GhMCS1, the Cotton Orthologue of Human GRIM-19, Is a Subunit of Mitochondrial Complex I and Associated with Cotton Fibre Growth

GRIM-19 (Gene associated with Retinoid-Interferon-induced Mortality 19) is a subunit of mitochondrial respiratory complex I in mammalian systems, and it has been demonstrated to be a multifunctional protein involved in the cell cycle, cell motility and innate immunity. However, little is known about the molecular functions of its homologues in plants. Here, we characterised GhMCS1, an orthologue of human GRIM-19 from cotton (Gossypium hirsutum L.), and found that it was essential for maintaining complex integrity and mitochondrial function in cotton. GhMCS1 was detected in various cotton tissues, with high levels expressed in developing fibres and flowers and lower levels in leaves, roots and ovules. In fibres at different developmental stages, GhMCS1 expression peaked at 5–15 days post anthesis (dpa) and then decreased at 20 dpa and diminished at 25 dpa. By Western blot analysis, GhMCS1 was observed to be localised to the mitochondria of cotton leaves and to colocalise with complex I. In Arabidopsis, GhMCS1 overexpression enhanced the assembly of complex I and thus respiratory activity, whereas the GhMCS1 homologue (At1g04630) knockdown mutants showed significantly decreased respiratory activities. Furthermore, the mutants presented with some phenotypic changes, such as smaller whole-plant architecture, poorly developed seeds and fewer trichomes. More importantly, in the cotton fibres, both the GhMCS1 transcript and protein levels were correlated with respiratory activity and fibre developmental phase. Our results suggest that GhMCS1, a functional ortholog of the human GRIM-19, is an essential subunit of mitochondrial complex I and is involved in cotton fibre development. The present data may deepen our knowledge on the potential roles of mitochondria in fibre morphogenesis.


Introduction
Mitochondria in plants, similar to other eukaryotes, are highly specialised and dynamic double membranous organelles that play a pivotal role in energy production through oxidative phosphorylation (OXPHOS) [1]. The OXPHOS system is localised to the mitochondrial inner membrane, and it is composed of five complexes (I to V) that work in concert to drive the aerobic synthesis of ATP [2]. Of these complexes, complex I or NADH:ubiquinone oxidoreductase (E.C. 1.6.5.3) is the first and largest enzyme of the OXPHOS system that contains over 40 different subunits in eukaryotes, including mammals and plants [3][4][5]. It catalyses the transfer of electrons from NADH to ubiquinone in a process coupled to proton transport across the mitochondrial inner membrane. Obviously, the maintenance of its integrity and function depends on the contribution of all of the proteins that are encoded by both the nuclear and mitochondrial genomes. It has been reported that its defect in complex I is the most common cause of mitochondrial disorders, leading to disruption of cell function and subsequent disease or even death [4,6]. Mutations have been found in both mitochondrial and nuclear genes, such as GRIM-19 (gene associated with retinoid-interferon-induced mortality -19), which leads to the complete abolishment of complex I assembly when deleted; additionally, its knockout causes early embryonic lethality in mice [4,7].
Strikingly, although GRIM-19 was initially identified as an interferon-β and retinoic acidinducible gene with apoptotic effects in human cancer cell lines [7], recent studies have demonstrated that GRIM-19 is a nuclear-encoded subunit of complex I that is critical for assembly and enzymatic activity of mitochondrial complex I in mice [8]. Further, it has been shown that GRIM-19 is involved in the cell cycle, cell motility and innate immunity in mammalian cells [9,10]. Therefore, it appears to be a multifunctional protein involved in cell death, maintenance of mitochondrial metabolism, growth regulation and innate immunity. Although this protein has been studied in humans and other mammals, as well as in some insects and fish [11,12], it has not been described in plants to date.
The cotton (Gossypium spp.) fibre consists of a single epidermal cell derived from the ovule, and it plays a significant role in the global textile industry [13][14][15]. After differentiation (prior to the day of anthesis), the emerged fibre cells on the surfaces of cotton ovules undergo rapid elongation [5-20 days post anthesis (dpa)], secondary cell wall synthesis (15-45 dpa) and dehydration (40-60 dpa), finally forming mature long fibres that are harvested from seeds and used in textile products [13,16,17]. These economically important single cells dramatically elongate after anthesis, typically reaching up to 30 mm in length in most commercially important species. Therefore, the fibre cells can serve as the excellent single-celled model system for studying cell differentiation and growth [13,18].
It has been well-established that various fibre-specific and fibre-enriched genes are involved in the different stages of cotton fibre development, such as cellulose synthase [19], endo-1,4-βglucanase [20], sucrose synthase [21], αand β-tubulin [22,23], reversibly glycosylated polypeptide [24], β-galactosidase [25], glucuronosyltransferase [26], actin [27], the MYB transcription factors [28,29], phosphatise and annexin [30,31]. Over the past decade, genome-wide transcriptomic and proteomic studies have increased the understanding of the molecular basis of fibre development [13,14,18,[32][33][34]. These studies have revealed that many important molecular processes, such as glycolysis, the tricarboxylic acid (TCA) cycle, redox homeostasis, cytoskeleton dynamics and anthocyanidin metabolism, are differentially regulated during fibre initiation and elongation. Among these processes, the TCA cycle, which is one of the three main stages of cellular respiration, occurs in the matrix of the mitochondrion, while OXPHOS is the final step of cellular respiration that results in the production of ATP molecules [4]. Thus, it is well known that OXPHOS is critical to the integrity of fibre mitochondrial function and even to the development and growth of cotton fibre cells. However, there are few reports of the potential influences of some components of the cotton fibre mitochondrial OXPHOS system on the integration of major fibre metabolism as well as fibre cell growth.
To investigate the function of the cotton orthologue of human GRIM-19, the full-length cDNA sequence of GhMCS1 was cloned from cotton fibre cells and characterised. GhMCS1 is highly expressed in developing fibres and flowers and exhibits a specifically temporal expression pattern correlating with fibre growth. Biochemical detection of GhMCS1 showed its mitochondrial localisation and key role in complex I assembly in cotton mitochondria. GhMCS1 expression was closely correlated with fibre mitochondrial activity and development, which indicates that GhMCS1, as a subunit of mitochondrial complex I, is essential for mitochondrial function and cotton fibre morphogenesis.

Plant materials
Upland cotton (Gossypium hirsutum L. cv. CRI35) was grown in the field under standard conditions at the Tsinghua University in China as described previously [35]. The seeds were kindly provided by the Cotton Research Institute, Chinese Academy of Agricultural Science. When cotton plants were in full bloom (approximately 90 days after planting), we collected different cotton tissues, including roots, stems, leaves, flowers, and intact ovules. Cotton fibers were harvested at 5,10,15,20,25,30,35, and 40 days post anthesis (dpa). All of these samples were immediately frozen in liquid nitrogen and then stored at -80°C until extraction.

Cloning of GhMCS1 gene
Genomic DNA was extracted from young cotton leaves with a DNeasy Plant Kit (Qiagen, Valencia, CA, USA). Total RNA was isolated from cotton seedlings and various tissues using the TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) following purification using an RNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA). For the samples of 30~40-dpa fibres, before purification with the kit, the crude RNA was precipitated in 1/3 volume 8M LiCl at 4°C overnight to remove the contaminants (eg. sugars, phenols, etc.) [37]. First-strand cDNA synthesis was performed with Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA, USA).
A rapid amplification of cDNA ends (RACE) system (SMART TM RACE cDNA Amplification Kit, Clontech, Palo Alto, USA) was used to isolate the GhMCS1 cDNA. The gene-specific primers P1 (5 0 -CTTCACACTCGCCATTCCT-3 0 ) and P2 (5 0 -GTCACTGTTGAGCTGTTTT CAC-3 0 ) were designed to target the known GhMCS1 sequence fragment and were used for the 5 0 -RACE and 3 0 -RACE assays, respectively. The PCR-amplified products were cloned into a pMD18-T vector (TakaRa, Tokyo, Japan) and sequenced. According to the RACE results, the primers FP1 (5 0 -ATGACTGAGGCAGTG ATAAG-3 0 ) and FP2 (5 0 -TCACCAAACTTCA GGCG-3 0 ) were designed and used to amplify the full-length cDNA with a cotton cDNA library as the template.

RNA analysis
For northern blot, 30 μg of total RNA were size fractionated on 1.2% agarose MOPS-formaldehyde gels and then transferred to Hybond-N + nylon membrane (Amersham, Piscataway, NJ, USA) and hybridised with GhMCS1 full-length cDNA labelled with [α-32 P]dCTP as described previously [36].

Generation of Arabidopsis transgenic plants
To produce Arabidopsis transgenic plants overexpressing GhMCS1, the Xho I-Spe I fragment of the GhMCS1 cDNA was ligated into a plant binary vector, pTA7002, allowing for the DEXinducible expression of GhMCS1 [38]. The plants carrying the empty pTA7002 vector was used as a control. For the PGhMCS1::GUS construct, the 1,714-bp Hind III-Xba I fragment of the GhMCS1 promoter was ligated with the GUS reporter gene and then inserted into the corresponding sites of a pBE2113 vector to replace the CaMV35S promoter, producing a pGhMCS1:: GUS plasmid. The recombinant plasmids were introduced into Agrobacterium tumefaciens GV3101 by the liquid nitrogen freeze-thaw method [39] and then transferred into Arabidopsis plants by the floral dip method as described previously [40,41]. Homozygous plants were selected from the T 2 progenies and confirmed in the T 3 generations on the basis of hygromycin or kanamycin resistance. These homozygous plants were utilised in the described experiments.

Subcellular localisation of GhMCS1 protein
Young leaves from 20-d-old cotton or Arabidopsis seedlings were used for isolation of subcellular fractions as described previously [42,43]. Briefly, fresh leaves were washed in distilled water and gently homogenised with a mortar and pestle in extraction buffer (450 mM sucrose, 15 mM MOPS-KOH, pH 7.4, 3 mM EDTA, 0.2% BSA, 0.6% PVP-40, 10 mM DTT, and 0.2 mM PMSF). The homogenate was first filtered through 4 layers of pledget to remove large contaminants, small cell debris and nuclei by centrifugation at 100×g for 5 min. The supernatant was resuspended and served as the total protein.After recentrifugation at 3,500×g for 10 min to pellet crude chloroplasts, the supernatant was centrifuged at 17,000×g for 15 min to pellet mitochondria. The final supernatant was regarded as the cytosol fraction. This final pellet was resuspended with wash buffer (300 mM sucrose, 15 mM MOPS-KOH, pH 7.5, 3 mM EDTA, and 0.2 mM PMSF) and carefully pipetted onto the surface of a 40%/23%/18% Percoll step gradient. After centrifugation at 40,000×g for 45 min, the mitochondria appeared as a brown band at the surface of the 40%/23% Percoll layers, while the chloroplasts appeared as a bright yellow band at the 23%/18% Percoll interface. The subcellular fractions were carefully transferred to fresh tubes and washed with the wash buffer.
For immunolocalisation, the intact mitochondria, chloroplast and cytosol fractions were separated by SDS-PAGE and subjected to Western blot analysis. Western blot was performed as described previously [40], using anti-GhMCS1 serum as the probe for the GhMCS1 protein, while antibodies directed against the wheat subunit 9 of NADH dehydrogenase (NAD9) [42] and Arabidopsis photosystem II core protein CP43 [44] were used as controls for the mitochondrial and chloroplast fractions, respectively. To determine the submitochondrial localisation of GhMCS1, the mitochondrial samples were treated with proteinase K (100 ng/ml) for various durations in the absence or presence of 0.5% Triton X-100. Then, the mitochondrial proteins were subjected to SDS-PAGE and Western blot.
BN-PAGE was used for separation of different mitochondrial complexes on a 5 to 18% polyacrylamide gradient [45]. The respiration complexes were observed directly in the gel and were also subjected to Western blot analysis using antibodies against GhMCS1 or NAD9.

Respiratory activity assay
Respiratory activity was determined as reported previously with some modifications [46]. Briefly, the fresh plant samples were incubated for 8 h in 5.0 ml of 2,3,4-triphenyl tetrazolium chloride (TTC, 0.1%) at 30°C in the dark. After incubation, the samples were washed with sterilized water for three times and then grinded. The formazan was extracted with 3.0 ml of acetone for 5 min. The optical density at 485 nm (A 485 ) of the acetone solution was measured spectrophotometrically, the fresh weights (FW) of the samples were determined, and the A 485 per gram FW was calculated as respiratory activity. The samples without TTC incubation were used as negative control.

Statistical analyses
All data are expressed as the mean ± sd. Statistical analysis was performed with a two-tailed, unpaired, Student's t-test. All differences were considered significant at P<0.05.

Sequence characterization of GhMCS1
In our previous study, a novel cDNA has been detected to be preferentially expressed in developing fibre cells of upland cotton (GenBank Accession No. AJ311660) by fluorescence differential display PCR [24]. Partial amino acid sequence alignment has indicated that it shares a high level of sequence similarity with human GRIM-19. In the present study, we cloned its fulllength cDNA sequence by the RACE-PCR and designated it as GhMCS1 (Gossypium hirsutum mitochondrial complex I subunit 1). By mapping to the NBI-Gh genome database (http:// www.cottongen.org), the GhMCS1 gene was matched to the Gh_A10G1065. As shown in S1 Fig, the open reading frame of GhMCS1 contained 432 bp, encoding a protein of 143 amino acids with a predicted molecular mass of 16 kD and a theoretical pI of 9.83. Nuclear DNA sequence analysis illustrated that it was interrupted by two introns and that the first (4,844 bp) was much larger than the second (90 bp). The sequences surrounding the exon-intron junctions in the GhMCS1 gene conformed to the "GT-AG" rule for donor/acceptor sites (S1 Fig) [47]. This gene structure was greatly similar to those of its homologues from other plant species, including Arabidopsis and rice. The long intron (4844 bp) indicated the possibility of alternative splicing. However, no alternative transcript was detected in the RT-PCR sequencing, suggesting that alternative splicing might not occur during GhMCS1 transcription, or the expression of alternative transcripts was too low to be detected. BLAST analysis indicated that the deduced GhMCS1 protein shared at least 50% amino acid sequence homology with GRIM-19 homologues from animals and plants, including human, bovine, dog, mouse, maize, rice and Arabidopsis (Fig 1). In particular, GhMCS1 was highly similar to two GRIM-19-like protein homologues in Arabidopsis (At1g04630 and At2g33220) that shared 87.94% and 88.65% identities, respectively (Fig 1). Collectively, this sequence information suggests that this gene may encode a GRIM-19-like protein in cotton.
To test whether GhMCS1 was present as multiple copies in tetraploid cotton, we searched the genome sequence of hirsutum (AADD) using GhMCS1 as query [48]. Totally, we found two copies in A subgenome and D subgenome (S2 Fig). All of the four copies showed very high similarities with GhMCS1 in amino acid sequence and gene structure, although some SNP differences were detected (S2 Fig).

Spatial and temporal expression patterns of GhMCS1
To examine the tissue-specific expression patterns of GhMCS1, Northern blot analysis was performed using a probe specific for the GhMCS1 gene. As shown in Fig 2A, GhMCS1 transcripts were detected at significantly high levels in the fibre cells compared with the naked ovules at 10 dpa. These results are consistent with our previous results [24]. In addition, GhMCS1 transcripts were also observed in the leaves, stems and roots, with the highest levels occurring in the flowers at 0 dpa, suggesting that GhMCS1 possessed constitutive expression patterns. During the fibre morphogenetic processes, GhMCS1 transcript abundance was maintained at a high level in the rapidly elongating fibre cells (5-15 dpa), but it decreased significantly at 20 dpa ( Fig 2B). GhMCS1 mRNA was almost undetectable at 25 dpa and diminished at 35 dpa (Fig 2B), indicating that the GhMCS1 gene was predominantly expressed during the fibre elongation stage but not during the subsequent secondary cell wall synthesis stage. It should be noted that because of the high sequence similarity and identical transcript sizes of the four GhMCS1 copies, the spatial and temporal expression pattern is actually from all of four copies.
To gain a better understanding of the regulatory mechanism of the cotton GhMCS1 gene, the 1,741-bp promoter sequence upstream of GhMCS1 was cloned, and the promoter::GUS chimeric construct (pGhMCS1::GUS) was introduced into Arabidopsis. Histochemical GUS staining was performed in various organs throughout plant development. As shown in Fig 2C, GUS activity was detected in 1~3-d-old transgenic seedlings. GUS staining was strong in the cotyledons and hypocotyls but relatively weaker in the radicles; in which activity could be detected only at the root tips (Fig 2C and 2A-2C). In the 5-and 10-d-old seedlings, GUS expression was maintained in the cotyledons and hypocotyls (Fig 2C, 2D and 2F), but in the roots, only trace activity was present in the vascular tissues (Fig 2C, 2E, 2H and 2I). No activity was detected in the root hairs (Fig 2C, 2E, 2H and 2I). Interestingly, GUS expression was also observed in the new true leaves of 10-d-old seedlings, in which the trichomes could be stained (Fig 2C and 2G). Furthermore, in the rosette leaves of 20-d-old soil-grown seedlings, GUS activity could be detected only in the trichomes and veins (Fig 2C, 2J and 2K). In the cauline leaves of 35-d-old transgenic plants, a decreased number of trichomes was stained, and GUS activity was relatively weaker (Fig 2C, 2L and 2M). In the stems of soil-grown plants, no GUS activity could be detected except at the cutting sites (Fig 2C and 2N). The trichome and vascular tissues-preferential GUS activity further supported the notion that GhMCS1 was associated with the elongation of cotton fibres (Fig 2B).
To detect GUS expression in the reproductive organs of transgenic Arabidopsis plants, whole flowers and siliques were also stained ( Fig 2C and 2O-2S), showing intense GUS activity in the stigmas (Fig 2C and 2P). In the stamens, staining could be observed in the anthers ( Fig  2C and 2Q). GUS expression was also detected in the sepals (Fig 2C and 2S) but not in the petals (Fig 2C and 2R). In the young siliques, some weak GUS staining was observed (Fig 2C and  2T), which decreased in intensity as the siliques matured (Fig 2C and 2U). Considering the northern blot and histochemical GUS assay results, GhMCS1 was highly expressed in the trichomes of the transgenic Arabidopsis plants and the elongating fibre cells, which are trichomes of cotton ovules, as well as in the flower tissues, and it is likely that it is precisely regulated during plant development.

Mitochondrial localisation of GhMCS1 protein
It is well known that proteins must localise to their appropriate subcellular compartments to perform their desired functions. To examine the cellular localisation of GhMCS1, cotton leaves were lysed and divided into cytoplasmic, mitochondrial, and chloroplast fractions, and then GhMCS1 protein levels were detected by Western blotting with a rabbit polyclonal antibody against cotton GhMCS1. As shown in Fig 3A, GhMCS1 was detected in the mitochondrial but not in the chloroplast or soluble cytoplasmic fractions. Detection of the mitochondrial marker protein NAD9 in the mitochondrial fraction and the chloroplast marker protein CP34 in the chloroplast fraction indicated ideal fraction separation. The results presented here support the notion that GhMCS1 indeed localises to cotton mitochondria, similar to the GRIM-19 protein in animals [8].
As reported previously, mitochondrial outer membrane proteins, such as BCL2, maybe digested by proteinase K in the absence of Triton X-100, whereas COX IV localised to the inner membrane or Hsp60 localised to the mitochondrial matrix cannot be digested under the same conditions [8]. However, all three proteins are digested by proteinase K in the presence of Triton X-100 [8]. To investigate sub-mitochondrial localisation of GhMCS1 protein, the mitochondrial protein fraction was treated with proteinase K in the presence or absence of 0.5% Triton X-100. As shown in Fig 3B, the GhMCS1 protein and the NAD9 marker could not be digested by proteinase K within the various treatment durations, even after 20 min of treatment, in the absence of Triton X-100. In contrast, both proteins were digested after a 5 min treatment when mitochondria were disrupted in the presence of Triton-X 100. These results suggest that GhMCS1 is localised either to the inner membrane or matrix but not to the outer membrane of mitochondria.
To obtain direct evidence that GhMCS1 is a component of mitochondrial complex I in cotton, a Blue native (BN)-PAGE assay was performed together with Western blot analysis. Mitochondria were isolated from cotton leaves, and OXPHOS complexes were resolved with BN-PAGE. Five major bands representing the enzyme complexes I to V were observed in the gel (Fig 3C). The identity of the target complex was verified by Western blot analysis with corresponding antibodies. As shown in Fig 3C, complex I, which was the largest complex, was detected with the antibody against the mitochondrial marker NAD9 protein. More importantly, the same complex was also recognised by the anti-GhMCS1 antiserum, whereas the other complexes were not detected (Fig 3C), indicating that GhMCS1 is physically present in cotton mitochondrial complex I in its native form.

Changes in mitochondrial activity in GhMCS1-expressing transgenic Arabidopsis
To further investigate its potential biological roles, a plasmid harbouring GhMCS1 cDNA and a dexamethasone (DEX)-inducible promoter were introduced into Arabidopsis (S3 Fig) [37]. Assayed by semi-quantitative RT-PCR, six independent homozygous lines with high levels of transgene expression were obtained (S3 Fig). Three of these transgenic lines (lines 2, 11 and 13) were randomly selected for subsequent biochemical assays. For each transgenic line, the results of the Western blot showed that the GhMCS1 protein could be induced by DEX, and its levels increased as DEX concentrations increased from 0 to 10 μM (S3 Fig). All of the accumulated GhMCS1 was localised to the mitochondria, even in the presence of DEX concentrations as high as 10 μM (S3 Fig) To examine the effects of GhMCS1 expression on mitochondrial activity, respiratory activity was assayed in transgenic Arabidopsis seedlings without or with DEX induction. As shown in Fig 4, an approximate two-fold increase in respiratory activity was detected in both the shoots ( Fig 4A) and roots (Fig 4B) of the transgenic plants compared to those of the control plants or non-DEX-induced seedlings. Further analysis revealed that respiratory activities increased with   increasing DEX concentrations; however, the activities in both the shoots and roots declined when the DEX level reached 10 μM (Fig 4C and 4D). These decreases were presumably due to deficits in other mitochondrial components. The data presented here indicate that respiratory activity in plants is closely associated with GhMCS1 protein concentration within a certain range.
As described above, the GhMCS1 protein was identified to be an important component of cotton mitochondrial complex I (Fig 3). To further understand the effect of GhMCS1 levels on mitochondrial activity, BN-PAGE was used together with Western blotting to analyse the assembly of intact multi-subunit complexes of the mitochondrial respiratory chain. As expected, a distinct increase in GhMCS1 protein abundance in mitochondrial complex I was detected in the GhMCS1-expressing transgenic Arabidopsis compared to the control plants harbouring empty vectors (Fig 4E). Strikingly, increased numbers of complex I could be detected by the anti-NAD9 antibody in the three transgenic lines (Fig 4E). Interestingly, a gradually increasing band representing complex I was observed along with increased levels of GhMCS1 (Fig 4F), indicating the efficient assembly of intact multi-subunit complexes of the mitochondrial respiratory chain in the transgenic Arabidopsis plants. Taken together, our data suggest that GhMCS1 physically resides in complex I and plays an essential role in the assembly of this complex and mitochondrial respiration. Nevertheless, no significant phenotypic differences were observed between the transgenic over-expressing plants with an approximate twofold increase in respiratory activity and the control plants, neither during the early stage of vegetative growth nor during the reproductive stage (data not shown).
Phenotypic changes and mitochondrial activity in Arabidopsis with At1g04630 mutation As indicated above, no distinct phenotypic changes were observed when GhMCS1 was overexpressed in Arabidopsis. One of the most direct approaches for assessing the function of a gene is to observe the effects on an organism that occur when that gene is missing (loss-of-function). To circumvent difficulties in regeneration of cotton plants, many fibre-specific or preferentially expressed genes or promoters have generally been analysed in model plants, such as Arabidopsis [28,49] or tobacco [26]. As shown in Fig 1, GhMCS1 has two highly similar Arabidopsis homologues, At1g04630 and At2g33220, which shared 87.94% and 88.65% amino acid sequence identities with GhMCS1, respectively. As an initial step towards investigating biological function, their expression levels were examined. RT-PCR showed that the transcript levels of At1g04630 and/or At2g33220 in the leaves and stems were much higher than those in the roots (S4 Fig). In the reproductive tissues, the transcript levels were somewhat lower, and the levels in the flowers were higher than those in the siliques and seeds (S4 Fig). This expression seedlings that had not been induced by DEX were used as controls. pTA7002 represents the Arabidopsis seedlings carrying the empty pTA7002 as negative controls. The numbers over the bars indicate the ratios of respiratory activity of the plants with and without DEX induction. (C, D) Dose dependencies of respiratory activity in the shoots (C) and roots (D) in transgenic lines 2 and 13. 10-day-old seedlings of different lines grown on non-inductive medium were transferred to medium containing each concentration of DEX and cultured at 23°C for 3 days. (E) Effects of GhMCS1 on the assembly ability of mitochondrial complex I. 10-day-old seedlings were lysed following 3 d of exposure to 5 μM of DEX, and the isolated mitochondria were subjected to BN-PAGE and assessed immunologically with anti-GhMCS1 and anti-NAD9. (F) Dose dependency of mitochondrial complex I assembly. Seedlings of the transgenic line 2 were used for mitochondrial isolation and Western blot analysis. Complex V was used as a loading control for the BN gel. The numbers under the bands indicate the relative abundance of NAD9 and GhMCS1 proteins. The relative abundance was statistically analyzed by BandScan 5.0. Error bars indicate the sd (n!15) from three independent experiments. *, P < 0.05 and **, P < 0.01; both compared with the corresponding samples without DEX induction. pattern was very similar with that of GhMCS1 in cotton (Fig 2A), especially the GUS distribution pattern in the transgenic Arabidopsis plants harbouring pGhMCS1::GUS (Fig 2C), suggesting that At1g04630 (and/or At2g33220) presents with similar tissue distributions as GhMCS1 in cotton.
For loss-of-function analysis, Arabidopsis mutants with T-DNAs or transposons inserted into At1g04630 and At2g33220 were searched for in the ABRC stocks (Ohio State University, Columbus, OH). Fortunately, two homozygous mutant alleles, At1g04630-1 (SALK_020599C) and At1g04630-2 (CS26138) were obtained. For the At2g33220 gene, a T-DNA-inserted mutant, SALK_054968, was found. However, no homozygous line was separated. We ultimately chose At1g04630-1 and At1g04630-2 for subsequent assays. As shown in Fig 5A, both mutants carried a T-DNA or transposon insertion in the first intron of At104630 that separated the transcriptional from the translational start sites. RT-PCR analysis showed very low At1g04630 transcript levels (Fig 5B), but no change in expression was found for the At2g33220 mRNA (Fig 5B). Western blot analysis also showed that At1g04630 protein levels decreased significantly compared to those in the wild type plants, and in At1g04630-2 mutant, the protein was expressed at the lowest levels ( Fig 5C). In terms of phenotypes, both the At1g04630-1 and At1g04630-2 mutants were shorter and smaller than the wild type seedlings under normal growth conditions. During the whole vegetative growth stage, the At1g04630-2 mutant plants were smaller than At1g04630-1 (Fig 5D). Similar results were observed with regard to seed size. More specifically, the seeds of the two mutants were significantly smaller and poorly developed, especially for the mutant At1g04630-2 ( Fig 5D).
The At1g04630-knockdown mutants possessed true leaves of different sizes and shapes compared to the wild type seedlings (S5 Fig). The leaves of the mutants were smaller (Fig 5F) and more rotund (smaller length:width ratio, Fig 5G). The petiole lengths of the mutants were also significantly shorter (Fig 5H). Considering the fifth leaf as an example, its lengths in At1g04630-1 and At1g04630-2 were only 69.7% and 43.9%, respectively, of that of the wild type ( Fig 5H). Interestingly, a close-up view revealed that the leaf trichomes of both mutants were less numerous and were produced later compared with those of the wild type ( S5 Fig and Fig  5E). Furthermore, significant declines in respiratory activities were detected in the shoots and roots of both mutant plants (Fig 6A and 6B). Western blot analysis also indicated that the amount of complex I was drastically decreased in both mutants, especially in the mutant At1g04630-2 (Fig 6C), which showed decreases in both respiratory activity and complex I abundance of over 50%. Our data indicate that knockdown of At1g04630, which is a homologue of GhMCS1 in Arabidopsis, leads not only to serious decreases in mitochondrial respiratory activity and complex I assembly but also to significant changes in mutant phenotypes, including fewer trichomes, smaller whole-plant architecture, and poorly developed seeds.

Correlation between GhMCS1 level and respiratory activity during cotton fibre growth
To reveal the correlations among GhMCS1 protein level, mitochondrial activity and fibre development, biochemical and cellular assays were performed to evaluate the developing cotton fibres. As shown in Fig 7A, the GhMCS1 protein accumulated at a relatively high level in the elongating fibres (5-15 dpa, elongation stage), which was in good agreement with the northern blot results (Fig 2B). Thereafter, GhMCS1 decreased to a trace level (25 dpa, early secondary cell wall synthesis stage), while no bands could be detected in the mature dehydrated fibres (45 dpa, dehydration stage). Very similar results were observed for the mitochondrial marker protein NAD9, which is a symbolic component of mitochondrial complex I (Fig 7A), suggesting that GhMCS1 expression is tightly controlled at the protein level during the process of cotton fibre development and growth. TTC assays also showed a drastic decline in respiratory activity from 10 dpa to 25 dpa along with a decrease in the GhMCS1 protein level (Fig 7B). Not surprisingly, no activity could be detected in the dehydrated fibres (45 dpa), indicating an important correlation of respiratory activity with cotton fibre growth. Taken together, these data indicate that the GhMCS1 protein level is closely correlated with the respiratory activity of cotton fibres and also with fibre growth.

Discussions
In this study, a cotton GhMCS1 gene encoding a subunit of mitochondrial complex I was characterised. GhMCS1 shares high sequence similarities with GRIM-19 proteins, which have been extensively studied in animals (Fig 1). To determine whether GhMCS1 functions as a bona fide subunit of complex I in cotton, we performed a series of biochemical and molecular assays. Using BN-PAGE and Western blot assays, GhMCS1 was found to be localised to the mitochondria and to colocalise with OXPHOS complex I (Fig 3). Inducible overexpression of GhMCS1 in Arabidopsis resulted in enhancements in respiratory activity and complex I assembly (Fig  4). Although loss-of-function analysis of GhMCS1 was not performed due to difficulties involved in the production of transgenic cotton, Arabidopsis mutants with knockdown of At1g04630 (the closest orthologue of GhMCS1) were identified. The At1g04630 mutation resulted in a significant reduction in both respiratory activity and complex I assembly (Fig 6). Obviously, GhMCS1 functions as a critical subunit involved in mitochondrial complex I assembly and in cotton fibre growth.
As mentioned above, GRIM-19 proteins have been detected in animal mitochondrial complex I and are essential for complex I assembly and electron transfer activity [8,50]. More importantly, the functional domains in GRIM-19 proteins have also been defined in mammalian cells. As previously reported, the N-terminal domain (1-60 aa) is required for mitochondrial localisation and complex I incorporation, the middle region (70-100 aa) is essential for electron transfer activity of complex I (ΔCm), and the 10-aa C-terminal region may promote GRIM-19-mediated assembly of complex I [51]. Based on sequence alignment analysis, GhMCS1 as well as its Arabidopsis orthologue At1g04630 possess conserved middle regions (Fig 1), indicating their potential roles in maintaining ΔCm. However, these plant orthologues of GRIM-19 lack C-terminal regions (Fig 1), suggesting a specific assembly mechanism of mitochondrial complex I mediated by GhMCS1 and its orthologues in plants. Further research is required to elucidate these mechanisms.
GhMCS1 transcripts were detected to be enriched in developing fibres during the elongation stage (Fig 2B). Both Arabidopsis trichomes and cotton fibres shared the similar process of cell elongation. The coss-species gene over-expression and complementation experiments have identified the common transcriptional regulatory network between them [52]. Additionally, the xylems can also serve as the model for the late stages of cotton fiber development [53]. In Arabidopsis, the GhMCS1 promoter may drive GUS reporter gene expression in the trichomes and vascular tissues of leaves ( Fig 2C). Besides, the At1g04630 mutant alleles produced less trichomes than the wild type plants (Fig 5). Furthermore, a close correlation was found among GhMCS1 protein level, respiratory activity and fibre developmental phase (Fig 7), indicating that GhMCS1 may participate in cotton fibre development by regulating mitochondrial activity. These results suggested that GhMCS1 is involved in cotton fibre growth and development. However, further evidence is necessary to elucidate the detailed underlying mechanisms.
Pang et al. have reported that an additional mitochondrial protein, the δ1 subunit of ATP synthase (GhATPδ1), is related to cotton fibre growth [54]. GhATPδ1 expression is significantly up-regulated during fibre cell elongation, showing expression patterns similar to those of GhMCS1. During fibre cell wall formation and elongation, high energy is required. In fibre cells, GhATPδ1 expression may result in an elevated ATP:ADP ratio, while exogenous application of inhibitors of ATP synthase in ovule culture media may lower the ATP:ADP ratio, resulting in shorter fibres. In mitochondria, concomitant proton translocation by the membrane arm of complex I is coupled to OXPHOS, generating ATP [55]. This is the major source of intracellular ATP, especially in non-photosynthetic tissues. Furthermore, the transport of H + and the exchange of ATP/ADP across the inner membrane are driven by ΔCm. As discussed previously, GhMCS1 may play an essential role in maintaining ΔCm [51]. Therefore, GhMCS1 may function to limiting mitochondrial ATP generation. During cotton fibre growth, GhMCS1 expression may ensure for the precise assembly of mitochondrial complex I and maintenance of ΔCm across the inner membrane, leading to coupled ATP synthesis, although further evidence is necessary to verify this hypothesis. The ΔCm value maintained by GhMCS1 and the resulting ΔpH can also drive the export of organic acids from the mitochondrial matrix to the cytosol. Organic acids, including citrate, succinate, malate, and oxaloacetate, are metabolites of the TCA cycle. They can be translocated across the inner mitochondrial membrane by a series of specific transporters. During cotton fibre elongation, these organic acids can function as osmotic regulators to maintain cellular turgor pressure, leading to unidirectional outgrowth from the epidermis of the developing ovule [56,57]. To further reveal the relationship of mitochondrial complex I and respiratory activity in cotton fibre growth, the effects of respiratory inhibitors, such as rotenone and piericidin A, would be detected in cotton ovules culture.
In addition to its potential roles in cotton fibre growth, other functions of GhMCS1 and its orthologs have not been excluded. As described in this study, the Arabidopsis mutants with knockdown of At1g04630, the GhMCS1 ortholog, showed smaller seedling phenotypes and seed sizes. Additionally, smaller and rounder leaves and shorter petioles were found in these mutants (Fig 5, S4 Fig). In another study involving a large-scale mutant screen of Ds transposon insertion lines, At1g04630 was identified as AtMEE4 (maternal effect embryo arrest 4), and its disruption resulted in the arrest of endosperm development at the one-cell zygotic stage [58]. Similarly, in mice, GRIM-19 knockout has been shown to cause early embryonic lethality [8], and its knockdown in Xenopus has been demonstrated to lead to failure of early heart development [59]. These results indicate that GhMCS1 and its orthologs might also be critical for both seedling morphogenesis and embryonic development in plant. However, whether these functions are associated with their roles in mitochondrial complex I assembly are largely unknown.

Conclusions
In conclusion, the current work provides a comprehensive description of the GhMCS1 gene, including its expression pattern, subcellular localisation and potential roles in cotton fibre growth. These results collectively suggest that GhMCS1 is a conserved subunit of mitochondrial complex I in cotton. Both gain-of-function and loss-of-function transgenic assays suggest that GhMCS1 can facilitate the assembly of complex I and efficiently maintain the integrity and activity of mitochondria. We also have shown that GhMCS1 is associated with cotton fibre growth. Further investigations are expected to reveal the detailed mechanisms of GhMCS1 in complex I assembly and cotton fibre growth.