is it possible to isolate a unknown compound through HPLC

[debadutta bhoi]

i have a crude extract of some plants if i want to know what kinds of
compounds are there, is it possible to isolate each compounds from the
crude extract through HPLC ?
please help

[Dakota Hamill]

Hi, long story short, yes it is.  However, you may not find it as simple as just just injecting a mystery liquid and collecting nice clear fractions of separate compounds.   Big questions would be, what solvent to use? what column to use? How do you discern what fractions are "known" compounds and what are "unknown" even if you do get nice fractions?

I'd maybe start with a TLC plate, unless you have a MALDI-MS or another LCMS then perhaps you can just start firing away and let the database do the hard work.

Even after HPLC, you'd still have to run IR, NMR, MS to then figure out what it is you even got, and if it's new, but, that sounds like a fun challenge to me!  

Here is a quick return on a google search for screening for new compounds from plant extracts.

http://www.wiley-vch.de/books/sample/3527315306_c01.pdf

http://old.iupac.org/symposia/proceedings/phuket97/hostettmann.pdf

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3218439/

Perhaps someone can give you a better, more detailed explanation of the steps to take, good luck, and is this for school or for a personal project? 

[Nathan McCorkle]

Isolating the compounds is pretty easy, using HPLC, TLC, GC, etc...
For HPLC you just run your sample through the system, and at the end
of the line there is a spectrometer, when the signal changes the
liquid get diverted into a clean test tube, every time the signal
changes, a new test tube is brought into place to accept this
'fraction'.... I'm inclined to call this 1D (1 dimensional)
chromatography, because the separation of compounds depends on the
solvent/separation system used.... so you could in fact have multiple
compounds in each fraction. To separate those from each other, you'd
have to run each fraction through a different separation system than
the first one was (a la 2D gel electrophoresis).

Identifying these compounds is either expensive, or potentially a PhD
thesis sized project... but certainly do-able.

--
Nathan McCorkle
Rochester Institute of Technology
College of Science, Biotechnology/Bioinformatics

[mad_casual]

More specifics would help. When you say 'crude extract' do you mean 'crushed or lysed plant cells in water' or do you mean 'salt precipitation and ethanol extraction of plant material that's been pushed through a .22 um filter'? Crude extract meaning the extraction was poor/simple or meaning that it was a good/complex extraction just done by hand. Also, when you say 'each compounds' do you have something specific you're looking for or are you expecting an HPLC-detector system to point you in a direction. Because, depending on the detector system, you're going to get pointed in about 100 different directions. As for actually doing the deed, you can easily overload an HPLC column, you can equally easily overload a detector. Given that you haven't overloaded the HPLC or the detector, then the answers spit out are always relative, especially for unknown complex mixtures. As a rule of thumb, 1000:1 relative concentrations are 'saturated'. Using a mass detector with a 1mL sample with 100 mg/mL of unkown solute, a competently designed modern system should be able to detect e.g. 1 mg of protein in 99 mg of "stuff", etc. (~99:1). The well designed systems should be able to see 100-10 ug in a 100 mg sample (~1000:1). It's not too difficult to see low nano- to  high pico- levels of compounds in a background of low micro-level contaminants. There are rather easy and even simple techniques for getting well beyond the 1000:1 level. Most LC-NMR and LC/LC-MS/MS/MS systems can be tailored to do this, bead based affinity systems can be done DIY. However, without knowing exactly what you're looking for, what Nathan says about getting into $100K+ instruments and Ph.D.-candidate level work is fairly cogent.