Overexpression of a truncated CTF7 construct leads to pleiotropic defects in reproduction and vegetative growth in Arabidopsis

Eco1/Ctf7 is essential for the establishment of sister chromatid cohesion during S phase of the cell cycle. Inactivation of Ctf7/Eco1 leads to a lethal phenotype in most organisms. Altering Eco1/Ctf7 levels or point mutations in the gene can lead to alterations in nuclear division as well as a wide range of developmental defects. Inactivation of Arabidopsis CTF7 (AtCTF7) results in severe defects in reproduction and vegetative growth. To further investigate the function(s) of AtCTF7, a tagged version of AtCTF7 and several AtCTF7 deletion constructs were created and transformed into wild type or ctf7+/–plants. Transgenic plants expressing 35S:NTAP:AtCTF7∆299–345 (AtCTF7∆B) displayed a wide range of phenotypic alterations in reproduction and vegetative growth. Male meiocytes exhibited chromosome fragmentation and uneven chromosome segregation. Mutant ovules contained abnormal megasporocyte-like cells during pre-meiosis, megaspores experienced elongated meiosis and megagametogenesis, and defective megaspores/embryo sacs were produced at various stages. The transgenic plants also exhibited a broad range of vegetative defects, including meristem disruption and dwarfism that were inherited in a non-Mendelian fashion. Transcripts for epigenetically regulated transposable elements (TEs) were elevated in transgenic plants. Transgenic plants expressing 35S:AtCTF7∆B displayed similar vegetative defects, suggesting the defects in 35S:NTAP:AtCTF7∆B plants are caused by high-level expression of AtCTF7∆B. High level expression of AtCTF7∆B disrupts megasporogenesis, megagametogenesis and male meiosis, as well as causing a broad range of vegetative defects, including dwarfism that are inherited in a non-Mendelian fashion.

Arabidopsis contains a single CTF7 ortholog (AtCTF7), which can complement the temperature-sensitive yeast ctf7-203 mutant [16]. AtCTF7 encodes a 345 amino acid protein, containing a PIP box at residues 82 to 86, a C 2 H 2 zinc finger motif at residues 92-130 and an acetyltransferase domain from residues 184 to 335. The acetyltransferase domain can be further separated into three motifs: D (amino acids 184-204), A (amino acids 266-301) and B (amino acids 311-335). Motif D provides the framework of the acetyltransferase domain. Motif A participates in acetyl-CoA binding and is critical for catalytic activity, while motif B is the most Cterminal region and participates in substrate recognition and catalytic activity regulation [28,29]. Variations in AtCTF7 expression have been shown to result in a wide range of defects. Plants heterozygous for a T-DNA insertion in AtCTF7 grow normally but their siliques contain approximately 25% arrested seeds, consistent with the belief that CTF7 is an essential protein [16]. However, Atctf7 homozygous plants can be detected at very low frequencies [17]. Atctf7 plants exhibit a wide range of developmental defects, including extreme dwarfism and sterility with Atctf7 meiocytes exhibiting abnormal chromosome segregation and defective sister chromatid cohesion. Knockdown of AtCTF7 mRNA levels using RNAi leads to growth retardation and defective sister chromatid cohesion [18]. Finally, overexpression of full length AtCTF7 with the CaMV 35S promoter leads to ovule abortion, but plant growth is not affected [16].
To further investigate the roles of CTF7 in Arabidopsis and to identify AtCTF7 interacting proteins, transgenic plants expressing AtCTF7 constructs that encode either TAP-tagged [30] or non-tagged versions of full length or truncated AtCTF7 were generated and analyzed. Transgenic plants that express a truncated version of AtCTF7, missing motif B, from the 35S promoter exhibited impaired reproduction and vegetative growth. Ovules in 35S:NTAP:AtCTF7ΔB plants displayed alterations in cell identity and the timing of megasporogenesis and megagametogenesis, while male meiocytes exhibited alterations during meiosis II. The transgenic plants also displayed alterations in vegetative growth that were not inherited in a Mendelian fashion.

Results
Arabidopsis plants expressing high levels of NTAP: AtCTF7ΔB display reduced fertility An AtCTF7 construct missing the C-terminal 46 amino acids (Δ299-345) was generated, fused with NTAP [30] and expressed from the CaMV 35S promoter in wildtype Columbia plants (35S:NTAP:AtCTF7ΔB; Additional file 1: Figure S1). Twenty out of the 36 independent lines examined exhibited reduced fertility, with fertility levels varying significantly between the lines. Plants exhibiting a weak phenotype, which accounted for two of the 20 reduced fertility lines, produced shorter siliques with reduced numbers of seeds, but the seeds appeared normal (Figure 1Aii). For example Line 19 produced 39.2 ± 3.9 seeds per silique (n = 35) compared to wild type plants that produce 54.2 ± 4.1 seeds per silique (n = 35). Anthers from Line 19 plants were smaller and contained reduced numbers of pollen (574 vs 1175 in wild type), but the pollen appeared viable (Figure 1Bii). The other 18 reduced fertility lines exhibited more severe defects. These plants produced siliques containing large numbers of unfertilized ovules and aborted seeds (Figure 1Aiii). Unfertilized ovules appeared as white dots, resembling the situation in atctf7-1 plants [17]. Aborted seeds appeared white and plump, similar to seeds containing arrested embryos in Atctf7-1 +/− plants [16]. Seed set varied considerably between the lines, ranging from 13.3 ± 5.6 seeds per silique to 34.7 ± 8.4 seeds per silique. Anthers from these plants typically contained reduced numbers of pollen, much of which was not viable. For example, anthers from Line 11 produced on average approximately 210 pollen, of which only 20% was viable ( Figure 1Biii). Given that most lines exhibited severe fertility defects, one representative line (#11) was chosen and characterized in detail.
The reduced fertility phenotype of 35S:NTAP:AtCTF7ΔB plants suggested that co-suppression may be lowering AtCTF7 transcript levels in the lines. Therefore, quantitative reverse transcription polymerase chain reaction (qRT-PCR) was conducted to examine transcript levels of native AtCTF7 as well as the CTF7 transgene. Surprisingly, total AtCTF7 (35S:NTAP:AtCTF7ΔB and native AtCTF7) transcript levels were approximately seven-fold higher in 35S: NTAP:AtCTF7ΔB plants than CTF7 transcript levels in wild type plants, while native AtCTF7 transcript levels were almost three-fold higher in the transgenic plants (Figure 1C). Therefore, the transgenic lines contained high levels of AtCTF7ΔB transcripts along with elevated levels of native AtCTF7 mRNA, indicating that the observed phenotypes are not the result of reduced AtCTF7 expression. Instead these results suggest that the 35S: NTAP:AtCTF7ΔB construct may be exerting a dominant negative effect.
Reciprocal crossing experiments were carried out to determine the effect of the 35S:NTAP:AtCTF7ΔB construct on male and female gametes. The 35S:NTAP: AtCTF7ΔB construct was transmitted at reduced levels through both male and female gametes (Additional file 1: Table S1), consistent with our preliminary results that both male and female fertility were affected in 35S:NTAP:AtCTF7ΔB plants. Nonviable seeds were obtained when 35S:NTAP:AtCTF7ΔB plants were used as either the male or female parent; however the relative proportion of nonviable seed was three times greater (9.3% verses 31.3%) when 35S:NTAP:AtCTF7ΔB plants served as the female. The number of nonviable seeds was greatest (65%) in self-pollenated plants. Therefore, while both male and female gametes are affected, the 35S:NTAP:AtCTF7ΔB construct has a greater effect on female reproduction. Inheritance of 35S:NTAP:AtCTF7ΔBassociated defects from both parents has a compounding effect on seed production.

Male meiotic chromosome segregation is altered in 35S: NTAP:AtCTF7ΔB plants
Previous studies showed that inactivation of AtCTF7 via T-DNA insertion or reduction in AtCTF7 mRNA levels via RNAi disrupts sister chromatid cohesion, resulting in uneven chromosome segregation during meiosis and ultimately reduced pollen viability [17,18]. The reduced fertility observed in 35S:NTAP:AtCTF7ΔB plants suggested that meiosis might also be affected in these lines. Therefore, chromosome spreading experiments were carried out to examine the effect of 35S:NTAP:AtCTF7ΔB expression on male meiosis. Male meiocytes in 35S:NTAP:AtCTF7ΔB plants resembled wild type during early stages of meiosis, with normal chromosome morphology during pachytene (Figure 2A,E), diakinesis ( Figure 2B,F) and metaphase I ( Figure 2C,G). The first noticeable defect was observed at telophase I when lagging chromosomes were observed ( Figure 2D,H), followed by mis-segregated chromosomes at prophase II ( Figure 2M). More than twenty individual chromosomes were typically observed in meiocytes beginning at metaphase II ( Figure 2N), indicating that sister chromatid cohesion was prematurely lost. Chromosomes did not segregate evenly at anaphase II ( Figure 2O) resulting in the production of polyads with varying DNA contents at tetrad stage ( Figure 2P). Similar to atctf7 plants, a small number of relatively normal meiocytes were also observed throughout meiosis with fewer normal-appearing meiocytes in later stages of meiosis. For example, the percentages of defective meiocytes observed at various stages of meiosis in Line 11 were: metaphase I: 0% (0/41), telophase I: 6.8% (7/87), prophase II: 23 Immunolocalization experiments were then carried out to examine the distribution of the meiotic cohesin protein SYN1 [31] in 35S:NTAP:AtCTF7ΔB male meiocytes. Consistent with meiotic chromosome spreading experiments the overall distribution of SYN1 was not dramatically affected in meiocytes of 35S:NTAP: AtCTF7ΔB plants (Additional file 1: Figure S2). SYN1 exhibited diffuse nuclear labeling during interphase with strong signal observed on the developing chromosomal axes from early leptotene into zygotene (Additional file 1: Figures S2A,D). During late zygotene and pachytene the protein lined the chromosomes (Additional file 1: Figures S2B,E,C and F). SYN1 was released normally from the condensing chromosomes during diplotene and diakinesis and similar to the situation in wild type, SYN1 was barely detectable on chromosomes by prometaphase I. Therefore, cohesin appears to load and be removed normally from meiotic chromosomes in 35S: NTAP:AtCTF7ΔB plants.  Previous studies showed that ovule development is very sensitive to AtCTF7 levels [16][17][18]. Approximately 50% of the ovules from Atctf7-1 +/plants contain nondegenerated antipodal nuclei [16], while ovules from atctf7 plants degenerate early [17]. RNAi directed reduction of AtCTF7 mRNA levels, as well as 35S-mediated increases in AtCTF7 transcript levels result in ovule arrest at Female Gametophyte (FG) 1 stage [16,18].
To elucidate the function(s) of 35S:NTAP:AtCTF7ΔB in ovule development, ovules from wild type and Line 11 plants were analyzed by differential interference contrast (DIC) microscopy. In wild type siliques, archesporial cells are specified from the subepidermal cell layers and differentiate into megaspore mother cells, which are initially unpolarized and become polarized prior to the start of meiosis (Additional file 1: Figure S3A). Two rounds of meiosis produce a tetrad of four haploid megaspores ( Figure 3C,D; Additional file 1: Figure S3B-E) [32]. Prior to FG1, the megaspore mother cell, dyad and tetrad are all adjacent to L1 cells (Figures 3A-D) [33]. When the ovule reaches FG1, the megaspore at the chalazal-end differentiates into the functional megaspore while the other three megaspores undergo programmed cell death (Additional file 1: Figures S3F, G; Figures S4B,  C) [34,35]. In wild type siliques, adjacent ovules are similar in size and point in opposite directions ( Figures 3A,B).
In mutant plants, defects were observed in ovules very early in development. Approximately 30.4% (41/135) of the ovules observed were defective at the premeiosis stage. Abnormally enlarged sub-epidermal cells were observed adjacent to normal looking megaspore mother cells in some pre-meiotic ovules ( Figure 3E,F'). Adjacent ovules were often different in size and sometimes pointed in the same direction (Figure 3 F,F'). A functional megaspore was not identified in some ovules even though their shape and size was beyond FG1. Instead, large cells with prominent nuclei and megaspore-like characteristics [36] were found in the position of a normal embryo sac (Figures 3G-I).
Ovules containing either one megaspore-like cell ( Figure 3G,H) or two megaspore-like cells with dyad characteristics ( Figure 3I) were also observed. This suggested that meiosis is delayed, altered and arrested in most megaspores. Extra cells were also observed between the megaspores and the L1 layer of cells ( Figure 3G-I). In WT plants, a small number (6.7%, 8/120) of ovules were found to contain twin megaspore mother cells; however no defects were observed during or after meiosis. In Line 11 plants, 14.1% (19/135) of the ovules contained multiple megaspore mother-like cells at pre-meiosis. Furthermore, 72.5% (50/69) of ovules with dyads contained extra cells between the dyad and L1 layer, and 27.5% (19/69) the dyads were in elongated shape. Ultimately 50.4% (64/127) of the ovules observed were defective at FG1 stage. Therefore, alterations in archesporial cell differentiation, the onset and progression of meiosis and possibly somatic cell identity are observed in ovules of 35S:NTAP:AtCTF7ΔB plants.
The effects of 35S:NTAP:AtCTF7ΔB on megagametogenesis were investigated by confocal laser scanning microscopy (CLSM) and DIC microscopy. During wild type megagametogenesis, the functional megaspore undergoes three rounds of mitosis accompanied by nuclear migration, fusion, degeneration and cellularization to form the final embryo sac (Additional file 1: Figures S3H-L; S4D-J). At FG1, the functional megaspore undergoes mitosis to produce a two-nucleate embryo sac (FG2; Additional file 1: Figure S3H). Formation of a vacuole between the two nuclei marks stage FG3 (Figure 4,D and L; Additional file 1: Figures S3I, S4D). During FG3 the ovule becomes curved and the inner integument embraces the nucellus. A second round of mitosis produces a four-nucleate embryo sac (FG4; Figure 4E,M; Additional file 1: Figures S3J, S4E). This is followed by migration of the two chalazal nuclei from an orthogonal orientation to a chalazal-micropylar orientation. After nuclear migration, a third round of mitosis gives rise to eight nuclei in a 4n + 4n configuration (FG5; Additional file 1: Figure S3K). The two polar nuclei, one from each side, meet at the embryo sac's micropylar half and fuse to form the central cell, while the antipodal nuclei start to degenerate (Additional file 1: Figure S4H). The central cell has formed and the antipodal nuclei are completed degenerated by FG7 (Additional file 1: Figures S3K, S4I). Prior to fertilization, one synergid nucleus degenerates, such that the embryo sac consists of one egg cell, one central cell and one synergid nucleus (FG8; Additional file 1: Figures S3L,  S4J).
Ovule development is typically synchronous in wild type sliques with predominately one or two developmental stages present in a given pistil (Additional file 1: Table S2). In contrast, 35S:NTAP:AtCTF7ΔB ovule development appeared slowed and asynchronous (Additional file 1: Table  S3). Female gametophytes in the same pistil were often at several different stages, indicating that the synchrony of gametophyte development was disturbed and embryo sac maturation was delayed. Ovules containing a megaspore ( Figures 3G,H'), FG1 embryo sacs with a functional megaspore-like cell [37] and degrading megaspores ( Figures 4G,H,N), and abnormal FG2 and FG3 embryo sacs were commonly observed in the same slique ( Figure 4I,O-P). Ovules containing degraded/degrading megasporocytes, and degrading FG1 embryo sacs were also observed (Additional file 1: Figures S5A-D). Alterations in nuclear division appeared to precede arrest in some megaspores (Additional file 1: Figure S5E-H). Most embryo sacs arrest at FG2/FG3; although some terminal ovules appeared to progress beyond FG3 (Additional file 1: Table S3). Common phenotypes included degraded/degrading nuclei (Additional file 1: Figure S5J-K), degenerated embryo sacs (Additional file 1: Figure S5M), polar nuclei fusion defects (Additional file 1: Figure S5N) and vacuole development defects (Additional file 1: Figure S5N-P).
To determine if the alterations observed in Line 11 are representative of 35S:NTAP:AtCTF7ΔB plants in general, ovules from other lines exhibiting severely reduced fertility (#13 and #15) were examined. Similar to the situation in Line11, pre-meiotic ovules from Line15 contained abnormally enlarged sub-epidermal cells adjacent to megaspore mother-like cells ( Figures 5A-C). During meiosis, ovules contained abnormal cells adjacent to degenerated megaspores ( Figures 5D-Q). Some ovules contained functional megaspore-like cells ( Figure 5F-H,Q). Mature ovules contained differentiated functional megaspore-like cells (Figures 5R,S) and female gametophytes with various defects ( Figures 5U,V). Similar to Line 11, female gametophytes from Lines 13 and 15 developed slowly and asynchronously; embryo sacs arrested at FG1 and FG3 (Additional file 1: Figure S6A,C). Additional alterations not observed in Line 11 were also identified, including some ovules that appeared to contain two functional megaspore-like cells (Additional file 1: Figure S6B). In some ovules the middle megaspores appeared to differentiate into functional megaspore-like cells while the megaspores at the chalazal-end degenerated (Additional file 1: Figure S6D), suggesting that ovule polarity was disrupted. All together, 30.4% (48/158) ovules examined in Line 15 plants displayed alterations at pre-meiosis; 66.0% (68/103) of the ovules were defective at meiosis and 79.3% (73/92) of the ovules were defective at FG1. Therefore, common defects associated with NTAP:AtCTF7ΔB include a delay and alterations in both megasporogenesis and megagametogenesis with embryo sacs arresting at various stages of development.

Expression of 35S:NTAP:AtCTF7ΔB causes pleiotropic growth defects
During the analysis of 35S:NTAP:AtCTF7ΔB reduced fertility lines, plants displaying vegetative defects began to appear in the T2 or T3 generations. Specifically, later generations grew progressively worse in 12 of the 18 independent severely reduced fertility lines examined. The remaining six lines continued to display reduced fertility, but did not exhibit vegetative defects through the seventh generation. A wide range of morphological defects was observed in the 12 lines ( Figure 6). The defects varied between lines and between progeny of the same line. The observed vegetative abnormalities included dwarf plants, fused stems and disruption of phyllotaxis ( Figures 6C-E).
The proportion of plants exhibiting vegetative alterations increased in successive generations. Likewise, the severity of the vegetative alterations also became successively worse in subsequent generations. For example, the frequency of dwarf plants increased from approximately 18% in the third generation to 82% by generation six (Figure 6; Table 1). The dwarf phenotype also became more severe with later generations containing smaller, more defective plants. Dwarf plants varied in morphology and exhibited a range of alterations, including acaulescent plants, floral abnormalities, homeotic changes and irregular leaves ( Figure 6F-I). While some of the most severe 35S:NTAP: AtCTF7ΔB dwarf plants resembled atctf7-1 plants ( Figure 6J), in general the phenotype was less severe than atctf7-1 plants. Furthermore, while most atctf7-1 plants exhibit early senescence, 35S:NTAP:AtCTF7ΔB plants typically did not. In order to determine if the phenotypic alterations were due to increased 35S:NTAP:AtCTF7ΔB expression, native CTF7 and total CTF7 levels were measured in both dwarf and non-dwarf plants of three different generations of 35S:NTAP:AtCTF7ΔB plants, which showed progressively more severe phenotypes. Native Ovule with a FG3 embryo sac. FMLs are identified as having distinctly bright nuclear autofluorescence and DMs contain a diffuse signal throughout the cells, but no clearly defined nucleus, defined according to Barrell and Grossniklaus [37]. Size bar = 10 μm. Developmental stages in CLSM and DIC are defined according to Christensen et al. [34,35] and Schneitz et al. [32], respectively. CTF7 levels ranged from 2.6-3.7 fold above wild type in the different plants, while total CTF7 transcript levels were elevated 5.4-7.6 fold relative to wild type. However, no consistent difference was observed between dwarf and nondwarf plants or between one generation and the other. Therefore, the phenotypic differences observed are not due to dramatic changes in AtCTF7ΔB transcript levels.
Rather, and most surprisingly, the dwarf phenotype was not inherited in a Mendelian fashion (Table 1). When dwarf plants were selfed they produced a mixture of dwarf and  non-dwarf, reduced fertile plants. The frequency of dwarf plants produced from selfed dwarf plants was similar to the frequency of dwarf plants resulting from selfing a nondwarf plant. The stochastic appearance of the dwarf phenotype and variation in phenotypes suggested that the alterations could be the result of epigenetic changes. In order to investigate this possibility, qRT-PCR was carried out to measure the expression levels of several epigenetically regulated transposable elements (TEs), including MU1, COPIA 28 and solo LTR [46], as well as several genes associated with epigenetic events [47][48][49][50]. Expression levels of MU1, COPIA 28 and solo LTR were increased between five (MU1) and 24 fold (COPIA 28) in 35S:NTAP:AtCTF7ΔB plants ( Figure 7A). Subtle changes were also observed in the transcript levels of several siRNA associated genes ( Figure 7B). ARGONAUTE1 (AGO1), RDR2 and mir156 transcript levels were reduced approximately 40-60% while AGO4 transcripts were elevated slightly. Transcript levels of HDA19 and RDM4 were also decreased approximately 50% ( Figure 7B), while transcript levels of the canonical DNA methylation genes, MET1 and DMT7, did not vary significantly ( Figure 7B).
Transcript levels of several cell cycle (RBR, CYCB1.1 and CYCA1.1) and DNA repair (BRCA1 and BRCA2B) genes have previously been shown to be elevated in atctf7 plants [17]. These genes were tested and found to also be elevated in 35S:NTAP:AtCTF7ΔB plants ( Figure 7C). The similarities in morphological defects and expression patterns observed between 35S:NTAP:AtCTF7ΔB and atctf7-1 plants [17] are consistent with the hypothesis that the 35S:NTAP:AtCTF7ΔB construct is exerting a dominant negative effect.
Finally, experiments were carried out to determine if the 35S:NTAP:AtCTF7ΔB-associated defects are due to the presence of the N-terminal tag or the absence of the acetyltransferase B motif. A 35S:AtCTF7ΔB construct was generated and transformed into wild type Columbia plants (Additional file 1: Figure S1). Eight out of the 13 35S: AtCTF7ΔB transgenic lines examined exhibited reduced fertility, with the 35S:AtCTF7ΔB defects typically appearing more severe than those in 35S:NTAP:AtCTF7ΔB plants (Additional file 1: Figure S8). Like 35S:NTAP:AtCTF7ΔB plants, dwarf plants were not observed until the third generation in 35S: AtCTF7ΔB plants. However, dwarf plants appeared at higher frequencies and their phenotypes were more varied than 35S:NTAP:AtCTF7ΔB plants (Additional file 1: Figure S8B). For example, the viability of Line 21 decreased with successive generations, such that seeds from this line were not viable by the 4th generation. Likewise, while abnormalities associated with 35S:AtCTF7ΔB plants were similar to those observed in 35S:NTAP:AtCTF7ΔB plants, additional alterations were also present. For example, 35S:AtCTF7ΔB Line 24 segregated for two types of plants, those without an inflorescence and "normal" reduced fertility plants (Additional file 1: Figure S8C). 35S: AtCTF7ΔB Line 29 plants produced siliques that pointed downward and contained fewer seeds (46.4 ± 2.5 versus 54.2 ± 4.1 per silique in WT; n = 35) (Additional file 1: Figure S8D). This phenotype is similar to bp/knat1 mutations [51].
Finally, in order to investigate which aspect(s) of the 35S: NTAP:AtCTF7ΔB construct was causing the fertility and growth defects, several additional constructs were generated and introduced into wild type or Atctf7-1 +/− plants. Wild type plants transformed with a 35S:NTAP:AtCTF7 construct (Additional file 1: Figure S1) resembled 35S: AtCTF7 plants [16]. Specifically, the plants grew normally and produced pollen, but exhibited reduced female fertility. A CTF7:AtCTF7ΔB construct (Additional file 1: Figure S1) was also created and transformed into Atctf7-1 +/plants. The native CTF7 promoter is expressed at low levels throughout the plant [17]. Wild type plants containing the CTF7:AtCTF7ΔB construct exhibited normal growth and development and normal fertility levels (52.0 ± 2.2 seed/silique, n = 33 versus 54.2 ± 4.1 seed/silique in wild type, n = 35). No alterations were observed in six different transgenic lines over six generations. atctf7-1 plants containing the CTF7:AtCTF7ΔB construct were obtained in T2 populations, but at frequencies (7%, 6/86) much lower than expected (25%) if the construct complemented the atctf7-1 mutation. While these plants were dwarf, they grew better than atctf7-1 plants (Additional file 1: Figure S8E). Somewhat similar to 35S:AtCTF7ΔB plants, plant morphology varied between plants with some plants appearing acaulescence or producing fewer rosette leaves. The plants produced approximately 40 ovules per silique; however they failed to set seed; siliques contained aborted ovules resembling atctf7-1 plants (Additional file 1: Figure S8G). Therefore, the defects observed in 35S:NTAP:AtCTF7ΔB plants

Discussion
The acetylation of cohesin complexes at conserved lysine residues by CTF7/Eco1 plays an essential role in the establishment of cohesion during S phase and therefore nuclear division. Consistent with this, CTF7/Eco1 null mutations are typically lethal. Organisms expressing altered CTF7/Eco1 levels or point mutations in the protein typically display relatively normal levels of cohesin during nuclear division, but exhibit a range of developmental alterations. For example, mutations in the N-terminus of the protein typically lead to defects in cohesion and often chromosome loss during mitosis [52]. In contrast certain mutations in the C-terminal acetyltransferase domain of yeast CTF7/ECO1 have little effect on S-phase cohesion and chromosome segregation, but cause an increased sensitivity to DNA-damaging agents. Likewise, Roberts Syndrome in humans has been linked to point mutations in ESCO2. Cells from patients with Roberts Syndrome are typically hypersensitive to DNA-damaging agents and show premature centromere separation; however, only 10-20% of cells show abnormal mitosis [27,53]. Finally, numerous studies have shown that cohesin mutations or reductions in cohesin levels result in transcriptional alterations that can have far-ranging developmental consequences [54]. Generally similar results have been obtained from studies on CTF7 in plants. Arabidopsis plants heterozygous for a T-DNA insertion in AtCTF7 grow normally but produce approximately 25% aborted seeds, consistent with the conclusion that CTF7 is an essential protein [16]. Likewise knockdown of AtCTF7 mRNA levels leads to growth retardation and defective sister chromatid cohesion [18]. However, unlike other organisms, homozygous Atctf7 plants have been detected at very low frequencies [17]. Atctf7 plants exhibit a wide range of developmental defects, including extreme dwarfism and sterility.
In our current study we show that high-level expression of a truncated form of AtCTF7 results in reduced fertility and dramatic alterations in vegetative growth. Specifically, plants that express a 35S:NTAP: AtCTF7 Δ299-345 construct exhibit defects in male and female meiocytes, with female reproduction being affected more dramatically. Male meiocytes exhibited chromosome fragmentation and uneven chromosome segregation during meiosis II that resulted in abnormal pollen development and ultimately pollen abortion. Ovules contained abnormal megasporocyte-like cells during pre-meiosis, megaspores that experienced elongated and aborted meiosis and defective megaspores and embryo sacs that arrested at various stages. A broad range of vegetative defects was also observed beginning in T 2 generations of AtCTF7ΔB transgenic plants. The appearance of these defects was stochastic and inherited in a non-Mendelian fashion. Comparison of AtCTF7ΔB transgenic plants with AtCTF7 RNAi and atctf7 plants and CTF7/ Eco1 mutants in other organisms suggests that CTF7 may have multiple roles in the cell.
Reproductive defects in 35S:NTAP:AtCTF7ΔB plants differ from those in AtCTF7RNAi and atctf7 plants Inactivation of AtCTF7 by T-DNA insertion or a reduction in AtCTF7 levels by AtCTF7-RNAi lead to alterations in chromosome condensation and sister chromatid cohesion during early meiotic prophase followed by defects in homologous chromosome pairing and segregation later in meiosis [17,18]. The effect of 35S:NTAP:AtCTF7ΔB on male meiocytes was less severe and occurred later in meiosis. The first noticeable defect was the appearance of lagging and broken chromosomes during telophase I ( Figure 2H). Twenty or more individual chromosomes were often observed beginning at metaphase II, suggesting that cohesion might be prematurely lost. Overexpression of AtCTF7ΔB did not have a noticeable effect on the initial establishment of cohesion, as the distribution of SYN1 on meiotic chromosomes was normal throughout prophase (Additional file 1: Figure S2). Likewise, chromosome condensation, sister chromatid cohesion and homologous chromosome pairing were normal during male meiotic prophase in 35S: NTAP:AtCTF7ΔB plants. This suggests that AtCTF7ΔB does not affect the bulk of meiotic cohesin complexes to a significant extent, but rather may alter centromeric cohesin levels, or possibly cohesin interactions with SGO1 or PATRONUS [55].
In contrast to male reproduction, the effect(s) of 35S: NTAP:AtCTF7ΔB on female reproduction are observed earlier and are more variable than those in atctf7 and AtCTF7 RNAi plants. Over expression of AtCTF7 from the 35S promoter or knockdown of AtCTF7 using RNAi blocks early ovule development, typically at FG1 or FG2 [16,18]. In contrast, atctf7 ovules in AtCTF7 +/plants develop normally, but arrest soon after fertilization [16]. In all three situations the alterations are relatively uniform with arrest occurring at a specific developmental stage. In contrast, 35S:NTAP:AtCTF7ΔB causes pleiotropic ovule/seed defects. NTAP:AtCTF7ΔB lines displayed a wide range of defects, including additional abnormal cells adjacent to gametic cells, delayed/arrested meiosis, the production of functional megaspore-like cells of which some are mis-positioned, and delayed and altered embryo sac development. Although alterations were commonly first observed prior to and during meiosis, most megaspores progressed to FG2 or FG3 before arresting (Additional file 1: Table S3).
35S:NTAP:AtCTF7ΔB leads to defects consistent with epigenetic alterations 35S:NTAP:AtCTF7ΔB lines exhibited relatively normal vegetative growth and development for the first two generations. However, severe vegetative abnormalities began to appear starting in the T2 or T3 generations of different lines. The defects, which included dwarf plants, fused stems and disrupted phyllotaxis ( Figure 6C-E), varied between lines and between progeny of the same line. The proportion of plants exhibiting vegetative alterations as well as the severity of the vegetative alterations increased in successive generations. It is interesting to note that phenotypic variability is very common in RBS patients [26,27].
The delayed appearance of vegetative defects and the increased frequency of defects in subsequent generations could result from the accumulation of defects in 35S: NTAP:AtCTF7ΔB plants. Consistent with this possibility is the observation that ctf7/eco1 mutations are commonly associated with sensitivity to DNA damaging agents [56]. As expected, both 35S:NTAP:AtCTF7ΔB and atctf7 plants contain elevated transcript levels for DNA repair and recombination genes (Figure 7; Additional file 1: Figure S7) [17]. While the vegetative defects in 35S:NTAP:AtCTF7 plants may result from spontaneous mutations, the situation is clearly more complex as the dwarf phenotype is not inherited in a Mendelian fashion (Table 1). When dwarf plants were selfed they produced a mixture of dwarf and non-dwarf, reduced fertile plants. Further, the frequency of dwarfs in the progeny of selfed dwarf plants was similar to the frequency of dwarf plants resulting from selfing of nondwarf plants. This raised the possibility that epigenetic alterations may be present in 35S:NTAP:AtCTF7ΔB plants. Consistent with this possibility is our observation that transcript levels of MU1, COPIA 28 and solo LTR were increased between five (MU1) and 24 fold (COPIA 28) in 35S:NTAP:AtCTF7ΔB plants ( Figure 7A). Subtle changes were observed in the transcript levels of several siRNA associated genes. AGO1, RDR2 and mir156 transcript levels were reduced approximately 40-60% while AGO4 was elevated slightly. HDA19 and RDM4 transcripts were also decreased approximately 50% ( Figure 7B).
Our observation that epigenetic alterations may be present in 35S:NTAP:AtCTF7ΔB plants is consistent with the alterations we observe in female reproduction. Recent studies have shown that embryo sacs are enriched for transcripts of proteins involved in RNA metabolism and transcriptional regulation, and that they display distinct epigenetic regulatory mechanisms [57][58][59]. Disruption of genes in small RNA regulatory pathways, such as AGO1, AGO9, DICER-LIKE1 (DCL1) and MEIOSIS ARRESTED AT LEPTOTENE1 (MEL1), leads to multiple gametic cells at premeiosis, abnormal meiotic divisions, gametic cell fate alterations and twin female gametophytes [33,59,60]. For example, mutations in AGO9, which participates in small RNA silencing by cleaving endogenous mRNAs, results in additional gametic cells in pre-meiotic ovules, which may skip meiosis and twin female gametophytes in postmeiotic ovules [33]. Several of these defects are observed in 35S:NTAP:AtCTF7ΔB ovules before and during meiosis. Moreover, AGO9 participates in the epigenetically regulated silencing of TEs [33]. At this time it is not clear if the apparent epigenetic alterations we observed are the direct result of high-level NTAP:AtCTF7ΔB expression or a secondary effect. For example, it is possible that NTAP: AtCTF7ΔB expression directly affects the expression of genes involved in epigenetic regulation. The involvement of cohesin complexes in transcriptional regulation is well documented in other organisms [54]. It is also possible that the changes we observe are an indirect effect of NTAP:AtCTF7ΔB expression. For example, previous studies have shown that eco1/ctf7 mutations result in defects in nucleolar integrity, rRNA production, ribosome biogenesis and protein biosynthesis in Saccharomyces cerevisiae and human [25,61]. In Arabidopsis, mutations in genes participating in mRNA production and rRNA/ribosome biogenesis slow mitotic progression in female gametophytes and result in pleiotropic defects in embryo sacs [58,59,[62][63][64][65][66]. For example, mutations in SLOW-WALKER1 (SWA1), which participates in 18S pre-rRNA processing, results in asynchronous megagametophyte development, and embryo sac arrest over a wide range of stages [62]. Likewise, mutations in ribosomal protein genes lead to defects in inflorescence, leaf and plant stature in Arabidopsis, similar to those observed in 35S:NTAP:AtCTF7ΔB plants [66][67][68][69]. Therefore, many of the alterations we observe in 35S: NTAP:AtCTF7ΔB plants could be the result of alterations in rRNA biogenesis or ribosome biogenesis, which in turn could indirectly impact epigenetic pathways.

AtCTF7ΔB likely acts on several levels
The alterations observed in 35S:NTAP:AtCTF7ΔB plants appear to result from the presence of high levels of AtCTF7ΔB and not a reduction of native AtCTF7 levels or the presence of the NTAP. 35S:AtCTF7ΔB plants exhibit similar, if not more dramatic phenotypes than 35S: NTAP:AtCTF7ΔB plants, indicating that the NTAP is not responsible for the observed phenotypes. Likewise, expression studies show that in addition to high AtCTF7ΔB transcript levels, transgenic plants also contain elevated levels of native AtCTF7 transcripts ( Figure 1C). Therefore, the 35S:AtCTF7ΔB construct does not cause co-suppression. Consistent with this are the apparently normal SYN1 cohesin patterns observed in male meiocytes (Additional file 1: Figure S2). Interestingly, the elevated levels of native AtCTF7 transcript suggest that the cellular cohesion status is altered to some extent in AtCTF7ΔB plants and that a feedback loop exists to monitor and maintain cohesion levels in plants.
High-level expression of AtCTF7ΔB appears to exert a dominant negative effect, resulting in a relatively wide range of alterations. Less clear is how high level expression of AtCTF7ΔB exerts its effect or if the alterations we observe are all related. Deletion of the last 46 amino acids of the acetyltransferase domain is expected to eliminate most of the actyltransferase activity. High-level expression of the protein may directly compete with native AtCTF7 for cohesin substrates resulting in an overall reduction or redistribution of cohesin levels throughout the genome. These changes could in turn result in a wide range of transcriptional alterations, similar to the situation observed in other organisms [54]. While this is the most-likely effect, it may not explain all of the observed alterations. For example, to our knowledge apparent epigenetic alterations have not been observed in either AtCTF7 RNAi or atctf7 plants [17,18]. Therefore, it is also possible that the 46 amino acid deletion alters acetyltransferase specificity such that the protein acts on off targets. For example, CTF7 has been shown to not acetylate histones; however if altered substrate specificity resulted in the acetylation of histones, then changes in chromatin structure could produce some of the alterations we observe. Finally, the possibility also exists that the deletion may alter the interaction of AtCTF7 with other proteins, either directly or indirectly involved in maintaining chromatin structure. Further experiments are required to determine how specifically AtCTF7ΔB is acting, why male and female reproduction respond differently to alterations in AtCTF7 levels and what role, if any AtCTF7 plays in epigenetic regulation.

Conclusions
Proper levels of AtCTF7 are critical for proper plant growth and development with female gametophytes being most sensitive to changes in AtCTF7 activity. High level expression of NTAP:AtCTF7ΔB results in pleiotropic defects in reproduction and vegetative growth. High levels of AtCTF7ΔB may affect small RNA processing, which in turn appears to result in epigenetic alterations. These results indicate that CTF7 may play multiple roles in plant cells.

Plant material and growth conditions
Wild type Arabidopsis thaliana plants (ecotype Columbia), the SALK_059500 (ctf7-1) insertion line and all transgenic plants described in this report were grown in Metro-Mix200 soil (Scotts-Sierra Horticultural Products; http:// www.scotts.com) in a growth chamber at 22°C with a 16-hlight/8-h-dark cycle as described [16]. T-DNA insertion and transgenic plants were genotyped by PCR with primer pairs specific for the T-DNA and wild-type loci.

Cloning procedures for construction of the transgenic plants
The AtCTF7ΔB cDNA fragment (1-894 bp nucleotides, AtCTF7 Δ299-345 ) was digested with NdeI/HindIII and cloned into pIADL14 as described [16]. The 35S:NTAP:AtCTF7ΔB construct was generated using Gateway-compatible binary vectors containing the NTAP tag ((NTAPi) [30]; a gift from Dr. Qinn Li lab, Miami Univeristy). AtCTF7ΔB, 1-894 bp cDNA nucleotides, was first PCR-amplified with primers (1111/1201) and cloned into the pENTR vector, which was then fused with the binary vector by LR recombination reactions (Invitrogen) to make the final 35S:NTAP:AtCTF7ΔB construct. 35S:AtCTF7ΔB was generated by cloning the AtCTF7 cDNA (1-894 bp) as a NcoI/SpeI fragment into the binary vector pFGC5941. The 35S:NTAP:AtCTF7 construct was generated by the Gateway method with the corresponding primers. The AtCTF7 promoter (1.3 kb upstream of the ATG) was amplified, digested and cloned into the 35S: AtCTF7ΔB construct. All the constructs were confirmed by DNA sequencing. The primers used are listed in Additional file 1: Table S4.
Each construct was mobilized into Agrobacterium tumefaciens strain AGL-1 and transformed into Arabidopsis thaliana using the floral dip method [70]. Transgenic plants were screened by BASTA and further confirmed by PCR.

Quantitative real-time RT-PCR (qRT-PCR)
Buds of wild type and 35S:NTAP:AtCTF7ΔB Line 11 4th generation plants were harvested and pooled separately. For 35S:NTAP:AtCTF7ΔB samples, buds were only harvested from reduced fertile, non-dwarf plants or dwarf planys separately. Total RNA was extracted with the Plant RNeasy Mini kit (Qiagen, Hilden, Germany), and 10 μg of RNA was treated with Turbo DNase I (Ambion, http:// www.invitrogen.com/site/us/en/home/brands/ambion.html) and used for cDNA synthesis with an oligo(dT) primer and a First Strand cDNA Synthesis Kit (Roche, http://www. roche.com). PCR was performed with the SYBR-Green PCR Mastermix (Bio-Rad, Hercules, CA, USA) and amplification was monitored on a MJR Opticon Continuous Fluorescence Detection System (Bio-Rad). Expression was normalized against β-tubulin-2. At least three biological replicates were performed, with two technical replicates for each sample. Student's t-test was conducted to identify transcripts that exhibit statistically significant variation at the 95% confidence level. Sequences of primers used in these studies are presented in Additional file 1: Table S4.