Improved resistance to controlled deterioration in transgenic seeds

We show that seed-specific overexpression of the sunflower ( Helianthus annuus ) HaHSFA9 heat stress transcription factor (HSF) in tobacco ( Nicotiana tabacum ) enhances the accumulation of heat-shock proteins (HSPs). Among these proteins were HSP101 and a subset of the small HSPs (sHSPs), including proteins that accumulate only during embryogenesis in the absence of thermal stress. Levels of Late Embryogenesis Abundant proteins or seed oligosaccharides, however, were not affected. In the transgenic seeds, a high basal thermotolerance persisted during the early hours of imbibition. Transgenic seeds also showed significantly improved resistance to controlled deterioration in a stable and transgene-dependent manner. Furthermore, overexpression of HaHSFA9 did not have detrimental effects on plant growth or development, including seed morphology and total seed yield. Our results agree with previous work tentatively associating HSP gene expression with phenotypes important for seed longevity. These findings might have implications for improving seed longevity in economically important crops. enzymatic determination of sugar composition, and showed that raffinose was the sole oligosaccharide from the raffinose family (RFO) detected in mature tobacco seeds. Using HPLC, we determined that the sucrose content in transgenic seeds averaged 1.38% ± 0.1% (sugar contents expressed as percent of total seed fresh weight). A similar amount of sucrose was measured in non-transgenic seeds (1.36% ± 0.2%). The raffinose content in transgenic seeds was 0.33% ± 0.03%, with a similar abundance of


INTRODUCTION
Mature seeds of most plants (the so called orthodox seeds) withstand extreme desiccation and temperature conditions, but only when preserved in the very dry state reached during zygotic embryogenesis (typical moisture content fresh weight [MCFW] ≤ 5%). The industrial conservation of seeds is detrimentally influenced by unintended rehydration and temperature increases (McDonald, 1999;Halmer, 2000), which also lead to stress-induced damage and to inefficient germination. Germination efficiency and seed longevity involve the expression of multiple genes. Attempts to identify such genes have involved the use of mutants (Ooms et al., 1993;Clerkx et al., 2004a;Sattler et al., 2004), or analyses of allelic variation in model plants (Clerkx et al., 2004b) and crops (Miura et al., 2002). These studies revealed the genetic complexity of these traits, as well as the regulatory genes (Ooms et al., 1993;Clerkx et al., 2004a) and enzymatic activities (Sattler et al., 2004) involved. To date, only genes that reduce seed longevity have been described, including mutants in Arabidopsis thaliana with pleiotropic defects in developing seeds and during germination.
However, there is considerable variation in seed longevity among accessions of Arabidopsis (Clerkx et al., 2004b). Therefore, natural genetic diversity for seed longevity exists, and this diversity could be exploited to improve longevity.
Alternatively, and towards this aim, key transcription factors with specific and multiple effects on the genes involved in longevity are good candidates to be tested in transgenic approaches. Among the genes with potential roles in seed longevity are those coding for small heat-shock proteins (sHSPs) (Scharf et al., 2001), since they contribute to different processes that have been associated with seed longevity (Wehmeyer and Vierling, 2000;Sun et al., 2002;Tsvetkova et al., 2002) such as thermotolerance, tolerance to embryo-desiccation, membrane stabilization and oxidative stress resistance. Furthermore, mutants with reduced seed longevity also show impaired expression of sHSP genes in embryos (Wehmeyer and Vierling, 2000;Sun et al., 2002). Previously, we have shown the transcription factor HaHSFA9 to be specifically involved in the developmental regulation of sHSP genes in sunflower (Helianthus annuus) embryos (Almoguera et al., 2002).
Here, we have tested the effects of seed-specific overexpression of promoter and additional 5'-and 3'-flanking sequences from HaDS10G1 (DS10), which is an unusual Late Embryogenesis Abundant (LEA) gene -of group 1 (Wise 2003)-expressed in sunflower seeds from mid-maturation. The DS10 promoter is not only highly efficient in seeds, but also is seed-specific except for a marginal expression in pollen (Prieto-Dapena et al., 1999;Rousselin et al., 2002, and our unpublished observations). We found that the commonly used cauliflower mosaic virus 35S (CaMV35S) promoter confers constitutive expression levels that, in seeds, are two orders of magnitude lower than when the DS10 promoter is used. Using the DS10 promoter, we observed specific changes in gene expression induced by overexpression of HaHSFA9.
These changes resulted in an increase of seed longevity, as determined by controlled deterioration tests (CDT, McDonald, 1999;Halmer, 2000;Clerkx et al., 2004b). The observed phenotypes were stable over at least two generations, and they segregated with the DS10:HaHSFA9 transgene. Adverse effects on plant growth, morphology or seed production were not observed. The specific, and harmless, effects of HaHSFA9 overexpression represent a novel example of genetically improved resistance to CDT of seeds. We discuss how the observed effects involve the temporal extension of the high thermotolerance of mature dry seeds to the early stages of seed imbibition. The identification of HaHSFA9 as a transcription factor with positive effects on resistance to CDT opens new possibilities for improving seed longevity.  The DS10:A9 plants did not show ectopic HSP expression in seedlings 3 to 4 weeks after germination (data not shown). However, HaHSFA9 induced the overexpression of different HSP genes in mature seeds, including CI and CII sHSP genes (that encode two classes of cytosol-localized proteins, Scharf et al., 2001), andHSP101 (Queitsch et al., 2000). The transgene dependence of such an effect is illustrated with seeds from two different homozygous lines compared to their respective non-transgenic siblings (Fig. 1). We further show that HaHSFA9 caused the overexpression of the genes encoding all CI proteins also present in the non-transgenic mature seeds under control conditions (Fig.   1B). Comparison with heat-stressed samples from seedlings indicated the presence in seeds of specific sHSPs that were absent or barely detected after heat stress. These seed-specific polypeptides were upregulated in seeds of the DS10:A9 plants, as exemplified for one CI sHSP (arrow in Fig. 1). We also found that the transgene induced the accumulation of most of the CII sHSPs present in seeds. However, in this case the accumulation changes induced by HaHSFA9 were not as extensive as shown for the CI sHSPs, and the levels of the CII seed-specific polypeptides were unchanged in the transgenic seeds ( Fig. S2, spots labeled s). The effects of HaHSFA9 on gene expression were very specific. As such, the DS10:A9 transgene did not affect the levels of dehydrin proteins or the total amount or composition of soluble sugars in seeds ( Fig. 2). We did not find any significant difference among transgenic and nontransgenic seeds for sugar contents or for any of the carbohydrate content estimates (F < 1.304, P > 0.25, 1 and 14 degrees of freedom [df], for all possible a posteriori comparisons). High-performance liquid chromatography (HPLC) analyses confirmed the enzymatic determination of sugar composition, and showed that raffinose was the sole oligosaccharide from the raffinose family (RFO) detected in mature tobacco seeds. Using HPLC, we determined that the sucrose content in transgenic seeds averaged 1.38% ± 0.1% (sugar contents expressed as percent of total seed fresh weight). A similar amount of sucrose was measured in non-transgenic seeds (1.36% ± 0.2%). The raffinose content in transgenic seeds was 0.33% ± 0.03%, with a similar abundance of this sugar also observed in non-transgenic seeds (0.29% ± 0.03%). Glucose was at concentration undetectable by our HPLC assay.

Overexpression of HaHSFA9 in transgenic tobacco plants
Additional analyses using specific probes for LEA-protein (Wise, 2003) mRNAs of groups 2 (dehydrins), 1, 3, or 4 did not reveal transgene effects on their regulation. Total proline content in seeds was also unaltered (data not shown).

Persistence of basal thermotolerance in the DS10:A9 seeds
We first investigated whether the specific effects of HaHSFA9 on gene expression modified the basal thermotolerance of imbibing seeds from T 0 lines.
As negative controls, we used seeds either from non-transgenic plants, from lines with different, unrelated transgenes, and from 35S:A9 lines. All control lines contained the same marker gene (conferring kanamycin-resistance). The persistence of basal thermotolerance was assayed by determining germination percentages after the high-temperature treatments (4h at 50ºC). In the dried state reached upon natural seed maturation, 100% of the seeds from the control lines resisted the 50ºC treatments (the same was true for the DS10:A9 seeds, data not shown). The difference between both kinds of seeds was revealed after a short rehydration, which raised MCFW to 41.3% ± 0.4%. Control seeds lose their thermotolerance but the DS10:A9 seeds retain it substantially. Mendelian segregation analyses strongly suggested that the thermotolerant phenotype was linked to the DS10:A9 transgene (Fig. 3). The analyses with seeds from the T 0 plants indicated a clear effect of the DS10:A9 transgene. The percent of germination of control seeds was reduced from 95-100% (observed before treatment) to 0-6% (Fig. 3B). This reduction was observed 4 to 7 days after transferring the treated seeds to germination conditions following the 50ºC treatment. In contrast, the DS10:A9 seeds resisted the treatment and germinated much better, reaching average germination of 24% ± 5% after 7 days (Fig. 3A). The observed differences were significant (F = 6.99, P = 0.015, 1 and 11 df). Resistance to kanamycin in transgenic seeds was evaluated before and after the 50ºC treatments (with the seeds that completed germination). Figure 3B shows that values close to the expected 3:1 ratio, between antibiotic resistance and sensitivity, were observed before the treatment both for control and DS10:A9 seeds. However, segregation  The expression of HaHSFA9 from the 35S promoter caused minor effects on HSP accumulation only observed after transgene homozygosis (Fig. 3C). These effects were much smaller than observed for the DS10:A9 lines (compare Fig.   3C with Fig. 1A).
The persistent-thermotolerant phenotype was confirmed after transgene segregation in the subsequent generation. This was demonstrated with the same material used for the gene expression analyses (compare Figs. 1 and 4).
Furthermore, the levels of HSP accumulation observed for transgenic and nontransgenic seeds were unaltered by the 50ºC treatments. Therefore, these treatments did not induce additional HSP accumulation in either seed type ( Fig.   S3). Seeds carrying the transgene resisted the 50ºC treatments ( Fig. 4), but sibling seeds without the transgene did not as previously observed with control seeds (see also Fig. 3A). We found highly significant differences among transgenic and control plants for germination percentage (F = 44.93, P < 0.0001, 1 and 66 df) and cotyledon expansion (F = 45.32, P < 0.0001, 1 and 11 df). Additional analyses indicated that the seeds that do not complete germination after the treatments were dead; as for example these seeds did not stain using tetrazolium (data not shown). Among the seeds that survived the 50ºC treatments, transgenic seeds completed germination earlier than their respective non-transgenic siblings. This indicates a faster recovery after damage induced by the treatment, which is consistent with other seedling establishment parameters, such as the higher percentage of cotyledon expansion observed for the transgenic seeds (Fig. 4A

Resistance to controlled deterioration of the DS10:A9 seeds and lack of adverse phenotypic effects caused by overexpression of HaHSFA9
Resistance of seeds to a controlled deterioration treatment (CDT) has been successfully used for the rapid evaluation and prediction of seed longevity (Powell, 1995;TeKrony, 1995;McDonald, 1999;Halmer, 2000;Clerkx et al., 2004aClerkx et al., , 2004bSattler et al., 2004). CDT is achieved after the exposure of rehydrated seeds to high temperatures, which leads to rapid decline of seed vigor and loss of viability (determined as a negative effect on seed germinability). Conceptually, CDT resembles assays for basal thermotolerance performed with imbibing seeds. The difference between them is the temporal separation between seed rehydration and the incubation at high temperatures.
That separation is not required for basal thermotolerance assays, but it is usual in CDT performed with small-sized seeds (Powell, 1995;McDonald, 1999; see

Materials and Methods for details).
Seeds from the same lines analyzed for basal thermotolerance in Figure 4 were subjected to CDT suited for tobacco seeds, under conditions similar to those reported in the literature for other small-sized seeds (Powell, 1995; CDT). Statistical analyses confirmed significant differences between the germination percentages of transgenic and non-transgenic seeds during the whole time-span of the experiment (4 to 16 days after CDT, [F = 68.77, P < 0.0001, 1 and 240 df; repeated-measures ANOVA]). CDT had a milder effect on seed germination percentages than the treatments used for the basal thermotolerance assays, and the difference between the transgenic and sibling non-transgenic lines was therefore more evident (compare the results of Figs. 4 and 5 and the statistics for the data in each Figure). Therefore, resistance to CDT is also associated with inheritance of the DS10:A9 transgene.
The longevity of seeds from transgenic and sibling non-transgenic lines was estimated by performing CDT in which we exposed seeds at 50ºC for different times between 0.5 and 3 days. This allowed us determining their respective LD 50 (the number of days of CDT required to decline to 50% germination), a single parameter related to longevity. Additional homozygous transgenic (DS10:A9#19-4, DS10:A9#22-11) and sibling non-transgenic lines  in differences between the total sugar contents (or the dehydrin accumulation patterns) of the transgenic and non-transgenic seeds (Fig. 7).
Plants carrying the DS10:A9 transgene did not show differences in either reproductive or vegetative growth, from wild type or control plants (data not shown). In particular, no differences in seed yield, size or morphology were observed when we compared sibling lines with or without resistance to CDT (Fig. S4).

DISCUSSION
Plants show two conditions of tolerance to high temperature, basal and acquired thermotolerance, which appear in the absence of and after heat stress treatment at a sub-lethal temperature, respectively. The natural high basal thermotolerance of mature seeds is lost shortly after rehydration. For example, this happens upon seed imbibition (during germination), or if seeds are accidentally rehydrated during (or after) storage. Imbibed seeds are therefore thermo-sensitive because they lose basal thermotolerance (as shown for the control seeds in Figs. 3A and 4). Our results with the DS10:A9 seeds imply the temporal extension of basal thermotolerance to thermo-sensitive stages of seed germination. Basal thermotolerance persisted, to a significant extent, after the controlled rehydration of the DS10:A9 seeds (Fig. 4). This correlated with major, and specific, increases in HSP accumulation (for example, see Fig. 1). In contrast, we could detect only minor HSP accumulation and deteriorationresistance modifications in seeds when HaHSFA9 was expressed under the CaMV35S promoter (Fig. 3). The thermotolerant phenotype of the rehydrated DS10:A9 seeds requires gene expression modifications during embryogenesis that must persist in seeds within the early hours of imbibition. Our results demonstrate that this novel phenotype requires the very high expression level of HaHSFA9 conferred in seeds by the DS10 gene regulatory sequences. We should note that increases in basal thermotolerance have been previously described in transgenic plants overexpressing different HSFs; but the phenotype was observed in vegetative tissues well after seed germination (i.e., Prändl et al., 1998;Mishra et al., 2002). In these previous reports, the persistence of thermotolerance in imbibed seeds was not analyzed. Because these studies employed the CaMV35S promoter, our own observations with the

13
CaMV35S:A9 seeds suggest that we would not expect significant seed thermotolerance modifications to occur (Fig. 3).
The specific effects of HaHSFA9 could be also required for the persistence of basal thermotolerance in germinating seeds. In plants, sHSPs and HSP101 contribute to basal and acquired thermotolerance (Queitsch et al., 2000;Sun et al., 2002). Our analyses indicate that  HSPs are involved in the persistent-thermotolerant phenotype (compare Figs. 1 and 4). 2D-immunoblots showed that HaHSFA9 mainly induced the accumulation to higher levels of a subset of the CI sHSPs (Fig. 1). This included all proteins from that class expressed in seeds, some of which are exclusively present in seeds. HaHSFA9 also increased, although to lower levels, the accumulation of HSP101 and that of some CII-sHSP polypeptides (see Figs. 1 and S2). In contrast, the accumulation of dehydrin proteins was not increased where such genes exist and could perform seed-specific functions. For example, the sHSP genes that are upregulated by HaHSFA9 might contribute to the desiccation tolerance of mature seeds. In agreement with this suggestion, it is worth mentioning previous reports on maturing seeds indicating a strong association between longevity and the ability of seeds to tolerate desiccation (Ellis and Hong, 1994;Hay and Probert, 1995;Bruggink et al., 1999). In addition, Arabidopsis mutants that at the same time show decreased seed longevity and low accumulation levels of seed sHSPs, produce seeds that are desiccation-intolerant (Wehmeyer and Vierling, 2000  Furthermore, Bettey and Finch-Savage (1998) showed that a rapid aging treatment that reduced the germination performance of Brassica oleracea seeds also reduced the amount of HSP17.6 (a CI sHSP). In the same study, dehydrins did not show a positive correlation with seed performance.
We also demonstrate that HaHSFA9 induces a novel seed-deteriorationresistant phenotype. The CDT conditions for tobacco seeds required adaptation to the high CDT-resistance already showed by the non-transgenic material.
Such resistance has precedents in other small-sized seeds from Solanaceae as tomato (Argerich et al., 1989). Our conditions to achieve substantial deterioration of tobacco seeds in a reasonable time, are not very different from those generally used for less resistant seeds (45ºC, MCFW up to 24%, Powell, 1995). Furthermore at the temperature used in our experiments (50ºC), the absence of a HS response (Fig. 7)  better field emergence under stress conditions (Powell, 1995;reviewed by McDonald, 1999). In spite of this, we should add caution to any inference from our results of improved seed shelf life, as in other cases longevity predictions based on results on CDT experiments might be controversial (reviewed in McDonald, 1999). However even if similar predictions have not been performed with tobacco seeds, or with small-sized seeds from Solanaceae, it is worth mentioning that tomato seeds are highly resistant to CDT (Argerich et al., 1989).
Tomato seeds also have been independently shown to have a high relative

Protein electrophoresis and western blot analysis.
Gel electrophoresis and the conditions for western blot analysis were as described previously (Almoguera et al., 2002)  were extracted using the phenol buffer method (Lehmann et al., 1995) and the protein samples were resuspended in 2D-sample buffer (9M Urea, 2% [w/v] CHAPS, and 1% [w/v] DTT). For one-dimensional gels, the buffer conditions were adjusted by the addition of an equal volume of 2X Laemmli buffer and 45 µg of total protein was loaded per lane. SDS-PAGE gels were 15% (w/v) (for CI and CII sHSPs), 12.5% (w/v) (for dehydrins), or 8% (w/v) polyacrylamide (for HSP101). For two-dimensional gels, ampholytes (0.2% [w/v] Bio-Lyte 3-10 buffer, Bio-Rad) were added to the resuspended protein samples, 140 µg or 300 µg for seedling and seed samples, respectively. Isoelectrofocusing (IEF) was performed using the Protean IEF Cell system (Bio-Rad) and the IEF strips (ReadyStrip™ IPG Strips, 7 cm, pH 5-8, Bio-Rad) were subjected to active rehydration for 12 to 16 h at 50 V and 20ºC. The IEF program was: 15 min at 250 V, followed by a slow voltage rise to 4,000 V for 2 h and a final rise to 20,000 V. The strips were used to separate the proteins in the second dimension on 15% (w/v) polyacrylamide SDS gels (Almoguera et al., 2002). For western detection, the samples were transferred to Hybond-P membranes (Amersham Biosciences) after electrophoresis. The membranes were stained with Ponceau S, and incubated with primary antibodies at the following dilutions: anti-sHSP CI (Coca et al., 1994), 1/1,000; anti-sHSP CII (Coca et al., 1994), 1/2,000; anti-HSP101 (Queitsch et al., 2000), 1/5,000; or anti-DHN (Close et al., 1993), 1/1,000. The secondary antibody (anti-rabbit IgG, peroxidase-linked) was used at a dilution of 1/20,000. The immunoreaction was detected by using the ECL Plus system (Amersham Biosciences).

Basal thermotolerance assays and controlled deterioration of seeds.
Tobacco plants were grown under controlled environment, as described previously (Carranco et al., 1999). Mature seeds were harvested at 35 days post anthesis, and stored with silica gel at 4ºC in sealed containers until use.
Seeds produced and stored in this way consistently showed germination percentages ≥ 99% over the time span of the studies carried out here.
For basal thermotolerance assays, dry seeds were imbibed in water under controlled conditions and then subjected to a short incubation at 50ºC. Assays were performed on 3 to 4 replicates, each involving 30 mg of dry seeds (≈ 400-500 seeds) per plant line. The MCFW of dry mature seeds was 4.8 ± 0.2%, expressed as the percentage mass fraction of water of the total tissue mass (fresh weight). Rehydration was achieved by the sequential incubation of the seeds in 1 mL of the sterilizing solution (1% [v/v] sodium hypochlorite, for 20 min) and 1 mL of sterile water (40 min) at 25ºC. The MCFW of seeds after rehydration was 28.1 ± 0.7%. The rehydrated seeds were then incubated for 4 h at 50ºC in a water bath, in direct contact with 1 mL of sterile water. At this step, the MCFW of seeds increased to 41.3 ± 0.4%. After this treatment, germination was scored by placing individual seeds at defined points in grids drawn on Petri dishes with 1X Murashige and Skoog basal medium (MS Salts; Duchefa Biochemie, 3% [w/v] sucrose and 0.8% [w/v] phytoagar). The dishes were incubated at 25ºC (day) and 20ºC (night) in a 16 h photoperiod at 100 µmol m -2 s -1 , and photographed at different times after deterioration. For scoring the germination percentage, only the seedlings that kept growing after radicle protrusion during the experiment were considered. For the transgene segregation analyses performed with seeds from heterozygous T 0 lines, the seedlings were transplanted to MS medium with 300 µg mL -1 kanamycin ≈ 15 days after germination. Seedlings that remained green 3 weeks after transplanting were scored as positive for transgene inheritance.
Controlled deterioration tests (CDT) were performed on 4 to 6 experimental replicates, each involving 10 mg of dry seeds per plant line. Seeds were rehydrated before their exposure to high temperatures for different time lengths. MCFW was raised in controlled conditions by adding 20 µL of water to the seeds in Eppendorf tubes. After 2h at 25ºC, the excess of water is removed by briefly placing the rehydrated seeds on filter paper. The seeds were then placed in a single layer and are sealed in plastic bags. Caution was taken to remove as much air as possible before sealing the seeds in the bags.
Deterioration was achieved by incubation of the bags with the rehydrated seeds at 50ºC, in pre-warmed sealed boxes (12 x 7.5 x 6 cm) containing eight wet paper towels (with 50 mL H 2 0) placed within a water bath. The incubation at 50ºC was usually performed for 48h (conditions found to be the most discriminating CDT between transgenic and non-transgenic material). MCFW of seeds was 28.1 ± 0.5% and 27.9 ± 0.5%, respectively determined immediately before or after CDT. Different 50ºC treatments were performed to estimate seed longevity as a single parameter: LD 50 the number of days of CDT required to decline to 50% germination (as for example in Clerkx et al., 2004b).

Soluble carbohydrate and RFO content.
Total soluble extracts were prepared from mature seeds. Samples of 30 mg were homogenized in 3 mL of 80% (v/v) ethanol, and incubated for 20 min at 90ºC. The insoluble residue was removed by centrifuging at 5,000 g for 10 min and the extracts were air-dried to completely remove the ethanol. The total carbohydrate content was determined by the phenol-sulphuric acid method (Dubois et al., 1956). The total glucose, sucrose and RFO contents were measured using the raffinose series oligosaccharide assay from Megazyme.
We followed the manufacturer's instructions except that the quantities of reagents were scaled down starting from Milli-Q water). We also determined the sucrose and raffinose content by HPLC.
Samples (≈ 40 µg sugars in 40 µL Milli-Q water) were filtered by centrifugation through 0.22 µm Ultrafree-MC filters (Millipore) before HPLC injection. We used a Waters HPLC system (600E + 700 Satellite WISP) equipped with a Luna NH2 100A column (15 x 0.46 cm, 5 µm) and a light-scattering detection system (LSD Sedex 45 SEDERE). The mobile phase was a gradient of acetonitrile and water, and the flow rate was 1 mL min -1 under operating conditions at room temperature. The carbohydrate standards were from Sigma.

Statistical Analysis.
We tested for differences between the transgenic and control groups of contrasts) for multiple comparisons between the same data set. Unless stated otherwise, means ± 1SE are reported. We used SAS and JMP (SAS Institute, 1990) statistical packages to perform the analyses. Figure S1 Seed-specific overexpression of HaHSFA9 mRNAs from the DS10 promoter in transgenic tobacco.

Figure S2
2D-western analyses using CII sHSP antibodies in DS10:A9 seeds.  analysis using CI sHSP antibodies after 2D-electrophoresis of the same samples. The asterisk marks a heat-stress specific polypeptide not detectable in transgenic seeds. The pH range for IEF is indicated at the bottom. The arrow indicates a seed-specific polypeptide resolved by 1D-and 2D-gels, molecular mass standards on the left.    Table,