cot1: A Regulator of Arabidopsis Trichome Initiation

Corresponding author: M. David Marks, University of Minnesota, Department of Genetics and Cell Biology, 250 Biological Sciences Center, 1445 Gortner Ave., St. Paul, MN 55108-1068, dmarks{at}biosci.cbs.umn.edu (E-mail).

Communicating editor: E. MEYEROWITZ

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In Arabidopsis, the timing and spatial arrangement of trichome initiation is tightly regulated and requires the activity of the GLABROUS1 (GL1) gene. The COTYLEDON TRICHOME 1 (COT1) gene affects trichome initiation during late stages of leaf development and is described in this article. In the wild-type background, cot1 has no observable effect on trichome initiation. GL1 overexpression in wild-type plants leads to a modest number of ectopic trichomes and to a decrease in trichome number on the adaxial leaf surface. The cot1 mutation enhances GL1-overexpression-dependent ectopic trichome formation and also induces increased leaf trichome initiation. The expressivity of the cot1 phenotype is sensitive to cot1 and 35S::GL1 gene dosage, and the most severe phenotypes are observed when cot1 and 35S::GL1 are homozygous. The COT1 locus is located on chromosome 2 15.3 cM north of er. Analysis of the interaction between cot1, try, and 35S::GL1 suggests that COT1 is part of a complex signal transduction pathway that regulates GL1-dependent adoption of the trichome cell fate.

TRICHOMES are specialized hair cells that are found on the epidermal surface of almost all terrestrial plants (JOHNSON 1975 ). Trichome structure is diverse, ranging from simple, unicellular spikes to multicellular, glandular organs. The functional roles for trichomes have been suggested to include environmental adaptation and plant defense against herbivory (LEVIN 1973 ). On Arabidopsis leaves, trichomes are a nonessential cell type and exist as large unicellular structures that contain three or four branches (ROLLINS and BANERJEE 1976 ; HAUGHN and SOMERVILLE 1988 ). In Arabidopsis, trichome initiation is integrated with the leaf developmental program. Trichome initiation first occurs at the leaf tip after the leaf primordium reaches a length of ~100 µm (PYKE et al. 1991 ; LARKIN et al. 1996 ). As the leaf expands, trichome initiation proceeds basipetally, and subsequent initiations become restricted to the leaf base or intercalary zones between existing trichomes (MARKS et al. 1991 ; HULSKAMP et al. 1994 ; LLOYD et al. 1994 ). This article addresses the genetic control of the competency of epidermal cells to enter the trichome pathway.

At least 24 genes are known to regulate aspects of trichome initiation and morphogenesis (reviewed in MARKS 1997 ). The TRANSPARENT TESTA GLABRA (TTG) and GLABROUS1 (GL1) genes are required for normal trichome initiation. Mutations in these genes result in a complete loss of trichomes on most plant surfaces. The deduced amino acid sequence of the TTG gene contains seven WD-40 repeat motifs (M. WALKER and J. GRAY, personal communication). Mutations in TTG also cause defects in seed coat mucilage development, anthocyanin production, and root hair initiation (KOORNNEEF 1981 ; MASUCCI et al. 1996 ). The maize R gene that encodes a myc-like transcription factor can suppress all aspects of the ttg mutant phenotype (LLOYD et al. 1994 ). This suggests that TTG may regulate the activity of a myc-like gene needed for trichome initiation. The GL1 gene encodes a putative myb-type transcription factor, and the only known phenotype of gl1 null plants is defective trichome initiation (OPPENHEIMER et al. 1991 ). The GLABRA2 (GL2) gene is required for normal trichome morphogenesis and has been analyzed as a potential molecular target for GL1 (SZYMANSKI et al. 1998 ). Overexpression of GL1 is necessary but not sufficient for ecotopic GL2 expression. The maize R gene can provide additional activity, such that overexpression of GL1 and R leads to ectopic GL2 transcription (SZYMANSKI et al. 1998 ). These results indicate that GL1 does not act alone to promote trichome initiation but instead requires interaction with additional factors.

GL1 expression is tightly regulated in wild-type plants (LARKIN et al. 1993 , LARKIN et al. 1994 ). Expression appears uniform in the epidermis of the young leaf primordium but is restricted to trichomes during later leaf development. Expression in trichomes is transient, with no GL1 expression detected in trichomes on the mature leaf. To examine the effects of GL1 expression throughout leaf development, the cauliflower mosaic virus 35S promoter (35S::GL1) was used to overexpress the GL1 gene. mRNA in situ hybridization (LARKIN et al. 1994 ) and immunohistochemistry experiments (D. SZYMANSKI, unpublished results) confirmed that the GL1 transgene was highly expressed and translated. Overexpression of GL1 resulted in the initiation of a few ectopic trichomes on the cotyledon and abaxial surface of the first leaf pair of wild-type plants (LARKIN et al. 1994 ), but surprisingly led to a decrease in trichome number on the adaxial leaf surface, suggesting that an overabundance of GL1 had an inhibitory effect on initiation (LARKIN et al. 1994 , LARKIN et al. 1997 ). A "squelching" mechanism has been proposed in which overexpression results in excess GL1 protein that can titrate other required proteins away from gene targets needed for trichome initiation (LARKIN et al. 1994 ). However, interactions between 35S::GL1 and additional loci that regulate trichome initiation are not consistent with the squelching model. Additional genetic experiments with 35S::GL1 suggested that the inhibition of trichome initiation in the leaf requires both TTG and TRYPTICHRON (TRY). 35S::GL1 plants heterozygous for ttg show an increase in adaxial trichome numbers similar to wild type and a breakdown of trichome spacing control (LARKIN et al. 1994 ). Because TTG activity positively regulates trichome initiation and potentially the hypothetical squelched factor, an increase in trichome initiation with reduced TTG gene dosage is not consistent with the squelching model. Similarly, try mutants that overexpress GL1 show enhanced trichome initiation. TRY has been proposed to have an inhibitory function during trichome development (HULSKAMP et al. 1994 ; LARKIN et al. 1997 ). try mutations cause trichome clustering and extra branch formation, and affect the DNA content of mutant trichomes (HULSKAMP et al. 1994 ; FOLKERS et al. 1997 ). 35S::GL1 plants homozygous for try exhibit an increase in adaxial leaf trichomes and ectopic trichomes (D. SZYMANSKI, unpublished results). Unless TRY directly affects the balance of GL1 and the hypothetical squelched factor, results with try are also inconsistent with the squelching model. The above genetic data and the results in this article are most consistent with GL1 overexpression leading to an inhibited state, during which subsequent trichome initiation is not observed.

The extent of trichome initiation in 35S::GL1/- try and 35S::GL1/- TTG/ttg plants is limited: only a small fraction of epidermal cells enter the trichome pathway. This suggests that trichome initiation is under complex control and that additional factors limit trichome initiation in the epidermis. To identify additional genes that effect the abundance and spacing of trichomes in 35S::GL1 plants, 35S::GL1 homozygous seeds were mutagenized and screened for suppressors or enhancers. This article reports the characterization of the cot1 mutant that was identified in a mutagenized 35S::GL1 population. The cot1 mutation increases ectopic and leaf trichome initiation in 35S::GL1-plants, but cot1 plants have no detectable phenotype in the absence of the 35S::GL1 transgene. In the presence of 35S::GL1, cot1 interacts with try and leads to a synergistic enhancement of the cotyledon trichome phenotype and reduced seedling viability. The timing and spatial arrangement of cot1-dependent trichomes are consistent with the presence of multiple points of trichome initiation control.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant strains, growth conditions and mutagenesis:
Columbia (Col), Wasselewskija (WS), and Niederzenz (Nd-O) ecotypes were used for crosses. The gl2-3 mutant was isolated from a WS T-DNA transformed line (RERIE et al. 1994 ). The 35S::GL1 4-1 transformant was made in the Col background (OPPENHEIMER et al. 1991 ) and has been maintained as a homozygous line. The 35S::GL1 Cla4A construct was derived from 35S::GL1. The Cla4A construct lacks the 3' GL1 sequence downstream of the ClaI site, which is located 82 bases downstream of the GL1 stop codon (OPPENHEIMER et al. 1991 ). This construct was transformed into the WS background. The 35S::GL1 Cla4A phenotype was indistinguishable from 35S::GL1 4-1 (M. D. MARKS, unpublished results).

Seeds were sown on 2 inches of Peters Redi-Earth (Grace Sierra, Milpitas, CA) that was layered over coarse vermiculite. The soil was initially saturated with a complete nutrient solution (FELDMANN and MARKS 1987 ). Plants were propagated in growth chambers at 22° under constant illumination with fluorescent lights (190 µE m^-2 sec^-1).

For mutagenesis of 35S::GL1 4-1 plants, seeds homozygous for 35S GL1 4-1 were treated with 0.3% ethyl methanesulfonate (EMS) for 8 hr. M₂ seeds were collected from 104 M₁ pools, each consisting of ~30 M₁ plants. Phenotypic and segregation analyses were performed on plants into which the cot1 mutation had been outcrossed at least twice. Trichome numbers on wild-type and mutant plants are reported as mean number trichomes ± standard deviation.

Microscopy and histochemistry:
Plants were processed for histochemical GUS staining and trichome counting in leaf primordia as previously described (LARKIN et al. 1994 ). The only modification was that plants were not vacuum infiltrated prior to staining. The MR GL2 promoter construct expression profile is similar to the full-length GL2 promoter (SZYMANSKI et al. 1998 ) and was crossed into the mutant background. For embedding samples in LR White (Electron Microscopy Sciences, Ft. Washington, PA), plants were fixed overnight in 4% buffered paraformaldehyde and dehydrated in an ethanol series. Samples were moved into 100% LR White and polymerized overnight at 60°. Sections were cut using a glass knife on a RMC MT 7000 microtome (Research Manufacturing Corp., Tucson, AZ). For microscopic observations, samples were viewed using a Nikon SMU-Z stereoscope or a Nikon Diaphot-200 inverted microscope with DIC optics (Nikon Corp., Tokyo, Japan). Images were collected on color slides and transformed into digital images using Adobe Photoshop (Adobe Systems Inc., Mountain View, CA). Images were only adjusted for size, brightness, and contrast. Powerpoint 4.0 (Microsoft Corporation, Redmond, WA) was used to construct the final figures. For scanning electron micrograph (SEM) analysis (AHLSTRAND 1996 ), soil-grown plants were frozen in liquid nitrogen and mounted onto the cold stage of a Philips 500 scanning electron microscope (Philips Electron Optics, Eindhoven, Netherlands). Digital images were collected at 1.5 kV using National Institutes of Health Image.

DNA isolation, PCR and mapping:
Genomic DNA was isolated from either a detached leaf or whole seedlings 7 days after germination. Tissue was frozen in liquid nitrogen and ground in a microfuge tube with a small pestle. Two hundred microliters (µl) of DNAZOL (Life Technologies, Gaithersburg, MD) was added to the powdered tissue, and the suspension was mixed thoroughly. After centrifugation for 10 min at 4°, DNA in the supernatant was precipitated with one-half volume of 100% ethanol. The DNA pellet was resuspended in 20 µl of 50 mM Tris/HCl pH 7.5, and 1 µl was used for PCR. Simple sequence length polymorphisms (SSLP) and cleaved amplified polymorphic sequences (CAPS) primer sequences and PCR conditions were as previously described (KONIECZNY and AUSUBEL 1993 ; BELL and ECKER 1994 ) except that MgCl₂ concentration was optimized for each primer pair. The sequence of primer pairs for the markers m322, athbio2, and er were obtained from the Arabidopsis thaliana Database (http://genome-www.stanford.edu/Arabidopsis/Ecker homepage). Linkage assignments and chromosome positions were assigned using three-point analysis and Mapmaker 3.0 (LANDER et al. 1987 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The overexpression phenotype:
In wild-type plants the initiation of trichomes in the first leaf pair is tightly coupled with the stages of leaf development (PYKE et al. 1991 ; LARKIN et al. 1996 ). Leaf primordia must reach a minimum size of ~100 µm before trichome initiation begins (Figure 1A). Trichome formation at the leaf base and within intercalary zones between existing trichomes continued until the leaf reached ~900 µm. The mature leaf had a mean of 28 ± 0.97 trichomes on its adaxial surface (Table 1). Trichome initiation in 35S::GL1 plants mirrored the behavior of wild type up to about the 400-µm stage, after which initiation was inhibited (Figure 1B). Trichome number at this stage is not significantly different from the trichome number on a fully expanded first leaf (Table 1). This behavior is very similar to that of RTN_ler in Col background (see Figure 7 in LARKIN et al. 1996 ) and is consistent with 35S::GL1 inhibition throughout the entire leaf after a defined developmental stage. The spatial arrangement of trichomes on 35S::GL1 leaves represents the final position of the epidermal cells that had differentiated into trichomes at or before the 400-µm stage. The presence of 9.7 ± 1.0 trichomes within 0.3 mm of the leaf margin of 35S::GL1, compared to 15 ± 1.3 in wild type, is also consistent with inhibition of intercalary trichome initiation in the population of epidermal cells near the leaf margin. It is not known if the 35S::GL1-dependent inhibition is mediated by cell-cell contact and/or organ/tissue level processes.

View larger version (24K): In this window In a new window Download PPT slide   Figure 1. —Relationship between leaf length and trichome number in wild-type Col and in homozygous lines of Col that were transformed with the 35S::GL1 transgene. Trichome number for individual leaves at defined lengths was measured and plotted on the histogram. (A) Col. (B) 35S::GL1. The mean trichome number ± SD for mature leaves is indicated with vertical error bars for each genotype. The mean final leaf length for Columbia is 5.4 ± 0.5 mm and for 35S::GL1 is 5.3 ± 0.6 mm.

View larger version (229K): In this window In a new window Download PPT slide   Figure 2. —Scanning electron micrographs of the aerial surfaces of wild-type, 35S::GL1, and Cot1--G Arabidopsis seedlings at the four-leaf stage. (A–D) Wild type. (E–H) 35S::GL1. (I–L) cot1/cot1 35S::GL1/35S::GL1. Arrows indicate representative trichomes and support cells from each genotype: t, wild-type trichome; ut, unbranched trichome; sc, support cell; pt, pavement cell trichome; st, support cell trichome; et, expanding trichome. Bars: A, B, E, F, I, J, K, 50 µm; D, H, L, 20 µm; C, G, 100 µm. (A, E, I) Hypocotyl. (B, F, J) Cotyledon. (C, G, K) Adaxial leaf surface of first leaf pair. (D, H, L) Trichome and support cell structure.

View larger version (70K): In this window In a new window Download PPT slide   Figure 3. —Semithin longitudinal sections through wild-type, 35S::GL1, and Cot1--G trichomes. (A) Wild type. (B) 35S::GL1. (C) cot1/cot1 35S::GL1/35S::GL1. Solid arrows indicate support cells. Open arrows indicate trichomes. Bars: A, B, C, 50 µm.

View larger version (24K): In this window In a new window Download PPT slide   Figure 4. —Histogram indicating the effects of cot1 and 35S::GL1 transgene dosage on the density of cotyledon trichomes in Cot1--G seedlings. (A) Frequency distribution of cotyledon trichomes in a 35S::GL1/35S::GL1 plant. (B) Frequency distribution of cotyledon trichomes in an F3 population derived from a cot1/cot1 35S::GL1/35S::GL1 individual. (C) Frequency distribution in an F3 population derived from a cot1/cot1 35S::GL1/- individual. (D) Frequency distribution in an F3 population derived from a COT1/cot1 35S::GL1/35S::GL1 individual. Dashed lines in C and D indicate putative divisions between genotypic classes. Segregation ratios and calculated chi-squared values are shown in C and D. G represents 35S::GL1.

View larger version (15K): In this window In a new window Download PPT slide   Figure 5. —Diagram of the genetic intervals on chromosomes I and II that contain 35S::GL1 and COT1, respectively. Loci are shown to the right of each chromosome, and the genetic intervals as calculated by Mapmaker 3.0 are shown to the left. The 35S::GL1 and COT1 loci are indicated in bold.

View larger version (84K): In this window In a new window Download PPT slide   Figure 6. —Activity of a GL2::GUS reporter construct in 35S::GL1 and Cot1--35S::GL1 seedlings. (A–C) 35S::GL1. (D–H) cot1/cot1 35S::GL1/35S::GL1. The boxes in B and E indicate the regions that are shown under higher magnification in C and F, respectively. The closed arrow in G indicates the expanding tip of the pavement cell trichome, and the open arrow indicates the cell border at the base. sc, support cell; st, support cell trichome; ct, cotyledon trichome; ht, hypocotyl trichome. Bars: C, F, G, 50 µm. (A–G) Adaxial surface of first leaf pair 14 days after germination. (H) Cotyledon and hypocotyl 3 days after germination.

View larger version (149K): In this window In a new window Download PPT slide   Figure 7. —Scanning electron micrographs of the adaxial surface of cotyledons. (A) 35S::GL1/35S::GL1 try/try. (B) Putative 35S::GL1/35S::GL1 cot1/cot1 try/try double mutant. Bar, 50 µm.

Analysis of the cot1 phenotype:
Overexpression of GL1 results in the initiation of a few ectopic cotyledon trichomes and in a reduced number of adaxial leaf trichomes. To identify enhancers or suppressors of this dominant phenotype, homozygous 35S::GL1 (transformant 4-1) seeds were mutagenized with EMS (see MATERIALS AND METHODS). An M₂ population representing the equivalent of ~3300 M₁ seeds was screened to identify plants with altered trichome distribution. The mutagenesis was successful because additional alleles of ttg (LARKIN et al. 1994 ), dis, and zwi were obtained (data not shown). In the unmutagenized 35S::GL1 4-1 line used in this experiment, ~10% of plants normally contain cotyledon trichomes; however, plants with five or more trichomes on both cotyledons have not been seen in this population (M. D. MARKS, unpublished results), and the presence of at least five trichomes on each cotyledon was used as the criterion to identify mutations. The cot1 mutation was isolated based on its cotyledon trichome phenotype. The severity of the cot1 phenotype (Cot1^--G) was sensitive to the gene dosage of cot1 and 35S::GL1, and will be discussed below. Homozygous cot1 35S::GL1 plants often displayed a strong Cot1^--G phenotype, which is defined by the presence of 20 or more cotyledon trichomes.

To compare the epidermal anatomy of wild-type and mutant leaves, SEMs of wild-type, unmutagenized homozygous 35S::GL1 and homozygous cot1 35S::GL1 plants were obtained (Figure 2). In wild-type Arabidopsis seedlings, the hypocotyl (Figure 2A) and cotyledons (Figure 2B, Table 1) are glabrous. The adaxial surface of the first leaf of Col plants contains an average of 28 trichomes that are evenly spaced (Figure 2C, Table 1). A mature trichome and a developing trichome are indicated in Figure 2C. Mature trichomes have three or four branches and a concentric ring of support cells. The support cells are clearly visible after the associated trichome has initiated branch formation (Figure 2D).

The presence of a 35S::GL1 transgene causes an alteration in the distribution of trichomes on several aerial organs. In contrast to its dosage sensitivity in the presence of cot1, the 35S::GL1 cotyledon and first leaf pair phenotypes are dominant in wild-type background (LARKIN et al. 1994 ). 35S::GL1 plants have an apparently normal hypocotyl epidermis (Figure 2E), but ~10% of the plants have a few ectopic trichomes on the adaxial surface and the margin of the cotyledon (Figure 2F). Cotyledon trichomes were defined as spiked or unbranched epidermal cells with a length of ~100 µm. Cotyledon trichomes were not present immediately postgermination but were first visible as small projections along the cotyledon margin ~2 days after germination. The cotyledon trichomes were usually unbranched and did not acquire a papillate surface (Figure 2F). The adaxial surface of 35S::GL1 leaves has a reduced number of trichomes (Table 1), and the trichomes were most prevalent near the leaf margin (Figure 2G). The reduced number and altered distribution of trichomes on the leaves form the basis for the 35S::GL1 phenotype. Under higher magnification, images of mature 35S::GL1 trichomes indicated that the trichome and support cell structure was indistinguishable from the wild type (Figure 2H).

cot1 35S::GL1 homozygous plants had ectopic trichomes on the hypocotyl (Figure 2I) and cotyledon (Figure 2J, Table 1). The presence of hypocotyl trichomes was variable and not a reliable marker for the Cot1^--G phenotype. The cotyledon phenotype was the most reliable marker. Homozygous cot1 35S::GL1 plants always produced cotyledon trichomes. Most Cot1^--G hypocotyl and cotyledon trichomes were unbranched, and a papillate surface was not observed on these cells. An enhanced number of trichomes was also observed on the adaxial and abaxial surface of cauline leaves in Cot1^--G plants compared to 35S::GL1 plants (data not shown). The distribution of root hairs in Cot1^--G plants could not be distinguished from those of 35S::GL1 plants (data not shown).

In cot1 35S::GL1 homozygous plants, the 35S::GL1-dependent inhibition of trichome initiation is partially suppressed (Table 1); however, enhanced trichome initiation in the first two Cot1^--G leaves was limited to support cells and regions near the leaf margin and reflected the spatial distribution of trichomes that was observed in 35S::GL1 plants (Figure 2K). Two classes of mutant Cot1^--G leaf trichomes were observed: (1) trichomes that appeared to differentiate from support cells (Figure 2K and Figure L), and (2) aberrant trichomes that emerged from epidermal pavement cells (Figure 2K). Pavement cell trichomes were not detected in leaves <500 µm in length, were usually unbranched, and occasionally acquired a papillate surface late in leaf development. The base of pavement cell trichomes was irregular in shape and was not encircled by a highly ordered ring of support cells. Although the first leaves in Cot1^--G plants almost always contained support cell trichomes, the frequency of support cell trichome initiation within a restricted region of the leaf was variable. In some cases, all of the support cells associated with a particular trichome appeared normal, and in other cases several trichomes originated from a single ring of support cells (Figure 2L). Initiation was asynchronous because a range of developmental stages was often observed within a single cluster of support cell trichomes. Like the pavement cell trichomes, ectopic support cell trichomes were only detected on leaves longer than 500 µm. Support cell trichomes always emerged from a ring of cells that had clear support cell structure, and the associated trichome was always fully developed. To determine if support cell trichomes were unique to Cot1^--G, wild-type and 35S::GL1 plants were carefully reexamined. No support cell trichomes were observed on wild-type plants. Examination of 30 35S::GL1 plants at the six leaf stage detected one support cell trichome.

To confirm that the support cell trichomes originated from support cells and were not an outgrowth from the base of a mature trichome, the structure of Cot1^--G trichome clusters was compared to the structures of wild-type and 35S::GL1 trichomes. Longitudinal sections (1 µm) through wild-type (Figure 3A) and 35S::GL1 (Figure 3B) trichomes showed that the basal cell wall perimeter was appressed to an associated support cell. Support cells appeared like cellular shims that extended beneath the trichome base. Cot1^--G support cell trichome clusters were extremely fragile, and intact samples were obtained with 400 nm sections. Longitudinal sections through Cot1^--G trichomes showed similar support cell appression along the trichome base, but a subset of the support cells expanded perpendicularly to the leaf surface (Figure 3C). SEM on whole plants and light microscopy of sectioned trichomes indicated that Cot1^--G support cells expand abnormally and display the visual hallmarks of trichome morphogenesis.

Genetic interactions between cot1 and 35S::GL1:
The genetic basis of the Cot1^--G phenotype was examined in outcrossed F₂ populations that were segregating for Cot1^--G trichomes. To rule out the possibility that Cot1^--G is due to a modified allele of 35S::GL1, several phenotypically wild-type plants from an outcrossed F₂ population segregating for cot1 were crossed to an unmutagenized 35S::GL1 4-1 homozygous line (Table 2, cross 1) or to the 35S::GL1 Cla4a line (Table 2, cross 2) that overexpressed GL1 derived from a different genomic fragment (see MATERIALS AND METHODS). F₂ populations from each cross were characterized, and the strong Cot1^--G phenotype was observed in progeny from both crosses (Table 2). Selfing of the wild-type parent in these crosses resulted in exclusively wild-type progeny; thus, the Cot1^--G phenotype is not due to a modified 35S::GL1 4-1 transgene, but the phenotype requires the overexpression of GL1. The strong Cot1^--G phenotype, defined as cotyledons having at least 20 trichomes, segregated as 1 (strong Cot1^--G):15 (Cot1^--G weak plus wild type) in each F₂ population (Table 2). These data are consistent with cot1 and 35S::GL1 being unlinked and for the requirement of homozygosity at both loci in order to observe the strong Cot1^--G phenotype.

In the F₂ populations that were examined, many seedlings had cotyledon trichome numbers that were intermediate to 35S::GL1 and strong Cot1^--G plants. The weak Cot1^--G phenotype is defined as plants that have between five and 19 cotyledon trichomes. To better understand the relationship between cot1, 35S::GL1, and the severity of the Cot1^--G phenotype, the segregation of 35S::GL1, wild type, and Cot1^--G was analyzed in an F₂ population (Table 2, cross 3). If cot1 is recessive, unlinked, and synthetic with the 35S::GL1 transgene, a 9 (35S::GL1):4 (wild type):3 (strong Cot1^--G plus weak Cot1^--G) ratio is expected. However, in the analysis of one F₂ population, 182 35S::GL1, 56 wild-type, and 28 Cot1^--G plants were observed. These data do not conform to the hypothesized inheritance of 150 35S::GL1, 66 wild type, and 50 Cot1^--G (² = 18; P < 0.001). The wild-type phenotype did segregate 1:3 in this population as expected (² = 2.0; P > 0.1), and the departure from expected values was due to a depletion of the Cot1^--G category.

To better characterize the genetic contribution to variation in cotyledon trichome number, the number of cotyledon trichomes was analyzed in several families derived from a genetically defined parent. A limited number of cotyledon trichomes was observed in the presence of 35S::GL1 alone (Figure 4A). The cotyledon trichome numbers ranged from zero to three, with most plants lacking cotyledon trichomes. In the cot1 35S::GL1 double homozygote, the cotyledon trichome numbers were variable and ranged from 13 to 44 (Figure 4B). In the cot1/cot1 35S::GL1/-derived population, cotyledon trichome numbers ranged from zero to 32 (Figure 4C). The severity of the Cot1-G phenotype in this population segregated 1 (wild type):2 (intermediate Cot1-G):1 (strong Cot1-G), and the differences in cotyledon trichome number in Figure 4C were consistent with 35S::GL1 gene dosage affecting cotyledon trichome number. Seeds were collected from 12 F₃ individuals indicated in Figure 4C that had between six and nine trichomes. Progeny testing showed that 11 of 12 F₃s were heterozygous for 35S::GL1, and the other was homozygous for the transgene (data not shown). In the COT1/cot1 35S::GL1/35S::GL1 derived population, cotyledon trichome number ranged from zero to 39 (Figure 4D). Assuming that the zero cotyledon trichome category corresponded to 35S::GL1/35S::GL1 COT1/COT1, the severity of the Cot1-G phenotype segregated 1 (wild type):2 (intermediate Cot1-G):1 (strong Cot1-G). Despite significant overlap with the cotyledon trichome density observed with 35S::GL1 alone, the distribution shown in Figure 4D suggested that cot1 dosage also affected cotyledon trichome number.

The above results indicate that the expressivity of the Cot1^--G phenotype depends on the dosage of both 35S::GL1 and cot1. The strongest Cot1^--G phenotype is only seen in individuals homozygous for both cot1 and 35S::GL1. The inheritance of the strong Cot1^--G phenotype in segregating F₂ populations was also consistent with a requirement for homozygosity at cot1 and 35S::GL1 (Table 2). While enhanced cotyledon trichomes could be detected when either or both 35S::GL1 and cot1 were heterozygous, the phenotype (as defined by five or more cotyledon trichomes) was not always expressed in these individuals. This is clearly seen in Figure 4C, in which six of the 78 individuals with a 35S::GL1 leaf phenotype had fewer than five cotyledon trichomes even though the genetic data indicated that the family was homozygous for cot1. This lack of penetrance likely accounts for the skewed ratio of phenotypes in the F₂ population derived from a cot1 35S::GL1 double heterozygote.

An examination of the genotype of F₂ Cot1^--G plants derived from a cross between a wild-type Col plant and a severe Cot1^--G plant also suggested a gene dosage dependency of the Cot1^--G phenotype. Of the 42 selected F₂ individuals, 22 were strong Cot1^--G and 20 had between five and 10 cotyledon trichomes. Progeny testing of the 22 strong Cot1^--G F₂ plants indicated that all of these lines were homozygous for 35S::GL1 and cot1. Eight of the 20 F₃ families derived from F₂s with fewer cotyledon trichomes contained only 35S::GL1 and Cot1^--G plants and were therefore homozygous for 35S::GL1 and segregating for cot1. Ten other F₃ families were segregating for wildtype, 35S::GL1, and Cot1^--G, which indicated that the F₂ parents were heterozygous for both genes. Two F₃ families contained only Cot1^--G and wild-type plants, and, as such, were derived from cot1/cot1 35S::GL1/- individuals. Thus, the Cot1^--G phenotype can be observed in all genetic combinations of cot1 and 35S::GL1 in which at least one allele is cot1 and one copy of 35S::GL1 is present.

cot1 homozygous mutants do not have a phenotype in wild-type background:
The wild-type members in the F₃ families derived from cot1/cot1 35S::GL1/- parents should all be cot1/cot1-/-. The possibility that these cot1 homozygotes have an epidermal phenotype in the absence of the 35S::GL1 transgene was examined. Wild-type plants from two of these F₃ families were analyzed for cotyledon and leaf trichome number, and neither differed from wild type (Table 1). SEM indicated that the adaxial surface of cot1/cot1 cotyledons and leaves did not differ from wild type (data not shown).

Mapping the 35S::GL1 transgene and cot1:
To establish the chromosomal location of COT1, it was first necessary to identify the location of the 35S::GL1 4-1 transgene that cosegregated with the Cot1^--G phenotype. Analysis of previous genetic crosses suggested that this transgene may be linked to GL2 on Arabidopsis chromosome 1 (J. LARKIN, unpublished results). So the 35S::GL1 locus was mapped independently of cot1 by crossing unmutagenized 35S::GL1 4-1 with gl2-3 and allowing the F₁ to self. The 35S::GL1 4-1 transgene and gl2-3 allele are in the Col and WS backgrounds, respectively, which harbor abundant polymorphisms that have been genetically mapped. The resulting F₂ population was scored using as visible markers Gl2^-, the 35S::GL1 trichome phenotype, and kanamycin resistance, which is physically linked to 35S::GL1. Strong linkage was detected between 35S::GL1 4-1 and GL2 (Table 3). GL2 is located at position 115.9 on chromosome 1. Because 35S::GL1 is a dominant transgene over wild type, wild-type and Gl2^- plants in the F₂ population lack Col DNA at the 35S::GL1 locus. DNA was isolated from 23 wild-type or Gl2^- plants, and the SSLP markers nga280 and nga111 were used to confirm linkage with chromosome 1. Three-point linkage analysis (LANDER et al. 1987 ) placed 35S::GL1 4-1 between nga280 and nga111.

The chromosome 1 segment shown to contain the 35S::GL1 transgene also cosegregated with the Cot1^--G phenotype in an F₂ mapping population derived from a cross between cot1/cot1 35S::GL1/35S::GL1 (Col) x Nd-O/Nd-O (Table 3). From this population, DNA was isolated from 44 Cot1^--G seedlings whose cotyledons had at least 20 trichomes; these plants were putatively homozygous for 35S::GL1 and cot1. Through the use of the SSLP markers nga280 and nga111, we found that a Col chromosome 1 segment from 25.6 cM south of nga280 and 18.6 cM north of nga111 cosegregated with the Cot1^--G phenotype (Figure 5). The position of this region is in agreement with the map position of 35S::GL1 4-1 defined in the gl2-3 35S::GL1 4-1 mapping population.

The cot1 locus was mapped using the same F₂ population. Strong linkage was detected with chromosome 2 using a combination of SSLP and CAPS markers (Table 3). No significant linkage was detected with chromosomes 3, 4, or 5 (data not shown). Three-point linkage analysis of the mapping data placed cot1 15.3 cM north of er (Figure 5). The calculated map was 4.8 x 10⁴ times more likely than any other map that interrupted the interval COT1 to ER with one of the other markers.

Molecular markers of Cot1^--G trichomes:
To begin to address the molecular mechanisms of ectopic trichome initiation in Cot1^--G, a transgene containing a GLABRA2 promoter fragment fused to ß-glucuronidase (GL2::GUS) was crossed into the mutant background, and its expression pattern was analyzed in a segregating F₂ population (MASUCCI et al. 1996 ; SZYMANSKI et al. 1998 ). GL2::GUS is first expressed in fields of protodermal cells containing initiating trichomes. As the leaf matures, expression becomes limited to trichomes and is detected throughout their development (SZYMANSKI et al. 1998 ). A similar GL2::GUS expression pattern is observed in 35S::GL1. The expression of the GL2::GUS transgene in the developing leaves of 35S::GL1 and cot1 35S::GL1 homozygotes was compared (Figure 6). 35S::GL1 plants were stained histochemically for GUS activity when the first leaf pair reached a length of ~2 mm. At this stage of leaf development, the GL2::GUS activity in 35S::GL1 was detected primarily in trichomes (Figure 6A); however, weak diffuse staining could be detected in some areas along the leaf margin after 24 hr of histochemical staining (data not shown). A closer examination of the margin of 35S::GL1 leaves showed that GL2::GUS could be detected in several trichomes (Figure 6B), and that the support cells of 35S::GL1 leaves at this stage of development often did not display detectable GUS activity (Figure 6C). An analysis of Cot1^--G leaves at a similar developmental stage consistently detected GL2::GUS staining in trichomes and near the leaf margin (Figure 6D). Higher magnification of Cot1^--G leaves revealed persistent GL2::GUS expression in epidermal cells along the leaf margin (Figure 6E). Support cell (Figure 6F) and pavement cell (Figure 6G) trichomes showed elevated GL2::GUS transcription relative to adjacent cells. GL2::GUS activity was not detected in cotyledon and hypocotyl trichomes in plants at the four-leaf stage (Figure 6D); nevertheless, GL2 transcription in hypocotyl and cotyledon trichomes was easily detected in seedlings 3 days after germination (Figure 6H). GL2::GUS staining was only faintly detected in spiked cotyledon trichomes compared to branched cells.

Interaction between cot1 and try:
Trichome initiation in Cot1^--G cotyledons was enhanced compared to what was observed with 35S::GL1 alone. The try mutation also interacts with the 35S::GL1 transgene to enhance cotyledon trichomes. An analysis of 13 F₃ populations that were homozygous for try and segregating for 35S::GL1 indicated that 35S::GL1 try/try plants initiate cotyledon trichomes with a mean number of 11 ± 3.2 per cotyledon. No cotyledon trichomes were observed in try plants in the absence of 35S::GL1. Most cotyledon trichomes on 35S::GL1 try/try plants were branched, but as found on Cot1^--G plants, most trichomes were restricted to the cotyledon perimeter (Figure 7A). The interaction between cot1 and try was examined in an F₂ population segregating for cot1, try, and 35S::GL1. A unique phenotype was identified in the F₂. All of these individuals had at least one copy of the 35S::GL1 transgene and had massive clusters of trichomes covering much of the adaxial cotyledon surface (Figure 7B). To eliminate the dosage effects of 35S::GL1, the segregation of the novel phenotype was analyzed in an F₃ population derived from a 35S::GL1/35S::GL1 COT1/cot1 TRY/try F₂ plant. The inheritance of the novel phenotype was 12 (super cot):163 (intermediate to 35S::GL1 phenotype). This conformed very well to the 11 (super cot):164 (intermediate to 35S::GL1 phenotypes) that would be expected if the enhanced phenotype required homozygosity at cot1, try (² = 0.09, P > 0.9). Cotyledon trichome density was so high that accurate counts could only be obtained using SEM. Using the same criterion for cotyledon trichomes, the putative double mutants had 64 ± 4.4 (N = 11) trichomes per cotyledon. This is significantly greater than a cotyledon trichome number of 39 ± 8.8 that would be expected if the effects of cot1 and try were additive.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic studies have shown that the GL1 and TTG genes are required for trichome initiation. The competency of a leaf epidermal cell to enter the trichome pathway is under complex regulation and is likely to involve the activity of many genes. Overexpression of GL1 is not sufficient for widespread trichome initiation. Decreased TTG gene dosage and the try mutation alter the distribution and increase the number of leaf trichomes in the 35S::GL1 background. In the presence of 35S::GL1, try also gives rise to ectopic trichomes on the cotyledon and hypocotyl. These observations suggest that TTG and try are required to inhibit trichome initiation when GL1 is not limiting. The COT1 locus is a gene that was identified in a screen for additional mutations that interact with the 35S::GL1 transgene and alter trichome initiation.

COT1 maps to a distinct region 15.3 map units north of er on chromosome 2. The cot1-dependent phenotype is synthetic with 35S::GL1. In 35S::GL1 plants, cot1 enhances trichome initiation in the hypocotyl, cotyledon, and limited regions of the leaf, whereas cot1 in a wild-type background produces no detectable phenotype. Homozygosity of cot1 and the 35S::GL1 transgene leads to a dramatic increase in cotyledon trichome initiation. The expressivity of the Cot1^--G phenotype is variable but is positively correlated with cot1 and 35S::GL1 gene dosage. This is similar to the 35S::GL1 dosage dependence inhibition of trichome initiation on the stem, but differs from the dominant inheritance of the 35S::GL1 dependent leaf and cotyledon phenotype (LARKIN et al. 1994 ). It is possible that COT1 and GL1 gene dosage regulates trichome initiation, and the concentration of both proteins affects the probability that a given cell enters the trichome pathway. The semidominant behavior of cot1 complicates the interpretation of the function of the wild-type gene, and many models are consistent with the genetic data. If the cot1 allele is a loss-of-function mutation, COT1 may inhibit initiation in differentiated epidermal cells. Inhibition in nontrichome cells could be mediated by stable repression of GL1-regulated promoters of trichome-specific genes such as GL2 or by indirect effects on parallel pathways that control the competency of a cell to enter the trichome pathway. Because 35S::GL1 is required for the Cot1^--G phenotype, COT1 might repress trichome initiation when GL1 protein is not limited. Functional redundancy or the tight control of GL1 expression may mask the effects of the cot1 mutation in wild-type plants. Similar functional redundancy has been observed with the APETALA1 and CAULIFLOWER genes that regulate cell fate during floral development (BOWMAN et al. 1993 ). It is also possible that cot1 is a gain-of-function mutation and increases the probability that a cell enters the trichome pathway. Similar arguments regarding direct or indirect activation of trichome initiation would apply. The isolation of additional alleles of cot1 and additional mutants involved in trichome spacing control should clarify this issue.

The ectopic cotyledon trichomes in Cot1^--G plants display elevated GL2::GUS transcription relative to surrounding cells and apparently use elements of normal trichome transcription control. Unlike the persistent GL2 transcription detected in leaf trichomes, GL2::GUS transcription is only transiently detected in spiked cotyledon trichomes. Cot1^--G cotyledon trichome structure resembles that of gl2 trichomes (RERIE et al. 1994 ); most elongated cells are unbranched and rarely acquire a papillate surface. This phenotype would be expected if mistimed or inadequate GL2 expression leads to trichome structure defects.

The presence of trichomes on the cotyledon of Cot1^--G plants does not appear to mimic previously identified leafy cotyledon (lec) mutations that transform cotyledons into leaf-like organs (MEINKE et al. 1994 ; WEST et al. 1994 ). The cotyledon trichomes on lec1 plants develop during embryogenesis, whereas Cot1^--G cotyledon trichomes appear postgermination. In addition, cot1 and lec1 map to different linkage groups. In Cot1^--G plants, cotyledon vasculature, shape, and adaxial epidermis appear normal. Therefore, the ectopic expression of GL1 appears to allow a subset of cotyledon epidermal cells to enter the trichome morphogenesis pathway rather than affect organ identity, and cot1 enhances that effect.

Two subclasses of ectopic epidermal trichomes were found on Cot1^--G leaves: pavement cell and support cell trichomes. Pavement and support cell trichomes appeared after most trichome initiation was completed and apparently developed from differentiated epidermal cells. Pavement cell trichomes initiated late in leaf development and were first detected as small projections on irregularly shaped pavement cells. The surrounding pavement cells were adherent but lacked the characteristic support cell structure because they had already expanded to near wild-type size. Similarly, support cell trichomes developed from specialized support cells late in leaf development. The cot1 mutation did not alter potential cell-cell signaling that controls the initial support cell specification, but rather affects the persistent repression of trichome initiation in the cell type.

Because only a small fraction of epidermal cells enter the trichome pathway in Cot1^--G plants, positional information and the expression of factors besides GL1 and COT1 are likely to determine the competency of a cell to enter the trichome pathway. Enhanced trichome initiation in the first leaf pair of Cot1^--G plants is limited to support cells and to regions near the leaf margin. This indicates that the cot1 mutation is not sufficient to overcome the 35S::GL1-dependent inhibition of trichome initiation in cells that give rise to the leaf mid-blade. This contrasts with the effect of try on the 35S::GL1 phenotype, in which enhanced initiation is not restricted to specific regions of the leaf (D. SZYMANSKI, unpublished results). It is possible that wild-type trichomes and cells along the leaf margin provide a locally acting trichome promoting signal and that additional leaf trichomes in Cot1^--G leaves result from the inability to attenuate that signal. The predisposition of support cells to enter the trichome pathway is consistent with this possibility. Alternatively, the spatial limitations of trichome initiation in the Cot1^--G leaf could reflect the stable competency of the population of cells that were present in the 100 to 400 µm stages of leaf development throughout which initiation occurs in 35S::GL1. Persistent GL2::GUS staining along the perimeter of Cot1^--G leaves could result from the inability of cot1 to limit GL2 transcription in the developing leaf.

Regardless of the mechanism of cot1 enhancement of trichome initiation, it is clear that a limited number of epidermal cells in Cot1^--G leaves enter the trichome pathway. The parameter that controls the late entry of specific epidermal cells is not known. The correlation between increased DNA content and Arabidopsis trichome morphogenesis has been made (MELARAGNO et al. 1993 ; HULSKAMP et al. 1994 ), and increased DNA content has also been positively correlated with age in Arabidopsis leaves (GALBRAITH et al. 1991 ). It is possible that factors that regulate DNA synthesis influence the timing and number of cells that enter the trichome pathway in Cot1^--G plants. The synergistic interaction between try and cot1 in the 35S::GL1 background is consistent with the idea that some component of DNA replication contributes to the severity of the Cot1^--G phenotype. The try mutation affects both trichome initiation and the DNA content of mature trichomes (HULSKAMP et al. 1994 ). Loss of try function may enhance the cot1 phenotype by removing a point of negative regulation of trichome initiation that is linked to DNA replication. The reduced seedling viability in putative cot1 try mutants in the 35S::GL1 background is similar to what is observed when both GL1 and the maize R gene are overexpressed and trichome initiation is globally enhanced in shoot-derived tissues (LARKIN et al. 1994 ). It is not known if cot1 and try operate within the same or in parallel pathways. A thorough examination of the relationship between try, 35S::GL1, and trichome initiation is in progress.

The identification and characterization of the cot1 locus begin to clarify some of the mechanisms that control epidermal cell fate in Arabidopsis. Overexpression of GL1 is not sufficient for widespread trichome initiation, and it appears that cot1 is another gene that regulates trichome initiation. The spatial and temporal restrictions of Cot1^--G trichome initiation suggest that many genes regulate competency, and additional genes that positively and negatively regulate trichome initiation have yet to be identified. Identification and characterization of the cot1 gene will contribute to understanding how epidermal cell fate is controlled.

	FOOTNOTES

¹ Present address: Hughes Institute, St. Paul, MN 55108.
² Present address: Botany Department, Louisiana State University, Baton Rouge, LA 70803.

	ACKNOWLEDGMENTS

Thanks to BOB HERMAN, MIKE SIMMONS, BETH KENT, JASON HILL, and SCOTT SATTLER for critical comments on the manuscript, and SUSAN POLLOCK for editorial help. The University of Minnesota Imaging and Microscopy Consortium provided excellent technical support. This research was supported by National Science Foundation grant NSF/IBN-9506192 to M.D.M.

Manuscript received January 12, 1998; Accepted for publication March 4, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AHLSTRAND, G. G., 1996 Low-temperature low-voltage scanning microscopy (LTLVSEM) of uncoated frozen biological materials: a simple alternative, p. 918 in Proceedings of Microscopy Microanalysis, 1996, edited by G. W. BAILEY, J. M. CORBETT, R. V. M. DIMLICH, J. R. MICHAEL and N. J. ZALUZEC. San Francisco Press, San Francisco.

BELL, C. J. and J. R. ECKER, 1994  Assignment of 30 microsatellite loci to the linkage map of Arabidopsis. Genomics 19:137-144[Medline].

BOWMAN, J. L., J. ALVAREZ, D. WEIGEL, E. M. MEYEROWITZ, and D. R. SMYTH, 1993  Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development 119:721-743[Abstract].

FELDMANN, K. A. and M. D. MARKS, 1987  Agrobacterium-mediated transformation of germinating seeds of Arabidopsis thaliana: a non-tissue culture approach. Mol. Gen. Genet. 208:1-9.

FOLKERS, U., J. BERGER, and M. HÜLSKAMP, 1997  Cell morphogenesis of trichomes in Arabidopsis: differential control of primary and secondary branching by branch initiation regulators and cell growth. Development 124:3779-3786[Abstract].

GALBRAITH, D. W., K. R. HARKINS, and S. KNAPP, 1991  Systemic endopolyploidy in Arabidopsis thaliana. Plant Physiol. 96:985-989[Abstract/Free Full Text].

HAUGHN, G. W. and C. R. SOMERVILLE, 1988  Genetic control of morphogenesis in Arabidopsis. Dev. Genet. 9:73-89.

HÜLSKAMP, M., S. MISÉRA, and G. JÜRGENS, 1994  Genetic dissection of trichome cell development in Arabidopsis. Cell 76:555-566[Medline].

JOHNSON, H. B., 1975  Plant pubescence: an ecological perspective. Bot. Rev. 41:233-258.

KONIECZNY, A. and F. AUSUBEL, 1993  A procedure for mapping Arabidopsis mutations using co-dominant ecotype-specific PCR-based markers. Plant J. 4:403-410[Medline].

KOORNNEEF, M., 1981  The complex syndrome of ttg mutants. Arab. Inf. Serv. 18:45-51.

LANDER, E. S., P. GREEN, J. ABRAHAMSON, A. BARLOW, and M. DALY et al., 1987  MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174-181[Medline].

LARKIN, J. C., D. G. OPPENHEIMER, S. POLLOCK, and M. D. MARKS, 1993  Arabidopsis GLABROUS1 gene requires downstream sequences for function. Plant Cell 8:513-524.

LARKIN, J. C., D. G. OPPENHEIMER, A. LLOYD, E. T. PAPAROZZI, and M. D. MARKS, 1994  The roles of GLABROUS1 and TRANSPARENT TESTA GLABRA genes in Arabidopis trichome development. Plant Cell 6:1065-1076[Abstract].

LARKIN, J. C., N. YOUNG, M. PRIGGE, and M. D. MARKS, 1996  The control of trichome spacing and number in Arabidopsis. Development 122:997-1005[Abstract].

LARKIN, J. C., M. D. MARKS, J. NADEAU, and F. SACK, 1997  Epidermal cell fate and patterning in leaves. Plant Cell 9:1109-1120[Medline].

LEVIN, D. A., 1973  The role of trichomes in plant defense. Q. Rev. Biol. 48:3-15.

LLOYD, A., M. SCHENA, V. WALBOT, and R. W. DAVIS, 1994  Epidermal cell fate determination in Arabidopsis: patterns defined by a steroid-inducible regulator. Science 266:436-439[Abstract/Free Full Text].

MARKS, M. D., 1997  Molecular genetic analysis of trichome development in Arabidopsis. Annu. Rev. Plant Physiol. Plant Mol. Biol. 48:137-163.

MARKS, M. D., J. ESCH, P. HERMAN, S. SIVAKUMARAN and D. OPPENHEIMER, 1991 A model for cell-type determination and differentiation in plants, pp. 77–87 in Molecular Biology of Plant Development, edited by G. JENKINS and W. SCHUCH. Company of Biologists Limited, Cambridge, UK.

MASUCCI, J. D., W. G. RERIE, D. R. FOREMAN, M. ZHANG, and M. E. GALWAY et al., 1996  The homeobox gene GLABRA2 is required for position-dependent cell differentiation in the root epidermis of Arabidopsis thaliana. Development 122:1253-1260[Abstract].

MEINKE, D. W., L. H. FRANZMANN, T. C. NICKLE, and E. C. YEUNG, 1994  Leafy Cotyledon mutants of Arabidopsis. Plant Cell 6:1049-1064[Abstract].

MELARAGNO, J. E., B. MEHROTA, and A. W. COLEMAN, 1993  Relationship between endoploidy and cell size in epidermal tissue of Arabidopsis. Plant Cell 5:1661-1668[Abstract].

OPPENHEIMER, D. G., P. L. HERMAN, J. ESCH, S. SIVAKUMARAN, and M. D. MARKS, 1991  A myb-related gene required for leaf trichome differentiation in Arabidopsis is expressed in stipules. Cell 67:483-493[Medline].

PYKE, K. A., J. L. MARRISON, and R. M. LEECH, 1991  Temporal and spatial development of the cells of the expanding first leaf of Arabidopsis thaliana (L.) Heynh. J. Exp. Bot. 42:1407-1416[Abstract/Free Full Text].

RERIE, W. G., K. A. FELDMANN, and M. D. MARKS, 1994  The GLABRA2 gene encodes a homeo domain protein required for normal trichome development in Arabidopsis. Genes Dev. 8:1388-1399[Abstract/Free Full Text].

ROLLINS, R. C., and U. C. BANERJEE, 1976 Trichomes in studies of the Cruciferae, pp. 145–166 in The Biology and Chemistry of the Cruciferae, edited by J. G. VAUGHN, A. J. MACLEOD and B. M. G. JONES. Academic Press, New York.

SZYMANSKI, D. B., R. A. JILK, S. POLLOCK, and M. D. MARKS, 1998  Control of GL2 expression in Arabidopsis leaves and trichomes. Development in press.

WEST, M. A. L., K. M. YEE, J. DANAO, J. L. ZIMMERMAN, and R. L. FISCHER et al., 1994  Leafy Cotyledon is an essential regulator of late embryogenesis and cotyledon identity in Arabidopsis. Plant Cell 6:1731-1745[Abstract/Free Full Text].

This article has been cited by other articles:

				A. Schnittger, U. Folkers, B. Schwab, G. Jürgens, and M. Hülskamp Generation of a Spacing Pattern: The Role of TRIPTYCHON in Trichome Patterning in Arabidopsis PLANT CELL, June 1, 1999; 11(6): 1105 - 1116. [Abstract] [Full Text]

				D Luo and D. Oppenheimer Genetic control of trichome branch number in Arabidopsis: the roles of the FURCA loci Development, January 12, 1999; 126(24): 5547 - 5557. [Abstract] [PDF]

				D. B. Szymanski and M. D. Marks GLABROUS1 Overexpression and TRIPTYCHON Alter the Cell Cycle and Trichome Cell Fate in Arabidopsis PLANT CELL, December 1, 1998; 10(12): 2047 - 2062. [Abstract] [Full Text]