Chromatography Forum

Hi guys,

I am currently working on a GC-MS metabolomics method for human urine and would like to know your opinion regarding the best way to tackle this task..

I am currently treating thawed centrifuged urine with acetonitrile to deproteinize and then drying the supernatant under N2. I then oximate for 16 hours at room temp and then trimethylsylilate with BSTFA for 1 hour at 70 °C. Another drying and resuspension in hexane and then I inject split mode (10:1) in the GC..

I keep getting yellow crystallization in the vial and that doesn't want to dissolve in the hexane. I tried doing a water-hexane extraction after derivatization, but the resulting chromatogram is too empty for metabolomics method (almost all I get are derivatized sugars).. the yellow gunk does however dissolve very nicely ni the water phase, but I am afraid that some of the TMS-derivatives also go into the water phase and then I lose it.

Any suggestions please?

Hi Cynthia

The yellow deposit is probably all the salts, urea and other GC-intractable muck that is dissolved in urine, and you do not want it dissolved in the sample that you inject to the GC. Adding water will hydrolyse the derivatives.

Are you getting sensible looking chromatograms from the hexane phase ?. What makes you think that anything interesting is in the yellow solids ?

Peter

Hi Peter,

The chromatogram I am getting now only contains the sugars and sugar phenols, the rest of the BSTFA derivatives is hydrolysed in the water phase (after some research I came to know that BSTFA derivatives are still susceptible to hydrolysis..)

A great thing about the picture is that there is no urea after water extraction.. but alas, all other classes of metabolites that I want to see are not present. This simply will not do for a metabolomics method...

What I want after derivatization is a two phase system in the vial with ALL the derivatized compounds in the hexane phase and the salts and other underivatized muck in the other phase (what if I use something like an alcohol instead of pure water?)

The other thing I am going to try now is to sonicate the sample after derivatization and addition of hexane (no water).. If I can get all the derivatized stuff in the hexane without getting crystallized muck in there I will be very pleased indeed.

After a quick slow centrifugation I can then get the hexane out (without the solids) and inject that (if it works)..

As this is a metabolomics method with no pre-extraction of a specific class of metabolites, I am encountering many obstacles.. but I am also doing an LC-TOF-MS method on the untreated urine. Together, I want to use the data from GC and LC methods to draw my conclusions (One has a stronger argument for metabolomics when one utilizes more than one type of analyses.)

Thanx for your reply! I hope to reach a breakthrough with this GC method soon!

Hi Cynthia

By derivatising the samples you are already selecting a sub-class of compounds to concentrate on - so you are not really doing metabolomics in the sense of blindly seeking differences between chemical fingerprints. What class of compounds are you interested in ? Once you have decided on that you can optimise the sample prep and GC conditions to work properly for those - there is no GC method that will work for all the organics in urine, and no single derivatisation that will make all the GC-unfriendly ones tractable.

Peter

Hi Peter,

We are not making any hypotheses beforehand, so this study is more like a discovery than a targeted test.

Thus we chose to derivatize with BSTFA as this is the type of reagent that derivatizes the most classes of compounds. I am fully aware that one GC method alone cannot be called metabolomics as we already introduce bias with the derivatization steps.

But this way we can get (theoretically) the most compounds in one method.

I am going to try to add pyridine today before the derivatization to act as solvent, so hopefully I can get the important compounds to dissolve and to be available for derivatization then. Afterwards I can still add some hexane and inject that.

With regards to what classes of compounds we are interested in: I think a fatty acid profile on serum will give us the answers we need, maybe by doing a Bligh and Dyer (or Folch) extraction followed by methylation...


Cindy

Just a small side remark, I am surprised to see the Bligh and Dyer method mentioned here. It is inherently better than the Folch method for many if not all applications, as it starts out with a homogenous soluton (no different liquid phases). Yet it seems to take a "back seat" to the Folch.

Hi Cynthia

Natural materials are so complex that the limiting factor on how many compounds you can see in one run is set by the resolving power of the GC - even if you limit the scope to those compounds that are volatile and stable enough to get through a GC column there is no column on Earth that can separate all the compounds in complex natural materials.

To see what I mean extract say 100 ml of urine with 3x 10 ml of clean dichloromethane. Concentrate the solvent down to about 1 ml, and inject 1 ul splitless with a temperature programme of 5C/min from 40 to 240C. The whole chromatogram is filled with peaks, blow up any section of it and you find more peaks and more overlaps, dig around with an MS on what looks like baseline and you will find traces of peaks that are individually below the TIC baseline. You have more peaks with this simple sample prep than you can possibly hope to analyse statistically, even if you could measure their areas accurately.

Because there are more compounds in the sample than you can possibly analyse you have to select what you are going to look at. If the selection is arbitrary; let's see what differences this makes to the volatiles/acids/sugars or whatever, you are probably less likely to make discoveries related to whatever biological difference you are interested in, but any discovery you do make is quite likely to be outside the box. If the selection is based on what is already known; e.g. lets look at hydroxy acids because other similar drug treatments are known to affect hydroxy acids, you are probably more likely to find something, but what you do find will be within the established paradigm.

For twenty years or so researchers have been looking for mammal semiochemicals using an unselective metabolomics approach to GC fingerprinting; extract, analyse, multivariate statistics. No semiochemicals have been found by that route - those that have been identified have been tracked down by painstaking selective analyses informed by animals' responses to different fractions.

So you have to be selective in one way or another - even if you turn down the sensitivity of the analysis so that you have fewer peaks that do not overlap, you are slecting those compoiunds thatare present at the highest concentration.

Peter

To add to Peter's comments on quantity of compunds - I have looking at a solvent extraction of human urine by GCxGC-TOFMS. With the greater peak capacity, I see thousands of peaks in a single chromatoram. And burrried under those peaks are more peaks...

While you can try for the blind search, if you have some hypothesis about the metabolism you are trying to study, it can help you select a class of compounds - as you can look only at a limited number of compunds in any chromatogram. And, I am frustrated by being limited to "only" several thousand.

By derivatising the samples you are already selecting a sub-class of compounds to concentrate on - so you are not really doing metabolomics in the sense of blindly seeking differences between chemical fingerprints.

Peter

Second to Peter's comment above. Because you will use LC-MS, a better way to do your work is to use GC-MS for only volatile and semi-volatile component with no derivatization. A simple extraction with non-polar GC-friendly solvent for GC analysis, send rest to LC.

Cynthia,

You are right on this point: One has a stronger argument for metabolomics when one utilizes more than one type of analyses. Metabolomics seeks to determine as many small molecules as possible in a sample, it's a non-targeted, non-biased, comprehensive analytical approach, no one single techique can do "real metabolomics" under current state of analytical science.