created: 10/09/97, last updated: 10/09/97,© 1997 ABRF
Michael T. Davis Beckman Research Institute
Introduction
The practical application of off- and on-line micro electrospray techniques for the analysis of simple and complex peptide mixtures are presented along with detailed instructions for the construction of the respective interfaces. Particular attention is given to the construction of the microspray compatible micro capillary column and on-column flowcell as well as the modification of a Finnigan MAT LCQ quadrupole ion trap mass spectrometer for micro electrospray analyses. The relevant advantages and disadvantages of the nanospray and microspray LC/MS approach to peptide mixture analysis are illustrated using a simple standard mixture (peptide digest of cytochrome c) and a complex in-gel digestion extract. For the purpose of clarity, the terms nanospray and microspray, respectively, refer to the off- and on-line forms of micro electrospray.
The advent of electrospray ionization techniques has made the practical coupling of liquid chromatography to mass spectrometry an almost pedestrian effort (1). Instrumentation development has progressed (or regressed, depending upon your point of view) to the point where column flowrates up to the ml/minute range can be tolerated by today's mass spectrometers while still obtaining high quality mass spectral information. It's unfortunate, however, that the samples of the highest biological interest are nearly always present in limited amounts and are not usually amenable to analysis by the prevailing chromatographic formats. Microcapillary liquid chromatography with less than 0.5 mm ID HPLC columns, is an efficient format which provides improved sensitivity, decreased sample requirements, and decreased solvent consumption. The availability of commercial microcapillary columns along with compatible sample injectors and UV detectors has outpaced the development of truly micro flow solvent delivery systems. Although several commercial solvent delivery systems now promise high performance at low microliter per minute flowrates, most users of these formats must resort to stream splitting techniques to reduce the output from their HPLC systems to the desired level. The recent development of on-line micro electrospray techniques (2,3) have placed further demands on the instrumentation available from commercial suppliers and are generally being performed using individually customized equipment.
Stream splitting can be accomplished in a variety of ways but can generally be defined as being either pre-injector or post-column. Pre- injector splitting maximizes the efficient use of trace samples because 100% of the injected sample is directed to a micro column and ultimately to the mass spectrometer, but wastes solvent. Post-column splitting allows the user to operate a larger column format at its optimal flowrate but ultimately wastes sample because only a portion of the effluent is delivered to the mass spectrometer. The bulk of the column effluent may be collected, however, for orthogonal analyses such as Edman sequencing, AAA, or bioactivity assays. The needs of the project, the type of analyses desired, and the amount of available sample dictates which approach is selected.
Interchangeable interfaces for on- and off-line micro electrospray analyses are presented in Figure 1, along with a detailed diagram of the internal structure of the microspray needle. The salient features of both interfaces are a one to two micron orifice and a integral platinum electrode. They differ only in the manner in which the sample is delivered to the tip, and because a continuous solvent stream is delivered through the microspray needle, an integral filter must be included to prevent clogging. Nanospray (4) techniques are best suited for the analysis of single components that have been previously purified in an off-line fashion or for the partial characterization of complex mixtures. Microspray LC/MS analysis is suitable for these types of samples as well as the comprehensive analysis of complex mixtures. The common features of a micro electrospray interface is the delivery of a minute amount of sample at a reduced flowrate (nanoliters per minute and less) without the assistance of an organic make-up solvent to stabilize the spraying process.
Figure 1. Interchangeable micro electrospray interfaces for on-line (A) and off-line (B) applications. The internal detail of the microspray needle illustrates the placement of an integral filter. The best performance in terms of spray stability, performance over a range of solvent compositions (gradients), and the lowest possible electrospray potential (important for positioning relative to the MS) is accomplished using the smallest tip feasible. Our standard tip has a 1-2 µm orifice and an optimal electrospray potential between 550-700 volts.
The relevant HPLC system parameters for microcapillary columns are presented in Table 1. The larger formats (320 - 530 µm ID) are ideally suited for sample purification and off-line peptide mapping experiments with manual fraction collection. The resulting fractions are amenable to any of the orthogonal methods described above and direct mass spectral analysis using Nanospray. The mid-size formats are recommended for conventional LC/MS techniques involving sheath liquids and nebulizing gases, although the smaller formats can also be used. The 100-mm ID column represents the largest recommended format for sheathless LC/MS analysis using a micro electrospray interface.
|
Column ID |
Flow (µl/min.) |
Sample Amount1 |
Flowcell2 |
Transfer lines |
|
530 µm |
15 - 20 |
10 - 100 pmoles |
LC Packings "U" cell |
50 µm ID |
|
320 µm |
6 - 8 |
2 - 50 pmoles |
250 µm "O-C" |
25-50 µm ID |
|
250 µm |
2 - 3 |
0.25 - 10 pmoles |
150 µm "O-C" |
25 - 50 µm ID |
|
150 µm |
1 - 2 |
0.10 - 5 pmoles |
150 µm "O-C" |
25 µm ID |
|
100 µm |
< 0.3 |
0.05 - 1 pmole |
100 µm "O-C" |
20 µm ID |
Table 1. Microcolumn HPLC System Parameters
Syringe Pumps Recommended - Pumps with 10-ml syringes require stream splitting for all column formats (pre-column split recommended). Pumps with 2-2.5 ml syringes can be used directly as low as 2-3 ml/min - stream splitting is recommended below 5 ml/min. 1Sample amounts are recommended by experience and do not reflect the theoretical values that would be calculated from the ratios of the cross-sectional areas of the respective columns. In all cases, column selection is a compromise between column capacity, resolution, flowrate, and the goal of the separation (fraction collection or LC/MS). 2"O-C" refers to on-column flowcells using a PE/ABD HT capillary flowcell holder.
All of the presented formats permit UV detection by either a commercial micro flowcell or by using an on-column flowcell/holder design. Useful UV absorbance information can be obtained to approximately 50-100 femtomoles of material. The UV trace is always valuable when using the extremely low flowrates because it is the only reliable means of monitoring changes in the delivery of the gradient and, hence, changes in flow.
Variable Flow Chromatography
An alternative solvent delivery system optimized for the delivery of micro scale gradients has been developed at the City of Hope and has been described in detail elsewhere (5). This system is distinguished by its use of a preformed gradient for column elution and the use of pressure programming to regulate column flow. Samples are injected onto the column at a ballistic flowrate at 4000 psi. The micro gradient is formed and stored in an off-line loop during the sample injection and is in turn "injected" onto the column immediately following the sample slug. The system pressure is reduced following the injection of the gradient in a controlled fashion down to the running flowrate. The time compression advantage of the variable flow system is illustrated in Figure 2 using a 250-µm ID column and a running flowrate of 2 ml/min (20 ml/min loading flowrate). By using pressure programming, the last component in the cytochrome c digest standard has eluted in less than 25 minutes in Figure 2A while the first component has yet to elute using constant flow conditions (Figure 2B). The periodic cycling of the system pressure is controlled and has little effect on column lifetime or performance. These systems are compatible with all the column formats described in Table 1 and require minimal modification for optimal performance between formats. Furthermore, the ability to perform a stop-flow (peak parking) analysis under a single eluting peak has been demonstrated using the microspray interface to extend the analysis time available to the mass spectrometer without compromising the chromatographic resolution (3,6). The operative range for the design is, therefore, between 20 nanoliters and 20 microliters per minute depending upon the column format and the type of analysis required. Retention time (RT) reproducibility was examined using the 250-µm ID column format described above with fair results (5). The concept of RT reproducibility is cast aside during peak parking since time is no longer a constant.
Figure 2. The use of pressure programming in the variable flow LC system to eliminate the effects of system dead volumes and their associated delays. Sample injection was at time zero while gradient injection (GI) was delayed until the UV baseline had stabilized following the injection artifact. (A) Pressure programmed variable flow - sample loading at 4000 psi (20 ml/min), gradient injection at loading pressure with immediate ramp to running pressure and flowrate (200 psi and 2 ml/min). (B) Sample loading and gradient injection under continuous running pressure/flowrate conditions (2 ml/min). A 10 pmole aliquot of an endo-Lys C digestion of equine cytochrome c and a 250-µm ID column format was used for these comparisons. See Davis, et al. (5) for details.
Micro Capillary Column Construction
The 100-µm micro column/flowcell format recommended for variable flow LC/MS analyses is diagrammed in Figure 3. This format performs optimally at flowrates from 150 - 300 nanoliters/minute with sheathless ionization using a microspray interface. The column and flowcell are constructed from 100-µm ID x 350-µm OD fused silica capillary (FSC) and are connected using 20-µm ID x 90-µm OD FSC transfer lines. The standard column length is 15 cm and the length of the transfer line is the distance from the end of the column to the UV detector, plus 5-10 cm. The top of the column is trimmed periodically to remove debris and the extra length allows for column shortening in the future. Column performance is maintained through several cycles of shortening. With reasonable care and periodic trimming, most columns can last several months depending upon the amounts and types of samples analyzed. Teflon (Zitex), Hydrophilic-PVDF (Duropore) and GF/A membranes are all suitable for use as frits with equivalent sample recoveries. The frit is constructed using the column as a "cookie" cutter to cut a circular portion of the frit membrane and positioned within the column to a depth of 1 cm using the post-column transfer line as a ramrod. A small spot of two-part epoxy is applied to the point of insertion to seal the connection. A thin coat of epoxy is applied over the overlapped area to strengthen the assembly (this area is extremely fragile and breaks easily if bent). The minimum column curing time is approximately one hour, but most columns are allowed to cure overnight.
Figure 3. Micro capillary column for variable flow micro electrospray analyses. All connections are sealed with a standard two-part epoxy.
The columns are packed at 6000 psi using acetonitrile as the packing solvent and any pump that can run at its preset pressure limit. The pressure is maintained against a closed SSI two-way valve. A Valco ZDV reducing union (1/16" - 1/32") with a 0.4-mm ferrule serves as the column fitting. The slurry reservoir is constructed from a 10-cm length of 1/16" x 0.02" ID stainless steel tubing with each end terminating in a Valco fitting (column volume:reservoir volume = 1:10). A 1/16" ZDV union is connected to the "pump" side of the reservoir. Column packing is as follows:
1. The slurry reservoir is "backfilled" using a 1-cc syringe fitted with a blunt 16-gauge needle and an Upchurch fingertight fitting. The reservoir is full when slurry is observed in the syringe. A simple slurry of two volumes of methanol to one part silica (Vydac, C18, 5 µm) is sufficient and should be briefly vortexed before use.
2. The filled reservoir is rapidly attached to the column fitting (fingertight at first) and the syringe is removed. The coupled reservoir - column assembly is then attached to the SSI valve (closed against 6000 psi) and the connections are wrench tightened,
3. Open the valve and pack the column. The column should pack almost instantaneously. Turn the pump off and allow the pressure to bleed out through the column once the column has packed. A rapid decompression will disrupt the column bed.
On-Column Flowcell Construction
Although an excellent micro flowcell is available commercially, we use the PE/ABD HT capillary flowcell holder with an on-column flowcell. The flowcell is repairable, disposable and, in our opinion, nearly as sensitive as the LC Packings "U" cell (7) at a much lower cost. The on-column flowcell is constructed from a 3 to 4 cm section of 100-µm ID FSC (the same material as the column) with a 2 to 3 mm section of the polyimide coating removed to create the UV lightpath. As with the micro column, all joints are sealed with epoxy. The transfer lines are opposed across a 0.6 to 1 mm gap within the flowcell lightpath to create the actual flowcell. The polyimide is removed from the tips of the transfer lines to prevent creeping of the coatings into the flowcell lightpath. The end of the transfer line can be "butted" using a sleeve of FSC identical to the material used for the flowcell. Restoring the transfer line to 350-µm OD simplifies downstream connections to unions, microspray electrodes, etc.
At this point you have constructed and packed a capillary column, attached the on-column flowcell and the transfer line which ultimately interfaces with the mass spectrometer. The PE/ABD HT flowcell is self-aligning in one axis but needs to be centered within the transfer line gap. This can be performed under a low power (8x) dissecting scope, or with an 8x ocular loop over a light box.
On-Line Microspray Needle Construction
The on-line microspray needle is best drawn using a microprocessor controlled laser-based micropipette puller (Sutter Instrument Co., Novato, CA). Suitable tips may be pulled by hand, but reproducibility is difficult to achieve. Although their analytical performance is equivalent, the hand-pulled tip typically features thicker walls and is sufficiently robust to withstand a careful manual cleaning. The laser-pulled tip is thinner and all contact with the tip must be avoided, but it can be reliably produced at the touch of a button. The art of manually pulling 1 to 2 µm tips is an acquired skill and not generally recommended. There are commercial sources of pulled tips (e.g., New Objective, Cambridge, MA) so the frugal mass spectrometrist might pursue this avenue. The needle is drawn from 150-µm ID x 350-µm OD FSC and is fitted with a discontinuous transfer line (25-µm ID x 150-µm OD) to position a membrane frit and to consume the otherwise fatal deadvolume (see Figure 1). The critical step in assembling the finished needle is the insertion of the first section of transfer line. Because this section will be in front of the filter, it has to be inserted cleanly without chipping or fraying the polyimide coating or the tip will clog. The clean assembly of microspray needles is best performed under a low power (6 to 10x) dissecting scope.
Cut and position the frit in the manner described for the micro column using the hydrophilic PVDF or Zitex membrane only; GF/A can fragment and clog the tip.
Insert another transfer line to cut the frit and seat the first piece within the taper of the tip. Retracting the second transfer line slightly before cutting will leave a small gap between the frit and the second transfer line upon reinsertion . The membrane frit will last longer if it is left uncompressed and a small gap (less than 0.5 mm) will have a minimal effect on peak broadening. The finished needle is used "as is" without an epoxy seal. Although this seems to be counterintuitive, the tight tolerance between the internal transfer line and the pulled tip limits the side-streaming of solvent that one might expect to occur. Unfortunately, the best tips feature the tightest tolerance and can be difficult to make; although the tolerance of all tips improve with use due to swelling of the polyimide cladding over the internal FSC. Also, direct solvent contact with the epoxy cement can produce an overwhelming chemical background, especially if it is near the electrospray electrode.
Tip lifetime is variable and dependent upon the usual, sample related, parameters. Typical lifetimes of 1 to 2 days are routine with some tips lasting upward of 50 hours of use (more than one week). Poor assembly technique (chipped transfer line) is the root cause of most tip failures and occurs immediately upon use at 4000 psi.
The Liquid-Metal Interface
The interface serves to connect the LC transfer line to the microspray needle while providing a liquid - metal junction for the application of the electrospray potential. We use a platinum sheath electrode (22-gauge tube, about 400-µm ID, Hamilton, Reno, NV) secured over the end of the LC transfer line using the two-part epoxy. By positioning the transfer line two to three mm short of flush within the distal end of the electrode, the transfer line - microspray needle interface is contained within the platinum tube. The 50-µm tolerance between the wall of the electrode and the FSC needle and transfer line has an unnoticeable effect on peak width as long as the needle-transfer line junction is tight. The electrode is 2.5-cm long and sealed with a minimal amount of epoxy. As mentioned above, the epoxy cement will contribute an overwhelming chemical background if it contacts the liquid stream, so the minimal amount is recommended. The union itself can be floated (2) but may produce iron adducts over time or contribute a strong, complex, background from the graphite ferrules. This liquid-metal interface has been used for several months with dozens of needle changes without producing the chemical background previously attributed to graphite ferrules or epoxy. The platinum tube is easily recycled by burning away the epoxy resin with a Micronox torch (Alltech Associates) followed by flushing with methanol.
Nanospray Needle and Interface Assembly
The nanospray source design of Wilm and Mann is commercially available but incompatible with the on-line microspray interface described above. We currently use a simple nanospray interface which is interchangeable with our on-line device (refer to Figure 1B). The modified interface uses the same glass micropipettes popularized by Wilm and Mann but utilizes an internal platinum wire electrode. Samples can be loaded into the tip using "gel-loading" Eppendorf tips, and ionization is initiated without pneumatic assistance due to the internal electrode. Also, the tip is never intentionally broken. There is no overwhelming reason to use an internal electrode rather than the "coated" needles of Wilm and Mann, except the needles can be used immediately. Since the electrode is internal, solvent out-gassing can be observed if the electrospray potential is too high. In our lab, the best performance is typically obtained from tips that "optimize" (high SNR) at an electrospray potential between 600-700 volts. Bubble formation within the taper may be fatal to this interface since it lacks the pneumatic assistance feature of the Wilm and Mann design.
Instrument Modifications&emdash;Finnigan MAT LCQ
The modification of the normal LCQ ESI interface for microspray analysis involves the removal of the standard front-end, and its replacement with a multi-axis translational stage for precise and stable control of the needle position. The tolerance between the needle and the sample inlet is typically 0 to 200 µm and the tips are extremely fragile. We use a 460A-XYZ translational stage with SM-13 micrometers (Newport Corp., Newport Beach, CA) mounted onto a custom dove-tailed platform (Figure 4). A video system has been incorporated to provide a remote, magnified, image of the needle position. The system mounted on the LCQ uses a Sony SSC-20 CCD camera coupled through a 15-mm extension tube to a 18 to 108 mm, f2.5 C-mount video lens. The video output is displayed using a Sony PVM-1390 color monitor. A Schott #1150 fiber optic light source with a single light guide is used for illumination. This system provides a safe working distance of 6 to 8 centimeters and a final magnification of 40 to 50 fold. The high voltage connection has been modified to terminate in an alligator clip to facilitate its connection to the micro electrospray interfaces.
Figure 4. Modified API interface for micro electrospray analyses using the Finnigan Mat LCQ quadrupole ion trap mass spectrometer. The sheath and auxiliary gas lines are sealed together in a single FPLC fitting to maintain gas pressure at the sensor without actual gas flow. The video signal is fed to a 13 inch monitor which rests on top of the mass spectrometer (not shown).
Off and On-Line Micro Electrospray MS/MS Analyses
The comparative sensitivities of the nanospray and microspray (off- vs. on-line) analysis of a simple digest mixture at the low femtomole level is presented in Figure 5. The nanospray analysis was performed using the unseparated cytochrome c digest mixture at a concentration of 100 fmoles/ml (1 ml loaded). The same amount was injected onto the micro column for the on-line analysis. The "Zoom Scan" ( a narrow mass range, higher resolution, scanning method used by the LCQ to obtain accurate mass information) analysis of the 2+ ion at m/z 675.8 from 100 fmoles of a standard equine cytochrome c digest was obtained at a signal to noise ratio (SNR) of 3.5:1 using nanospray (Figure 5A). The SNR of the same analysis is improved 20-fold in the on-line result (Figure 5C) using variable flow techniques and "peak parking" (stop-flow). The CID spectra obtained by both methods (Figure 5, B and D) are identical, but an order of magnitude greater ion injection time was required for the nanospray analysis. Since the LCQ features "Automatic Gain Control" (AGC), these analyses were actually performed using the same number of ions and, therefore, produced nearly identical spectra. The longer injection time required to fill the trap with the targeted number of ions during the nanospray analysis reflects the weaker signal of the off-line method. The flowrate during each analysis presented in this figure was approximately 10-20 nanoliters/minute. Also, the ion at m/z 675.5 in Figure 5B is likely to be a singly charged contaminant ion and not residual parent species since both of the MS2 experiments were performed under identical conditions. The relative collision energy put into the system was below the threshold for inducing the fragmentation of the singly charged ion so the remainder of the spectra is unaffected. This contaminant has been separated from the ion of interest in the on-line analysis. Overall, the improved SNR of the "Zoom Scan" analysis and the lower injection time of the MS2 experiment are in good agreement with the predicted sample concentration effect of the reverse-phase column.
Figure 5. The comparative sensitivity of a nanospray and variable flow microspray analyses at the 100-femtomole level. The nanospray analysis was performed using 1 µl of a 100-femtomoles/µl dilution of the standard cytochrome c digest mixture in 1% acetic acid. The variable flow analysis was performed using a 0.4 µl injection of a 250-femtomole/µl dilution of the same sample. Zoom scan (higher resolution, narrow mass range scans) and MS2 analysis of the 2+ ion at m/z 675.5 were performed using the nanospray (A and B, respectively) and variable flow techniques (C and D). The number of scans averaged for the "Zoom" spectra was 14 (A and C), while 10 scans were averaged for the MS2 result (B and D). The average ion-injection time was calculated from the data provided within the spectrum scan header. The ion at 675.5 m/z observed in the nanospray MS2 experiment (B) is suspected to be a singly charged contaminant.
The nanospray "Zoom Scan" analysis (Figure 5A) includes a singly charged ion at m/z 678.3 which illustrates the downside of analyzing an unseparated mixture. As the complexity of the sample increases, the probability of observing multiple components at the same m/z increases as well. Although the selective fragmentation of ions of the same m/z but different charge state and, hence, different mass can be achieved by careful selection of the CID parameters, this will not work with isobaric ions. As the complexity of the analysis increases, the sequence coverage obtained from an unseparated analysis should decrease. The proof of this hypothesis is presented in Figure 6. The variable flow analysis of an in-gel digestion mixture from a single Coomassie blue stained spot detected five ions in the region of m/z 689 and obtained CID spectra for four of them. Each of the CID spectra were matched via a Sequest database search to either of two 58 kDa yeast proteins (6). The unfragmented singly charged peptide at m/z 688.3 in Figure 6G could have been sequenced by directing the CID analysis to its 2+ charge state and none of these sequences could have been derived from an unseparated mixture.
Figure 6. The mass redundancy of an in-gel digest mixture observed under variable flow conditions. Four peptides of overlapping m/z were separated and identified using the variable flow technique. High resolution Zoom scans followed by MS2 analysis confirmed the sequence of a 1+ ion at m/z 689.3 (A and B, respectively), a 2+ ion at m/z 688.3 (C and D), a 1+ ion at m/z 688.3 (E and F) and a 2+ ion at 687.8 (G and H). The 1+ ion at m/z 688.3 in G was unfragmented in the MS2 experiment (H).
Conclusions
Micro electrospray ionization techniques can be used in either on-line or off-line formats for the analysis of peptide digest mixtures present in low femtomole amounts. The limit of detection is improved 10 to 20 fold using an on-line concentration/separation technique over nanospray although the analysis time is limited to the width of the LC peak. The available analysis time can be extended using a variable flow approach to provide sufficient time for parent ion selection and CID parameter optimization to produce high quality spectra. The selection of which micro electrospray method to use will depend upon the goal of each analysis. The identification of proteins contained within a 2-D gel is almost universally approached using the nanospray technique due to it's inherent simplicity (8). Any peptide sequence can be used in a database search. However, if specific structural information is required, then, in theory, the on-line approach has a higher probability of success. In summary, microspray and nanospray are not mutually exclusive methods for the analysis of peptide mixtures. Each method has its merits and its drawbacks, but they are complementary and it would be prudent to develop the ability to perform both.
Acknowledgments
The author wishes to acknowledge the contributions of Douglas Stahl, Kristine Swiderek, and Terry Lee for their input over the course of several years of collaboration, and Lowell Ericsson, Carol M. Beach, and Clayton Naeve for their assistance with this manuscript.
References
Mike Davis may be contacted at the Division of Immunology, Beckman Research Institute, City of Hope, 1450 East Duarte Road, Duarte CA, E-mail: mdavis@smtplink.coh.org.
The following additional illustrations have been provided to supplement the above article:
Return to the The ABRF Home Page