Reply to Bagni: On BC1 RNA and the fragile X mental retardation protein

In Iacoangeli et al. (1), five independent groups report that results published by Zalfa et al. (2) are not reproducible. Bagni now suggests (3) that different reagents, antibodies, or procedures might explain this lack of reproducibility. Iacoangeli et al. replicated the experimental conditions reported by Zalfa et al. whenever possible, as indicated. In several cases, however, reagents used by Zalfa et al. were not available. For instance, Zalfa et al. generated antibody rAM1 and used it to probe BC1–FMRP interactions in supershift and other assays. The antibody did not produce a supershift in brain extracts—as one would have expected if BC1 RNA did in fact bind to FMRP in vivo—but abolished the regular mobility shift (2). Despite repeated requests by several of the undersigned, antibody rAM1 was not provided. We urge the Bagni group (2) to make antibody rAM1 available for independent examination. In contrast, Iacoangeli et al. used two established anti-FMRP antibodies that have been independently validated and are publicly available. Bagni suggests that two other groups (4, 5) have published data in support of her claims. Both articles were quoted and discussed by Iacoangeli et al. (1). One of the undersigned (E.W.K.) coauthored one of these articles (4) and has confirmed that, although BC1 RNA does bind to FMRP in vitro, this binding is entirely nonspecific because it is completely abrogated by competitor tRNA (1). Bagni does not mention that Gabus et al. (4) reported a K_d of FMRP for tRNA of 25 nM, the same as for BC1 RNA. The second article (5) reported a weak BC1 RT-PCR signal in FMRP cross-linked immunoprecipitates and a stronger signal in MAP2 cross-linked immunoprecipitates. However, considering that MAP2 is not a known RNA binding protein, mere copurification cannot be taken as evidence for physical association (1). Thus, there is no confirmation, independent of the Bagni group, of a specific physical link between FMRP and BC1 RNA, as posited by Zalfa et al. (2). Similarly, the Zalfa et al. claim that FMRP is not associated with polyribosomes in neurons could not be confirmed in subsequent work (6, 7).