In Chlamydomonas chloroplasts, atpB pre-mRNA matures through a two-step process. Initially, endonuclease cleavage occurs 8 -10 nt downstream of the mature 3 end, which itself lies at the end of a stem-loop-forming inverted repeat (IR) sequence. This intermediate product is then trimmed by a 3 3 5 exonuclease activity. Although the initial endonucleolytic cleavage by definition generates two products, the downstream product of atpB pre-mRNA endonucleolytic processing cannot be detected, even transiently. This product thus appears to be highly unstable, and it can be hypothesized that specific mechanisms exist to prevent its accumulation. In experiments described here, the atpB 3 maturation site was placed upstream of reporter genes in vivo. Constructs containing both the IR and endonuclease cleavage site (ECS) did not accumulate the reporter gene mRNA, whereas constructs containing only the IR did accumulate the reporter mRNA. The ECS alone gave an intermediate result, suggesting that the IR and ECS act synergistically. Additional secondary structures were used to test whether 5 3 3 and/or 3 3 5 exonuclease activities mediated degradation. Because these structures did not prevent degradation, rapid endonucleolytic cleavages most likely trigger RNA destruction after ECS cleavage. On the other hand, fragments resulting from cleavage within the endogenous atpB mRNA could occasionally be detected as antisense transcripts of the adjacent reporter genes. Because endonuclease cleavages are also involved in the 5 maturation of chloroplast mRNAs, where only the downstream cleavage product accumulates, it appears that chloroplast endoribonuclease activities have evolved mechanisms to selectively stabilize different ECS products.Maturation of mRNA can involve multiple steps in both prokaryotic and eukaryotic systems, and results in functional transcripts with a stability and subcellular localization appropriate to their functions. RNA sequence and secondary structure, along with ribonucleases and a variety of accessory factors, are used to achieve these goals. Built into this process is the necessity to recognize nonfunctional RNAs and eliminate them; destruction of nonsense codon-containing mRNAs is a good example of the complexity of such surveillance mechanisms (reviewed in Ref. 1).Our laboratory has focused on 5Ј end and 3Ј end maturation of chloroplast mRNAs, using both vascular plants and the unicellular green alga Chlamydomonas reinhardtii as models. In this respect, chloroplast transcripts have mostly prokaryotic features such as the lack of a trimethylguanosine 5Ј cap, a 3Ј stem-loop-forming inverted repeat (IR) 1 structure, and they are destabilized by polyadenylation (reviewed in Refs. 2 and 3). Processing of chloroplast mRNAs, with rare exceptions, depends on nucleus-encoded proteins, several of which have been identified genetically through screens for plants or Chlamydomonas strains unable to carry out photosynthesis. The phenotypes suggest that defects in intercistronic processing of polycistronic transcripts...