INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

IT IS RECOGNIZED THAT AN INCREASE in collagen concentration is observed as an integral part of extracellular matrix (ECM) remodeling that takes place in left ventricle (LV) consequent to the natural aging process (12, 15, 33) as well as in response to a variety of pathologies resulting in hypertrophy of this chamber (4, 8, 10, 17, 20, 21, 28). However, as first pointed out by Bartosova et al. (1), disproportionate collagen accumulation and/or interstitial fibrosis are not obligatory responses of the ECM to left ventricular hypertrophy (LVH), as normal collagen concentrations are seen in several other LVH models including anemia, arteriovenous fistula, and exercise training. The functional significance of myocardial fibrosis was elucidated by Boluyt et al. (5) in a pressure-overload LVH model using the spontaneously hypertensive rat. In this study, transition from stable hypertrophy to failure was associated with a four- to fivefold increase in expression of procollagen type I and III mRNAs. Although Boluyt et al. (5) did not evaluate the quality (degree of cross-linking) of the hypertrophied myocardial ECM, Norton et al. (32) attributed the observed increase in myocardial stiffness in this same model to alterations in collagen cross-linking profile as opposed to changes in total collagen or phenotype ratios.

In the present study, we evaluated the single and interactive effects of aging and exercise training on gene expression for procollagen types I and III in rat LV. Exercise training has previously been shown to increase myocardial prolyl 4-hydroxylase, an enzyme with an activity level that correlates positively with collagen biosynthesis (37). Because training failed to alter amount of LV collagen (hydroxyproline) in this same study, the findings were interpreted as being indicative of an increase in collagen turnover rate that could potentially alter collagen characteristics even if concentration remains unchanged. This hypothesis was confirmed by the finding that the principal nonreducible collagen cross-link in LV, hydroxylysylpyridinoline (HP), was significantly reduced as a result of exercise training in old rats (38). This result indicated that exercise training affects LV collagen metabolism posttranslationally in the myocardial ECM. Determining whether or not this same intervention impacts any age-associated changes in gene expression of the two principal fibrillar collagens found in the heart was the primary purpose of the present investigation. In this manner, we sought to determine whether a point of regulation of interstitial collagen deposition and cross-linking, and by implication its maturation and stiffness characteristics, occurs at the pretranslational level.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Young (10-wk-old), middle-aged (12-mo-old), and old (24 mo old) female Fischer 344 pathogen-free rats were obtained from the National Institute on Aging colony. Animals were housed in a climate-controlled room (25°C, 12-h light-dark cycle) and allowed food and water ad libitum.

Training. Rats within a given age group were weight-matched and randomly assigned into exercise training and sedentary control groups. This resulted in the following groups: young control, young trained, middle-aged control, middle-aged trained, old control, and old trained. Before training, all rats were familiarized with a Stanhope small animal treadmill by walking/running for 10 min/day for 5 days. Training was carried out in a manner similar to that previously performed in our laboratory (45). In brief, young, middle-aged, and old rats assigned to the training groups commenced training at 15, 13, and 10 m/min, respectively, for 50 min/day. This was progressively increased throughout the 10-wk training program, so that at its termination rats in the three age groups were running continuously at 30, 24, and 15 m/min for 60 min/day, 5 days/wk. This training program has previously been shown to produce LVH as well as to increase aerobic enzyme activity in locomotor muscle of the rat (38). Sedentary groups walked/ran on the treadmill 10 min daily to ensure similar handling and to maintain familiarity with the treadmill for subsequent testing of maximal oxygen uptake (O_2 max) (see below).

O_2 max. To assess the efficacy of the training program, O_2 max was determined via flow-through spirometry techniques employing a small-animal metabolic chamber, as described previously (45).

Tissue removal. Forty-eight hours after the conclusion of the 10-wk training program, rats were deeply anesthetized with pentobarbital sodium (50 mg/kg ip). With the use of sterile techniques, a pneumothorax was performed, and the animals were killed by removal of the heart. Both atria and right ventricle were immediately trimmed off, and the LV was divided into free wall (LVF) and septum (LVS) (28) and weighed before being flash-frozen and maintained in liquid nitrogen.

Total RNA isolation. Total RNA was extracted using the methods described by Strohman et al. (36) and MacDonald et al. (25). Frozen tissue was homogenized, on ice, in 4 M guanidine thiocyanate. For every 750 µl of homogenate, the following substances were added, followed by vortexing, in order: 50 µl of 2 M sodium acetate, 500 µl phenol, and 100 µl of a 49:1 chloroform-isoamyl alcohol mixture. After centrifugation (10,000 g for 20 min at 4°C), the supernatant was transferred to a fresh tube and 500 µl of isopropanol were added. The mixture was frozen (20°C) for at least 1 h and then recentrifuged. The pellet was collected and dissolved in 150 µl of 4 M guanidine thiocyanate; 500 µl of isopropanol were then added, and the solution was again frozen (20°C) for at least 1 h. Total RNA was pelleted by centrifugation, purified by ethanol precipitation, and quantified by absorbance at 260 nm, assuming 40 mg/ml for each unit of absorbance.

Northern hybridization. Messenger RNA levels were determined by Northern and dot-blot hybridization analyses by the use of methods adapted from those of Lehrach et al. (24) and Goldberg (16). Total mRNA (10 µg) was denatured in 50% formamide, 17.5% formaldehyde, and 1× MOPS buffer (20 mM MOPS; 5 mM sodium acetate, pH 6.5; 1 mM EDTA, pH 8.0; pH total solution to 7.0), electrophoresed through a 1.2% agarose gel, and transferred to a nylon membrane (Biotrans, Pall BioSupport, East Hills, NY). The membrane was baked (2 h at 80°C) in a vacuum oven. Prehybridization of blots took place for 8 h at 42°C in prehybridizing buffer (5× standard saline citrate; 0.1% sodium dodecyl sulfate; 5× Denhardt's solution; 50% deionized formamide; 0.1 M phosphate buffer, pH 7.0; and 100 mg/ml salmon sperm DNA). Hybridization was performed at 42°C using the same buffer, without salmon sperm DNA, containing the appropriate radioactive probes to obtain 3 × 10⁶ counts · min¹ · ml hybridization medium¹.

1

1

Dot blot hybridization. For dot blot hybridization, 5 µl of total RNA were incubated (15 min at 65°C) in 50% formamide, 7% formaldehyde, and 1× standard saline citrate. The mixture was spotted onto the nylon membrane using a Minifold I following the procedure described by Sambrook et al. (35). Membranes were baked before prehybridization as described above. Dots were quantified for relative amounts of mRNA by densitometry.

Hydroxylysylpyridinoline. Tissue samples homogenized in 4 M guanidine thiocyanate to isolate RNA were centrifuged, and the supernatant was removed. The remaining solid was lyophilized and stored at 70°C for subsequent analysis. The degree of collagen cross-linking was assessed by using the reverse-phase high-performance liquid chromatography methods of Eyre et al. (13) with modifications described previously by us (28).

Hydroxyproline. Collagen concentration was calculated from colorimetric determination of hydroxyproline concentration by the method of Woessner (42). Collagen concentration was calculated assuming that collagen weighs 7.25 times hydroxyproline and has a molecular weight of 300,000.

Statistical analysis. Data were analyzed by two-way ANOVA for aging and training main effects as well as for aging-training interactions. Post hoc contrasts to determine differences between means were also carried out when appropriate with  set at 0.05. The statistical program SuperANOVA (Abacus Concepts, Berkeley, CA) was used for all analyses.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Body and heart weights. Whole body weight (BW), LV weight, and LV-to-BW ratio for the six groups are shown in Table 1. Although exercise had no effect on BW, an overall increase with aging was observed. Although there were significant aging and training main effects for both LV weight and LV-to-BW ratios, significant age-by-training interactions for these parameters (both P < 0.05) also occurred. Post hoc analysis indicated that the effects of training on the heart were more pronounced in younger and middle-aged animals, compared with older animals (Table 1).

O_2 max. Treadmill running at age-adjusted speeds for a period of 10 wk significantly increased whole body O_2 max in all three trained groups compared with age-matched sedentary counterparts (Fig. 1).

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Effects of 10-wk treadmill running on maximal oxygen uptake (O2 max) in young, middle-aged, and old Fischer 344 rats. Values are means ± SE (n = 5-10 rats per group). * Trained rats significantly higher than corresponding sedentary control group (P < 0.05).

LV collagen characteristics. Percent collagen in LVF and LVS from young, middle-aged, and old sedentary and trained rats is shown in Fig. 2. Percent collagen in LV ECM increased progressively from young adulthood to senescence so that values at 26 mo were higher than those at both 5 and 15 mo, which in turn were also significantly different from each other. Training had no effect on LV collagen concentration at any age level. As shown in Fig. 3, there were significant overall aging-related increases in HP cross-linking for both LVF (P < 0.005) and LVS (P < 0.01). Exercise modulated this effect, but only in LVF, in which the training-induced attenuations were significant for the old trained vs. sedentary comparison (P < 0.05, Fig. 3).

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Percent collagen in left ventricular (LV) free wall (A) and LV septum (B) from young, middle-aged, and old sedentary and trained rats.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Hydroxylysylpyridinoline (HP) cross-linking concentration in LV free wall (A) and LV septum (B). * Significant difference between old trained and sedentary rats in LV free wall (P < 0.05).

LV collagen mRNA expression. A significant overall decline in expression of mRNAs for both procollagen type I (P < 0.005) and type III (P < 0.001) was seen with aging in LVF but not in LVS. Although exercise training did not prevent the overall decline seen in both procollagen types I and III in LVF, it had the effect of attenuating this decline. Thus LVF procollagen types I and III levels in middle-aged trained rats were significantly elevated over age-matched sedentary peers and no different from levels seen in young sedentary controls. Likewise, mRNA levels for the two principal fibrillar collagen phenotypes were significantly elevated in LVF from old trained rats compared with old sedentary animals and no different from those seen in middle-aged sedentary rats (Figs. 4 and 5).

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Effects of training on type I procollagen mRNA in LV free wall (A) and LV septum (B) from young, middle-aged, and old rats. * Values in LV free wall of middle-aged and old trained rats significantly elevated compared with age-matched sedentary controls (P < 0.05).

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Effects of training on type III procollagen mRNA in LV free wall (A) and LV septum (B) from young, middle-aged, and old rats. * Values in LV free wall of middle-aged and old trained rats significantly elevated compared with age-matched sedentary controls (P < 0.05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We were interested in elucidating the relationship between alterations in pretranslational levels of expression for the principal fibrillar collagens in LV and changes in ECM characteristics of this same chamber. The principal findings from the study were that a relatively short period of exercise training (10 wk), but one of sufficient intensity and duration to produce a mild LVH, reverses aging-associated declines in myocardial fibroblast expression for procollagen types I and III in LVF of middle-aged and old rats. Additionally, it attenuates the increases in ECM collagen cross-linking seen in the same region of the LV in old sedentary rats. These changes occur without affecting the overall increase in collagen concentration seen in this chamber with aging.

Relationship between collagen gene expression and collagen cross-linking. The deposition of collagen in the ECM depends on the production and secretion of properly transcribed, translated, and processed protein molecules and on cellular, gene-, and enzyme-mediated processes. Collagen cross-linking is obligatorily an ECM event, not directly under gene control. Enzyme-derived cross-links occur between collagen molecules and fibrils, and their formation is dependent on relationships among and between fibrillar collagen molecules. Cross-linking residues on adjacent collagen molecules or fibrils must be precisely aligned for cross-linking to proceed (13). Such a requirement indicates that spatial relationships between type I and III collagens are likely a controlling factor in cross-linking formation (34, 40). Populations of collagen molecules and fibrils may exist in different arrays and varying frequencies. All known lysine-derived cross-links are restricted to four sites per collagen molecule, and fibril orientation could affect both divalent and trivalent cross-link formation (13, 18). Thus perturbations in mRNA levels for the fibrillar collagens I and III suggest ongoing anabolic activity that would be reflected in protein deposition in the ECM, and such activity would be expected to affect cross-linking. However, there may be positive or negative relationships between variations in gene expression and cross-linking. For example, in pathological models of LVH, e.g., myocardial infarction, there are significant increases in collagen gene expression concomitant with increased collagen deposition and increased cross-linking (unpublished findings). The significance of increased protein synthesis and ECM deposition is the potential to also alter cross-linking profiles. We have discussed nonreducible collagen cross-linking and its potential regulation in several recent reviews (29, 30, 34).

LV collagen metabolism and aging. The present findings extend our understanding of aging and its effects on collagen metabolism in the normal healthy heart (2, 26, 27, 29, 38). Our results with respect to mRNAs for type I and III collagen in LVF, which makes up ~70% of the mass of the entire LV in the Fischer 344 strain of rat employed (28), are in general agreement with those of Besse et al. (2). These investigators reported significant declines in both type I and type III procollagen mRNA levels in rat LV from 3 to 24 mo of age. In the present study, although reductions in procollagen types I and III levels were seen in LVF with aging, they were maintained in LVS, again pointing out regional differences in collagen metabolism within the normal left ventricle (28) and in response to myocardial infarction (6). Simultaneously, percent LV collagen increased with age, as did HP cross-linking for both free-wall and septal regions of this chamber. These results imply reduced collagen turnover with aging, with degradation rates falling faster than simultaneously declining synthesis rates (4, 23), a process that allows for greater maturity of collagen as indicated by higher concentrations of the HP cross-link (34). The ECM remodeling associated with both aging and training in the present study thus differs from that seen in a variety of pathological cardiac hypertrophy models.

Collagen metabolism in other LVH models. Alterations in collagen metabolism and associated fibrosis are seen in a variety of heart pathologies. These include both pressure- (8, 10, 17, 20, 21) and volume-overload (17, 19) models of LVH. Alterations in wall stress resulting from myocardial infarction (33) are also thought to provide a potent stimulus to the ECM, resulting in a reactive fibrosis of viable myocardium quite separate from the reparative fibrosis associated with tissue necrosis and infarct formation (28, 39, 41). In the same model, 5- to 15-fold increases in type I and III procollagen mRNAs have been observed in both infarcted and noninfarcted (septal) areas of LV with the specifics and time course of the response differing somewhat between regions (9). The increase in collagen concentration and altered stiffness characteristics of the left ventricle in these and other experimental models (7) are thought to play a role in the simultaneously observed depression of selected contractility indexes, which may ultimately result in heart failure (5, 6, 10). In the pressure-overload model of LVH, as well as that produced by norepinephrine (3, 8, 11), the observed fibrosis is accompanied by an increase in mRNA for pro-₂ (I) collagen and transforming growth factor-₁, implying that collagen biosynthesis is being regulated, at least in part, at the level of the gene by cardiac fibroblasts (11).

Exercise training and collagen metabolism. The effects of exercise training on expression of message for the two principal collagens differ from those seen in the aforementioned pathological hypertrophy models in terms of both extent and duration of response. The principal effect of training seems to be one of preserving the decline in expression of mRNAs for type I and III collagen seen with aging, in contrast to the 5- to 15-fold increase in these same mRNAs associated with myocardial infarction, for example (9). Likewise, dramatic increases in these mRNAs have largely returned to control levels in functioning myocardium by 13 wk postinfarction, whereas in the abdominal aortic constriction model (8), transient sixfold increases in levels of mRNAs for both collagen type I and III had declined to control levels 7 days postbanding. In the present study, the stimulus for increased expression of the mRNAs is apparently maintained even 10 wk after inception of the training program, perhaps because of the progressively increasing intensity of exercise employed throughout the duration of the program. Whether it is alterations in hemodynamic load on the heart that is providing the stimulus for the observed elevation in LVF collagen mRNAs in trained compared with sedentary rats in the older age group, or some other exercise-related alteration in hormonal milieu that is affecting rates of myocardial collagen synthesis and degradation, is an interesting question posed by the findings of the present study.