- THIS ARTICLE
-
Abstract
- Full Text (PDF)
- Alert me when this article is cited
- Alert me if a correction is posted
- SERVICES
- Similar articles in this journal
- Similar articles in PubMed
- Alert me to new issues of the journal
- Download to citation manager
- Reprints & Permissions
- CITING ARTICLES
- Citing Articles via Google Scholar
- GOOGLE SCHOLAR
- Articles by Perazza, D.
- Articles by Bonneville, J.-M.
- Search for Related Content
- PUBMED
- PubMed Citation
- Articles by Perazza, D.
- Articles by Bonneville, J.-M.
Trichome Cell Growth in Arabidopsis thaliana Can Be Derepressed by Mutations in at Least Five Genes
Daniel Perazza1,a, Michel Herzoga, Martin Hülskampb, Spencer Brownc, Anne-Marie Dornea, and Jean-Marc Bonnevilleaa Laboratoire de Génétique Moléculaire des Plantes, CNRS-Université J. Fourier, 38041 Cedex 9, Grenoble, France,
b Lehrstuhl für Entwicklungsgenetik, Universität Tübingen, D-72076 Tübingen, Germany
c Institut des Sciences Végétales, CNRS, 91198 Gif-sur-Yvette, Cedex, France
Corresponding author: Jean-Marc Bonneville, Laboratoire de Génétique Moléculaire des Plantes, CNRS-Université J. Fourier, BP 53, 38041 Cedex 9, Grenoble, France., jean-marc.bonneville{at}ujf-grenoble.fr (E-mail)
Communicating editor: J. CHORY
 | ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Leaf trichomes in Arabidopsis are unicellular epidermal hairs with a branched morphology. They undergo successive endoreduplication rounds early during cell morphogenesis. Mutations affecting trichome nuclear DNA content, such as triptychon or glabra3, alter trichome branching. We isolated new mutants with supernumerary trichome branches, which fall into three unlinked complementation groups: KAKTUS and the novel loci, POLYCHOME and RASTAFARI. They map to chromosomes IV, II, and V, respectively. The trichomes of these mutants presented an increased DNA content, although to a variable extent. The spindly-5 mutant, which displays a constitutive gibberellin response, also produces overbranched trichomes containing more nuclear DNA. We analyzed genetic interactions using double mutants and propose that two independent pathways, defined by SPINDLY and TRIPTYCHON, act to limit trichome growth. KAKTUS and POLYCHOME might have redundant actions mediating gibberellin control via SPINDLY. The overall leaf polysomaty was not notably affected by these mutations, suggesting that they affect the control of DNA synthesis in a tissue- or cell type-specific manner. Wild-type tetraploids also produce overbranched trichomes; they displayed a shifted polysomaty in trichomes and in the whole leaf, suggesting a developmental program controlling DNA increases via the counting of endoreduplication rounds.
TRICHOMES are epidermal hairs found on the aerial surfaces of nearly all plants; in different species, they adopt different shapes and play a variety of functions, including acting as glandular secretor organs and protecting against insect predators (
Arabidopsis is a highly polysomatic species: many cells, in rosette leaves and in hypocotyls in particular, contain nuclear DNA amounts more than four times higher than C, the haploid DNA content (
The dispensable nature of trichomes for the laboratory life of Arabidopsis has facilitated elaboration of a genetic system to analyze endoreduplication control. Mutants in more than 20 loci are known to affect trichomes and have been recovered in several screens using ethyl methanesulfonate (EMS;
Trichome initiation is impaired by mutations affecting the TRANSPARENT TESTA GLABRA or the GLABROUS1 genes. While the ttg syndrome is pleiotropic, affecting also seed coat and root epidermis (
Trichome branch number is altered together with DNA content in several mutants. In the glabrous3 mutants (gl3), trichomes are smaller, producing at most two branches (
Trichome branching can also be affected without apparent modification of the final cell volume or of the nucleus size: architecture mutants with a reduced branching define four loci (angustifolia, stachel, stichel, and zwichel;
Both initiation and morphogenesis of trichomes are under gibberellin hormone (GA) control. The expression of the GL1 gene is positively regulated by GAs (
The implication of several unlinked mutations in the successive events of trichome initiation and of cell morphogenesis suggests that both processes are controlled by overlapping mechanisms, where gibberellins play an important role. In this work, we describe the production of overbranched trichomes by new mutants as well as by polyploid lines. The mutant lines define three genes, KAKTUS (KAK), RASTAFARI (RFI), and POLYCHOME (PYM). Analysis of the genetic interactions in double mutants give new insights on how trichome growth is controlled in Arabidopsis. Overbranched trichomes contained more nuclear DNA than wild type, and our results give new evidence for a link between processes controlling endoreduplications and cell morphogenesis in trichomes.
| MATERIALS AND METHODS |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Arabidopsis stocks:
The original tetraploid line derived from Columbia (Col) ecotype (4n Col) was isolated from a population mutagenized using Agrobacterium tumefaciens T-DNA and in vitro regeneration from roots, kindly provided by Csaba Koncz (Max-Planck Institut, Köln, Germany), in a screen for seedlings germinating on uniconazol; the trichome phenotype of this line was discovered fortuitously. Triploid plants are the F1 plants from a 2n Col x 4n Col cross. These seeds are fully viable, despite endosperm genetic imbalance; they are fully fertile, but their F2 offspring contain many abortive or strongly deformed plants (aneuploids). The tetraploid line derived from Landsberg erecta and used in Figure 1, Figure 5, and Figure 7 is a kind gift from Dr. Jeff Leung (CNRS, Gif-sur-Yvette, France). The spy-5 mutant is in a Ler background (
|
|
|
|
|
|
4-branch trichomes were selected for wild type and mutants, respectively, DAPI-stained, and treated as for
|
Mutant screening:
Approximately 16,000 individual F2 lines descending from Ler seeds mutagenized with EMS according to
Plant growth and observation:
Plants were grown in soil under long day conditions (16/8 hr) and observed under low magnification (x20–50) with a binocular lens. All visible trichome branches were counted, irrespective of their size; 
2 tests were performed using Microsoft Excel5 software to compare distributions, and trichome populations were considered different when the associated probability was lower than 0.01. For example, in Figure 2, trichomes with 3:4:>4 branches were 101:127:34 for F1 try x pym vs. 103:99:8 for F1 try x Ler (P < 10-3).
For scanning electron microscopy, plants were vacuum infiltrated and fixed for 4 hr at room temperature with 4% glutaraldehyde in phosphate buffer, pH 7.0, 0.02% Tween-20. Third leaves were dissected, mounted on specimen stubs, dehydrated in an ethanol series, critical point dried, and sputter-coated with 200 Å gold-palladium.
Genetic mapping:
Homozygous mutant plants (Ler background) were crossed to Col plants and about 20 F2 plants with a strong overbranched phenotype were selected for each cross. DNA was extracted either from single F2 inflorescences or from F3 pools and characterized by PCR using simple sequence length polymorphism (SSLP) markers (
Construction of double mutants:
Homozygous mutants were crossed pairwise and the F1 plants were allowed to self. For seven crosses (corresponding to the double mutants on Figure 4), F2 plants fell into three classes of increasing phenotypic strength. The weakest class produced a majority of three branch trichomes, i.e., a phenotype compatible with wild-type or single heterozygous genotypes. The second class produced overbranched trichomes resembling the homozygous parental lines. The third class produced very large trichomes bearing at least seven (try x pym, try x rfi, and try x spy-5 crosses) or eight (other crosses) branches, termed maxichomes. The F2 progeny from most crosses between unlinked mutations (try x kak-1, try x kak-2, try x kak-3, try x kak-4, try x pym, pym x rfi, kak-2 x pym, kak-3 x pym) contained a high proportion of plants bearing maxichomes, compatible with either 3/16 or 5/16. F2 plants with maxichomes were selfed, and at least 12 plants were observed for each of the resulting F3 progenies. The maxichome phenotype was transmitted to the F3, either stable or in segregation with plants resembling simple mutants. Stable F3 families, devoid of plants with a trichome branching equivalent to either parental simple mutant, identified the pym rfi, kak-2 pym, try pym, and try spy-5 double homozygous mutants. For try kak-2, neither of the two harvested F3 families was stable, but a stable F4 family was obtained. The production of nested, adjacent trichomes was an additional stable trait in all try double mutants.
The kak-3 rfi double mutant was isolated as a stable F3 family showing a synergetic effect on stems. The kak-3 stems were nearly wild type with a majority of nonstellate (and very rare three-branch) trichomes. On rfi stems, most trichomes had two or three branches; nonstellate trichomes were rare, and four-branch trichomes were never observed. One F3 progeny of the kak-3 x rfi cross produced stem trichomes with four and five branches on all 12 plants observed, identifying the double homozygous. Leaf trichomes bearing eight or more branches were present (but rare) on nearly every kak-3 rfi plant; they were absent from rfi and kak-3 leaves.
In the cross between the linked try and rfi mutations, 143 F2 plants were observed: 9 plants were noted as wild type, 125 with a phenotype compatible with a try/+, rfi/+, try or rfi genotype, and 9 as maxichome bearing; phenotypic segregation in F2 thus confirmed that these mutations are not allelic. None of the nine F3 families harvested was stable for the maxichome trait, but stable F4 families have been isolated.
Trichome DNA staining and measurements:
Trichomes were isolated from the third leaf of 20- to 25-day-old soil grown plants under binocular lens using dissection forceps, fixed for 4 hr in 3.7% paraformaldehyde in phosphate buffer, pH 7, 0.02% Tween-20, and washed three times in fixative-free buffer. Nuclei were stained for 15 min at room temperature with fresh 4',6-diamidinophenylindole (DAPI) solution (1 µg/ml) in phosphate buffer, pH 7, 0.02% Tween-20 containing 0.1 mg/ml p-phenylene diamine as antifading agent. After three washes in phosphate/Tween-20 buffer, samples were mounted between slide and coverslip in 50% glycerol containing 0.1 mg/ml p-phenylene diamine and visualized using epi-fluorescence microscopy. Microfluorometry was carried out using the SAMBA device (System for Analytical Microscopy in Biomedical Applications, UNILOG, Meylan, France). The hard- and software packages of the system were as described (
To measure the DNA content of guard cells, whole leaves from wild-type tetraploids were processed as detached trichomes as used in whole mount. Fluorescence units for trichome nuclei were converted into C values by taking the mean fluorescence of 4n stomatal guard cell nuclei as 4C (an external standard representing the lowest C content on the leaf;
Flow cytometry:
A fresh, fully expanded Arabidopsis leaf was chopped together with a 25-mm2 piece of tomato leaf (internal standard). The chopping buffer (pH 7) contained 45 mM MgCl2, 30 mM sodium citrate, 20 mM 4-morpholinepropane sulfonate, and 1 mg/ml Triton X-100, as described by
| RESULTS |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
A screen for mutants with large, overbranched trichomes:
Wild-type, diploid A. thaliana plants produce branched trichomes on rosette leaves. In many ecotypes, and in particular Landsberg erecta (Ler), the three branch trichomes predominate (Figure 1); the stalk and the three branches regularly present a nearly tetrahedral geometry. Since an increase in trichome DNA content appears to lead to an increased branch number in a try mutant, we reasoned that other mutants presenting an overdevelopment of trichomes might be defective in endoreduplication control. We had previously observed that a reduction in GA biosynthesis can mimic the gl3 phenotype (
When comparing trichomes of equal branch number, the repertories of cell shapes produced by all single mutants appeared to overlap deeply. Branching points often appeared farther apart from each other than on wild-type trichomes, resulting in forked branches (Figure 1); this trend was more pronounced on five- and six-branch trichomes. The phenotype of single mutants showed some extent of variability in successive batches of soil-grown plants. In at least four independent measurements, the proportion of three-branch trichomes varied from 78 to 92% for Ler, 7 to 40% for pym, 13 to 30% for try, and 2 to 8% for rfi. The hierarchical order of phenotypic strengths for branch number, Ler < pym 
kak-2, kak-3, try < rfi, was nonetheless always respected when plants were grown side by side.
In addition to trichome overdevelopment, the rfi mutation also led to a small but consistent and significant increase in trichome clustering. On rfi plants, 2.5% trichomes were found within a nest, i.e., 22 trichomes out of 884 had another trichome as neighbor cell vs. 0.1% for Ler (2/1456) and 10.5% for try (43/411). Trichome clustering was not increased in spy-5, pym, and kak mutants. The number of trichomes per leaf was somewhat reduced in try, pym, and rfi, but not in the kak mutants (data not shown). The distribution of trichomes on the leaf blade appeared otherwise normal in all mutants. On stems, the majority of wild-type trichomes were nonstellate (i.e., unbranched), and only 10–20% had two branches. No major increase in trichome stem branching was observed for the pym and spy-5 mutants. The kak mutants presented a very weak stellate phenotype on stems (data not shown). By contrast, nearly all stem trichomes from try and rfi mutants were stellate, presenting two or three branches, and with stalks typically shorter than on leaves.
To define the formal nature of their genetic defects, mutant lines were backcrossed to their wild-type ecotype, Ler. Heterozygous F1 plants produced by crosses with kak-2, kak-3, kak-4, and pym were wild-type in phenotype (Figure 2B). In addition, the F2 populations derived from crosses of these four mutant lines with either Ler or Col wild-type plants segregated for wild type and strongly overbranched phenotypes in a ratio close to 3:1 (data not shown). These data indicate the presence of a single, recessive Mendelian mutation for kak-2, kak-3, kak-4, and pym. By contrast, lines try, spy-5, kak-1, and rfi yielded F1 plants with a weakly overbranched phenotype (Figure 2B). The semidominant nature of the try mutation on trichome branching is in agreement with
Genetic mapping defines five complementation groups:
Due to the fact that some mutations are semidominant, it may be misleading to class mutants into complementation groups on the sole phenotype of F1 plants from crosses between homozygous mutants. As a first step to exclude possible allelic relationships, we therefore mapped the mutations, using DNA polymorphic markers (Figure 3). Mutants were crossed to wild-type plants of a different ecotype, Col, F2 plants showing a strong overbranched phenotype were selected, and their DNA was analyzed (MATERIALS AND METHODS). Whereas most markers yielded Ler and Col chromatides in 1:1 ratio, we noted a skew in favor of Col in a few cases (MATERIALS AND METHODS); most of these markers also show a Col excess in the Col x Ler recombinant inbred (RI) population (
|
To summarize, our genetic screen allowed the definition and mapping of three loci, KAK, PYM, and RFI, where mutations result in a trichome overdevelopment. These phenotypes are strongly reminiscent of both try and spy plants.
Double mutants define distinct pathways regulating trichome cell growth:
To define whether the five wild-type genes SPY, TRY, KAK, PYM, and RFI act in a single, linear pathway or in several parallel ones, double mutants were constructed to define their additive or epistatic interactions. Each of the 10 possible pairwise combinations between two different complementation groups was represented by at least one cross. In several cases, trans-heterozygous F1 plants showed overbranched trichomes, as illustrated on Figure 2C for four crosses. Mutations try and rfi are both semidominant; the trans-heterozygous try/+ rfi/+ plants could not be distinguished from either homozygous parent (Figure 2C and A). The three other crosses involved at least one (kak-3 x rfi, pym x try) or even two recessive mutations (kak-2 x pym). Again, the corresponding trans-heterozygous plants showed a stronger phenotype than plants heterozygous at a single locus (compare Figure 2C and Figure B). According to our mapping data (Figure 3), the mutations combined in these four crosses are not allelic. Indeed, in the last three cases, the phenotype of plants heterozygous at two distinct loci was less marked than that exhibited by the homozygous parents (Figure 2A and Figure C). These data therefore indicate that some cumulated haplo-insufficiencies can lead to a synthetic phenotypic enhancement.
In seven crosses (corresponding to the double mutants on Figure 4), some F2 plants produced clearly larger trichomes bearing more than seven branches, hereafter referred to as maxichomes. Maxichomes from three different genotypes are illustrated on Figure 4. Branches ramified twice or more were frequent, resulting in tree-shaped cells. The relative size of branches was variable, and the extreme situation of a bump on a main branch could be found (Figure 4C): branch initiation is therefore maintained until late stages of mutant trichome cell growth. Maxichomes were never observed in the homozygous parental lines, so they must represent a synthetic phenotype. In the try x spy-5 cross, about 1/16 of the F2 plants showed such a maxichome phenotype (Table 2), indicating that this additive interaction requires both loci to be homozygous to be detected. Other crosses between unlinked mutations produced significantly more than 1/16 F2 plants bearing maxichomes (see list in MATERIALS AND METHODS; data not shown), suggesting that some genotypes with one homo- and one heterozygous mutant locus lead also to an additive phenotype.
|
Homozygous double mutants were isolated on the basis of their enhanced phenotype, which was stabilized in their offspring (MATERIALS AND METHODS). Trichome branching was quantified for each double mutant and compared to parental single mutants. This revealed five strongly additive interactions (Figure 4D) and two weakly additive ones (Figure 4E). The try mutation was strongly additive in double mutants with spy-5, pym, or kak-2 (Figure 4D). This was characterized by an increase in branch number in the double mutants higher than the sum of the increases observed in the parental single mutants. Maxichomes were also frequently observed in the F2 progeny of try x kak-1, try x kak-3, and try x kak-4 crosses (data not shown), indicating that a strong additivity is not due to specific interactions with a given kak allele. The simplest explanation to these strongly additive combinations is that TRY defines a repression route independent of SPY, KAK, and PYM. Besides increased trichome branching, we also noted that the four double mutants try spy-5, try pym, try kak-2, and try rfi frequently produced clustered trichomes (e.g., see Figure 4B and Figure C). They also contained a small proportion of trichomes with one very basal branch emerging very low on the stalk, a feature also observed on the try simple mutant but on no other genotype. On both try pym and try rfi double mutants, we sometimes noted swellings affecting the upper part of the stalk or one main branch: the swollen part could be twice as wide as the stalk.
The quantitative effects of the rfi mutation on trichome branching varied sharply in three different genetic combinations. The pym rfi double mutant presented the highest score (Figure 4D), suggesting that RFI and PYM act in independent pathways. By contrast, the try rfi and kak-3 rfi double mutants only showed a partial additivity (Figure 4E). In the try rfi double mutant, the increase in trichome branching was clearly less than the added contributions of the two single mutations. A similar situation was observed for the kak-3 rfi double mutant: when kak-3 rfi was compared to rfi, the global increase in leaf trichome branching was extremely low, although an additive phenotype was also observed on kak-3 rfi stems (MATERIALS AND METHODS). The kak-2 x rfi cross produced F2 plants similar to the kak-3 x rfi cross (data not shown).
The pym mutation showed a strong additivity in double mutants with try and rfi, and also with kak-2 (Figure 4D). Some of the plants observed in the F2 progeny of kak-4 x pym and kak-3 x pym crosses were very similar to the kak-2 pym double mutant, suggesting that a strong additivity does not require a specific kak allele (data not shown). The simplest explanation for these additive phenotypes is that the wild-type PYM gene represses branch formation through a pathway independent of both KAK and TRY.
By contrast with the seven genetic combinations described above, where at least a qualitative increase could be found, three crosses with spy-5 produced F2 populations devoid of plants with an increased trichome branching: spy-5 x kak-2, spy-5 x pym, and spy-5 x rfi. Table 2 shows that the complete absence of F2 plants defining an additive class is highly significant for unlinked mutations. The simplest explanation for each of these three pairs of mutations is that they both act to block (or downregulate) the same genetic pathway.
Tetraploid lines produce overbranched trichomes:
A Col tetraploid derivative was isolated during our initial screening for lines showing overbranched trichomes, as mentioned above. Among Ler offspring, we also isolated a tetraploid line with a similar phenotype (MATERIALS AND METHODS), suggesting that trichome overbranching is a general feature of genetically polyploid lines. Examination of another Ler tetraploid derivative confirmed the prominent presence of four-branch trichomes (Figure 1B), i.e., a leaf phenotype very close to kak and pym mutants. We quantified the increase in trichome branching with ploidy, which was clear in both Ler and Col ecotypes (Figure 5A). Interestingly, triploid plants presented a trichome branching intermediate between diploids and tetraploids, with about as many trichomes bearing three branches as four. Arabidopsis stocks described as polyploid and available from the Ohio Arabidopsis Biological Resource Center (http://aims.cps.msu.edu/aims/) were scored for trichome branching and assayed for their genomic content by flow cytometry. The two lines confirmed to be true tetraploids, CS3151 (another Col derivative) and CS3432, also produced mainly four- and five-branch trichomes (data not shown). Thus a general consequence of polyploidy is to raise the probability for a leaf trichome to make more than three branches.
As trichome cells are known to undergo endoreduplications, we studied the trichome nuclear DNA content in diploid, triploid, and tetraploid plants. DAPI-stained trichomes were examined by microfluorometry (Figure 5B). Trichome nuclei from wild-type diploid plants presented a broad distribution ranging from about 16C to about 64C. For triploid and tetraploid lines, trichome nuclei distributions were shifted to higher values, and the mean DNA fluorescence values increased with the initial ploidy level. In genetically polyploid plants, an increase in trichome branching is therefore correlated with an increased trichome DNA content.
More DNA in mutant trichomes with supernumerary branches:
We then asked whether trichomes from mutant lines also contained an increase in nuclear DNA content. Figure 6 shows that a shift toward higher DNA contents was observed for all single mutants; the extent of this shift, however, varied among mutants. The pym mutant showed the strongest increase in DNA fluorescence. An intermediate increase was observed for the spy-5, kak-1, and kak-2 mutants, as well as for the kak-3 mutant (compared to Ler in an independent experiment; data not shown). The increase in DNA content was the weakest, and clearly less than twofold for the kak-4 and rfi mutants. Compared to wild type, the increases in mean trichome nuclear fluorescence were nonetheless significant: for rfi vs. Ler, P < 10-4 and for kak-4 vs. Ler, P < 10-3 in Student's t-tests. Trichome nuclear areas from mutants also showed a shift toward higher values, with mean values significantly higher than those for wild type (data not shown). The observed trend thus suggests that the Arabidopsis genome, as a part or as a whole, undergoes additional replication during development of overbranched trichomes.
Arabidopsis leaves are known to be highly polysomatic, containing only a minority of 2C cells and high proportions of cells with nuclear contents of 4C, 8C, and 16C (
We finally examined whether an increased trichome branching corresponds to a further increase in DNA content in a double mutant situation. Trichomes from wild-type, pym, rfi, and pym rfi plants were stained with DAPI, and their nuclear fluorescence was recorded (Figure 8). As described above, trichome nuclei in either simple mutant were larger on average than in wild type. The double mutant exhibited a very broad distribution, with 8% nuclei having less than one time and 25% nuclei having more than four times the wild-type trichome mean DNA content (6 and 19 nuclei out of 76, respectively). This class of very large nuclei was absent in rfi and much less abundant in pym trichomes (8%, 6/85). The mean nuclear fluorescence was significantly higher for the double mutant than for either single (i.e., pym rfi vs. pym: P < 0.05 in a Student's t-test), and both single mutants were higher than wild type. Therefore, at least for this double mutant, the presence of trichomes bearing more branches correlates with the presence of larger nuclei with a brighter fluorescence. These data are compatible with the idea that a fraction of the pym rfi maxichomes has undergone two more endoreduplication rounds than wild-type trichomes.
|
| DISCUSSION |
|---|
| TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
New large overbranched trichome mutants:
Previous studies have identified three types of mutants producing large overbranched trichomes: try, kak (
A single mutant allele only has been recovered for each of the RFI and PYM genes, and it may be that mutations in more than five genes lead to a similar phenotype. Trichomes from the mutants described in this study differ from those produced by the nok-122 mutant, which also have supernumerary branches, but whose cell and nuclear size are similar to wild-type trichomes (
SPY and TRY act by independent pathways:
The five genes TRY, SPY, KAK, PYM, and RFI act on trichome growth, as mutant trichomes bear extra branches and appear larger than wild type. Trichome growth derepression is never a fully dominant trait in the corresponding mutants; it is recessive in pym and three of the kak mutants. Other traits associated with spy-5 and try mutations are also recessive (
The trichome overbranching phenotype provoked by single mutations described in this study was strongly enhanced in some double mutant combinations and weakly or not at all enhanced in others. A nonadditive phenotype in a double mutant shows the epistasy of one mutation over the other and indicates that the two corresponding genes act in a common, linear pathway. A weak, less than arithmetically additive phenotype can be assimilated with epistasy (e.g., see
A common feature to the mutations used in this study is that the phenotype they confer to single mutants could always be strongly enhanced in at least one double mutant combination (e.g., spy-5 was enhanced by try). This observation implies that no single mutation is sufficient to fully derepress trichome cell growth. The first possible explanation is that all mutations are leaky. The spy-5 mutation is a missense (
A SPINDLY-dependent pathway can be defined from the epistatic relations of spy-5 with pym, kak, and rfi in double mutants (Table 2). The kak rfi double mutants presented an extremely weak phenotypic additivity (Figure 4E), which further suggests that the SPY, KAK, and RFI genes act within a linear pathway. However, the kak pym and pym rfi double mutants presented a strongly additive phenotype (Figure 4D), and this suggests a redundancy between PYM and KAK/RFI to mediate SPY action. The recessive mutations kak-2 and pym displayed nonallelic noncomplementation (Figure 2C), and a possible explanation is that KAK and PYM functions are largely overlapping. At present, we have no genetic evidence to order genes within the proposed SPY/KAK/RFI and SPY/PYM pathways. However, the highly pleiotropic phenotype of spy compared to kak, rfi, and pym mutants makes SPY likely to act upstream from KAK, RFI, and PYM genes. Whereas SPY modulates most gibberellin responses (
A SPINDLY-independent pathway is suggested primarily by the additive phenotype between spy-5 and try (Table 2; Figure 4D). This proposal implies that TRY is not acting within the SPY-PYM linear pathway and is confirmed by the strongly additive try pym phenotype. Similarly, TRY is not cross-talking with SPY/KAK since the four try kak combinations also show strong additivity. On the other hand, only weak additivity was observed between try and rfi, which contrasts with the other try double mutants. This suggests that TRY could exert part (or all) of its effect through the RFI gene; it would be consistent with the occurrence of clustered trichomes on leaves and of highly branched trichomes on stems in try and rfi mutants only. As mentioned above, RFI also appears to interact with KAK, and this would mean that the TRY and KAK pathways are convergent, with RFI acting at (or downstream of) the convergence point.
Taken together, our data are compatible with a model in which the TRY and SPY genes limit trichome growth by distinct pathways. The SPY-dependent response may involve the parallel actions of PYM and KAK. The TRY and KAK pathways would converge on RFI.
Derepressed DNA synthesis in mutants:
An increase in nuclear DNA content was observed in pym, spy-5, kak, and rfi trichomes. This increase was a plain doubling for pym but surprisingly less for the spy, kak, and rfi overbranched trichomes. A less than twofold increase in nuclear DNA fluorescence has also been reported for try trichomes (
Interestingly, mutations leading to an increased DNA content in trichomes did not lead to a general derepression of endoreduplications in all plant tissues, as indicated by leaf polysomaty distributions close to wild type (Figure 7). This suggests that nuclear DNA synthesis is triggered differently in trichomes and in the major tissues of the leaf blade (i.e., in mesophyll cells). This may involve distinct mechanisms, or different threshold levels of a common mechanism in different cell types. Our flow cytometry histograms of bulk leaf tissue, however, may not detect changes in the polysomaty of a minor class like epidermal pavement cells. The issue of a cell-type or tissue-specific action of the mutated genes therefore remains open.
Gibberellin hormones have recently been shown to be involved in both trichome initiation and trichome branching control (see Introduction). We show here that spy-5 trichomes contain nearly twice as much DNA as wild-type trichomes. So far, all processes affected by the spy mutations are known to be GA-controlled (
A patterning defect in rfi and try mutants:
The rfi mutant presented an increase not only in trichome branching but also in trichome clustering: whereas wild-type trichomes are evenly spaced on the epidermis, rfi trichomes were found at a higher frequency with another trichome as neighbor cell. This pleiotropic phenotype has been noted earlier for the three try alleles (
Leaf cells count the number of endocycles:
Diploid, triploid, and tetraploid plants were found to produce trichomes containing increasing nuclear DNA amounts. This strongly suggests a developmental program in which the cell specified to become a trichome counts the number of endocycles, independently of its initial chromosome number. Flow cytometry analysis of whole leaves of diploid and tetraploid lines indicates that the ratios of leaf cells having undergone 0, 1, 2, or 3 endoreduplication cycles are also independent of the initial genome size. Our data rule out a feedback mechanism warning the nucleus about its genomic content past a critical threshold. An equivalent conclusion has been reached by comparing hypocotyl cells on diploid and tetraploid seedlings (
More DNA in larger cells:
The increased trichome branch number in polyploid vs. diploid lines suggests a coupling between cell size and nuclear DNA content in wild-type trichomes. Such a coupling has been documented in other Arabidopsis tissues and in many other eukaryotic organisms. Arabidopsis epidermal pavement cells, for instance, show a linear correlation between cell volume and nuclear DNA content (
The spy-5, kak, rfi, and pym mutants showed an increase in both trichome nuclear content and branch number. A first possibility is that increases in DNA content are the consequences of an increase in cell volume. This is, however, quite unlikely because the opposite has been described in hypocotyl cells: when seedlings are allowed to germinate, endoreduplications take place before cell elongation (