predicting retention in ion exchange hplc

[DJ]

Hydrophobicity, at least superficially, seems like a more difficult-to-estimate parameter than "charge." Yet, while there are numerous algorithms for predicting relative retention in RP-HPLC for proteins, peptides, ion exchange retention calculators are not nearly as commonly encountered. Why is it this? Is it because retention in ion exchange is genuinely difficult to predict, or rather, because RP-HPLC is all the rage compared to IEX, and the folks in the retention time calculation business tend to focus more on RP?

[tom jupille]

Actually, for fully-ionized species in "pure" ion exchange it's very simple: the log of k' is a linear function of the log of the buffer ionic strength, and the slope (the "z" value) is the ratio of the effective charge on the analyte to the charge on the driving ion.

In real life, many separations are effectively "mixed mode", which complicates things.

[Kostas Petritis]

I actually developed one for peptide retention time prediction in strong cation exchange.

You can find the actual utility here: http://omics.pnl.gov/software/NETPredictionUtility.php

As Tom mentioned, there are mixed effects and as we recently found, some unexpected ones (see Anal Chem paper in the link below).


http://dx.doi.org/10.1021/ac100651k

[DJ]

I am familiar with this paper. There are instances when Arg deletion peptides were -more- well retained on WCX than the same peptide +Arg. How does one even begin to go about building an algorithm that would predict this

[Kostas Petritis]

That shouldn't be happening and my first reaction would be to ask if you are positive that this is really what you observe.

Then we will need much more information in terms of operation conditions to maybe try to explain it.

Are those synthetic peptides (are you positive about their structure, MS confirmation)? How long is your peptide? Number of acidic/basic residues?

What are the operation conditions of your LC? Have you run the same samples in any SCX or other WCX?

[DJ]

This is not my work..

Page 3, top chromatogram

http://www.westernanalytical.com/pdf/cat_227.pdf

"The Sample was the main fraction collected from Reversed Phase step. The product was purified
satisfactorily at the cation exchange step. Selectivity is remarkable; deletion of a Thr- residue sufficed to
afford nearly baseline resolution from the main product. The elution order was as expected in some cases;
Loss of -PO4 or Ser(PO4) would decrease electrostatic repulsion and increase retention time, while loss of
two Arg- residues decreased retention. However, the loss of all three Arg- Residues had little effect on
retention, while loss of a single Arg- actually increased retention. The indicated deletions were identified by
ES-MS. Again , one cannot specify which particular residue is missing if there is more than one possiblity."

[Kostas Petritis]

I would guess that in this case as you do not have any organic solvent in the mobile phase you get a lot of hydrophobic interactions on top of the ion-exchange effects. The model I developed used 25% acetonitrile to eliminate those effects. I used the SCX version from PolyLC, which have been shown that at >10% acetonitrile you have eliminated hydrophobic interactions...

[DJ]

Since the columns from Poly LC are silica-based, I was under the impression that, at high organic (>60%), hydrophilic interaction would be superimposed on ion exchange retention mechanism. So hydrophobic interactions can still be significant even with a charged stationary phase bound to silica? Interesting.

I've always added at least 25% MeCN to the mobile phase. Alpert reported (in '88, I believe) increasing concentrations of MeCN improved peak shape, even selectivity. Do you think this is a result of diminished hydrophibic interactions, increased hydrophilic interactions, or, something else-- perhaps, altered secondary structure?

[Kostas Petritis]

Of course you do not want to go at very high % of organic solvent as you do start to have hydrophilic interactions. So 25% is a good percentage to all the reasons that I and you mentioned. As Andy is following the forum, I will let him respond at your question.

[Andy Alpert]

Having been summoned, I emerge from the lamp...

Concerning hydrophobic effects, there's charged coatings and then there's charged coatings. First, take a look at my paper in J. Chromatogr. 359 (1986) 85. This demonstrates that a single methylene group makes a substantial difference in the hydrophobic character of a coating in the hydrophobic interaction chromatography (HIC) mode. That means methyl- vs. ethyl- vs. propyl. Same goes for an ion-exchange material. That's why we make our SCX material, PolySULFOETHYL A, with a sulfoethyl- ligand instead of the sulfopropyl- (= SP) ligand used by nearly everyone else. It makes a difference. In A. Holm et. al., Anal. Bioanal. Chem. 382 (2005) 751, they compared four different SCX materials regarding degree of hydrophobic interaction with the peptide angiotensin II. See pg. 755; 10% ACN eliminated any vestige of hydrophobic interaction in their test when using PolySULFOETHYL A. It took 20-25% to do so with the other three SCX materials.

There's been a number of papers in the literature recently extolling the benefits of using a fairly hydrophobic material in the HILIC mode, implying that you can get hydrophobic and hydrophilic interactions operating simultaneously to effect a separation you can't get any other way. Retention of solutes as a function of % organic solvent is a U-shaped curve; the hydrophobic interactions almost always are reduced to zero or close to zero before you get to a level of organic solvent where the hydrophilic interactions start to become significant. Result: The bottom of the U is flat, not sharp, with the two regions separated widely. If a material has no significant amount of hydrophobic character, then the curve of retention is a backwards L shape, not a U.

DJ: The application you cited on May 31st was my work from 1996, performed with some peptides from your lab at U. Nevada-Reno. This synthetic peptide was an unholy mixture of failure sequences, making it an ideal standard for testing the selectivity of the experimental method in question. This was the uncharging of a weak cation-exchange (WCX) material with a decreasing pH gradient, leading peptides to elute in a totally volatile solvent from an ion-exchanger (this method has been revived for top-down proteomics, most recently with histones). If you want the full poster, download it from the Literature section of our web site (www.polylc.com); it's the ISPPP 1996 poster. I did note the novelty of the Arg- deletion peptide eluting later than the version with all of the Arg residues. However, there was ample opportunity for internal salt bonds. Also, per the paper that Kostas posted the link to in his posting of May 27, not all the charged residues in a peptide have access simultaneously to a stationary phase surface. That means it's unclear how much effect putting in or taking out a particular arginine should have. If taking one out results in the peptide assuming a conformation in which more of the basic residues have access to the surface than was the case before, then this could provide a rationale for an increase in retention. All in all, it's a complicated scenario. If you're trying to build a model for interactions in chromatography for purposes of prediction of retention times, start with simple standards first.