LOP1: a gene involved in auxin transport and vascular patterning in Arabidopsis

The role of translocatable signal molecules in plant development was first proposed by Darwin in the nineteenth century as an explanation for phototropism (Lomax et al., 1995). It was later demonstrated that this capacity for directional growth in higher plants was indeed governed by lateral redistribution of indole-3-acetic acid (IAA), a naturally occurring plant hormone in the auxin family. Auxin has since been shown to influence other aspects of plant development, most notably the differentiation of vascular tissue. A large body of evidence shows that auxin synthesized by leaf primordia is an inductive signal for vascular differentiation in the shoot (Shininger, 1979). Surgical manipulation of the shoot apex revealed that leaf primordia are essential for basipetal development of primary xylem (Young, 1954; Wangermann, 1967). The inductive effect of the leaf can be replaced by application of auxin or with auxin and sucrose (Young, 1954; Jacobs and Morrow, 1957). Similarly, shoot tips grafted to undifferentiated callus induce xylem formation, and the response is mimicked by application of auxin (Wetmore and Rier, 1963). Auxin appears to play a similar role in differentiation of the primary phloem (LaMotte and Jacobs, 1963). Manipulation of IAA levels in transgenic plants also supports a primary role for this hormone in vascular differentiation. Transgenic petunia plants with ten-fold overproduction of IAA show an increase in production of secondary xylem and phloem (Klee et al., 1987), and conversely, auxin inactivation in transgenic tobacco produces an inhibition of vascular differentiation (Romano et al., 1991).

Although it is clear that auxin plays an essential role in formation of vascular tissue, the mechanisms for spatial regulation of this process and ultimate elaboration of reproducible venation patterns in the leaf remain largely unexplored. Our current insights have arisen mostly from experiments involving redifferentiation of xylem elements in plant stems following application of IAA (Sachs, 1991). This work demonstrates that experimental redirection of auxin flow around wound sites produces a reorientation in the polarity of cell elongation and a correlated change in the path of vascular differentiation. This is thought to begin with formation of specialized, elongated cell files for basipetal auxin transport in a process known as ‘canalization’. According to the canalization hypothesis (Sachs, 1991), formation of a specialized path for auxin flow leads directly to axial elongation of transporting cells and their subsequent differentiation into vascular elements. The molecular basis for basipetal auxin transport is thought to involve establishment of an intracellular asymmetry for membrane-localized protein carriers with specific roles in IAA uptake and efflux (Lomax et al., 1995). Localization of IAA efflux carriers at the basipetal end of the cell is considered to be the primary mechanism establishing the polarity of auxin transport.

Identification of genes involved in the positioning of these specialized channels for auxin transport will be an important step toward understanding the molecular basis of vascular patterning. Mutations at the PIN-FORMED locus on chromosome 1 in Arabidopsis produce a disruption in IAA transport (14% of wild-type) and a deficiency in levels of free IAA (8% of wild-type) (Okada et al., 1991). Mutant pin1 plants have an abnormal spike inflorescence lacking floral buds and fused cotyledons, suggesting that this genes plays a role in flower initiation and during embryogenesis. Mutant leaves also show split midveins, supporting a role for this gene in leaf vascular patterning. In the absence of any other mutations in the transport pathway, however, our knowledge of the process remains quite rudimentary.

We have initiated a mutational analysis in Arabidopsis aimed at understanding the molecular basis for vascular patterning in dicot leaves. Here we report the recovery and characterization of a recessive mutation (lop1) that produces a major defect in basipetal transport of IAA. Mutant midveins show disoriented axial growth, and bifurcation into twin veins that are frequently rotated out of the normal dorsal/ventral axis of the leaf. In addition mutants show abnormal patterns of cell expansion in the midrib cortex and in the root epidermis that are consistent with a disruption in IAA transport.

The T-DNA collection, the ecotype Landsberg erecta, and the axr1-3 and axr1-12 mutants were obtained from the Arabidopsis Biological Resource Center (ABRC, Ohio State). Plants were grown in Metro Mix 200 (Grace) at 22°C and 60-70% relative humidity with 16 hours illumination. For plating seed on media, the seeds were surface sterilized by immersion in 20% hypochlorite/0.05% Tween followed by several rinses with sterile water. The seeds were allowed to air dry and then plated on an MSG medium (Murashige and Skoog, 1962) at 0.5× major salts, 1% sucrose, and no hormones. Plants were grown at 27°C under continuous illumination. The T-DNA collection was screened from plants grown in Metro Mix as described above.

A homozygous lop1 plant (ecotype Ws) was used as the pistillate parent in a cross with wild-type Landsberg erecta. Seed was collected from individual F1 plants yielding a polymorphic F2 population. Arabidopsis genomic DNA was isolated from individual F2 plants grown in soil or from pools of >20 individuals in an F3 family grown on MS plates. All wild-type F2 plants (LOP1/LOP1 and LOP1/lop1) were progeny tested providing an informative population for mapping purposes. Because lop1 is male sterile, the F3 pools consisted of plants either homozygous or heterozygous for the wild-type allele.

For PCR analysis, DNA was isolated using the method of Dellaporta et al. (1994). Amplification was performed in a Perkin Elmer thermal cycler as described by Bell and Ecker (1994) although more consistent reactions were obtained when approximately 25 ng of DNA was used in the reaction. For nga76, the annealing temperature was lowered from 55°C to 50°C. PCR products were resolved by electrophoresis on either 2% ultrapure agarose (Gibco BRL)/2% Nusieve, 1× TAE or 3-4% Metaphor (FMC, Rockland, ME) in 1× TBE (89 mM Tris borate, 2 mM EDTA) depending on the size of the different alleles. Microsatellite primers were obtained from Research Genetics (Huntsville, AL). For Southern blot analysis, plant DNA was isolated as above except extractions with phenol followed by phenol/choloroform/isoamyl alcohol (50:48:2) were included prior to precipitation of DNA with isopropanol. Standard methods were used for plasmid and cosmid DNA preparations, Southern blots and hybridizations with probes labeled by random priming (Sambrook et al., 1989). Chromosome five RFLP markers were obtained from the ABRC (Ohio State). The cosmid RFLP marker 558A had been subcloned and converted to an ARMS marker (Fabri and Schaffner, 1994) providing a marker in plasmid form. For initial mapping studies using the microsatellite primers, a population consisting of 29 plants (58 chromosomes) was analyzed. For further localization of lop1 using additional chromosome 5 markers, a population consisting of 46 plants (92 chromosomes) was examined.

Multipoint linkage analysis was performed using a Macintosh version of MAPMAKER (version VI, Lander et al., 1987). Recombination frequencies from multipoint analysis were converted into map distances (centiMorgans) using the Kosambi function (Kosambi, 1944) and integrated with the existing molecular genetic map (Hauge et al. 1993).

Mutant strains homozygous for the axr1-3 and axr1-12 mutations were employed in a double mutant analysis. Under our growth conditions, both axr1 mutants had similar phenotypes. Both mutants were hybridized with LOP1/lop1 heterozygotes. The analysis of lop1, axr1-3 and lop1, axr1-12 F2 progenies was conducted separately, but gave similar results and the data were subsequently pooled. Double het-erozygotes were identified as F1 individuals segregating the axr1 and lop1 mutations, along with a rare novel phenotype. The frequency of F2 phenotypic classes was scored in MSG plates and subjected to χ2 analysis.

Plant materials were prepared for sectioning by overnight fixation in ethanol-chloroform-glacial acetic acid (6:3:1 v/v), followed by stepwise infiltration first with butanol and then with melted paraffin at 60°C in a vacuum oven. Sections (8 μm) were cut from paraffin blocks, fixed to glass slides and dried overnight prior to staining with hemalum and safranin as previously described (McHale, 1993). Wild-type and mutant leaves were cleared by fixation in 70% ethanol and dehydration in 100% ethanol, followed by overnight incubation in saturated chloral hydrate.

Samples were prepared for scanning electron microscopy by overnight fixation in ethanol-chloroform-glacial acetic acid (6:3:1) that was followed by stepwise transfer to 100% ethanol, and critical point drying with liquid carbon dioxide in a Polaron pressure chamber. Dried samples were mounted on stubs, sputter coated, and examined with an ISI-SS40 scanning electron microscope.

To compare levels of free IAA in mutant and wild-type plants, we grew an F3 family segregating the lop1 mutation on MS medium with 2% sucrose in Petri plates, and harvested the aerial portion of mutants and normals at 11 days postgermination. Triplicate samples were frozen in liquid N2 and ground in a mortar (65% isopropanol in 0.2 M imidazole, pH 7.0). Samples were spiked with 100 ng [13C6]IAA and [3H]IAA (26 Ci/mmol), diluted to 10% isopropanol, applied to amino ion exchange columns (PrepSep 0.3g Fisher), washed (sequen-tially with 2 ml hexane, ethylacetate, acetonitrile, methanol), eluted with 3 ml methanol/2% acetic acid, dried and resuspended in 100 μl 50% methanol for HPLC (Phenomenex, UltraCarb 30), using 20% methanol in H2O/1% acetic acid for elution. Samples were assayed by mass spectrometry by Dr Jennifer Normanly (Department of Bio-chemistry and Molecular Biology, University of Massachusetts) by the method of Chen et al. (1988).

Auxin response was assayed by measuring root growth of mutant and wild-type seedlings in the presence of exogenous 2,4D at 0.5, 1, 3 and 6×10−8 M. An F3 family segregating the lop1 mutation was grown for 6 days on an MS medium with 2% sucrose. Mutant and wild-type siblings were transferred separately to media containing 2,4D, and to MS control plates with no 2,4D. The root system was positioned on the agar surface to allow measurement of new growth (downward) at daily intervals up to 12 days postgermination.

Auxin transport was assayed in excised inflorescence stems of soil-grown mutant and wild-type siblings as described by Okada et al. (1991). The [14C]IAA was purchased from the Sigma Chemical Company, St. Louis, MO.

Our strategy for the genetic analysis was to isolate putative vascular mutants generated by T-DNA insertion. We screened 100,000 plants in pools of T4 seed (Feldmann collection: CS 2606-2654 from ABRC) representing 5,000 independent T-DNA insertion events, and self-pollinated or outcrossed the aberrant leaf variants for further analysis. One variant with normal cotyledons and narrow, deformed leaves was male-sterile, but outcrossed as pistillate parent to wild-type Landsberg erecta. Pooled data from three independent F2 families showed segregation ratios conforming to the 3:1 pattern (272 normal: 90 mutant) expected for a single recessive mutation. Because mutant leaves displayed sporadic occurrence of bladeless regions along the midrib, the mutation was named lopped and the mutant allele designated lop1. Southern blot analysis using various T-DNA probes revealed that lop1 did not contain a T-DNA insert. Microsatellite primers were used to position lop1 on the molecular genetic map. lop1 was found to be linked to marker nga129 at the bottom of chromosome five, as shown in Fig. 1. Additional RFLP molecular markers (g4130, m558, and g3844) further localized lop1 to the position 83.0±2.0, as described in Materials and Methods.

To facilitate the comparison of mutant leaf phenotypes to wild-type, stages in the development of wild-type rosette leaves of Arabidopsis were assigned based on leaf length and appearance of distinct morphological landmarks, as described in Table 1 and shown in Fig. 2A-E.

At stage 1, leaf primordia are 0.05-0.1 mm in length and curved over the shoot apical meristem (Fig. 2A). At stage 2, primordia are 0.1-0.2 mm long and show an upright stature, and trichomes are initiated on the dorsal surface toward the tip of the blade (Fig. 2A). At stage 3 (0.2-0.5 mm), developed trichomes are visible at the leaf tip, and primary lobes of the blade are initiated (Fig. 2B). The midrib also becomes visible at this stage through expansion of ventral epidermal cells subtending the developing midvein, and elongated epidermal cells appear at the margin of the leaf blade. At stage 4 (0.5-1.0 mm), the tip of the leaf curves away from the shoot axis, and trichomes are observed from the tip to the base of the blade. Petiole elongation begins, and secondary and tertiary lobes (below the primary lobes) are initiated at the blade/petiole junction (Fig. 2C). During stage 5 (>1 mm, Fig. 2D,E) petiole elongation continues, and files of expanded ventral epidermis associated with the midrib extend from the leaf base upward through approximately 75% of the total leaf length.

Anatomical analysis of leaf midrib development was carried out by examination of serial transverse sections along the rib axis from tip to base in stage 4 leaves, as shown in Fig. 2F-I. Sections through the tip of the midvein show that the earliest evidence of rib development is vacuolation and expansion of ventral mesophyll cells subtending the vein (Fig. 2F). The upper mesophyll is unaffected at this stage, and there is no evidence of epidermal cell expansion. Protrusion of the midrib is first observed on the lower surface of the leaf (Fig. 2G). This process involves localized division and expansion of vein-proximal cells in the ventral cortical region, along with expansion of the subtending epidermis (Fig. 2E,G). The next stage in rib development is marked by a similar pattern of cortical development on the dorsal side of the vein (Fig. 2H). Vein proximal cells of the upper mesophyll show vacuolation and expansion, periclinal cell divisions have generated an additional layer of cortical cells, and epidermal cell expansion is visible. Sections through the region at the blade/petiole junction show that this developmental pattern persists in the later stages of rib development (Fig. 2I). The ventral cortical region shows 4-5 layers of vacuolated cells that account for protrusion of the rib on the underside of the leaf. Formation of 2-3 layers of dorsal cortex generates a visible rib on the upper surface as well. At this stage, vacuolated cortical cells are also observed on the lateral flanks of the midvein, generating a radial pattern of rib growth. Cell size increases with increasing distance from the vein in all directions, suggesting that addition of cortical layers involves division of the vein proximal cells.

Morphological and anatomical aberrations associated with the lop1 mutation are shown in Fig. 3. Mutant plants germinate normally and have normal cotyledons, but all of the true leaves are deformed and dwarfed (Fig. 3B) relative to wild-type (Fig. 3A). The placement and timing of leaf initiation is normal in the mutant, suggesting that the basic programs of the shoot meristem are not affected by the mutation. Mutant plants show a variety of morphological aberrations including twisting of the leaf petiole (Fig. 3E) and the style of the flowers (Fig. 3C). In extreme cases, twisting of the leaf petiole results in a 180° rotation of the leaf away from the shoot axis. Mutant leaves show disruptions in development of the midvein, including general disorientation in the direction of growth (Fig. 3D), and bifurcation into twin veins that fail to initiate the normal series of lateral branches (Fig. 3G). This is in contrast to wild-type leaves which show the typical netted venation of dicots, with a fixed set of main lateral branches joining a single central midvein (Fig. 3F). The lop1 mutation also produces abnormal development of the root system. In wild-type plants, lateral roots are initiated at right angles to the primary root, as shown in Fig. 3I. In contrast, lateral roots of the mutant show random patterns of curvature during initiation (Fig. 3J), and sometimes encircle the primary root before establishing growth in a normal downward direction.

The developmental morphology and anatomy of lop1 mutant leaves are shown in Fig. 4.

Stage 1 primordia have a normal morphology, but stage 2 primordia show emergence of marginal protrusions (Fig. 4A) that develop as independent growing points in the expanding blade (Fig. 4C). These protrusions typically emerge in the position for the primary lobes of the leaf, although well ahead of the stage when they normally appear in wild-type (stage 3, Fig. 2B). Newly initiated protrusions can be observed in the blade of stage 4 leaves, indicating that this process extends over several stages of leaf development (Fig. 4D). Early stages of midrib development appear normal in stage 3 leaves (Fig. 4B), but disorientation of midrib growth is evident by stage 4 (Fig. 4C). In mutant leaves at stage 5, there are two abnormal midrib phenotypes that are associated with abnormal vascular patterning. The midrib can develop as a centrally located oversized structure (Fig. 4E), or it can show bifurcation and develop as two distinct ribs located at the leaf margins (Fig. 4G). Trans-verse sections through the blade/petiole junction of these leaves is shown in Fig. 4F and H, respectively. In both cases, twin vascular regions are observed indicating a bifurcation of the developing midvein. The twins are usually dis-oriented and in some cases are rotated 90° relative to one another (Fig. 4F). In contrast, vascular tissue in a wild-type midvein shows a uniform alignment along a single dorsal/ventral axis in the leaf, with the xylem and phloem poles facing the dorsal and ventral surfaces, respectively (Fig. 2I).

The transverse section in Fig. 4I shows stage 2 and stage 3 wild-type leaves. The stage 2 leaf shows meristematic divisions throughout the mesophyll, and no evidence of cortical cell expansion. At stage 3, radial expansion is visible in the ventral cortical cells subtending the midvein. In contrast, the mutant stage 2 leaf in Fig. 4J shows precocious expansion of cortical cells on the ventral surface of the primordium. Similarly, the mutant stage 3 leaf shows a broad zone of precocious cortical expansion on the ventral side of the midrib. Median transverse sections through mutant leaves at stage 4 (Fig. 4K) frequently show a one-sided blade morphology that is associated with precocious expansion of cortical cells along the lateral flank (compare to median wild-type sections, Fig. 2G,H).

To quantitate patterns of cortical cell expansion in the midrib, cellular dimensions were measured in mutant leaves and matched wild-type controls, as shown in Table 2. Cortical cell number and size and overall rib dimensions were examined at progressive stages of rib formation (tip to base) in stage 4 and stage 5 leaves. The data show that development of the wild-type midrib cortex involves both cell division and cell expansion. In stage 5 leaves, there is a doubling of cortical cell layers (ventral) and a 2.5-fold increase in cortical cell size at the midrib base relative to the tip. Data from mutant leaves (stage 4 and 5) show that the tip of the midrib is significantly larger than the wild-type control both in the dorsal/ventral and the lateral axis. Cortical cell data show that cell number is unchanged relative to wild-type, and that the increase in rib size reflects an increase in cell size. The transect diameter of the largest cortical cell in the mutant leaves (20 μm) is double that of wild-type. A similar pattern is observed at the base of the midrib. Mutant ribs are larger than wild-type, again reflecting primarily an increase in cortical cell size. The number of cortical cell layers on the dorsal and ventral sides of the midvein is unaffected.

The lop1 mutation also produces abnormal patterns of cell expansion in epidermal cells of the root elongation zone, as shown in Fig. 5. In wild-type roots, expanding epidermal cells in the elongation zone remain aligned in distinct longitudinal files (Fig. 5A), whereas mutant roots show an irregular pattern of cell expansion in this region, and a deviation of cell files from their normal axial alignment into a spiralling pattern (Fig. 5B). This pattern persists in the elongation zone, ultimately generating lateral roots with a twisted morphology (Fig. 5D). During subsequent maturation, mutant roots become flattened and oblong in transverse view, and frequently show two parallel vascular systems (Fig. 5F). This appears to result from formation of new lateral roots that grow initially along the same axis as the primary root before emerging as a separate structures. In contrast, wild-type roots retain their cylindrical shape during maturation, and initiate new lateral roots at right angles to the primary axis (Fig. 5C,E).

Initially it seemed possible that the mutation was affecting the ethylene pathway, as this hormone can affect radial cell expansion (Abeles et al., 1992), and xylem differentiation (Aloni, 1987). Since ethylene overproducing mutants of Ara-bidopsis (eto1; Guzman and Ecker, 1990) show the well-known triple response phenotype (exaggerated apical hook, stunting, radial swelling), we examined hook and hypocotyl phenotypes in dark-grown lop1 mutants and wild-type siblings, but found no observable difference (not shown). In addition, the ethylene inhibitors AgNO3 and alpha amino isobutyric acid which abolish the triple response of eto1 did not affect the lop1 mutant phenotype. Thus it appeared from preliminary assays that the lop1 phenotype was not the result of ethylene overproduction.

Our evidence for disruptions in vascular patterning and cell expansion in mutant plants raised the possibility that lop1 affects a step in the auxin pathway. To determine whether the mutation affects endogenous levels of auxin, we assayed free IAA in mutant seedlings and wild-type controls at 11 days post-germination by GC-SIM-MS (Chen et al., 1988). Free IAA levels were virtually identical (1.25±0.09 ng IAA/mgfwt) in triplicate samples of mutant and wild-type seedlings. The majority of IAA in plants is conjugated to sugars, peptides, or amino acids by ester or amide linkages, and gradually released as free IAA, which is generally assumed to be the biologically active form (Normanly et al., 1995). Our results do not exclude the possibility that lop1 is affecting the size or dynamics of these conjugated pools, but this would be expected to produce some alteration in levels of free IAA. A highly sensitive IAA assay produced no evidence for this, suggesting that lop1 is not a biosynthesis mutant.

We also followed a genetic approach to this question, taking advantage of available mutants of Arabidopsis that are defective in auxin response. Two Arabidopsis genes involved in auxin response (AXR1, AXR2) have been identified by selection for mutations conferring resistance to exogenous 2,4D (Estelle and Somerville, 1987; Lincoln et al., 1990; Timpte et al., 1994). Sequence analysis of the AXR1 gene shows that it encodes a protein related to the ubiquitin-activating enzyme E1 (Leyser et al., 1993). The axr1 mutations were identified by auxin-resistant growth of the root system, but the mutations also generate curled, serrated rosette leaves with short petioles, and a highly branched inflorescence indicating that the gene has a function in the aerial portion of the plant. We constructed axr1-3,lop1 and axr1-12,lop1 double mutants to test the idea that lop1 is an auxin overproducer. We reasoned that if the lop1 phenotype resulted from overproduction of auxin, then the response defect of axr1 would abolish this phenotype in double mutants. Double heterozygotes were constructed and their F2 progeny segregated each of the single mutants and a rare class representing the axr1,lop1 double mutants, as shown in Table 3. The rosette leaves of double mutants were curled and serrated (axr1-like) and twisted (lop1-like), showing evidence of both mutant phe-notypes. We also observed inhibition of root growth characteristic of lop1 in the double mutants, as shown in Table 3. These results show that axr1 does not abolish the lop1 phenotype, and together with the GC-SIM-MS data indicate that lop1 is not an overproducer of IAA.

Inhibition of primary root growth by exogenous auxin is a well-characterized trait employed in identification of Ara-bidopsis mutants defective in auxin response (Lincoln et al., 1990). Our dose response curves with wild-type Arabidopsis indicated that root growth in the interval between 6 and 12 days post-germination was reduced 50% by 3×10−8 M 2,4D. To determine whether the lop1 mutation affected auxin response, growth of primary roots was compared in mutant and wild-type seedlings in the presence and absence of 3×10−8 M 2,4D, as shown in Fig. 6. Wild-type control roots grew to an average length of 6.3 cm by 12 days, as opposed to 2.9 cm in the presence of 2,4D. Mutant roots grew slower than wild-type on the control medium, reaching an average length of only 1.7 cm, but in the presence of 2,4D, root length was further reduced by 50% to 0.7 cm. Further, it was observed that mutant and wild-type seedlings similarly displayed cessation of root growth at 6×10−8 M 2,4D (data not shown). Thus, auxin perception does not appear to be altered in the lop1 mutant.

Auxin is produced primarily in young leaf primordia at the shoot apex and transported in a basipetal direction (downward) in the stem. Excised stem segments retain this pattern of basipetal transport, translocating labeled IAA preferentially from the apical end to the basal end of the segment (Okada et al., 1991). Non-specific acropetal transport from the basal to the apical end provides the appropriate control. Data in Fig. 7 show that basipetal IAA transport in the mutant is only slightly above non-specific acropetal transport whereas, wild-type controls show about a ten-fold increase in basipetal vs. acropetal transport. These results indicate that the lop1 mutant is defective in basipetal transport of IAA.

The ability to impose precise controls over the extent and the polarity of cell expansion is one of the most fundamental elements of pattern formation in higher plants. It is well known that auxin is a key player in this aspect of cellular growth, and accordingly this plant hormone affects a very broad range of developmental pathways. Even its effect on differentiation of vascular tissue is thought to be associated with the induction of polarized cell elongation and subsequent formation of specialized channels for auxin transport (Sachs, 1991). The principal challenge with the auxin pathway has been to determine the molecular basis for its specificity, and this picture is now emerging from the identification of genes and proteins involved in auxin synthesis (Boerjan et al., 1995), transport (Lomax et al., 1995; Okada et al., 1991) and perception (Leyser et al., 1993). In particular, there has been progress toward understanding mechanisms for auxin transport. Directional flow appears to be established by localization of IAA efflux carriers at the basipetal end of trans-porting cells (Lomax et al., 1995), and there is some evidence that the process is regulated by flavonoids (Rubery and Jacobs, 1990) and other regulatory proteins (Lomax et al., 1995). There are as yet, however, no molecular probes or loss-of-function mutations identified for these components of the transport pathway.

Our histological and biochemical analysis of lop1 mutant plants provide several lines of evidence that this mutation generates a primary defect in the auxin transport pathway. Foremost is the transport assay itself showing that mutants stems are defective in basipetal movement of labeled IAA (Fig. 7), whereas auxin perception (Fig. 6) and overall levels of free IAA appear normal. Further, the disoriented growth of the midvein in lop1 mutants (Figs 3D, 4C) and particularly the midvein bifurcation (Fig. 3G) are mutant phenotypes that are consistent with an early disruption in patterns of auxin transport in young primordia. It remains unclear how a transport defect would generate the vein rotation observed in mutant petioles (Fig. 4F,H). Transport patterns may have some role in determination of dorsal/ventral orientation of developing veins, but alternatively it is possible that this phenotype is a secondary consequence of abnormal growth in the surrounding cortical tissue.

The abnormal patterns of cell expansion observed in lop1 mutants also seem consistent with a primary defect in auxin transport. In dicot stems, the elongated parenchyma cells associated with veins are the main conduit for auxin transport (Jacobs and Gilbert, 1983), and lateral movement of IAA from these cells is known to affect cell expansion in the surrounding cortex (Sanchez-Bravo et al., 1991). We observed formation of oversized midribs in mutant leaves that results primarily from precocious radial expansion of the cortical cells (Table 2). His-tological analysis revealed that abnormal expansion patterns are detectable at early stages in leaf development (Fig. 4J), particularly along the lateral flanks of the primordium and often in association with a loss of blade formation (Fig. 4K). Abnormal auxin transport by vascular-associated parenchyma cells may lead to auxin accumulation and excess lateral movement into the cortex where it generates precocious cell expansion. It is also possible that vein rotation is partially responsible for abnormal patterns of cortical development. Our analysis of wild-type midribs shows that cortical expansion always begins with cells adjacent to the phloem pole on the ventral side of the midvein (Fig. 2G-I), suggesting that phloem-associated parenchyma may be more active in IAA transport. Vein rotation shifts the orientation phloem poles out of the normal dorsal/ventral axis in mutant midribs, as illustrated in Fig. 8. This could result in abnormal cell expansion patterns in the cortex, having the net effect of increasing the circumferential extent of rib formation at the occasional expense of blade development.

Finally, the abnormal development of lateral roots (Figs 3J and 5D), and disorganized patterns of epidermal cell expansion in the root elongation zone (Fig. 5B) of the mutant are also consistent with a disruption in the auxin pathway. Epidermal cells of the elongation zone are well known target sites for auxin-mediated growth regulation, and further it is known that there are two distinct auxin transport streams in plant roots, one moving acropetally through the stele, and another moving basipetally in the epidermis (Lomax et al., 1995; Jones, 1990; Tsurumi and Ohwaki, 1978). Application of exogenous IAA has shown that elongation growth of roots is controlled primarily by basipetal transport through the epidermis (Davies et al., 1976). Defective basipetal transport in epidermal cells thus represents a reasonable explanation for the abnormal patterns of cell expansion and general disorganization of the lop1 root elongation zone. A disruption in epidermal transport might lead to localized accumulation of IAA, and associated abnormalities in basic patterns of cell expansion. This epidermal disruption is correlated with a twisted morphology in the roots, and may be the cause of this abnormal growth pattern. But just how adjacent epidermal cells manage to elongate collectively in the wrong direction to generate a repeating dextral or sinistral spiral remains an unanswered question.

Although these shoot and root phenotypes in the mutant point to a defect in auxin transport, it is still uncertain as to whether the LOP1 gene product actually plays a direct role in this process. For example it is possible that the lop1 mutation disrupts some fundamental aspect of cellular organization which then produces the transport defect as a secondary con-sequence (Gersani et al., 1986). Nonetheless it appears that lop1 has a rather specific effect on transport, as opposed to mutations at the PIN-FORMED locus (chromosome 1) in Ara-bidopsis, which produce a disruption in transport (14% of wild-type) and a deficiency in levels of free IAA (8% of wild-type; Okada et al., 1991). The pin1 mutant plants generate abnormal flowers, or a spike inflorescence lacking floral buds altogether. Mutants also have fused cotyledons, and split midveins in the leaves, indicating a role for this gene in early stages of embryo-genesis and in subsequent vegetative development. The lop1 mutation produces split midveins like pin1, but otherwise the phenotypes are quite distinct. In particular, lop1 has no effect on cotyledon morphology, and allows a normal pattern of floral initiation. This may be related to the different levels of free IAA in these mutants. Free IAA levels are normal in lop1, but drastically reduced in pin1, which may account for the broad spectrum of abnormalities observed there. The IAA deficiency in pin1 was considered a secondary consequence of the transport lesion (Okada et al., 1991), perhaps due to feedback regulation linking IAA biosynthesis and transport. Our results with lop1 show normal biosynthesis in spite of a major transport defect, which argues against feedback regulation.

Genetic approaches are certain to play an indispensable role in dissecting the myriad aspects of development connected to auxin biology. Mutations knocking out one function at a time may provide the only straightforward means of identifying functionally distinct branches of this complicated pathway. The lop1 mutant, for example, has abnormal shoots and roots but its cotyledon phenotype is normal. Since polar transport is an essential feature of normal embryogenesis (Cooke et al., 1993), this suggests that LOP1 functions primarily in the post-embryonic phase of development. In contrast, the pin1 mutation appears to have its most prominent effect in the embryo and in the emerging inflorescence. Molecular analysis of LOP1 and PIN1 gene products, together with continued genetic dissection should lead eventually to a better understanding of the auxin pathway and its far reaching effects in plant development.

We thank Dr Jennifer Normanly (Dept. of Biochemistry and Molecular Biology, University of Massachusetts) for generous advice and assistance with the GC-SIM-MS auxin assay, Dr Stephen Della-porta (Dept. of Biology, Yale University) for his participation in the T-DNA screen, and Dr Timothy Nelson (Dept. of Biology, Yale Uni-versity) for invaluable counsel. We gratefully acknowledge the technical assistance of Regan Huntley and Ruben Ruiz. This work was partially supported by a USDA/NRI grant to N. A. M.