Anal. Chem. 2003, 75, 791-797
Capillary Liquid Chromatography/Electrospray Mass Spectrometry for Analysis of Steroid Sulfates in Biological Samples Suya Liu,* William J. Griffiths, and Jan Sjo 1 vall
Department of Medical Biochemistry and Biophysics, Karolinska Institutet, SE-171 77 Stockholm, Sweden
A new procedure for capillary liquid chromatographyelectrospray (CLC-ES) mass spectrometry is described. Using this procedure, coupling of a CLC column to a lowflow-rate ES interface is made simple. A 5-cm precolumn and a 35-cm analytical column, both fused-silica capillaries with an i.d. of 100 µm and packed with 3-µm octadecylsilane-bonded material, are coupled in series to a sheathless ES emitter. One solvent splitter is positioned between the loop injector and the precolumn, and a second, between the precolumn and the analytical column. By opening and closing the splitters in the appropriate order, this arrangement permits the injection of 1-20 µL of sample solution with analyte focusing onto the top of the precolumn, followed by isocratic or gradient elution at a flow rate of 0.2-0.3 µL/min through the analytical column. The relative standard deviation of the retention times of reference compounds was 1 µL/min. The flow splitter can be placed before or after the injector; however, in most published systems, it is positioned before the injector to avoid sample loss. Positioning of the splitter before the injector does not cause great inconvenience if only a few nanoliters of sample are to be injected. However, it creates a problem when injection of large sample volumes is desired. For example, at a postsplitter flow rate of 0.2 µL/min, it will take 100 min for the solvent to traverse a 20 µL sample loop. For the rapid injection of large sample volumes, a precolumn must be integrated into the system, usually by column switching where an additional pump is used.6,7,9,11 The back-flushing of trapped sample from the precolumn into the analytical column, which is the most common way used in these systems, may cause Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
793
column deterioration as a result of the presence of insoluble components in the biological samples. Back-flushing may also disturb the column packing of the precolumn. In the system described in this paper, two splitters were used. Splitter A was placed between the injector and the precolumn and was closed during sample injection, whereas splitter B was positioned between the precolumn and the analytical column and was opened during sample injection. During this operation pump A was run under pressure control giving a flow rate of 2-4 µL/ min through the precolumn. After analytes were sorbed onto the precolumn but before starting the gradient program, splitter A was opened, and splitter B was closed. Then gradient elution was carried out at typical total flow rates of 20-30 µL/min (at the end of splitter A), while the flow rates through the columns were ∼0.2-0.3 µL/min. In this way, it only takes 1 min for solvent in the mixer to reach the top of the precolumn as the pump is delivering solvent at a rate of 20-30 µL/min through the injector. Elution can also be performed in a stepwise manner in which sample injection is made in A solvent and elution is with B solvent. In this case, the pump is run at a pressure between 200 and 300 bar under pressure control so that a flow rate of 20-30 µL/min is generated through the injector and splitter A, and a flow rate of ∼0.2-0.3 µL/min is achieved through the analytical column. An advantage with the double splitter system is that samples with a high content of inorganic and polar organic contaminants can be desalted and purified with A solvent alone or with a low percentage of B solvent. The washings are diverted through splitter B, thus protecting both the analytical column and the ES interface. By performing the desalting on-line as the last step in the sample preparation, a conventional initial solid-phase extraction step for desalting is not required. Instead, contaminants less polar than the analytes can be removed by a solid-phase extraction prior to injection of the sample to the precolumn. This is illustrated by the application of the system to the analysis of steroid sulfates in plasma (see the Experimental Section and below). The two splitters also provide the possibility of selective analysis of ranges of compounds that vary with respect to polarity (or charge, if the precolumn is an ion exchanger column). In the analysis of plasma steroid sulfates (see below), the precolumn was first washed with A solvent while splitter B was open in order to remove inorganic salts and polar organic metabolites. Then splitter B was closed and the steroid sulfates and less polar compounds were admitted into the analytical column. It is also possible to start the gradient elution while splitter B is still open to remove additional compounds of higher polarity than the analytes. When the selected polarity range has been reached, the analytes can be admitted to the analytical column by closing splitter B, which can then be opened again at a desired point of polarity. Nonpolar compounds remaining on the precolumn will then be removed by the gradient of strongly eluting solvent. If desired, a very strong solvent (e.g., methanol/2-propanol) can be injected into the sample loop and passed through the precolumn with splitter B open and then through the analytical column with both splitters closed. This will prolong the lifetime of the analytical column and prevent contamination of the interface to the mass spectrometer. The conditions for optimal sample loading onto the precolumn were evaluated using the test mixture containing [3H4]-DHEA sulfate. The occurrence of any sample breakthrough from the 794 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003
Figure 2. Effect of the introduction of splitter B and the precolumn on separation efficiency. Analyses were carried out using the Quattro Ultima instrument, and total ion chromatograms (TIC) are shown. TIC (a) was obtained with the analytical column directly connected to splitter A; TIC (b) was obtained after insertion of splitter B and the precolumn between splitter A and the analytical column (see Figure 1). In both cases, isocratic elution was performed with 30% A and 70% B solvent (i.e., 62% methanol with 10 mM ammonium acetate) at a flow rate of 20 µL/min through the open splitter A (split ratio ∼1: 100). One microliter of the test mixture containing 10 ng of each steroid sulfate was injected with splitter A open, that is, 100 pg was admitted to the column(s). Time is given in minutes.
precolumn depends on the capacity factors (k′) for the analytes in the injection solvent and the plate height of the precolumn. In reversed-phase liquid chromatography, more polar compounds (low k′) are more likely to be lost as a result of their inefficient absorption onto the precolumn. For this reason [3H4]-DHEA sulfate was chosen as a test compound because it is a polar steroid sulfate. The optimal condition, that is, minimum time for sample loading with a convenient operating pressure, was found to exist when the pump was operated at a constant pressure of 200 bar using A solvent (10% methanol containing 10 mM ammonium acetate) as injection solvent. This generated a flow rate of 2 µL/ min through the precolumn (splitter A closed, splitter B open), so it takes only 10 min for 20 µL of sample to pass through the precolumn. It is important to note that the sample loading conditions are dependent on analytes, precolumn, injection solvent, and flow rate. The presence of the two splitters in the system introduces the possibility of sample loss by diversion through the splitters. However, the experiments with [3H4]-DHEA sulfate showed that the loss of sample due to misdirection through splitter A is insignificant, provided that the capillary is carefully closed and a short period of pressure equilibration is allowed. It is important to avoid formation of an air cushion in the Valco union when it is closed. The experiments with [3H4]-DHEA sulfate showed an average recovery of 85% at the end of the analytical column. The loss of sample through splitter B during the sample injection was
