Senescence, aging and death - conceived of in the past as inevitable, negative processes - are now considered an integral part of differentiation and development. Leaf senescence is one of the most conspicuous processes to have been studied in the context of plant aging and senescence, and has an important impact on agriculture, affecting crop yield and the shelf life of leafy vegetables 12. This terminal phase of leaf development cannot be described simply as a collection of passive and deteriorative processes during which a gradual decline in vital systems takes place. Extensive physiological and biochemical studies on leaf senescence in the last three decades have suggested that it is a highly regulated and active process, which is characterized by differential and sequential changes in almost every subcellular compartment. Leaf senescence involves diverse metabolic changes associated with multiple biochemical pathways 123. Guo et al. 4 now report a comprehensive study on the transcriptional program of Arabidopsis leaf senescence. This study not only provides genetic confirmation for many previous biochemical and physiological findings, but also adds significant new information indicating that leaf senescence, like other developmental processes, is a very dynamic, complex and active program that requires the activation of many genes.

In both plants and animals, programmed cell death (PCD) plays a crucial role, mainly during development and differentiation. Although there is insufficient information, at this stage, to postulate that animals and plants share common and basic regulatory mechanisms of PCD, similar biochemical and cellular changes are often displayed in both systems. An example of a process involving PCD in plants is the formation of the xylem elements that die and lose their content in order to conduct water and solutes. The autolysis of cells in roots, cell death during pollination, embryo development and seed maturation are also referred to as depending on PCD. Leaf senescence displays the three criteria suggested by Barlow 5 to be necessary to define it as a plant PCD process: first, cells die at a predictable time and location; second, death has some beneficial effect on plant development; and third, cell death is encoded in the hereditary material. This definition excludes necrotic cell death due to accidental damage, or injury as a result of exposure to a toxic environment.

Although leaf senescence can generally be defined as a late developmental process leading to death, the primary molecular pathway of the senescence program is not known. Leaf yellowing due to chlorophyll degradation is often considered to be the main marker for leaf senescence. Chlorophyll breakdown may also indicate the early disintegration of the photosynthetic machinery localized in the chloroplast. The drastic biochemical transition characterizing the onset of leaf senescence, however, requires more than 'just' deterioration of the activity of existing proteins or downregulation of gene expression. Indeed, recent genomic and molecular studies support the notion that the onset of senescence involves de novo synthesis of proteins and the expression of a complex array of genes whose products are involved in, and are responsible for, the multitude of senescence-related biochemical and cellular changes 4678.

The comprehensive transcriptome study by Guo et al. 4 provides the largest available list (6,200) of senescence-associated expressed sequence tags (ESTs), representing approximately 2,500 genes, in Arabidopsis leaves. It has been hypothesized that the ESTs found in senescing tissues represent genes that are expressed in the fully senescent leaf, when all transcripts present in non-senescent leaves have already been degraded. Given that the current database of Arabidopsis ESTs does not represent senescing tissues, it is perhaps not surprising that an additional 100 new genes, not found in the public databases, were identified in this study.

As in many other genomic studies, the biological relevance of the senescence-associated transcriptome study of Guo et al. 4 depends on our ability to predict the function of individual genes. Functional assignment of the genes represented in their senescent-leaf EST collection was carried out using the sequence annotation and classification of genes in the Arabidopsis databases. The general picture emerging from this study is that leaf senescence is indeed, like any other developmental process, a very dynamic and complex phenomenon that involves the activation of a large number of genes representing a broad spectrum of functional categories. The relative abundance of transcripts represented by the various categories during senescence differs substantially compared to those represented at other developmental stages, however. For example, massive degradation of cellular components, a distinct feature of leaf senescence, is reflected by the high ratio of the number of genes for primary catabolism over those for anabolism found in the senescence EST database (1.84) as compared to this ratio for the entire Arabidopsis genome (0.57).

The main functional categories of the senescence-related ESTs as reported by Guo et al. 4 are summarized in Table 1. These data provide important clues to the regulatory and metabolic processes associated with leaf senescence. Macromolecule degradation is reflected by the upregulation of several groups of genes whose products are responsible for the intensive degradation of proteins, lipids, nucleotides and polysaccharides during senescence. Seven percent of the ESTs are from genes involved in proteolysis; among these, those encoding cysteine and other types of proteases are prominent. Upregulation of the ubiquitin/polyubiquitin genes and genes whose products are associated with the activation and ligation reactions of the ubiquitination pathway is evidence for proteolysis via the ubiquitin pathway 467. Genes encoding the components of the proteosome have also been demonstrated to be actively upregulated during leaf senescence 7. These results support the suggestion that senescence may be regulated by the ubiquitin pathway through the breakdown of negative regulatory molecules 9.

The dramatic biochemical shift from a photosynthetically active organ into a senescing leaf is induced by an array of endogenous factors, such as age and hormonal regulation, as well as by external factors, such as biotic and abiotic stresses. It is not known how these signals are perceived by the plant, but the study of Guo et al. 4 identified 182 genes encoding apparent components of signal perception and transduction pathways. The putative senescence-induced signals are, as in other developmental processes, likely to be perceived by signaling molecules belonging to various classes. Candidates include plant receptor kinases - transmembrane kinases that have been implicated in ligand perception 10. Genes encoding senescence-associated receptor-like kinases (SARK and SIRK) have also been identified in bean and Arabidopsis 1112. Among the 610 genes encoding receptor-like proteins found in the Arabidopsis genome, 44 are expressed in the senescing leaf 4.

Signal-transduction pathways linked to the components that perceive the senescence signals are predicted to trigger a cascade of events inside the cells. Signaling cascades frequently involve the addition and removal of phosphate groups (phosphorylation and dephosphorylation) from cellular proteins. Indeed, a limited number of genes encoding components of protein-phosphorylation cascades have been identified in the senescing leaf, and prominent among these are genes encoding members of the mitogen-activated protein (MAP) kinase cascade 4. MAP kinase cascades are known to link extracellular stimuli to a wide range of cellular responses in animal cells and yeast, and may be involved in the senescence program as well.

Following the signaling pathways downstream, the targets are likely to be transcription factors that can act as switches to initiate differential gene expression upon binding to specific cis-elements of target-gene promoters. The 134 genes encoding transcription factors identified by Guo et al. 4 represent 5.4% of the total number of senescence-associated genes. These genes provide a key to the understanding of the regulatory pathways of the senescence program. Furthermore, subsets of target genes are regulated by specific transcription factors, helping, in turn, to identify the genes expressed as the final output of the pathway. The two largest groups of senescence-related transcription factors are NAC transcription factors, which exist exclusively in plants and have previously been shown to control organ development and response to pathogens, and the WRKY transcription-factor group, which are also known to control the response to pathogens and are elicited by salicylic acid. The involvement of WRKY6, WRKY18, WRKY22/29 and WRKY53 in leaf senescence has already been demonstrated, and the expression of members of this group is upregulated both in defense responses and during leaf senescence 12. Interestingly, zinc-finger proteins and other transcription-factor families have been identified in the senescence EST collection of Guo et al. 4, whereas none of the MADS-box transcription factors known to participate in flower development were found to be expressed during senescence.

Some orthologs of the senescence proteases found by Guo et al. 4 have also been identified in autumn leaves of the Populus tree 6. A recent genomic study employing EST sequencing and microarrays of gene expression during autumn leaf senescence of Populus trees suggests that, as during leaf senescence of annual plants, there is a dramatic shift in gene expression reflecting the transition from anabolic to catabolic processes, chlorophyll degradation, oxidation of fatty acids and nutrient mobilization 8.

In addition to the significant information regarding the existence and nature of the biochemical pathways and regulatory mechanisms involved in leaf senescence, the vast collection of genes described in these recent transcriptome studies 4678 also provides the basis for future reverse-genetic studies of the senescence program. We can look forward to insights into the molecular basis for leaf senescence and, ultimately, elucidation of the complex pathways involved.