Help with Gel Filtration

D.K. dk at
Fri Oct 25 21:04:27 EST 2002

In article <20021026004236.28950.qmail at>, dustjohn at wrote:
>Trying to separate a 60kDa His-tagged protein from a ~26kDa Ecoli protein that
> purifies along with it on the His-column.  Had no luck separating the two with
> a Sephadex G-75 column (bed length ~25cm, column diameter ~1.5cm).  Switched
> to G-50, same bed length and, again, no separation.  Should I run a longer
> column (~90cm bed length)? 

60K vs 26K should be relatively easy. I don't have any experience with 
Sephadex G-75. G-50 defintely won't work. I think you need higher pores
than usual G-75. Superose 12 or Superdex 200, both availbale in "Prep"
grades, will resolve very easily. 

>Should I increase/decrease the filtration rate (usually ~1ml/min).

Ouch! It's way, way too fast. Resolution of gel-filtration is extremely 
sensitive to flow rate, generally the lower the better since usually you 
cannot realistically approach a limit where diffusion would be 
detrimental to resolution. And this, obviously, is strongly related
to bead size (Which is rather large even for "fine" grade Sephadexe.
in 90 cm column you are unlikely to use "fine" grade though;
I'd guess "medium" which is larger). 

Here is an example:  for 1 cm column filled with really small particles 
with minimal dispersion (ca 13 um and sigma ~ 2), the maximum 
recommended flow rate is 0.5 ml/min. In fact, 0.2 ml/min gives 
considerably better results. Your flow rate corresponds to 0.44 ml/min 
for a 1 cm column. At the same time, your gel has about 6X larger 
particles with a hugely larger dispersion (~ +/- 50%). In first 
approximation, optimal flow rate is reverse linear function of particle 
diameter. That means you'd have to decrease you flow rate 6X to 
even get close to resonable range. In reality, it has to be even lower
because your particle dispesion is a lot higher - optimal rate is 
a reverse function of sigma squared. 

So you can start with using 0.1 ml/min and it probably will work
(Assuming the column is packed well - this is the single most 
important factor! Did you run something colored as a check?).
If it does not, you need to change matrix. 


More information about the Proteins mailing list