To determine whether there are structural differences in two topologically separated, biochemically defined mitochondrial populations in rat heart myocytes, the interior of these organelles was examined by high-resolution scanning electron microscopy. On the basis of a count of 159 in situ subsarcolemmal mitochondria (SSM, i.e., those that directly abut the sarcolemma), these organelles possess mainly lamelliform cristae (77%), whereas the cristae in in situ interfibrillar mitochondria (IFM, i.e., those situated between the myofibrils, n = 300) are mainly tubular (55%) or a mixture of tubular and lamelliform (24%). Isolated SSM (n = 374), similar to their in situ counterparts, have predominantly lamelliform cristae (75%). The proportions of crista types in isolated IFM (n = 337) have been altered, with only 20% of these organelles retaining exclusively tubular cristae, whereas 58% are mixed; of the latter, lamelliform cristae predominate. This finding suggests that, in contrast to SSM, the cristae in IFM are structurally plastic, changing during isolation. These observations on >1,000 organelles provide the first quantitative morphological evidence for definitive differences between the two populations of cardiac mitochondria.
- high-resolution scanning electron microscopy
the mitochondria of cardiomyocytes consist of two spatially disparate populations: one abuts the sarcolemma, and the other is trapped within the contractile apparatus. Techniques that permit the separate isolation of these two populations have led to the finding that in several mammalian species the two show significant physiological differences (6, 7, 10). Despite these documented biochemical differences, studies based on conventional transmission electron microscopy (TEM) have failed to reveal morphological disparities between the two populations. We have used a recently developed osmium-extraction technique for scanning electron microscopy (SEM) that allows inspection at high resolution (HRSEM) of the interior of cells (16). With this methodology, we found that cardiac mitochondria have different cristae that characterize each population. These differences are maintained, albeit to a lesser degree, in isolated mitochondria separately derived from the two sets of cardiac mitochondria. High-voltage electron microscopic (HVTEM) tomography (3, 8, 9, 12–14) has paved the way for a reexamination of crista structure in mitochondria in other tissues, but HRSEM permits quantitative estimation of crista morphology in a very large sample compared with the very limited number of tomographic reconstructions.
MATERIALS AND METHODS
Six Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA; 330–690 g body wt, 2.5–6 mo of age) were killed by decapitation, and the hearts were extirpated. The protocol and animal care were reviewed and approved by the Institutional Animal Care and Use Committees of the Louis Stokes Veterans Affairs Medical Center Medical Research Service and Case Western Reserve University School of Medicine.
Mitochondrial isolation and biochemistry.
A portion of the left ventricle from each heart was retained for microscopic examination. The remainder (left and right ventricles, but not atria) of each heart from four of the six rats was used to isolate subsarcolemmal mitochondria (SSM) and interfibrillar mitochondria (IFM) according to Palmer et al. (10) except Chappell-Perry buffer was used. Oxidative phosphorylation was studied as previously described (2, 10).
For HRSEM (16), fresh, full-thickness 3 × 5-mm strips of left ventricular tissue were fixed for 15 min at room temperature in 0.5% glutaraldehyde-0.5% paraformaldehyde in 0.1 M cacodylate buffer. The tissue was rinsed in three changes of PBS, each for 10 min. Specimens were postfixed for 2 h in the dark at 4°C in 1:1 2% OsO4-1.25% potassium ferrocyanide. After another rinse in three changes of PBS, the tissue was embedded in 1% agarose and cut with a TC2 Sorvall tissue sectioner into 150-μm-thick sections. The sections were rinsed three times in PBS and subjected to a second postfixation in ferrocyanide-reduced osmium for 1 h in the dark and rinsed three times in PBS. The sections were macerated (osmium extracted) for 44–48 h in 0.1% OsO4 at 25°C and then rinsed in PBS. They were dehydrated in ascending concentrations of acetone and then critical point dried using carbon dioxide. Mitochondrial pellets were processed in almost the same fashion as were solid tissues, except the former were embedded in agarose and fractured with liquid nitrogen. The sections or pellet fragments were coated with platinum in an Emitech 575 sputtering apparatus and examined in an FE Hitachi S4000 scanning electron microscope operating at 15–20 kV.
To classify and quantitate in situ mitochondrial types according to their crista structure, only transversely sectioned cardiomyofibers with intact sarcolemmas were included. Mitochondria were considered subsarcolemmal only if they abutted the sarcolemma. IFM were some distance from the sarcolemma. Two observers independently counted those mitochondria with intracellular location that was clear cut according to the foregoing criteria. In the case of the mitochondrial pellets, mitochondria were counted in micrographs of random fields.
Results of oxidative phosphorylation are expressed as means ± SD. Differences between the IFM and SSM were compared using Student's t-test. For the SEM data, a 2 × 2 χ2-test was performed comparing lamellar vs. tubular cristae in SSM and IFM in situ and in isolated mitochondria. P < 0.05 was considered significant.
Figure 1 a shows a cardiac muscle fiber in transverse section; the transected sarcolemma is retained. The contractile apparatus has been extracted in its entirety; the gaps between columns of mitochondria represent the space originally occupied by myofibrils. Those mitochondria immediately subjacent to and abutting the sarcolemma correspond to SSM (Fig. 1b) on the basis of TEM. In contrast, those mitochondria situated amid the gaps correspond to the IFM of TEM (Fig. 1c). To quantitate types and subcellular location of in situ mitochondria, two investigators counting independently scored only transversely sectioned cardiomyofibers. To identify SSM vs. IFM, only fibers with clearly visible sarcolemma (Fig. 1a) were examined. Furthermore, only mitochondria with unambiguous intracellular position were counted.
On the basis of examination of multiple sections of four hearts, 77% of the SSM had lamelliform cristae (Fig. 2, a and b, and Fig. 3A). Lamelliform cristae are broad and flat, may show through-and-through fenestrations, and are joined to the boundary membrane by numerous crista junctions, to use the terminology of Perkins et al. (14). Some of the lamelliform cristae have small vesicular protuberances along their free edges; these may represent transected crista junctions (Fig. 2b).
Crista structure in in situ IFM differs from that in SSM. On the one hand, some mitochondria possess only tubular cristae (Figs. 2c and 3A), whereas the remainder contain one or several lamelliform cristae interspersed among the tubular cristae (Fig. 2d) or may have only lamelliform cristae (Fig. 3A). Such flat cristae display an abundance of crista junctions. Crista junctions as such are not identifiable in the tubular cristae, perhaps because the overall diameter of these cristae is a constant, so that even if crista junctions are present, they do not appear as a constricted portion of these inner membrane structures. The tubular cristae frequently branch and occasionally anastomose with one another, sometimes forming a lattice (Fig. 2d). Fifty-five percent of the IFM have tubular cristae exclusively, although another 24% have tubular cristae, among which are lamelliform cristae (Fig. 3A). Some of the tubular cristae are fingerlike, whereas others are disposed in lattices. Twenty-one percent of the IFM had only lamelliform cristae.
In summary, our observations show that, although most cristae in in situ SSM are lamelliform, an occasional tubular crista occurs in these organelles. In contrast, the polar opposite of this observation occurs in IFM, where flat cristae occur among the predominant tubular cristae (Fig. 3A).
Our metabolic studies of isolated cardiac mitochondria paralleled our previous published work (10). These data are shown in Table 1. The yield of mitochondrial protein in the SSM and IFM as well as their state 3 and 4 rates of oxidation (data not shown), respiratory control ratios, ADP-to-O ratios, and maximal ADP-stimulated rates of oxidation are virtually identical to values reported earlier in adult rats (2, 10).
To quantitate the morphology of isolated mitochondria of each population, micrographs of random fields were scored in terms of crista organization in the same fashion as the in situ organelles. The structure of isolated SSM is nearly identical to that of in situ SSM, with most (75%) having lamelliform cristae (Figs. 3B and 4 a). Among the remaining isolated SSM, most had a mix of lamelliform and tubular cristae (Fig. 4b). Compared with the in situ situation, the percentage of IFM with tubular cristae (Fig. 4c) has been halved (20%), whereas the percentage of mitochondria with heterogeneous cristae (Fig. 4d) has doubled (58%; Fig. 3, A and B). Virtually the same percentage of IFM with lamelliform cristae as that prevailing in situ is retained in the isolated organelles (Fig. 3, A and B).
HRSEM combined with improved osmium extraction techniques have led to the discovery of certain structural features of rat heart mitochondria not previously reported by TEM. SSM have predominantly lamelliform cristae, whereas interfibrillar organelles have cristae that mainly are tubular or have a mixture of tubular and lamellar cristae. Our study indicates that, to a large extent, isolated SSM retain cristae that are virtually identical, i.e., lamelliform, to those they exhibit in situ. In contrast, in isolated IFM, there is a doubling in the number of those organelles that have heterogeneous cristae and a reciprocal decrease in the number of mitochondria with only tubular cristae. One interpretation of these observations is that cardiac mitochondrial cristae are interconvertible in terms of structure. If this is true, then what is the signal for conformational change? Are changes in crista structure and mitochondrial function in lock step? Model systems exist that could be used to address this issue. In hearts of cardiomyopathic hamsters (6) and aged rats (2), IFM exhibit a selective decrease in oxidative phosphorylation, but TEM has failed to detect ultrastructural alterations in this mitochondrial population; HRSEM examination of cristae in osmium-extracted IFM in aged hearts is under way.
In brown fat mitochondria, which previously have been examined by HVTEM (8), our results obtained with HRSEM (16) are identical to those obtained with HVTEM. In such mitochondria, we have noted the same arrangement of cristae and crista junctions with the same resolution as in HVTEM tomography. A major advantage of our methodology is that numerous (literally, hundreds) organelles can be observed in a single specimen without the laborious reconstructions (that limit sample size) required for HVTEM tomography. Because osmium extraction is carried out on thoroughly fixed sections, rather than whole specimens, there is no concentration gradient of osmium during extraction; instead, all areas of a given cell are simultaneously exposed to this agent, thus eliminating the distance from the extracellular space as a factor in the morphological differences that we report here. We and others using HRSEM examined the ultrastructure in other tissues of an assortment of membranous organelles, including rough endoplasmic reticulum (15, 16, 18), Golgi apparatus (16, 18), annulate lamellae (16), and secretory granule (15, 16), and, in every case, we found these organelles to precisely match their TEM counterparts. Moreover, we used our method to observe mitochondria in a very large number of organs and have concluded that osmium extraction is no more likely to produce artifacts than is any other ultrastructural technique in use. So, in a sense, HRSEM and HVTEM tomography are not only compatible, they are complementary as well.
The morphological differences we describe between the two populations of cardiac mitochondria from normal hearts parallel the biochemical differences in these two sets of organelles. The rate of oxidative phosphorylation in the IFM is ∼150% of that observed in the SSM (10). Both populations of isolated mitochondria are well coupled and have excellent ADP-to-O ratios, marking these mitochondria as functionally normal. An additional biochemical difference is that the specific activity for many, but not all enzymes, also is 150% greater in the IFM than in the SSM (10). How do these biochemical differences relate to the morphological observations? Although IFM have higher specific activities of some enzymes, these enzymes are not controlling for oxidative phosphorylation; thus the basis for the enhanced activity of the IFM must lie elsewhere. What advantages are conferred on the IFM by having tubular, rather than flat, cristae? One possibility is that a reduction of the intracristal space of tubular cristae leads to a higher concentration of protons within these structures, thus enhancing ATP synthase activity, which facilitates oxidative phosphorylation. A case in point is the alteration in mitoplast morphology occasioned by changing oxidative state (13). An additional possibility is that key molecular interactions within the electron transport chain complexes situated in crista membranes are affected by membrane conformation. For example, Schägger (17) described “supercomplexes,” i.e., assemblies of the electron-transport chain complexes that require close association to operate at maximum efficiency. The possibility also exists that the biochemical composition of the two types of cristae differs to some extent. The phospholipid or protein composition of their membranes might play an important role in determining crista morphology. Dynamin (1, 4) and ATP synthase (11) have been shown to influence crista morphology in yeast and in several mammalian tissue culture cell types. The present emphasis on mitochondrial proteomics and lipidomics could lead to data relevant to this point.
This work was supported in part by the Italian Ministry of Health FIRB Program, National Institute on Aging Grant P01 AG-15585, and the Department of Veterans Affairs Medical Research Service.
We thank Drs. Giacomo Diaz and Alan Davis for statistical assistance and Drs. Edward Lesnefsky, John Mieyal, and Medhat Hassan for reviewing the manuscript. Gabriele Conti, Michela Isola, Francesco Loy, Marco Piludu, and Rachel Floyd provided technical assistance.
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. Section 1734 solely to indicate this fact.
- Copyright © 2005 by the American Physiological Society