The phenotypes of the flowers in both the sterile and maintainer lines were similar except for the number of pollen grains. The anthers in the sterile line hardly produced pollen grains, whereas the maintainer anthers were full of pollen grains (Figure 1).

To study the relationship of premature PCD on the abortion of pollen in HW203A (sterile line), we compared the characteristics of pollen development with those from the HW203B (maintainer line). So far, studies of pepper anther development have been performed in many types of CMS system; moreover, the onset of abnormal development was reported to be varied from one system to another [18,38–40]. Shifriss and Frankel [38] and Horner and Rogers [18] observed that the sterile line could form an abnormal tetrad but could not produce normal pollen for the development of tapetal cells. Kaul [39] and Bhargawa [40] both thought that pollen development was abnormal at the beginning of meiosis. It is necessary to examine the pollen-abortive stage of our CMS line to further analyze gene expression and enzyme activity. Most prior reports focused on the initial cytological abnormalities during anther development in the sterile line to understand the potential causes of the CMS phenotype in pepper because the tetrads cannot then produce a mononuclear microspore. For our study, we first performed a paraffin section of the anther to observe its development. All the flower buds were divided into eight grades (Figure 2) according to the morphology of the buds [41], such as the length and the width of the pepper florets.

In histologically stained anther sections that were observed by using light microscopy, the abnormal development of CMS pepper became evident at the tetrad stage, when the tapetal cells started to degenerate. At the tetrad stage (grade 4 and 5), the microspores in the sterile line appeared normal relative to those of the maintainer line, though the tapeta began to disappear from the sterile line at this stage, whereas they persisted in the maintainer line (Figure 3II). When the florets of the CMS lines had reached a length that corresponded to grades 6 and 7, the tapetal cells thoroughly degenerated and the central nucleus microspores were released from the tetrad. The microspores were compressed in the center of the locule by the enlarged middle layer cells (Figure 3III(A3)). In the maintainer line, the tapeta also disappeared, and the microspores were uniformly distributed in the locule (Figure 3III(B3)). At grade 8, the middle layer cells further enlarged in the sterile line, and the sterile microspores were degraded and adhered to the residuum of the tapetal cells and were crushed into a lump of condensed material by the enlarged middle layer (Figure 3IV(A4)). Mature pollen grains were produced in the maintainer line at the same stage, and the middle layer cells were degraded (Figure 3IV(B4)).

To find if premature programmed cell death (PCD) caused the precocious callose degradation that was noted by Binh et al.[42], we stained the microspore with aniline blue. As shown in Figure 4, the stained microspores of HW203A and B had practically no differences between them, which implied that the premature PCD had nothing to do with the callose.

Together, these results show that the tetrad in HW203A could release single nucleus microspores, whereas no normal pollen grains were produced in the end, just as described by Wang et al.[43]. Although HW203A can produce single nucleus microspores, the tapetal cells were degraded early at the start of meiosis and had nearly completed its process by the tetrad stage. During normal pollen development, the anther tapetum starts to degrade at the tetrad stage and completely finishes when mature pollen forms [32]; the whole process is called programmed cell death (PCD). The programmed degradation of tapetal cells provides nutrients for pollen development such as proteins, lipids and others. Finally, the tapetal cell remnants and liposome are released to form a coat around the pollen surface. Therefore, we inferred that the premature PCD in HW203A might lead to a lack of nutrients and energy that is needed for normal pollen development, especially at the later stage, as is already known in sunflower [8]. The nutrient and energy deficiency of pollen might result in a precocious degradation, which would lead to the abortion of pollen at the mononuclear microspore stage in HW203A. During the development of the microgametophyte, the callose plays an important role in protecting the microspore and supporting the formation of the pollen exine wall [44]. Similar roles for the callose wall at the tetrad and mononuclear microspore stages in HW203A and HW203B indicated that the premature PCD did not affect the callose wall formation, and perhaps the abortion of pollen in the sterile line has a weak relationship to callose.

Kim et al.[35] and Gulyas et al.[6] found that orf507 and the Ψatp6-2[36] are associated with pollen formation in the pepper CMS line. Within the CMS line, orf507, which is a novel ORF at the 3′-end of the cox II gene, is co-transcribed with cox II, and Ψatp6-2 is a pseudogene that is truncated at the C-terminal coding region when compared with the normal atp6-2 from the maintainer line. Kim [35] showed that orf456 was expressed in leaves and flowers, but it is unclear if the two genes could not or could only be slightly expressed in some certain tissues of the flower, and if they were temporally expressed. In our experiment, we analyzed the spatiotemporal expression patterns of orf507 andΨatp6-2 in the sterile and maintainer lines to find if the two genes were related to the abortion of pollen grains by using reverse transcription PCR (RT-PCR) and real time PCR (qRT-PCR).

Gene expression analysis was then performed to investigate the differences in expression levels between CMS and the maintainer lines at different anther development stages, including the MMC (microspore mother cell), tetrad and mononuclear microspore stages. The cDNAs were reverse-transcribed from the total RNA of HW203A and HW203B florets at the three different development stages listed above. Pollen abortion started at the tetrad stage. The results showed that no or very faint expression products of orf507 were detected in CMS flowers at the mononuclear microspore stage and that in the maintainer line, high-intensity expression was found at the tetrad and MMC stages (Figure 5). Gene expression levels in CMS line reached their highest peaks at the tetrad stage, which were significantly higher than at the mononuclear microspore stage; while they were a little higher than at the MMC stage, but with no significant difference. The high intensity expression before pollen abortion indicated that orf507 was transcriptionally down-regulated and might be related to the induction of abortion.

To ascertain whether the gene was expressed in specific plant tissues, cDNAs from the petals, anthers, stigmas, placentas and leaves were used. The orf507 transcripts were undetected or nearly so in the stigmas from the CMS line (Figure 6). The expression of orf507 in the other four tissues had no significant difference, and they were highest in the anthers and placentas. This pattern indicated that orf507 was not constitutively expressed in the flower and further demonstrated that the expression of this novel gene might cause injury to the anther or the microspore to some degree, with the highest expression in the anther.

PCR primers were designed to amplify ψatp6-2 in CMS plant. The primer set is expected to amplify atp6-2 in maintainer plant. The expression of atp6-2 in the maintainer plants was highest at the mononuclear microspore stage; however, it was highest at the MMC stage in the CMS line, and both were significantly higher than at other development stages. Meanwhile, the changing trend of gene expression in the CMS line was more or less contrary to what was found in the maintainer line. In concert with pollen development, the gene expression level was gradually decreased in the CMS, but it was dramatically increased in the fertile line after the tetrad stage (Figure 7). The opposing trends indicated that the Ψatp6-2 in CMS was down-regulated, and conversely, the atp6-2 was up-regulated in the maintainer line. Furthermore, spatial expression analysis showed that the pseudogene transcripts were not or were only weakly detected in the stigmas, petals and leaves both in the CMS and maintainer lines. However, the expression of atp6-2 in anthers was significantly higher than in other tissues from the maintainer line. In contrast, the expression of Ψatp6-2 in anther was relatively higher than other tissues in the CMS line and was slightly lower than in the placenta, as shown in Figure 8. We speculated that the atp6-2 might have something to do with reproductive development, such as pollen development, and that Ψatp6-2 might have a little opposite impact on pollen development when compared with atp6-2 in normal cytoplasm.

In summation, both the CMS and maintainer lines showed that the two genes of interest were not, or were only weakly expressed in the stigma, which might imply that they did not affect the stigma during pollen development. Moreover, temporal expression study showed that the two genes were down-regulated in the CMS line. In the maintainer line, orf507 was not expressed, whereas the opposite expression trend occurred for the atp6-2 gene. Meanwhile, orf507 and Ψatp6-2 or atp6-2 in normal cytoplasm were all highly expressed in the anther. Following these results, we speculated that orf507 andΨatp6-2 might be mainly expressed in the tapetal cell or microspore cell. Their exact tissue specificity remains to be seen and could be further explored by RNA gel blot.

CMS is closely connected with mitochondrial dysfunction in higher plants [45], especially with regards to energy metabolism. In rice, CMS-HL, orfH79 transcripts have been shown to reduce ATPase stability and activity [9], but in rice, the CMS-ZA and orfZ79 transcripts are associated with deficient NADH dehydrogenase activity but normal ATPase activity [10]. In a new orf220-type CMS line of Brassica juncea with down-regulated altered atpA gene, mitochondrial ATP synthesis decreased more than one-fold. [46]. However, few reports on mitochondrial enzyme activity in the CMS line of pepper have been published, although many novel mitochondrial genes that are caused by the rearrangement of DNA have been found and identified in this line. Given our current level of knowledge, we did not know if the novel genes were associated with a mitochondrial function for the corresponding enzyme. Kim et al.[35] transformed orf456 into Arabidopsis thaliana by using tapetum-specific promoter TA29, and they observed a resulting defect in pollen-like grains in the exine wall. They speculated that the expression of this novel gene might cause premature programmed cell death in tapetal cells, which provided lipids and protein components for the formation of the pollen exine wall, especially callose. In addition to high orf507 expression in buds before the tapetum degradation, we supposed that orf507 had a bearing on premature PCD through an effect on the energy supply.

To demonstrate the relevance of this gene to its relative enzyme activity, we analyzed cytochrome c oxidase and ATPase activity in the CMS and maintainer lines at different pollen development stages. Zhang et al.[9] demonstrated that orfH79 contributes to a reduction in ATPase activity when co-transcribed with atp6. In the pepper CMS line, orf507 is co-transcribed with cox II (which encodes a subunit of cytochrome c oxidase called COX II), just like the orfH79 in rice. As shown in Figure 9, the enzyme activity of cytochrome c oxidase in the maintainer line is minimal at the tetrad stage, which is significantly lower than at the MMC and mononuclear microspore stages. An opposite trend was found in the CMS line. When the tapetum started to degenerate at the tetrad stage, the enzyme activity of cytochrome c oxidase reached the matrix. All these findings implied that the sterile line was cytochrome c oxidase-deficient. By comparative analysis of orf507 gene expression level and cytochrome c oxidase activity, we speculated that the expression of orf507 may affect the normal function of cytochrome c oxidase.

The function of COX II, with its Cu_A binuclear center, is to accept an electron from reduced cyt c and transfer it to cytochrome a in the respiratory chain [24]. It was hypothesized that ORF507 might interfere with the electron transmission from reduced cyt c to cyt a, or reduce the stability of the enzyme by affecting the interaction of COX II with other subunits of cytochrome c oxidase, such as the binding of COX II to COX I. Nevertheless, further evidence is still needed to define the relationship between this gene and cytochrome c oxidase. In the future, we can demonstrate the interactions between the orf507 protein and cytochrome c oxidase using yeast two-hybrid assay.

In higher plants, F₁F_o-ATP synthase produces the bulk of the ATP. The two F₁ and F₀ parts are coupled in cells, and the intact F₁F_o-ATP synthase functions as ATPase and ATP synthase as regulated by the H⁺ concentration [47]. ATP6 is a subunit of F₁, a water-soluble unit that contains the catalytic binding sites for ATP synthesis or hydrolysis [48,49]. This subunit plays an important role in the assembly of F_o with F₁ because ATP6 has five transmembrane domains with an N terminus on the periplasmic side of the plasma membrane and a C terminus in the matrix compartment [49]. It was reported that mutations of atp6, such as the atp6-orfH79 in Honglian rice and a single amino acid mutation in yeast mitochondria, could induce some interaction changes between ATP6 and other subunits of F₁F_o-ATP synthase; moreover, these interactions are needed to stabilize the assembly of F_o with F₁[50].

There are two copies of atp6 in pepper, from which the sequence of atp6-1 in the CMS line is the same as it is in the maintainer line. The other atp6-2 has a different sequence in the CMS line, with a truncated 3′ region (Ψatp6-2) and an intact sequence in its counterpart.

In a previous study, we showed that the transcription of Ψatp6-2 was down-regulated in different pollen development stages, whereas the atp6-2 was up-regulated in the maintainer line. With respect to the ATPase activity at the initial microspore mother cell stage, the ATPase activity in the maintainer line was the highest (F1) and was much higher than in the sterile line (S1). The enzyme activities in both the sterile and maintainer lines were significantly decreased after the MMC stage. After the tetrad stage, when the anther started to abort in the CMS line, the enzyme activity in the CMS line still gradually decreased, but it increased in the maintainer line. To analyze the trend as a whole, the up-regulated pattern of atp6-2 and the gradually decreased ATPase activity might indicate that the normal gene might be involved in the inhibition of ATPase activity. The changing ATPase activity that was down-regulated in the CMS line was similar to the gene expression of Ψatp6-2, meaning that Ψatp6-2 might increase the ATPase activity and promote ATP hydrolysis (Figure 10). These results showed that atp6-2 and Ψatp6-2 encoded different proteins with opposite effects on the function of ATPase. On the other hand, the changing ATPase activity happened after the beginning of the anther abortion of the CMS and maintainer lines, which may demonstrate that the mitochondrial F₁F_o-ATPase was aberrant in the CMS line, compared with in the maintainer line, and the dysfunction of the enzyme had something to do with the expression of Ψatp6-2.

The reverse ATP hydrolysis and synthesis of F₁F_o-ATP synthase is regulated by membrane electrochemical potential [25]. In yeast mitochondria F₁F_o-ATP synthase, the ATP6 of F_o forms the proton-conducting interface with the c9-15 ring. For functional F₁F_o-ATP synthase, ATP6 and subunits β and α consist of static subunits and the subunits γ, ɛ and c consist of rotating subunits. The reversible rotation of γ–ɛ–c subunits relative to the β and α catalytic subunits of F₁ takes place during ATP synthesis (clockwise) and ATP hydrolysis (counterclockwise). The ATP6 and subunit c work as a stator to anchor the catalytic subunits of F₁ to the membrane [51]. In view of our results, we supposed that ATP6-2 may be involved in switching ATP hydrolysis to ATP synthesis. The protein structure of F₁F_o-ATP synthase in the CMS line may have been changed by the expression of Ψatp6-2, and the changed structure may inhibit the coupled enzyme when switching to a clockwise direction to perform ATP synthesis, but still persists in hydrolyzing ATP. When compared with the increased ATPase hydrolysis activity in the maintainer line, the decreased ATPase activity following the tetrad stage may lead to insufficient ATP supplies for normal tetrad microspore or mononuclear microspore development, which ultimately results in the abortion of pollen in the CMS line. Nevertheless, more evidence is required to address this hypothesis. In view of the idea that ATP hydrolysis is not a simple reversible reaction of ATP synthesis from F₁F_o-ATP synthase, further demonstrations of ATP synthase activity between the CMS and maintainer lines will be performed in the future.

Pepper CMS line HW203A and its corresponding maintainer line HW203B, which are near-isogenic lines with similar nuclear genotypes and different cytoplasms, were provided by the Horticultural College of Northwest A&F University. Plants were grown in an unheated greenhouse during the summer, and the florets were harvested after flowering at different anther development stages.

The florets of fertile line HW203B and CMS line HW203A were divided into eight grades according to the length and width of the sepals and petals. The different floret grades were fixed, embedded, and sectioned using a procedure that was adapted from Peng et al.[41]. The paraffin section samples were observed under a light microscope. After the tetrad stage the microspore was stained with 0.02% aniline blue in 0.1 M PBS to observe the accumulation of callus around the microspore, which was examined with a fluorescence microscope under UV light.

The florets in the pollen mother cell, tetrad and single nuclear microspore stages were harvested, ground with a pestle in cold extraction buffer (0.4 M mannitol, 1 mM EDTA, 25 mM MOPS-KOH (pH 7.8), 10 mM KH₂PO₄, 0.1% BSA, 1% PVP 40 and 0.2% β-mercaptoethanol). The homogenate was filtered through a 200 mesh nylon strainer and then subjected to gradient centrifugation according to a procedure by Millar et al.[52]. The mitochondria pellets were resuspended in 1 mL of medium (0.4 M mannitol, 1 mM EDTA, 10 mM KH₂PO₄, 0.1% BSA). Protein yield was determined by UV-Vis spectroscopy.

To analyze gene-specific expression, total RNA was isolated from the florets in the pollen mother cell, tetrad and single nuclear microspore stages from different tissues including the petals, anthers, stigmas, placentas and leaves using the Trizol method and was then purified with DNase I. The mRNA from each sample was subjected to reverse transcription using the PrimeScript™ RT-PCR Kit (Takara Biotech., Shiga, Japan) by combining the use of random hexamers and oligo dT primer. The primers that were used for RT-PCR and qRT-PCR were designed as indicated in Table 1, whereas coxII was used as the reference gene [6]. The primers that were used for gene expression analysis of Ψatp6-2 and atp6-2, which is known here as ratp6-2, were designed according to the homologous sequences of the two genes. The qRT-PCR was performed using a SYBR Green Kit (Sigma, St. Louis, MO, USA). Reverse-transcription PCR (RT-PCR) was performed just like normal PCR. The PCR protocol was as follows: an initial step of 94 °C for 5 min, followed by 28 cycles of 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, followed by a final elongation step of 72 °C for 10 min. The PCR products were detected by electrophoresis on a 1.5% agarose gel.

The mitochondria pellets of florets at three different stages were lysed by freeze-thaw cycles with the addition of 0.05% Triton X-100 [32] and the lysate was used for the spectrophotometric analyses of enzyme activities. ATPase activity was estimated according to the method of Hans Bisswanger [53]. The activity of cytochrome c oxidase was analyzed as described by Prasad et al.[54], whereas the reaction mix contained 50 mM phosphate buffer, pH 7.0, 10 μM reduced cyt c (reduced with 10 mM sodium ascorbate) and 10 μL of protein in a final volume of 1.5 mL.