We used 2D 1H-15N heteronuclear single quantum correlation (HSQC) spectroscopy to monitor interactions between BAF and its putative binding partners by NMR. In general, each cross-peak in the 1H-15N HSQC spectrum corresponds to one amino acid in the 15N-labeled protein. Changes in local environment caused by altered conformation or interaction with other proteins shift or alter the intensity of related cross-peaks. The HSQC spectrum therefore represents a fingerprint of the system and can be used to map binding interfaces. shows the changes in the 1H-15N HSQC spectrum of LEM domains upon interaction with BAF. shows the 1H-15N HSQC spectrum of the LEM domain of Emerin. Upon incubation with BAF there are major changes in many of the peaks corresponding to residues that interact with BAF (). Comparison of the 1H-15N HSQC spectrum of the 15N-labeled LEM domain of MAN1 () with the same domain bound to BAF () also reveals considerable differences.
1H-15N HSQC spectrum of the Emerin (panel A and B) and MAN1 (panel C and D) LEM domains in the absence and presence of BAF.
We initially set out to probe the interaction surfaces of BAF and HIV-1 MA by NMR. shows the 1
N HSQC spectrum of 15
N-labeled MA. To our surprise, addition of BAF to the MA sample did not result in any changes in the spectrum even at the protein concentrations used for the NMR measurements. We conclude that BAF and MA do not directly interact. Then how might the previously reported interactions 
be explained? BAF and MA both bind DNA and we propose the protein preparations contained sufficient DNA to allow an apparent interaction between BAF and MA through DNA binding. Consistent with this interpretation, the low micromolar range reported apparent affinity of MA for BAF is similar to the affinity of MA for DNA 
. BAF binds DNA more tightly and DNA condensation can present a kinetic barrier to dissociation 
. In our hands, both BAF and MA tend to co-purify with DNA and extensive washing of columns at high ionic strength is required to remove all traces of DNA during purification. Some of the previously reported experiments 
were also carried out with BAF synthesized in an in vitro
coupled transcription/translation system and this protein will have contained carried over template plasmid DNA. As expected, when DNA is added to the mixture of BAF and MA the spectrum radically changes (). We note that the number of cross-peaks is greatly diminished in contrast with addition of DNA to MA, alone which shifts and broadens a subset of peaks without reducing their number 
. The DNA used for this experiment was a 16 mer duplex oligonucleotide which demonstrates that BAF and MA can simultaneously bind to very short DNAs.
1H-15N HSQC spectrum of: (A) 0.5 mM 15N-labeled free HIV-1 MA, (B) 0.5 mM 15N-labeled MA plus 1 mM unlabeled BAF, (C) 0.5 mM 15N-labeled MA plus 1 mM unlabeled BAF and 2 mM 16 mer DNA.
Many of the proteins reported to interact with BAF are DNA binding proteins. We therefore decided to reexamine their binding in light of our experience with HIV-1 MA. Cone-rod homeobox (Crx) protein is a transcription factor that was identified as a BAF interacting protein in a yeast two-hybrid screen 
. Co-immunopreciptation and pull-down assays supported the conclusion that BAF and Crx interact directly. show the 1
N HSQC spectrum of 15
N labeled Crx in the absence and presence of BAF, respectively. The spectra are identical, unambiguously indicating that BAF and Crx do not interact directly. Addition of a 16 mer duplex DNA to the Crx results in a shift of a subset of the cross-peaks as expected for binding of DNA by Crx (, compare the spectrum in the presence of DNA (black) with the superimposed spectrum in the absence of DNA (red)). Addition of both DNA and BAF to Crx results in disappearance of most of the cross-peaks (), indicative of the formation of large complexes of Crx, BAF and DNA.
-1H-15N HSQC spectra of Crx homeodomain.
MAN1 is an inner nuclear membrane protein that contains a LEM domain near its N-terminus that binds BAF. The C-terminal domain of MAN1 (MAN1-C) has been reported to independently bind BAF 
in addition to the transcription factors GCL and Btf. We labeled MAN1-C with 15
N and recorded the 1
N HSQC spectrum (). Addition of BAF to the MAN1-C resulted in no significant change in the spectrum () demonstrating that these proteins do not interact directly. Interestingly we find that MAN1-C binds DNA (). Addition of DNA and BAF to MAN1-C results in additional disappearance and shifts in peaks demonstrating that BAF and MAN1-C form a large complex with DNA, although BAF and MAN1-C do not interact directly.
To confirm that DNA contamination can confound the interpretation of pull-down assays for protein-protein interactions we carried out such an assay for BAF and MA interaction in the absence and presence of DNA. His-tagged BAF was bound to a Ni chelating sepharose column. MA was then added to the column in the absence or presence of DNA. After extensive washing BAF was eluted with imidazole. In the absence of DNA only BAF eluted from the column (, lane 2). However, in the presence of DNA, MA co-eluted with the BAF (lanes 3 and 4).
Co-elution of MA with BAF in pull-down assays in the presence of DNA.
The mutation of Ala12Thr in BAF has been identified as the cause of a human Hereditary Progeroid Syndrome 
. Ala12 is surface exposed but does not map to either the DNA or LEM domain binding surfaces of BAF, so the effects of this mutation are unlikely to involve disruption of DNA or LEM domain binding. It was proposed that the mutation might affect the interaction of BAF with other proteins, its subcellular localization or stability 
. Indeed Ala12 lies on the surface of BAF that was implicated in binding MAN1-C 
and disruption of the MAN1-C/BAF interaction would have been a reasonable candidate for the primary effect of the mutation. Our finding that MAN1-C does not interact with BAF eliminates this model. Although we cannot ignore the possibility that Ala12Thr disrupts an interaction with a factor yet to be identified, the reduced abundance of BAF in the mutant cells 
suggests a primary effect on protein stability.
Pull-down assays and co-immunoprecipitation are commonly used to screen for protein-protein interactions because of their simplicity and convenience. However putative interactions identified by such assays need to be confirmed and substantiated by more direct biochemical and biophysical methods. We conclude that, contrary to previous reports, BAF does not interact with HIV-1 MA, Crx, or MAN1-C.