Proline is required for male gametophyte development in Arabidopsis
© Mattioli et al.; licensee BioMed Central Ltd. 2012
Received: 18 June 2012
Accepted: 3 December 2012
Published: 12 December 2012
Skip to main content
© Mattioli et al.; licensee BioMed Central Ltd. 2012
Received: 18 June 2012
Accepted: 3 December 2012
Published: 12 December 2012
In crosses between the proline-deficient mutant homozygous for p5cs1 and heterozygous for p5cs2 (p5cs1 p5cs2/P5CS2), used as male, and different Arabidopsis mutants, used as females, the p5cs2 mutant allele was rarely transmitted to the outcrossed progeny, suggesting that the fertility of the male gametophyte carrying mutations in both P5CS1 and P5CS2 is severely compromised.
To confirm the fertility defects of pollen from p5cs1 p5cs2/P5CS2 mutants, transmission of mutant alleles through pollen was tested in two ways. First, the number of progeny inheriting a dominant sulfadiazine resistance marker linked to p5cs2 was determined. Second, the number of p5cs2/p5cs2 embryos was determined. A ratio of resistant to susceptible plantlets close to 50%, and the absence of aborted embryos were consistent with the hypothesis that the male gametophyte carrying both p5cs1 and p5cs2 alleles is rarely transmitted to the offspring. In addition, in reciprocal crosses with wild type, about 50% of the p5cs2 mutant alleles were transmitted to the sporophytic generation when p5cs1 p5cs2/P5CS2 was used as a female, while less than 1% of the p5cs2 alleles could be transmitted to the outcrossed progeny when p5cs1 p5cs2/P5CS2 was used as a male. Morphological and functional analysis of mutant pollen revealed a population of small, degenerated, and unviable pollen grains, indicating that the mutant homozygous for p5cs1 and heterozygous for p5cs2 is impaired in pollen development, and suggesting a role for proline in male gametophyte development. Consistent with these findings, we found that pollen from p5cs1 homozygous mutants, display defects similar to, but less pronounced than pollen from p5cs1 p5cs2/P5CS2 mutants. Finally, we show that pollen from p5cs1 p5cs2/P5CS2 plants contains less proline than wild type and that exogenous proline supplied from the beginning of another development can partially complement both morphological and functional pollen defects.
Our data show that the development of the male gametophyte carrying mutations in both P5CS1 and P5CS2 is severely compromised, and indicate that proline is required for pollen development and transmission.
In addition to its role as proteinogenic amino acid, and as a molecule involved in responses to a number of biotic and abiotic stresses, proline has been implicated in plant development, particularly flowering and reproduction [1–3]. The first convincing evidence that proline may play a key role in plant reproduction under normal unstressed conditions, came from measurements of free proline content in a number of plant species, which revealed low levels of proline in vegetative tissues followed, after flower transition, by proline accumulation in reproductive tissues and organs [4–8]. Chiang and Dandekar , for example, reported that free proline accumulates in Arabidopsis reproductive tissues up to 26% of the total amino acid pool, while in vegetative tissues represents only 1-3%. Among floral organs, different authors [7–11] pointed out that the floral organ with the highest proline content is pollen, where proline may represent more than 70% of the total amino acid content .
It is not clear, to date, the reason for such a massive proline accumulation in pollen. Because pollen grains undergo a process of natural dehydration, a role of compatible osmolyte capable of protecting cellular structures from denaturation, has been proposed by some authors [7, 12, 13], while others  have postulated a role for proline as a source of energy or as metabolic precursor to support the rapid and energy-demanding elongation of the pollen tube. On the other hand, the rapid elongation of the pollen tube requires extensive synthesis of cell wall proteins , some of which are rich in proline or hydroxyproline stretches, and proline accumulation may be needed to sustain the synthesis, at high levels, of proline-rich cell wall proteins .
Irrespective of its function, proline may accumulate in pollen due to an increased transport from external sources, or to an increased ratio between synthesis and degradation of endogenous proline, or because of a combination of the two, but no conclusive evidence has been produced, as yet, to distinguish among these alternative models. Long distance transport of proline through phloem vessels has been documented [17, 18] and since AtProT1 (AT2G39890), a gene encoding an amino acid carrier recently shown to mediate proline uptake in plants, is highly expressed in mature pollen , transport has been proposed to account for proline accumulation in pollen grains. However single, double, and triple knockout mutants for all the genes belonging to the AtProT family are available, and none of them show difference, compared to wild type, neither in proline content, nor in pollen germination efficiency , raising the possibility that endogenous proline synthesis may be responsible for, or contribute to proline accumulation in pollen.
In higher plants proline synthesis proceeds from glutamate that is converted to proline in a two-step pathway catalyzed by the enzymes Δ1-pyrroline-5-carboxylate ynthetase (P5CS), and Δ1-pyrroline-5-carboxylate reductase (P5CR). The existence of an alternative route for proline synthesis, converting ornithine to proline by the action of δ-ornithine-amino-transferase (δ-OAT; At5g46180) and P5CR, has been hypothesized by some authors [20, 21]. However the relevance of this pathway for proline synthesis has been recently questioned  and glutamate may be the only precursor of proline synthesis in plants. P5CS, regarded as the rate-limiting enzyme for proline biosynthesis in plants, is encoded in Arabidopsis by the two paralog genes P5CS1 (At2g39800) and P5CS2 (At3g55610) , while no paralog genes have been described for P5CR (At5g14800).
T-DNA insertional mutants have been characterized [1, 3, 24] for both P5CS1 (SALK_063517, p5cs1-4) and P5CS2 (GABI_452G01, p5cs2-1;FLAG_139H07, p5cs2-2), providing hints for assigning gene functions. P5CS1 is responsible for abiotic stress-induced proline accumulation, as homozygous p5cs1 mutants do not accumulate proline upon stress induction and are hypersensitive to environmental stresses [3, 24], while P5CS2 is necessary for embryo development, as homozygous p5cs2 mutants are embryo lethal and the p5cs2 mutant allele can be propagated only in p5cs2/P5CS2 heterozygous mutants [1, 24]. In addition, both genes have been shown to modulate flower transition in Arabidopsis, as the flowering time of mutants homozygous for p5cs1 and heterozygous for p5cs2, is more delayed than that of the single p5cs1 mutant [1, 2].
In the course of a genetic screen designed to identify the floral pathway(s) proline interacts with in Arabidopsis (Mattioli et al., in preparation), we found that the p5cs2 mutation was rarely transmitted to the offspring when the proline-deficient p5cs1 p5cs2/P5CS2 was used as a pollen donor suggesting yet another role for proline in affecting male fertility. This prompted the analysis presented in this work, aimed to evaluate the role of endogenous proline in pollen development and fertility. We show here that the development of the male gametophyte carrying mutations in both P5CS1 and P5CS2 is severely compromised, indicating a role for proline in pollen function and development.
In crosses between p5cs1 p5cs2/P5CS2, used as male, and Arabidopsis flowering time mutants, used as females, aimed to understand the flowering pathway proline interacts with (Mattioli et al., in preparation), the p5cs1 mutant allele was always transmitted to the outcrossed progeny, while the transmission frequency of the p5cs2 mutant allele was exceedingly low (in average 0.8 ± 0.1%). Since no obvious gametophytic defects have been ever noticed neither on p5cs1 nor on p5cs2 single mutants, this result suggests that a male fertility defect may be linked to pollen grains bearing mutations in both P5CS1 and P5CS2 genes.
Reciprocal crosses between p5cs1 p5cs2/P5CS2 mutants and wild type plants
Genotype of the progeny
p5cs1 p5cs2/P5CS2 p5cs1 p5cs2/P5CS2
wt × p5cs1 p5cs2/P5CS2
131/132 (99.15 %)***
p5cs1 p5cs2/P5CS2 × wt
On the basis of crosses showing that pollen with genetic defects in both P5CS1 and P5CS2 is almost completely unable of a successful fertilization, a requirement for proline in male gametophyte viability was hypothesized and demonstrated here, by means of genetic, developmental and molecular evidence.
Massive accumulation of proline has been reported in anthers and pollen by different authors in a number of species [7–11, 14], but it is not yet clear why such high amount of proline is required. Functions as diverse as free radical scavenger , protector of membranes and cellular structures , energy source , metabolic precursor , and main amino acid constituent of hydroxyproline-rich cell walls  have been proposed but none of these possibilities has attained wide acceptance, and the interesting hypothesis that multiple functions may be accounted for by proline action has been suggested .
Evidence provided in this work indicates that pollen from mutants homozygous for p5cs1 and heterozygous for p5cs2 can transmit the p5cs2 mutation with an overall frequency of about 0.8%, and that mutants with decreasing levels in proline content [2, 3, 24] have increasing problems in pollen viability indicating that proline is required for male fertility.
Apart from the defects in pollen development described in this work, and from a delay in flower transition described by Mattioli et al. , the vegetative and reproductive growth of mutants homozygous for p5cs1 and heterozygous for p5cs2, including the development of the female gametophyte, is essentially normal. This evidence implies that the small amount of proline coming from the activity of the wild type P5CS2 allele, always present in the sporophytic tissues of p5cs1 p5cs2/P5CS2, is sufficient for normal growth and reproduction, but not for proper pollen development.
The specific requirement of proline for pollen development and function, is confirmed by the high amount of proline found in pollen by different authors [7–11, 14], by the very low level of proline measured in mutant pollen, and by the partial rescue of pollen defects obtained by exogenous proline treatment. While this complementation provides direct proof that proline is required for pollen development and function, it gives no indication whether the required proline derives uniquely from endogenous synthesis inside the pollen or also from proline synthesized in nearby mother cells and transported or diffused inside pollen grains. This point clearly needs to be understood in future works.
However, since pollen from p5cs1 p5cs2/P5CS2 is infertile, a possible transport or diffusion of proline synthesized in surrounding sporophytic cells by the residual P5CS2 allele is obviously insufficient. Furthermore, single, double and triple knockout mutants of the AtProT genes (At2G39890, At3G55740, At2G36590) responsible for proline transport in plants, have been isolated and characterized but none of them revealed alterations, compared to wild type, neither in proline content nor in pollen germination efficiency . Although the possibility that different carriers, such as AtLHT5 (At1g67640)  or AtLHT7 (At4G35180) , may compensate, overlap to, or substitute for AtProTs, current evidence does not support a role of transport for proline accumulation in pollen grains. In addition, microarray data indicate that all the genes involved in proline synthesis are strongly expressed in pollen [33, 34], and data from Szekely et al. , who detected in the pollen of Arabidopsis the expression of both AtP5CS1–GFP and AtP5CS2–GFP, confirm the presence of the P5CS protein in the male gametophyte. Overall these data suggest that proline is actively synthesized in pollen.
The question whether proline may be synthesized directly in pollen grains has been the object of controversial discussions, because some authors could detect low [8, 32] or no expression of P5CS in pollen , while others reported the presence of an AtP5CS-GFP protein in the pollen of Arabidopsis, suggesting that biosynthesis of proline takes place in this organ . However, the discrepancy in the expression levels of P5CS gene as observed by different authors [8, 19, 24, 32], may depend on the developmental stage in which pollen was analyzed, and we may speculate that P5CS1/2 genes could be expressed only in particular stages of pollen development, and still accumulate enough P5CS enzyme to satisfy overall proline demand for pollen maturation. Temporal discrepancies between transcript and protein levels have been reported in pollen also for other genes, such as AtSUC1 (AT1G71880), whose transcript level is high at tricellular stage and low in mature pollen  and AtSTP9 (AT1G50310), whose gene product can only be detected by immunofluorescence microscopy after the onset of germination . In addition, as above stated, high levels of expression of either P5CS1, P5CS2, and P5CR are detected in pollen by microarray analysis [33, 34], directly confirming that endogenous proline synthesis from glutamate takes place in pollen grains. Unexpectedly, microarrays analysis also detects the expression of δ-OAT in pollen, although ornithine pathway seems not able to compensate P5CS deficiency in pollen from p5cs1 p5cs2/P5CS2. These contrasting pieces of evidence can be reconciled if ornithine pathway does not contribute to proline synthesis. Incidentally, this evidence supports the finding of Funck et al.  who demonstrated that the ornithine pathway is essential for arginine catabolism but not for proline synthesis.
Overall, proline accumulation in pollen may rely essentially on endogenous proline synthesis, although is yet to be understood whether proline derives uniquely from endogenous synthesis inside the male gametophyte or also from proline synthesized in nearby sporophytic cells and transported or diffused inside pollen grains. A likely hypothesis is that, as pollen lose desmosomal connections to surrounding sporophytic cells, becomes dependent on endogenous proline synthesis, consistent with the late appearance (stage 11) of visible aberrations in developing pollen from p5cs1 p5cs2/P5CS2, and with the absence of obvious defects in female gametopytes, always embedded in sporophytic cells.
Intriguingly, two Arabidopsis mutants bearing T-DNA insertions on P5CR, emb-2772-1 and emb-2772-2 exhibit an embryo lethal phenotype, similarly to p5cs2 mutants, halting embryo development at a preglobular stage. In sharp contrast, however, no gametopytic defects have been associated, so far, to these mutants.
While it is not surprising that lesions in P5CR and P5CS2, two genes coding for proline synthesis enzymes belonging to the same pathway, may lead to similar defects in embryo development, it is puzzling that, contrary to p5cs2, emb-2772 exhibits no gametophytic defects. In Arabidopsis a number of mutants have been described by Muralla et al.  with defects in embryo but not in gametophyte development.
To explain this apparent paradox, the authors propose that gene products derived from transcription of wild type alleles in heterozygous sporocytes may compensate the deficiency of the mutant gametophytes, and that embryo lethality results when these products are eventually depleted. Likewise, we may speculate that, contrary to P5CS2, the P5CR transcript and/or protein, synthesized in heterozygous sporocytes, is stable enough to sustain pollen but not embryo development.
The data presented here suggest that proline is required for pollen development, but gives no indication on the role of proline in pollen development. We know from histological analysis (Figure 5) that a fraction of pollen grains begins to look shriveled and shrunk from stage 11, when, after completion of the two mitotic divisions, the microspores start their maturation to pollen grains. As pollen development proceeds, it becomes more and more desiccated, and increasing amounts of proline may be needed to avoid protein denaturation and preserve cellular structures, including nuclei, as hypothesized by Chiang and Dandekar . A role for proline in the protection of cellular structures from denaturation has been proposed by different authors either as compatible osmolyte , scavenger of free radicals [12, 24], or as protector of membranes and cellular structures . Although, from available data, a defect in mitotic divisions cannot be ruled out, the degeneration of the cellular structures observed in pollen grains of p5cs1 p5cs2 genotype, may be caused by the irreversible damages on cellular membranes caused by the process of dehydration in absence of the protective action of proline.
Once pollen has reached full maturation, accumulated proline is catabolized and serves as source of energy - to fuel the rapid and energy-demanding elongation of the pollen tube [15, 39] - and/or as metabolic precursor of γ-amino butyric acid (GABA), the catabolism of which has been shown essential for late pollen tube elongation and guidance .
In the future it will be interesting to address this issue by uncoupling these two putative functions, for example targeting in developing pollen grains from p5cs1 p5cs2/P5CS2 plants non-metabolizable compatible osmolyte, such as glycine betaine. Equally interesting it will be to dissect the role of proline synthesized in the haploid male gametophyte from that synthesized in diploid sporophytic tissues of the anther.
We show here that in mutants homozygous for p5cs1 and heterozygous for p5cs2, defective in proline synthesis, the development of the male gametophyte with mutations in both P5CS1 and P5CS2 is severely compromised, and provide genetic evidence that proline is needed for pollen development and fertility.
Wild-type and mutant Arabidopsis thaliana from Columbia-0 (Col-0) ecotype used in this work were grown in a growth chamber at 24/21°C with light intensity of 300-μE·m-2·s-1 under 16 h light and 8 h dark per day. Arabidopsis homozygous for p5cs1 (SALK_063517), originally obtained from the SALK collection, are knockout insertional mutants described in  containing a pROCK-derived T-DNA within exon 14. Arabidopsis heterozygous for p5cs2 (GABI_452G01), originally obtained from the GABI-Kat collection, are insertional mutants described in , containing a PAC161- derived T-DNA within exon 18. As reported in [1, 3]p5cs2/P5CS2 is embryo lethal in homozygous state and must be propagated in heterozygous state. Arabidopsis homozygous for p5cs1 and heterozygous for p5cs2 (p5cs1 p5cs2/P5CS2), have been characterized and described elsewhere [1, 3]. For segregation analysis, seeds from a self-fertilized p5cs1 p5cs2/P5CS2 plant were stratified for three days at 4°C, surface-sterilized, and germinated on MS1/2 plates supplemented with 12 μg/ml sulfadiazine. Segregation ratios were calculated by scoring the number of resistant over susceptible plantlets, and confirmed by PCR analysis of random samples. using primers 5’-CAAGCAATGGTGGAAGAGTAAA-3’ and 5’- CGGGGCTCAAGAAAAATCC -3’ for the sulfadiazine resistance gene. For embryo analysis, siliques derived from self-fertilized wild types, p5cs2/P5CS2 mutants, or p5cs1 p5cs2/P5CS2 mutants, were dissected and analyzed under a Zeiss Stevi SV 6 light stereomicroscope (Carl Zeiss Microimaging GmbH, Jena, Germany). Digital images were acquired with a Jenoptik ProgRes® C3 digital camera (Jenoptik, Jena, Germany). All the analyses have been repeated at least four times. Statistical significance was inferred from percentage data by using χ2 analysis.
In crosses between p5cs1 p5cs2/P5CS2 and flowering time mutants the F1 generation was allowed to self-fertilize and the presence of the p5cs2 mutant allele was assessed from the F2 generation, by sulfadiazine selection or by PCR genotyping of the sulfadiazine resistance gene. To confirm the data and rule out any possible interference of the flowering time mutant genotypes in the transmission of the p5cs2 mutation, reciprocal crosses between p5cs1 p5cs2/P5CS2 and wild type were performed. Transmission of the T-DNA insertion on P5CS2 gene was assessed, either by sulfadiazine selection of outcrossed seeds germinated on sulfadiazine-containing solid medium, or by PCR genotyping of sulfadiazine resistance gene on plantlets grown without selection Statistical significance was inferred from percentage data by using χ2 analysis.
For orcein staining, pollen was collected by dabbing mature flowers, from four weeks old plants, on a microscope slide. After a brief incubation in 1% acetic orcein, the pollen grains were rinsed in 50% acetic acid and examined under a Leitz Laborlux D light microscope (Leitz, Wetzlar, Germany) equipped with a Jenoptik ProgRes® C3 digital camera (Jenoptik, Jena, Germany). For histological analysis, floral buds, of different developmental stages, were embedded in Technovit 7100 (Kulzer), and 3-mm cross-sections were stained with 1% Toluidine blue as described in  and analyzed under a Leitz Laborlux D light microscope (Leitz, Wetzlar, Germany). For evaluation of pollen vitality, flower buds or isolated anthers were collected, fixed overnight in Carnoy’s fixative (6 alcohol:3 chloroform:1 acetic acid), and stained with a modified Alexander’s stain as described by Peterson et al. (2010). For DAPI (4’, 6-diamidino-2-phenylindole) staining , mature pollen was collected, incubated 30’ in DAPI staining solution (0.1 M sodium phosphate buffer, pH 7, 1 mM EDTA, 0.1% Triton-X-100 and 0.4 μg/ml DAPI) and analyzed with a Zeiss Axioskop 2 plus microscope (Carl Zeiss Microimaging GmbH, Jena, Germany) equipped with a DAPI filter set consisting of an excitation filter (BP 365/12 nm), a beam splitter (395 nm), and an emission filter (LP 397 nm). Acquisition of digital images was made with a Jenoptik ProgRes® C3 digital camera (Jenoptik, Jena, Germany).
Pollen grains were separated by size (large and small) under a Zeiss Stevi SV 6 light stereomicroscope (Carl Zeiss Microimaging GmbH, Jena, Germany) and pools of about 500 pollen grains were prepared and frozen at −20°C. Pollen DNA was extracted from these samples with a modified CTAB (cetyl trimethylammonium bromide) protocol according to . Because of the presence of a tough pollen coat, major modifications were introduced in the initial steps consisting in squashing pollens between a microscope slide and a cover slip, retrieving them with 50 μl CTAB buffer and heating the solution for 15’ at 95°C. PCR conditions were 3’ at 94°C followed by 35 cycles of 30” at 94°C, 1’ at 58°C, and 50” at 72°C. The primer pairs used were 5’-CAAGCAATGGTGGAAGAGTAAA-3’ and 5’-CGGGGCTCAAGAAAAATCC-3’ for the sulfadiazine resistance gene, 5’-GGAGCAGAATGGTTTTCTCG-3’ and 5’-TATCTGGGAATGGCGAAATC-3’ for the T-DNA insertion on P5CS2, 5’-GGAGCAGAATGGTTTTCTCG-3’ and 5’-TGGAAAACAGCAGCACTGTC - 3’ for the gene P5CS2, 5’-CTGTTGGGGGTAAACTCATTG-3’ and 5 -GCGTGGACCGCTTGCTGCAACT-3’ for the T-DNA insertion on P5CS1, 5’-CTGTTGGGGGTAAACTCATTG-3’ and 5’-CTCTGCAACTTCGTGATCCTC-3’ for the gene P5CS1.
Mature pollens from stage 13 flowers  were collected, transferred to glass slides coated with freshly prepared germination medium (5 mM CaCl2, 1 mM MgSO4, 5 mM KCl, 0.01 mM H3BO3, 10% sucrose and 1.5% agarose, pH 8.0), and kept overnight in a moist chamber at 21°C. To minimize in vitro germination variability, pollen from wild type plants was always included nearby pollen from p5cs1 p5cs2/P5CS2 plants on the same microscope slide.
For pollen germination analysis, the slides were examined under a Leitz Laborlux D microscope (Leitz, Wetzlar, Germany) and digital pictures of randomly chosen fields were acquired with a Jenoptik ProgRes® C3 digital camera (Jenoptik, Jena, Germany). To determine pollen germination efficiency, the number of germinated and non-germinated pollens was scored from five, randomly chosen, fields per replica in four independent experiments. Statistical significance was inferred from percentage data by using χ2 analysis.
Proline measurements were modified from  as follows: Mature pollen grains (~ 5–10.000) were collected on a microscope slide, smashed with a slide cover glass, and retrieved with 30 μl of a 3% (w/v) aqueous solution of sulfosalicylic acid. After centrifugation (12.000 ×g for 10’), the supernatant was added to 30 μl of acid-ninydrin and 30 μl of glacial acetic acid and let react for 2 hours at 80°C in a micro-test tube made up by the tip of pasteur pipette heat-sealed at the two ends. The reaction mixture was extracted with 60 μl of toluene and its optical density measured at 520 nm with a NanoDrop 2000 micro-spectrophotometer (Thermo Scientific, Wilmington, USA). Proline concentration was determined from a standard curve built with L-proline. In vitro complementation was attempted by spotting mature pollens in microscope slides coated with germination medium supplemented with 10 μM L-proline. In vivo complementation was performed by daily spraying early developing inflorescences with 10 μM proline. Mature pollens from stage 13 proline-treated flowers were collected and analyzed as described above.
This work was partially supported by research grants from MIUR, FIRB, and ERA-PG to PC and by a research grant from Università La Sapienza (Progetto di Ateneo) to MT.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.