Regulated expression of proteins involved in mammalian iron metabolism is achieved in part through the interaction of the iron regulatory proteins IRP1 and IRP2 with highly conserved RNA stem-loop structures, known as iron-responsive elements (IREs), that are located within the 5' or 3' untranslated regions of regulated transcripts. As part of an effort to determine the structures of the IRP-IRE complexes using crystallographic methods, we have developed an efficient process for obtaining functionally pure IRP1 and IRP2 that relies upon the improved overexpression (>10 mg of soluble IRP per liter of culture) of each human IRP in the yeast Pichia pastoris and large-scale purification using RNA affinity chromatography. Despite the utility of RNA affinity chromatography in the isolation of RNA-binding proteins, current methods for preparing RNA affinity matrices produce columns of low capacity and limited stability. To address these limitations, we have devised a simple method for preparing stable, reusable, high-capacity RNA affinity columns. This method utilizes a bifunctional linker to covalently join a 5'-amino tethered RNA with a thiol-modified Sepharose, and can be used to load 150 nmole or more of RNA per milliliter of solid support. We demonstrate here the use of an IRE affinity column in the large-scale purification of IRP1 and IRP2, and suggest that the convenience of this approach will prove attractive in the analysis of other RNA-binding proteins.

Much of the nonrandom usage of V, D, and J genes in the Ab repertoire is due to different frequencies with which gene segments undergo V(D)J rearrangement. The recombination signal sequences flanking each segment are seldom identical with consensus sequences, and this natural variation in recombination signal sequence (RSS) accounts for some differences in rearrangement frequencies in vivo. Here, we have sequenced the RSS of 19 individual V(H)7183 genes, revealing that the majority have one of two closely related RSS. One group has a consensus heptamer, and the other has a nonconsensus heptamer. In vitro recombination substrate studies show that the RSS with the nonconsensus heptamer, which include the frequently rearranging 81X, rearrange less well than the RSS with the consensus heptamer. Although 81X differs from the other 7183-1 genes at three positions in the spacer, this does not significantly increase its recombination potency in vitro. The rearrangement frequency of all members of the family was determined in mu MT mice, and there was no correlation between the in vitro recombination potential and VH gene rearrangement frequency in vivo. Furthermore, genes with identical RSS rearrange at different frequencies in vivo. This demonstrates that other factors can override differences in RSS potency in vivo. We have also determined the gene order of all V(H)7183 genes in a bacterial artificial chromosome contig and show that most of the frequently rearranging genes are in the 3 ' half of the region. This suggests that chromosomal location plays an important role in nonrandom rearrangement of the V(H)7183 genes.