Peptides as signals - Intercellular Communication in Plants - page 42

indicates that many more peptide signals in plants are yet to be discovered. For
example, plants possess a large number of receptor-like kinases (RLKs). Six hundred
ten RLKs have been identified through analysis of the Arabidopsis genome (Shiu &
Bleecker, 2001). Among these, 216 contain extracellular LRR domains typical of
peptide-binding motifs (Kobe & Deisenhofer, 1994, 1995). The receptors for CLV3,
PSK, and systemin are all the members of LRR-containing RLKs. However, to date,
ligands have only been identified for very few of these plant RLKs.
One of the hurdles in identifying peptides that function as signal molecules is
their low abundance. For example, over 60 lb of tomato leaves was used in puri-
fying the tomato systemin for sequence determination and characterization (Pearce
et al. , 1991). Another hurdle is the lack of an assay for their biological activity.
The discovery of the TomHypSys and TobHypSys systemins was facilitated by the
availability of the alkalinization assay. Without a convenient assay, it would be very
difficult to isolate a new peptide signal using a biochemical approach.
Twoofthe peptide signals, SCR and CLAVATA3, were identified through genetic
approaches. Mutational analysis will continue to contribute to discovery of new
peptide signals. However, most genes encoding peptide signals are likely small.
This small target size reduces the chance to introduce a mutation in those genes.
Even if some of them do encode large precursors, the active form of a peptide signal
could be a short peptide such as PSK. Therefore, a large portion of mutations in
those genes will be silent. As a result, a relatively large mutagenized population
needs to be screened in order to isolate a loss-of-function mutation in a peptide
signal-encoding gene.
Gain-of-function mutagenesis such as activation tagging (Weigel et al. , 2000)
could be a powerful alternative genetic approach to isolate genes that encode peptide
signals. Such an approach has led to the recent identification of an Arabidopsis
gene ( DEVIL1 ) encoding a novel 51-amino acid polypeptide whose gain-of-function
mutation causes changes in a variety of developmental processes, including the leaf
shape, plant stature, and silique development (Wen et al. , 2004). The Arabidopsis
genome encodes 20 DVL1-like proteins, many of which appear to play a biological
role similar to DVL1. The exact biological functions of the members of this gene
family have yet to be determined. Besides, further studies are needed to prove that
DVL1 and its homologues act as signal molecules.
The availability of whole genome sequences for a growing list of plant species
as well as advances in functional genomics and proteomics technology provide a
unique opportunity to identify new peptide signals in a high-throughput fashion. An
in silico search of putative peptide signal-encoding genes is the first step toward this
goal. However, the commonly used gene prediction algorithms predict genes on the
basis of presence of a significant ORF of at least 100 amino acids (Harrison et al. ,
2002). Similar gene prediction algorithms are also used to predict genes from the
Arabidopsis and rice genomes (MacIntosh et al. , 2001; Goff et al. , 2002). Undoubt-
edly, a large number of genuine genes with small ORFs could not be predicted from
those genomes. Therefore, many small ORFs that have been predicted as noncoding
sequences could encode peptide signals.