ARTICLE

TOP ARTICLE LITERATURE CITED

We have identified a large family of genes with sequence similarity to the cell wall-associated kinase (WAK) genes (He et al., 1999). Like the WAKs, these genes exist in multiple gene clusters, and our analyses suggest that they encode functional protein kinases that are associated with the cell wall. The WAKs represent a unique class of receptor-like kinase genes. Each encodes a transmembrane protein with a cytoplasmic Ser/Thr kinase (STK) domain and an extracellular region with similarity to vertebrate epidermal growth factor (EGF)-like domains. WAKs are thought to physically link the extracellular matrix and the cytoplasm and to serve a signaling function between them (He et al., 1996; Kohorn, 2000). Consistent with these ideas, WAK1 is covalently bound to pectin in the cell wall (Wagner and Kohorn, 2001) and can form an approximately 500-kD protein complex via interactions with a Gly-rich extracellular protein, AtGRP-3 (Park et al., 2001) and a cytoplasmic type 2C protein phosphatase, KAPP (Anderson et al., 2001).

Previous findings indicate that WAKs are involved in the response to pathogens (He et al., 1998). Induction of WAK1 is required for plants to survive Pseudomonas syringae infection. In addition, WAK1 is induced by salicylic acid in an NPR1-dependent manner (nonexpresser of pathogenesis-related genes), demonstrating that it is a pathogenesis-related gene. Moreover, WAK1 is up-regulated during systemic acquired resistance (Maleck et al., 2000) and is induced by the fungal pathogen Alternaria brassicicola and the defense-related signaling molecules methyl jasmonate and ethylene (Schenk et al., 2000).

WAKs have recently been shown to be required for cell expansion (Lally et al., 2001; Wagner and Kohorn, 2001). For example, Lally et al. (2001) used a glucocorticoid-inducible system to control expression of a WAK4 antisense gene. Induction of WAK4 antisense expression caused a decrease in WAK protein levels that could be controlled by the concentration of inducer applied. Reduction of WAK protein levels resulted in inhibition of cell elongation, and the degree of inhibition was correlated with the concentration of inducer applied (Lally et al., 2001).

Reiterative database searches (BLAST) using the WAK1 cDNA or WAK1 protein sequences as queries identified a large family containing 22 genes in Arabidopsis similar to WAKs (Table I). We have called these sequences WAK-like genes (WAKLs), as suggested by Shiu and Bleecker (2001). Intron-exon junctions were verified by identifying and sequencing cDNA clones for each of the WAKL genes. With the exception of three sequences (see below), all the intron-exon predictions were correct.

WAKL3 (At1g16140) is predicted to have three introns; however, our analysis shows that there are only two. The first intron (17 bp), predicted to reside between nucleotides 64,385 and 64,403 on the bacteria artificial chromosome (BAC) clone (T24D18), was not present. This results in a shift in the reading frame at the 5' end of the first exon. As such, the first "ATG" codon in the longest open reading frame (ORF) is predicted to reside at nucleotide 64,155. WAKL6 (At1g16110) was predicted to have a stop codon at nucleotide 55,719 on BAC T24D18; however, sequencing of a WAKL6 cDNA shows that the stop codon was not present. The stop codon for WAKL6 resides at nucleotide 55,984, resulting in a 2,196-bp ORF. WAKL8 (At1g16260) is predicted to have three exons and two introns, similar to other WAKLs. Sequencing of a WAKL8 cDNA shows that only the first intron is present. This intron is predicted to extend from nucleotides 16,177 to 16,098 (3'-5') on BAC F3O9; however, our analysis shows that it extends from 16,177 to 16,120. In addition, a stop codon is present at nucleotide 16,093, giving rise to a truncated ORF of 879 bp. These corrected sequences were used in all subsequent analyses.

Five of the WAKL genes (WAKL7, WAKL8, WAKL12, WAKL16, and WAKL19) are predicted to encode abbreviated WAKL proteins. The remaining 17 genes share similar intron-exon organization with the WAKs, with each having three exons and two introns (Fig. 1). Each of the non-truncated WAKLs is predicted to encode a transmembrane protein containing a cytoplasmic STK (ProSite: PS00108. ProSitea database for protein families and domainshttp://www.expasy.ch/prosite/) and an extracellular region containing a calcium-binding EGF-like domain (EGF-Ca²⁺; PS01187) and/or an EGF2-like domain (EGF2; PS01186). In some cases (see below), the EGF-like domains were slightly degenerate. In these cases, all the Cys residues predicted to be involved in the formation of disulfide bridges were conserved; however, the spacing between them was off by one to two residues or other hallmark residues characteristic of EGF-like domains were changed. Because EGF domains have not been well characterized in plants, it is not clear what effect, if any, this degeneracy has on their potential function.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Intron-exon structure is conserved between the WAK and WAKL genes. The diagram shows a standardized depiction of a WAK or WAKL gene from each of the four groups (I-IV). Exons are represented by boxes. Introns are represented as `V's. Regions of each gene encoding functional domains are indicated with shaded boxes: N-terminal signal sequence (black), EGF2-like domain (red), calcium-binding EGF domain (blue), transmembrane domain (green), and Ser/Thr protein kinase active site (orange).

We have divided the WAKs and WAKLs into four groups based on pairwise comparisons of their predicted protein sequences (Fig. 2). Group I contains five members, WAK1 through WAK5 (He et al., 1999). WAK1 through WAK5 all have an EGF-Ca²⁺ domain and an overlapping Asn hydroxylation site (PS00010). Immediately following the EGF-Ca²⁺ domain is an EGF2 domain, which is degenerate in both WAK4 and WAK5. These domains are completely encoded by the second exon (Fig. 1). Group II includes seven members, WAKL1 through WAKL6 and WAKL22. Their predicted proteins all contain EGF-Ca²⁺ and EGF2 domains (both of which are degenerate in WAKL1, WAK6, and WAKL22). In addition, the domains are separated by a short gap of 15 to 18 amino acids and are in reversed order relative to Group I. The regions of the genes encoding them are split by the first intron (Fig. 1). Group III contains six members: WAKL9, WAKL10, WAKL11, WAKL13, WAKL17, and WAKL18. Their corresponding proteins all contain EGF-Ca²⁺ and EGF2 domains, and they are structurally similar to the Group II WAKLs. In WAKL13, the EGF-Ca²⁺ domain is degenerate. With the exception of WAKL17, all have degenerate EGF2 domains (Fig. 1). Group IV contains four members: WAKL14, WAKL15, WAKL20, and WAKL21. Each has an EGF2 domain encoded by the first exon. This domain is degenerate in both WAKL20 and WAKL21. All four members lack the EGF-Ca²⁺ domain (Fig. 1). In addition, each has a cytoplasmic protein kinase ATP-binding domain (PS00107).

View larger version (22K): [in this window] [in a new window]   Figure 2.   Dendrogram showing the relationships between the various WAKs and WAKLs. The tree was generated by ClustalW analysis with the corrected full-length WAK/WAKL protein sequences using the MacVector software package (version 7.0; Accelrys, San Diego). The bar beneath the dendrogram represents a distance of 0.1 change per amino acid.

The remaining sequences (WAKL7, WAKL8, WAKL12, WAKL16, and WAKL19) are predicted to encode abbreviated WAKL proteins. WAKL7, WAKL8 and WAKL19 are similar to various other WAKLs in their extracellular regions, and lack a transmembrane domain. WAKL8 and WAKL9 both contain an EGF-Ca²⁺ domain and WAKL19 contains a degenerate EGF2 domain. Neither of these domains is present in WAKL7. WAKL12 also contains an EGF-Ca²⁺ domain, but unlike WAKL8, it contains a transmembrane domain. WAKL16 contains a transmembrane domain, an STK domain that is most similar to WAK3, and a short extracellular domain of eight amino acids that lacks both of the EGF-like domains. Recent evidence form our laboratory shows that at least three of these genes (WAKL7, WAKL8, and WAKL19) are expressed (J.A. Verica and Z.H. He, unpublished data). Their protein products are likely secreted from the cell. This raises the possibility that they may play a role in the formation of an active WAKL receptor complex. A similar role has been proposed for SCR/SP11 in the SRK receptor complex in self-incompatible Brassica spp. (Schopfer et al., 1999; Takayama et al., 2000). Alternatively, these sequences may encode nonfunctional WAKL isoforms. Potential roles for the remaining two sequences (WAKL12 and WAKL16) remain to be determined.

Expressed sequence tags (ESTs) have been identified for many of the WAK and WAKL genes (Table I). In addition, we have isolated cDNAs for all of the non-abbreviated members. Bacterial expression of several WAKL members shows that they have autophosphorylation activity, and analysis of several of the WAKL proteins shows that they localize to the cell wall (J.A. Verica and Z.H. He, unpublished data). This suggests that the WAKLs, like the WAKs, are protein kinases that are tightly associated with the cell wall.

The WAK and WAKL genes are distributed among all five chromosomes (Fig. 3), with the majority (19) being present on chromosome I. Sixteen of the genes (WAK1-WAK5, WAKL1-WAKL8, and WAKL11-WAKL13) are located on the upper arm within a region spanning less than 12 cM. Moreover, WAK1 through WAK5 and WAKL1 through WAKL7 are both present as two separate clusters of tandemly arrayed genes (He et al., 1999). WAKL9, WAKL10, and WAKL22 are located on the lower arm, with WAKL10 and WAKL22 being adjacent to the telomere. The remaining six genes are distributed more or less evenly among chromosomes II through V. Chromosome II contains only one gene on its lower arm, WAKL14. Chromosome III contains two genes (WAKL15 and WAKL16), one on each arm. Chromosome IV contains three genes. WAKL19 is located on the upper arm adjacent to the telomere. WAKL17 and WAKL18 are located on the lower arm as two tandem genes. Chromosome V contains two genes, WAKL20 and WAKL21, that are located at opposite ends of the chromosome in regions directly adjacent to the telomeres.

View larger version (22K): [in this window] [in a new window]   Figure 3.   The WAK and WAKL genes are distributed among all five chromosomes. Chromosomes (I-V) are indicated by the vertical bars. Centromeres are indicated by the darkened circles. Horizontal bars indicate the location of each of the WAKL/WAK genes, and their physical position is given in centiMorgans.

Closer examination of the Arabidopsis genome reveals several potential explanations for the expansion of the WAK gene family. For example, the arrangement of and the high sequence similarity between the genes in the WAKL1 through WAKL6 cluster suggest that they may have arisen via tandem duplication. Likewise, the genes in the WAK1 through WAK5 cluster may have independently arisen via a similar scenario, although other possibilities cannot be ruled out. For example, the regions flanking the WAK1, WAK2, WAK4, WAK5, WAKL11, and WAKL16 contain sequences that are similar to the copia-like retrotransposon Hopscotch (White et al., 1994) or non-long terminal repeat retroelements (Kumar and Bennetzen, 1999). These retroelements transpose via RNA intermediates that are generated by reverse transcription. In some cases, adjacent genes (or portions of genes) may be acquired by a retrotransposon, resulting in their amplification and dispersal throughout the genome (Kumar and Bennetzen, 1999). If these retroelements are functional in Arabidopsis, this could be one mechanism for the expansion of the WAKL gene family. A third possibility arises from the observation that the Arabidopsis genome contains extensive duplications of large chromosomal segments (Arabidopsis Genome Initiative, 2000; Blanc et al., 2000). Several of the WAKL genes are contained within these regions. For example, the region overlapping WAKL11 and WAKL13 on the lower arm of chromosome I is duplicated on the upper arm in the region overlapping the locations of WAKL9 and WAKL10, suggesting that these genes could have arisen via segmental duplication. The high degree of sequence similarity between these genes is consistent with this possibility.

The existence of the WAKs and WAKLs is not unique to Arabidopsis. For example, protein gel blots using the WAK1 antibody as a probe revealed immunologically related proteins in pea (Pisum sativum), tobacco (Nicotiana tabacum), and maize (Zea mays; He et al., 1996; Gens et al., 2000). Expressed sequence tags for WAK-like genes have been identified in tomato (Lycopersicon esculentum; GenBank accession nos. AW220490, AW455238, and AW622503) and soybean (Glycine max; GenBank accession no. BE473800). In addition, GenBank searches have revealed that there are at least 10 WAK-like genes in rice (Oryza sativa; accession nos. AC079685, AF327447, AL442007, and AP003021). Like the Arabidopsis WAKs, many of the rice WAKs are clustered. Moreover, they are also flanked by transposon-like elements (MITEs, miniature inverted-repeat transposable elements; Bureau et al., 1996). These genes are more similar to WAK genes than they are to any of the WAKL genes, suggesting that the WAKLs may be unique to Arabidopsis. Completion of the rice genome and further sequencing of other plant genomes may help to resolve this issue.

The large number of genes in this family may provide Arabidopsis with the potential to recognize and respond to a diverse array of ligands. The observations that WAKs play roles in both the pathogen response and cell elongation suggest that they function in some manner that is common to both processes. For example, both processes involve alteration of the cell wall (Cosgrove, 1997; Grant and Mansfield, 1999). As such, the tight association of WAKs and WAKLs to the extracellular matrix could allow them to function by responding to architectural changes that occur within the wall.