Neither SDS PAGE nor the SEC is absolute measurement of molecular mass. I would say in response to your question, there may be multiple reasons/assertions to be considered. In addition to what Dr. Butnev has explained, in SEC, the time of elution is also depend upon the conformation of your protein, means how compact or extended it would be in that particular solvent conditions. For instance, when we teach to students we give example of circular and linear forms of the peptide of identical mass and we ask them, which one will come out of the column first. The answer is the circular one. Therefore, molecules  comprising compact conformations run faster even having similar molecular weights. 

Hi Nikolai!

I cannot give you an answer but I have seen a similar behavior for a DNA/RNA-binding protein with a high pI (around 8.5-9). We used to get very broad peaks in SEC, corresponding to much lower MW than expected. It was solved by adding some arginine to the SEC buffer (50mM I think). We never really figured out what was happening but I believe it had something to do with non-specific interactions (the resin in that case was superdex).

Hi Marcin! Nice to hear from you. I saw your question, also on Researchgate, about a wide profile and the suggestion to use arginine. Very interesting to hear from you now that it worked. Thank you. I will try to use it next time. I'm wondering also whether this can be an intrinsic property of the studied protein...

Another possible reason that you might want to consider is the intrinsic properties of your particular protein. When I purified one of the lysosomal enzyme having the affinity to some carbohydrates using the Superdex resin, I also observed this phenomenon. It was explained by the affinity interaction between the protein and the chromatography resin matrix. I would suggest to run your protein structure on computer against all affinity databases for the search of possible affinity of your protein to the resin (including all lectins).

Neither SDS PAGE nor the SEC is absolute measurement of molecular mass. I would say in response to your question, there may be multiple reasons/assertions to be considered. In addition to what Dr. Butnev has explained, in SEC, the time of elution is also depend upon the conformation of your protein, means how compact or extended it would be in that particular solvent conditions. For instance, when we teach to students we give example of circular and linear forms of the peptide of identical mass and we ask them, which one will come out of the column first. The answer is the circular one. Therefore, molecules  comprising compact conformations run faster even having similar molecular weights. 

Thank you Dr. Butnev for the interesting suggestion. I would appreciate if you could specify your point concerning the databases and probably provide me with some relevant link?

Thank you Dr. Akhter for your consideration. I think my question is a bit more complicated than the comformation of two peptides of different shape. The SDS-PAGE result here is just to confirm that the protein is complete and column calibration by globular proteins (approximated with spheres of distinct sizes with some average density, more or less typical for a protein) means that a protein being eluted later is MORE compact than an average globular protein taken for column calibration. This observation is also confirmed by the RI measurements (calibrated by BSA and dn/dc = 0.185), again showing lower mass and size than expected for the complete monomer. This really indicates something interesting happens to my protein, and I'm curious what could this be. Thank you again.

Hi Nikolai! Did you tried to look at the disorder propensity of this protein?

Hello, Vladimir! Yes, it should be kind of normal one, with pretty well defined structure but with a central cavity surrounded by protein matter. This means that its gyration and hydrodynamic radii should probably be higher than for another protein of the same mass, but in fact it elutes later and with somewhat underestimated Mw...

Hi Nikolai! what about the surface distribution charge of your protein? SDS-PAGE or SEC is not the better choice to estimate the mass of the mass! Do ESI.

If you can coupled your SEC to a MALS or/and a DLS module, even if your UV-elution profile is not the expected, and you can obtain a good mass and Rh, you can sleep peacefully! There are many proteins in literature with anomalous SEC profiles....it is not a surprise

Hi Diego! Thanks for your input. Well, it seems I have to add a bit more information as my initial question was not that naive. I did even better, not SEC-MALLS, but SEC-SAXS/RALS/RI/UV analysis of the protein. It behaves like a particle of really smaller size (in terms of later elution, Rg, Rh, apparent Mw, etc.) but... according to MSMS termini are present and the SAXS curve could be reasonably well described with Xtal structure (e.g., by crysol3 - with atomic representation of a water shell)... So, the question again is why can a globular pretty well-folded protein display hydrodynamic characteristics of a smaller one?

To me it seems like you have eliminated all the options, besides the interaction. Try to include various additives and/or run on different column type.

I think that Tomas is right and some protein-column interactions are present. You may try to increase salt concentration, say to 0.5M NaCl or use some other additives. Conformation per se can definitely be excluded, since your protein runs as more compact than expected for a typical globular protein of its size. This means that the protein's chromatographic behavior in this case is not determined by its shape and hydrodynamic volume (as you know, any type of partial or complete unfolding of a globular protein will lead to the increase in its apparent molecular mass).

I did not see this last addition by Nikolai, namely that the smaller size of a mystery protein is confirmed by multiple techniques (SEC-SAXS/RALS/RI/UV). This definitely makes question much more interesting. Did you try to evaluate packing density of this protein? Do you know its structure? Can this be somehow related to the abnormal hydration? You may try to look at this by NMR. Another possibility would be to look at your protein by SANS, where you would be able to completely exclude water by playing with contrast variation. Just a thought...

Thank you for the great point, Vladimir, concerning the hydration of the protein. In fact, I think this can be somehow involved here, as it is a soluble lipid binding protein with a cavity. I think I will look at this point more closely. Regarding packing density, how can it be estimated? SANS would be definitely great to try, although it is very protein- and time- demanding..

For more ideas on how to deal with protein packing density, please have a look at these publications: PMID: 10388571; http://www.stat.rutgers.edu/home/rongchen/publications/03JCP_packing-density.pdf; and http://www.cell.com/biophysj/abstract/S0006-3495(01)75739-6. In two last papers, you will see that there is scaling of the protein packing density with chain length. Here, small proteins are packed denser than larger proteins, because larger proteins have larger number of voids, pockets and cavities. The simple explanation based on these observations would be that the "protein part" of your protein is packed tighter than it is expected for the protein of its size (say, as tight as one would expect for a tightest packing of a small protein, with almost no voids, pockets and cavities) because it has a lipid binding cavity. If this functional lipid binding cavity of your protein has a volume smaller than a volume that could be taken by all those voids that would be present in a protein of its molecular mass, then the resulting volume of the protein will be smaller than the volume of a "normally packed" globular protein of its molecular mass. Hope that this makes sense. I will try to formalize this idea in the following message.

Here is promised formalization: Lets assume that a total volume of a protein, V(tot), is a sum of two components, V(prot), a volume taken by polypeptide chain, and V(void), a volume taken by all the voids. Lets assume that V(prot) represents a volume taken by a perfectly packed polypeptide chain (with no voids). This means that V(prot) is directly proportional to protein molecular mass (PMM). Therefore, the density of a perfectly packed protein defined as PMM/V(prot) does not change with the increase in PMM (as in the case of the tightest packing of balls). In reality, protein packing density scales with PMM, getting smaller for larger proteins, since the real packing density is defined as PMM/V(tot)=PMM/(V(prot)+V(void)) and not as PMM/V(prot). Obviously, information on the PMM dependence of V(void) can be extracted from the analysis of packing densities of a set of globular proteins with increasing PMM. For a protein with functional cavity, V(tot)’=V(prot)+V(void)’+V(cavity). Then you have several possible scenarios:

Case 1. V(void)’ of such a protein represents a result of normal V(void) scaling with PMM (V(void)’=V(void)). In this case, V(tot)’ would be larger than V(tot) of a globular protein of given molecular mass by the value of V(cavity).

Case 2. V(void)’ of such kind of proteins with functional cavities scales with PMM faster than the V(void) of the normal globular proteins does (V(void)’>V(void)). In this case, V(tot)’ would be even larger than that described in the case 1.

Case 3. V(void)’ of such kind of proteins with functional cavities scales with PMM slower than V(void) of the normal globular proteins does (V(void)’V(void). Then, V(tot)’>V(tot).

Case 6. V(void)’=0. V(cavity)