authors request that the sequence shown in Fig. 2 of the original report should be corrected as follows. As shown schematically here in Fig. 1 A, nucleotides 1452-1504 of the original report have to be replaced by a novel sequence shown here in Fig. 1B. This correction pertains to the 5Ј-flanking region of the gene and does not affect the coding region nor any of the conclusions drawn in the original report. The fact that there was a missing sequence element was discovered and pointed out to us by Tracy L. Bale in the laboratory of Daniel M. Dorsa, Departments of Psychiatry and Behavioral Sciences and Pharmacology, University of Washington, Seattle. As illustrated in Fig. 1 A, the novel sequence has to be inserted at the location of a dinucleotide repeat, (GT) 26 , located 89 nucleotides 5Ј to the main transcriptional initiation site. Resequencing of a newly generated phage subclone as well as Southern blot and PCR analyses (not shown) confirmed that the sequence presented here is indeed part of the genomic sequence. Since this novel sequence element is itself flanked by two dinucleotide repeats, (GT) 20 and (GT) 24 , respectively, a likely explanation is that this segment was spliced out during subcloning due to recombination between the two dinucleotide repeats. This idea is further supported by the fact that dinucleotide repeats that have the potential of forming Z-DNA structures have been shown to enhance recombination in extrachromosomal DNA up to 20-fold (1). Despite the recurrence of dinucleotide repeats around chromosomal rearrangement breakpoints, their role in mediating recombination on intact chromosomes remains, however, uncertain (2).We thank Tracy L. Bale and Daniel M. Dorsa for pointing out the error and for their help and collaboration in its correction. Fig. 3, A and C, are of (M ϩ 33H) 33ϩ .17. Mariman, E. C. M., Broers, C. A. M., Claesen, C. A.