The specific hypotheses of how FKF1 and ZTL affect flowering time and circadian clock function can be tested at the biochemical level. For example, by determining whether these proteins bind flavin and act directly as blue light receptors. Also the proteins that interact with the kelch repeats of FKF1 and ZTL need to be identified, to test whether they are targeted for degradation. This might also address the specificity of FKF1 and ZTL, because although they have similar arrangements of motifs and related mutant phenotypes, there are significant differences in the flowering time, circadian clock and light response phenotypes that might be explained by the specificity of the protein partners. Similarly, what is the role of the third homologue, FKL1/LKP2 (Refs 8.xZEITLUPE encodes a novel clock-associated PAS protein from Arabidopsis. Somers, D.E. et al. Cell. 2000; 101: 319–329Abstract | Full Text | Full Text PDF | PubMedSee all References, 9.xFKF1, a clock-controlled gene that regulates the transition to flowering in Arabidopsis. Nelson, D.C. et al. Cell. 2000; 101: 331–340Abstract | Full Text | Full Text PDF | PubMedSee all References) for which a mutant has not yet been described?The patterns of inheritance of fkf1 and ztl are slightly different: deletion of FKF1 causes a recessive mutation whereas ztl is co-dominant. This might be because the reported ztl alleles are loss-of-function alleles and ZTL is haploinsufficient (i.e. dominant simply because the heterozygote carrying one active copy of the gene does not produce sufficient ZTL protein). However, an alternative possibility is that the specific ztl alleles described, both of which affect the kelch repeats, generate a gain-of-function phenotype8xZEITLUPE encodes a novel clock-associated PAS protein from Arabidopsis. Somers, D.E. et al. Cell. 2000; 101: 319–329Abstract | Full Text | Full Text PDF | PubMedSee all References8. Similarly, plants overexpressing ZTL and plants homozygous for the ztl mutant alleles are late flowering9xFKF1, a clock-controlled gene that regulates the transition to flowering in Arabidopsis. Nelson, D.C. et al. Cell. 2000; 101: 331–340Abstract | Full Text | Full Text PDF | PubMedSee all References9, and this might also suggest that the ztl mutations are not loss-of-function alleles.More generally, although ZTL and FKF1, as well as those genes shown in Table 1Table 1, identify common components in the regulation of the circadian clock and flowering time, it is often not clear that their effects on circadian clock regulation are the cause of the flowering time defects. The genes might act independently on both processes. For example, ztl and fkf1 have similar effects on flowering time, but different effects on the expression of circadian-clock-regulated genes. Similarly, the gi-1 and gi-2 mutations have similar effects on flowering time, but one causes a short period phenotype and the other a long period phenotype in circadian clock-controlled-gene expression7xControl of circadian rhythms and photoperiodic flowering by the Arabidopsis GIGANTEA gene. Park, D.H. et al. Science. 1999; 285: 1579–1582Crossref | PubMed | Scopus (346)See all References7. Identification of genes that act downstream of the circadian clock to promote flowering, and analysis of their expression in different mutant backgrounds will help to address these issues. Such genes might be identified by other mutations that disrupt the control of flowering by daylength, or might emerge from new approaches, such as the use of microarray technology16xChasing the dream: plant EST microarrays. Richmond, T. and Somerville, S. Curr. Opin. Plant Biol. 2000; 3: 108–116Crossref | PubMed | Scopus (134)See all References16, to identify all circadian-clock-regulated genes.