Organization and Expression of the Mitochondrial Genome in the Nicotiana sylvestris CMSII Mutant

Corresponding author: Rosine De Paepe, Institut de Biotechnologie des Plantes, Bât 630, Université Paris-Sud, 91405 Orsay Cedex, France., depaepe{at}sidonie.ibp.u-psud.fr (E-mail).

Communicating editor: K. J. NEWTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Previous analyses suggested that the Nicotiana sylvestris CMSII mutant carried a large deletion in its mitochondrial genome. Here, we show by cosmid mapping that the deletion is 60 kb in length and contains several mitochondrial genes or ORFs, including the complex I nad7 gene. However, due to the presence of large duplications in the progenitor mitochondrial genome, the only unique gene that appears to be deleted is nad7. RNA gel blot data confirm the absence of nad7 expression, strongly suggesting that the molecular basis for the CMSII abnormal phenotype, poor growth and male sterility, is the altered complex I structure. The CMSII mitochondrial genome appears to consist essentially of one of two subgenomes resulting from recombination between direct short repeats. In the progenitor mitochondrial genome both recombination products are detected by PCR and, reciprocally, the parental fragments are detected at the substoichiometric level in the mutant. The CMSII mtDNA organization has been maintained through six sexual generations.

THE mitochondrial (mt) genome in higher plants is larger and more complex than in other eukaryotes (SCHUSTER and BRENNICKE 1994 ). According to cosmid mapping, it may be represented as a single circular map, the so-called master molecule (PALMER and SHIELDS 1984 ). This molecule often contains repeated sequences, some of which are able to recombine, generating subgenomic molecules in various amounts, depending on the balance between their recombination and replication rates (LONSDALE et al. 1988 ). However, neither the master genome nor the predicted subgenomic molecules have been physically identified in any species, and images from electron microscopy and pulsed-field electrophoresis suggest that most of the mt genome may consist of linear molecules (BACKERT et al. 1995 ; OLDENBURG and BENDICH 1996 ). A particular mt genome is thus characterized by both the nature and the relative stoichiometry of its molecules. From the molecular analysis of different maize cytotypes, SMALL et al. 1987 proposed that mtDNA evolution could result from changes in the proportions of substoichiometric molecules. However, experimental evidence for such an evolutionary mechanism is scarce, as mt genomes are in general stably inherited in a maternal fashion. Some cases of spontaneous changes have been described, for example, those associated with the reversion to fertility of cytoplasmic male sterile (CMS) plants in CMS-S maize (SCHARDL et al. 1985 ; SMALL et al. 1988 ) and CMS bean (MACKENZIE et al. 1988 ; MACKENZIE and CHASE 1990 ). Mitochondrial reorganizations have often been observed in in vitro cultured cells (OZIAS-AKINS et al. 1988 ; SHIRZADEGAN et al. 1991 ; KANAZAWA et al. 1994 ) and, more rarely, in the regenerated plants themselves (VITART et al. 1992 ; HARTMANN et al. 1994 ). A well-characterized system is the in vitro-induced reversion of the maize Texas CMS (ROTTMANN et al. 1987 ). A few examples of nuclear-directed changes of mt organization have been described, for example, in maize (NEWTON et al. 1990 ), bean (MACKENZIE et al. 1988 ; HE et al. 1995 ), and Arabidopsis (MARTINEZ-ZAPATER et al. 1992 ).

In most variant systems analyzed so far, recombination events have been involved in the observed mtDNA structural changes (BELLIARD et al. 1979 ; YOUNG and HANSON 1987 ; FAURON et al. 1990 ), suggesting recombination and differential replication of subgenomic molecules are indeed implicated in the evolution of mt genomes. Concerning the reversion of the CMS-T cytoplasm in maize, FAURON et al. 1990 proposed a model implicating two recombination events, one occurring frequently in planta, and a second rare event that occurred during the in vitro culture. We have previously described several mtDNA reorganizations in Nicotiana sylvestris mutants regenerated by protoplast culture from a fertile T line (LI et al. 1988 ; VITART et al. 1992 ). Two of these mt mutants, named CMSI and CMSII, regenerated from the same protoplast-derived callus, presented similar morphological abnormalities and male sterility. Restriction patterns and Southern hybridizations suggested the presence of deletions in the mtDNA of both mutants, more than 50 kb in CMSII and ~15 kb in CMSI (CHETRIT et al. 1992 ). The deletions were associated with two recombination events, rec1 found in both CMSI and CMSII, and rec2 specific to CMSI. For each recombination event, only one of the recombinant fragments could be detected by Southern hybridization, while the other recombinant fragment and parental fragments were undetectable. Target repeats of both recombinations have been characterized. Repeats of 102 bp (EMBL number X96741) were found at the recombining sites of rec1 (GUTIERRES et al. 1997A ; P. CHÉTRIT and S. GUTIERRES, unpublished results), while rec2 occurred between two short repeats of 65 bp (PLA et al. 1995 ). At last, several expression changes common to both CMSI and CMSII have been found. First, the complex expression pattern obtained using a probe overlapping one of the rec1 sites was reduced to a single signal in the mutants (CHETRIT et al. 1992 ). Second, the NAD7 polypeptide, a subunit of the first complex of the respiratory chain (complex I), could not be detected in either mutant (PLA et al. 1995 ; GUTIERRES et al. 1997B ). In addition, NAD9 and the nuclear-encoded 38-kD complex I subunit could not be detected, suggesting some misassembly of the complex. This complex I defect results in a dramatic decrease in its activity, and this has been postulated to account for reduced growth and sterility.

The aim of this article is to address the following points: (1) What is the extent and gene content of the mtDNA deletion in CMSII, and are the changes in the expression of nad7 solely responsible for the abnormal phenotype? (2) Is the rec1 recombination only involved in appearance of the deletion, did it specifically occur during the in vitro culture or did it preexist in the progenitor line, and what is the stability of the CMS II mtDNA organization through sexual generations?

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant material:
The fertile N. sylvestris T line is a botanical line provided by the Institut des Tabacs (SEITA, Bergerac, France) and maintained by self-pollination. The cytoplasmic male sterile (CMSII) line was regenerated after leaf mesophyll protoplast culture from the T line (LI et al. 1988 ) and maintained by backcrossing with T plants as male. Plants were grown in a greenhouse under a 16-hr photoperiod at 24° (day) and 17° (night).

Mitochondrial DNA isolation:
mtDNA was isolated from young leaves as described in CHETRIT et al. 1992 .

Mitochondrial RNA isolation and analysis:
Leaf mtRNAs were prepared from sucrose gradient-purified mitochondria (STERN and NEWTON 1985 ). RNA electrophoresis, Northern transfer, and hybridization were performed according to SAMBROOK et al. 1989 .

Mitochondrial gene probes:
The heterologous probes used were atpA (Oenothera, SCHUSTER et al. 1986 ), atp6 (maize, DEWEY et al. 1985 ), atp9 (N. sylvestris, PLA et al. 1991 ), coxI and coxIII (Oenothera, HIESEL et al. 1987 ), coxII (maize, FOX and LEAVER 1981 ), cob (maize, DAWSON et al. 1984 ), exons B, C, and E of nad1 (wheat, CHAPDELAINE and BONEN 1991 ), exon D of nad1 (wheat, HAOUAZINE et al. 1993 ), exon 3 of nad4 (wheat, LAMATTINA and GRIENENBERGER 1991 ), nad4L (Arabidopsis thaliana, BRANDT et al. 1992 ), all exons of nad5 (wheat, PEREIRA DE SOUZA et al. 1991 ), nad6 (wheat, HAOUAZINE et al. 1993 ), nad9 (wheat, LAMATTINA et al. 1993 ), rps12 (wheat, GUALBERTO et al. 1988 ), orf25 (maize, DEWEY et al. 1986 ), rrn26 (wheat, FALCONET et al. 1985 ), and rrn18 (wheat, FALCONET et al. 1984 ). Nicotiana sylvestris nad1/A, nad3, and nad7 coding sequences were cloned in the laboratory.

Cosmid libraries:
Cosmid libraries of T mtDNA were prepared using the SacI site of a modified pHC79 cosmid vector as previously described (CHETRIT et al. 1992 ). Two T cosmid libraries were used (clones named 1–400 and T1–T350). Cosmid DNAs were prepared according to BIRNBOIM and DOLY 1979 .

Colony hybridization and cosmid mapping:
Screening of cosmid libraries was performed according to GRUNSTEIN and HOGNESS 1975 , and clones were mapped using the SacI restriction enzyme (Boehringer Mannheim, Indianapolis).

DNA analysis:
DNA restriction, agarose gel electrophoresis, Southern transfer on Hybond N+ (Amersham, Arlington Heights, IL) membranes, and hybridization using radiolabeled probes were performed as described in SAMBROOK et al. 1989 . mtDNA restriction fragments used as probes for screening of cosmid libraries were purified from preparative agarose gels according to ZHU et al. 1985 and labeled using random priming kits (Boehringer).

cDNA probe:
Ten micrograms of T mtRNA was incubated at 37° for 30 min in 40 µl of a mixture containing 200 units of reverse transcriptase (Stratascript) with the appropriate 10x buffer, 4 mM pd(N)6 (random hexanucleotide primers provided by Pharmacia, Piscataway, NJ), 0.5 mM dATP, dGTP, and dTTP, 12 units of RNase Inhibitor (Promega, Madison, WI), and 5 µl of [-³²P]dCTP (3000 Ci/mmol, Amersham). After ethanol precipitation, the reaction product was used as a probe on Southern filters carrying SacI-restricted cosmid DNA.

Cloning and sequence analysis:
DNA fragments were subcloned into pBluescript KS+ vector (Stratagene, La Jolla, CA) as described in SAMBROOK et al. 1989 and sequenced using the T7 sequencing kit (Pharmacia).

PCR amplification:
PCR experiments were performed according to the following procedure: 20–50 ng of DNA template was mixed with 50 pmol of each primer, 0.2 mM each dNTP, one unit of Taq DNA polymerase (Appligene), and the Taq polymerase buffer in a total volume of 25 µl. Forty amplification cycles were performed, each cycle consisting of 1 min at 92°, 3 min at the primer annealing temperature, and 1 min at 72°. PCR products were electrophoresed in a 1% agarose gel, and Southern transfer and hybridization were performed as in SAMBROOK et al. 1989 . mtDNAs of plants corresponding to six sexual generations of the T and CMSII lines (1989–1995) were used as templates in the PCR experiments. Sequence of the oligonucleotides: O1-5' ACTAGTGAGAGGGCAAAAATT3'; O2-5' AACAGCTATGTACATTTTTTTTT3'; O3-5' ACCGCAGATGGTAGTCTGG3'; O4-5' CGTTAGCGGACCGACTTGAC3'; O5-5' ATTGAAAACGTCGAAGTAAG3'; O6-5' TGAGCAACTCAAGCG3'; O7-5' TATATAGGCTCGCTC3'.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Characterization of the CMSII mt deletion:
Previous results suggested that the reorganization of the CMSII mt genome was associated with a homologous recombination event named rec1, involving two SacI restriction fragments of 19 and 5.4 kb (CHETRIT et al. 1992 ). The 19- and 5.4-kb SacI fragments were undetectable by Southern hybridization in the CMSII mtDNA, whereas a recombinant 11.8-kb SacI fragment appeared. The reciprocal recombinant fragment, 12.6 kb in size, was detectable in neither the fertile nor the mutant line. In addition, six other SacI fragments were undetectable in CMSII. From these results, it was proposed that CMSII mtDNA carried a large deletion of about 50 kb, located between the two target sites of rec1 (CHETRIT et al. 1992 ). To confirm this model, we realized a partial physical map of the T mt genome by the overlapping cosmid method, using as initial probes the 5.4-kb SacI fragment and a 9-kb SacI fragment located near the 19-kb SacI fragment. The results of the walking experiments are presented in Figure 1A. Parental fragments of rec1 were found to be located 60 kb apart, delimiting a region named containing 15 SacI fragments. Several coding sequences are located between the rec1 target sites. First, the region contains the 12.3- and 2.9-kb SacI fragments previously found to carry the four exons of the cis-spliced nad7 gene coding for the NAD7 subunit of complex I (PLA et al. 1995 ). Second, an orf87-nad3-nad1/A cluster is located in the 5.4-kb SacI fragment (GUTIERRES et al. 1997A ), the rec1 target site being located within the nad3-nad1/A intergenic region (Figure 1A). Third, the 4.3- and 1.6-kb fragments showed homology with rrn26 and with exon B of nad1.

View larger version (84K): In this window In a new window Download PPT slide   Figure 1. Cosmid mapping of the region located between the two parental fragments of rec1. (A) Partial map obtained after screening the fertile mtDNA cosmid libraries with the 5.4-, 9-, 1.7-, and 1.6-kb SacI fragments as probes. Horizontal arrows, 102-bp direct repeats involved in rec1; black boxes, potentially transcribed regions determined using a T cDNA probe; gray boxes, regions homologous to known mt genes; A, B, and C, SacI subfragments homologous to expressed regions of unknown function. (B) Hybridization of T48, T75, T162, and T158 cosmid DNA on fertile (T) and CMSII (SII) mtDNA SacI restriction patterns.

To determine if the region was deleted in CMSII, cosmids containing all fragments spanning this region were used as probes on T and mutant Southern blots (Figure 1B). The six fragments located between the 19-kb SacI fragment and the 5.1-kb SacI fragments in the T75 cosmid (0.6, 2.9, 12.3, 9, 6.8, and 4.8 kb; 36 kb total) were undetectable in CMSII. In contrast, the nine adjacent fragments located between the 5.1-kb and the 5.4-kb fragments (5.1, 1.7, 1.5, 2, 2.8, 4.3, 1.6, 1.4, and 3.5; 24 kb total) were visible in CMSII using T75, T162, and T158 cosmids as probes. Screening T libraries with the 1.7-kb (from T75) and the 1.6-kb (from T158) fragments as probes showed that these fragments are in fact present in the T genome (Figure 2) in several cosmid organizations: (1) the 1.6-kb probe detected two additional groups of overlapping cosmids also containing the 2- to 3.5-kb fragments (Figure 2A). A first group carried nad3 closely associated to rps12 as already described in GUTIERRES et al. 1997A , while fragments of the other group showed homology with coxII and atp6 (Figure 2B); (2) using the 1.7-kb probe, we found a single additional series of cosmids, containing a region homologous to cob (Figure 2C).

View larger version (28K): In this window In a new window Download PPT slide   Figure 2. Additional groups of overlapping T cosmids containing the 1.6- (A and B) and 1.7-kb (C) fragments (white boxes). Fragments of the region detectable in the CMSII mtDNA (24 kb total) are boxed. Gray boxes represent known mt genes or exons located within these cosmid groups.

All three alignments were found in CMSII cosmid libraries (data not shown), which could explain presence of the corresponding fragments in CMSII mtDNA, even if the whole region is deleted. Such a deletion can be explained by considering the fact that, in the progenitor mtDNA, the 102-bp rec1 repeats (arrows in Figure 1A) are in direct orientation on a molecule called P. Recombination between these repeats would give rise to two recombinant subgenomic molecules, the C molecule carrying the 11.8-kb SacI recombinant fragment and the molecule carrying the 12.6-kb SacI recombinant fragment (Figure 3). The progenitor T line would contain only the P molecule, while the CMSII would only contain the C molecule. The molecule, 72.6 kb in size and carrying the region, would be absent in both lines. According to this model, in CMSII, the progenitor P genome has been replaced by the C subgenome lacking the region. However, due to the presence of several repeated regions in the T mtDNA, the nad7 gene only appears to be lost in CMSII.

View larger version (22K): In this window In a new window Download PPT slide   Figure 3. Model of the rec1 recombination. Two subgenomic molecules, C and D, are generated through the rec1 recombination between repeats in direct orientation (horizontal arrows in Figure 1A) in the P molecule. The black regions are the SacI fragments involved in rec1: the parental 19- and 5.4-kb fragments in P and the recombinant 11.8- and 12.6-kb SacI fragments in C and D, respectively. The T mt genome contains only the P molecule, while the CMSII mt genome contains only the C molecule, in which the region located between the 19- and 5.4-kb SacI parental fragments is lacking. Gray areas, SacI fragments of the region present in unique copy in the P molecule. Striped areas, repeated fragments. White boxes, known genes carried by these fragments. Arrowheads, location of primers designed for amplification of rec1 parental and recombinant fragments in PCR experiments (see Figure 6).

Transcriptional analysis of the CMSII mt genome:
Results of Northern experiments are shown in Figure 4. As previously found in CMSI (PLA et al. 1995 ), nad7 transcripts are undetectable even after overexposure in CMSII. In contrast, although the orf87-nad3-nad1/A cotranscripts present in the T line (GUTIERRES et al. 1997A ) are lost following deletion of the orf87-nad3 region, these last sequences are still transcribed from the orf87-nad3-rps12 cluster, and transcripts corresponding to nad1/A, which is not part of the deletion, remain present (data not shown). Similarly, as expected from their repetition in T mtDNA regions maintained in CMSII, rrn26 and exon B of nad1 gave the same patterns in both T and CMSII. We checked for other expressed sequences in the region deleted in CMSII. Southern hybridizations using a cDNA probe corresponding to total T mtRNAs revealed that in addition to the above fragments, the 2.8-, 2-, 9-, and 19-kb SacI fragments contained transcribed regions (data not shown). The adjacent fragments of 2 and 2.8 kb gave the same major signals when used as probes on T transcripts, suggesting that both fragments are overlapped by a single unidentified ORF called C (Figure 1A), the sequence of which was not determined. As expected from the presence of these two fragments in CMSII (Figure 2), no transcription changes as compared to T could be detected using them as probes (Figure 4). The transcribed regions carried by the 19- and 9-kb fragments were located in two subfragments called A (2-kb ApaI/ClaI subfragment) and B (1.8-kb BamHI/HindIII subfragment), respectively. The genomic sequences (GenBank accession numbers 72588 and 72609) are interrupted by several stop codons and show no significant similarity with any known mt genes or ORFs. Although the 19- and 9-kb SacI fragments are undetectable in the mutant mtDNA, the complex transcription patterns obtained using subfragments A and B as probes did not significantly differ between T and CMSII (Figure 4), suggesting total or partial duplications of the A and B regions in T and their presence in CMSII. This hypothesis was further supported by the use of A and B as probes on T Southern blots. Two or three signals were revealed with all three restriction enzymes used, the stronger signals corresponding to the 19- and 9-kb SacI fragments (Figure 5). In conclusion, transcriptional changes corresponding only to nad sequences could be shown in CMSII.

View larger version (45K): In this window In a new window Download PPT slide   Figure 4. Transcription analysis of the region: Comparison between T and CMSII Northern profiles using mt genes and ORFs found in the region as probes: nad7, rrn26, nad1/B, and A, B, and C subregions.

View larger version (59K): In this window In a new window Download PPT slide   Figure 5. Southern hybridizations of A (2-kb ApaI/ClaI subfragment) and B (1.8-kb BamHI/HindIII subfragment) on T mtDNA digested by SacI, BamHI, and EcoRI restriction enzymes.

PCR analysis of parental and recombinant fragments of rec1:
The origin of the mt genome reorganization in CMSII was investigated. Two main possibilities were considered: either the rec1 event was induced by the in vitro culture or it preexisted in the parent line, but at a very low level as the recombinant fragments were not detectable by Southern hybridization. Reciprocally, we wondered whether the parental fragments of rec1 could be present at a substoichiometric level in CMSII. In order to answer these questions, PCR experiments were performed with the two lines using primers specific to parental and recombinant fragments (see MATERIALS AND METHODS; Figure 3 and Figure 6). The stability of both T and CMSII mtDNA organizations was tested over six sexual generations, with several plants per generation. Results from three of the generations tested are presented in Figure 6.

View larger version (35K): In this window In a new window Download PPT slide   Figure 6. PCR amplification of parental and recombinant fragments in T and CMSII mtDNAs. PCR experiments were performed on different templates using primers specific to (A) the 5.4- and 19-kb rec1 parental fragments (primers O1–O2 and primers O3–O4, respectively), to (B) the 11.8- and 12.6-kb recombinant fragments (primers O2–O3 and primers O1–O5, respectively), and to (C) the 12.3-kb fragment undetectable by hybridization in the CMSII line (primers O6–O7). Primers are also located on the map in Figure 3. The blotted amplification products were hybridized with the corresponding restriction fragment as a probe, except for the 12.6-kb recombinant fragment for which the 5.4-kb parental fragment was used as probe. Schematic position of the primers in the restriction fragments and expected sizes of PCR products are indicated. PCR experiments were carried out using mtDNAs of several generations of the T (years 1989, 1993, and 1995) and CMSII lines (years 1989, 1992, and 1995) as templates. -DNA, negative control performed without DNA template in the reaction. Negative and positive controls were also carried out using the parental and/or recombinant fragments as templates: the 5.4-kb parental fragment (5.4), the T48 cosmid containing the 19-kb fragment (T48), the CMSII C8J cosmid carrying the 11.8-kb fragment (C8J, CHETRIT et al. 1992 ), and the 12.3-kb fragment (12.3). L, 1-kb Ladder (GIBCO BRL, Gaithersburg, MD).

In all T mtDNAs analyzed, regions specific to 11.8- and 12.6-kb recombinant fragments were amplified by PCR (Figure 6B). Reciprocally, regions specific to both parental fragments and the 12.6-kb recombinant fragment were found in CMSII whatever the generation tested (Figure 6A and Figure B). No amplification products were obtained in any control experiments. As PCR reactions are not quantitative in our conditions, the variation observed in the different amplifications was not representative of differences between samples, and no changes through sexual generations could be demonstrated in either cytotype. To check for the presence of the nad7 gene at a substoichiometric level in CMSII, additional PCR experiments were carried out using primers designed for the 12.3-kb SacI fragment (Figure 6C). A PCR product of the expected size was obtained, strongly suggesting that the nad7 gene is maintained at a low level in CMSII, as was previously found for CMSI (PLA et al. 1995 ). This was confirmed by RT-PCR experiments, revealing the presence of edited nad7 mature transcripts in either CMS (PLA et al. 1995 ; P. CHÉTRIT and S. GUTIERRES, unpublished results).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The aim of this work was to characterize the alterations in structure and expression of the CMSII mt genome. The results presented here allow us to propose a model for the transition from T to CMSII mtDNA organization, involving rec1 as a unique recombination event and resulting in expression changes of only nad genes.

Structural organization of the CMSII mt genome: a subgenomic molecule has emerged as the new "master" molecule in the mutant:
Cosmid mapping showed that the two parental SacI fragments involved in rec1, 19 and 5.4 kb in size, respectively, are physically linked and separated by a 60-kb region called (Figure 1). The region contains two types of SacI fragments: some are detectable in CMSII (24 kb total) and others are not (36 kb total). We showed that the fragments still found in CMSII are in fact present in at least two copies in T mtDNA. These results strongly suggest that the whole region is deleted in the mutant mtDNA. This deletion can be explained by considering the fact that in the progenitor mtDNA, the rec1 repeats are in direct orientation in a molecule called P (Figure 1 and Figure 3). Recombination between these repeats would give rise to two recombinant subgenomic molecules called C and . According to this model, in CMSII, the progenitor P genome would be replaced by the C subgenome in which the region is lacking.

Whether the whole mt genetic information of N. sylvestris is carried by the P molecule or by a putative multipartite structure generated by recombining repeats is not known. Such uncertainty is typical for all plant mt genomes, even when they have been fully mapped. For instance, YAMATO et al. 1992 suggested that the rice mt genome may be represented by two circular molecules instead of a master circle carrying a large duplication, and in Petunia, it was proposed that atp9 and pcf, the chimeric gene associated with sterility, could be carried by two different molecules (FOLKERTS and HANSON 1991 ). In N. sylvestris, three long repeats, sharing 12 kb in common and containing sequence homology to rrn26 and nad1/B, are present, but no recombinant borders were found. Short repeats involved in the emergence of new mtDNA organizations after protoplast culture were found to be located close to the rec1 target sites: (1) a 404-bp repeat involved in the rec3 recombination was found in the so-called U organization (VITART et al. 1992 ) at ~1 kb from rec1 in the 19-kb SacI fragment; (2) a 65-bp repeat involved in the rec2 event specific for the CMSI organization (PLA et al. 1995 ) is carried by the 12.3-kb SacI fragment. A short motif of 6 bp (TGCAAG) is shared by the three repeats, but whether this sequence could be the target site for recombination is not known. A 6-bp motif has also been suggested to be involved in homologous recombination in maize (NEWTON et al. 1990 ). Whatever the case, it appears that in N. sylvestris several sets of short repeats exist that are potentially able to produce new mtDNA organizations after in vitro culture.

Origin and stability of the CMSII mt genome:
Southern and cosmid analyses suggested that the region is completely absent in CMSII mtDNA. However, PCR experiments revealed the presence of rec1 parental fragments in the mutant and both recombinant fragments in the T line. Although PCR amplifications are sometimes artefactual as recombinant fragments may be produced in vitro ("jumping PCR"), the fact that rec1 parental and recombinant fragments were systematically found in different experiments strongly favors the hypothesis that they are also present in vivo. Presence of the region at the substoichiometric level in CMSII could also be deduced from two other experiments: (1) a PCR product of the expected size was obtained using primers designed for the 12.3-kb SacI fragment carrying the nad7 gene, and (2) RT-PCR experiments revealed the presence of edited mature nad7 transcripts in either CMS mutant. However, such transcripts are at levels too low to produce detectable amounts of the NAD7 polypeptide (PLA et al. 1995 ; P. CHÉTRIT and S. GUTIERRES, unpublished results).

Taken together, these results strongly suggest that all parental and recombinant molecules involved in rec1 are present in T and CMSII, the two genomes differing only in the proportions of these molecules. SMALL et al. 1987 originally proposed that stoichiometric variations could be a mechanism for plant mt genome evolution. This has since been confirmed by experimental situations generally also associated with in vitro culture (SHIRZADEGAN et al. 1991 ; HANSON and FOLKERTS 1992 ; HARTMANN et al. 1994 ; KANAZAWA et al. 1994 ). However, to our knowledge, a symmetrical situation as described here, where both parental and recombinant fragments are present in wild-type and mutant plants, has not been reported yet, except in the N. sylvestris U mt organization generated by rec3, where substoichiometric fragments are detectable by Southern hybridization (VITART et al. 1992 ). In the other cases described to date, only one of the recombinant fragments could be found, even when PCR experiments were performed (HARTMANN et al. 1994 ; KANAZAWA et al. 1994 ).

The question remains as to how such sudden changes in molecule stoichiometry may occur during culture and be maintained in the regenerated plants. This may result from either higher recombination rates or preferential replication of certain subgenomic molecules, or both mechanisms. It is likely that the P molecules are not yet the major replicating form in CMSII, as they are present in less than 1% of the normal stoichiometry, as judged by reconstruction experiments (data not shown). Differences in rates of replication of mt molecules have been suggested to account for the reversion to fertility in CMS bean (JANSKA and MACKENZIE 1993 ). In this species, loss or maintenance of the pvs sequence associated with CMS is under the control of the Rf gene (HE et al. 1995 ), and in A. thaliana, a mutation at the CHM locus promotes mtDNA reorganizations (MARTINEZ-ZAPATER et al. 1992 ). A still unidentified nuclear mutation could account for mt changes in the CMSII protoplast-derived plant (a particularly high nuclear mutation rate was observed in protoplast-derived plants. PRAT 1983 ; DE PAEPE et al. 1990 ), but not for the maintenance of the mutant organization during sexual reproduction. Indeed, CMS plants were backcrossed with the T line as male, and after six generations they should possess a near isogenic wild-type nuclear genome. A possible alternative would be the presence of an unidentified mtDNA mutation, either preventing efficient replication of the P molecule or favoring that of the C subgenome. Whatever the case, the new equilibrium has been maintained through sexual reproduction. No phenotypic reversion was observed among the hundreds of CMSII plants examined, none of them presented heterogeneous leaf sectors or produced fertile flowers. Several dozen of these plants (on average a dozen per generation) were chosen at random for Southern analysis: the parental fragments were never detected in any experiment, whatever the tissue analyzed (leaf, bud, or anthers) or the stage of plant development, from seedlings to flowering plants. In addition, several plants from each generation were analyzed by PCR; regions specific to parental fragments were amplified in all samples analyzed. Thus, as substoichiometric fragments never disappeared (PCR experiments) nor became detectable by Southern hybridization, they must be maintained in the same average range, less than 1% of the normal stoichiometry as judged by reconstruction experiments (data not shown). Given that some plant cells contain only a few hundred mt genomes (LAMPPA and BENDICH 1984 ), one would expect substoichiometric molecules to be rapidly lost from some cells after a few divisions, assuming random segregation of mtDNA between daughter cells. Female gametes derived from these fully homoplasmic cells would give rise to offspring free of the substoichiometric molecules, which was never observed, in contrast with the segregation observed in maize NCS plants (NEWTON et al. 1989 ). These observations strongly suggest that substoichiometric molecules are continuously produced by recombination (and then the recombinants greatly outnumber parentals in mutant CMSII).

The only unique gene that is deleted in CMSII mtDNA is nad7:
Results presented here show that the region deleted in CMSII contains the whole cis-spliced nad7 gene (PLA et al. 1995 ) and the orf87-nad3 part of an orf87-nad3-nad1/A cotranscription unit carried by the 5.4-kb SacI fragment (GUTIERRES et al. 1997A ). As expected, nad7 transcripts and the orf87-nad3-nad1/A cotranscripts are lacking, while a single signal likely to correspond to nad1/A alone remains present (CHETRIT et al. 1992 ; P. CHÉTRIT and S. GUTIERRES, unpublished results). All the other genes and putative ORFs, namely orf87-nad3, rrn26, and nad1/B exon present in the region are duplicated in T mtDNA regions maintained in CMSII, from which they are still expressed. Repetition of rrn26 is typical in higher plant mtDNA (FALCONET et al. 1985 ), while other duplications, such as that of nad3 and nad1/B, seem to be specific to Nicotiana (GUTIERRES et al. 1997A ). No other mt transcriptional changes have been found in CMSII using the mt probes listed in MATERIALS AND METHODS (results not shown). It is also unlikely that the deletion contains unique tRNAs, because CMS and T in organello mt synthesis patterns only present a limited number of differences, while absence of any tRNA would dramatically disturb translation. (LI et al. 1988 ; DE PAEPE et al. 1990 ).

All the available evidence thus supports the hypothesis that the only unique gene that is deleted in CMSII is nad7. This is also likely to be the case for the CMSI mutant, of which the 15-kb deletion is part of the region (CHETRIT et al. 1992 ) and contains the last two exons of nad7 (PLA et al. 1995 ). However, lack of NAD9 and of the 38-kD nuclear-encoded subunit of complex I was observed in both CMS, probably due to a misassembly of the complex (GUTIERRES et al. 1997B ), and recent results suggested that the NAD1 polypeptide is also lacking (P. CHÉTRIT and S. GUTIERRES, unpublished results). In plants, as in other organisms, complex I is composed of more than 30 subunits, most of them nuclearly encoded, and we do not know what subunits actually remain in either CMS mutant. It is possible that post-transcriptional and/or post-translational changes affecting other gene products, either nuclear or mitochondrial, may occur as a secondary effect of the initial mt mutation.

Whatever the exact composition of complex I in CMS mutants, its activity is dramatically decreased, if not completely absent, and this is likely to be the basis for the abnormal phenotype (GUTIERRES et al. 1997B ). Similar situations have been documented in fungi, where absence of the 51-kD subunit, the NADH-binding subunit (FECKE et al. 1994 ), or of the 21-kD subunit involved in complex I assembly (NEHLS et al. 1992 ; VIEIRA DA SILVA et al. 1996 ), are compatible with cell survival, although associated with respiration defects and reduced growth. Complex I mt mutations are also well documented in humans, in which they are often associated with degenerative diseases (see reviews in WALLACE 1993 , WALLACE 1994 ). Such mutations are usually heteroplasmic, the severity of the defect being correlated with the proportion of mutant mt genomes as compared to normal ones. However, some LHON (Leber's hereditary optic neuropathy) mutations approach the homoplasmic stage. The N. sylvestris CMS are the first near homoplasmic complex I mutants to be described in higher plants. The only other plant complex I mutants are the NCS mitochondrial mutants of maize, which are thought to survive only thanks to their heteroplasmic state (NEWTON et al. 1989 ). In CMSII as in CMSI, it is possible that very low amounts of normal complex I are formed, due to the presence of subliminal P molecules. Accordingly, presence of nad7 transcripts has been evidenced by RT-PCR in CMSI (PLA et al. 1995 ). However, in our opinion, CMS survival could better be accounted for by the increased activity of the other NADH-dehydrogenase activities existing in plants (GUTIERRES et al. 1997B ). Such alternative respiration routes also exist in fungi and are probably involved in the survival of the Neurospora complex I mutants.

In summary, the deletion carried by the CMSII mtDNA, one of the largest described in higher plants in terms of genetic information, involves one single recombination event, in contrast to most other mt variants that often reveal complex and multiple rearrangements (HANSON and FOLKERTS 1992 ). Both N. sylvestris CMSI and CMSII mutants can be classified as loss-of-function mutants, in contrast to most other CMS that are thought to be gain-of-function mutants. Their survival seems to be assured thanks to two properties shared by other higher plant species: at the molecular level, redundance of mt genetic information, and at the physiological level, presence of alternative respiration routes able to compensate, at least partially, for defects in the main pathway. However, N. sylvestris CMS plants present severe developmental defects, namely those concerning reproductive characters, and it is very unlikely that such mutants could survive in natural conditions. Selection of unfavorable new equilibria of mt molecules following in vitro cultures remains a mystery, showing that we are far from even a limited understanding of how mt genomes are maintained or evolve, in coordination (or conflict) with their nuclear partners.

	FOOTNOTES

¹ These authors have contributed equally to this work.

	ACKNOWLEDGMENTS

We thank P. H. Gouyon, I. Small, C. White, C. Small-Colas des Francs, B. Lejeune, and J. W. Tregear for helpful discussions and comments.

Manuscript received February 1, 1998; Accepted for publication June 16, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BACKERT, S., P. DÖRFEL, and T. BÖRNER, 1995  Investigation of plant organellar DNAs by pulsed-field gel electrophoresis. Curr. Genet. 28:390-399[Medline].

BELLIARD, G., F. VEDEL, and G. PELLETIER, 1979  Mitochondrial recombination in cytoplasmic hybrids of Nicotiana tabacum by protoplast fusion. Nature 281:401-403.

BIRNBOIM, H. C. and J. DOLY, 1979  A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucleic Acids Res. 7:1513-1523[Abstract/Free Full Text].

BRANDT, P., S. SUNKEL, M. UNSELD, A. BRENNICKE, and V. KNOOP, 1992  The nad 4L gene is encoded between exon c of nad 5 and orf 25 in the arabidopsis mitochondrial genome. Mol. Gen. Genet. 236:33-38[Medline].

CHAPDELAINE, Y. and L. BONEN, 1991  The wheat mitochondrial gene for subunit-I of the NADH dehydrogenase complex—a trans-splicing model for this gene-in-pieces. Cell 65:465-472[Medline].

CHÉTRIT, P., R. RIOS, R. DE PAEPE, V. VITART, and S. GUTIERRES et al., 1992  Cytoplasmic male sterility is associated with large deletions in the mitochondrial DNA of two Nicotiana sylvestris protoclones. Curr. Genet. 21:131-137[Medline].

DAWSON, A. J., V. P. JONES, and C. J. LEAVER, 1984  The apocytochrome b gene in maize mitochondria does not contain introns and is preceded by potential ribosome binding site. EMBO J. 3:2107-2113[Medline].

DE PAEPE, R., P. CHÉTRIT, V. VITART, F. AMBARD-BRETTEVILLE, and D. PRAT et al., 1990  Several nuclear genes control both male sterile sterility and mitochondrial protein synthesis in Nicotiana sylvestris protoclones. Mol. Gen. Genet. 222:206-210[Medline].

DEWEY, R. E., C. S. LEVINGS, III, and D. H. TIMOTHY, 1985  Nucleotide sequence of ATPase subunit 6 gene of maize mitochondria. Plant Physiol. 79:914-919[Abstract/Free Full Text].

DEWEY, R. E., C. S. LEVINGS, III, and D. H. TIMOTHY, 1986  Novel recombinations in the maize mitochondrial genome produce a unique transcriptional unit in the texas male sterile cytoplasm. Cell 44:439-449[Medline].

FALCONET, D., B. LEJEUNE, F. QUETIER, and M. W. GRAY, 1984  Evidence for homologous recombination between repeated sequences containing 18S and 5S ribosomal RNA genes in wheat mitochondrial DNA. EMBO J. 3:297-302[Medline].

FALCONET, D., S. DELORME, B. LEJEUNE, M. SÉVIGNAC, and E. DELCHER et al., 1985  Wheat mitochondrial 26S ribosomal RNA has no intron and is present in multiple copies arising by recombination. Curr. Genet. 9:169-174.

FAURON, C. M. R., M. HAVLIK, and R. I. BRETTELL, 1990  The mitochondrial genome organization of a maize fertile CMS T revertant line is generated through two sets of repeats. Genetics 124:423-428[Abstract].

FECKE, W., V. D. SLED, T. OHNISHI, and H. WEISS, 1994  Disruption of the gene encoding the NADH-binding subunit of NADH: ubiquinone oxidoreductase in Neurospora crassa. Eur. J. Biochem. 220:551-558[Medline].

FOLKERTS, O. and M. R. HANSON, 1991  The male sterility-associated pcf gene and the normal atp9-1 gene in Petunia are located on different mitochondrial DNA molecules. Genetics 129:885-895[Abstract].

FOX, T. D. and C. J. LEAVER, 1981  The Zea mays mitochondrial gene coding cytochrome oxidase subunit II has an intervening sequence and does not contain TGA codons. Cell 26:315-323[Medline].

GRUNSTEIN, M. and D. HOGNESS, 1975  Colony hybridization: a method for the isolation of cloned DNAs that contain a specific gene. Proc. Natl. Acad. Sci. USA 72:3961-3965[Abstract/Free Full Text].

GUALBERTO, J. M., H. WINTZ, J. H. WEIL, and J. M. GRIENENBERGER, 1988  The genes coding for subunit 3 of NADH dehydrogenase and for ribosomal protein S12 are present in wheat and maize mitochondrial genomes and are co-transcribed. Mol. Gen. Genet. 215:118-127[Medline].

GUTIERRES, S., C. LELANDAIS, R. DE PAEPE, F. VEDEL, and P. CHÉTRIT, 1997a  A mitochondrial sub-stoichiometric orf87-nad3-nad1 exonA co-transcription unit present in solanaceae was amplified in the genus Nicotiana. Curr. Genet. 31:55-62[Medline].

GUTIERRES, S., M. SABAR, C. LELANDAIS, P. CHÉTRIT, and P. DIOLEZ et al., 1997b  Lack of mitochondrial and nuclear encoded subunits of Complex I and alteration of respiratory chain in Nicotiana sylvestris mitochondrial deletion mutants. Proc. Natl. Acad. Sci. USA 94:3436-3441[Abstract/Free Full Text].

HANSON, M. R. and O. FOLKERTS, 1992  Structure and function of the higher plant mitochondrial genome. Int. Rev. Cytol. 141:129-172.

HAOUAZINE, N., A. TAKVORIAN, M. F. JUBIER, M. FRANCOIS, and B. LEJEUNE, 1993  Nad6 gene and nad1 exon D are co-transcribed in wheat mitochondria. Curr. Genet. 24:533-538[Medline].

HARTMANN, C., H. RECIPON, M. F. JUBIER, C. VALON, and E. DELCHER-BESIN et al., 1994  Mitochondrial DNA variability detectable in a single wheat regenerant involves a rare recombinant event across a short repeat. Curr. Genet. 25:456-464[Medline].

HE, S., A. LYZNIK, and S. MACKENZIE, 1995  Pollen fertility restoration by nuclear gene Fr in CMS bean: nuclear-directed alteration of a mitochondrial population. Genetics 139:955-962[Abstract].

HIESEL, R., W. SCHOBEL, W. SCHUSTER, and A. BRENNICKE, 1987  The cytochrome oxidase subunit I and subunit III genes in Oenothera mitochondria are transcribed from identical promoter sequences. EMBO J. 6:29-34[Medline].

JANSKA, H. and S. A. MACKENZIE, 1993  Unusual mitochondrial genome organization in cytoplasmic male sterile common bean and the nature of cytoplasmic reversion to fertility. Genetics 135:869-879[Abstract].

KANAZAWA, A., N. TSUTSUMI, and A. HIRAI, 1994  Reversible changes in the composition of the population of mt DNAs during dedifferentiation and regeneration in tobacco. Genetics 138:865-870[Abstract].

LAMATTINA, L. and J. M. GRIENENBERGER, 1991  RNA editing of the transcript coding for subunit 4 of NADH dehydrogenase in wheat mitochondria: uneven distribution of the editing sites among the four exons. Nucleic Acids Res. 19:3275-3282[Abstract/Free Full Text].

LAMATTINA, L., D. GONZALEZ, J. GUALBERTO, and J. M. GRIENENBERGER, 1993  Higher plant mitochondria encode an homologue of the nuclear-encoded 30 kD subunit of bovine mitochondrial complex I. Eur. J. Biochem. 217:831-838[Medline].

LAMPPA, G. K. and J. A. BENDICH, 1984  Changes in mitochondrial DNA levels during development of pea (Pisum sativum L.). Planta 162:463-468.

LI, X. Q., P. CHÉTRIT, C. MATHIEU, F. VEDEL, and R. DE PAEPE et al., 1988  Regeneration of cytoplasmic male sterile protoclones of Nicotiana sylvestris with mitochondrial variations. Curr. Genet. 13:261-266.

LONSDALE, D., M. T. BREARS, T. P. HODGE, S. E. MELVILLE, and W. H. ROTTMANN, 1988  The plant mitochondrial genome: homologous recombination as a mechanism for generating heterogeneity. Phil. Trans. R. Soc. Lond. B 319:149-163.

MACKENZIE, S. A. and C. D. CHASE, 1990  Fertility restoration is associated with loss of a portion of the mitochondrial genome in cytoplasmic male sterile common bean. Plant Cell 2:905-912[Abstract/Free Full Text].

MACKENZIE, S. A., D. R. PRING, M. J. BASSET, and C. D. CHASE, 1988  Mitochondrial DNA rearrangement associated with fertility restoration and cytoplasmic reversion to fertility in cytoplasmic male sterile Phaseolus vulgaris L. Proc. Natl. Acad. Sci. USA 85:2714-2717[Abstract/Free Full Text].

MARTINEZ-ZAPATER, J. M., P. GIL, J. CAPEL, and C. SOMERVILLE, 1992  Mutations at the Arabidopsis CHM locus promote rearrangements of the mitochondrial genome. Plant Cell 4:889-899[Abstract/Free Full Text].

NEHLS, U., T. FRIEDRICH, A. SCHMIEDE, T. OHNISHI, and H. WEISS, 1992  Characterization of assembly intermediates of NADH:ubiquinone oxidoreductase (complex I) accumulated in Neurospora mitochondria by gene disruption. J. Mol. Biol. 227:1032-1042[Medline].

NEWTON, K. J., E. H. COE, S. GABAY-LAUGHNAN, and J. R. LAUGHNAN, 1989  Abnormal growth phenoytpes and mitochondrial mutants in maize. Maydica 34:291-296.

NEWTON, K. J., C. KNUDSEN, S. GABAY-LAUGHNAN, and J. R. LAUGHNAN, 1990  An abnormal growth mutant in maize has a defective mitochondrial cytochrome oxidase gene. Plant Cell 2:107-113[Abstract/Free Full Text].

OLDENBURG, D. J. and A. BENDICH, 1996  Size and structure of replicating mitochondrial DNA in cultured tobacco cells. Plant Cell 8:447-461[Abstract].

OZIAS-AKINS, P., Z. TABAEIZADEH, D. R. PRING, and I. K. VASIL, 1988  Preferential amplification of mitochondrial DNA fragments in somatic hybrids of the Graminae. Curr. Genet. 13:241-245.

PALMER, J. D. and C. R. SHIELDS, 1984  Tripartite structure of the Brassica campestris mitochondrial genome. Nature 307:411-415.

PEREIRA DE SOUZA, A., M. F. JUBIER, E. DELCHER, D. LANCELIN, and B. LEJEUNE, 1991  A transplicing model for the expression of the tripartite nad5 gene in wheat and maize mitochondria. Plant Cell 3:1363-1378[Abstract/Free Full Text].

PLA, M., C. MATHIEU, F. VEDEL, and J. DURAND, 1991  Oligomycin/venturicidin-resistance is due to a nuclear modification in a somatic hybrid between Nicotiana tabacum and a variant line of Nicotiana sylvestris. Plant Sci. 78:53-60.

PLA, M., C. MATHIEU, R. DE PAEPE, P. CHÉTRIT, and F. VEDEL, 1995  Deletion of the last two exons of the mitochondrial nad7 gene results in lack of the NAD7 polypeptide in a Nicotiana sylvestris CMS mutant. Mol. Gen. Genet. 248:79-88[Medline].

PRAT, D., 1983  Genetic variability induced in Nicotiana sylvestris by protoplast culture. Theor. Appl. Genet. 64:223-230.

ROTTMANN, W. H., T. BREARS, T. P. HODGE, and D. M. LONSDALE, 1987  A mitochondrial gene is lost via homologous recombination during reversion of CMS-T maize to fertility. EMBO J. 6:1541-1546[Medline].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SCHARDL, C. L., D. L. PRING, and D. M. LONSDALE, 1985  Mitochondrial DNA rearrangements associated with fertile revertants of S-type male sterile maize. Cell 43:361-368[Medline].

SCHUSTER, W. and A. BRENNICKE, 1994  The plant mitochondrial genome: physical structure, information content, RNA editing and gene migration to the nucleus. Annu. Rev. Plant Physiol. Plant Mol. Biol. 45:61-78.

SCHUSTER, W., R. HIESEL, P. G. ISAAC, C. J. LEAVER, and A. BRENNICKE, 1986  Transcript termini of messenger RNAs in higher plant mitochondria. Nucleic Acids Res. 14:5943-5954[Abstract/Free Full Text].

SHIRZADEGAN, M., J. D. PALMER, M. CHRISTY, and E. D. EARLE, 1991  Patterns of mitochondrial instability in Brassica campestris cultured cells. Plant Mol. Biol. 16:21-37[Medline].

SMALL, I., P. ISAAC, and J. C. LEAVER, 1987  Stoichiometric differences in DNA molecules containing the atpA gene suggest mechanisms for the mitochondrial genome diversity in maize. EMBO J. 6:865-869[Medline].

SMALL, I., E. D. EARLE, C. J. ESCOTE-CARLSON, S. GABAY-LAUGHNAN, and J. R. GABAY-LAUGHNAN et al., 1988  A comparison of cytoplasmic revertants to fertility from different CMS-S maize sources. Theor. Appl. Genet. 76:609-618.

STERN, D. B. and K. J. NEWTON, 1985  Mitochondrial gene expression in Cucurbitaceae: conserved and variable features. Curr. Genet. 9:395-405[Medline].

VIEIRA DA SILVA, M., P. S. ALVES, M. DUARTE, N. MOTA, and A. LOBO-DA-CUNHA et al., 1996  Disruption of the nuclear gene encoding the 20.8 kDa subunit of NADH:ubiquinone reductase of Neurospora mitochondria. Mol. Gen. Genet. 252:177-183[Medline].

VITART, V., R. DE PAEPE, C. MATHIEU, P. CHÉTRIT, and F. VEDEL, 1992  Amplification of substoichiometric recombinant mitochondrial DNA sequences in a nuclear, male sterile mutant regenerated from protoplast culture in Nicotiana sylvestris. Mol. Gen. Genet. 233:193-200[Medline].

WALLACE, D. C., 1993  Mitochondrial diseases: genotype versus phenotype. Trends Genet. 9:128-133[Medline].

WALLACE, D. C., 1994  Mitochondrial DNA mutations in diseases of energy metabolism. J. Bioenerg. Biomembr. 26:241-250[Medline].

YAMATO, K., Y. OGURA, T. KANEGAE, Y. YAMADA, and K. OHYAMA, 1992  Mitochondrial genome structure of rice suspension culture from cytoplasmic male sterile line (A-58CMS): reappraisal of the master circle. Theor. Appl. Genet. 83:279-288.

YOUNG, E. G. and M. R. HANSON, 1987  A fused-mitochondrial gene associated with cytoplasmic male sterility is developmentally regulated. Cell 50:41-49[Medline].

ZHU, K. W., W. KEMPENAERS, D. VAN DER STRACTEN, R. CONTRERAS, and W. FIERS, 1985  A method for fast and pure DNA elution from agarose gels by centrifugal filtration. Biotechnology 3:1014-1016.