Transcriptomic analysis links gene expression to unilateral pollen-pistil reproductive barriers

Background Unilateral incompatibility (UI) is an asymmetric reproductive barrier that unidirectionally prevents gene flow between species and/or populations. UI is characterized by a compatible interaction between partners in one direction, but in the reciprocal cross fertilization fails, generally due to pollen tube rejection by the pistil. Although UI has long been observed in crosses between different species, the underlying molecular mechanisms are only beginning to be characterized. The wild tomato relative Solanum habrochaites provides a unique study system to investigate the molecular basis of this reproductive barrier, as populations within the species exhibit both interspecific and interpopulation UI. Here we utilized a transcriptomic approach to identify genes in both pollen and pistil tissues that may be key players in UI. Results We confirmed UI at the pollen-pistil level between a self-incompatible population and a self-compatible population of S. habrochaites. A comparison of gene expression between pollinated styles exhibiting the incompatibility response and unpollinated controls revealed only a small number of differentially expressed transcripts. Many more differences in transcript profiles were identified between UI-competent versus UI-compromised reproductive tissues. A number of intriguing candidate genes were highly differentially expressed, including a putative pollen arabinogalactan protein, a stylar Kunitz family protease inhibitor, and a stylar peptide hormone Rapid ALkalinization Factor. Our data also provide transcriptomic evidence that fundamental processes including reactive oxygen species (ROS) signaling are likely key in UI pollen-pistil interactions between both populations and species. Conclusions Gene expression analysis of reproductive tissues allowed us to better understand the molecular basis of interpopulation incompatibility at the level of pollen-pistil interactions. Our transcriptomic analysis highlighted specific genes, including those in ROS signaling pathways that warrant further study in investigations of UI. To our knowledge, this is the first report to identify candidate genes involved in unilateral barriers between populations within a species. Electronic supplementary material The online version of this article (doi:10.1186/s12870-017-1032-4) contains supplementary material, which is available to authorized users.


Background
Reproductive barriers are critical for maintaining species integrity, and their emergence between populations is associated with decreased gene flow and increased genetic divergence, ultimately leading to speciation. In a number of plant families, including the Solanaceae, a unidirectional post-mating prezygotic barrier, termed unilateral incompatibility (UI) occurs at the level of pollen-pistil interactions. UI has most often been studied between genera or species [1][2][3][4][5], but there is also evidence of UI between populations of the same species [1,[6][7][8][9].
The unidirectionality of UI in crosses between species has been linked to plant mating system and specifically to the self-incompatibility response. Self-incompatible (SI) species are obligate outcrossers, whereas self-compatible (SC) species are capable of producing offspring through self-pollination. Gametophytic SI is the most common system of incompatibility in flowering plants [10] and in many families it is dependent on the activity of pistilexpressed S-locus ribonucleases, or S-RNases [11,12]. Distinct from the sporophytic SI of the Brassicaseae in which pollen is rejected at the stigma surface [13,14], in gametophytic SI growing pollen tubes are actively rejected within the style [11,[15][16][17]. This type of S-RNase-based SI is genetically determined by the polymorphic S-locus, which harbors pistil-(S-RNase) and pollen-(S-locus Fbox) expressed factors that are required for the specificity of the SI response [18][19][20][21]. Additional factors have also been implicated in SI, although they do not determine specificity. These include pistil-expressed HT-proteins and 120 K glycoproteins [22][23][24], as well as pollen-expressed components of the E3 ubiquitin ligase complex (Cullin1 and SKP1) [15,[25][26][27][28][29].
UI generally follows the SI x SC rule, wherein pollen of SC species is unable to fertilize SI ovules, but the reciprocal cross is compatible [4]. Because of this link between SI and UI, initial studies suggested pleiotropic effects of the S-locus on interspecific incompatibilities. In fact, more recent studies have shown that the proteins S-RNase, HT-protein, S-locus F-box 23 and Cullin1 all function in both the SI and UI response, demonstrating mechanistic overlap between these two types of reproductive barriers [26,28,[30][31][32][33]. However, it is also clear that genetic factors other than those involved in SI function in UI.
Crosses between SC species or populations (many of which do not express S-RNase) exhibit unexpected incompatibilities [1-4, 7, 31, 34, 35]. This suggests that there are at least two mechanisms underlying UI, one of which is S-RNase based and the other of which is S-RNase independent. Most SC species exhibit full crossability with other SC species, but recently evolved SC populations of SI species exhibit UI [2,4]. This suggests that the mechanistic factors underlying S-RNase independent UI are also present in populations that contain S-RNase. However, the number and type of factors involved in UI remains unclear and most models suggest roles for multiple UI genes of both large and small effect [2,5,8,36].
The existing data on SI and UI in the Solanaceae can be described under a functional architecture in which pollenside resistance factors are required to overcome pistil-side barriers [37]. In this sense, the presence of barriers in the pistil (such as S-RNase and HT) render it competent to reject pollen tubes, whereas resistance factors in pollen (such as SLFs and CUL1) render it competent to overcome pistil-side barriers [37]. As many of the factors involved in UI remain uncharacterized, this architecture of pistil barriers and pollen resistances provides a useful framework in which to view pollen-pistil interactions.
The tomato clade (Solanum section Lycopersicon) offers ample opportunities to further understand and identify the genes involved in UI [38]. This small clade has recently diverged from a common SI ancestor [39,40] over the course of~2.5 million years [41]. The clade harbors six SC species, and seven SI species, three of which contain both SI and SC populations [42]. A number of studies have examined both the physiological aspects [1,5,43,44] and genetic basis [26-28, 30, 31, 33] of UI between members of the clade. In addition, a small number of studies have assessed UI between SI and SC populations within a species [1,[6][7][8][9]44].
The wild tomato Solanum habrochaites is an ideal species in which to study both interspecific and interpopulation UI within the context of recently diverged populations. S. habrochaites has undergone at least two independent transitions from the ancestral SI state to SC at both the northern and southern species range margins [6,7,9,45]. The loss of SI at the northern range margin is correlated with the loss of pistil-side SI factor S-RNase [6,31,46]; whereas southern S. habrochaites SC populations express an S-RNase protein with little or no RNase activity [31]. The genetic structure of S. habrochaites is consistent with the spread of populations both north and south from a central origin, with central SI populations showing the highest levels of diversity [45,47]. Interestingly, the types and strengths of reproductive barriers displayed by a number of populations also follow this same latitudinal axis of divergence [6,45].
Previous studies have demonstrated variation in the strength of both intra-and interspecific UI in different S. habrochaites populations [1, 6-9, 31, 35]. For instance, a central S. habrochaites SI population (LA1777) rejects pollen tubes from a subgroup of northern S. habrochaites SC populations, including LA0407; whereas LA0407 and other SC populations accept LA1777 pollen tubes [1,6].
In interspecific interactions, when either SC S. lycopersicum or SC S. neorickii (as female) is crossed with S. habrochaites (as male), crosses result in fruit [48,49]. In the reciprocal cross, pistils of SI and southern SC populations of S. habrochaites rapidly reject interspecific SC pollen after a few millimeters of growth through the style, demonstrating a classic UI response [1,6,31]. However, interspecific pollen tube rejection is delayed in many northern SC populations of S. habrochaites, [1,6,31], and one unique SC population (LA1223) is unable to reject both S. lycopersicum and S. neorickii pollen tubes [6]. LA1223 is the only known S. habrochaites population that does not accumulate the pistil HT-protein [6,31], which may explain this extraordinary lack of UI at the pollen-pistil level.
Although elegant mapping, biochemical and transgenic studies have revealed that SI factors are important players in UI [12,23,26,28,30,31,33,50], the full suite of molecular factors involved in the complex UI response are only beginning to be characterized. The evidence for both interspecific and interpopulation UI in S. habrochaites suggests that factors other than those involved in SI are involved in UI. This is consistent with transgenic studies in tomato showing that together S-RNase and HT-protein can act to reject pollen from only a subset SC species, and even then pollen tube rejection is delayed [33].
Transcriptome profiling can provide a broad view of the factors involved in pollen-pistil interactions and represents a powerful tool for candidate gene identification. Studies in Arabidopsis have identified large numbers of genes that are specific to pollen or pistil, as well as those that are specifically induced upon the interaction between the two in compatible pollinations [51][52][53][54]. However, very few studies have investigated pollen-pistil transcriptomes in response to UI. In a recent study, Pease et al. [55], identified a number of interspecific UI candidate genes by comparing stylar transcriptomes of UI-competent S. pennellii and UI-deficient S. lycopersicum, two species which likely diverged over 2 mya [41]. Their results identified a large number of loci that varied in expression between styles of these two species, and highlighted the importance of HT-protein in UI [55].
Here, we used a similar transcriptomics approach to broadly characterize the molecular factors involved in UI in S. habrochaites at the level of pollen-pistil interactions. The populations selected for study have recently diverged (< 0.8 mya, [41]), and exhibit differences in both interspecific and interpopulation reproductive barriers. To our knowledge, this is the first report using transcriptomics to identify novel candidate genes associated with reproductive barriers between diverging populations within a species.

Pollen-pistil interactions in SI LA1777 and SC LA0407
Using pollen tube growth assays, we confirmed that all LA1777 (SI) individuals used in RNA-seq experiments rejected both self-pollen tubes and those of LA0407, but accepted pollen from other LA1777 individuals (intrapopulation). Further, we confirmed that each LA0407 (SC) individual accepted pollen tubes from LA1777, self, and intrapopulation crosses.
In compatible crosses, pollen tubes generally reach the ovary within 24 h [1,31]. However, incompatible crosses may differ in the timing of pollen tube rejection [1]. To identify the post-pollination time-point at which pollen tube rejection occurs in the incompatible cross between LA1777 (female) and LA0407 (male), pistils were pollinated and harvested over a time course. Pistils of LA1777 were actively rejecting pollen of LA0407 at 12 h post-pollination, and full rejection occurred by 48 h (Fig.  1). Although important changes in gene expression may occur immediately after pollination and at early timepoints in pollen tube rejection, we chose to harvest stylar tissue 16 h post-pollination for subsequent RNA-seq experiments in an attempt to capture changes in gene expression corresponding to active pollen tube rejection in the style. At this point, incompatible pollen tubes will have traversed~20% of the style length ( Fig. 1), whereas compatible pollen tubes will have traversed~60% of style length (Fig. 1, [31]).

Differential gene expression between UI-competent and compromised populations
We sampled both pollen and stylar tissue to better characterize differences in transcriptomes between UIcompetent LA1777 and UI-compromised LA0407 (Fig. 2). Styles were either unpollinated (UP) or subject to three different treatments including self, intrapopulation and interpopulation pollination; whereas pollen was subjected to a single treatment (ungerminated) (Fig. 2). We considered pistils of LA1777 to be "UI-competent" because they contain functional barriers that allow for the rejection of self, interpopulation and interspecific pollen tubes [1,6,31]. Alternatively, we considered pistils of LA0407 to be "UIcompromised" because a loss of barriers has resulted in the inability to reject self, interpopulation and some types of interspecific pollen tubes [1,6,31]. In the case of pollen, we also considered LA1777 to be "UI-competent" because it harbors pollen-side resistance factors that render it competent to overcome pistil-side barriers, whereas LA0407 pollen is considered "UI-compromised" because it is rejected by all SI pistils [1,6,31].
We analyzed a total of 33,055 genes, out of which 7358 (22.3%) were specific to pollen (that is, they were upregulated in pollen with respect to all styles), and 4793 (14.5%) were specific to styles. Additionally, we found 1492 genes that were highly upregulated only in tissues of LA1777 and 1382 genes highly upregulated only in LA0407 (4.5% and 4.2% of genes analyzed, respectively).

Pollination type significantly influences a small number of genes
We expected that stylar transcriptomes would exhibit differences due to the type of pollination and whether or not a pollination was compatible. We performed three separate pairwise comparisons to look at differential gene expression in UP LA1777 styles compared to different pollination types (intrapopulation, self and interpopulation). Pairwise comparisons between pollinated and UP LA1777 styles revealed only a small number of differentially expressed genes, all of which were expressed at very low levels (< 0.3 counts per million (cpm); Additional file 1: Table S1 (self vs UP), Additional file 1: Table S2 (intrapopulation vs UP) and Additional file 1: Table S3 (interpopulation vs UP)). In all pollination treatments we identified pollinationinduced differential expression of a number of hypothetical proteins, RNA-directed DNA Polymerases and retrovirus-related Reverse Transcriptases. However, only three genes showed similar regulation among all pollination treatments: a hypothetical protein (Sopen0 5g010690) and two genes that have high homology to a Copia-like retrotransposon (Sopen01g019140 and Sop en02g008750).
Five genes were similarly differentially expressed in incompatible crosses with LA1777 as female (self and interpopulation crosses; Additional file 1: Table S1 and  Table S3) versus compatible crosses (intrapopulation; Additional file 1: Table S2), three of which encoded hypothetical proteins. In addition, a cystathionine beta synthase (CBS) domain-containing protein (Sopen06g019460) was downregulated 3-fold in incompatible crosses. It has been proposed that CBS proteins are redox regulators involved in the modification of cell wall composition, and in Arabidopsis, changes in CBS expression can reduce self-fertility [56]. A homolog of Defective meristem silencing 3 (DMS3, Sopen03g019410), a gene that is involved in silencing and epigenetic modification [57], was also downregulated 3fold in incompatible crosses.

Differential gene expression related to UI-competence in styles
Although a number of potentially interesting candidates involved in UI were identified from our pairwise comparisons, they were expressed at relatively low levels, often showed only low fold-changes, and had relatively high pvalues (p < 0.05). A principal components analysis (PCA) of all stylar treatments showed that most of the variation between samples could be explained by source population, and that there was no consistent grouping of styles due to pollination treatment (Additional file 2: Figure S1). The fact that few significant differences were detected between UP styles and pollination treatments may reflect results of previous studies indicating that the genes involved in UIcompetence are expressed in styles regardless of pollination status [31,38,55]. In other words, the UI-competent styles appear 'primed' to reject incompatible pollen, as large changes in gene expression are not seen between UP styles and those pollinated with compatible versus incompatible pollen parents. Because of these results, we also performed an analysis in which all stylar treatments within a population (UP, pollinated with self, intrapopulation and interpopulation pollen) were pooled to capture differential expression between UI-competent (LA1777) versus UIcompromised (LA0407) styles.
We identified 179 genes that were significantly upregulated in UI-competent versus compromised styles, and 179 that were significantly downregulated (Additional file 1: Table S4 and Table S5). However, we focused our interest predominantly on genes that showed a mean expression level over 3 normalized cpm and were upregulated over 10-fold in UI-competent versus compromised styles. The top 25 genes showing the largest upregulation in UIcompetent styles are shown in Table 1.
Within the top 25 genes showing the highest upregulation in UI-competent versus compromised styles, we found ten that are involved in oxidation-reduction reactions ( Table 1). These included three Cytochrome P450 genes (Sopen02g037430, Sopen06g021780, Sopen07g028940), an Alkenal Reductase (Sopen12g006330), a Glutathione S-Transferase (Sopen12g026980), an NAD(P)H-Dehydrogenase (Sopen05g003640), and an NAD(P)-Oxidoreductase (Sopen12g021490). In addition, we identified two genes involved in early steps of flavonoid biosynthesis: Chalcone Synthase 2 (Sopen05g032070) and Chalcone-Flavanone Isomerase (Sopen05g030780). Finally, we discovered that the gene showing the highest upregulation in UIcompetent styles (hypothetical protein, Sopen12g0 17530) contains a heavy metal-associated domain (Additional file 2: Figure S2), suggesting that it also may be involved in oxidation-reduction reactions.
In UI-competent styles, we also identified three upregulated genes that are putatively involved in defense responses. These included an Endo-Chitinase involved in jasmonic acid and ethylene signaling (Sopen02g027710) that was highly expressed (182 cpm) and upregulated over 20-fold, a Beta-Glucosidase (Sopen11g004500) and a Disease Resistance Protein (Sopen10g025030) ( Table 1). These genes are of interest, as many molecular components involved in plant-pathogen interactions show significant overlap with those involved in pollen tube growth and guidance [65].
One gene of particular interest that was highly upregulated in UI-competent styles is the hypothetical protein Sopen02g033850 (Table 1), which shows 80% amino acid identity to the peptide hormone Rapid ALkalinization Factor (RALF) from the wild potato, S. chacoense. Small secreted peptides including RALFs are involved in a wide variety of plant functions including development and immunity [66], and a pollen-specific RALF from S. lycopersicum (SlPRALF) was found to negatively regulate pollen tube elongation in vitro [67]. We also identified a protein involved in oligo-peptide transport (Sopen05g001950) which was upregulated >28-fold in UI-competent styles and may be involved in the transport/secretion of peptide hormones such as RALFs.
Another gene of interest that was highly upregulated in UI-competent versus compromised styles was a Kunitz family protease inhibitor (Table 1). In Nicotiana, the Kunitz Family Protease Inhibitor, NaStep, is highly expressed in the pistils of SI species and is thought to stabilize HT-proteins, although the mechanistic basis of this interaction remains unknown [68]. The NaStep protein is taken up by both compatible and incompatible pollen tubes, and the transgenic suppression of NaStep in SI Nicotiana species compromises rejection of both self-and some types of interspecific pollen tubes [68]. The reduced expression of this gene in UI-compromised LA0407 may reflect this population's lack of ability to reject some types of interspecific pollen tubes [1,6].
Although we were most interested in genes showing the highest upregulation in UI-competent styles, we also analyzed genes that are highly downregulated. We found that three of the top 25 most highly downregulated genes in UI-competent styles were involved in oxidation-reduction reactions, and one was putatively involved in defense response (Additional file 1: Table S5). Notably, five of the top 25 genes downregulated in UI-competent styles are putatively involved in cell wall modification (Additional file 1: Table S5). These include two Pectin Lyases (Sopen12g009500 and Sopen01g038750), a Glucosyltransferase (Sopen00g008620) and two Glycosyl Hydrolases (Sopen04g034210 and Sopen07g001240), all of which were downregulated over 13-fold. Eight additional genes involved in cell wall modification were downregulated in UI-competent styles, although to lesser extents (Additional file 1: Table S5).
In addition to our analysis of genes that are highly upor downregulated in UI-competent versus compromised styles, we investigated the expression of 33 a priori candidates (Additional file 1: Table S6) based on reports in the literature of stylar genes that may be involved in UI [26,28,31,55]. These candidates included genes that were identified in a recent study comparing stylar transcriptomes between UI-competent S. pennellii LA0716 and the cultivated tomato S. lycopersicum M82 which shows no UI response [55]. Only three of the 33 a priori candidates showed over a 2-fold increase in styles of UIcompetent LA1777 versus UI-compromised LA0407 (Additional file 1: Table S6). These included a Pectin-Methylesterase Inhibitor (Sopen04g027820, upregulated 2.4-fold) and two Glucosyltransferases (Sopen08g002330 and Sopen08g002350, both upregulated over 8-fold); however all were expressed at low levels, < 1 cpm. HTprotein, a small asparagine-rich protein required for S-RNase-based UI [33] was highly expressed in styles from both populations and was downregulated slightly in LA1777 (HT-A, 1.3-fold). Interestingly HT-B transcript was expressed in both populations, although neither population accumulates functional protein due to an early stop codon in the transcript [31]. Each individual of LA1777 harbored two unique S-RNase alleles as expected (data not shown), whereas LA0407 did not express S-RNase transcript, as has been documented in previous studies [31].

Differential gene expression related to UI-competence in pollen
As coordinated interactions between pollen and pistil are required for successful fertilization, we also compared pollen transcriptomes between populations. Pollen tubes of LA1777 reach ovaries of all SI and SC species within the tomato clade [1]. However, LA0407 pollen tubes are rejected by all SI species, including by SI S. habrochaites LA1777 [1]. The inability of LA0407 pollen to traverse the styles of LA1777 and other SI Solanum species suggests that it has lost a pollen factor(s) required for S-RNase resistance. For our analysis, we therefore considered LA1777 pollen to be UI-competent, and LA0407 pollen to be UI-compromised. We used dry mature pollen in our RNA-seq analysis because genes previously identified in UI in tomato (CUL1 and SLF23) have high levels of gene expression in dry/ungerminated pollen [26,28]. However, it is acknowledged that some pollen-specific genes important in UI may require induction through hydration or interaction with pistil tissues, and these would not have been captured in our analyses.
We identified 90 genes that were upregulated in UIcompetent pollen (Additional file 1: Table S7) and 99 that were downregulated (Additional file 1: Table S8). Ten of the top 25 genes upregulated in UI-competent versus UIcompromised pollen were annotated as hypothetical proteins (Table 2; Additional file 1: Table S7). Upon further analysis, the hypothetical gene with the highest upregulation in UI-competent pollen (Sopen12g014190) likely encodes an Arabinogalactan Protein (AGP) (Additional file 2: Figure S2). In Arabidopsis, pollen AGPs are required for proper pollen tube development and growth [69][70][71], and pistil AGPs are known to stimulate pollen tube growth [70][71][72][73]. We also identified a Rab-GTPase (Rab4A, Sopen01g033860) that was upregulated nearly 200-fold in UI-competent pollen. Pollen specific proteins from this family have been found to promote pollen tube tip growth and play a role in the ability of the pollen tube to sense directional cues [74,75]. Our analysis of pollen also identified differentially expressed genes that may be involved in protein degradation pathways (Table 2; Additional file 1: Table S7 and  Table S8). These genes are of interest, as components of a pollen SCF (Skp-Cullin-F-box) E3 ubiquitin ligase complex have been implicated in both SI and UI, and are potentially involved in detoxifying pistil side factors such as S-RNases through the proteasomal degradation pathway [11,15,21,26,27,29,76]. We found two F-box/Skp2 -like genes (Sopen05g003280 and Sopen12g022970) that were upregulated over 150-fold in UI-competent pollen ( Table 2). Another gene involved in protein degradation, an aspartyl protease (Sopen01g032940) was upregulated 189-fold in UI-competent pollen ( Table 2). No previously identified SLFs [28] were significantly differentially regulated, nor was the pollen UI factor Cullin1 [26] (Additional file 1: Table S9).
We identified two protein kinases that were upregulated over 50-fold in UI-competent pollen, one of which encodes a calcium binding serine/threonine wall-associated kinase (Sopen10g028190), and the other a cysteine-rich receptor like kinase (RLK) (Sopen02g013470) ( Table 2). Because RLKs have a proven role in pollen tube growth [77], these genes are of potential interest. Further, pollenexpressed RLKs are involved in reception of peptide and hormone signals from the pistil [78][79][80].
Genes involved in transcriptional regulation are likely to be important components of pollen tube growth. We identified two types of transcription factors known to play key roles in stress response [81,82] that were highly upregulated in UI-competent pollen: a NAC-domain containing protein (Sopen03g020850) and a bZIP family protein (Sopen12g006170) that has previously been localized to pollen (Table 2; Additional file 1: Table S7).
An analysis of genes that were highly downregulated in UI-competent pollen identified an F-box Protein (Sopen11g004020) and a Cysteine-rich RLK (Sopen05g 014070) that were downregulated over 45-and 100-fold respectively (Additional file 1: Table S8). Genes showing over 100-fold decreases in UI-competent pollen also included two Histone Deacetylases (HDACs, Sopen11g 004040 and Sopen11g004050; Additional file 1: Table  S8). The proper function of HDACs has been linked to successful pollen tube germination and tip growth in Picea willsoni [83].

UI competence in additional S. habrochaites populations
Our initial analyses comparing transcriptomes of LA1777 and LA0407 reproductive tissues led to some intriguing candidates that might be integral to UI. In an attempt to further narrow down candidate gene and to expand our analysis to a broader range of S. habrochaites populations, we performed a second RNA-seq experiment that included UP styles and pollen from S. habrochaites populations with selected phenotypes (Table 3; Additional file 1: Table S10).
First we confirmed that the UI phenotypes of the individuals tested reflected previous results [6] (phenotypes summarized in Table 3). A PCA of all samples shows that much of the sample variation (1st PC) can be explained by tissue type, whereas a smaller percentage of variation (2nd PC) is explained by source population and potentially mating system (Fig. 3). Interestingly, the additional populations selected for transcriptome analysis, which lie between LA1777 and LA0407 geographically, cluster between these two populations in the PCA.
For each gene, we trained a linear discriminant function for UI-competence on the expression values for LA1777 and LA0407, and then classified gene expression in other populations as UI-competent or UI-compromised. We took into account the UI patterns of these populations with LA0407, as well as previously reported results on  [6,45] c interpopulation and interspecific UI as reported in Broz et al., [6] d pollen of the LA1223 individual used in this experiment was not accepted by LA1777, Broz et al., [6] reports variation in individual LA1223 phenotypes for this cross interspecific UI ( [6], summarized in Table 3). Because styles of LA2119 are unique in that they accept interpopulation LA0407 pollen while rejecting interspecific pollen, we interpret the information in Table 4 to reflect the variation in UI phenotype of LA2119 styles: i.e., genes that are up in LA2119 are top candidates in interspecific UI, whereas genes that are down in LA2119 are top candidates in interpopulation UI.
The top candidates for interspecific UI-competence in styles identified in the linear discriminant analysis (LDA) included the hypothetical protein Sopen01g011020, that is highly expressed and upregulated >100-fold (Table 4). Six genes potentially involved in oxidation-reduction reactions were identified including the NAD(P)-Linked Oxidoreductase (Sopen12g021490), a Cytochrome P450 (Sopen07g030750), a Lipid Oxidoreductase (Sopen06g 026260), a Heavy Metal Transport Protein (Sopen0 5g028380), a Peroxisomal Membrane Protein (Sopen03 g005470), and the Photosystem I Reaction Center (a Ferredoxin Oxidoreductase, Sopen12g006800).
The LDA of pollen genes was more straightforward in that LA2119 and LA1264 were expected to be UIcompetent (i.e., similar to LA1777) whereas LA1223 is UI-compromised and may be missing pollen factor(s) required to traverse SI styles. As shown in Table 5, using the LDA we identified 22 genes that were upregulated in UI-competent pollen. Two genes, upregulated >25-fold, encode F-box proteins that are both putatively involved in the ubiquitin ligase complex (Sopen03g040880 and Sopen01g027170). In addition, a RAPTOR/KOG kinase homolog that is associated with the CUL4 RING ubiquitin ligase complex (Sopen10g027930) was upregulated 12-fold in UI-competent pollen, and an annexin-like calcium binding protein (Sopen05g030530) putatively involved in calcium signaling and polysaccharide transport was upregulated over 4-fold. Two transcription factors were also identified in the analysis: Sopen12g027920 encodes a suppressor of FRI1 that may act to recruit Histone H3 Methyltransferases and Sopen11g020250 encodes a Zinc Finger Transcription Factor.

Discussion
Although UI is widespread in plant families, the underlying molecular basis of this unidirectional reproductive barrier is not well understood. Here, using a transcriptomic approach, we identified genes in both pollen and stylar tissues that represent strong candidates for involvement in UI. Overall, our analyses identified a large number of differentially expressed genes that are involved in oxidation-reduction reactions and ROS signaling. Oxidative stress response is involved in a variety of plant physiological processes, including reproduction [58,60,[86][87][88]. The production of ROS in plants generally results in one of two outcomes: adaptation to stress or programmed cell death [86]. The balanced interplay between pollen and pistil ROS production can display either of these results: signaling and detoxification are required for successful fertilization (adaptation to stress) and can also function in the incompatible (SI) response (cell death of incompatible pollen tubes). For example, in the Papaveraceae, ROS are recognized as key regulators of programmed cell death in incompatible (self) pollen tubes [64,89] and recent studies of Rosaceae family member Pyrus pyrifolia have demonstrated a link between ROS accumulation, Ca 2+ signaling, calmodulin levels and actin filament depolymerization during self-pollen tube rejection [63,90]. One of the first indications that ROS might be involved in pollen tube growth and crosscompatibilities in the Solanaceae came from histochemical staining in styles of Petunia hybrida, which demonstrated that peroxidase activity is found in unpollinated styles and decreases following compatible pollinations, but remains high during incompatible self-pollinations [91]. This pattern is also observed in an analysis of cytochrome P450 (CYP51G1-Sc) in the wild potato S. chacoense, where in compatible pollinations, mRNA levels of CYP51G1-Sc declines, but in incompatible (self) pollinations, levels of this cytochrome remain stable.
In our pairwise comparisons investigating changes between UP and pollinated styles, we found increases in ROS pathway members in incompatible interpopulation pollinations (Additional file 1: Table S3) but not incompatible self- A linear discriminant function was trained on the expression values of LA1777 and LA0407, and then used to classify other accessions as UI or non-UI pollinations (Additional file 1: Table S1) or compatible intrapopulation pollinations (Additional file 1: Table S2). For instance, an H 2 O 2 transporter was upregulated over 16-fold only in incompatible interpopulation pollinations. These transporters generally pump reactive H 2 O 2 into the apoplast, an acidic environment with low numbers of ROS scavengers, resulting in oxidative stress [86]. A K + channel and a Ca 2+ -Binding Annexin Protein, both of which were annotated as being pollen-expressed were also upregulated in interpopulation interactions, as was a Rab-GTPase. All three of these proteins can play important roles in ROS generation and signaling. In pollen tubes, annexins may provide an important link between Ca 2+ , the membrane and the cytoskeleton [92]. Further, many annexins form Ca 2+ channels which are vital not only for the oscillating Ca 2+ influx associated with pollen tube tip growth, but also for facilitating ROS signaling, cell elongation and cell wall remodeling [86,93]. Interestingly, SKOR K + channels like the one identified in our analysis can also act as ROS-activated Ca 2+ channels [93]. Small GTPases, including Rabs, increase NADP(H)-oxidase activity in a Ca 2+ -dependent manner [59,94], and their proper function is required for pollen tube growth [59,61,74,75,95]. In Nicotiana, the overexpression of both active and mutant forms of the RAB11 protein leads to the inhibition of pollen tube growth [75] suggesting that the correct balance of multiple Rabs is required for effective pollen tube growth. In UI-competent styles, ten of the top 25 most highly upregulated genes are putatively involved in ROS generation and/or signaling, including an NADP(H)-oxidase and multiple cytochrome P450s (Table 1). We also identified two highly upregulated gene candidates in the flavonoid pathway, which could produce flavonoid compounds to act as pro-or anti-oxidants (Table 1). In our subsequent analysis using transcriptome data from additional S. habrochaites populations, we found that six of the nine candidate stylar genes that may be involved in interspecific UI were ROS pathway genes (Table 4). Surprisingly, only two ROS-linked genes were upregulated in interpopulation UI-competent styles, one of which encodes a DELLA-like transcription factor that indirectly inhibits ROS accumulation and restrains cell expansion [84,85]. In sum, these results suggest that ROS production and signaling between pollen and pistil must be held in a tight balance for pollen tubes to successfully grow through styles to reach the ovary. The generation of ROS is linked to cell expansion, growth, cell wall cross linking and callose deposition [96], and our analysis identified genes putative involved in cell wall modification. Pollen tube walls consist of a number of polymers (callose, cellulose and pectin) that are highly cross-linked to each other; however the mechanisms by which cell wall modification is regulated remains unclear [97][98][99]. Studies using microarray analysis have found an upregulation of cell-wall modification genes in pollinated versus unpollinated styles [52]. However, few have specifically investigated specificity to the SI or UI response (but see Pease et al. [55]). Using electron microscopy, de Nettancourt et al. [100] found differences in callose deposition between SI and UI crosses in S. peruvianum wherein SI crosses showed high levels of callose deposition at the pollen tube tips, but interspecific crosses did not [100]. Our pairwise comparisons in UP versus pollinated LA1777 styles did not reveal any genes involved in cell wall modification. However, we did identify a number of differentially expressed genes involved in cell wall modification in our larger analysis of stylar tissue (Additional file l: Table S4 and Table S5), most of which were highly downregulated in UI-competent styles.
One of the most intriguing gene candidates from our analysis of stylar tissue included a putative RALF peptide hormone (Sopen02g033850) that was upregulated over 22fold in UI-competent styles. Peptide hormone signaling is involved in numerous processes during pollen-pistil interactions, from pollen hydration to fertilization [78]. A tomato RALF has been shown to reduce pollen tube growth during specific windows of development [67], and therefore a stylar-secreted RALF could play a role in the rejection of interspecific pollen tubes. Another interesting style-expressed candidate to pursue is the Kunitz-like Protease Inhibitor that was upregulated in UI-competent styles. The Nicotiana NaStep protein from this family is required for the rejection of some (but not all) interspecific pollen [68], and this protease inhibitor may play a similar role in S. habrochaites UI. Finally, the most highlyupregulated stylar gene identified in UI-competent styles encodes a putative prenylated heavy-metal binding protein (Additional file 2: Figure S2), that may be worthy of further investigation. Other proteins of this type have been identified in a variety of tissues, and the few that have been characterized have been implicated in stress responses [101].
In UI-competent pollen, we identified a highly upregulated putative AGP (Sopen12g014190) ( Table 2) that contains multiple defining characteristics of an AGP (Additional file 2: Figure S2; [71,102]). This protein contains a signal peptide, a hydrophobic C-terminal domain, eight prolinecontaining dipeptides and is composed of >35% Pro/Ala/ Ser/Thr (PAST) amino acids (Additional file 2: Figure S2). In Arabidopsis, pollen AGPs are required for pollen tube growth, and may be involved in complex signaling cascades [69][70][71], and pollen AGPs have been localized to pollen tube tips in some species [103]. Another gene that warrants further study is a Rab-GTPase (Sopen01g033860) that was upregulated nearly 200-fold in UI-competent pollen. It will be interesting to see if increasing the expression of these candidate genes in UI-compromised pollen is able to increase rates of pollen tube growth through SI styles.

Conclusions
Our analyses revealed differentially expressed genes that may contribute to reproductive incompatibility between populations of S. habrochaites. This work represents an important first step in understanding how unilateral barriers might arise between populations, and how they are maintained during speciation. The variability in UI responses between S. habrochaites populations provides an exciting opportunity to further analyze these candidate genes and link them to specific UI phenotypes.

Solanum habrochaites plant material and growth
Solanum habrochaites (S. Knapp & D. M. Spooner) is a wild relative of tomato that demonstrates variability in mating system [42,45], as well as interspecific [31,35] and interpopulation [1,[6][7][8] cross-compatibilities. Seeds from the S. habrochaites accessions (referred to hereafter as populations) used in this study (Table 3) were acquired from the C.M. Rick Tomato Genetic Resource Center (TGRC) at University of California, Davis (http://tgrc.ucdavis.edu, [104]), sterilized according to recommendations from the TGRC, grown under greenhouse conditions as previously described [1] for approximately 3 weeks, and transplanted in covered outdoor agricultural field plots at Colorado State University. All agronomic experiments were conducted in accordance with local legislation and no permitting was required. Plants used in the study were randomized within a single block, and remained large and healthy throughout the experiment. Specific populations were chosen based on reproductive characteristics as described in Broz et al., [6], see Table 3 for more information.

Pollen tube growth phenotypes
Pollen tube growth through the style was assessed for self, intrapopulation and reciprocal interpopulation crosses of S. habrochaites individuals, although for LA0407, self and intrapopulation crosses consistently resulted in fruit-set, so pollen tube growth was not assessed for every individual. For all crosses, buds were emasculated 1 day prior to anthesis (day −1), hand-pollinated 24 h later (day 0, budbreak), and harvested into fixative (1:3 acetoethanol) 48 h after pollination. For crosses between LA1777 (female) and LA0407 (male), styles were harvested at various time points after pollination (12, 24 and 48 h) to determine the time at which pollen tube rejection occurred. Pollinations were typically performed in the late afternoon and collected the following morning. However, pollen tube growth through styles was similar for pollinations performed in both the morning and the afternoon (data not shown).
Pollinations were covered with mesh bags to prevent unintended pollen deposition by pollinators. Pollen tube growth was assessed using fluorescence microscopy as previously described [31], and the length of styles and the point in the style at which no more than three pollen tubes passed were measured using ImageJ 1.47v (https://imagej.nih.gov/ij/; [105]).

Tissue collection, RNA extraction and library construction
The primary RNA-seq experiment was performed to identify genes involved in interpopulation pollen tube rejection observed in crosses between LA1777 females and LA0407 males. Samples consisted of unpollinated styles, pollinated styles (self-pollination, intrapopulation pollination, or interpopulation pollination), and ungerminated pollen ( Fig. 1; Additional file 1: Table S10)resulting in five treatment/tissue types for each individual. Three individuals from each population were used as biological replicates, resulting in a total of 30 libraries. In a follow-up experiment designed to narrow the list of genes involved in interpopulation interactions, RNA samples from between one and three individuals were pooled at an equimolar ratio before library creation (Additional file 1: Table S10).
For all style samples, flowers were emasculated and pollinated as described above for pollen tube growth experiments and harvested 16 h post-pollination (the approximate time at which pollen tube rejection occurred in interpopulation crosses). Unpollinated controls underwent the same treatment (emasculation at day −1), except they were left unpollinated. For each individual plant, approximately 30 styles were collected for each treatment and pooled before RNA extraction. To minimize variation due to environmental conditions, all treatments were conducted on the same days and harvested at approximately the same time of day. Styles (including stigmas) were harvested directly into RNALater (Qiagen) and stored at 4°C for 1 week, after which styles were blotted dry and immediately frozen at −80°C until processing. Approximately 100 mg of dry pollen was harvested from each individual plant and immediately frozen at −80°C. Ungerminated pollen was used for RNA extraction.
Tissues were ground using the Tissue-lyser (Qiagen), RNA was extracted using the Qiagen RNeasy Plant mini-kit and brought to a final concentration of 70-200 ng/uL. A subset of RNA samples was checked for quality by agarose gel electrophoresis and visualized by ethidium bromide staining. Sample quality was further evaluated using the Agilent 2200 RNA TapeStation system before library creation. Stranded, paired-end libraries of total RNA were generated for each sample using Illumina Truseq Stranded mRNA sample preparation kits. Libraries were pooled and distributed evenly across two lanes of Illumina HiSeq™ 2000 (Illumina Inc., San Diego, CA, USA). RNA quality control, library preparation, and pooling were performed by the Indiana University Center for Genomics and Bioinformatics. The raw transcriptome data is available on the NCBI SRA database (BioProject PRJNA310635 at https:// www.ncbi.nlm.nih.gov/bioproject/310635).

RNA-seq read processing and mapping
Prior to mapping and assembly, reads were trimmed and filtered using the SHEAR program (http://www.github.com/jbpease/shear; [55]). Briefly, SHEAR first uses the Scythe algorithm (https://github.com/vsbuffalo/scythe; [106]) to remove adapters from the 3′ end, and then filters low quality reads (mean Q < 10), reads with >7 ambiguous bases (N's), reads <50 bp, and repetitive reads with mutual information score > 0.5. The program then trims reads on both ends by removing low quality bases (Q < 20), poly-A or poly-T runs of n ≥ 12, and ambiguous bases. The appearance of AGATC at the 3′ end was also removed as we presumed it was an adapter fragment. We removed both reads in a pair if either one of them failed the filters. On average, 2.9% of all reads failed to pass the filter (min 2.5%, max 3.4%). The full command and parameters used for SHEAR can be found in Additional file 2: Method S1.

Differential gene expression analysis
For all tests of differential expression we used linear models implemented by the limma package [111,112] and modules from the edgeR package [113] in R [114]. First, we normalized and transformed the raw reads with voom, a weighted transformation based on the expected relationship between expression mean and variance [112]. We computed t-statistics on the transformed expression values for each gene using an empirical Bayes adjustment of standard errors with the eBayes function [111].
We initially searched for differential expression among stylar tissues by carrying out separate pairwise comparisons. Specifically, we compared the following pairs of style treatments in LA1777: intrapopulation-pollinated (compatible) against unpollinated (UP) styles, self-pollinated (incompatible) against UP styles and interpopulation (incompatible) against UP styles.
We visualized the genome-wide patterns of expression through a PCA of the normalized mean read counts per gene (cpm reads in the library) with the prcomp function (implemented in R; [114]). The PCA showed that there were negligible differences at the genome-wide scale among all style treatments within a population ( Fig. 3; Additional file 2: Figure S1). Because of these high consistencies in their gene expression profiles, all style treatments (UP, self-, intra-and interpopulation pollinated) within a population were considered identical in our linear models that focus on the differences between UI-competent (LA1777) and UI-compromised (LA0407) styles.
We identified genes that are differentially expressed in UI-competent tissues using a linear model with a single fixed effect (collinear with the population of origin) for pollen (n = 6) and styles (n = 24) separately. From these models, we took genes as differentially expressed if they showed large differences in expression (> 3-fold change), with statistical significance at a false discovery rate (FDR) of 5%. Further, we considered only tissue-specific genes: we required genes to have a significant (FDR < 5%) tissue effect in a general linear model Y~P + T + e (where P is a population effect, T is a tissue effect, and e is the error term). For stylar-side factors, we included genes as differentially expressed if they were upregulated in styles with respect to pollen, and vice versa for pollen-side factors. This last filter ensured that differences in styles were unlikely to be contributed by pollen in the styles of pollinated samples.
A number of a priori pollen and pistil UI candidate genes were selected based on information from previous publications [23,26,28,31,55]. Some of the SLF genes have not yet been annotated in the S. pennellii genome, so we used locus numbers and sequences from Li and Chetelat [28] to find these genes in our dataset. We identified expression levels of all a priori candidates, and carried out statistical tests similar to the ones described above to determine whether they were differentially expressed.

Additional files
Additional file 1: This excel file contains all of the additional Tables (Table S1-Table S10) associated with the manuscript. Each Table is in a  different tab. Table numbers and titles are listed below: Table S1. Genes showing differential regulation in a pairwise comparison of LA1777 selfpollinated versus unpollinated (UP) styles. Table S2. Genes showing differential regulation in a pairwise comparison of LA1777 intrapopulation pollinated versus unpollinated (UP) styles. Table S3. Genes showing differential regulation in a pairwise comparison of LA1777 interpopulation (LA0407) pollinated versus unpollinated (UP) styles. Table S4. All significantly upregulated genes in styles of LA1777 vs. LA0407. Table S5. All significantly downregulated genes in styles of LA1777 vs. LA0407. Table S6. Expression of patterns of a priori gene candidates in styles of LA1777 vs. LA0407. Table S7. All significantly upregulated genes in pollen of LA1777 vs. LA0407. Table S8. All significantly downregulated genes in pollen of LA1777 vs. LA0407. Table S9. Expression of patterns of a priori gene candidates in pollen of LA1777 vs. LA0407. Table S10. Solanum habrochaites samples used to create RNA-seq libraries. (XLSX 140 kb) Additional file 2: This PDF contains all of the additional material ( Figure S1, Figure S2 and Method S1) associated with the manuscript. Figure/Method numbers and titles are listed below. Figure S1. The genome-wide patterns of expression in styles from two populations of Solanum habrochaites.