Links Between Morphology and Function of the Pollen Wall: an Experimental Approach

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/291420840

Links between morphology and function of the pollen wall: An experimental approach

Article in Botanical Journal of the Linnean Society · April 2016 Impact Factor: 2.53 · DOI: 10.1111/boj.12378

7 authors, including:

Christian Raquin François Brisset French National Centre for Scientific Resea… Université Paris-Saclay

42 PUBLICATIONS 507 CITATIONS 112 PUBLICATIONS 1,123 CITATIONS

SEE PROFILE SEE PROFILE

Beatrice Albert Pierre-Henri Gouyon Université Paris-Sud 11 Muséum National d'Histoire Naturelle

23 PUBLICATIONS 364 CITATIONS 156 PUBLICATIONS 4,557 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Alexis Matamoro-Vidal letting you access and read them immediately. Retrieved on: 20 April 2016 Botanical Journal of the Linnean Society, 2016, 180, 478–490. With 6 ﬁgures

Links between morphology and function of the pollen wall: an experimental approach

ALEXIS MATAMORO-VIDAL1,2,*, CHRISTIAN RAQUIN2, FRANCß OIS BRISSET3,HEL ENE COLAS1, BENJAMIN IZAC2,BEATRICE ALBERT2† and PIERRE-HENRI GOUYON1†

1Departement Systematique et Evolution, Museum national d’Histoire naturelle, UMR 7205 MNHN- CNRS, Paris, 75005, France 2Universite Paris Sud-11. Ecologie, Systematique et Evolution, UMR 8079 CNRS-AgroParisTech, Orsay, 91405, France 3Universite Paris Sud-11. ICMMO, UMR 8182 CNRS, Orsay, 91405, France

Received 17 April 2015; revised 3 December 2015; accepted for publication 18 December 2015

The wall of pollen grains exhibits morphological variation in many features including apertures, ornamentation and thickness, but the function of these characters remains to be clariﬁed. It has been suggested that they are involved in the accommodation of volume changes (harmomegathy). To investigate this further, we developed a protocol that induces a controlled hydration of the pollen without affecting its metabolism and we applied it to six species differing in their pollen wall morphology. The entry of water caused pollen swelling and volume increase leading to breakage of the wall and/or of the plasma membrane, such that the per cent of intact grains was negatively correlated with the level of hydration. Qualitative and quantitative differences were observed between the species. Breakage of the exine was observed only in pollen lacking apertures and with thin exine. Variation in the exine ornamentation and thickness could explain the interspeciﬁc differences observed for the rates of breakage of the plasma membrane. Our results suggest that pollen wall morphology matters for survival and maintenance of pollen integrity further to volume increase due to hydration. We propose a rationale for future studies that should allow disentangling the contribution of different pollen morphological and physiological features to harmomegathy. © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016, 180, 478–490.

ADDITIONAL KEYWORDS: aperture pattern – exine ornamentation – harmomegathy – natural selection.

INTRODUCTION drates and eventually germinates (Edlund, Swanson & Preuss, 2004). Additional hydration–dehydration Plant male success depends largely on the viability phases can occur before or during the dispersal of of pollen until the pollen tube reaches and fertilizes the grain, depending on the relative humidity of the ovules. This makes the understanding of how pollen environment and of the time of dispersal (Lisci, grains adapt to environmental ﬂuctuations an Tanda & Pacini, 1994; Pacini, 2000; Franchi et al., important issue for plant evolution. During its life 2011). cycle, a pollen grain can go through several hydra- Many structural, physiological and molecular tion and dehydration phases. When pollen reaches mechanisms are used by pollen grains to adjust to maturity in the anther, it is transferred from a liq- changes in water content and to maintain internal uid environment to an atmospheric environment. In stability (Firon, Nepi & Pacini, 2012). The external most angiosperm species, pollen dehydrates and wall of the pollen (the exine) is made of a highly remains in this state until it lands on a compatible resistant polymer called sporopollenin. Mechanisms stigma. Once it is on the stigma, the pollen rehy- allowing changes in the shape and in the volume of the wall in order to accommodate the variation in *Corresponding author. E-mail: [email protected] the volume of the cytoplasm caused by changing †These authors contributed equally. hydration are thus necessary to avoid pollen

478 © 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016, 180, 478–490 POLLEN WALL MORPHOLOGY AND FUNCTION 479 breakage. The term harmomegathy was proposed by uniformly thin. In these pollen grains, the whole sur- Wodehouse (1935) to qualify such an accommodation face of the wall can be considered as an aperture. process, which is necessary to allow a retraction This type of pollen is thus referred to as omniapertu- (bending) of the wall during dehydration or an rate (Thanikaimoni et al., 1984). At the other extension (stretching) during hydration. Here, we extreme, there are pollen grains with a uniformly present an experimental approach aimed at thick exine. studying in a quantitative manner how pollen mor- The innermost layer of the pollen wall, the intine, phological and physiological characteristics may is beneath the exine and borders the surface of the help the grain to accommodate a volume increase cytoplasm. The intine is composed of pectin and cel- and avoid pollen breakage further to hydration. On lulose and it is much more capable of stretching the basis of the data obtained from this approach and contraction than the exine (Heslop-Harrison & and from published work, we propose a hypothetical Heslop-Harrison, 1982). Variability in the thickness model predicting response of the grain to hydration of the intine wall might affect the stretching pro- depending on its morphological and physiological cess and the efficiency of volume-change accommo- characteristics. dation. Two characteristics of the pollen wall, aperture Differences in aperture number, wall ornamenta- sites and exine ornamentation, have been suggested tion, and thickness, are likely to produce differences to be involved in volume-change accommodation on in harmomegathic efficiency. We studied whether the basis of comparative and theoretical approaches pollen morphology may affect the capacity of the (Wodehouse, 1935; Payne, 1972; Heslop-Harrison, grain to accommodate a volume-increase resulting 1979b; Muller, 1979; Blackmore & Barnes, 1986; from hydration. For this, we developed a protocol Thanikaimoni, 1986; Scotland, Barnes & Blackmore, that allows the hydration of the grains in a dose- 1990; Halbritter & Hesse, 2004; Chichiricco, 2007; dependent manner, without changing metabolism: Katifori et al., 2010; Volkova, Severova & Polevova, pollen grains were placed in solutions with different 2013). Apertures are sites on the wall where the concentrations of a non-metabolic sugar, which cre- exine is thin or absent (Fig. 1A–D). During dehydra- ates different levels of hydration without affecting tion of the grain, the membrane of the aperture sites the metabolism of the grains. We scored the rates folds inward, so that the edges of each aperture are of breakage in the exine and in the plasma mem- touching each other (Volkova et al., 2013), closing up brane of pollen grains exposed to four levels of the aperture site. It has been shown by mathemati- hydration. This was done for pollen grains of six cal modelling that the area and the shape of the species that differ in their aperture sites, wall orna- apertures contribute to harmomegathy by reducing mentation and thickness. The percentage of intact the necessity of the wall to stretch and bend in order pollen grains was found to be negatively correlated to accommodate volume-changes (Katifori et al., with the level of hydration. In addition, qualitative 2010). However, even if the apertural sites may and quantitative differences between the species accommodate a part of the volume-changes, some were observed. We summarize in a model how the flexibility of the wall is required: several species pro- morphological properties of the pollen wall could duce pollen with tiny apertures or totally lacking explain these differences. apertures (Fig. 1E, F), and they are still able to accommodate volume changes. This additional flexibility might be provided by other properties of the MATERIAL AND METHODS exine. Exine ornamentation is the pattern of the outer LIVING MATERIAL wall of the pollen (Fig. 1A0–F0) and is variable. For Pollen grains of six species were studied (Table 1). example, some species produce pollen with network- Plants were grown in the laboratory greenhouse like exine patterns made of wide spaces bordered by (Orsay, France) using material obtained from the ridges of exine narrower than these spaces (reticu- Parc Botanique de Launay (PBL) for the following late pattern, Fig. 1D0,E0). Other exine patterns are species: Carica papaya L. (voucher PBL-000536), Iris made of exine units (pilea) that have loose connec- germanica L. (voucher PBL-011139), Jatropha inte- tions with each other (crotonoid pattern, Fig. 1F0). gerrima Jacq. (voucher PBL-018013), Nicotiana syl- These exine units may move apart from each other vestris Speg. & Comes (voucher PBL-019066), and facilitate volume-changes (Katifori et al., 2010). Ricinus communis L. (voucher PBL-020382). For Their separation has been considered as adaptation Matthiola tricuspidata (L.) R.Br., living material was to volume change during water uptake (Hesse, obtained from seeds provided by Royal Botanic Gar- 1999). There is also appreciable variation in exine dens, Kew (voucher K 000984541, seed bank acces- thickness. Some pollen grains have an exine that is sion number 60501).

AA′ BB′

CC′ DD′

EE′ FF′

Figure 1. Pollen morphology of the species investigated (scanning electron microscope images). A, A0, Ricinus communis. Pollen (A) has three apertures (triaperturate). Exine ornamentation (A0) consists of more or less rounded depressions < 1 lm in diameter and the distance between the depressions is equal or greater than their diameter (perforate). B, B0, Carica papaya. Pollen (B) is triaperturate and exine ornamentation is perforate (B0). C, C0, Nicotiana sylvestris. Pollen (C) is triaperturate. Ornamentation (C0) is perforate. D, D0, Iris germanica. Pollen (D) has a unique aperture (monosulcate). Ornamentation (D0) is a network-like pattern made of wide spaces bordered by ridges of exine narrower than these spaces (reticulate). E, E0, Matthiola tricuspidata. Pollen (E) has no apertures (inaperturate) and exine (E0)is reticulate. F, F0, J. integerrima. Pollen (F) has no visible apertures, but the exine is uniformly thin such that the entire wall functions like an aperture (omniaperturate). Ornamentation (F0) is made of exine units (pilea), which have loose connections with each other (crotonoid).

POLLEN MORPHOLOGY cleaned using an ultrasonic cleaner and critical point Pollen morphology was observed with an FEG dried with an Emitech K850 critical-point dryer scanning electron microscope using experimental con- Quorum Technologies Ltd. Lewes, East Sussex, UK. ditions chosen to avoid the use of a metallic coating For each species, exine thickness (here taken as and deep penetration in the organic sample. To the thickness of the wall and of the ornamentation) achieve that, a 1-kV high voltage was used with a low was measured for ten grains from one individual, probe current, permitting observation without any acetolyzed according to the technique of Erdtman fatal charging effect and allowing a detailed sample (1969) and photographed under a light microscope. surface view. Before observations, pollen grains were Measurements were made on the section where the

f diameter of the grain was highest using ImageJ Soft- ware (U.S. National Institutes of Health, Bethesda,

Grains scored Maryland, USA).

EXPOSURE OF POLLEN GRAINS TO DIFFERENT LEVELS OF HYDRATION Freshly opened ﬂowers were collected and immediately taken to the laboratory. All the plants were grown in the same plot of the greenhouse so they were

Total 24,718 exposed to similar conditions at the time of flower col- lection. In the laboratory, opened anthers were removed and gently rubbed on a cellophane tape (Hutchinsonâ, Maplewood, MN, USA.), until an appreciable amount of pollen was spread on the tape. The cellophane tape was then placed in a Petri dish, Number of individuals Replicates/individual filled beforehand with six rounds of laboratory filter paper imbibed with the osmotic medium. As the cello- m) l phane tape is permeable, pollen was in contact with 0.14 1 1 1500 0.30 3 1 4500 0,06 1 3 4420 0.130.12 6 1 1 1 7293 1500 0.19 4 1the osmotic medium, 5505 and could easily be removed by taking out the cellophane tape without touching directly the grains. Petri dishes were sealed and incu- Exine thickness ( bated for 12 h at 28 °C. To stop the reaction, the cellophane tape containing pollen was placed on a slide with Alexander’s stain and covered with a cover glass. Alexander’s stain stains the cytoplasm pink and the exine green. For each slide, a number of grains (averaging 274 36 grains per slide; N = 90 slides) (Sup- Exine ornamentation porting Information, Table S1) was monitored for cell membrane and wall disruption using a light microscope. One of the three following states was attributed to each grain: intact; disruption of the plasma membrane; or breakage of the exine (with unimpaired plasma membrane). Note that in inaperturate pollen Aperture pattern the disruption of the plasma membrane occurred always through a local breakage of the exine. For each experiment, a sample of pollen collected from a mix of flowers of an individual was distributed among four Petri dishes, each of which had a different osmotic concentration. To control for the Solanaceae Triaperturate Perforate 1.62 Iridaceae Monosulcate Reticulate 1.69 Caricaceae Tripaerturate Perforate 1.56 Euphorbiaceae Omniaperturate Crotonoid 0.54 Brassicaceae Inaperturate Reticulate 2.38 Euphorbiaceae Triaperturate Perforate 1.71 – – effects of manipulation, a portion of the sample was – – – – placed with the cellophane tape directly on a slide with Alexander’s stain and covered with a cover

Eudicots glass (control media). We studied as many individuals and replicates as possible given material avail- ability at the time of the experiments. The number of individuals studied per species and the number of

(L.) R.Br Eudicot replications are given in Table 1. Jacq. Eudicot The osmotic medium was adapted from a medium L. Eudicot used routinely to germinate pollen grains in vitro L. Monocot L. Eudicot (Bergamini-Mulcahy & Mulcahy, 1983). This mineral

salt solution contained 1.62 mM H3BO3, 1.27 mM Ca List of the species investigated with their morphological characteristics, number of individuals studied, number of replicates, and total number o (NO3)2.4H2O, 0.81 mM MgSO4.7H2O. The solution was buffered to pH 6 with 0.2 mM KH2PO4 and 0.05 mM K2HPO4.3H2O. To avoid bias caused by the Speg. & Comes Carica papaya Iris germanica Table 1. grains scored Species Taxonomy Jatropha integerrima Matthiola tricuspidata Nicotiana sylvestris Ricinus communis For the exine thickness, the mean made across 10 grains and the standard deviation are given. physiological process of pollen tube growth, we did

© 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016, 180, 478–490 482 A. MATAMORO-VIDAL ET AL. not add any metabolic sugar to the solution, but RESULTS added instead mannitol, which is a non-metabolic DESCRIPTION OF POLLEN MORPHOLOGY sugar. Four different osmotic levels (0, 0.20, 0.45 and Pollen morphology of the species under investigation 1.00 mol L 1) were obtained by adding respectively is shown on Figure 1 and Table 1. The pollen wall of 0, 1.09, 2.46 and 5.47 g of mannitol to 3 mL of the Ricinus communis, Carica papaya and Nicotiana syl- mineral salt solution, in a final volume of 30 mL. vestris has three apertures (triaperturate pollen). Pollen grains placed in these media were subjected Pollen of Iris germanica has a single broad aperture to hypotonic conditions in this way (the lower the (monosulcate). Jatropha integerrima and Matthiola concentration of mannitol, the higher the level of tricuspidata produce pollen with no visible apertures hydration to which the pollen was exposed). (inaperturate). Three different exine ornamentations were observed. Ricinus communis, N. sylvestris, and STATISTICAL ANALYSES C. papaya have pollen with a perforate pattern con- To avoid bias due to breakage of pollen caused by sisting of more-or-less rounded depressions lacking manipulations or by Alexander’s stain, the percent- exine (Punt et al., 2006). These depressions are ages of intact pollen grains and of pollen breakage < 1 lm in diameter and the distance between the were, for each experiment, corrected with the rates depressions is equal or greater than their diameter. of pollen breakage observed in the control media. Some areas of the pollen of N. sylvestris have elon- This correction had only a slight effect on the results gated elements > 1 lm long arranged in an irregular because the percentages of intact pollen in the con- pattern (rugulate pattern), but the dominant pattern trol experiments (Alexander Solution) were always was perforate. For I. germanica and M. tricuspidata, high (mean = 95.9%; SD = 4; N = 18). the exine exhibits a reticulate pattern (Punt et al., We tested for differences between species in the 2006) consisting of spaces > 1 lm bordered by ridges percentage of intact pollen, the percentage of pollen (muri) of exine narrower than these spaces. The with disruption of the plasma membrane and the exine surface of J. integerrima has a crotonoid pat- percentage of pollen with exine breakage using the tern. This pattern is like a reticulate pattern except Kruskal–Wallis (KW) test. When this test yields sig- that the network is composed of separated exine ele- nificant results, then at least one of the samples is ments (pila) instead of muri (Punt et al., 2006). different from the other samples. To test for an effect The results of the measurements of exine thick- of the aperture structure (also using the KW test), ness are given in Table 1. The six species may be species were classified in four aperture patterns arranged into three classes according to their exine (omniaperturate, triaperturate, inaperturate and thickness. The first class contains J. integerrima monosulcate; Table 1). In the same way, differences with a relatively thin exine, averaging 0.54 lm. Car- between pollen ornamentations in the percentage of ica papaya, N. sylvestris, R. communis and I. ger- pollen with disruption of the plasma membrane were manica form the second class as the mean of the tested using the KW test by classifying the species in exine thickness of these species ranges between 1.56 three different patterns (perforate, reticulate and and 1.71 lm. Matthiola tricuspidata forms the third crotonoid; Table 1). The data could not be analysed class, with a relatively thick exine, averaging statistically for each of the four levels of hydration 2.38 lm. separately because this resulted in low sample sizes. Thus, the analyses were performed with the data pooled for all the levels of hydration, such that we POLLEN HYDRATION RESULTS IN BREAKAGE OF THE PLASMA could test for differences between species. Neverthe- MEMBRANE AND/OR OF THE EXINE IN A DOSE-DEPENDENT less, the plots of the raw data for each level of hydra- MANNER tion are shown in order to provide global trends. The percentage of intact pollen increases with the Tests for correlations between the percentage of concentration of mannitol (i.e. with the diminution of intact pollen and the hydration level and between hydration levels) (Fig. 2A). The two variables are the percentage of pollen with disruption of the significantly correlated (Kendall test; s = 0.473; plasma membrane and exine thickness were per- P < 0.001). When looking at each species separately formed using a Kendall test. Non-parametric statisti- (Fig. 2B), the same trend was observed, but the cor- cal tests were used as the assumptions of the relation could not be tested because of low sample parametric methods do not apply to our data (the sizes. Nevertheless, the data suggest that for all the residuals were not normally distributed). All the species, the lowest performance occurs for the analyses and graphs were performed using R Soft- high levels of hydration (i.e. for low mannitol concen- ware (R Development Core Team, 2010). trations).

Figure 2. Proportion of intact pollen as a function of the level of hydration. In all the plots, circles represent means, error bars represent 95% conﬁdence intervals and ‘N’ is the number of observations. The data are shown for all the species pooled together (A) and for each species separately (B). The statistics for the result of the Kendall test are shown.

Exposure to hydration induced breakage of the grains remained intact (Fig. 3H, J), whereas other plasma membrane in all species studied, and break- had only their exine broken (Fig. 3K), and others age of the exine only in species producing pollen had both their plasma membrane and exine broken lacking apertures. For the species producing pollen (Fig. 3I, L). In the last case, the disruption of the with apertures (R. communis, C. papaya, N. sylve- plasma membrane occurred through a local breakage stris and I. germanica), only two kinds of pollen were of the exine. observed following hydration: intact pollen and pollen with a breakage of the plasma membrane (Fig. 3A–G). The disruption of the plasma membrane PERCENTAGE OF INTACT POLLEN occurs through the apertures (Fig. 3B, D, F, G). The percentage of intact pollen differs between species Breakage of the exine was never seen in these four (KW = 22.2; d.f. = 5; P-value 0.0005) (Fig. 4A). species. In the species producing pollen lacking aper- Removing R. communis from the dataset eliminates tures (J. integerrima and Matthiola tricuspidata), these differences (KW = 5,67; d.f. = 4; P-value < 0.25). the exposure of pollen to hydration resulted in three The mean over all the media of the percentage of different effects on the pollen (Fig. 3H–L): some intact pollen is > 60% for C. papaya, N. sylvestris and

ABCD

EFGH

IJKL

Figure 3. Effects observed on the exine and on the plasma membrane, produced by the swelling of the grain. Pollen is stained with Alexander’s stain, which colours the exine green and cytoplasm pink. A, B, Ricinus communis. A, Intact hydrated pollen. B, Pollen with unimpaired exine and disrupted plasma membrane. The cytoplasm moves through the aperture. C, D, Carica papaya. C, Intact pollen. D, Pollen with disruption of the plasma membrane. As in B, the cytoplasm moves through apertures. E, F, Iris germanica. E, Hydrated pollen. F, Pollen with disruption of the cytoplasmic. G, Pollen of Nicotiana sylvestris with disrupted plasma membrane. H, I, Matthiola tricuspidata. H, Intact pollen. I, Pol- len with disrupted plasma membrane. The cytoplasm moves through a tiny breakage of the exine. J–L, J. integerrima. J, Intact pollen. K, Pollen with exine breakage, but unimpaired plasma membrane. L, Pollen with breakage of the exine and disruption of the plasma membrane.

J. integerrima. For M. tricuspidata and I. german- PERCENTAGE OF POLLEN WITH DISRUPTED PLASMA ica, the mean is between 40% and 60%. Ricinus com- MEMBRANE < munis has an average of 40% of intact pollen, the We found signiﬁcant differences between species in lowest value of all the species investigated. The data the percentage of pollen with disruption of the cell of intact pollen as a function of hydration level for membrane (KW = 30.35; d.f. = 5; P-value < 0.001) each species separately (Fig. 2B) show that R. com- (Fig. 4A). These differences hold if R. communis or < munis has a relatively poor performance ( 50% of J. integerrima are removed from the dataset intact grains) at all the levels of hydration except for (KW = 13.3; d.f. = 4; P-value < 0.01 for R. commu- the lowest (1 M mannitol); M. tricuspidata and nis, and KW = 18.5, d.f. = 4, P 0.001 for J. inte- < I. germanica are 50% for the high levels of hydra- gerrima), but not if both R. communis and > tion (0 and 0.2 M mannitol), but 50% for the low J. integerrima are removed (KW = 2.83; d.f. = 3; ones; and C. papaya, N. sylvestris and J. integerrima P-value < 0.5); therefore, these two species account > are somewhat 50% of intact pollen for all the for the differences observed. The data for each level media.

A % All media Pollen intact or 100 Pollen with disruption of the cell membrane

Pollen with broken exine and unimpaired cell membrane 80

n = 16 n = 24 n = 12 n = 4 n = 12 n = 4 Ricinus MatthiolaIris Nicotiana Jatropha Carica

B Disruption of the cell membrane or % 100

0 n n = 4 n = 6 n = 3 n = 1 n = 3 n = 1 n = 4 n = 6 n = 3 n = 1 n = 3 n = 1 n = 4 n = 6 n = 3 n = 1 n = 3 n = 1 n = 4 n = 6 n = 3 n = 1 n = 3 n = 1

a a s a a l Iris us ris na ha us n c Iris o I Iri n a i a pha cin ti o i Carica Carica tr Cari Ricinus Carica Ricinus R Ric a Jatropha Matthiola Nicotiana Jatropha Jatrop J Matthiola Nicotiana Matthi Nico Matthiol Nicoti 0 0.2 0.45 1 Mannitol (M)

Broken exine and unimpaired plasma membrane C %

n = 4 n = 4 n = 4 n = 4 n = 6 n = 6 n = 6 n = 6 n = 3 n = 3 n = 3 n = 3 n = 1 n = 1 n = 1 n = 1 n = 3 n = 3 n = 3 n = 3 n = 1 n = 1 n = 1 n = 1 0 0.2 0.45 1 0 0.2 0.45 1 0 0.2 0.45 1 0 0.2 0.45 1 0 0.2 0.45 1 0 0.2 0.45 1 Mannitol (M) Ricinus MatthiolaIris Nicotiana Jatropha Carica

Figure 4. Proportion of intact pollen (white circles), of pollen with disruption of the plasma membrane (black circles), and of pollen with breakage in the exine only (grey circles) as a function of the species. For each species, the data are shown for all the levels of hydration pooled together (A) and for each level of hydration separately (B, C). of hydration separately (Fig. 4B) suggest that the qualitatively similar for the high levels of hydration, relative differences between species in the percent- as compared with the full dataset (Fig. 4A). It could age of pollen with disrupted plasma membrane are be that these interspeciﬁc differences decrease for

© 2016 The Linnean Society of London, Botanical Journal of the Linnean Society, 2016, 180, 478–490 486 A. MATAMORO-VIDAL ET AL. low levels of hydration (0.45 and 1.00 M mannitol), with the plasma membrane disrupted relatively to but this could not be tested rigorously because of low the perforate and reticulate patterns (data not sample sizes. shown). There was a significant positive correlation between exine thickness and the percentage of pollen PERCENTAGE OF POLLEN WITH BROKEN EXINE AND with disruption of the plasma membrane (Kendall UNIMPAIRED PLASMA MEMBRANE test; s = 0.278; P-value < 0.002) (Fig. 5B). This ten- There were also interspecific differences in the per- dency remained when looking at each level of hydra- centage of pollen with broken exine and unimpaired tion separately (data not shown) and it could be that plasma membrane (KW = 53.6; d.f. = 5; P- the effect of exine thickness is bigger for the high value < 0.001) (Fig. 4A), due to the fact that this levels of hydration (0 and 0.2 M mannitol) but this effect was observed only in J. integerrima and could not be tested because of the resulting low M. tricuspidata. These two species have nevertheless sample sizes. different behaviours (Fig. 4C): J. integerrima has on average much higher percentage than M. tricuspi- = = data (J. integerrima: mean 15.42; SD 7.6; M. tri- DISCUSSION cuspidata: mean = 2.7; SD = 3.44). In addition, M. tricuspidata has a roughly stable percentage for Understanding the relationship between structure all the levels of hydration, whereas it is variable for and function of morphology is a longstanding ques- J. integerrima (Fig. 4C). tion in biology (e.g. Lauder, 1981). In the case of pollen, the involvement of wall morphology in the accommodation of volume change was recognized RELATIONSHIPS BETWEEN POLLEN MORPHOLOGY AND THE long ago (Wodehouse, 1935). As the gametophytic EFFECTS INDUCED BY THE HYDRATION phase represents a crucial stage in the life cycle of The diminution of intact pollen grains was due to plants, it is of interest to know how its fitness is breakage of the plasma membrane for all the species affected by environmental fluctuations. This study and to breakage of the exine in the inaperturate spe- falls into such an approach, by allowing the mea- cies (J. integerrima and M. tricuspidata) only surement of how the structure of the pollen wall (Fig. 4). Leaving aside exine breakage, which is may or may not accommodate changes in the volume observed only in species lacking apertures, we tested of the grain further to hydration and avoid pollen for an effect of the morphological properties on the breakage. percentage of disruption of the plasma membrane. Samples that were exposed to high levels of hydra- We found significant differences in the percentage tion had lower percentages of intact grains than of pollen with disruption of the plasma membrane samples that were exposed to lower levels (Fig. 2). due to aperture pattern (KW = 18.15; d.f. = 3; P- This indicates that our experimental conditions effec- value < 0.001), but these differences were completely tively induce a stress to the pollen in a dose-depen- attributable to the omniaperturate pattern (J. inte- dent manner: the higher the level of hydration the gerrima) (KW after removal of J. integerrima from lower the percent of intact grains. The effects the dataset = 3.97; d.f. = 2; P-value < 0.15). observed are thus due to the hydration induced by We found differences between the exine ornamenta- our experimental conditions and not by an experi- tions patterns in their percentage of pollen with the mental artefact. This stress might be similar to those plasma membrane disrupted (KW = 18.1; d.f. = 2, P- encountered by pollen grains in natural conditions, value < 0.001) (Fig. 5A). The percentage is highest for example, when they are exposed to rainfall or for the perforate ornamentation (C. papaya, covered in dew. Pollen viability is affected by such N. sylvestris and R. communis), intermediate for the environmental fluctuations (Lisci et al., 1994) and it reticulate pattern (I. germanica and M. tricuspidata) may be that pollen has been adapted to these varia- and it attains its lowest value with the crotonoid pat- tions. One way to achieve this is to evolve a morphol- tern (J. integerrima). Testing differences between ogy that allows accommodating the increase of pairs of ornamentation patterns systematically volume induced by abrupt hydration and swelling. resulted in significant differences: reticulate/perforate Under this scenario, differences in the response to (KW = 3.9; d.f. = 1; P-value < 0.05); perforate/croto- the hydration are expected between pollen grains noid (KW = 14.5; d.f. = 1; P-value < 0.001); reticulate/ exhibiting morphological differences. crotonoid (KW = 11.3; d.f. = 1; P-value < 0.001). By The exposure to hydration caused inter-specific dif- considering each level of hydration separately we ferences in the rates of wall breakage and/or of dis- found that, for all the levels of hydration, the croto- ruption of the cell membrane breakage (Figs 3, 4). noid pattern has still the lowest percentage of pollen Exine breakage was seen only in pollen grains

n = 24

n = 36 membrane (%) Disruption of cell 20 40 60

n = 12 10 30 50 70

Crotonoid Reticulate Perforate (Jatropha) (Iris + Matthiola) (Carica + Nicotiana + Ricinus) Exine ornamentation

n = 16

n = 12 n = 24

n = 4

membrane (%) n = 4 Disruption of cell n = 12 0 20406080

0.54 1.56 1.62 1.69 1.71 2.38 (Jatropha) (Carica) (Nicothiana) (Iris) (Ricinus) (Matthiola) Exine thickness (µm)

Figure 5. Proportion of pollen grains with disruption of the plasma membrane, as a function of exine ornamentation pattern (A) and exine thickness (B). Circles are means made across all the media. Similar results were obtained by considering each level of hydration separately (not shown). lacking apertures (M. tricuspidata and J. inte- The interspecific differences observed for the rate gerrima), never in aperturate pollen. This suggests of breakage of the plasma membrane might be that apertures contribute to the accommodation of related to exine ornamentation and thickness. There volume change due to abrupt hydration by avoiding is a positive correlation between the rate of cell breakage of the exine. The percentage of pollen with membrane disruption and the thickness of the exine. exine breakage and unimpaired cell membrane was Thin exine walls are more flexible (Katifori et al., much higher in J. integerrima than in M. tricuspi- 2010; and references therein) and thus could exert data, suggesting that the wall of J. integerrima pol- fewer constraints on the plasma membrane during len is much more delicate than the wall of hydration. Regarding ornamentation patterns, the M. tricuspidata. Accordingly, we found that the highest rate of breakage of the plasma membrane exine of J. integerrima is nearly five times thinner was found for the perforate pattern, followed by the than that of M. tricuspidata. We may conclude that reticulate pattern and the crotonoid pattern. The the thin wall of J. integerrima does not tolerate an outer exine of the crotonoid pattern is made of units increase of the volume of the cytoplasm. of exine (pilea) that are not closely connected with The rate of breakage of the plasma membrane of each other. This facilitates a deformation that makes J. integerrima was the lowest of all the species. The the surface of the grain convex, as it is the case pollen of J. integerrima is omniaperturate. This type when pollen swells (Katifori et al., 2010). Conversely, of pollen has an exine that is uniformly thin and an the perforate pattern is composed of a network of intine that is uniformly thick (Thanikaimoni et al., ridges of exine with tiny openings. A high stiffness is 1984; Furness, 2007). Thickness of the intine wall of thus expected for this morphology. The reticulate J. integerrima could help to maintain integrity of the pattern may be expected to have an intermediate plasma membrane further to hydration. flexibility, as it is made of ridges of exine that are

firmly connected, but with relatively large openings. content (Firon et al., 2012) and on the water It could be that such differences in exine flexibility exchanges between the cytoplasm and the environ- play a role in the avoidance of pollen breakage. ment, which are in part regulated by the apertures In this study, we have shown that there are quali- (Heslop-Harrison, 1979a). If there is water influx, tative and quantitative differences between species then a swelling of the grain is expected. In this case, in how their pollen responds to hydration. In a quali- if the wall is plastic (i.e. capable of deformation tative point of view, it seems clear that the inapertu- through stretching), it will accommodate the swelling rate, thin and crotonoid-like exine of J. integerrima of the cytoplasm (Fig. 6B). Such plasticity is expected pollen differs from the others species in way that in species with apertures, loose exine ornamentation make it much more delicate. However, it remains to pattern and thin exine (Katifori et al., 2010). If the be established whether the quantitative difference wall is not plastic, the swelling of the cytoplasm observed among all species are attributable to differ- might be constrained by the wall (Fig. 6C). In this ences in pollen morphology. This could not be estab- case, two configurations should occur, depending on lished unambiguously in this study because the the strength of the exine and the intine wall. If the sampled species differ from each other by too many exine is delicate, the exine will break because of the variables (aperture pattern, exine ornamentation stress induced by the swelling of the cytoplasm and thickness). Moreover, pollen water content plays (Fig. 6D). This effect should be amplified if the intine also a crucial role in pollen response to hydration. wall is thick and robust. Such exine rupture due to a The water content of the pollen after anther anthesis sudden and large expansion of the grain has been is variable between species. In some species, pollen reported in Cupressus L. and in Montrichardia is released with a content of water greater than 30% Crueg., pollen of which has a thin exine and a thick [pollen partially hydrated, according to the terminol- intine (Weber & Halbritter, 2006; Chichiricco et al., ogy of Franchi et al. (2002)], whereas in other species 2009; Danti et al., 2011) and is also observed here in pollen has < 30% of water (pollen partially dehy- pollen of J. integerrima which presents similar char- drated). The response of the grains to hydration is acteristics. Alternatively, if the exine is rigid, it will likely to depend on the initial water content: par- constrain the membrane of the cytoplasm and the tially hydrated pollen having higher initial water intine (Blackmore & Barnes, 1986) and this will pro- content than partially dehydrated pollen, the latter voke the disruption of the plasma membrane might be subject to higher stress than the former. (Fig. 6E, E0) if the difference in osmotic pressure on Thus, the possibility that the effects observed result the different sides of the membrane is high enough. both from the morphological properties of the pollen This phenomenon, comparable to blowing a balloon wall and from the initial water content cannot be confined in a rigid box, should be amplified if the ruled out. Further studies controlling for initial intine is thin and delicate. In the case of aperturate water content and pollen morphology are required to pollen, the disruption of the plasma membrane disentangle the relative contribution of these two fac- should occur through an aperture (Fig. 6E), whereas tors to the capacity of survival of the grain further to in the case of inaperturate pollen, it should occur hydration. We propose a model predicting pollen through local breakage of the exine (Fig. 6E0). response to hydration depending on pollen morphology and water content. The validity of this model remains to be tested, but it provides a roadmap for CONCLUSIONS future experiments aiming to answer the question of the contribution of pollen morphological and physio- We present a method that allows to induce a stress to logical features to the accommodation of a volume pollen in a dose-dependent manner. This stress, a increase and to the avoidance of pollen breakage hydration resulting in a swelling of the grain, is simi- further to hydration. lar to those encountered by pollen grains in natural The model (Fig. 6) states that increased hygro- conditions (Lisci et al., 1994; Pacini, 2000). Using this metry in pollen environment during the pollen dis- method to realize experimental observations on a persal phase may have different effects on the grain, group of species differing in their pollen morphology depending on water exchanges (driven by pollen allowed us to demonstrate different responses, which water content and hydrodynamics between the cyto- depend to some extend on the morphology of the wall. plasm and the environment) and on the rigidity and In particular, the presence of apertures and the thick- the flexibility of the wall (driven by wall morphologi- ness of the wall seem to facilitate pollen volume cal properties). When pollen is placed in hypotonic increase and avoidance of wall breakage. From an conditions, cytoplasm eventually hydrates and its evolutionary perspective, pollen wall morphology volume increases (Fig. 6A). Such hydration and might have played a role in the evolutionary success volume increase would depend on the initial water of eudicots (Furness & Rudall, 2004; Matamoro-Vidal

Favoured by: Low pollen initial water content Low Hydration Permeability of the pollen wall High

Plastic exine Cytoplasm A B Exine Intact pollen

Aperture cell membrane + intine C Delicate exine Non-plastic & robust intine exine D Exine disrupted

E E’ Rigid exine Cell membrane & delicate intine disrupted Cell membrane disrupted and local breakage of the exine

Figure 6. Model predicting possible responses of pollen grains following hydration. A, During hydration, water eventually enter into the pollen grain and causes a swelling. B, If exine is plastic enough, the cytoplasm and the exine will grow and reach an equilibrium state. All the structures remain intact. C, If the exine lacks flexibility, it constrains the swelling of the plasma membrane and intine. D, If the level of hydration is high enough, the exine delicate and the intine resistant enough, then the exine will break. E, E0. In the case that the exine is resistant and the intine is delicate, then the plasma membrane breaks through an aperture (E) or through a local breakage of the exine (E0). et al., 2012). At least two selective pressures are useful comments and corrections on an earlier version known to act on pollen morphology: one related to the of the manuscript. We thank E. Couradeau; C. Dje- efficiency of pollen germination and survival (Dajoz, diat; A. Dubois, C. Rausch, C. Sanchez, L. Saunois Till-Bottraud & Gouyon, 1991, 1993; Till-Bottraud and the ‘Plateforme de Microscopie et d’Imagerie du et al., 1999) and another one related to the retraction MNHN’ for valuable technical assistance. We thank of the wall during dehydration (Halbritter & Hesse, the Royal Botanic Gardens, Kew, and the Parc Botani- 2004; Volkova et al., 2013). We propose the existence que de Launay, Orsay, for providing seeds and plant of an additional selective pressure that might be act- material. This work was supported by the ‘Action ing on pollen wall morphology through its influence Transversale du Museum Formes Possibles, Formes on survival upon hydration and swelling of the grain. Realis ees’ (Museum national d’Histoire naturelle).

ACKNOWLEDGEMENTS REFERENCES We thank T. Deroin, J. Dupont, T. Giraud, M. Lopez- Bergamini-Mulcahy G, Mulcahy DL. 1983. A comparison Villavicencio, A. Ressayre and G. Restoux for helpful of pollen tube growth in bi- and trinucleate pollen. In: discussions and advice. J. Doyle, C. Furness and J. Mulcahy DL, Ottaviano E, eds. Pollen: biology and implica- Moustakas-Verho and an anonymous reviewer made tion for plant breeding. New York: Elsevier, 29–33.

Blackmore S, Barnes SH. 1986. Harmomegathic mecha- Katifori E, Alben S, Cerda E, Nelson DR, Dumais J. nisms in pollen grains. In: Blackmore S, Ferguson IK, eds. 2010. Foldable structures and the natural design of pollen Pollen and spores: form and function. Linnean Society Sym- grains. Proceedings of the National Academy of Sciences of posium Series. London and New York: Academic Press, the United States of America 107: 7635–7639. 137–148. Lauder GV. 1981. Form and function: structural analysis in Chichiricco G. 2007. Post-shedding dehydration and in vivo evolutionary morphology. Paleobiology 7: 430–442. temporal germination of Crocus pollen. Grana 44: 142–147. Lisci M, Tanda C, Pacini E. 1994. Pollination ecophysiol- Chichiricco G, Spano L, Torraca G, Tartarini A. 2009. ogy of Mercurialis annua L. (Euphorbiaceae), an anemophi- Hydration, sporoderm breaking and germination of Cupres- lous species flowering all year round. Annals of Botany 74: sus arizonica pollen. Plant Biology 11: 359–368. 125–135. Dajoz I, Till-Bottraud I, Gouyon P-H. 1991. Evolution of Matamoro-Vidal A, Furness CA, Gouyon P-H, Wurdack pollen morphology. Science 253: 66–68. KJ, Albert B. 2012. Evolutionary stasis in Euphorbiaceae Dajoz I, Till-Bottraud I, Gouyon P-H. 1993. Pollen aper- pollen: selection and constraints. Journal of Evolutionary ture polymorphism and gametophyte performance in Viola Biology 25: 1077–1096. diversifolia. Evolution 47: 1080–1093. Muller J. 1979. Form and function in angiosperm pollen. Danti R, Della Rocca G, Calamassi R, Mori B, Lippi Annals of the Missouri Botanical Garden 66: 593–632. MM. 2011. Insights into a hydration regulating system in Pacini E. 2000. From anther and pollen ripening to pol- Cupressus pollen grains. Annals of Botany 108: 299–306. len presentation. Plant Systematics and Evolution 222: Edlund AF, Swanson R, Preuss D. 2004. Pollen and 19–43. stigma structure and function: the role of diversity in polli- Payne WW. 1972. Observations of harmomegathy in pollen nation. Plant Cell 16: S84–S97. of Anthophyta. Grana 12: 93–98. Erdtman G. 1969. Handbook of palynology. New York: Haf- Punt W, Hoen PP, Blackmore S, Nilsson S, Le Thomas ner Publishing Co. A. 2006. Glossary of pollen and spore terminology. Review Firon N, Nepi M, Pacini E. 2012. Water status and associ- of Palaeobotany and Palynology 143: 1–81. ated processes mark critical stages in pollen development Scotland RW, Barnes HB, Blackmore S. 1990. Harmo- and functioning. Annals of Botany 109: 1201–1213. megathy in the Acanthaceae. Grana 29: 37–45. Franchi GG, Nepi M, Dafni A, Pacini E. 2002. Partially R Development Core Team. 2010. R: a language and envi- hydrated pollen: taxonomic distribution, ecological and evo- ronment for statistical computing. Vienna: R Foundation for lutionary significance. Plant Systematics and Evolution Statistical Computing. 234: 211–227. Thanikaimoni G. 1986. Pollen apertures: form and func- Franchi GG, Piotto B, Nepi M, Baskin CC, Baskin JM, tion. In: Blackmore S, Fergusson IK, eds. Pollen and spores: Pacini E. 2011. Pollen and seed desiccation tolerance in form and function. London: Linnean Society, 120–134. relation to degree of developmental arrest, dispersal, and Thanikaimoni G, Caratini C, Nilsson S, Grafstrom€ E. survival. Journal of Experimental Botany 62: 5267–5281. 1984. Omniaperturate Euphorbiaceae pollen with striate Furness CA. 2007. Why does some pollen lack apertures? A spines. Bulletin du Jardin Botanique National de Belgique review of inaperturate pollen in eudicots. Botanical Journal 54: 105–125. of the Linnean Society 155: 29–48. Till-Bottraud I, Vincent M, Dajoz I, Mignot A. 1999. Pol- Furness CA, Rudall PJ. 2004. Pollen aperture evolution – len aperture heteromorphism. Variation in pollen-type pro- a crucial factor for eudicot success? Trends in Plant Science portions along altitudinal transects in Viola calcarata. 9: 154–158. Comptes Rendus de l’Academie des Sciences – Series III – Halbritter H, Hesse M. 2004. Principal modes of infoldings Sciences de la Vie 322: 579–589. in tricolp(or)ate angiosperm pollen. Grana 43: 1–14. Volkova OA, Severova EE, Polevova SV. 2013. Structural Heslop-Harrison J. 1979a. An interpretation of the hydrody- basis of harmomegathy: evidence from Boraginaceae pollen. namics of pollen. American Journal of Botany 66: 737–743. Plant Systematics and Evolution 299: 1769–1779. Heslop-Harrison J. 1979b. Pollen walls as adaptive systems. Weber M, Halbritter H. 2006. Exploding pollen in Mon- Annals of the Missouri Botanical Garden 66: 813–829. trichardia arborescens (Araceae). Plant Systematics and Heslop-Harrison Y, Heslop-Harrison J. 1982. The Evolution 263: 51–57. microfibrillar component of the pollen intine: some struc- Wodehouse R. 1935. Pollen grains, their structure, identifi- tural features. Annals of Botany 50: 831–842. cation and significance in science and medicine. New York: Hesse M. 1999. Pollen wall stratification and pollination. In: McGraw-Hill. Dafni A, Hesse M, Pacini E, eds. Pollen and pollination. Vienna: Springer, 1–11.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Table S1. Raw data for the analyses presented in this study.