guide fibroblasts so new collagen is oriented to counteract the stretch. To investigate these potential mechanisms, we infarcted three groups of pigs. Group 1 served as infarct controls. Group 2 had the endocardium slit longitudinally to alter local systolic deformation. Group 3had a plug sectioned from ischemic tissue and rotated 90°. The slit altered systolic deformation in the infarcted tissue, changing circumferential strain from expansion to compression and increasing radial strain and shears and the variability of collagen fiber angles but not the mean angle. In the plug pigs, when deformation, matrix orientation, and continuity are altered in the infarct area, the result is complete disarray in the organization of collagen within the infarct scar.

  • cardiac strain
  • infarct healing
  • myocardial infarction
  • collagen fiber angles

shortly after acute cessation of coronary blood flow a complex reparative process is initiated, which includes tissue necrosis, removal of necrotic tissue, replacement fibrosis, scar contraction, and evolution of a highly organized scar with unique material properties (15, 16, 18-20). In a mature infarct scar, as much as 25% of the dry weight may consist of collagen (25). Large collagen fibers [picrosirius-red stained red or orange, 1.6–2.4 μm in diam (10)] within an infarct scar are highly organized in each layer. The mean fiber angles in those layers are biased toward a circumferential orientation but retain the organized counterclockwise rotation from epicardium to endocardium characteristic of muscle fibers in the left ventricle (LV) (18). However, the factors responsible for forming this carefully organized scar are poorly understood.

Fibroblasts are the chief source of type I and III collagen in the heart (12, 13) and are known to be regulated by a variety of cytokines and growth factors (4, 31). Fibroblasts appear to deposit collagen in the direction of migration (2), and the path of fibroblast migration has been shown to be influenced by local structure (11). It also has been known for some time that stretch can influence fibroblast orientation (5) and migration (17) and can upregulate type III collagen production (8). Thus the dynamic environment of the ischemic heart could potentially direct the migratory and translational characteristics of cardiac fibroblasts.

In the present study, we examined three potential mechanisms for the organization of the large collagen fibers in the scar. First, we propose that the original extracellular matrix acts as a scaffold, along which fibroblasts migrate, laying down new collagen in the same orientation as the original matrix. Second, for the extracellular matrix to act as a scaffold, there must be continuity between the matrix of the infarct and that of the nonischemic myocardium, which is the source of the migrating fibroblasts. Third, we propose that passive stretch of the infarcted tissue (20, 26) regulates the preferred direction of fibroblast migration. Although the necrotic myocytes are resorbed postinfarction, there is evidence that a substantial portion of the preexisting matrix remains. Collagen degradation may occur early postinfarction (27). However, the degree to which early degradation removes the original matrix is unclear. Some investigators have found a substantial, although not complete, early collagen loss within the ischemic area (9), whereas others have found less degradation (33). More recent evidence suggests that the collagen compartment is relatively stable following coronary artery ligation. In fact, Wiggers and colleagues (34) found no evidence of collagen degradation after 6 h of coronary artery occlusion. Collagen degradation due to ischemia appears to preferentially affect smaller collagen fibers, whereas larger fibers are less affected (35). Thus available evidence would support the contention that at least a majority of the original extracellular matrix remains intact following porcine coronary artery ligation.

Further evidence to support the first hypothesis comes from structural studies of viable myocardium and infarct scar. It has been known for some time that large collagen fibers are aligned parallel to cardiomyocytes in vivo (9). Collagen fiber angles from infarct scar are similar to muscle fiber angles in viable LV (18). Thus it seems likely that these preexisting large fibrils may form the underlying structure that results in a dense and highly organized mature cardiac scar.

The third hypothesis requires that fibroblasts respond to the deformation of the ischemic area during the cardiac cycle. The porcine ischemic area is slightly stretched circumferentially (seen as bulging) during systole (20). Fibroblasts will orient themselves in the direction of tension generated in a collagen gel (29), suggesting the capacity to recognize stretch. By following the direction of stretch within the ischemic region, fibroblasts could orient collagen parallel to the direction of that stretch.

In the present study, we investigated the role of the preexisting matrix and local deformation following coronary ligation in infarct scar organization using two separate protocols, one which changed only deformation in the ischemic area and the other which changed deformation, the orientation of the preexisting matrix, and the continuity between the preexisting matrix and the normal myocardium. The results indicate that the preexisting matrix and systolic wall deformation influence the organization of large collagen fibers in the infarct scar.


Surgical preparation.

All experiments were conducted in accordance with American Association for Accreditation of Laboratory Animal Care guidelines for the use of animals in research and were approved by the University of California San Diego Animal Subjects Committee (protocol S-98041).

Twenty-one Hampshire farm-raised pigs weighing ∼20 kg were sedated with ketamine (24 mg/kg) and pretreated with atropine (0.03 mg/kg). Anesthesia was induced with isoflurane (4–5%) by mask. All animals were then intubated, ventilated, and maintained with a mixture of isoflurane (1%), nitrous oxide, and oxygen. The heart was exposed via a fifth interspace thoracotomy. The pericardium was opened, and the heart was exposed. Arterial pressure was monitored throughout via a right femoral catheter and pressure transducer. Cardiac support (dobutamine hydrochloride, 0.33 g ⋅ kg−1 ⋅ min−1) and antiarrhythmics (lidocaine hydrochloride, 2%, 0.04 ml/kg) were administered as necessary during the procedure to maintain femoral artery pressure above 65 mmHg and to prevent prolonged episodes of ventricular tachycardia. Inotropic support and lidocaine were discontinued as soon as the chest was closed, and no animals received postoperative support.

In slit and control animals, three columns of three radiopaque gold beads (0.8–0.9 mm diam) were implanted transmurally within the LV wall in tissue supplied by either the first or the second branch of the left circumflex coronary artery (Fig.1 A). Beads were implanted by temporarily suturing a platform onto the epicardium above the desired location. The platform had three holes, allowing access to the heart and providing a template for the insertion of a trocar. A sharp hollow metal probe (trocar) was inserted to a predetermined depth as controlled by a threaded gauge. A gold bead was inserted into the probe and gently pushed to the end of the probe with a blunt plunger of the same length. The depth gauge was then threaded in 5 mm and a second bead was placed. This was repeated until three to four beads were in the myocardium and three columns of beads were completed. After removal of the platform, a larger gold bead (1 mm diam) was sutured on the epicardium to mark each transmural column. Beads were also sutured to the LV apex and the bifurcation of the left coronary artery to define an apex-base cardiac coordinate system for the transmural bead set. The dissected artery was then ligated, and the three-dimensional location of each bead was assessed by biplane cineradiography. Ischemia was verified visually by tissue blanching and akinesia. In slit animals, a no. 12 scalpel blade was inserted through the apex and into the LV chamber. Using anterior-posterior axis X-ray to view the blade, an incision was made in the endocardium from the basal to the apical margin of the bead set 7–9 mm deep (dashed line on Fig.1 A). Care was taken to avoid incising the epicardium. Biplane X-ray was then used to confirm altered wall deformation following the slit through comparison of the pre- to postslit images using the finite element method described inStrain Calculation. X-ray images were taken 5 min postligation (preslit) and within 20 min of the slit (postslit). Of the eight animals to undergo this procedure, three died of surgical complications.

Fig. 1.

Schematic representation of surgical procedures.A: endocardial slit protocol. The slit was made parallel to the apex-base axis in nonischemic tissue.B: plug rotation protocol. The plug was cut from ischemic tissue but not separated from the endocardial surface, rotated 90°, and reapproximated. LAD, left anterior descending coronary artery; LCX, left circumflex coronary artery.

In a third group of pigs (plug), the first or second branch of the left circumflex coronary artery was dissected free and ligated. A tissue block (∼1.5 × 1.5 × 0.8 cm) within the ischemic region was separated from the surrounding wall but not from the endocardium and rotated 90° (Fig. 1 B). The epicardium of the tissue block was then reapproximated to the surrounding epicardium with 5-0 silk suture. It was not technically possible to place a full set of 12 gold beads in the tissue plug. Of the eight animals to undergo this procedure, two died due to coronary artery ligation. Overall, we experienced a 39% intraoperative mortality from all procedures.

After each procedure, the infarct region was covered by the pericardium when possible and the chest was closed. Air was evacuated from the chest, anesthesia was discontinued, and animals were allowed to recover. All pigs received sulfamethoxazole-trimethoprim (400:80 mg) in oral suspension for 5 days following surgery.

Strain calculations.

After each biplane X-ray experiment, a calibration phantom with beads embedded in known locations was imaged and used for perspective calibration (30). Digitizing the two views of the phantom allowed calculation of a transformation matrix from the known phantom bead positions and the digitized coordinates. A least-squares finite-element method was used to calculate nonhomogeneous distributions of three-dimensional finite strain as described elsewhere (24). Briefly, arterial pressure tracings and electrocardiogram synchronized with a frame grabber (Data Translation DT2651) were used to define end-diastolic and end-systolic frames. At least two frames were averaged to obtain each end-diastolic or end-systolic view and the three-dimensional coordinates of each bead were computed from the transformation matrix. The coordinates were then converted to cardiac coordinates defined by the apex and base beads. Three normal strains (E11, E22, and E33) quantified stretch or shortening along local circumferential (X 1), longitudinal (X 2), or radial (X 3) axes. Three shear strains (E12, E13, and E23) quantified changes in angle between pairs of initially orthogonal axes. End-systolic strains and shears were therefore calculated as the bead set deformed between the diastolic and systolic configurations. Deformation data from the control animals have been reported previously (20). The resolution of this technique has been calculated to be 0.2 mm (30).


Three weeks postsurgery, all animals were again anesthetized and a medial sternotomy was performed. Once the heart was exposed and any noncardiac tissue was dissected away, an overdose of pentobarbital was administered. The heart was brought to anoxic arrest through ligation of the superior and inferior vena cavae and main pulmonary vein. Hearts were dissected free and weighed, and the great vessels, atria, and valves were trimmed. The right and left coronary arteries were cannulated via their respective ostia, and hearts were fixed by perfusion at 120 mmHg with buffered Formalin (Fisher Scientific) while suspended in saline. The right ventricle wall was removed and a plastic rod was inserted through the apex, which extended beyond the base along the apex-base axis. The LV chamber was filled with silicone, which was allowed to set for 48 h. Hearts were placed in a cutting apparatus supported by the rod. The apparatus allowed cuts to be made in the LV wall perpendicular to the apex-base axis. Tissue blocks from the center of the infarct scar or in the slit pigs containing gold beads were removed for analysis.

Tissue blocks were dehydrated, embedded in paraffin, and sectioned at 10 μm thickness in the epicardial tangent plane. Sections were mounted and stained for collagen with picrosirius red. Images of each slide were oriented circumferentially and acquired at 200 total magnification using a color video camera (Sony CCD/RGB) mounted on a microscope and connected to a Macintosh computer (Quadra 900 with National Institutes of Health Image v.1.61 software). Images were downloaded to a Unix (Silicon Graphics, O2) workstation and analyzed with an automated intensity gradient technique for determining fiber orientation (23) relative to the circumferential axis of the heart. Tissue sections from control animals had been used in a previous study (18) but were independently reanalyzed for this project using the gradient density technique. Mean orientation and angular deviation for each image were calculated using circular statistics (14). The two most endocardial images for each pig were averaged to give endocardial collagen fiber angle and dispersion. Midwall data were obtained from one or two images depending on the thickness of the scar. Epicardial collagen fiber angle and angular dispersion were measured from sections 1.0 and 1.5 mm from the epicardium.


The effect of the slit on systolic strains was calculated using ANOVA with repeated measures for depth (SuperANOVA v.1.1, Abacus Concepts). Changes in mean angle and angular deviation between the three groups at different depths were assessed by two-way ANOVA. Significance was set at P ≤ 0.05.


Strain data.

The adjacent endocardial slit altered normal systolic strains in the infarcted area as shown in Fig.2 A. The slit converted mean circumferential strain (E11) from slight expansion (0.011 ± 0.005, means ± SE) to slight compression (−0.020 ± 0.004, P = 0.0003) compared with preslit. The slit procedure also increased mean radial strain (E33, 0.003 ± 0.010 preslit vs. 0.034 ± 0.015 postslit,P = 0.0011) and significantly altered shear strain (Fig. 2 B). Mean shear strains in both the circumferential-longitudinal and circumferential-radial planes were increased following slitting of the adjacent endocardium (E12, −0.006 ± 0.003 vs. 0.020 ± 0.006,P = 0.008; E13, −0.004 ± 0.006 vs. 0.014 ± 0.008, P = 0.05).

Fig. 2.

Normal (A) and shear (B) strains following coronary artery ligation pre- and postendocardial slit. Values are means ± SE. Significant differences between pre- and postslit values were found. * P ≤ 0.05. † P ≤ 0.001. ‡ P ≤ 0.0005.

Histological data.

There was a significant interaction of treatment and tissue depth on collagen fiber angle in the infarct scars. At both the epicardium and endocardium, collagen fiber angle was significantly altered in the plug but not the slit pigs compared with controls (Fig.3). Plug pig mean fiber angles clustered around zero degrees at all depths (epicardium 3.77 ± 4.89, midwall 0.41 ± 15.26, endocardium 11.76 ± 12.80), with a high standard deviation of the mean angle measurements in each image (Fig.4). Infarct scar collagen fiber angles in the control and slit animals are in good agreement with past measurements made from pigs in this laboratory from a similar infarct protocol (18).

Fig. 3.

Effect of endocardial slit and tissue plug rotation on infarct scar collagen fiber angles relative to circumferential axis. Values are means ± SE; † significantly different from control (P < 0.001).

Fig. 4.

Representative light microscopy images of control (1A), slit (2A), and plug (3A) pig scar tissue and histograms of fiber angles measured (1B,2B,3B, respectively). Mean angle and angular deviation of each image are shown within histogram. Note higher degree of disorder (angular deviation) in both2 and3 relative to1. Y-axis of histograms represents number of angles measured within each 10° bin. Sections selected have mean angles near 0° for ease of comparison.

Collagen disarray in the infarct scar was increased in both the slit and plug pigs when assessed by angular deviation about a mean angle calculated by circular statistics using our automated method (Fig. 4). There was a main effect of the treatment on collagen fiber dispersion in infarct scars irrespective of tissue depth. The slit procedure significantly increased collagen disorganization relative to the infarcted control animals, whereas the plug procedure was associated with even greater disarray as measured by angular deviation (control 11.72 ± 0.975° vs. slit 18.47 ± 1.60° vs. plug 27.65 ± 1.76°; Fig. 5). Collagen fiber data from plug animals were examined for the presence of a bimodal distribution. A bimodal distribution may have indicated two distinct groups of collagen fibers, one produced before infarction and rotation and the other produced after the plug procedure. Our distributions were unimodal and did not delineate newer vs. preexisting collagen.

Fig. 5.

Effect of endocardial slit and tissue plug rotation on angular dispersion of infarct scar collagen. Plug tissue exhibited significantly greater dispersion than control († P ≤ 0.001) and slit (* P ≤ 0.05) at all depths. Slit tissue was more highly disordered than control at midwall and endocardium (‡ P ≤ 0.05).


In the present study, we examined three hypotheses to explain the organization of the large collagen fibers in the scar. First, we propose that the original extracellular matrix acts as a scaffold along which fibroblasts migrate, laying down new collagen in the same orientation as the original matrix. Second, for the extracellular matrix to act as a scaffold, there must be continuity between the matrix of the infarct and that of the nonischemic myocardium, which is the source of the migrating fibroblasts. Third, we propose that passive stretch of the infarcted tissue (20, 26) regulates the preferred direction of fibroblast migration. The results of the present work suggest a predominant role for the continuity and orientation of the preexisting extracellular matrix in the subsequent organization of myocardial scar collagen. They also demonstrate that changes in systolic deformation do not effect mean collagen fiber angle within the scar. These results provide support for the premise that wall deformation and the original orientation and continuity of the preexisting extracellular matrix all influence infarct scar organization.

Fibroblasts exert tremendous influence over the organization and orientation of developing collagenous tissues (2). Fibroblasts migrate on type I collagen gels (1) and will preferentially move in the direction of collagen fibril orientation (11). Fibroblast interaction with the collagen matrix most likely involves collagen binding to members of the β1-integrin subfamily (for review see Refs. 21 and 22), of which α1 and α2β1-integrins are expressed on cardiac fibroblasts (6, 7). Migration on type I collagen has been shown to be partially mediated through the α2-subunit in melanoma cells (28). It has been suggested from observations made in developing corneal tissues that fibroblasts migrate along an existing collagenous matrix and deposit collagen that is aligned with the original stroma (2). In the normal myocardium, large collagen fibers are aligned parallel to cardiomyocytes in vivo (9). Collagen fiber angles from infarct scar are similar to muscle fiber angles in viable LV (18). Therefore, the loss of continuity of the extracellular matrix, as in the plug protocol, may impede fibroblast migration and impair the deposition of collagen in an organized fashion.

The hypothesis of fibroblast migration along the original collagen matrix requires that the original matrix remain in the ischemic tissue in some form following infarction. Degradation of collagen can begin within 20 min of coronary artery occlusion (27) and may result in the loss of up to 50% of the collagen in the ischemic area within 3 h of infarction, and by 24 h there is substantial degradation of collagen (9). However, light microscopy data do not support the contention that a large amount of collagen is lost from the ischemic region postinfarction (33). In fact, some evidence disputes whether early degradation even occurs in the pig. For example, no evidence of collagen degradation was found, whether measured biochemically or histologically, after 6 h of left anterior descending coronary artery occlusion (34). Most of the early collagen disruption due to ischemia appears to affect smaller collagen fibers, whereas larger “cables” are less affected (35). At the least, evidence would suggest that early degradation does not eliminate all collagen within the ischemic area.

It is possible that cytokines may determine the direction of fibroblast migration. The predominant source of cytokines in the scar is inflammatory cells that migrate in from the edge of the infarct. However, because the scar is not organized in a radial fashion from the center of the infarct, cytokines seem unlikely to be a factor controlling the organization of the scar, although they clearly influence the marked increase in the concentration of collagen in the scar.

Both the slit and plug protocols showed increased variance in collagen fiber orientation; however, only the plug pigs had a different mean fiber orientation. Inasmuch as both treatment groups may have experienced altered wall deformation compared with control hearts, stretch may play a role in determining collagen fiber organization. Fibroblasts are the primary collagen-producing cells in the heart (12,13). Mechanical load alters fibroblast phenotype and the mosaic of gene expression through a variety of potential transduction systems (for review see Refs. 3 and 4). In vitro studies indicate that stretch induces fibroblasts to migrate parallel to the direction of stretch (29) and causes rat cardiac fibroblasts to increase production of type III collagen mRNA (8). In the present study, there appeared to be no consistent relationship between the acute change in direction of deformation and the change in collagen fiber angle. We were able to directly measure a change in systolic wall deformation in the slit animals and document a significant change in radial and circumferential strain following the slit. Unfortunately, the fragility of the ischemic tissue following plug rotation precluded a direct measurement of wall deformation. However, it was apparent following approximation of the plug tissue that a difference in the wall motion existed between the plug and the rest of the ischemic area and thus we assume that deformation was altered in the center of the plug. It is our assumption then, that both surgical preparations altered wall deformation in the ischemic area. After the slit was produced, deformation in the scar changed without affecting the average direction of the large collagen fibers, whereas scar disorganization was moderately increased. Thus if the infarct matrix is intact and contiguous with the adjacent noninfarcted tissue, scar collagen mean angle is not altered. The plug protocol altered matrix continuity, orientation, and wall deformation with a subsequent change in mean collagen fiber angle and a dramatic increase in fiber disarray. The small mean angles at each depth are probably reflective of the high degree of disorder in the plug animals. In addition, although we did not directly measure the rotation of tissue through the depth of the plug, there must have been a continuum of rotational angles from 90 to 0° from epicardium to subendocardium. Interestingly, collagen fibers in the infarct scar plug are consistently disorganized at all depths of the plug. Thus the amount of rotation of the original tissue has no bearing on the disorganization of scar collagen. Although speculative, it appears that an intact and organized extracellular matrix may be necessary for the proper organization of infarct scar collagen.

Under ideal circumstances, it would be best to independently vary deformation and matrix orientation and continuity. Unfortunately, this is not possible in the intact myocardium. However, the slit studies in which deformation was markedly altered and mean fiber angle not changed strongly support the concept that the preexisting matrix is an important structural factor in determining the mean orientation of collagen fibers in the scar. The studies in which we changed the orientation of the preexisting matrix have two serious limitations. First, in addition to altering the direction of the matrix we also disrupted the continuity of the fibers. The fact that we saw a disorganized highly variable scar under these circumstances may indicate that continuity of fibers in the scar with adjacent normal fibers is necessary. This would further support the hypothesis that cardiac fibroblasts are using the preexisting matrix as a scaffold for migration. Studies in the plug pigs also have the limitation that the epicardium is incised, exposing the underlying myocardium to the wound healing process occurring in the mediastinum. This may have influenced the degree to which the preexisting matrix is reabsorbed in the plug. It is not possible to determine the effects of incising the epicardium in the present protocol, and performing a similar study from the endocardium would require cardiopulmonary bypass. Because we studied tissue remote from the slit, the slit protocol did not have a similar problem.

The results of this work suggest a role for both systolic wall deformation and extracellular matrix orientation and continuity on subsequent infarct scar organization in the porcine heart 3 wk postmyocardial infarct. Altering wall deformation alone decreases collagen fiber organization within an infarct scar while not changing the mean orientation large fibers. Altering wall deformation and the orientation and continuity of the original extracellular matrix to the degree they were in the present work results in a disorganized scar with a random orientation of collagen. Thus the preexisting matrix and systolic wall deformation appear to be important regulators of the organization of large collagen fibers in the infarct scar. The fact that the mean fiber angle was not changed in the slit pigs and was in the plug pigs leads us to speculate that the preexisting matrix plays an important role in infarct scar collagen organization.


  • Address for reprint requests and other correspondence: J. W. Covell, 9500 Gilman Dr., 0613J, Univ. of California, San Diego, La Jolla, CA 92093 (E-mail: jcovell{at}ucsd.edu).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract