A lysine-rich arabinogalactan protein in Arabidopsis is essential for plant growth and development, including cell division and expansion

Authors

  • Jie Yang,

    1. Department of Environmental and Plant Biology
    2. Department of Biological Sciences
    3. Molecular and Cellular Biology Program, Ohio University, Athens, OH 45701-2979, USA
    Search for more papers by this author
  • Harjinder S. Sardar,

    1. Department of Environmental and Plant Biology
    2. Department of Biological Sciences
    3. Molecular and Cellular Biology Program, Ohio University, Athens, OH 45701-2979, USA
    Search for more papers by this author
  • Kathleen R. McGovern,

    1. Department of Biological Sciences
    Search for more papers by this author
  • Yizhu Zhang,

    1. Department of Environmental and Plant Biology
    2. Department of Biological Sciences
    3. Molecular and Cellular Biology Program, Ohio University, Athens, OH 45701-2979, USA
    Search for more papers by this author
  • Allan M. Showalter

    Corresponding author
    1. Department of Environmental and Plant Biology
    2. Department of Biological Sciences
    3. Molecular and Cellular Biology Program, Ohio University, Athens, OH 45701-2979, USA
      (fax +1 740 593 1130; e-mail showalte@ohio.edu).
    Search for more papers by this author

(fax +1 740 593 1130; e-mail showalte@ohio.edu).

Summary

Arabinogalactan proteins (AGPs), a family of hydroxyproline-rich glycoproteins, occur throughout the plant kingdom. The lysine-rich classical AGP subfamily in Arabidopsis consists of three members, AtAGP17, 18 and 19. In this study, AtAGP19 was examined in terms of its gene expression pattern and function. AtAGP19 mRNA was abundant in stems, with moderate levels in flowers and roots and low levels in leaves. AtAGP19 promoter-controlled GUS activity was high in the vasculature of leaves, roots, stems and flowers, as well as styles and siliques. A null T-DNA knockout mutant of AtAGP19 was obtained and compared to wild-type (WT) plants. The atagp19 mutant had: (i) smaller, rounder and flatter rosette leaves, (ii) lighter-green leaves containing less chlorophyll, (iii) delayed growth, (iv) shorter hypocotyls and inflorescence stems, and (v) fewer siliques and less seed production. Several abnormalities in cell size, number, shape and packing were also observed in the mutant. Complementation of this pleiotropic mutant with the WT AtAGP19 gene restored the WT phenotypes and confirmed that AtAGP19 functions in various aspects of plant growth and development, including cell division and expansion, leaf development and reproduction.

Introduction

Arabinogalactan proteins (AGPs) are hyperglycosylated members of the hydroxyproline-rich glycoprotein superfamily, and decorate the surfaces of cells throughout the plant kingdom. The protein backbones of AGPs are rich in Pro/Hyp (hydroxyproline), Ser, Ala and Thr, and are modified by the addition of type II arabinogalactan polysaccharides and arabinose oligosaccharides (Gaspar et al., 2001; Showalter, 2001). AGPs react specifically with β-glucosyl Yariv reagent (Yariv et al., 1962, 1967), a chemical reagent used to precipitate AGPs as well as to probe their function.

AGPs are divided into several classes: classical AGPs, lysine-rich classical AGPs, AGP peptides, fasciclin-like AGPs (FLAs), and other chimeric AGPs. Classical AGPs are 85–151 amino acids in length and consist of an N-terminal signal peptide, a Pro/Hyp-rich AGP central domain and a C-terminal glycosylphosphatidylinositol (GPI) lipid anchor addition sequence (Schultz et al., 1998; Showalter, 2001). Upon GPI anchor cleavage by phospholipases, AGPs are released from the plasma membrane to the extracellular matrix. Lysine-rich AGPs are a subclass of the classical AGPs and have a small lysine-rich region within the classical AGP domain. The lysine-rich region is not glycosylated, which has allowed the production of peptide-specific antibodies (Gao and Showalter, 2000; Gao et al., 1999; Zhang et al., 2003). AGP peptides are distinguished by small AGP domains, which are typically 10–15 amino acids in length; these AGP domains are flanked by a signal peptide and in many cases by a GPI anchor addition sequence. FLAs contain both AGP and fasciclin-like domains (Schultz et al., 2002). In Arabidopsis, there are 14 classical AGPs, three lysine-rich AGPs, 12 AG peptides and 21 FLAs (Schultz et al., 2002).

In addition to FLAs, other chimeric AGPs exist (Borner et al., 2003; Schultz et al., 2002). Xylogens in Zinniaelegans and Arabidopsis contain both an AGP domain and a non-specific lipid transfer protein domain (Motose et al., 2004). In rice, two chimeric AGPs, an early nodulin-like protein and a lipid transfer-like protein, were recently identified (Mashiguchi et al., 2004).

While the structure and composition of AGPs are well characterized from biochemical and molecular biology studies (Gaspar et al., 2001; Kieliszewski and Shpak, 2000; Sun et al., 2004b, 2005; Zhao et al., 2002), their biological roles have remained elusive and are only now beginning to be understood. AGPs are implicated as functioning in many cellular and physiological processes (Majewska-Sawka and Nothnagel, 2000; Showalter, 2001), including somatic embryogenesis (van Hengel et al., 2002), cell proliferation (Langan and Nothnagel, 1997; Serpe and Nothnagel, 1994), cell expansion (Willats and Knox, 1996), xylem differentiation and development (Gao and Showalter, 2000; Motose et al., 2001; Zhang et al., 2003), programmed cell death (Gao and Showalter, 1999), pollen tube growth (Wu et al., 2000) and plant hormone actions (Suzuki et al., 2002).

AGP mutants represent a new, powerful and direct means to assign function to individual AGPs. RNA interference (RNAi) of AGP1 in moss (Physcomitrella patens) resulted in reduced cell length (Lee et al., 2005). When both Arabidopsis xylogen genes (AtXYP1 and AtXYP2) are knocked out, the mutant leaves have discontinuous veins (Motose et al., 2004). A mutation in AtFLA4 results in irregular cell expansion, thinner cell walls and increased sensitivity to salt (Shi et al., 2003). AtAGP30, a non-classical AGP containing six cysteines in the C-terminus, enhances the response to abscisic acid (ABA), and is required for root regeneration and seed germination (van Hengel and Roberts, 2003). In addition to AtAGP30, two lysine-rich AGPs, CsAGP1 from cucumber (Cucumis sativus) and LeAGP1 from tomato (Lycopersicon esculentum), also respond to phytohormones. CsAGP1 is responsive to gibberellin and implicated in stem elongation (Park et al., 2003). LeAGP1 is up-regulated by cytokinin, and its over-expression results in tomato plants that phenotypically resemble cytokinin-over-expressing plants (Sun et al., 2004a).

The three lysine-rich AGPs in Arabidopsis, AtAGP17, AtAGP18 and AtAGP19, are likewise important for plant growth and development. One incomplete knockout mutant of AtAGP17, designated as rat1 (resistance to Agrobacterium), is deficient in binding of Agrobacterium to its roots (Gaspar et al., 2004; Nam et al., 1999). AtAGP18 is essential for female gametogenesis, as functional megaspores in AtAGP18 RNAi mutants failed to enlarge and divide, resulting in ovule abortion and reduced seed set (Acosta-Garcia and Vielle-Calzada, 2004). In order to complement our previous work with a tomato lysine-rich AGP, namely LeAGP1 (Gao and Showalter, 2000; Gao et al., 1999; Li and Showalter, 1996; Sun et al., 2004a,b, 2005), and further our understanding of the lysine-rich AGPs in Arabidopsis, AtAGP19 was examined here in terms of its genetic expression and function in Arabidopsis growth and development, based upon mutant analysis and genetic complementation.

Results

AtAGP19 is classified as a lysine-rich AGP

The AtAGP19 gene (At1g68725) is 833 bp in length, and the predicted size of its mRNA is 744 nucleotides. While the consensus intron splicing site for most genes contains the sequence GT…AG, AtAGP19 has a single intron with a non-consensus intron splicing site of GT…AT. Seven lysine-rich AGPs from various plant species have been identified to date: LeAGP1 (Gao et al., 1999; Li and Showalter, 1996) in tomato, NaAGP4 in Nicotiana alata (Gilson et al., 2001), AtAGP17, 18 and 19 in Arabidopsis (Schultz et al., 2002), CsAGP1 in cucumber (Park et al., 2003) and PtaAGP6 in pine (Pinus taeda) (Zhang et al., 2003). All have an N-terminal signal sequence, a central AGP domain containing a small lysine-rich region, and a C-terminal GPI anchor addition sequence (Figure 1). While LeAGP1 and NaAGP4 share 80% amino acid sequence identity, other lysine-rich AGPs have lower amino acid sequence similarities and identities. For example, AtAGP17 and AtAGP18 are 64% and 54% similar and identical to each other, respectively. On the other hand, AtAGP19 has only 38% and 28% amino acid sequence similarity and identity to AtAGP17, and 45% and 34% similarity and identity to AtAGP18. Attention was focused on AtAGP19, as a mutation in this gene has the potential to provide a more dramatic phenotype and would be less likely to experience compensation by the two other subfamily members.

Figure 1.

 Gene structure of AtAGP19.
A T-DNA insertion in AtAGP19 (in SALK_038728) is shown and not drawn to scale.

AtAGP19 expression is tissue-specific and developmentally controlled

Northern blot analyses of Arabidopsis seedlings and mature organs using gene-specific probes revealed that expression of AtAGP17, 18 and 19 in Arabidopsis was organ-specific (Figure 2a). The AtAGP17 transcript was found in seedlings, rosette leaves, flowers and stems, but not roots. AtAGP19 transcript levels were high in stems, moderate in roots and flowers, low in seedlings, and barely detectable in mature rosette leaves. Similarly, AtAGP18 was expressed in roots, flowers and stems and weakly expressed in seedlings and rosettes.

Figure 2.

 Genetic expression of AtAGP19.
(a) Transcript abundance of AtAGP17, 18 and 19 determined by Northern blotting. Aerial parts of 10-day-old seedlings (SL) and 4-week-old roots (RT), rosettes (RS), flowers (FL) and stems (ST) were examined. rRNAs were stained with ethidium bromide to show equal loading. The sizes of AtAGP17, 18, and 19 mRNA are shown on the right.
(b) Expression pattern of AtAGP19 by microarray analyses. Values are means + SD.
(c) Expression of AtAGP19 detected by massively parallel signature sequencing. Roots and leaves were 21 days old, seedlings were 3 days old, and floral buds and developing siliques were harvested from 5-week-old plants.

To investigate the expression of AtAGP19 at the tissue level, expression of the GUS gene under the control of the AtAGP19 promoter was examined in transgenic Arabidopsis plants (Figure 3). GUS staining was consistently found in the vasculature of leaves, roots, stems and flowers (pedicels, sepals, petals and filaments). Both light- and dark-grown 3-day-old seedlings expressed GUS in the cotyledons, hypocotyls and roots, but not in root hairs or root tips. In 7-day-old-seedlings, the root staining pattern resembled that of the 3-day-old seedlings, although the staining in hypocotyls was stronger and broader, with the apical portion of the hypocotyl being more heavily stained than the basal portion. In rosette leaves, the vascular tissues in the blade and petiole were stained, with pronounced staining at the hydathodes, epidermal structures specialized for water secretion. Staining in new leaves was strong throughout, but decreased and became restricted to the vasculature as the leaves matured. Old or senescent leaves showed little or no staining, consistent with the Northern blot analysis of mature leaves. Staining in cauline leaves was similar to that in rosette leaves. GUS activity was also found in leaf and stem trichomes. Young stems were stained throughout their lengths, while older stems showed greatest staining in the apical portion (data not shown). Anthers lacked GUS staining, while the style, ovary walls and transmitting tract were stained. Siliques also demonstrated GUS staining. These results indicate that AtAGP19 expression is developmentally controlled, with young organs being preferentially stained, and organ- and tissue-specific, with the vasculature, style and developing siliques displaying the greatest amount of GUS staining.

Figure 3.

 GUS staining in transgenic plants harboring the PAtAGP19:GUS reporter gene construct.
(a,b) Light-grown 3- and 7-day-old seedlings, respectively. (c) Dark-grown 3-day-old seedling. (d) Rosette. Staining was restricted to the vasculature in mature leaves. The arrow indicates a hydathode. (e,f) Root and root tip. (g) Inflorescence including a cauline leaf. (h) Magnified view of the stem in (g). (i) Open flower. (j) Carpel in an unopened flower. Note staining of the style, ovary wall and transmitting tract. (k) Silique. Bars = 1 mm for (a)–(d) and (g), and 0.1 mm for (e), (f) and (h)–(k).

In addition to Northern blotting and GUS analyses, AtAGP17, 18 and 19 expression was also examined by accessing publicly available microarray data online (Zimmermann et al., 2004). Microarray data indicated that AtAGP18 was the most abundantly expressed lysine-rich AGP, followed by AtAGP17 and 19. The relative transcription levels of these three AGPs were corroborated by Northern blotting, RT-PCR and GUS staining (data not shown). Furthermore, our expression patterns of AtAGP17, 18 and 19 were largely consistent with the microarray data. Low to moderate levels of AtAGP19 transcription were detected in wild-type (WT) Arabidopsis plants and cell cultures, and Figure 2(b) shows the mean signal intensities for AtAGP19 in a pool of microarray experiments. When Arabidopsis seedlings of WT and various mutant backgrounds were treated with biotic and abiotic stresses (such as wounding, heat, hormones and hormone inhibitors, pathogen-derived elicitors), AtAGP19 mRNA levels did not change significantly (data not shown). Microarray data confirmed the widespread expression of AtAGP19, although there were some minor differences with regard to relative expression levels between the AtAGP18 and AtAGP19 microarray profiles and our results. Two reasons may account for such differences. First, the Arabidopsis plants used in microarray experiments were grown by different laboratories and might not be harvested at identical stages. Second, the detection sensitivity of microarray experiments is limited by RNA quantities (Meyers et al., 2004), and AtAGP19 transcript abundance is relatively low.

Another approach to analyze gene expression is by MPSS (massively parallel signature sequencing), which is more sensitive and accurate in quantifying gene expression at low levels compared to microarray experiments (Meyers et al., 2004). MPSS analysis of Arabidopsis revealed that AtAGP19 was most highly expressed in elongating siliques and immature floral buds. AtAGP19 transcripts were also found in germinating seedlings, and the abundance was low in roots and leaves (Figure 2c). Consistent with microarray data, MPSS showed that overall AtAGP19 mRNA levels were low.

Identification and genetic complementation of a null knockout mutant of AtAGP19

A reverse genetics approach was taken to determine the function of AtAGP19. Only one null T-DNA mutant line of AtAGP19 (SALK_038728) has been identified (Alonso et al., 2003) and was obtained from the Arabidopsis Biological Resource Center (ABRC). The location of the T-DNA insertion was verified by sequencing (Figure 1). RT-PCR demonstrated that AtAGP19 mRNA was absent in the homozygous atagp19 mutant (Figure 4a).

Figure 4.

 Expression of AtAGP17, 18 and 19 in the atagp19 mutant and complemented mutant plants.
(a) RT-PCR of total stem RNA showed that AtAGP19 mRNA was absent in the atagp19 mutant but present in complemented plants. Actin was used as the internal control.
(b) Northern blotting of AtAGP17 and 18 in the atagp19 mutant. Lane 1, wild-type (WT) seedling; lane 2, atagp19 mutant seedling; lane 3, WT rosette; lane 4, atagp19 mutant rosette; lane 5, WT stem; lane 6, atagp19 mutant stem; lanes 7–10, stem RNA of complemented atagp19 mutant lines 3, 6, 10 and 14. rRNAs were stained with ethidium bromide to show equal loading.

Although the heterozygous mutant was not phenotypically different from WT, the atagp19 homozygous mutant displayed considerable abnormal phenotypes with respect to various aspects of plant growth and development. Complementation of the mutant with the WT AtAGP19 gene under the control of its own promoter restored AtAGP19 mRNA (Figure 4a) as well as the WT phenotypes. This confirmed that the mutant phenotypes were caused by the knockout of AtAGP19. SALK lines of AtAGP17 (SALK_101062) and AtAGP18 (SALK_117268) were also examined, but neither showed any readily identifiable phenotypes compared with those observed in atagp19.

As there are three lysine-rich AGPs in Arabidopsis, it was important to know whether the two homologs of AtAGP19, namely AtAGP17 and 18, were compensating for AtAGP19 in the atagp19 null mutant. As shown in Figure 4(b), both AtAGP17 and 18 transcription was down-regulated in the mutant stem. In complemented mutants, AtAGP17 and 18 expression returned to normal levels. In atagp19 seedlings, while AtAGP18 transcription was elevated, the AtAGP17 mRNA level showed little change. In contrast, in mutant rosettes, AtAGP18 expression did not change, while the AtAGP17 mRNA level increased.

Beyond these molecular phenotypes, several remarkable phenotypes were readily apparent in the atagp19 mutant. These phenotypes were analyzed in detail and reported below.

Reduced hypocotyl cell length in the atagp19 mutant

atagp19 hypocotyls were 75% of the length of WT hypocotyls when grown under long-day conditions, which corresponded to a reduction in hypocotyl cell length but not cell number. However, hypocotyl length was not compromised when atagp19 seedlings were grown in the dark. These results indicate that AtAGP19 function is regulated by light. Moreover, although cell length was reduced in the light-grown mutant, hypocotyl diameter, cell width and cell layers were all similar to those of WT (Table 1).

Table 1.   Hypocotyl parameters of wild type (WT) and atagp19 mutant seedlings
ConditionsParametersWTatagp19
  1. Values are expressed as mean ±95% CI.

LightHypocotyl length (mm)2.8 ± 0.12.1 ± 0.1
 Hypocotyl cell number18.1 ± 0.819.8 ± 0.6
 Average epidermal cell length (μm)155.9 ± 6.4110.3 ± 8.4
 Hypocotyl diameter (μm)330.8 ± 14.2320.5 ± 6.7
 Average epidermal cell width (μm)13.2 ± 0.713.4 ± 0.6
DarkHypocotyl length (mm)16.2 ± 0.516.5 ± 1.3

The atagp19 mutant grows more slowly and has altered leaf morphology and pigmentation

During vegetative growth, atagp19 had fewer rosette leaves compared to WT plants at the same age. Moreover, atagp19 rosette leaves were smaller and more round and had shorter petioles than WT (Figure 5a,b).

Figure 5.

 Morphology and pigmentation differences between wild-type (WT) and atagp19 mutant leaves.
(a) Three-week-old rosettes of WT (left), atagp19 mutant (center) and complemented plants (right). Bar = 1 cm.
(b) Rosette leaves of WT (top), atagp19 mutant (center) and complemented plants (bottom). Bar = 1 cm.
(c) The atagp19 mutant had lower chlorophyll and anthocyanin levels than WT. Error bars indicate SE (n = 5).

atagp19 mutants, including leaves, stems and sepals, were lighter green than WT plants throughout the life cycle. Pigment content analyses demonstrated that atagp19 rosette leaves contained less chlorophyll and anthocyanin compared to WT (Figure 5c,d). Microscopic analysis showed that there was no significant difference in chloroplast numbers in mesophyll cells between the mutant and WT (data not shown).

In order to explore causes of the smaller leaf size, the abaxial epidermis of the first two true leaves of WT and atagp19 was studied. Both the size and number of the epidermal cells were found to be decreased in the mutant. However, stomata cells appeared normal in terms of morphology and fraction of total epidermal cells (Figure 6).

Figure 6.

 The epidermis of atagp19 mutant first leaves contained smaller and fewer cells.
(a) Three-week-old first leaves of the atagp19 mutant were only half the size of those of wild type (WT), consistent with smaller and fewer cells in the atagp19 abaxial epidermis. However, the atagp19 mutant and WT had similar stomata to total epidermal cell ratios. Error bars indicate SE (n = 9).
(b) Drawing tube images of WT (top) and atagp19 (bottom) abaxial epidermis. Bars = 50 μm.

As plants matured, WT rosette leaves curled down and atagp19 mutant leaves remained flat (Figure 7a). To determine the causes of flat mutant leaves, a differential interference contrast (DIC) microscope was used to visualize mesophyll cells. Palisade mesophyll cells were smaller in the mutant than in WT (Figure 7b,c), whereas spongy mesophyll cells in the atagp19 mutant were more regular in shape and more densely packed than in WT (Figure 7d,e); this abnormal cellular packing may result in the flat leaves in the atagp19 mutant. In contrast, spongy mesophyll cells in WT were irregular in shape and characterized by large intercellular spaces.

Figure 7.

 Mature atagp19 mutant rosette leaves were flat.
(a) 49-day-old rosettes of WT (left) and atagp19 mutant plants (right). The inflorescence stems were removed from the rosettes before the image was taken.
(b,c) Paradermal views of palisade mesophyll cells in wild type (WT) and the atagp19 mutant, respectively.
(d,e) Paradermal views of spongy mesophyll cells in WT and the atagp19 mutant, respectively. Gray represents cells, and white represents intercellular space. At least ten WT and atagp19 mutant leaves were examined, respectively.
Bars = 1 cm in (a), and 10 μm in (b)–(e).

Compromised fertility in the atagp19 mutant

The atagp19 mutant had shorter, more slender inflorescence stems with fewer auxillary branches and side bolts (Figure 8a and Table 2). atagp19 produced fewer flowers than WT (Figure 8b,c). atagp19 also had fewer and shorter siliques, fewer seeds per silique, and a higher percentage of sterile siliques (Table 2), resulting in less seed production. More than half of the atagp19 flowers were fertile, and, although they were smaller than WT flowers, they opened normally and had normal arrangements and numbers of floral organs (Figure 8d). Some atagp19 sterile flowers were open, while others remained closed (Figure 8e). One reason for sterility was the failure of the stamens to elongate beyond the pistil at floral stage 14 (Figure 8f,g), while, in WT, the stamens were longer than the pistil and brushed against the stigma to allow pollination and fertilization (Smyth et al., 1990). Siliques of atagp19 could be divided into three length ranges, with the longest ones being slightly shorter than WT siliques (Figure 8h). The shortest atagp19 siliques were completely sterile, while the medium-length siliques were semi-sterile, bearing fewer than five seeds per silique. Even when pollination occurred normally, there was a high percentage of aborted ovules without signs of early seed development (Figure 8i); therefore, shorter stamens did not seem to be the only cause of sterility. Moreover, reciprocal crosses suggested that female reproduction was possibly defective in atagp19, because few seeds were recovered when atagp19 plants were used as female parents, and normal fertility was observed with atagp19 plants as the male parents (data not shown).

Figure 8.

 Inflorescence and silique phenotypes of the atagp19 mutant.
(a) Inflorescence stems of wild-type (WT) (left), atagp19 mutant (center) and complemented plants (right).
(b,c) Inflorescences of WT and the atagp19 mutant, respectively.
(d) A WT flower (left) was larger than a fertile atagp19 mutant flower (right), but the floral organization was similar.
(e) Top view of a closed atagp19 flower. Sometimes the carpel extended beyond the sepals, as in this case.
(f) A dissected closed atagp19 flower showing abnormal petal and stamen development.
(g) A sterile atagp19 flower with the carpel longer than the stamens at floral stage 14, preventing pollination.
(h) Siliques of WT (top) and atagp19 (bottom) plants. The siliques shown were of representative lengths. The atagp19 mutant contained an abnormally high proportion of short/sterile siliques (Table 2).
(i) The atagp19 mutant (center and bottom) had higher aborted ovule rates than the WT (top). The center silique corresponded to longest atagp19 silique, and bottom silique was of the middle length range in (g).
Bars = 1 cm in (a) and (h), 1 mm in (b)–(d) and (i), and 0.5 mm in (e)–(g).

Table 2.   Stem and flower comparison between wild type (WT) and atagp19 mutant plants
ParametersWTatagp19
  1. Values are expressed in mean ± 95% CI (n > 30).

Stem length (cm)33.8 ± 1.120.1 ± 1.2
Stem width (mm)1.2 ± 0.10.8 ± 0.1
Silique number121.0 ± 21.419.4 ± 3.2
Silique length (mm)13.5 ± 0.410.8 ± 0.4
Seeds per silique54.0 ± 2.428.2 ± 4.6
Sterile siliques (%)4.0 ± 0.531.5 ± 5.1
Side bolts from shoot apex (>1 cm)1.3 ± 0.30.6 ± 0.3
Auxillary branches3.1 ± 0.22.2 ± 0.2

Complementation of the atagp19 mutant and restoration of WT phenotypes

As mentioned previously, complementation of the mutant with the WT AtAGP19 gene under the control of its own promoter restored AtAGP19 mRNA (Figure 4a) as well as the WT phenotypes. In particular, complemented plants displayed normal leaf size, shape, color and growth rate (Figure 5a,b). In addition, stem length and thickness as well as seed production were also restored in the complemented mutants (Figure 8a). These data confirmed that the mutant phenotypes were caused by knockout of AtAGP19.

Discussion

AtAGP19 is a lysine-rich, classical AGP family member

AtAGP19 is a lysine-rich, classical AGP that is predicted to be anchored to the plasma membrane by a GPI anchor (Borner et al., 2003; Schultz et al., 2002); the GPI anchor prediction is supported by the subcellular localization of AtAGP19 to the plasma membrane (J. Yang and A. Showalter, unpubl. results). This gene is related to two homologous AGP genes in Arabidopsis, AtAGP17 and 18, as well as to several other AGP genes characterized to varying extents in several other plant species; all are members of the lysine-rich, classical AGP subfamily. AtAGP19 has relatively low amino acid sequence similarities and identities with AtAGP17 and 18, and is a unique member of the lysine-rich AGP family in Arabidopsis as shown by the phylogeny tree and reflected in its predicted glycosylation pattern (Sun et al., 2005). In addition to arabinogalactan polysaccharide addition sites at non-clustered proline/hydroxyproline residues, AtAGP19 also contains a greater percentage of oligoarabinoside addition sites at clustered proline/hydroxyproline residues compared with its two homologs (Sun et al., 2005).

Expression of lysine-rich AGPs

GUS staining patterns of PAtAGP19:GUS transgenic plants are consistent with the Northern blot, microarray and MPSS data. GUS staining is under developmental control and strong in seedlings and young organs. For example, young leaves and young stems show strong GUS activity throughout. As the leaves mature, GUS staining becomes confined to the veins and gradually diminishes. In flowers, the style exhibits intense GUS activity. MPSS analysis further corroborates that AtAGP19 mRNA levels are high in flowers and developing siliques and low in rosette leaves. Nonetheless, the overall transcription levels of AtAGP19 are the lowest among the three lysine-rich AGPs.

Expression patterns of AtAGP18 and AtAGP19 are similar to that of LeAGP1, which is strongly expressed in flowers and young stems, moderately expressed in roots and young fruits, and weakly expressed in leaves and old stems (Li and Showalter, 1996). LeAGP1 and PtaAGP6 are both immunolocalized to differentiating stem xylem elements and functionally associated with secondary cell wall thickening and xylem development (Gao and Showalter, 2000; Zhang et al., 2003). Similarly, AtAGP19 promoter activity is tightly associated with the vascular tissues and is localized to differentiating, but not developed, xylem elements (J. Yang and A. Showalter, unpubl. results). LeAGP1 is also abundant in stylar transmitting tissues, with putative roles in guiding and nourishing pollen tube growth. In contrast, the CsAGP1 transcript is found throughout cucumber seedlings and may be involved in stem elongation (Park et al., 2003). These proposed functions of other lysine-rich AGPs, which are based largely on expression patterns, have provided guidance in elucidating the roles of AtAGP19 and in studying the atagp19 mutant.

AtAGP19functions in plant growth and development, including cell division and expansion

The atagp19 mutant displays multiple and dramatic mutant phenotypes that are statistically different from WT. These phenotypes occur during both vegetative and reproductive growth, and include altered leaf shape and size, lighter color, slower growth, shorter hypocotyls and inflorescence stems, and compromised fertility. Phenotypes of the atagp19 mutant correspond to the position and timing of AtAGP19 expression and indicate that AtAGP19 is essential for normal plant growth and development, including leaf development, stem elongation and seed production. It is not yet clear how loss of AtAGP19 results in lower pigment levels and a slower growth rate in atagp19. With respect to reproduction, it is likely that the high percentage of aborted ovules seen in atagp19 siliques are a result of several factors, such as abnormal stamen development, compromised female gametogenesis, and poor pollen tube guidance in the style.

AtAGP19 function is regulated by light with respect to seedling growth. In light, atagp19 mutant hypocotyls are approximately 75% as long as WT, but there is no length difference when grown in the dark. AtAGP19 expression, as determined by GUS staining, is found in both light- and dark-grown Arabidopsis seedlings. Similarly, LeAGP1 transcription is not affected by light (Li and Showalter, 1996). Thus, this light-regulated function of AtAGP19 occurs at a post-transcriptional level and probably involves interactions with other components that are more directly regulated by light to control hypocotyl length.

Plant growth involves cell division and cell elongation. The atagp19 mutant has fewer abaxial epidermal cells in rosette leaves, indicating a decrease in cell division. Histological analyses further indicate a reduction in cell division in stems (J. Yang and A. Showalter, unpubl. results). AtAGP19 also plays a role in cell expansion; this is seen from the shorter hypocotyl cells, smaller rosette epidermal cells and more regularly shaped spongy mesophyll cells in the atagp19 mutant. Therefore, the dwarf phenotype of atagp19 is a result of impaired cell division and cell expansion. Previous work has indicated that AGPs, either individually or collectively, function in cell division and expansion (Langan and Nothnagel, 1997; Lee et al., 2005; Majewska-Sawka and Nothnagel, 2000; Park et al., 2003; Serpe and Nothnagel, 1994; Shi et al., 2003; Willats and Knox, 1996), and the findings reported here support these functional roles.

The precise mechanism(s) for the observed function of AtAGP19 in plant growth and development, including cell division and expansion, remains to be elucidated. There are several indications, however, that AtAGP19 acts through cellular signaling pathways. First, the atagp19 mutant has multiple abnormal phenotypes, and the disruption of one or more signaling pathways in this mutant would provide a plausible explanation to account for these pleiotropic effects. Second, many of the phenotypes observed are related to alterations of phytohormone pathways. Third, AtAGP19 has a predicted GPI anchor, and GPI-anchored proteins play putative roles in cellular signaling and communication (Schultz et al., 1998; Youl et al., 1998).

Functional relationships of AtAGP17, 18 and 19

Despite some changes in AtAGP17 and 18 mRNA accumulation in the atagp19 mutant and the similar expression patterns of AtAGP18 and 19, there appears to be little functional compensation. This idea is further supported by the observation that no additional mutant phenotypes are seen in atagp17 atagp19 and atagp18 atagp19 double mutants produced by crossing the SALK_101062 and SALK_117268 lines with atagp19, respectively (J. Yang and A. Showalter, unpubl. results). In AtAGP17 or AtAGP18 T-DNA insertion mutants and AtAGP18 RNAi mutants, no gene compensation by the other lysine-rich AGPs occurs in the organs examined (Acosta-Garcia and Vielle-Calzada, 2004; J. Yang and A. Showalter, unpubl. results). Overall, the picture that emerges from the above expression data and from characterization of mutants of the three lysine-rich AGPs is that all three lysine-rich AGPs have unique and independent functions in Arabidopsis. Future experiments to test this idea might involve determining whether the other two lysine-rich AGP genes can complement a specific lysine-rich AGP mutant (e.g. atagp19) when expressed under the control of the promoter of the mutated AGP gene (e.g. PAtAGP19:AtAGP17 and PAtAGP19:AtAGP18).

AGP mutants provide insight into AGP function

AGP mutants are helping to identify the function of AGPs and provide a framework to explore the underlying mechanisms responsible for AGP function. A study of the over-expression of LeAGP1 connected this AGP to cytokinin signaling (Sun et al., 2004a). In contrast, the atagp19 mutant responds to cytokinin in a way similar to that in WT; for example, chlorophyll content, primary root length and lateral root numbers of both WT and mutant seedlings decrease in response to cytokinin 6-benzyl-amino-purine concentrations higher than 0.005 μm. These observations imply that the function of AtAGP19 may be different from that of LeAGP1 (J. Yang and A. Showalter, unpubl. results). Analysis of a null mutant of AtAGP30, a non-classical AGP gene, suggests that it plays a role in ABA responses and root regeneration (van Hengel and Roberts, 2003). The implication of AGPs in phytohormone pathways is in line with our hypothesis that AtAGP19 is involved in cell signaling pathways. Considering the other two lysine-rich AGPs in Arabidopsis, an insertion in the promoter region of AtAGP17 results in reduced binding to Agrobacterium (Gaspar et al., 2004; Nam et al., 1999), while AtAGP18 RNAi mutants display high ovule abortion rates (Acosta-Garcia and Vielle-Calzada, 2004). Similar to the AtAGP18 RNAi mutants, atagp19 mutant siliques contain many aborted ovules; the reason for this remains unknown. On the other hand, roots of the atagp19 mutant have normal Agrobacterium-binding capacities (J. Yang and A. Showalter, unpublished results).

The defects in cell expansion in the atagp19 mutants are reminiscent of those in three other AGP mutants. Moss AGP1 RNAi plants represent one such mutant and display reduced apical cell extension (Lee et al., 2005). A mutant of AtFLA4 in Arabidopsis shows abnormal cell expansion (Shi et al., 2003), while tobacco plants over-expressing CsAGP1 are taller due to greater stem elongation (Park et al., 2003). Together with this study, there is now compelling evidence that AGPs are required for normal cell expansion. In addition, a double xylogen mutant in Arabidopsis (atxyp1 atxyp2) has shown discontinuous leaf venation patterns (Motose et al., 2004). Although this particular phenotype is not observed in the atagp19 mutant, atagp19 has smaller vascular cylinders in the mature roots compared to WT. The smaller vascular region in the mutant is due to fewer cells and not to changes in cell size, shape and wall thickness of the xylem elements, which appear normal (J. Yang and A. Showalter, unpubl. results). Considering all AGP mutants examined to date, the atagp19 mutant shows the broadest and most readily apparent phenotypes without the need to resort to special screens. The underlying mechanisms explaining how AtAGP19, as well as the other AGPs from characterized mutants, regulate various cellular and physiological processes remain to be elucidated, but efforts are underway to address this important and challenging issue.

Experimental procedures

Bioinformatics

The similarities/identities of lysine-rich AGPs were obtained with MatGAT (matrix global alignment tool), version 2.01, using the BLOSUM62 algorithm (http://www.angelfire.com/nj2/arabidopsis/MatGAT.html) (Campanella et al., 2003).

Microarray data for gene expression in WT Arabidopsis under normal conditions were obtained from https://http://www.genevestigator.ethz.ch/(Zimmermann et al., 2004). Microarray data for gene expression under stresses were retrieved from TAIR Microarray Expression Search (http://www.arabidopsis.org/servlets/Search?action=new_search&type=expression) and analyzed. Signature MPSS data were from http://mpss.udel.edu/at/ (Meyers et al., 2004), and transcript per million values for 17 bp and 21 bp signatures were averaged and then plotted.

Plant material and growth conditions

WT and atagp19 Arabidopsis plants were Columbia-0 ecotype. Plants were grown under 16 h light/8 h dark conditions in soil or on MS plates consisting of 1 × MS (Murashige and Skoog) medium, 1% sucrose and 0.8% agar.

Mutant verification and sequencing

In order to isolate and verify the atagp19 mutant, two AtAGP19-specific primers (5′-CAACAAGATGCATCAAGTCTTACC-3′ and 5′-GTGCTGGTGGTGGTGATACAG-3′), and one T-DNA-specific primer (5′-TGGTTCACGTAGTGGGCCATCG) were designed using the SIGnAL T-DNA verification primer design tool (http://signal.salk.edu/cgi-bin/tdnaexpress). Sequencing reactions were performed using a BigDye Terminator version 3.1 cycle sequencing kit (Applied Biosystems, http://www.appliedbiosystems.com) according to the manufacturer's instructions, and run on an ABI 310 genetic analyzer (Applied Biosystems).

RNA extraction, Northern blotting and RT-PCR

Total RNA was extracted using the RNeasy plant mini kit (Qiagen Inc., http://www.qiagen.com), and on-column DNase digestion was conducted for RT-PCR purposes. Northern blotting was carried out according to the method described by Sun et al. (2005). RT-PCR was performed using a OneStep RT-PCR Kit (Qiagen Inc.) using AtAGP19 mRNA-specific primers 5′-CTGCTCTCATCTCTTCCTTTAGTG-3′ and 5′-ATTGAGCCACATTACTGCTCTTCC-3′.

Transformation

Agrobacterium strain LBA4404 was transformed by electroporation using the GENE PULSER II system (Bio-Rad, http://www.bio-rad.com) according to the manufacturer's instructions. Arabidopsis plants were transformed by the floral dip method (Clough and Bent, 1998).

GUS staining

For the PAtAGP19:GUS construct, the 1.8 kb AtAGP19 promoter sequence was amplified using the primer combination 5′-CCGGTTAATTAATGAAACTGCCTAGTCGGAACCTGA-3′ and 5′-AAATGGCGCGCCTGTGTTGTGGAGGAAGCTACAAGA-3′, and inserted into PacI and AscI sites of the binary vector pMDC164 (Curtis and Grossniklaus, 2003). The same promoter sequence was also used for the complementation experiment described below; this complementation experiment indicated that this promoter successfully drove AtAGP19 expression.

At least one plant from each of the 15 independent transgenic lines harboring the PAtAGP19:GUS fusion was tested, and representative GUS staining patterns are presented. Samples were incubated in a GUS staining solution (0.5 m sodium phosphate, 0.2 mm potassium ferricynide, 0.2 mm potassium ferrocynide, 0.5% Triton X-100 and 2 mm X-GlcA, pH 7.2) at 37°C for 12–24 h. After staining, samples were cleared in absolute ethanol, and images were taken under the ACE light source (Schott-Fostec L.L.C., Auburn, NY, USA) using a Nikon COOLPIX 5400 digital camera connected to a Nikon binocular stereo dissecting microscope SMZ1500 (Nikon, http://www.nikonusa.com).

Phenotypic analyses

Chlorophyll levels in 14-day-old WT and atagp19 plants and anthocyanin content in 24-day-old WT and atagp19 plants were assayed according to published methods (Arnon, 1949; Laby et al., 2000).

Reproduction was examined in mature WT and atagp19 plants immediately after they finished flowering. Stem height was measured with a ruler, and stem width was measured using image-pro software (Media Cybernetics Inc., http://www.mediacy.com). Flowers and fully expanded siliques from same positions on WT and atagp19 plants were compared, and the siliques close to either end of inflorescence stems were avoided.

In order to study leaf cells, fully elongated first leaves (21-day-old plants grown in soil) were cleared in methanol and then in lactic acid. Leaf abaxial epidermal cells and mesophyll cells were drawn using a DIC microscope equipped with a drawing tube. Analyses of epidermal cells were carried out according to the methods described by Autran et al. (2002).

After 7-day-old seedlings grown on horizontal plates were cleared as described above, digital pictures of hypocotyls were taken. Hypocotyls and hypocotyl epidermal cells were analyzed with image j software (http://rsb.info.nih.gov/ji/).

Statistics

Statistical analyses and data plotting and were performed using sigma plot 8.0 (SPSS Inc., http://www.SPSs.com).

Complementation of the atagp19 mutant

The WT AtAGP19 gene sequence, including its endogenous promoter and coding sequence (from the nucleotide at position 1836 upstream of the AtAGP19 start codon to the stop codon), was amplified using primers 5′-CCGGTTAATTAATGAAACTGCCTAGTCGGAACCTGA-3′ and 5′-CCCGCGAGCTCTTAGGCTGTCATAGCAAGTAGAAAG-3′, and cloned into the binary vector pMDC110 between the PacI and SacI sites (Curtis and Grossniklaus, 2003), followed by the nos terminator.

Acknowledgements

This work was supported by a grant from the National Science Foundation (IBN-0110413) and by a Baker Fund Award from Ohio University. T-DNA mutant lines of AtAGP17, 18 and 19 were obtained from the SALK Institute and ABRC. The authors would like to thank Dr Ming Chen, Dr Wenxian Sun and Dr Sarah Wyatt for suggestions and advice, and Dr Ahmed Faik for critical reading of the manuscript.

Ancillary