Regarding a recent article by Embree-Ku and Gruppuso, which appeared in the August 2005 issue of HEPATOLOGY,1 I am concerned the data presented in the paper do not justify the conclusion that the molecule nucleolin binds to the NFκB DNA binding motif.
The authors determined in Fig. 2 that nucleolin binds to this motif. They did this by excising an NFκB oligo gel shift band from a native PAGE gel shift gel and identifying the protein by MALDI analyses. They sought to confirm their conclusion in Fig. 3 by showing that immunodepletion of nucleolin decreases the intensity of the gel shift signal.
These data do not justify the conclusions because nuclear extract proteins often run in native PAGE gels with mobilities that are unaffected or minimally affected by the added DNA, even when bound to the DNA. Even if the authors were to have run their gel shift gel with nuclear extract and added NO DNA, nucleolin is likely to have had the same mobility. It is likely that the identification of nucleolin is a function of its mobility in the gel, not of its binding to DNA.
The authors could have attempted a Western on the band they excised from the EMSA gel to confirm the identity of nucleolin, although this would not have proven that nucleolin binds to the DNA, for the reasons cited in the previous paragraph. But it would have directly shown that nucleolin was in the band analyzed by MALDI, a necessary exercise since MALDI is a matching method with many false positives, especially for very crude samples such as nuclear extracts.
In Fig. 3, the authors did attempt to confirm the identity of nucleolin by comparing the intensities of the gel shift bands for anti nucleolin depleted samples to samples mock-depleted with non-immune IgG. However, nucleolin is such an abundant protein that it is hard to deplete; successful depletion should have been verified by immunoblotting of the supernatant with anti nucleolin to show that the amount of nucleolin was significantly reduced. Another reason this experiment is not conclusive is that the same result would have been obtained if nucleolin sequesters NFκB protein by complexing with it. In this scenario the reduced signal intensity in the gel shift would be attributable to reduced presence of NFκB, which of course binds to the NFκB motif. Nucleolin binds to a number of proteins including several other transcription factors and DNA binding proteins, so this is quite plausible.2–4
Taken together, the data presented do not justify the conclusion that nucleolin binds to the NFκB DNA motif. Nucleolin may bind to this motif, but additional data are needed before this conclusion is warranted.