Figure 1 Deletions in the nad7 region of mtDNA in N. sylvestris CMS mutants (see text). CMS I lacks exons III and IV; CMS II lacks the entire gene.

We present here an integrated study of the developmental, molecular, and functional alterations associated with mitochondrial deletions in CMS I and II mutants of N. sylvestris.

Plant Material. The N. sylvestris parental wild type (T) is a fertile botanical line of the Institut des Tabacs (SEITA, Bergerac, France). CMS I and II plants were derived from protoplast cultures and were maintained by backcrossing with T as the male. The plants were grown in a greenhouse under 16 h of fluorescent light at a day/night temperature of 24/17°C. For histology and fertility tests, plants were maintained in controlled chambers for 2 weeks, i.e., longer than the 10 days required for male gametogenesis; conditions were 9 or 16 h of lighting, each at 300 or 30 μmol of photosynthetically active radiation (PAR) photons m⁻²·s⁻¹. Cross-sections of anthers embedded in paraffin were stained with toluidine blue, and fresh pollen was stained with Alexander dye (13).

Western Blots and SDS/PAGE of Mitochondrial Proteins from Leaves or Pollen. For one-dimensional electrophoresis (14), a 10-μg protein was used. They were stained with 0.1% (mass/vol) Coomassie brilliant blue R-250 or were transferred for Western blotting (15) with antisera against Neurospora crassa, a 49-kDa subunit of complex I (16), wheat NAD9 (5) subunits of complex I, Sauromatum guttatum alternative oxidase (AOX) (17), or potato formate dehydrogenase (18) applied at 10⁻³ dilution (10⁻² for AOX). For two-dimensional SDS/PAGE, 100- or 10- to 20-μg proteins were used for Coomassie blue or silver nitrate (19) staining, respectively. All electrophoreses and Western blots were repeated at least five times.

Protein Microsequence Determination. Spots corresponding to the 38-kDa polypeptide were eluted from 40 dried Coomassie blue two-dimensional gels of fertile pollen mitochondrial proteins, rehydrated in distilled water, and submitted to N-terminal or internal sequence determination after digestion by endoprotease lysine C (14).

Purification of Leaf Mitochondria for Respiration Experiments. To obtain well coupled mitochondria, the standard protocol (20) was modified by raising the mannitol concentration to 0.6 M. All operations were carried out at 4°C. After grinding a few mature leaves, the filtered homogenate was centrifuged, and the crude mitochondrial pellet was purified on a self-forming 32% (vol/vol) Percoll gradient. Proteins were determined with BSA as standard (21, 22).

Respiration Measurements. Oxygen uptake at 25°C was measured with a Clark electrode (Rank Brothers, Cambridge, U.K.). The chamber contained 2–3 ml of the assay medium (0.6 M mannitol/20–30 mM KCl/5 mM MgCl₂/0.2 mM ATP/10 mM K phosphate, pH 7.2). A typical sequence of additions was: mitochondria (50- to 150-μg of protein ml⁻¹ final); substrate (10 mM glycine, 10 mM succinate, or 1 mM NADH); 50–80 μM ADP; 2 μM carbonylcyanide m-chlorophenlyhydrazone or carbonylcyanide p-trifluoro-methoxyphenylhydrazone as uncoupler; 50 μM rotenone as complex I inhibitor; 0.2 mM KCN as cytochrome oxidase inhibitor; and 50 μM n-propylgallate or 1 mM salicylhydroxamic acid as AOX inhibitors. Minute injections were made to avoid dilution and solvent effects. About 2 min were allowed after each addition to reach steady state. The potential difference across the mitochondrial inner membrane [the major (23) component of the electrochemical proton gradient in plant mitochondria] was estimated with a tetraphenylphosphonium⁺-sensitive electrode (24) inserted into the suspension.

CMS I and II Plants Present the Same Developmental Anomalies. CMS and T plants were compared throughout their life cycle (Table 1). Both mutants developed slowly (Fig. 2A), and, at the adult stage, their vegetative and floral organs were small. No seeds were produced during winter, neither spontaneously nor with manual pollination. Cross-sections indicated that, although the anther inner cell layer (tapetum) looked normal, only low amounts of degenerating microspores were present (Fig. 2 B and C). During summer, a few capsules were formed but with low seed mass. The proportion of viable pollen at anthesis varied between plants and flowers from a same plant but averaged 60% (Fig. 2 H and I).

Table 1 Phenotypical characteristics of the T line and CMS mutants of N.sylvestris

Figure 2 Morphological abnormalities and conditional male sterility in CMS genotypes. (A) Two-month old plants grown in the greenhouse under 16 h of artificial light; 1, CMS I; 2, CMS II. (B and C) Cross-section of T (B) or CMS I (C) half-anther at the microspore (more ...)

To better determine the environmental parameters of male sterility, assays were performed in controlled chambers (Fig. 2 D–G). CMS flowers shed pollen with both photoperiods at 300 μmol of PAR photons m⁻²·s⁻¹ but only with the 16-h photoperiod at 30 μmol of PAR photons m⁻²·s⁻¹. T plants were normally fertile in all conditions. Temperature had no effect over the range 17–30°C (not shown).

CMS Mitochondria Are Deficient in NAD7, NAD9, and a Nucleus-Encoded, 38-kDa Subunit of Complex I. According to Western blots with antiserum against the N. crassa 49-kDa subunit, the equivalent nad7 product (42 kDa) was missing in CMS I (12). We extended this observation to CMS II (Fig. 3). This result was expected from the deletion of all four nad7 exons in this mutant. Surprisingly, the 28-kDa signal seen on T Western blots with wheat NAD9 antiserum was invisible on CMS blots (Fig. 3).

Figure 3 Western blots of T and CMS mitochondrial proteins of complex I. Immunodetection was performed with antisera against the N. crassa 49-kDa subunit (NAD7) and the wheat NAD9. Potato formate dehydrogenase (FDH) served as the control. Lanes: 1, T; 2, CMS I; (more ...)

To look for other possible complex I defects, two-dimensional SDS/PAGE patterns of mitochondrial proteins from leaf and mature pollen were compared (Fig. 4). After silver staining, a 38-kDa spot, barely detectable with Coomassie blue (Fig. 4 a and b), was clearly visible on T patterns of leaf (not shown) and pollen (Fig. 4c), but it was missing in CMS I (Fig. 4d) and II (not shown).

Figure 4 Two-dimensional SDS/PAGE of mitochondrial proteins from fertile T and CMS plants. (a and b) Coomassie blue staining of T leaf (a) and T pollen (b). (c and d) Silver nitrate staining of T (c) and CMS I (d) mature pollen collected during summer (more ...)

The 38-kDa polypeptide was more abundant in pollen than in leaves (Fig. 4 a and b), allowing collection of sufficient amount for microsequencing. It showed significant homology with the 39- to 40-kDa complex I subunit, nucleus-encoded in all species thus far studied (25–28), especially with the related species S. tuberosum (Fig. 5). A nuclear location for the encoding gene in N. sylvestris could be inferred from two facts: (i) the N-terminal amino acid of mature protein is not a methionine, suggesting posttranslational processing, possibly involved in mitochondrial addressing; and (ii) the polypeptide does not correspond to a labeled spot on two-dimensional patterns of in organello synthesized T mitochondrial polypeptides (14).

Figure 5 Sequence homologies of the N. sylvestris 38-kDa polypeptide with complex I subunits of other species (25–28). N. s., Nicotiana sylvestris; S. t., Solanum tuberosum; V. f., Vicia faba; N. c., Neurospora crassa; B. t., Box taurus. Numbering refers (more ...)

In Isolated CMS Leaf Mitochondria, Oxygen Uptake with Glycine Is Dramatically Reduced, but Cyanide-Resistant Respiration Is Enhanced. The respiratory chain, with several alternative routes in plants (29–33), is schematized (Fig. 6). In addition to complex I (Fig. 6, [a]) inhibited by rotenone, at least three other NADH dehydrogenases have been described on inner (Fig. 6, [b]) and outer (Fig. 6, [c]) faces of internal membrane andon the outer face of external membrane (Fig. 6, [d]). Dehydrogenases channel electrons to ubiquinone, and [a]–[d] dehydrogenases potentially direct them to cytochrome oxidase. A second branching point occurs at ubiquinone, where electrons may pass either through the main cytochrome pathway or through AOX, which is insensitive to cyanide. Its involvement in the CMS trait has been suggested for various species (34–38).

Figure 6 Scheme of plant respiratory chain. Bold, main chain; I–IV, respiratory complexes; , coupled proton translocation; UQ, ubiquinone; [a]–[d], NADH dehydrogenation pathways. Arrows symbolize electron (more ...)

Complex I activity was estimated by oxidation of glycine, a priority substrate in green tissues (39, 40). NADH, oxidized at high rates by the various dehydrogenases (29, 31–33), andsuccinate, oxidized by complex II, were also used. Inhibition by rotenone enabled evaluation of the contribution of complex I to glycine andNADH oxidation.

Fig. 7 shows how, after stabilization after substrate injection, a first addition of ADP induced a phosphorylating “state 3” that shifted to “state 4” when ATP hydrolysis balances synthesis. This is visualized by acceleration, then slowing down of respiration, which result from the fall andthen rise of the proton gradient (Fig. 7, Inset). When an uncoupler was added, the proton gradient was fully dissipated, andthe chain functioned at its maximum rate. To avoid variable control of the redox chain by variable proton gradient, uncoupled rates are considered here.

Figure 7 Recorder traces of respiration measured with purified mitochondria from parental T andmutant CMS II leaves (same with CMS I). Mitochondrial proteins: 150 μg in 3 ml of the assay medium [4 mM glycine/500 μM ADP (more ...)

Table 2 summarizes the results. Succinate respiration of CMS andT mitochondria were close to each other. In contrast, there was a dramatic (two–thirds) decrease in both mutants with glycine. In addition, although in T mitochondria almost half of glycine oxidation passed through complex I, judging from rotenone inhibition, the CMS residual activity was much less (≈20%) sensitive to it.

Table 2 Respiration activities of the T line and CMS mutants of N.sylvestris

At variance with glycine, NADH respiration was 2- to 3-fold higher in mutants than in T; it was always well coupled (Fig. 7, Inset), andsome inhibition by rotenone was observed, suggesting partial contribution of complex I. The high rates of NADH oxidation indicated that respiration with glycine or succinate was likely limited by the initial dehydrogenation step, not by the redox chain.

Surprisingly, the cyanide-resistant oxygen uptake, low (succinate) or undetectable (glycine, NADH) in isolated mitochondria of fertile plants, was significantly higher in mutants, with succinate (up to 35% of total respiration in CMS I) andeven glycine (Table 2).

We evaluated AOX protein expression by immunodetection. Antibodies raised in Sauromatum guttatum (17) recognize AOX from other species, andthe corresponding cDNA sequences has been cloned (41). In several nonthermogenic plants, a dimeric enzyme (36-kDa monomer) was associated with AOX activity. Recently, a multigene family for the AOX was characterized in soybean (42). In N. sylvestris, the weak 36-kDa signal seen on Western blots of T leaf mitochondrial proteins, using the AOX antiserum, was noticeably increased in both CMS (Fig. 8).

Figure 8 Western blots of T andCMS AOX. Immunodetection with antisera against S. guttatum AOX andpotato formate dehydrogenase (FDH) serving as control. Lanes: 1, T; 2, CMS I; 3, CMS II. The T lane was overloaded to detect the weak AOX signal (more ...)

Until now, only two phenotypes have been associated with mtDNA reorganization in higher plants: (i) the widely distributed cytoplasmic male sterility trait that generally involves only male gametophytic development (43, 44), and (ii) the NCS variegated trait in maize (45). The N. sylvestris CMS I and II plants represent a new class of mitochondrial mutant.

CMS I and II Display Alterations in Vegetative and Reproductive Organs That Are Stably Transmitted Through Sexual and Somatic Generation. Some fertility could be restored by raising light, suggesting interactions between mitochondrial and chloroplast metabolism, which have been seen already for the maize NCS2 mutant (46). However, CMS I and II do not exhibit leaf variegation as NCS. This should result from the near homoplasmy of their genomes; indeed, the parental mtDNA fragments can only be detected by PCR amplification experiments (ref. 12 and unpublished work). No differences in fragment stoichiometry were found in leaf, bud, or pollen. On the contrary, NCS mutants are heteroplasmic; parental fragments can be detected by Southern hybridization, and plants give rise to both mutant and normal offsprings (9). In CMS, no phenotypic or molecular segregation was seen over eight sexual generations or after protoplast culture, suggesting that every CMS cell possesses a similar mtDNA organization. Heteroplasmy of NCS was proposed to account for their viability. In contrast, N. sylvestris CMS deletions, although associated with severe dysfunction, are compatible with plant survival.

CMS Mitochondria Are Deficient in Mitochondria and Nucleus-Encoded Complex I Subunits Other than NAD7. Even though the nad9 gene was not deleted and its transcription was unchanged (unpublished work), no significant amount of NAD9 was detected by Western blotting (Fig. 3). Furthermore, the nucleus-encoded, 38-kDa subunit was undetectable by two-dimensional SDS/PAGE of CMS mitochondrial proteins. The analyzed CMS I and II plants contained a near isogenic T nuclear genome because they have been maintained by backcrossing with the parental line. Therefore, the absence, or reduced level, of two complex I subunits other than NAD7 results from some change in the mitochondrial genome. Two mechanisms may intervene: altered control of gene expression or defects in complex I assembly with rapid proteolysis of the nonintegrated subunits. Both situations have been found in lower eukaryotes. Down-regulation of nuclear genes by the mitochondrial genome has been described in yeast “petite” mitochondrial deletion mutants (47), and assembly of complex I is impaired in the nuo21 disruption mutant of N. crassa (48). In this fungus, two assembly intermediates are formed: (i) a small, extrinsic subcomplex facing the matrix that consists of 13 nuclear-encoded subunits, including proteins of 49, 38 and 30 kDa; and (ii) a large, hydrophobic membraneous subcomplex that comprises all mitochondrial and the remaining nucleus-encoded subunits (16). An analogous structure was found in bovine heart mitochondria (2). In plants, the 49- and 30-kDa subunits are coded by mitochondrial genes nad7 and nad9 and, if the binary structure is as described above, N. sylvestris NAD7, NAD9, and 38-kDa subunits would belong to the extrinsic subcomplex. The purification of complex I from potato (25) and broad bean (26) mitochondria was recently reported, and its isolation from N. sylvestris T and CMS would help to clarify the role of missing subunits.

Complex I Activity Decreases While Functioning of Other NADH Dehydrogenases Increases in CMS. NAD7 is highly conserved among eukaryotes and prokaryotes, suggesting an important function (31). Thus, whatever the exact composition of complex I in CMS, its absence should be very deleterious. Indeed, glycine oxidation by purified CMS mitochondria dropped to less than one–third that of T controls (Table 2). A deficiency in glycine metabolism is unlikely because all implicated genes are nuclear-encoded, and, as we have stated already, CMS plants possess a near isogenic parental nuclear genome. A complex I dysfunction was also suggested in maize NCS2, deleted for nad4–nad7 sequences (9). The respiration remaining in CMS mitochondria with glycine (Table 2) may have two origins. Either defective complex I still functions, albeit slowly, or it is relayed by additional NADH dehydrogenase(s). The slight rotenone inhibition of this residual activity suggests that defective complex I partially functions, but probably most electrons pass through additional dehydrogenases.

Unlike succinate oxidation through complex II, which is comparable to T control, oxidation of exogenous NADH by CMS mitochondria is significantly faster. Moreover, it generates an important proton gradient (Fig. 7, Inset), involving complexes III and IV (via sites [c] and [d], Fig. 6). This suggests a partial compensation of complex I deficiency for respiration and phosphorylation. An analogous situation was described for the nuo51 mutant of Aspergillus niger that lacks functional complex I but maintains its respiration thanks to an alternative NADH pathway (49).

The mechanism of oxidation of cytosolic NADH by mitochondria in light is not well elucidated, and it has been proposed that, in some plant species, NADPH may be the source of redox equivalents (50). Consequently, a working hypothesis could be that the partial fertility of CMS mutants in normal light compared with low (Table 1; Fig. 2 D-G) is due to a sufficiently fast turnover of the photosynthetic chain to provide some extra ATP and NAD(P)H to cytosol and mitochondria, via the translocation of phosphorylated and reduced metabolites.

The Capacity of the AOX Pathway Is Increased in CMS Mitochondria. In the T line, a significant cyanide-resistant respiration was found only with succinate. In CMS mutants, it became detectable with glycine and was strongly enhanced with succinate, where it could reach 30% of total respiration (Table 2). Succinate is a ready source of pyruvate, so this is in agreement with the observation that AOX is activated by pyruvate (51). The present measurements are minimal estimates of AOX capacity because the best values are reached only by combining pyruvate and dithiothreitol (52, 53). In our CMS mutants, the larger capacity of the AOX route was associated also with more enzyme (Fig. 8), showing that it was not only due to regulatory effect. A similar correlation between activity and quantity of AOX was reported in tobacco cell cultures (54). Nevertheless, it is not known whether AOX capacity in vitro correlates with its engagement in vivo or whether the AOX competes more effectively with the cytochrome chain in CMS than in T. Isotope discrimination of oxygen uptake by both pathways would be a promising approach to this problem (55).

At variance with complexes III and IV, the AOX does not translocate protons and therefore is energetically futile (53). The role of the alternative pathway in nonthermogenic tissues is enigmatical. One possible function is to maintain the Krebs cycle if the cytochrome chain is defective. Overexpression of AOX was found in N. crassa chloramphenicol-poisoned cells (16) and in mitochondrial mutants (56) but not in the nuo21 complex I nuclear disruption mutant, in which the cytochrome pathway is unaffected (48). Unexpectedly, AOX activity is increased in N. sylvestris CMS although the majority of respiration still involves cytochrome oxidase (strong KCN effect; Table 2). However, in mitochondria from aged root slices of red beet, a correlation was noted (57) between enhanced AOX and NADH dehydrogenases activities, as in CMS (Table 2). It remains to be established if the increases in AOX capacity (Table 2) and protein content (Fig. 8) in CMS mutants have a physiological role. Our results clearly distinguish N. sylvestris mutants from other CMS mutants; the AOX pathway in the latter was unchanged or, at most, slightly inhibited when compared with fertile plants (34–38).

Conclusion. Although this cannot be strictly proven because of the absence of nuclear restorer genes, the sterility of N. sylvestris CMS I and II probably does not arise from the expression of abnormal chimeric mitochondrial genes, as in other CMS plants (44, 45), but rather from the deletion of coding sequences. Our recent analyses suggest that the only gene that is missing in these CMS mutants is nad7. Whatever the exact relationships between molecular and phenotypic alterations, male sterility in N. sylvestris CMS is clearly associated with respiratory dysfunction, a hypothesis long suggested (58) but scarcely supported by experimental data (59). This novel class of mutants should therefore be invaluable for studying the structure and functioning of complex I and the interactions between the different respiratory pathways in higher plants.

We thank Prof. H. Weiss (Universität Düsseldorf), Dr. J. M. Grienenberger (Institut de Biologie Moléculaire des Plantes, Strasbourg), Dr. T. E. Elthon (University of Nebraska), and Dr. R. Rémy (Institut de Biotechnologie des Plantes, Orsay) for the gift of antibodies against the N. crassa 49-kDa subunit, wheat NAD9, S. guttatum AOX, and potato formate dehydrogenase, respectively. Dr. M. Neuburger (CEA, Grenoble) is thanked for helpful initial advice, and we thank Dr. J. d’Alayer (Institut Pasteur, Paris) for internal microsequencing and Mrs. O. Robin for cytological work. This work was supported by the Centre National de la Recherche Scientifique.