Evidence that High Activity of Vacuolar Invertase Is Required for Cotton Fiber and 1 Arabidopsis Root Elongation through Osmotic Dependent and Independent Pathway, 2 Respectively

2 Vacuolar invertase (VIN) has long been considered as a major player in cell expansion. However, 3 direct evidence for this view is lacking due to, in part, the complexity of multi-cellular plant tissues. 4 Here, we used cotton fibers, fast growing single-celled seed trichomes, to address this issue. VIN 5 activity in elongating fibers was ~4-6-fold higher than that in leaves, stems and roots. It was 6 undetectable in fiberless cotton seed epidermis but became evident in initiating fibers and remained 7 high during their fast elongation and dropped when elongation slowed. Further, genotype with 8 faster fiber elongation had significantly higher fiber VIN activity and hexose levels than that 9 slow-elongating genotype. By contrast, cell wall or cytoplasmic invertase activities did not show 10 correlation with fiber elongation. To unravel the molecular basis of VIN-mediated fiber elongation, 11 we cloned GhVIN1 that displayed VIN sequence features and localized to vacuole. Once introduced 12 to Arabidopsis , GhVIN1 complemented the short root phenotype of a VIN T-DNA mutant and 13 enhanced the elongation of root cells in the wild-type. This demonstrates that GhVIN1 functions as 14 VIN in vivo . In cotton fiber GhVIN1 expression level matched closely with VIN activity and fiber 15 elongation rate. Indeed, transformation of cotton fiber with GhVIN1 RNAi or over-expression 16 constructs reduced or enhanced fiber elongation, respectively. Together, the analyses provide a set of evidence on the role of VIN in cotton fiber elongation mediated by GhVIN1. Based on the relative contribution of sugars to sap osmolality in cotton fiber and Arabidopsis root, we conclude that VIN regulates their elongation in an osmotic dependent and independent manner, respectively. invertase


Introduction 1
Sucrose is the principal end product of photosynthesis in higher plants and the major carbohydrate 2 translocated from source to sink tissues through phloem. Sucrose cleavage, serving as a starting 3 point for various carbohydrate metabolic pathways, is catalyzed by sucrose synthase (EC 2.4.1.13,4 Sus) and invertase (β-fructofuranosidase; EC 3.2.1.26). In contrast to the reversible reaction of Sus, 5 invertase irreversibly hydrolyzes sucrose to fructose and glucose. This hydrolysis step is required 6 for the development of many sink tissues and their responses to various stresses (Sturm, 1999; Importantly, in comparison with that of the short fiber genotype Gh, the higher VIN activity in Gb 1 (Fig 2A) corresponds to 50-60% increase in fiber glucose and fructose concentrations and 35% 2 increase in sucrose levels ( Fig 2B) at 10 DAA. The hexose levels were about halved at the slow 3 elongation phase of 15 DAA (Figs 2B). Noteworthy is that glucose and fructose were about equal in 4 concentration whereas sucrose was less than 10% of the hexoses (Fig 2B), consistent with an active 5 role of VIN in elongating fiber. The high VIN activity and hexose level in Gb corresponded 6 increased fiber length at 10 DAA onwards as compared to that of Gh ( Fig 2C). However, the faster 7 elongation rate in Gb, as indicated by a higher slope, appeared from 5 DAA onwards ( Fig 2C). The strong developmental and genotypic correlation between VIN activity and fiber elongation 1 1 (Figs 1 and 2) inspired us to explore the molecular basis of VIN activity in cotton fiber. To achieve 1 2 this, we screened a cotton cDNA library prepared from fiber and ovule and identified a full length 1 3 clone, designated GhVIN1. TBLASTN and BLASTP searches showed a much higher similarity of 1 4 GhVIN1 with plant VINs than that of CWINs. For example, GhVIN1 shared 70%, 68% and 66% 1 5 amino acid identities with VIN from Lagenaria siceraria, Pyrus pyrifolia and Prunus cerasus 1 6 (Genbank accession number of AF519809, BAG30919 and AAL05427, respectively) but only 37%  Further bioinformatics analyses revealed that GhVIN1 exhibited thirteen regions conserved in 2 0 known acid invertases, including the characteristic β -fructosidase motif (NDPD/NG) and cysteine 1 0 OsINV3 and ZmIvr1, had a site of "LL/ PLP", followed by a strong basic region, a hydrophobic 1 transmembrane segment, and a conserved motif (Supplemental Fig 2). These are VIN hallmarks 2 (Vitale and Chrispeels, 1992;Balk and de Boer, 1999;Sturm, 1999) which are absent from the two 3 CWINs, OsCWI1 and NtCWIN1, included in the comparison. Furthermore, GhVIN1 displayed five 4 distinctive amino acids (indicated by arrows in Supplemental Fig 2) that are highly conserved in 5 VINs but are different from that in CWINs (see Ji et al., 2005). 6 Consistently, phylogenetic analyses revealed that GhVIN1 was clustered in the plant VIN group, 1 1 To confirm its vacuolar localization, a GhVIN1:RFP fusion construct was co-bombarded with 1 a tonoplast magnesium transporter, ShMTP, fused with GFP (Delhaize et al., 2003) into onion 2 epidermis. Since these cells were heavily vacuolated, the tonoplast would be pushed closely against 3 the plasma membrane and the cell wall. To eliminate the possible ambiguity of signals in intra-and 4 extra-cellular spaces, bombarded samples were plasmolysed before confocal imaging. The 5 plasmolysis allowed the protoplasm to be pulled away from the cell wall that exhibited no 6 GhVIN1:RFP signals (Figs 3D and E). In contrast, the GhVIN1 signals were clearly detected in the  T-DNA insertion in the first exon of Atβfruct4 (Fig 4A) resulted in a short-root phenotype 1 9 ( Sergeeva et al., 2006), hence, representing an ideal system for testing the functionality of GhVIN1.

0
Here, a 35S:GhVIN1 over-expression (OE) construct was transformed into the homozygous KO 2 1 line (inv). The T-DNA insertion in Atβfruct4 was confirmed by PCR analysis (Fig 4B), as well as visible phenotype was observed in both complementation and over-expression lines.

1
To gain some insights into the GhVIN1-mediated effect on Arabidopsis root elongation, sugar 2 levels were assayed in roots. The analysis revealed lower hexoses and higher sucrose levels in the 3 KO mutant (inv-7) as compared to the wild-type plants. The ectopic expression of GhVIN1 4 recovered and increased hexose level, thus, the ratio of hexose to sucrose concentration, in the 5 mutant and wild-type background of C14 2-1 and OE 5-8, respectively (Figs 5A and B). planta. Further studies were therefore conducted to explore the role of GhVIN1 in cotton fiber 1 0 elongation. Semi-quantitative RT-PCR analyses showed that the GhVIN1 mRNA was highly 1 1 expressed in 10-d fibers and seeds, but was barely detectable in roots, stems and cotyledons (Fig compared to the sense control ( Fig 6F) at 0 DAA when fibers are initiating from the seed epidermis 1 7 as shown by toluidine blue staining ( Fig 6E). Importantly, the GhVIN1 mRNA signals were much 1 8 stronger in the initiating fibers than that in the adjacent non-differentiated epidermal cells or 1 9 underlying seed coat cells (Fig 6G), which is consistent with that of histochemical staining for 2 0 invertase activity (see Fig 1B and suppl Fig 1).

1
Developmentally, GhVIN1 transcript was more abundant in fibers at the fast-elongation phase 2 2 of 5-10 DAA than that at the slow elongation phase of 15-20 DAA ( Fig 6H). Genotypically, GhVIN1 was expressed more in faster elongating fibers from Gb than that from Gh with 1 slower-elongating fibers ( Fig 6I,  only construct was used as a void-vector control. compared to that in the micropyle end (Ruan, 2005). The randomization of the transformed and untransformed cells in the region further minimized the potential positional effect on the 1 comparison (see below).

2
In total, 50 and 36 transformed fibers derived from 25 and 27 cotton seeds, along with equivalent 3 number of adjacent untransformed cells, were identified and measured for fiber length and enzyme 4 activity after bombarded with RNAi and OE constructs, respectively (see Materials and methods for 5 details). Figure 7C shows that, in comparison to those untransformed fibers, transforming with 6 GhVIN1 RNAi construct reduced VIN activity by 33%, which led to a reduced fiber length ( Fig 7D) 7 On the other hand, fibers expressing GhVIN1 OE construct increased VIN activity by 45% (Fig 7C), 8 resulting in an increased fiber length ( Fig 7D). In contrast, transforming with the void-vector did 9 not evoke a statistical difference as compared to the non-transformed cells in either VIN activity or OE construct affects VIN but not CWIN or CIN activities in cotton fibers. Rapid fiber elongation requires high activity of VIN, probably mediated by GhVIN1 2 0 We present here several lines of evidence that high activity of VIN contributes to rapid cotton fiber 2 1 elongation. Developmentally, VIN activity (Fig 1) and GhVIN1 transcript level ( Fig 6H) was higher 2 2 at the rapid expansion phase of 5 and 10 DAA and became evidently lower at the slow elongation phase of 15-20 DAA. Consistent with the potential role of VIN in fiber elongation, a fiberless 1 mutant did not exhibit VIN activity in its seed epidermis ( Supplementary Fig 1). Genotypically, 2 higher fiber VIN activity (Fig 2A) and GhVIN1 mRNA level ( Fig 6I) correlate with faster fiber 3 elongation rate ( Fig 2C). Importantly, the correlation with fiber elongation is specific to activities of 4 VIN but not CWIN or CIN (Figs 1 and 2), highlighting the role of VIN in this cell expansion 5 process. Finally, the causality between VIN and fiber elongation was shown in bombardment 6 experiments, where transformation with GhVIN1-silencing or over-expression constructs reduced or 7 enhanced, respectively, VIN activity and fiber elongation (Fig 7). Collectively the data show that 8 VIN likely plays an important role in cotton fiber elongation.

9
The notion that GhVIN1 is probably a major vacuolar invertase underpinning the observed VIN 1 0 activity in cotton fiber is supported not only by developmental and genotypic correlation between 1 1 GhVIN1 transcript level (Fig 6) and VIN activity (Figs 1 and 2) as discussed above but also by the recovered the VIN activity to wild-type level and consequently complemented the short-root 1 9 phenotype (Fig 4).

0
The above findings are of significance for two reasons. First, VIN has long been considered as a the osmotic contribution of sucrose, thus has the potential to positively impact on cell turgor.

6
In this context, glucose and fructose concentrations ( Fig 2B) account for about ~50% and 40% of fibers from Gb than that of Gh since fiber water potentials are virtually identical between the two genotypes (Fang and Ruan, unpublished data). This degree of turgor increment could 1 significantly enhance cotton fiber elongation (Ruan et al., 2001) as evidenced by the faster 2 elongation observed in the Gb relative to Gh from 5 DAA onwards ( Fig 2C) when VIN activity was 3 significantly higher in the former than the latter (Fig 2A). 4 It seems paradox that the fiber VIN activity dropped after 5 DAA (Fig 1D), whereas rapidest 5 fiber elongation has been observed to occur from 10 to 15 DAA (Ruan et al., 2001). Here, despite 6 its decrease after 5-d, VIN activity remained ~ two-fold higher in 10-d fiber than that at 15-20 DAA 7 ( Fig 1D), hence could promote fiber elongation from 10-15 DAA through osmotic regulation as influx of water, hence contribute to the rapid elongation of fibers at 10-15 DAA. VIN regulates Arabidopsis root elongation independent of osmotic regulation 1 Our finding that the cotton GhVIN1 complemented the short-root phenotype of a vin knockout line 2 in Arabidopsis concurred with a previous report by Sergeeva et al., (2006). Notably, once 3 over-expressed in wild-type Arabidopsis, GhVIN1 enhanced root elongation and root hair size 4 (Supplemental Fig 4), indicating normal growth of Arabidopsis root is limited by VIN activity and 5 the utility of GhVIN1 for improving root development by using genetic engineering approach.

6
Regarding to the mechanism of VIN-regulated Arabidopsis root elongation, it has been proposed 7 that VIN may control the process through osmotic regulation (e.g. Sergeeva et al., 2006). This GhVIN1-overexpresion lines, respectively, which in turn exhibit short, medium and long root 1 5 length (Fig 4 and Supplemental Fig 4). These findings contradict to the hypothesis of VIN-mediated 1 6 osmotic regulation of Arabidopsis root elongation.

7
Cell expansion depends on relaxation of cell wall matrix once cell turgor is above a threshold 1 8 value. This raises a possibility that VIN may promote Arabidopsis root elongation by impacting on context, hexose and sucrose levels from mature leaves of the mutant plants were about 30% lower 8 and 25% higher, respectively than that the wild-type whereas over-expression of GhVIN1 increased 9 hexose level by ~33% in the mature leaves with no effect on sucrose levels (Supplemental Fig 5).
1 0 These changes might affect phloem loading and sucrose import to sinks such as roots, thereby Invertase activity assay and localization and sugar measurement 2 0 Fresh tissues were immediately frozen and ground to powder in liquid nitrogen, homogenized three 2 1 times with 1.5ml extraction buffer (see Tomlinson et al., 2004) in total, and centrifuged at 14000 g For invertase activity assay in transgenic cotton fibers after bombardment, fibers were detached 3 from cotton seeds on ice following GUS staining and photographing (for measuring fiber length, 4 see below). The transformed (indicated by GUS blue) and adjacent non-transformed fibers were 5 collected for assay. About 10-20 fibers for each of the two groups were sampled from 8-10 seeds as 6 one biological sample for VIN activity assay. The results of three biological replicates were used for 7 statistic analysis.

8
Sugar content was measured enzymatically as described by Stitt et al. (1989).

9
For invertase activity and sugar content assay, each biological replicate consisted of three 1 0 technical replicates. These experiments were repeated at least three times and similar results were 1 1 obtained. One representative set of data was used for analyses described in the paper. sucrose] in at 30 ºC for 1h. In control reactions, sucrose was omitted. Note, the assay does not 2 0 differentiate invertase activity from different sub-cellular compartments.

Measurement of cotton fiber length, Arabidopsis root cell and root-hair length and sap
1 osmolality 2 For fiber length comparison between genotypes Gb and Gh, the fibers were free-hand harvested 3 from seed at specified stages and measured after relaxing and straightening fibers according to root-hairs located in the middle of root-hair zones were measured using Image J program. About 1 5 120 cells were measured from six biological replicates (seedlings) for each case (see Fig 4E).

6
Sap osmolality of cotton fiber and Arabidopsis root was measured using a vapor pressure  Agrobacterium-mediated transformation using the floral dip method.

1
Primers used for making the above constructs were listed in Supplemental Table 2. In situ hybridization analyses of 0 and 1 DAA cotton seeds were performed according to Xu et al.

Semi-quantitative reverse transcription (RT)-PCR analyses 2 1
Total RNA was isolated from cotton fiber and other tissues according to Ruan et al., (1997). For 2 2 each reaction, 1μg RNA was reverse-transcribed to cDNA with an oligo (dT) primer. Gene specific primers (GhVIN1-RT) were designed to amplify a 684 bp fragment of GhVIN1. The cotton 1 18srRNA gene was used as an internal control, amplified with Gh18srRNA gene RT primers.

2
For RT-PCR analyses in Arabidopsis, total RNA was isolated using Trizol from Invitrogen and 3 reverse-transcribed as previously described. Atβfruct3 and Atβfruct4 transcript was determined 4 using Atβfruct3-RT and Atβfruct4-RT primers, with Arabidopsis tubulin1 (AtTUB) as an internal 5 control, amplified with AtTUB-RT primers.

6
All PCR conditions were optimized, and mentioned in Supplemental Table 2.  In each case, transformed fibers were identified by GUS staining (see Figs 7 A to C). Transformed 7 and adjacent untransformed fibers were dissected separately from the seed for measurements.  Each value in (A), (C) and (D) is the mean ± SE of at least four biological replicates. Different  (arrowheads) and cytoplasm adjacent to cell wall that appeared in rectangle shape (arrows).  calculated root length (each value is the mean ± SE of eighteen seedlings), respectively.
1 5 (E) The short root cell phenotype of inv-7 and C5 null was fully and partially rescued in 1 6 GhVIN1-complemented lines C7 3-2 and C14 2-1 respectively, compared to the wild-type level. activities remain almost unaltered among these lines. Each value is the mean Different letters in (D), (E) and (F) indicate significant differences at P ≤ 0.05 according to 1 randomization one-way ANOVA test. Each value is the mean ± SE of at least four biological replicates. Different letters indicate 1 0 significant differences at P ≤ 0.05 according to randomization one-way ANOVA test.  isc, inner seed coat; osc, outer seed coat. (A) Invertase activity in different cotton tissues. R, root; S, stem; C, cotyledon; L, source leaf. F, Sc and Ft represent 10-d fiber, seed coat and filial tissue, respectively; CIN, cytoplasm invertase; VIN, vacuolar invertase; CWIN, cell wall invertase. (B) Localization of invertase activity in whole cotton ovule and seed showing dark blue signals of invertase activity present in fiber-enriched epidermis at 0 and 1 DAA seeds but not in epidermis of -1 DAA ovule prior to fiber initiation (see Results for more detail). Bar= 1 mm (C) Invertase activity assay in ovule/ seed extracts from -1 to +1 DAA. (D) Invertase activity in fibers during elongation. Each value in (A), (C) and (D) is the mean ± SE of at least four biological replicates. Different letters in (C) indicate significantly different at P ≤ 0.05 according to randomization one-way ANOVA test.    (A) and (B). In (C) and (D), each value is the mean ± SE three biological replicates. Asterisks indicate significant differences (student-t test, * P ≤ 0.05; ** P ≤ 0.01) between transformed and non-transformed fibers.