Effect of Aperture Number on Pollen Germination, Survival and Reproductive Success in Arabidopsis Thaliana

Annals of Botany 121: 733–740, 2018 doi:10.1093/aob/mcx206, available online at www.academic.oup.com/aob

Effect of aperture number on pollen germination, survival and reproductive success in Arabidopsis thaliana Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 Béatrice Albert1,*, Adrienne Ressayre2, Christine Dillmann2, Ann L. Carlson3, Robert J. Swanson3, Pierre-Henri Gouyon4 and Anna A. Dobritsa5

1Ecologie Systématique Evolution, Univ. Paris-Sud, CNRS, AgroParisTech, Université Paris-Saclay, 91405 Orsay cedex, France, 2UMR de Génétique Végétale, Univ. Paris-Sud, INRA, CNRS, Université Paris-Saclay, Gif sur Yvette, F-91190, France, 3Biology Department, Valparaiso University, Valparaiso, IN 46383, USA, 4Institut de Systématique, Évolution, Biodiversité, ISYEB-UMR 7205-CNRS, MNHN, UPMC, EPHE, Muséum National d’Histoire Naturelle, Sorbonne Universités, 57 rue Cuvier, CP39, F-75005, Paris, France and 5Department of Molecular Genetics and Center for Applied Plant Sciences, The Ohio State University, 015 Rightmire Hall, 1060 Carmack Road, Columbus, OH 43210, USA *For correspondence. E-mail [email protected]

Received: 3 April 2017 Returned for revision: 27 September 2017 Editorial decision: 16 November 2017 Accepted: 15 December 2017 Published electronically 18 January 2018

• Background and Aims Pollen grains of flowering plants display a fascinating diversity of forms, including diverse patterns of apertures, the specialized areas on the pollen surface that commonly serve as the sites of pollen tube initiation and, therefore, might play a key role in reproduction. Although many aperture patterns exist in angiosperms, pollen with three apertures (triaperturate) constitutes the predominant pollen type found in eudicot species. The aim of this study was to explore whether having three apertures provides selective advantages over other aperture patterns in terms of pollen survival, germination and reproductive success, which could potentially explain the prevalence of triaperturate pollen among eudicots. • Methods The in vivo pollen germination, pollen tube growth, longevity and competitive ability to sire seeds were compared among pollen grains of Arabidopsis thaliana with different aperture numbers. For this, an arabidopsis pollen aperture series was used, which included the triaperturate wild type, as well as mutants without an aperture (inaperturate) and with more than three apertures. • Key Results Aperture number appears to influence pollen grain performance. In most germination and longevity experiments, the triaperturate and inaperturate pollen grains performed better than pollen with higher aperture numbers. In mixed pollinations, in which triaperturate and inaperturate pollen were forced to compete with each other, the triaperturate pollen outperformed the inaperturate pollen. • Conclusions Triaperturate pollen grains might provide the best trade-off among various pollen performance traits, thus explaining the prevalence of this morphological trait in the eudicot clade.

Key words: Pollen aperture, Arabidopsis thaliana, pollen germination, longevity, pollen performance, competitive pollination, reproductive success

INTRODUCTION performance, from survival to germination of pollen tubes (Wodehouse, 1935; Edlund et al., 2004; Halbritter and Hesse, Pollen grains, the male gametophytes of flowering plants, are 2004; Katifori et al., 2010; Vieira and Feijó, 2016). morphologically diverse across species, exhibiting differences Although a wide range of aperture patterns exists in angio- in size, shape, and macro- and ultrastructures of their pollen sperms (Erdtman, 1952; PalDat, http://www.paldat.org/), in walls (Erdtman, 1952; Heslop-Harrison, 1971; PalDat, http:// most species apertures exhibit one of the two predominant www.paldat.org/). One type of structure commonly present on patterns, considered to be two evolutionary stases (Bailey and the pollen surface are apertures, the areas characterized by a Nast, 1943; Matamoro-Vidal et al., 2016a). Basal angiosperms reduced deposition of the pollen wall exine (Hesse et al., 2009). and monocots mainly produce monosulcate pollen grains, char- Like other pollen features, aperture patterns are highly variable acterized by a single furrow-like aperture located at the distal among species: apertures can differ in morphology (pore, fur- pollen pole, while eudicots largely produce tricolpate pollen row or both), in number (from no apertures to >100) and in grains with three furrow-like, longitudinal apertures (Walker positions on the pollen surface (polar, equatorial or distributed and Doyle, 1975; Wortley et al., 2015), resulting in eudicots throughout the surface) (Erdtman, 1952). Apertures have been sometimes being referred to as the tricolpate clade (Donoghue proposed to play multiple roles: accommodate pollen deform- and Doyle, 1989; Judd and Olmstead, 2004). ation during changing hydration conditions; facilitate water The prevalence of the tricolpate, or more generally triapertu- and gas exchange; and participate in various aspects of pollen rate, pollen among eudicot species has long puzzled palynologists

© The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. 734 Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis and evolutionary biologists (Walker and Doyle, 1975; Furness pollen grains (Prieu et al., 2016; Reeder et al., 2016) (pollen and Rudall, 2004). Previous studies have proposed that the triap- length = 24.25 ± 2.00 µm; n = 11). The mutation in the OSD1 erturate stasis in eudicots is more likely to be caused by selective gene leads to the production of functional diploid pollen grains pressures favouring this particular pollen type than by develop- due to the absence of the second meiotic division (d’Erfurth mental constraints limiting other pollen morphologies (Albert et al., 2009). Unlike in the triaperturate and four-aperturate pol- et al., 2009; Matamoro-Vidal et al., 2012; Matamoro-Vidal len grains, in the six- and eight-aperturate pollen, apertures were et al., 2016a). This idea is consistent with the general trend in no longer distributed around the pollen equator. Instead, they Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 angiosperms towards an increase in aperture number over evo- usually formed the edges of a tetrahedron or a square-based pyra- lutionary time, suggesting that aperture patterns might have an mid, respectively (Prieu et al., 2016; Reeder et al., 2016). influence on pollen reproductive performance (Walker, 1974; Walker and Doyle, 1975; Van Campo, 1976; Doyle and Hotton, 1991; Furness and Rudall, 2004). Growth conditions Although several studies have previously looked at various aspects of pollen performance (Williams and Mazer, 2016), the Plants were grown in a climatic chamber under the follow- question of whether it is indeed influenced by aperture patterns ing conditions: 16 h of light at 20 °C; 8 h of darkness at 16 °C. has remained largely unanswered (Dajoz et al., 1991, 1993; Experiments were carried out as soon as plants produced flow- Till-Bottraud et al., 2001; Edlund et al., 2004). ers (approx. 6 weeks after sowing). Here, we took advantage of recently characterized arabidopsis mutants which, together with a wild-type arabidopsis, constitute a pollen aperture series. They produce pollen with zero, three, four and a mix of four to eight apertures (d’Erfurth In vivo germination and pollen tube growth assays et al., 2009; Dobritsa et al., 2011; Dobritsa and Coerper, 2012; The in vivo germination experiments (Fig. 1) were per- Reeder et al., 2016), thereby covering most of the range of formed using nearly open flowers with non-dehisced stamens aperture numbers found in eudicots and allowing us to use pol- [stage 13 flowers (Alvarez-Buylla et al., 2010)] as the source len from a single species to assess the role of aperture patterns. of receptive pistils. Each receptive flower was removed from We used this series to test if differences in aperture numbers the plant and emasculated. Stigmas were then pollinated with affect four specific aspects of pollen performance – on-stigma pollen from the flowers that opened that day (young pollen pollen germination, growth of pollen tubes, pollen longevity grains), or from flowers that opened 3 d previously (old pollen and ability to sire seeds in competitive pollinations – and, if so, grains). Pollination was carried out on receptive flowers of the whether triaperturate pollen, the predominant class among the same genotype as pollen. Pollen grains were carefully depos- pollen of eudicots, performs better than the other pollen types ited on stigmas using tweezers under a dissecting microscope in these assays. to ensure that each pollen grain was in contact with the stigma surface. Care was taken to place enough pollen to provide suffi- cient statistical power for the germination percentage analysis, MATERIALS AND METHODS but not so much that it would make scoring germination events difficult. The average number of pollen grains per stigma was 55 ± 24. The pollinated flowers were incubated on a Petri dish Arabidopsis lines lined with filter paper dipped in water for either 30 min or 4 h, We used several Arabidopsis thaliana mutants that differ in after which their stigmas were dissected, mounted in aniline their pollen aperture number from the wild type, which has blue solution (1 mm aniline blue, 80 mm K3PO4, 15% glycerol) triaperturate pollen grains. The inp1-1 mutant produces pollen and imaged using a ×40 objective on a Zeiss Axiophot micro- without apertures (Dobritsa et al., 2011; Dobritsa and Coerper, scope. This procedure was performed without any fixation to 2012) (pollen length = 18.07 ± 1.62 µm; n = 12). This mutant avoid dislodging the non-germinated pollen grains. Each pollen exhibits well-developed siliques and a normal number of seeds grain was scored as germinated/non-germinated by focusing on per silique, indicating that this morphological change has no it, and the percentage of germinated pollen grains was calcu- gross effects on fertility under laboratory conditions (Dobritsa lated. An event was scored as germination when the length of et al., 2011). The lsq6 mutant (Dobritsa et al., 2011; Reeder the pollen tube was at least equal to the diameter of the pollen et al., 2016) produces mostly four-aperturate pollen grains and grain. All the scoring was done blindly. The experiments were a few triaperturate grains (pollen length = 27.86 ± 2.02 µm; repeated at least six times for each genotype. n = 9). Since inp1 and lsq6 plants have the Columbia (Col-0) The in vivo pollen tube growth experiments were also per- genetic background, in all experiments their mutant pollen was formed using stage 13 receptive flowers prepared as described compared with the wild-type Columbia (wt Col) pollen (pollen above. Each receptive flower was pollinated with young pollen length = 23.49 ± 1.12 µm; n = 9). lsq6 plants are tetraploid and from flowers that opened that day (from 30 to 50 pollen grains). produce diploid pollen grains (Reeder et al., 2016). The pollinated flowers were incubated for 4 h on a Petri dish The osd1-1 mutant plants are in the Nossen (No-0) genetic with a filter paper dipped in water. The pistils were fixed in FAA background and were obtained from a segregating progeny of (50 % ethanol, 10 % formaldehyde, 5 % acetic acid) overnight the osd1-1/+ plants. The Nossen pollen, called wt No-0 here, at 4 °C, rinsed three times in water and incubated in 4 m NaOH was used as the wild-type control for the osd1 pollen (wt No-0 for 30 min. The pistils were again rinsed three times in water, pollen length = 22.24 ± 0.68 µm; n = 9). The osd1 mutants incubated in aniline blue solution for 2 d, mounted in 100 % (d’Erfurth et al., 2009) produce a mix of four- to eight-aperturate glycerol and imaged using a ×10 objective on a Zeiss Axiophot Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis 735 Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019

ABC

DEF

Fig. 1. In vivo germination experiment. (A) A stage 13 receptive flower. (B) A receptive flower with stamens removed. (C) Examples of a young flower (left) and an old flower (right) used as pollen donors. (D) A stigma with deposited pollen. (E and F) Germination of pollen grains on a stigma. (F) Germinated pollen grains are indicated by asterisks, and non-germinated pollen grains are indicated by arrows. microscope. The experiments were repeated at least 12 times pollen grains differing in aperture number. To compare the for each genotype. The mean pollen tube length in micrometres results for inp1, lsq6 and wt Col, we performed an analysis for each pistil was measured using the ROI manager tool in of variance (ANOVA) followed by the Tukey’s contrast test. NIH ImageJ. All the measurements were done blindly. A Student’s t-test was performed for the comparison between the osd1 and wt No-0 results.

Statistical analysis In vivo competitive pollinations R software (R Core Team, 2014) was used to carry out statistical analyses. In order to test the effect of aperture number In vivo competitive pollinations using triaperturate and inap- on germination, we analysed the number of germinated and erturate pollen were done as previously described (Carlson non-germinated young pollen grains with different numbers et al., 2009, 2013; Fitz Gerald et al., 2014). Col-NPTII, a of apertures, which had been on stigmas for either 30 min or derivative of Col-0 containing a homozygous intergenic kana- 4 h. Because the number of germinated and non-germinated mycin resistance marker NPTII that allows for easy paternity young pollen grains is a binomial variable, we used for ana- testing, was used as a source of triaperturate pollen and was lysis a generalized linear model (GLM; an extension of the considered to be equivalent to wt Col. It has been shown pre- linear model to a non-normal variable) with binomial distribu- viously that the presence of this marker does not affect com- tion and the classical logit link function. We tested the effects petitive ability of pollen and that Col-NPTII pollen behaves of aperture number, time that pollen spent on the stigma and indistinguishably from the Col-0 pollen in mixed pollinations the possible interaction between the aperture number and time. (Carlson et al., 2009, 2013) – we have also confirmed this result Pairwise comparisons between aperture numbers for each time here. Newly bolted primary inflorescences of Col-0 plants were on a stigma and between times on a stigma for each aperture used as maternal plants for all the crosses. Flower buds at number were conducted with Bonferroni’s multiple compari- developmental stages 11–12 (Alvarez-Buylla et al., 2010) were sons adjustment. emasculated and pistils were allowed to mature to stage 14 In order to test whether pollen age affects germination differ- before performing pollinations. Anthers from stage 14 flowers ently in lines with different aperture numbers, we analysed the were used as a source of pollen. We performed control competi- number of germinated and non-germinated young and old pollen tive pollinations between Col-NPTII and Col-0, as well as com- grains from different lines 4 h after pollination. We performed petitive pollinations between Col-NPTII and inp1. Saturating GLM analysis to evaluate the effects of aperture number, pol- amounts of pollen from one father were applied under a micro- len age and aperture number–pollen age interaction. Pairwise scope (Leica ZOOM2000) on one half of the stigma surface as comparisons between aperture number for each pollen age and described (Carlson et al., 2013), followed by the addition of between pollen ages for each aperture number were conducted pollen from the competing father on the remaining half of the with Bonferroni’s multiple comparisons adjustment. stigma. For each type of competitive pollination, we performed In order to test whether aperture number affects pollen tube a minimum of seven crosses. For the Col-NPTII vs. inp1 pol- growth, we compared the mean pollen tube length between linations, two independent trials (consisting of eight and seven 736 Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis crosses, respectively) were performed. On average, mixed pol- however, germination was significantly lower for pollen with linations produced 46.1 ± 1.8 seeds (mean ± s.e.) per silique, four apertures (Fig. 2; Table 2). Four hours after pollination, the which was within 1 SD of the seed sets produced by the parents triaperturate pollen grains had significantly higher germination Col-NPTII and inp1 in single pollinations, as well as of the seed than the other two types of pollen (Fig. 2; Table 2). sets produced by competitive crosses between Col-NPTII and a In the Nossen background comparison, the triaperturate wt variety of arabidopsis accessions (Carlson et al., 2013). Mature No-0 pollen showed a significantly higher germination than siliques were collected and seed paternity was assessed for indi- osd1 (four to eight apertures) after either 30 min or 4 h (Fig. 2; Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 vidual siliques by growing seeds on MS medium (Murashige Table 2). and Skoog, 1962) supplemented with 50 µg mL–1 kanamycin. For most pollen types from both backgrounds (inp1, wt Col, Seedlings grown from the seeds that inherited the NPTII gene wt No-0 and osd1), we found that more time allowed for ger- were resistant to kanamycin, while the progeny of Col-0 or inp1 mination led to higher germination: in all these cases, the per- parents were not. centage of germinated pollen grains was higher after 4 h than after 30 min of on-stigma time (Fig. 2; Table 2). This suggests that not all pollen grains capable of germination do so equally RESULTS quickly and that the extra time allows lagging grains to catch up. Interestingly, for the lsq6 pollen grains with four apertures, Effect of aperture number and time spent on stigma on pollen we did not observe any difference in germination after either germination 30 min or 4 h (Fig. 2; Table 2), suggesting that in this case, essentially all grains capable of germination do so soon after To look at the effect of aperture number on pollen germination, pollination. the germination of young pollen grains with different aperture numbers was determined at two time points (30 min and 4 h) after pollination. To test if aperture numbers and time spent on the stigma influenced pollen germination, we first performed a Germination of pollen with different aperture numbers as a global analysis of data using a GLM test. The GLM test compar- function of pollen age ing germination between pollen of the Columbia background To determine if differences in aperture numbers affect pol- with zero (inp1), three (wt Col) or four (lsq6) apertures demon- len longevity, we compared germination of young and old strated significant effects of aperture number and time allowed pollen from lines with different aperture numbers at 4 h after for germination, as well as the interaction between these two pollination. The GLM test comparing germination between the parameters (Table 1). The GLM test comparing pollen germin- Columbia background pollen with zero, three or four apertures ation between the Nossen lines with three (wt No-0) and four to showed a significant effect of aperture number, pollen age and eight (osd1) apertures also found a significant effect of aperture the interaction between these parameters (wt Col–inp1–lsq6; number and on-stigma time, but did not find an interaction between these parameters (Table 1). Young pollen 30 min To look in more detail at the effects of aperture number and Young pollen 4 h on-stigma time on pollen germination, we then performed pair- 100 3 ap wise comparisons for different variables. When we compared 90 3 ap germination of the Columbia background pollen grains with 0 ap zero (inp1), three (wt Col) and four (lsq6) apertures that were 80 4–8 ap kept on stigmas for only 30 min, no significant difference was 70 found between the pollen grains with zero and three apertures; 4 ap 60 Table 1. Summary of the GLM results testing the effects of aper- 50 ture number (Ap. number), on-stigma time and the aperture num- 40 ber–time interaction on pollen grain germination Germination (%) 30 Model d.f. Deviance Residual Residual P-value 20 d.f. deviance 10 wt Col-inp1-lsq6 0 Null 69 625 Ap. number 2 234 67 383 <2 × 10–16*** inp1 wt Collsq6 wt No-0 osd1 –5 Ap. number + time 1 19 66 363 1 × 10 *** Columbia Nossen Ap. number + time + 2 23 64 340 1 × 10–5*** Ap. number × time Fig. 2. Germination percentages of young pollen grains with different aperture osd1-wt No-0 numbers measured at two time points after pollination. Percentage germination Null 28 235 in young pollen of the Columbia and Nossen backgrounds, 30 min and 4 h –15 Ap. number 1 63 27 172 1.5 × 10 *** after pollination. The following genotypes were compared: inp1 (n = 17 and 11 –10 Ap. number + time 1 39 26 133 4.3 × 10 *** pollination events; the first number indicates the number of pollination events Ap. number + time + 1 3.3 25 130 0.069 evaluated for pollen germination at the 30 min time point, and the second indi- Ap. number × time cates the number of pollination events evaluated at the 4 h time point), wt Col (n = 16 and 7) and lsq6 (n = 14 and 10); and wt No-0 (n = 10 and 8) and osd1 ***P < 0.001. (n = 10 and 6). Error bars represent the 95 % confidence interval. Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis 737

Table 2. Pairwise comparisons of germination percentages for In both backgrounds, a significant effect of pollen age on young pollen grains that differed in their number of apertures and germination was found for pollen grains with zero and three in the time spent on stigmas apertures: in all these cases, young pollen exhibited a higher percentage of germination than old pollen (Fig. 3; Table 4). Comparisons χ2 d.f. P-value However, no effect of pollen age was found for pollen with four A lsq6 (4)-wt Col (3): 30 min 81 1 2 × 10–19 or four to eight apertures (lsq6 and osd1; Fig. 3; Table 4). –16

lsq6 (4)-inp1 (0): 30 min 65 1 6 × 10 Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 wt Col (3)-inp1 (0): 30 min 3.87 1 0.049 lsq6 (4)-wt Col (3): 4 h 108 1 3 × 10–25 lsq6 (4)-inp1 (0): 4 h 83 1 8 × 10–20 Pollen tube growth is largely unaffected by aperture number wt Col (3)-inp1 (0): 4 h 19 1 1 × 10–5 wt Col (3): 30 min–4 h 27 1 2 × 10–7 Another factor that can potentially affect pollen repro- inp1 (0): 30 min–4 h 9 1 0.0025 ductive performance is the rate of pollen tube growth. To see lsq6 (4): 30 min–4 h 0.069 1 0.79 if differences in aperture number impact this aspect of pol- B osd1 (4–8)-wt No-0 (3): 30 min 52 1 6 × 10–13 len performance, we determined the average length of pollen osd1 (4–8)-wt No-0 (3): 4 h 10 1 0.0014 wt No-0 (3): 30 min–4 h 9.9 1 0.0016 tubes produced by grains with different numbers of aper- osd1 (4–8): 30 min–4 h 31 1 3 × 10–8 tures 4 h after pollination. The ANOVA test performed on the data obtained for the Columbia background pollen with A, significant P-values (<0.004, Bonferroni’s multiple comparisons adjust- zero [inp1, pollen tube length (mean ± s.d.) = 378 ± 62 µm, ment) are denoted in bold; and B, significant P-values (<0.0125, Bonferroni’s number of pollinated pistils n = 12], three (wt Col, pol- multiple comparisons adjustment) are denoted in bold. len tube length = 416 ± 55 µm, n = 22) or four apertures (lsq6, pollen tube length = 464 ± 75 µm, n = 13) revealed Table 3). Similarly, the GLM test for lines of the Nossen back- a significant effect of aperture number (F = 5.94, d.f. = 2, ground with three (wt No-0) and four to eight (osd1) apertures P-value = 0.0052). The multiple comparisons of means also showed a significant effect of aperture number and age, but by the Tukey contrast test found no significant differ- no interaction between these two parameters (Table 3). ence between the triaperturate wt Col and the other two The pairwise comparisons of the germination percentages for lines (P-value = 0.22 for the wt Col–inp1 comparison and young pollen grains with the Columbia background showed a P-value = 0.08 for the wt Col–lsq6 comparison), but revealed significant difference between the lines with zero, three or four a significant difference between the pollen without apertures apertures (inp1/wt Col/lsq6; Fig. 3; Table 4). The percentage of and the pollen with four apertures (P-value = 0.004 for the germinated pollen grains was the lowest for young pollen with inp1–lsq6 comparison). four apertures and the highest for pollen with three apertures For the wt No-0–osd1 pair, the mean pollen tube growth (Fig. 3). For old pollen grains, no significant difference was rate was not significantly different (t = –1.13, d.f. = 27, found between the pollen grains with zero and three apertures P-value = 0.27) between the line with three apertures (pollen (Fig. 3; Table 4), whereas the germination percentage was still tube length = 475 ± 75 µm, n = 15) and the line with four to the lowest for old pollen with four apertures (lsq6) (Fig. 3). eight apertures (pollen tube length = 440 ± 90 µm, n = 14). In the Nossen background, a similar comparison between wt No-0 (three apertures) and osd1 (four to eight apertures) showed a significantly higher germination percentage in the triaperturate wt No-0 line than in osd1 for both young and old Triaperturate pollen outcompetes inaperturate pollen in mixed pollen grains (Fig. 3, Table 4). pollination assays Although inaperturate pollen is not very commonly found in Table 3. Summary of the GLM results testing the effect of aper- nature, it exhibited a surprisingly strong performance in most ture number (Ap. number), pollen age and aperture number–age assays described above, often behaving similarly to triapertu- interaction on pollen grain germination rate pollen. This prompted us to test the ability of inaperturate pollen to sire seeds when forced to compete in races with tria- Model d.f. Deviance Residual Residual P-value perturate pollen. We pollinated stigmas of Col-0 flowers with d.f. deviance both triaperturate Col-NPTII pollen and inaperturate inp1 pol- wt Col-inp1-lsq6 len and tested the paternity in seeds derived from such mixed Null 67 793 crosses using the inheritance of the NPTII gene, which provides Ap. number 2 159 65 634 <2 × 10–16*** resistance to kanamycin, as a paternity marker. Col-NPTII pol- –11 Ap. number + age 1 42 64 591 8 × 10 *** len exhibits identical behaviour to Col-0, and its progeny pro- Ap. number + age + 2 55 62 537 1 × 10–12*** Ap. number × age duced in competitive pollination assays segregates with Col-0 wt No-0-osd1 progeny at a 1:1 ratio (Table 5; Carlson et al., 2009, 2013). Null 39 187 We therefore considered Col-NPTII pollen to be equivalent to Ap. number 1 14 38 174 0.00023*** wt Col. In two separate trials performed with Col-NPTII/inp1 Ap. number + age 1 21 37 153 5 × 10–6*** Ap. number + age + 1 2.5 36 150 0.11 mixed pollinations, we found that, when forced to compete Ap. number × age with triaperturate pollen, inaperturate pollen sired significantly fewer seeds (Table 5), suggesting that apertures do in fact influ- ***P < 0.001. ence pollen reproductive success. 738 Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis

Young pollen 4 h germination, longevity and pollen tube growth, as well as com- Old pollen 4 h paring the general ability to sire seeds between two otherwise 100 3 ap well-performing types of pollen – triaperturate and inaperturate. 90 0 ap 3 ap 80 4–8 ap 70 Aperture patterns affect pollen germination and longevity

4 ap Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 60 Comparison of several arabidopsis lines differing in aperture 50 numbers revealed the effect of this morphological feature on 40 germination of young pollen grains. For pollen at 30 min post- pollination, the germination of the triaperturate and inaperturate Germination (%) 30 pollen was similar and they both outperformed pollen with higher 20 numbers of apertures. However, 4 h after pollination, the triaper- 10 turate pollen consistently outperformed all other pollen types. Comparison of germination between young and old pollen 0 inp1 wt Col lsq6 wt No–0 osd1 demonstrated that the young pollen grains with almost every aperture pattern exhibited higher germination than the old Columbia Nossen grains, suggesting a general loss of pollen viability over time. Fig. 3. Germination percentages of young and old pollen grains with differ- The reduction in germination effect, however, was much less ent aperture numbers. Percentage germination for young and old pollen of the pronounced in the pollen with more than three apertures, which, Columbia and Nossen backgrounds, 4 h after pollination. The following geno- as young pollen, already exhibited lower germination than other types were compared: inp1 (n = 11 and 14 pollination events; the first number pollen types. Although the young triaperturate pollen germi- indicates the number of pistils pollinated with young pollen and the second one indicates the number of pistils pollinated with old pollen), wt Col (n = 7 and nated better than the young inaperturate pollen, the germination 13) and lsq6 (n = 10 and 14); and wt No-0 (n = 8 and 16) and osd1 (n = 6 and was similar for the old pollen of both genotypes. In turn, old pol- 15). Error bars represent the 95 % confidence interval. len with either three or zero apertures outperformed the old lsq6 and osd1 pollen with more than three apertures. We note that a DISCUSSION potential caveat in the use of the lsq6 and osd1 lines is that their pollen is diploid: therefore, unlike with the inp1 mutant, which Previous studies have led to suggestions that the stasis of tria- produces haploid pollen and has no obvious differences from perturate pollen in eudicots is probably caused by selective the wild type other than the absence of apertures, the increased forces maintaining this pattern (Matamoro-Vidal et al., 2012, number of apertures may not be their only difference from the 2016a). Selection controlling aperture patterns may be sensi- triaperturate wild-type pollen. Yet, at the moment, these lines tive to several pollen traits that could be affected by the num- represent the best available tools for assessing performance of ber and morphology of apertures. Here, we used Arabidopsis pollen with more than three apertures: despite performing two thaliana mutants producing pollen with zero, three, four and a large-scale genetic screens in arabidopsis (Dobritsa et al., 2011; mix of four to eight apertures (d’Erfurth et al., 2009; Dobritsa M. Tan, P. Amom, and A. A. Dobritsa, unpubl. res.), we did not et al., 2011; Dobritsa and Coerper, 2012; Reeder et al., 2016). identify any mutants in which the number of apertures increased A monocolpate mutant would have completed the series, but no without a concomitant rise in pollen ploidy. such mutant has been identified yet. We tested the influence of The effect of pollen aperture pattern on germination was pre- aperture numbers on such important pollen performance traits as viously studied in the heteromorphic species Viola diversifolia and Viola calcarata which, respectively, produce pollen with Table 4. Pairwise comparisons of germination percentages for three and four or four and five apertures within the same anther pollen grains that differed in their number of apertures and age (Dajoz et al., 1991, 1993). In some of these experiments, the germination of young pollen grains was higher in the morphs Comparisons χ2 d.f. P-value with a higher aperture number – a finding inconsistent with our results on the arabidopsis aperture series. In most Viola experi- –25 A lqs6 (4)-wt Col (3): young 108 1 3 × 10 ments, however, no significant difference was found between lqs6 (4)-inp1 (0): young 83 1 8 × 10–20 wt Col (3)-inp1 (0): young 19 1 1 × 10–5 pollen with different aperture numbers (Dajoz et al., 1991, lqs6 (4)-wt Col (3): old 38 1 7 × 10–10 1993; Till-Bottraud et al., 1999). These experiments, however, lqs6 (4)-inp1 (0): old 39 1 4 × 10–10 were conducted primarily in vitro and achieved only a very low wt Col (3)-inp1 (0): old 0.003 1 0.96 germination (5–15 %), which could explain the discrepancy. wt Col (3): young–old 57 1 5 × 10–14 inp1 (0): young–old 23 1 2 × 10–6 Although an in vivo experiment in V. diversifolia was attempted lqs6 (4): young–old 0.0026 1 0.96 (Dajoz et al., 1993), it was performed with a very low number B osd1 (4–8)-wt No-0 (3): 10 1 0.0014 of pollen grains, making the results hard to interpret. young osd1 (4–8)-wt No-0 (3): old 7 1 0.009 wt No-0 (3): young–old 18.6 1 2 × 10–5 osd1 (4–8): young–old 3.4 1 0.066 Aperture patterns seem to have little effect on the in vivo pollen tube growth rate in arabidopsis A, significant P-values (<0.004, Bonferroni’s multiple comparisons adjustment) are denoted in bold; and B, significant P-values (<0.0125, Bonferroni’s Previous in vitro experiments on pollen tube growth in multiple comparisons adjustment) are denoted in bold. V. diversifolia indicated that the pollen tubes of triaperturate Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis 739

Table 5. Results of competitive pollinations: a null hypothesis of may negatively affect pollen viability or increase the amount of equal reproductive ability of different types of pollen and a 1:1 seg- water required for efficient rehydration. regation of alleles from both fathers in progeny was tested In addition to pollen germination and longevity, aperture patterns may also affect the ability of pollen to withstand volume Control: Col-NPTII Col-NPTII vs. changes associated with osmotic stresses or hydration status vs. Col-0 inp1 (Heslop-Harrison, 1979; Edlund et al., 2004). Mathematical modelling of the pollen folding under dehydrating conditions Trial 1 Trial 2 Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 highlighted the importance of apertures for achieving the pre- Pollen 1 Col-NPTII: 48.4 % Col-NPTII: 59.8 % Col-NPTII: 56.3 % dictable and reversible folding (Katifori et al., 2010). Our re- Pollen 2 Col-0: 51.6 % inp1: 40.2 % inp1: 43.7 % Seeds/silique 43.4 ± 2.6 47.3 ± 3.4 45.7 ± 4.2 cent studies on pollen from different species revealed that n 304 378 320 aperture patterns correlate with differences in the ability of pol- χ2 0.33 14.49 5 len to withstand osmotic stress (Matamoro-Vidal et al., 2016b). P-value 0.566 <0.001*** 0.025* In particular, using the same arabidopsis aperture series as in this work, we found that the increase in apertures is associated Seeds per silique are shown as the mean ± s.e. with increased pollen wall breakage, with inaperturate pollen n = total number of seeds from all the crosses for a particular trial. exhibiting the highest resistance to osmotic stress (Prieu et al., *Indicates statistical significance at the P < 0.05 level; ***indicates statis- 2016). tical significance at the P < 0.001 level. Interestingly, inaperturate pollen performed quite well in most of our assays, even though this pattern is fairly uncommon pollen grow longer compared with four-aperturate pollen after in nature (Furness, 2007). In many cases, inaperturate pollen is 2 h of germination, independently of pollen age (Dajoz et al., associated with plants that have functionally female flowers, 1993). Our current experiments in arabidopsis, comparing pol- are adapted to an aquatic lifestyle or have thin exine, any of len tube growth of young pollen after 4 h of on-stigma ger- which may lead to decreased reliance on apertures for pollen mination, did not reveal significant differences between the germination or hydration (Furness, 2007). The surprisingly triaperturate pollen and other pollen morphs. The difference strong performance of the inp1 pollen is likely to be linked between inp1 and lsq6 could potentially be attributed to some to the morphology and/or physiology of arabidopsis pollen delay in initiation of the inp1 pollen germination. The differ- and stigma. A recent study found that, even in the presence of ence between the Viola and arabidopsis results might be due to three apertures, pollen tubes in arabidopsis and in many other the difference between the in vitro and in vivo conditions or due Brassicaceae species exhibit a high frequency of non-apertural to interspecific differences. emergence, often choosing the most direct route to the stigma by breaking through the exine rather than taking an aperture detour (Edlund et al., 2016). The presence of a relatively thin (approx. Both triaperturate and inaperturate pollen grains exhibit increased 1 µm) exine in arabidopsis (Dobritsa et al., 2011; Edlund et al., longevity compared with pollen with higher aperture numbers 2016) may allow non-apertural breakouts and could explain the strong performance of inaperturate pollen in our germination In our experiments, the old triaperturate and inaperturate assays. In addition, it has recently been proposed that arabidop- pollen of Arabidopsis exhibited higher germination than the sis and the other Brassicaceae species capable of non-apertural four- or four- to eight-aperturate pollen, demonstrating that tube emergence may have specific enzymatic mechanisms act- pollen with fewer apertures has increased longevity compared ing at the pollen–stigma interface to facilitate exine degradation with pollen with more apertures. These results are consistent in the areas between apertures (Edlund et al., 2017). with previous findings in V. diversifolia, which showed that the Still, despite the ability of arabidopsis pollen tubes to break ability to germinate was lost faster in the pollen grains with four through the walls, none of the 307 recently surveyed arabidop- apertures than in those with three (Dajoz et al., 1993). sis ecotypes completely lacks apertures (Z. Weber and A. A. Dobritsa, unpubl. res.), arguing against the idea that apertures do not play an adaptive role in this species. Additionally, the Multiple aspects of pollen performance could be affected by results of our competitive crosses between triaperturate and aperture numbers inaperturate pollen indicate that apertures indeed provide selective advantage to arabidopsis pollen and facilitate its repro- Performance variation of pollen grains differing in aper- ductive success. ture number and location on the pollen surface could be due to Taken together, our results suggest that triaperturate pollen several factors which may be influenced by aperture patterns grains might provide the best trade-off among various pollen (Edlund et al., 2004): (1) the initial levels of hydration and fold- performance traits, ensuring strong germination ability, high ing of the pollen grain prior to its contact with the stigma; (2) longevity and a good enough capacity to accommodate volume the speed of water absorption and pollen hydration after pollen changes, thus potentially explaining the prevalence of this mor- arrival on the stigma; (3) the probability of contact between phological trait in the eudicot clade. an aperture and the stigmatic surface; and (4) the size of the pollen surface area that is available for pollen tube emergence. An increase in aperture number would be expected to increase ACKNOWLEDGEMENTS the probability of contact, the speed of water transfer and the area available for germination, but it would also be likely to in- We are grateful to R. Mercier (INRA, Versailles) for provid- crease the level of initial dehydration of the pollen grain, which ing osd1 and wt No-0 seeds. This work was supported by the 740 Albert et al. — Pollen aperture number influences pollen performance in Arabidopsis

Institut Diversité Ecologie et Evolution du Vivant [to B.A. and Halbritter H, Hesse M. 2004. Principal modes of infoldings in tricolp(or)ate A.R.] and by the US National Science Foundation [MCB- Angiosperm pollen. Grana 43: 1–14. Heslop-Harrison J. 1971. The pollen wall: structure and development. 1517511 to A.A.D.]. London: Butterworth. Heslop-Harrison J. 1979. Pollen walls as adaptive systems. Annals of the Missouri Botanical Garden 66: 813–829. LITERATURE CITED Hesse M, Halbritter H, Zetter R, et al. 2009. Pollen terminology: an illus- trated handbook. Wien: Springer-Verlag.

Albert B, Gouyon P-H, Ressayre A. 2009. Microsporogenesis variation in Judd W, Olmstead R. 2004. A survey of tricolpate (eudicot) phylogenetic Downloaded from https://academic.oup.com/aob/article-abstract/121/4/733/4816140 by Valparaiso University user on 05 February 2019 Codiaeum producing inaperturate pollen grain. Comptes Rendus Biologies relationships. American Journal of Botany 91: 1627–1644. 332: 507–516. Katifori E, Alben S, Cerda E, Nelson D, Dumais J. 2010. Foldable struc- Alvarez-Buylla E, Benitez M, Corvera-Poire A, et al. 2010. Flower develop- tures and the natural design of pollen grains. Proceedings of the National ment. The Arabidopsis Book 8: e0127. Academy of Sciences, USA 107: 7635–7639. Bailey I, Nast C. 1943. The comparative morphology of the Winteraceae. Matamoro-Vidal A, Furness C, Gouyon P, Wurdack K, Albert B. 2012. I. Pollen and stamens. Journal of the Arnold Arboretum 24: 340–3046. Evolutionary stasis in Euphorbiaceae pollen: selection and constraints. Carlson A, Telligman M, Swanson R. 2009. Incidence and post-pollination Journal of Evolutionary Biology 25: 1077–1096. mechanisms of nonrandom mating in Arabidopsis thaliana. Sexual Plant Matamoro-Vidal A, Prieu C, Furness C, Albert B, Gouyon P-H. 2016a. Reproduction1 22: 257–262. Evolutionary stasis in pollen morphogenesis due to natural selection. New Carlson A, Gong H, Toomajian C, Swanson R. 2013. Parental genetic dis- Phytologist 209: 376–394 tance and patterns in nonrandom mating and seed yield in predominately Matamoro-Vidal A, Raquin C, Brisset F, et al. 2016b. Links between selfing Arabidopsis thaliana. Plant Reproduction 26: 317–328. morphology and function of the pollen wall: an experimental approach. Dajoz I, Till-Bottraud I, Gouyon P-H. 1991. Evolution of pollen morphology. Botanical Journal of the Linnean Society 180: 478–490. Science 253: 66–68. Murashige T, Skoog F. 1962. A revised medium for rapid growth and bio Dajoz I, Till-Bottraud I, Gouyon P-H. 1993. Pollen aperture polymorph- assays with tobacco tissue cultures. Physiologia Plantarum 15: 473–497. ism and gametophyte performance in Viola diversifolia. Evolution 47: Prieu C, Matamoro-Vidal A, Raquin C, et al. 2016. Aperture number influ- 1080–1093. ences pollen survival in Arabidopsis mutants. American Journal of Botany, Dobritsa A, Coerper D. 2012. The novel plant protein inaperturate pollen1 103: 452–459. marks distinct cellular domains and controls formation of apertures in the R Core Team. 2014. R: a language and environment for statistical comput- Arabidopsis pollen exine. The Plant Cell 24: 4452–4464. ing. R Foundation for Statistical Computing, Vienna, Austria. URL http:// Dobritsa A, Geanconteri A, Shrestha J, et al. 2011. A large-scale genetic www.R-project.org/ screen in Arabidopsis to identify genes involved in pollen exine produc- Reeder S, Lee B, Fox R, Dobritsa A. 2016. A ploidy-sensitive mechanism tion. Plant Physiology 157: 947–970. regulates aperture formation on the Arabidopsis pollen surface and guides Donoghue M, Doyle J. 1989. Phylogenetic studies of seed plants and angio- localization of the aperture factor INP1. PLos Genetics, 12: e1006060 sperms based on morphological characters. In: Fernholm B, Bremer K, doi:10.1371/journal.pgen.1006060. Jornwall H, eds. The hierarchy of life: molecules and morphology in phylo- Till-Bottraud I, Vincent M, Dajoz I, Mignot A. 1999. Pollen aperture hetero- genetic analysis: Proceedings from Nobel Symposium 70. Amsterdam: morphism. Variation in pollen-type proportions along altitudinal transects Elsevier, 181–193. in Viola calcarata. Comptes Rendus Biologies 322: 579–589. Doyle J, Hotton C. 1991. Diversification of early angiosperm pollen in a cla- Till-Bottraud I, Gouyon P-H, Venable D, Godelle B. 2001. The number of distic context. In: Blackmore S, Barnes SH, eds. Pollen and spores: pat- competitors providing pollen on a stigma strongly influences intraspecific terns of diversification: Oxford: Clarendon Press, 169–195. variation in number of pollen apertures. Evolutionary Ecology Research Edlund AF, Swanson R, Preuss D. 2004. Pollen and stigma structure and 3: 231–253. function: the role of diversity in pollination. The Plant Cell 16: S84–S97. Van Campo M. 1976. Patterns of pollen morphological variation within taxa. Edlund AF, Zheng Q, Lowe N, et al. 2016. Pollen from Arabidopsis thaliana In: Ferguson I, Muller M, eds. The evolutionary significance of the exine. and other Brassicaceae are functionally omniaperturate. American Journal London: Academic Press. of Botany, 103: 1006–1019. Vieira A, Feijó J. 2016. Hydrogel control of water uptake by pectins during Edlund AF, Olsen K, Mendoza C, et al. 2017. Pollen wall degradation in the in vitro pollen hydration of Eucalyptus globulus. American Journal of Brassicaceae permits cell emergence after pollination. American Journal Botany 103: 437–51. of Botany, 104: 1266–1273. Walker J. 1974. Aperture evolution in the pollen of primitive angiosperms. Erdtman G. 1952. Pollen morphology and plant taxonomy: angiosperms: an American Journal of Botany 61: 1112–1136. introduction to palynology. Almqvist & Wicksell: Stockholm, Sweden. Walker J, Doyle J. 1975. The bases of angiosperm phylogeny, palynology. d’Erfurth I, Jolivet S, Froger N, Catrice O, Novatchkova M, Mercier R. Annals of the Missouri Botanical Garden 62: 664–723. 2009. Turning meiosis into mitosis. PLoS Biology, 7: 1–10. Williams J, Mazer S. 2016. Pollen – tiny and ephemeral but not forgotten: Fitz Gerald J, Carlson A, Smith E, et al. 2014. New Arabidopsis advanced new ideas on their ecology and evolution. American Journal of Botany intercross recombinant inbred lines reveal female control of nonrandom 103 365–374. mating. Plant Physiology 165: 175–185. Wodehouse R. 1935. Pollen grains: their structure, identification and signifi- Furness C. 2007. Why does some pollen lack apertures? A review of inaperturate cance. New York: Hafner Publishing Co. pollen in eudicots. Botanical Journal of the Linnean Society 155: 29–48. Wortley A, Wang H, Lu L, Li D, Blackmore S. 2015. Evolution of angio- Furness C, Rudall P. 2004. Pollen aperture evolution – a crucial factor for sperm pollen. 1. Introduction. Annals of the Missouri Botanical Garden eudicot success? Trends in Plant Science 9: 154–158. 100: 177–226.