Blue light inhibits stomatal development in soybean isolines containing kaempferol-3-O-2^G-glycosyl-gentiobioside (K9), a unique flavonoid glycoside

INTRODUCTION

Plants have developed a finely tuned system to control the balance between assimilation of CO₂ and transpiration of water vapour either by varying stomatal density or by adjusting guard cell turgor, which alters the aperture or stoma between adjacent guard cells (i.e. stomatal movement). Considerable knowledge has been gained over the years on stomatal movement ( Mansfield, Hetherington & Atkinson 1990; Zeiger 1994; Willmer & Fricker 1996; Assman & Shimazaki 1999), but very little is known about control of stomatal formation and density. Three mutants have been identified with irregular stomatal patterning and development ( Zeiger & Stebbins 1972; Yang & Sack 1995), but the genes that result in these phenotypes have not been characterized.

The following information about formation of the stomatal apparatus has been established (see recent reviews Chin et al. 1995 ; Larkin et al. 1997 ). First, the distribution of stomata over a leaf surface is not random. One hypothesis proposes that inhibitory chemicals generated by a developing stomatal complex block the formation of stomata in adjacent cells ( Bünning 1956; Larkin et al. 1997 ); however, the nature of these chemicals is unknown. Second, the formation of the stomatal apparatus is initiated by one or more asymmetric cell divisions and concluded by a symmetrical division that forms two guard cells. Cell lineage can be traced by following the development of a leaf. Third, stomatal development stops before 60–80% of leaf expansion occurs ( Gay & Hurd 1975; Rawson & Craven 1975; Turner & Heichel 1977), but stomatal formation is not synchronous in dicots ( Sachs 1979; Larkin et al. 1997 ). Therefore, stomata at different developmental stages can be found during the early phase of leaf expansion. However, there is a brief time during which stomatal development is most sensitive to environmental change ( Schoch et al. 1980 ).

Stomatal density is affected by environmental factors such as light ( Willmer & Fricker 1996). Most research has concentrated on light quantity. High light intensity is believed to promote high stomatal density, although this correlation is not always observed ( Knecht & O’Leary 1972; Turner & Heichel 1977; Kubinovä 1991). The effects of light quality have not been extensively studied. Kleiber & Mohr (1963) and Schoch et al. (1984) suggested that phytochrome controlled stomatal development. Blue light or far-red light appeared to decrease stomatal index when provided as a night-treatment at low fluence and compared with darkness or red light in Vigna sinensis ( Schoch et al. 1984 ). Lu, Quinones & Zeiger (1993) showed more than a three-fold increase in upper stomatal density in cotton leaves when the blue/red ratio was increased from 0·81 to 1·08. Alternatively, Rajapakse & Kelly (1993) reported a 10% decrease in stomatal density in Chrysanthemum when the blue/red ratio increased. Unfortunately, none of the studies in which blue light was varied maintained constant phytochrome photostationary states between treatments; therefore, possible effects of blue light due to phytochrome were not eliminated.

The presence of flavonoids is characteristic of the leaf epidermis. Most flavonoids are located in the central vacuoles of epidermal cells ( Stafford 1990), in which, it is suggested, they function as UV screens to protect underlying photosynthetic tissues ( Gitz, Liu & McClure 1998). Some flavonoids are located in epidermal cell walls ( Schnitzler et al. 1996 ), functioning possibly as antioxidants or as herbivore or insect repellent or both. Flavonoids are also characteristic of differentiated epidermal cells. Trichomes are often reported to contain large amount of flavonoids, including anthocynins ( Pedro, Campos & Pais 1990; Karabourniotis et al. 1992 ). Guard cells are distinguished by their high flavonoid content ( Weissenboeck, Hedrich & Sachs 1986; Weissenboeck et al. 1987 ), which has been reported to be different from the flavonoid components in neighbouring epidermal cells ( Donkin & Martin 1981; Schnabl et al. 1989 ). Although many external factors, such as wounding, pathogen invasion and herbivore damage, can induce flavonoid synthesis, light is the most influential factor ( Stafford 1990). Phytochrome, the blue/UV-A receptor and putative UV-B receptor(s) may all participate in induction of flavonoids ( Beggs & Wellman 1996).

Soybean isolines containing a unique flavonoid glycoside, kaempferol-3-O-2^G-glycosyl-gentiobioside (K9), exhibit lower photosynthetic rates, chlorophyll content and specific leaf weight, as well as extremely low adaxial stomatal density and moderately reduced abaxial stomatal density compared with isolines that do not contain K9 (Buzzell & Buttery & Buzzell 1976, 1987). For comparison, field-grown soybean lines that do not synthesize K9 have stomatal densities of over 130 mm⁻² for the upper epidermis, whereas densities of only 3–8 mm⁻² are typical for the K9-containing lines OX922 and OX941 ( Buttery & Buzzell 1987; Liu-Gitz unpublished observations). Cosio & McClure (1984) concluded that the effects on mesophyll chloroplast activity were indirect because K9 was localized exclusively in the epidermis. Buttery and coworkers ( Buttery & Buzzell 1987; Buttery et al. 1992 ) hypothesized that K9 directly affects formation of stomata and other features of leaf development and that the reduction of photosynthetic rate is the consequence of these effects. However, a shading experiment failed to change stomatal density in K9-containing lines, even though 80% neutral shade reduced flavonoid content in all lines and stomatal density in the non-K9 lines ( Buttery et al. 1992 ). The authors suggested that low, residual levels of K9 in shaded plants were sufficient to inhibit differentiation of the stomatal apparatus.

However, changes in shading will simultaneously alter photosynthetically active radiation (PAR), phytochrome, and blue-light photoreceptors, all of which have the potential to affect stomatal density. Since flavonoid levels are strongly influenced by short-wavelength portions of the spectrum, we reasoned that a method was needed to alter the blue light while having minimal effects on phytochrome and PAR. In this study, we have demonstrated that small amounts of blue light increased flavonoid accumulation but inhibited stomatal formation only in K9-containing isolines when total PAR and the phytochrome photostationary state were held constant.

MATERIALS AND METHODS

Four genes have been identified that control flavonoid glycoside synthesis in soybean, Fg_1,Fg₂, Fg₃, Fg₄ ( Buzzell & Buttery 1973, 1974). Seeds of soybean isolines OX921 (homozygous for the genes fg₁, fg₂, Fg₃, Fg₄), OX922 (Fg_1,fg_2,Fg_3,Fg₄), OX941 (Fg_1,fg_2,Fg_3,Fg₄) and OX943 (Fg_1,fg_2,fg_3,Fg₄) were obtained from the Department of Agriculture, Harrow, Ontario, Canada. OX921 and OX922 were derived from an F₄ plant of Blackhawk × Harosoy 63, OX941 and OX943 from F₇ plants of OX281 × OX922. These two pairs of isolines, which differ in their flavonoid glycoside patterns, are thought to have only one gene that is different within a pair ( Buttery et al. 1987 , 1992). OX922 and OX941 contain a unique flavonoid glycoside, kaempferol-3-O-2^G-glycosyl-gentiobioside (K9).

Plants were grown for 21 d in chambers (M-28, Environmental Growth Chambers, Chagrin Falls, OH, USA) at 27 C, approximately 65% relative humidity, 350 μmol mol⁻¹ CO₂, and a 14 h light : 10 h dark cycle with 14 h of light at total PAR of 750 μmol m⁻² s⁻¹. Light was supplied from low-pressure sodium (LPS) lamps (SOX 180 W, Philips N.A., Bloomfield, NJ, USA) with or without supplemental blue light supplied from special phosphor fluorescent lamps (F40T12/247; Sylvania-GTE, Danvers, MA, USA). Chambers were divided into two matched compartments with a reflective, aluminized barrier. One side received only LPS light and the other side LPS + blue (B). Air was circulated between the two compartments in a common plenum on the air return side and monitored to ensure equal temperature and humidity. Lamps were arranged in a cooled light cap and separated by acrylic barriers from the growing space. An amber barrier (No. 2422; 3·2 mm; Rohm & Haas, Philadelphia, PA, USA) was used on the LPS side to filter low levels of blue and UV-A light emitted by LPS lamps and stray light from the LPS + B side of the light cap. Light spectra in the growth chamber indicated that LPS lamps delivered a strong and narrow band of light primarily at 589 nm ( Fig. 1). Supplemental blue light (400–500 nm) contributed 45 μmol m⁻² s⁻¹ (about 7%) of total PAR, which was well below the blue contribution to sunlight. The blue-light-deficient side was exposed to less than 0·4 μmol m⁻² s⁻¹ of blue light, primarily stray light reflected from the common plenum. PAR was measured with a quantum sensor and data logger (Model 1000, LiCor, Lincoln, NE, USA). Light spectra were determined with a dual-grating spectroradiometer (OL754, Optronics Laboratories, Orlando, FL, USA).

LPS lamps were essentially red light sources with respect to phytochrome, because the estimated P_fr/P_tot ratio was about 0·91 ( Britz & Sager 1990). Previous studies calculated that supplemental blue light at the irradiances provided should have little direct effect on phytochrome (either photostationary state or cycling rate) or photosynthesis ( Britz & Sager 1990). Therefore, blue light effects should reflect primarily the photomorphogenetic action of a blue light photoreceptor. This assumption does not rule out cooperation between blue light and other photoreceptors, such as phytochrome ( Chory et al. 1996 ; Ahmad & Cashmore 1997).

The central leaflets of fully expanded first trifoliate leaves were selected for all measurements. Leaf impressions were made using a modification of the method described by Wilson, Pusey & Otto (1986). Leaflets were detached, coated with a layer of cyanoacrylate adhesive (DURO Super Glue-5, Loctite Corp., Rocky Hill, CT, USA) on the adaxial surface on one side of the midrib and then gently and evenly pressed against 50 mm × 75 mm glass microscope slides. After peeling a leaf, the resulting epidermal impression remained in hardened adhesive attached to the slide and was viewed under a light microscope (Zeiss, Axioscop, Oberkochen, Germany). The process was repeated for the abaxial surface on the opposite side of the midrib. With this method, both total and relative adaxial and abaxial stomatal densities for a single leaf were determined.

Stomatal densities were evaluated using a transect method ( Gitz 1992; Kubinovä 1994). Briefly, a linear reticule was used to define a known transect width (W = 0·164 mm) approximately 1 cm from the midrib and 3–4 cm from the pulvinus. As the leaflet impression was advanced toward the tip and parallel to the midrib, stomata were counted (n) and the distance travelled (L) was measured with the stage micrometer. Stomatal density (d) was then calculated as d = n/(W×L). Stomatal densities were always determined from the centre of the leaflet half. Large total areas were typically observed (2·7 mm²), enabling very reproducible measurements. Values were expressed as means ± one standard deviation for six replicate leaves. For presentation, data from a single experiment are shown. Similar results were obtained in at least three different experiments. Significance of the results between treatment means was determined by Student t-test (Sigma Stat, SPSS, Inc., Chicago, IL, USA) or analysis of variance ( ANOVA) (StatViewII, Abacus Concepts, Berkeley, CA, USA).

To examine the surface structure of the stomatal apparatus, matching segments of the leaves were placed in vials containing 3% glutaraldehyde in 0·05 M phosphate buffer pH 6·8 at 22 C. Chemical fixation for 24 h was followed by dehydration in an ethanol series and critical point drying from liquid carbon dioxide. The dried specimens were mounted on stubs and coated with gold-palladium in a sputter coater (Technics, Inc., Alexandria, VA, USA). The coated specimens were imaged and photographed with a scanning electron microscope (Hitachi Scientific Instrument, Mountain View, CA, USA) operating at 10 or 15 kV.

For high-performance liquid chromatography (HPLC) analysis of flavonoid glycosides, leaf discs were taken from the lateral leaflets of the same leaf where stomatal density was measured. The discs were ground in liquid N₂ and extracted twice with 50% methanol with 1% acetic acid at room temperature. Combined extracts were brought to 5 mL volume, filtered through 0·22 μm nylon membranes (Sigma, St. Louis, MO, USA), and injected into an HPLC injector (Waters, Medford, MA, USA) with a C18 Ultrasphere reverse-phase column (Alltech, Deerfield, IL, USA). Separation employed a linear gradient composed of acetonitrile with 0·5% phosphoric acid (solvent A) and water with 0·5% phosphoric acid (solvent B). The gradient started with 0% solvent A for the first 3 min, increased to 40% by 30 min, and then finished with 100% solvent A for washing. Sample elution was monitored with a photodiode array detector (Waters 996 PDA; Waters) and analysed by Waters Millinium^® software. Flavonoid glycoside peaks were integrated by absorbance at 340 nm, and identified using standards identified and purified by the thin-layer chromatography method described by Buzzell & Buttery (1973). Flavonoid glycoside quantification was performed using kaempferol-3-rutinoside and fisetin (INDOFINE, Somerville, NJ, USA) as external and internal standards, respectively.

RESULTS

The major flavonoid glycosides were quantified by HPLC for all isolines grown under high irradiance LPS illumination, either lacking (LPS) or supplemented with short wavelengths (LPS + B). Isolines OX922 and OX941 each contained one major glycoside, which is K9. OX921 contained one major glycoside, kaempferol-3-O-sophoroside (K6). OX943 had three glycosides in similar quantities, kaempferol-3-O-gentiobiose (K3), kaempferol-3-O-2^G-rhamnosyl-gentiobioside (K4) and kaempferol-3-O-glucose (K5). As blue light did not affect the relative content of the three glycosides in OX943, the combined contents of the three were shown for this isoline ( Table 1). All flavonoids, regardless of glycosylation, were increased about three-fold by supplemental blue light ( Table 1). Note that K9 was present in significant amounts in OX922 and OX941 even under LPS illumination, whereas OX921 and OX943 contained no detectable K9.

When OX922 and OX941 are grown under LPS conditions, they have relatively high stomatal densities (50–60 mm⁻²) that are slightly lower than those for lines OX921 and OX943, which contain flavonol glycosides other than K9 ( Table 2). If significant, the difference would suggest that the presence of even low levels of K9 in plants grown without supplemental blue light were correlated with alterations in guard cell development. Since stomatal densities for the lower epidermis were about three-fold greater and did not differ significantly between lines ( Table 2), the ratios of stomatal densities for the upper and lower epidermis were calculated ( Fig. 2). Normalization indicated the ratio was significantly lower for K9 lines compared with non-K9 lines.

To test whether stomatal development was affected by blue light, low irradiance supplemental blue light from special phosphor fluorescent lamps was added to high irradiance LPS background light. The result of this treatment on stomatal density depended on the presence or absence of genes for the K9 phenotype ( Table 2). The density of stomata on the adaxial epidermis remained high in lines that are unable to synthesize K9, but declined about 50% to less than 30 mm⁻² in lines with K9. In contrast, stomatal density on the lower epidermis increased slightly, but significantly, in response to supplemental blue light, when all lines were pooled together for ANOVA ( Table 2). Note that the ratio of stomatal density for upper and lower epidermis was unaltered by supplemental blue light in non-K9 lines, indicating that any changes in stomatal density were proportional for upper and lower epidermis. For K9 lines, the ratio was further reduced by the addition of blue light, indicating that these lines had proportionately fewer stomata on the upper epidermis in the presence of blue light.

Adaxial epidermal impressions of OX922 and OX941 plants (K9 lines) grown under LPS + B showed large numbers of undeveloped meristematic cells and abnormally shaped guard cells in mature leaves ( Fig. 3B). These features were not found on the adaxial surface of non-K9 lines ( Fig. 3A) or the abaxial epidermis of any isolines. Obvious differences between epidermal cells of the different isolines were not detected, indicating that changes in stomatal development were specific and not a general consequence of altered epidermal development. The surface morphology of guard cells was investigated in more detail by scanning electron microscope. A normal stomatal apparatus on an adaxial leaf surface (OX921 grown under LPS + B light) is shown in Fig. 4(A). It consists of the stoma or pore, two symmetric guard cells and two subsidiary cells. An undeveloped stomatal apparatus (OX922 grown under the same conditions) is shown in Fig. 4(B). The smallest central cell probably represents the mother cell or precursor of the guard cell that failed to complete a final cell division. This cell pattern is easily identified under light microscope ( Kagan & Sachs 1991; Chin et al. 1995 ) and is considered to represent arrested stomatal development, since fully matured leaves were examined and no further cell development was expected. Other stomatal complexes had guard cells that were distinct from subsidiary cells, but which did not appear fully formed or functional (i.e. discrete stomata were lacking and either the two guard cells did not appear fully separate from one another or were partially collapsed; Fig. 4c & d). We labelled these stomata as abnormal, since they are not found in leaves of any regular soybean isolines or cultivars. Note that stomatal densities in Table 2 included all normal and abnormal stomata, but not stomatal complexes in arrested development.

When the number of arrested stomata were included with the total count of stomata (potential total stomata), the difference between OX921 and OX922 under the LPS lamps was not significant ( Table 3). Similar results were obtained for OX941 and OX943. Note that the presence of arrested and abnormal stomata in OX922 from LPS conditions is consistent with the presence of low levels of K9 in these leaves. Addition of blue light caused a significant reduction in stomatal density in the upper epidermis of OX922 that resulted from both an inhibition of the initiation of stomatal development and a disruption the sequence of cell divisions as evidenced by an increase in cells with arrested development. These two processes were approximately equal in importance for the total reduction in stomatal density.

DISCUSSION

This study showed that in low irradiance blue light strongly inhibited stomatal development in the upper epidermis of lines that synthesize the flavonol glycoside K9, but not in isoline pairs that lack K9 and are nominally different by one gene ( Table 2). In contrast, a small, but significant, increase in stomatal density was observed in the lower epidermis. Stomatal behaviour in upper and lower leaf surfaces can be regulated separately ( Willmer et al. 1996 ), but it is not known whether the qualitative differences between upper and lower epidermis of K9 lines were due to different responses to light, different light distribution, different distribution of K9, or a combination of several factors. In lines that lack K9, blue light caused a proportional increase in stomatal density on both the upper and lower epidermis ( Table 2, Fig. 2).

Although supplemental blue light significantly increased leaf area for all lines ( Table 2), the reduction in stomatal density in K9 lines was clearly not a non-specific side-effect of blue light on leaf expansion. If the decrease in density was simply the result of the same number of stomata distributed over a larger surface, then similar blue-light effects on stomatal density would be expected for all lines and for both upper and lower epidermis. In any case, the decrease in adaxial stomatal density for K9 lines was much larger than the increase in leaf area.

Because phytochrome absorbs blue light as well as red and far-red light, blue light will induce some phytochrome activity. Blue light inhibition of stomatal density was reported in Vigna ( Schoch et al. 1984 ), but the comparison was done between blue light and darkness. Hence, effects on phytochrome cannot be excluded. We selected conditions that should minimize the differences in the phytochrome photostationary state between the two light treatments ( Britz & Sager 1990), so that treatment effects are unlikely to be related to phytochrome.

The abnormal stomatal morphology and development we reported in K9 lines is not found in any other soybeans under normal (i.e. field or controlled environment) growth conditions. However, we have noted abnormal stomatal development on the abaxial surface of K9 lines grown in the field. This observation merits further study. There is also one report stating that vitrified soybean microplants had similar stomatal features as those described herein ( Mohamed-Yasseen et al. 1992 ). However, these features do occur naturally in species of Cruciferae ( Pant & Kidwai 1967; Paliwal 1967), Begoniaceae and Gesneriaceae ( Inamdar, Bhatt & Patel 1973) as well as Tradescantia ( Boetsch, Chin & Croxdale 1995), Sansevieria trifasciata ( Kagan & Sachs 1991) and the moss Funaria hygrometrica ( Sack & Paolillo 1985).

Flavonol glycoside synthesis is reduced but not eliminated in the absence of blue light ( Table 2). The data are consistent with a connection between K9 and stomatal differentiation with a blue light photoreceptor affecting stomatal differentiation through the increased synthesis of K9. We are unaware any other species that also synthesizes K9 or other reports of connections between stomatal differentiation and flavonoid composition. The extensive inbreeding of the isolines in this study suggests that the isoline pairs may be useful in the genetic analysis of stomatal differentiation. This approach should be valid even if K9 is unique to soybean.

However, the connection between stomatal development and flavonol glycosides may be indirect. In Arabidopis, a Myc-like protein regulates both anthocyanin production (one type of flavonoid) ( Lloyd, Walbot & Davis 1992) and trichome development ( Larkin et al. 1997 ). A recent study of an Arabidopis mutant, CPC, suggests that both leaf trichome and root hair formation are regulated by com-mon pathways involving Myb-like proteins and that both processes are connected with anthocyanin accumulation ( Wada et al. 1997 ). Although epidermal cell differentiation is correlated with anthocyanin accumulation in both examples, it is not linked in a causal relationship. The differences in stomatal development between K9 and non-K9 lines may also be related to such gene transcription factors. Recent studies (Liu-Gitz et al. unpublished) show clearly that the effects of blue light on flavonoid accumulation and stomatal formation can be separated temporally during leaf development. Furthermore, the effects of blue light on leaf development extend to mesophyll cells where K9 is absent. These results are consistent with the absence of a causal relationship between K9 and stomatal formation.

In conclusion, the present study has shown that blue light inhibited stomatal development in two isolines of soybean that differ in genetic heritage and that both contain K9. Blue light did not inhibit differentiation of the stomatal apparatus in isolines differing nominally in only one gene and lacking K9. Differences in differentiation resulted under conditions in which photosynthetically active radiation and phytochrome photostationary state were similar, indicating that one or more blue-light receptors were involved in this process. Although very little is known about genes regulating stomatal formation, extensive knowledge has been gained about blue-light photoreceptor(s) and genes affected by blue light. The finding that blue light inhibits stomatal formation in K9 lines and is somehow linked with the presence of a unique flavonoid glycoside should prove useful in the search for clues regulating stomatal formation.

ACKNOWLEDGMENTS

We thank Dr D.C. Gitz for helpful discussions and for assistance with light spectra measurement, Drs B.R. Buttery and R.I. Buzzell for seeds of isolines, W. Harris for modifications to controlled environment chambers, F. Caulfield for assistance with raising plants, C. Pooley for assistance with photography, and C.A. Murphy for assistance with SEM. This research was funded by USDA-ARS-CWU 1270–11210– 005–00D.