Mobility Shift Assay

Harry Mangalam mangalam at SALK-SC2.SDSC.EDU
Fri Dec 18 17:11:54 EST 1992

Jorge L. Sepulveda warned me of ambiguity in my previous posting, regarding
my pseudo-diagram (reproduced below), and posed the questions below which
are perhaps wothwhile posting to supplement the original thread:

>Hi, Harry
>I am curious about your experiment with increasing concentrations of
>is your graph a picture of the gel or a plot of band intensity vs. competitor 
>(non-specific, I assume)? Also, what do you recomend for analysing protein/DNA 
>interactions, especially in crude extracts? Footprints? Southwesterns? I tried 
>those and I think they are more tricky than gel-shifts. 

>>             Competitor DNA (ng)
>> .01  .03  .1   .3    1    3    10   30  100  300 
>> -------------------------------------------------
>> ===  ===                           +++  +++  +++
>>           ===       ---  ---  ---  ---  ---  ---     
>>                ---            +++
>>           ---  ===       +++  
>> +++  +++  +++  +++  +++     
>>                     ===
>> ---  ---                 ===
>>                               ===  ===  ===  ===


Hi Jorge,

   The graph was supposed to be a 'picture' of the gel 8(, (actually a
merging of the results from several gels), the ===, ---, +++ representing
different bands across a gradient of (NONSPECIFIC) competitor DNA.  It
really looked amazing - and was one of the things that turned me off
gelshifts most convincingly. (And it was repeatable, unfortunately).
   And yes, it was non-specific DNA.
   In crude extracts, I would recommend SouthWesterns 1st.  Again, it
depends on your protein, but I know that a number of homeobox,
leucine-zipper, POU, HMG-box, and other proteins will bind DNA reliably
after transfer.  A number of Zn-finger proteins (typically the steroid
receptor family) and at least one HLH will not.
   Typically, I start with a phosphocellulose (PC - also stands for
politically correct 8)!  )-purified extract and very little competitor DNA,
increasing the purity of the protein and the competitor DNA as time goes
on.  PC is very cheap, has very capacity and doesn't clog very much; it
will also retain almost all DNA-binding proteins.  Heparin-Agarose is also
good, but much more expensive.  Hydroxyapatite is a good choice after an
initial purification because while it has high capacity and is relatively
cheap (just CaPO4), it clogs like an SOB unless you de-fine it carefully
and only feed it a cleaned extract (and because you elute with a gradient
of phosphate, you can't load samples of fractions without cleanup because
high [PO4] will precip the SDS in a loading buffer).

   Of course, depending on the protein 8), other matrices are good.  A
hydrophobic column can be a good choice, not only for it's separating
qualities, but it also can act like a desalting column (you load in high
salt, elute with decreasing salt).

   Footprinting is a decent test of binding, but it is more difficult than
either a SW or gelshift.  It is also very difficult to do with a crude
extract, but with a PC fractionated extract, it should really be no problem
unless your protein is present in vanishingly small amounts, in which case
you'll have to wait til you clone it until you can do the biochemistry 8(.

   If I were to do another complete characterization, I'd still use
gelshifts, but only in a secondary role, to check column fractions or in
combinations with other techniques like gel slice renaturations (in which
you run an SDS protein gel with large amounts of crude protein, cut it into
small slices, elute the protein from the individual slices and then use
those eluates in in a gelshift.  It sounds like (and is) a pain, but you
know that the gelshifts you are seeing are from proteins of a specific
size, which can be useful.


Harry Mangalam                                   Vox:(619) 453-4100, x250
Dept of Biocomputing                                   Fax:(619) 552-1546
The Salk Institute                        1'   mangalam at
10010 N Torrey Pines Rd                   2'        hjm at
La Jolla CA 92037                         3'         mangalam at salk.bitnet

More information about the Methods mailing list