We are pleased that Dr. Bernstein took an interest in the data identifying nucleolin presented in our paper.1 Nucleolin binding to the canonical NFκB target sequence was not a central theme of the paper, but rather an issue we felt compelled to address because its presence explained the only dark band on the EMSAs for NFκB using fetal nuclear extract. A considerable amount of time and effort was spent identifying nucleolin. However, because of space limitations and its not being the focus of the paper, we briefly summarized our work. We are sorry if this led to any confusion.

Dr. Bernstein may have misunderstood the methods used to purify and identify the protein contributing to the lower band on the NFκB EMSAs. We did not “excise the NFκB oligo gel shift band from a native PAGE gel shift gel”. Rather, we used a conventional DNA-affinity purification scheme. E19 nuclear extracts were subjected to heparin chromatography followed by DNA-affinity chromatography using the immunoglobulin κ promoter sequence (see Results, Identification of the lower band in NF-κB EMSAs1). The fractions were assayed using the NFκB EMSA, and those with the highest activity were run on SDS-PAGE. The most prominently silver-stained bands from the fractions with the highest EMSA binding activity were subjected to MALDI-TOF. Six bands were identified as nucleolin and its degraded products. However, NFκB was not one of the proteins identified by this survey of proteins purified by DNA affinity chromatography.

Additionally, although not described in detail in the paper, we performed UV cross-linking to identify the approximate size of the protein(s) of interest and mutational analysis of the target DNA used for the EMSAs to determine the sequence relevant for binding of the protein(s) producing the lower band. The UV cross-linking identified several proteins/complexes with molecular weights around 46, 48, 106 and 108 KDa, consistent with nucleolin and its degraded products.2 Mutational analysis of the target oligonucleotide used for the EMSAs revealed that all mutations in the canonical NFκB binding site and not the flanking DNA affected the binding of NFκB (the upper two bands on our EMSAs), In contrast, only the most 3′ mutations in the canonical NFκB binding sequence and some flanking DNA in the oligonucleotide affected binding of the lower band. These results, combined with the fact that the lower band on the gel did not supershift with any of the NFκB antibodies used, make it unlikely that the lower band contains NFκB. Finally, we were able to immunodeplete the protein responsible for the lower band using anti-nucleolin antibodies.

Given these results, we were left to conclude that the only protein species accounting for significant NFκB binding in the fetal liver extracts was nucleolin. Within the context of our studies, this was important because it supported the conclusion that NFκB is not active in late gestation fetal liver. We hope that our comments are adequate to address Dr. Bernstein's concerns.


  1. Top of page

Michelle Embree-Ku*, Philip A. Gruppuso*, * Department of Pediatrics, Rhode Island Hospital and Brown University, Providence, RI.