Protein precipitation leaves the phospholipids behind: how to see them in LC-MS and remove them

Protein precipitation is the fastest sample preparation there is for plasma. Add acetonitrile, vortex, centrifuge, inject the supernatant. It is the first method many bioanalytical labs try, and often the one they keep.

It does one job well: it removes protein. It leaves the phospholipids, and in electrospray LC-MS, the phospholipids are one of the main reasons a method loses signal.

I ran bioanalytical LC-MS/MS on plasma at a bioequivalence lab and at a CRO. This is the part of sample preparation I would most like a new analyst to understand before they choose the quick option.

What protein precipitation leaves behind

A 2020 tutorial review of matrix effects lists protein precipitation with organic solvents as “not effective for other interference removal like phospholipids, lipids, aminoacids” (Cortese et al. 2020, p. 16, Table 4). Acetonitrile is a better precipitating solvent than methanol (p. 15), but neither takes out the phospholipids.

Plasma has a lot of them. A 2023 study of chemicals in human plasma describes a 1000-fold concentration gap between endogenous substances, of which phospholipids are the main small molecules, and the environmental chemicals the authors wanted to measure (Sdougkou et al. 2023, p. 1). Your drug at its lower limit of quantitation is in the same position.

How to see them: m/z 184

You can watch phospholipids directly. Phosphatidylcholines, lyso-phosphatidylcholines and sphingomyelins share a phosphocholine head group, which gives an ion at m/z 184 in positive electrospray.

The technique usually cited is from Little and co-workers: monitor the transition 184 → 184 on a triple quadrupole, which uses in-source fragmentation (as summarised by Cortese, p. 16; Majors, p. 241). On a high-resolution instrument, the same fragment is at m/z 184.0733 (Sdougkou, p. 4).

Add that transition to your method, inject an extracted blank plasma, and you see where your phospholipids are. If they sit on top of your analyte, you have found a likely cause of suppression. This takes one injection.

Why the gradient alone may not fix it

The first idea is to move the analyte away from the phospholipids. That is worth trying, and it is the cheapest fix in matrix effect work generally.

But the phospholipids are not one peak. In the Sdougkou study, without phospholipid removal, the phosphocholine signal was present throughout the run, from 1.3 to 22 minutes, and was strongest after 15 minutes, in the high-organic part of the gradient (p. 4). Their conclusion was that the phospholipid mixture “could interfere with all other analytes, irrespective of RT.”

There is a second issue. Agilent’s sample preparation book notes that the lipophilic material left after precipitation can “lodge on the column and bleed later” (Majors, Sample Preparation Fundamentals, p. 240). So phospholipids from one injection can affect a later one.

How much signal is at stake

The Sdougkou group compared their method, with a phospholipid-removal step, against a control method without it. With removal, the average signal of the non-phospholipid compounds rose 6-fold in positive mode (up to 28-fold) and 4-fold in negative mode (up to 58-fold) (p. 5). They also detected 109% more non-phospholipid features in positive mode and 28% more in negative mode (p. 1).

That was an untargeted study, so not every analyte gained that much. But it shows the size of what phospholipids can hide.

What removes them

The 2020 review ranks the options for plasma (Cortese, pp. 16–17):

• Protein precipitation: fast, leaves phospholipids
• Supported liquid extraction (SLE): removes most of them with the right loading and elution
• Reversed-phase or cation-exchange SPE: cleaner extracts and less matrix effect than precipitation
• Mixed-mode SPE: the most effective, because it combines reversed-phase and ion-exchange retention
• Phospholipid-removal plates, which combine precipitation with a sorbent that holds the phospholipids

Much of that ranking comes from one widely cited comparison (Chambers and co-workers, 2007), which I know through the review and have not read in the original.

The removal plates are worth one sentence on mechanism. The HybridSPE type uses zirconia-coated silica. Zirconia acts as a Lewis acid and binds the phosphate group of the phospholipids (Ahmad et al. 2012, p. 5). The analyte passes through, if it does not bind as well.

Check your recovery before you switch

That last condition is where an independent result matters.

Sdougkou and co-workers did not follow the manufacturer’s protocol for their HybridSPE cartridges, “as these gave poor recoveries” for their analytes, and re-optimised it (p. 4). In preliminary tests, another commercial phospholipid-removal column gave recoveries that were “too low or negligible” for their perfluoroalkyl analytes and for late-eluting compounds (p. 4).

Vendor material for these products shows them beating precipitation, SPE and LLE in post-column infusion comparisons (Majors, pp. 240–242). It is often true for phospholipid removal. It is not a guarantee for your analyte, especially one with an acidic or phosphate-like group that the same sorbent can hold.

So before you change a validated method:

1. Run the m/z 184 transition on extracted blank plasma, with your current preparation, and see whether phospholipids overlap your analyte.
2. Measure the matrix effect and recovery separately, with the three-set experiment I described in the matrix effect essay.
3. If you try a removal plate, measure recovery for every analyte and the internal standard, at your low QC.
4. Keep the m/z 184 transition in development runs, so you can see phospholipid build-up across a batch.

Protein precipitation is a reasonable choice when the analyte is far from the phospholipids, sensitivity is not limiting, and the matrix effect has been measured across lots. When one of those is not true, it is the first thing I would change.

Sources

• Cortese M et al. Compensate for or minimize matrix effects? Strategies for overcoming matrix effects in LC-MS: a tutorial review. Molecules 2020; 25:3047. doi:10.3390/molecules25133047. A review; the Little and Chambers results are cited through it.
• Sdougkou K et al. Phospholipid removal for enhanced chemical exposomics in human plasma. Environ Sci Technol 2023; 57(28):10173–10184. doi:10.1021/acs.est.3c00663.
• Ahmad S et al. HybridSPE: a novel technique to reduce phospholipid-based matrix effect in LC–ESI-MS bioanalysis. J Pharm Bioallied Sci 2012; 4(4):267–275. A review built around one commercial product; used here for the mechanism only.
• Majors RE. Sample Preparation Fundamentals for Chromatography. Agilent Technologies, 5991-3326EN. A vendor book; product comparisons are vendor claims.

If your plasma method loses signal over a batch, or one lot of plasma behaves differently from the others, tell me the analyte and the preparation on LinkedIn and I will tell you where I would look first. The LC-MS troubleshooting cheatsheet covers the instrument side.