Many organisms, including higher plants, accumulate free proline (Pro) in response to osmotic stress. Although various studies have focused on the ability of Pro as a compatible osmolyte involved in osmotolerance, its specific role throughout plant growth is still unclear. It has been reported that Pro is synthesized from Glu catalyzed by a key enzyme, Δ1-pyrroline-5-carboxylate synthetase (P5CS), in plants. To elucidate essential roles of Pro, we generated antisense transgenic Arabidopsis plants with a P5CS cDNA. Several transgenics accumulated Pro at a significantly lower level than wild-type plants, providing direct evidence for a key role of P5CS in Pro production in Arabidopsis. These antisense transgenics showed morphological alterations in leaves and a defect in elongation of inflorescences. Furthermore, transgenic leaves were hypersensitive to osmotic stress. Microscopic analysis of transgenic leaves, in which the mutated phenotype clearly occurred, showed morphological abnormalities of epidermal and parenchymatous cells and retardation of differentiation of vascular systems. These phenotypes were suppressed by exogenous L-Pro but not by D-Pro or other Pro analogues. In addition, Pro deficiency did not broadly affect all proteins but specifically affected structural proteins of cell walls in the antisense transgenic plants. These results indicate that Pro is not just an osmoregulator in stressed plants but has a unique function involved in osmotolerance as well as in morphogenesis as a major constituent of cell wall structural proteins in plants.
Arabidopsis accumulates Pro in response to osmotic stresses due to drought, high salinity and chilling. The elevated expression of an AtP5CS gene encoding the P5CS protein in Arabidopsis precedes the accumulation of Pro in response to these stresses ( Yoshiba et al. 1995 ). To clarify the intrinsic roles of Pro in plants, we generated transgenic Arabidopsis plants with an AtP5CS antisense cDNA. These transgenics showed altered morphology and were susceptible to osmotic stress. Mutated phenotypes in both morphology and osmotolerance were suppressed by the application of exogenous l-Pro but not by d-Pro, suggesting that Pro has aspects other than as a compatible osmolyte with respect to osmotolerance in plants.
We further investigated a specific effect of Pro deficiency on protein biosynthesis in the transgenics. We found a significant reduction of Pro and hydroxyproline (Hyp) contents in hydrolysates of a purified cell wall fraction but not of Pro content of hydrolysates of soluble proteins in transgenic leaves. These results show that Pro deficiency specifically affects a defect in biosynthesis of cell wall structural proteins in the transgenic plant. We discuss roles of Pro as a major constituent of structural proteins of cell walls in osmotolerance as well as morphogenesis of plants.
Creation of transgenic lines with an AtP5CS antisense cDNA
Seventy independent transgenic lines with the AtP5CS antisense cDNA were established and T2 progeny were used for further analysis. When the Pro contents of these transgenic seedlings were measured under osmotic-stressed and unstressed conditions, several transgenic lines were found to accumulate significantly less Pro than wild-type plants. A variation of position effect may account for the difference in degree of Pro accumulation among transgenic lines. Among the lines, we selected three lines for further investigation: TF1, which accumulated Pro slightly in response to a change in humidity from 100% to 55%, and TF2 and TF3, which accumulated almost no Pro in spite of the treatment ( Fig. 1c). These three lines co-segregated with kanamycin resistance and susceptibility in a 3:1 ratio, which indicates T-DNA insertion at a single locus in the genome.
The effects of transformation with the AtP5CS antisense cDNA on the amounts of AtP5CS transcripts and AtP5CS protein were evaluated by Northern blot analysis with an AtP5CS antisense mRNA probe and by Western blot analysis, respectively. Although AtP5CS mRNA and protein accumulated in wild-type plants and in the vector-transformed control plants that responded to the stress treatment, they did not accumulate significantly in the TF1 and TF2 plants ( Fig. 2). These results confirm a reduction in free Pro level due to suppression of the AtP5CS protein synthesis by the effect of the antisense transgene. Similar results were obtained with the TF3 plants.
Recently, another P5CS gene (AtP5CS2) was isolated from Arabidopsis ( Strizhov et al. 1997 ). The expression of both of these P5CS genes would have been repressed in our AtP5CS antisense transgenics because of a high identity (82%) between the two AtP5CS cDNAs.
Morphological alterations of the AtP5CS antisense transgenics
The antisense transgenics, TF1, TF2 and TF3, showed mutated phenotypes in their morphology on GMK plates: repression of stem elongation and alteration in the shape of rosette leaves. All the kanamycin-resistant seedlings of TF1, TF2 and TF3 showed the mutated phenotype, suggesting a dominant effect of the transgene. Wild-type Arabidopsis plants in long-day photoperiods normally bolt and flower at the age of 3–4 weeks ( Fig. 3a). In contrast, the transgenic plants flowered without elongation of internodes even at the age of 4 weeks ( Fig. 3b). TF1 transgenics grown under 100% humidity for 9 weeks produced many short inflorescences and looked bushy ( Fig. 3c). Inflorescence internodes of the TF2 and TF3 plants never elongated; the siliques were shorter and fertility was extremely low. However, all the essential organogenesis occurred normally in all these plants. Leaf morphology was also affected in transgenic plants: adult rosette leaves of the transgenics were round with short petioles ( Fig. 3b), whereas those of the wild type were spatulate. Microscopic analysis of adult rosette leaves from wild-type and TF2 plants showed pleiotropic alterations in the epidermis, the mesophyll, and the vascular system of the transgenic leaf. Compared with the wild-type leaf, the TF2 leaves had significantly enlarged epidermal and palisade cells ( Fig. 3d,e,g,h). We also found that abnormal elongation of spongy cells caused greater intercellular spaces in the layer of the TF2 leaf ( Fig. 3f,i). This irregular expansion of leaf cells appeared to alter expansion of leaf blades. Moreover, differentiation of vascular bundles was incomplete and irregular in the TF2 leaf ( Fig. 3j,k). The TF3 transgenic plants showed similar phenotypes.
To analyze whether Pro starvation caused by the transgene led to these morphological alterations, the TF1 and TF2 seeds were germinated on l-Pro(+) medium and were then allowed to grow for 4 weeks. The mutated phenotypes were suppressed by the application of l-Pro: internodes of inflorescences elongated as much as those of wild-type plants, and rosette leaves of transgenics were indistinguishable from those of wild-type plants (data not shown).
We used the TF2 and TF3 plants for further analysis because their morphological alterations were more severe than those of the TF1 plants. Four-week-old TF2 and TF3 seedlings grown on GMK were transferred to GMK or l-Pro(+) medium and then grown aseptically for 8 days. Inflorescence internodes of the TF2 seedlings transferred to GMK did not elongate, whereas those transferred to the l-Pro(+) medium did elongate ( Fig. 4). d-Pro suppressed none of the mutant phenotypes. Similar results were obtained with the TF3 seedlings.
Osmotic stress sensitivity of the AtP5CS antisense transgenics
To elucidate the role of Pro in osmotolerance, we examined the AtP5CS antisense transgenics for their sensitivity to osmotic stress. The TF2 and TF3 plants were germinated and grown on GMK or l-Pro(+) medium at 100% humidity for 4 weeks and then subjected to low-humidity treatment (55% humidity) for 5 days. Although wild-type plants grew well during the treatment ( Fig. 5a,d), the TF2 seedlings grown on GMK showed hypersensitivity to the treatment, wilted and died within 5 days ( Fig. 5b,e). In contrast, the TF2 seedlings grown on the l-Pro(+) medium grew as well as the wild-type seedlings during the treatment ( Fig. 5c,f). Similar sensitivity to osmotic stress was observed in the TF3 plants.
To investigate how much exogenous Pro was accumulated in the TF2 seedlings subjected to the low-humidity treatment, we measured Pro contents in leaves. In response to the treatment, the Pro content in leaves of wild-type plants grown without Pro increased from 29 to 320 μg g–1 fresh weight. The endogenous Pro content in leaves of the TF2 seedlings grown on the l-Pro(+) medium increased from 5 to 20 μg g–1 fresh weight ( Fig. 6).
Next, we examined the effect of d-Pro on the suppression of hypersensitivity of the TF2 plants to osmotic stress. Exogenous d-Pro did not suppress the wilt-prone phenotype of the TF2 plants caused by the low-humidity treatment. The total l-Pro and d-Pro content in leaves of the TF2 plants grown on the d-Pro(+) medium increased from 13 to 47 μg g–1 fresh weight after treatment ( Fig. 6). The effects of l-Pro analogues and other amino acids (see Experimental procedures) on the mutated phenotype were examined. None of these compounds suppressed either osmo-hypersensitivity or morphological alteration in transgenics.
Effects of Pro deficiency on protein biosynthesis
To analyze the effect of Pro deficiency on protein biosyntheses, we measured contents of Pro and Hyp residues in hydrolysates of total soluble proteins and purified cell wall fraction. For these analyses, we used mainly unstressed rosette leaves of the TF3 transgenics grown on GMK plates for 3 weeks. This line had a T-DNA insertion at a single locus in the genome, accumulated little free Pro in response to the low-humidity treatment, and showed the severe morphological phenotype and hypersensitivity to osmotic stress as well as the TF2 plants (data not shown). The free Pro level in rosette leaves of the TF3 plants was markedly lower (4 μg g–1 fresh weight) than that of wild-type plants (44 μg g–1 fresh weight) ( Fig. 7a). However, no significant difference in the contents of total soluble proteins (data not shown) or Pro residues of the protein hydrolysates ( Fig. 7b) was found between wild-type and TF3 plants. In contrast, in the TF3 leaves, a significant decrease in both Pro and Hyp residues was observed in the cell wall preparation containing abundant Pro-rich and Hyp-rich cell wall matrix proteins ( Fig. 7c).
In this study, suppression of P5CS decreased Pro production in antisense transgenic Arabidopsis plants. This result supports previous reports that P5CS is a rate-limiting enzyme in the Glu pathway of Pro biosynthesis ( Hu et al. 1992 ; Igarashi et al. 1997 ; Kavi Kishor et al. 1995 ; Savouréet al. 1995 ; Yoshiba et al. 1995 ). Moreover, the Orn pathway may not play a key role in Pro biosynthesis in Arabidopsis. The result also led us to suppose that these P5CS antisense transgenic plants would provide a valuable approach for the study of essential roles of Pro in osmotolerance. The antisense effect on Pro biosynthesis clearly appeared in the plants even under a moderate decrease in humidity. Four-week-old wild-type Arabidopsis withstands a mild hypo-osmotic stress ( Fig. 5d). By contrast, the antisense transgenic plants deficient in Pro accumulation had decreased osmotolerance. The TF2 AtP5CS antisense transgenic plants, accumulating almost no Pro, showed significant hypersensitivity to the treatment and wilted rapidly ( Fig. 5e). The TF1 plants, which accumulated more Pro in response to osmotic stress than TF2 plants ( Fig. 1c), also showed hypersensitivity to low humidity. According to a quantitative assay for osmotolerance (monitoring changes in chlorophyll contents during the stress treatment), the TF1 plants were more stress-tolerant than the TF2 and TF3 plants (data not shown). This shows that Pro deficiency specifically decreases osmotolerance. In addition, the level of osmotolerance seems to be correlated with the ability of the plant to synthesize Pro.
To confirm the positive correlation between Pro and osmotolerance, we examined effects of the application of exogenous l-Pro on the phenotypes of the TF2 plants. The hypersensitivity of the transgenics was greatly suppressed by 1 m m l-Pro added in the agar medium ( Fig. 5f), which consolidates evidence for an essential role of Pro in osmotolerance. Many reports have shown positive correlations between the capacity for Pro accumulation and osmotolerance in plants ( Handa et al. 1986 ; Kavi Kishor et al. 1995 ; Van Rensburg et al. 1993 ). These previous results led us to believe that l-Pro-treated transgenics recovered osmotolerance by accumulating exogenous Pro at the same level as wild-type plants. To confirm this, we measured the level of free Pro content in l-Pro-treated antisense transgenics subjected to low humidity. Unexpectedly, l-Pro-treated transgenics contained only 7% of the Pro pool of the wild-type plants after the low-humidity treatment ( Fig. 6). Moreover, if exogenously supplied Pro imparted osmotolerance to the transgenics by acting as a mere osmotic adjuster, a difference in its stereospecificity would be irrelevant to its adaptive function. In fact, the application of exogenous d-Pro did not suppress the hypersensitivity of transgenics (data not shown). Although a similar small increase in free Pro content was observed in d-Pro-treated transgenic plants ( Fig. 6), this was probably caused by elevated imports of exogenous Pro via activated Pro permease ( Rentsch et al. 1996 ) or by a decrease in the activity of Pro degradation ( Kiyosue et al. 1996 ) in response to the low-humidity treatment. Thus, the endogenous level of free Pro is not correlated with hypersensitivity to the low-humidity treatment, and Pro may have functions other than a compatible osmolyte in stress tolerance (discussed below).
On the other hand, the AtP5CS antisense transgenics showed morphological abnormalities in rosette leaves and inflorescences ( Fig. 3b,c). The application of exogenous l-Pro suppressed these phenotypes but d-Pro did not ( Fig. 4), which indicates that l-Pro is specifically involved in morphogenesis in plants. Microscopic analysis indicated that morphological alterations observed in the antisense transgenic leaves are derived from abnormal cell morphology: both epidermal and mesophyll cells are significantly enlarged ( Fig. 3g,h). We also observed a defect in the vascular system of the TF2 leaf ( Fig. 3k). Such a defect in a conducting system must disturb water transport in plants and is thus likely to be a major reason for the hypersensitivity to osmotic stress in the transgenics. A difference in osmotolerance between TF1 and the other two lines may not result from a difference in the degree of ability to accumulate free Pro, but be correlated with a difference in their morphological properties.
Both physiological and morphological investigations led us to realize that Pro acts as a constituent of major proteins playing key roles in osmotolerance as well as in morphogenesis of leaves in plants. l-Pro is a component of proteins in plants, so a defect in Pro biosynthesis may broadly affect the biosynthesis of various proteins. However, considering that the antisense effect on Pro biosynthesis was leaky in our transgenics, it is reasonable to speculate that the synthesis of proteins containing Pro-rich or Hyp-rich motifs is more easily inhibited in the case of Pro deficiency. To test this possibility, we analyzed specificity of the effect of Pro deficiency on protein biosynthesis in the transgenics. There was no significant difference in the contents of total soluble proteins (data not shown) or those of Pro in the protein hydrolysates between wild-type and transgenic plants deficient in free Pro accumulation ( Fig. 7a,b). This shows that Pro deficiency in transgenics had little effect on biosynthesis of the cytosolic proteins by which each cell maintains activity vital for growth, supporting the fact that the essential organogenesis occurred normally in transgenics. In contrast, both Pro and Hyp contents decreased in the cell wall preparation of transgenic leaves ( Fig. 7c). This result indicates that Pro deficiency in transgenics specifically affected the biosynthesis of matrix proteins localized in the cell wall. Given the morphological alterations in transgenic leaves due to the abnormal growth of leaf cells, the most probable candidates responsible for the mutated phenotype are structural cell wall proteins such as Pro-rich proteins (PRPs) and Hyp-rich glycoproteins (HRGPs). These proteins, like extensins, contain abundant Pro-rich and Hyp-rich motifs. They have important roles in cell morphogenesis and provide mechanical support for cells under stressed conditions ( Muñoz et al. 1998 ; Showalter 1993; Tierney & Varner 1987). Moreover, by a pharmacological approach, Cooper et al. (1994) showed that HRGPs are important determinants of cell morphology and osmotic stability in tobacco protoplasts. Our results in planta may support their findings.
Deficiency of any amino acids would also change morphology or reduce growth. Nevertheless, some reports have shown the specific features of amino acid deficiency in Arabidopsis. Ravanel et al. (1998) presented methionine-deficient plants showing phenotypic abnormalities and discussed the metabolism and regulatory function of S-adenosylmethionine. Last et al. (1991) reported tryptophan-deficient mutants that exhibited altered leaf morphology due to a resulting defect in auxin biosynthesis. In this study, we have focused on specific effects of Pro deficiency on plant growth, and have consequently proved that Pro plays a vital role in the osmotolerance that wild-type plants possess by nature. Many workers regard Pro as a multiple contributor to stress tolerance in plants ( Delauney & Verma 1993; Hare & Cress 1997). We have shown that Pro is not just an osmoregulator or scavenger of free radicals in stressed plants but has a unique function involved in osmotolerance as well as in morphogenesis as a major constituent of cell wall structural proteins in plants.
Plant material and growth conditions
Arabidopsis thaliana (L.) Heynh. ecotype Columbia was used in this study. Wild-type seeds were sown on 90 mm × 20 mm Petri dishes containing solidified germination medium (GM) ( Van Valvekens et al. 1988 ); transgenic plants were grown on GM containing 30 mg l–1 of kanamycin (GMK). The plates were then incubated in a growth chamber under illumination of about lux with a 16 h light, 8 h dark cycle at 22°C.
Construction of transgenic plants
An AtP5CS cDNA ( Yoshiba et al. 1995 ) encoding Arabidopsis P5CS protein was reverse-fused to the BamHI site of the expression vector pBE2113 ( Mitsuhara et al. 1996 ), which contained a 35S cauliflower mosaic virus promoter ( Fig. 1b). The resultant plasmid was introduced into 6- to 8-week-old wild-type Arabidopsis seedlings grown in soil as described by Yoshiba et al. (1995) ; it was introduced by the modified vacuum infiltration method ( Bechtold et al. 1993 ; Wenck et al. 1997 ). Agrobacterium tumefaciens strain GV3101 (pMP90) was used. After vacuum infiltration, plants were grown for 4–6 weeks and T1 seeds were harvested. The seeds were then sown on 150 mm × 15 mm GMK plates containing 200 mg l–1 of claforan. T1 seedlings that showed kanamycin resistance were transferred to soil and grown as described by Yoshiba et al. (1995) for 4–6 weeks. Mature T2 seeds were harvested and used for subsequent experiments.
Low humidity treatment
Arabidopsis plants were grown on GM (wild-type) or GMK (transgenic) plates. The plates were then fully opened on a clean bench, and the seedlings were allowed to grow under continuous illumination of about lux at 22°C and 55% humidity. The agar was kept moist with added distilled water. Aerial parts of seedlings were exposed to a rapid change in humidity from 100% to 55%. This treatment was defined as the low humidity treatment.
RNA gel blot analysis
Total RNA was isolated as described by Nagy et al. (1988) from rosette leaves of plants that had been grown on GM or GMK plates for 3 weeks and then subjected to the low humidity treatment. Twenty micrograms of total RNA were fractionated by electrophoresis on a 1% agarose gel containing formaldehyde and blotted onto a nitrocellulose filter as described by Maniatis et al. (1982) . The membrane was probed with 32P-labeled AtP5CS antisense mRNA, which detects approximately 2.6 kbp AtP5CS sense mRNA. The AtP5CS antisense RNA probe (prepared by in vitro transcription using an RNA transcription kit (Stratagene) according to the manufacturer’s protocol) was used for Northern hybridization as described by Yoshiba et al. (1995) .
Protein extraction, SDS-PAGE and immunoblotting
Two grams (fresh weight) of 3-week-old rosette leaves were ground in 10 ml of homogenization buffer containing 100 m m Tris–HCl (pH 7.5), 10 m m MgCl2, 10 m mβ-mercaptoethanol, and 1 m m phenylmethylsulfonyl fluoride at 4°C. The homogenate was centrifuged at 10 000 g for 15 min at 4°C. Protein content of the supernatant was determined with a protein assay kit (Bio-Rad). The precipitate was used for the cell wall preparation and part of the protein extract was precipitated with 10% trichloroacetic acid (TCA) for 5 h and centrifuged at 10 000 g for 20 min. For biochemical analysis of total soluble proteins, the TCA precipitate was then hydrolyzed with acid as described below.
For immunodetection, protein extracts were separated by 10% SDS-PAGE gel and transferred electrophoretically onto a polyvinylidene difluoride membrane in a solution of 25 m m Tris, 192 m m glycine, and 20% (v/v) methanol. The membrane was blotted with 0.02% (v/v) anti-AtP5CS antibody raised in rabbit against a peptide corresponding to the 15 amino acids (EELDRSRAFARDVKR) of the AtP5CS N terminus; it was blotted in TBS containing 0.05% (v/v) Tween 20 (TBS-T) for 1 h at room temperature. The blot was washed with TBS-T and then agitated in a 0.1% solution (v/v) of horseradish-peroxidase-conjugated antibodies raised against rabbit Ig raised in donkey (Amersham) in TBS-T for 1 h at room temperature. The AtP5CS–antibody complex was made visible by ECL Western blotting analysis according to the manufacturer’s instructions (Amersham).
Rosette leaves used for microscopic analysis were obtained from 4-week-old plants grown on GM (wild-type) and GMK (transgenic) plates. Fully expanded 5th true leaves were fixed, washed and cleared as described by Tsuge et al. (1996) , and were then observed under a microscope (DAS Mikroskop DMR; Leica, Wetzlar, Germany) equipped with Nomarski differential interference-contrast optics.
Application of exogenous amino acids and Pro analogues
The AtP5CS antisense transgenics were germinated and grown on GMK plates supplemented with 10–5 to 10–2 M l-Pro, d-Pro, l-Glu, l-Orn, hydroxy- l-Pro or l-pipecolic acid or without supplement. GMK supplemented with 1 m m l-Pro or d-Pro was designated the l-Pro(+) medium or the d-Pro(+) medium, respectively. At 4 weeks the plants were transferred to 9 cm diameter × 10 cm high Biopots (Watanabe TAI Co., Ltd, Japan) containing the l-Pro(+) medium or the d-Pro(+) medium, and then grown for 8 days.
Cell wall preparation, hydrolysis and amino acid analysis
The precipitation of homogenates of the protein extraction described above was suspended in the homogenization buffer and centrifuged at 600 g for 10 min at 4°C. The pelleted material was washed four times with the homogenization buffer, twice with 0.5% Triton X100, three times with water, twice with 1 m NaCl, a further three times with water, and twice with acetone to remove contaminating cytoplasmic material. The insoluble material remaining was used as the cell wall preparation.
Each cell wall preparation containing dl-Norleucine as an internal standard was treated with 1 ml of 0.22 m Ba(OH)2 for 20 h at 90°C in a sealed tube. The total hydrolysate was then neutralized with 0.5 m H2SO4 and centrifuged at 1500 g for 10 min. The supernatant was filtered through cheesecloth and the filtrate was evaporated to dryness. The residue was redissolved in distilled water.
The alkaline hydrolysate was transferred to a small glass tube (8 mm × 50 mm) and the hydrolysate was evaporated to dryness. The small tube was then transferred into a sealed tube containing 250 μl of 6 m redistilled HCl with 1 μl of phenol on the bottom, and the alkaline hydrolysate was further hydrolyzed for 18 h at 110°C.
The acid hydrolysate was derivatized as described by Cohen & Strydom (1988), and the contents of Pro and Hyp were determined by HPLC (Waters Associates, Milford, MA, USA, model LC Module I plus).
We thank F. Hayashi for help with the Pro assay, S. Kawamura for preparation of the plants, and Y. Ohashi for providing the vector pBE2113. This work was supported by the Program for Promotion of Basic Research Activities for Innovative Bioscience, the Special Coordination Fund of the Science and Technology Agency, and a Grant-in-Aid from the Ministry of Education, Science and Culture of Japan.