Kurt NordstrBm Department of Microbiology Biomedical Center Uppsala University S-751 23 Uppsala Sweden Plasmids control their own replication; that is, during ex- ponential growth of a bacterial population, they are always present in defined copy numbers that are controlled by the plasmids themselves. The copy number may be different for the same plasmid in different hosts and, for the same plasmid-host combination, the copy number may be different under different growth conditions. At steady state, the number of replications per plasmid copy is exactly one per cell generation. This defines the steady state as shown in Figure 1. It should be stressed that while this is true for the population, it is not for in- dividual plasmids-these are selected randomly for repli- cation. Principles of Copy Number Control If the copy number is for some reason higher than the steady-state value, replication frequency is adjusted to be- come lower than one per plasmid copy (and cell genera- tion; Figure 1). At reduced copy numbers, the opposite is true; replication frequency exceeds one per plasmid copy and cell generation. Hence, the control curve falls within the hatched areas of Figure 1. This ensures that the steady-state value is retained and deviations are adjusted. Figure 1 illustrates that plasmids measure their copy number and use this measurement to adjust the replica- tion frequency to maintain a defined copy number. Both these processes use the same negative control circuit. The negative control element can be a repressor protein, antisense RNA, or direct repeats of DNA called iterons (see table; reviewed in Nordstrom, 1985). Copy Number Control by Repressor Protein Plasmid Xdv uses a repressor protein to adjust its replica- tion frequency (Figure 2A). Replication of hdv is limited by the availability of 0 protein, transcriptional activation of the replication origin (ori, which is situated within the 0 gene), or both. The negative control is exerted by the gene product of the first gene on the 0 operon, the Cro (or Tof) repressor. Hence, synthesis of Cro protein is autoregu- lated, which in turn ensures that copy number is con- trolled. The general properties of this control system have been discussed previously (Nordstrdm, 1985). Computer studies based on data about li indicate that the kinetics are essentially hyperbolic (Figure 1; Lee and Bailey, 1984). Using essentially the same data except for the values for some transcription and translation rate constants, a con- trol curve is predicted that is significantly less steep than that of Lee and Bailey (Figure 1; Womble and Rownd, 1988a); this would give a slower adjustment of the copy number from deviations back to the steady-state value than a hyperbolic response would.