Improving acquisition times of elemental bio-imaging for quadrupole-based LA-ICP-MS

Abstract

Elemental bio-imaging experiments by quadrupole-based LA-ICP-MS normally employ scan speeds where the distance traversed in one second is equal to or less than the diameter of the laser beam. Consequently, data for a higher-resolution (pixel size = 15 μm²) image of a 5 mm² tissue section can take upwards of 30 h to acquire. Appropriate laser scan speeds may be calculated by consideration of the relationship between laser scan speed, laser spot diameter and the total scan cycle of the quadrupole mass analyser. This paper presents a simple method to calculate the laser scan speeds capable of reducing the acquisition time by up to a factor of 5 whilst maintaining dimensional integrity of the image.

---

Introduction

Elemental Bio-Imaging (EBI) is an application of laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) that determines in situ trace element concentrations in thin sections of biological tissues.^1–3 Typical data acquisition times for a 5 mm² specimen vary from over two hours for a low resolution image (pixel size = 100 μm²) to more than 30 h for a more detailed, higher resolution image (pixel size = 15 μm²).^1–11 For many biological applications, high resolution images are desirable to show fine detail, such as the substantia nigra in brains of Parkinsonism mouse models.¹¹ Higher-resolution EBI images using LA-ICP-MS require significant reductions in analysis time in order to remain at the forefront of accessible μm-scale trace element imaging technology.

EBI utilises a focused laser beam directed onto the surface of a tissue sample. Material is ablated and swept to the ICP by a carrier gas where positive ions are produced and transferred to a mass analyser. An aperture wheel through which the incident beam is passed is used to alter the size of the laser beam spot, typically producing circular spots in the μm-diameter range. The sample stage is moved at a defined speed, passing the tissue surface through a series of continuous laser pulses. Therefore, the resulting ablated line is essentially rectangular in shape. A mass spectrum for each measured mass is recorded in a time-resolved manner, and these spectra are reduced to two-dimensional maps that represent relative signal intensity as a function of the image dimensions. Quantitative information can be collected using multi-point matrix-matched tissue standards that are used to convert the signal intensities to relative concentrations.¹¹

Quadrupoles are the most commonly used mass analysers for ICP-MS and quadrupole-based systems account for around 95% of all ICP-MS systems used today.¹² This type of mass analyser acquires data for an allocated mass to charge ratio (m/z) for a specified time period, or dwell time, before moving to the next m/z. This is continued until data has been acquired for all designated m/z. The scan cycle begins again and this continues for the length of each acquisition period, or in this case, the time required for the laser ablation system to complete a single ablation line. The quadrupole mass analyser is therefore not a simultaneous data collection technique, however it is able to rapidly acquire information for each m/z.^13,14 Quadrupoles are advantageous over simultaneous sector-field multi-collector instruments due mostly to their reduced cost and ability to analyse a wide range of masses in a single experiment.

Dwell time is a variable that can be adjusted in the ICP-MS software. Typical imaging experiments will measure several masses in the same acquisition. Different m/z values can be assigned specific dwell times independent of other measured masses. The scan cycle, or sampling period, is simply the sum of dwell times for each m/z plus the time it takes for the quadrupole to move between masses. For imaging experiments each scan cycle equates to one data point, with each data point corresponding to one pixel in the final image. Typical image construction methods are independent from dwell times for m/z, as each pixel is expressed as counts per second. Thus the pixel size will remain constant for all m/z. Triglav et al.¹⁵ recently described a comprehensive study of all variable laser parameters involved in imaging experiments with the aim of identifying ideal parameters for optimal imaging. Of these parameters, only spot diameter and laser scan speed determine the total acquisition time for an image.

This paper presents an investigation into the relationship between laser scan speed, quadrupole dwell time, and image resolution for the purposes of speeding up typical total acquisition times for image construction.

Experimental section

Instrumental

All experiments were performed on an Agilent 7500ce Series ICP-MS (Forrest Hill, Victoria, Australia) coupled to a New Wave UP-213 laser ablation unit (Kennelec, Mitcham, Victoria, Australia), equipped with an Nd:YAG laser emitting in the fifth harmonic at 213 nm. A New Wave Large Format Cell (LFC) with a x-y-z stage was used. To increase sensitivity, the ce lens assembly was replaced with a cs model in the ICP-MS interface. All experiments used high purity argon (Ace Cryogenics, Castle Hill, New South Wales, Australia) as the carrier gas. For maximum sensitivity and to ensure low levels of oxides, the LA-ICP-MS was tuned daily using NIST 612 Trace Elements in glass. Low oxide production was assured by an m/z 248/232 ratio (representing ²³²Th¹⁶O⁺/²³²Th⁺) that was consistently less than 0.3%. Solution nebulisation ICP-MS was performed on the Agilent 7500ce Series ICP-MS fitted with an I-AS autosampler (Agilent, Australia). Typical operational parameters for both SN- and LA-ICP-MS are given in Table 1.

Table 1 Typical SN/LA-ICP-MS Operational Parameters

Agilent 7500ce ICP-MS			New Wave UP-213 laser ablation
	LA	SN
RF power, W	1250	1500	Wavelength, nm	213
Cooling gas flow rate, L min^−1	15.0	15.0	Repetition frequency, Hz	20
Carrier gas flow rate, L min^−1	1.20	1.20	Laser energy density, J cm^−2	0.3 to 0.5
Sample depth/mm	4.0	8.0	Spot diameter, μm	15; 30; 65; 100
QP Bias, V	−3.0	−3.0	Line spacing, μm	≥15; 30; 65; 100
OctP Bias, V	−6.0	−6.0	Monitored in brain section analysis, m/z	13, 31, 55, 56, 57, 63, 66
Dwell time, s per m/z	Varied	0.1
Extracts 1;2, V	5; −100	0; −140	Monitored in all other experiments, m/z	13, 24, 31, 43, 44, 55, 56, 57, 59, 60, 63, 66, 85, 88
Collision gas	None	He

---

Tissue standard preparation

Quantification experiments were run using 30 μm thick sections of standard chicken breast tissue, adapted from a method described previously.¹¹ Chicken breast was removed of any fat or connective tissue and subsequently spiked with standard Mg, Ca, Mn, Fe, Co, Ni, Cu, Zn, Rb and Sr solutions. Solutions were prepared using appropriate soluble metal salts (Sigma-Aldrich, Castle Hill, NSW, Australia) dissolved in 1% HNO₃ and made up to concentrations of approximately 100 000 μg mL⁻¹ and 10 000 μg mL⁻¹. Aliquots of the chicken breast were then spiked with varying concentrations of each of the elements and homogenised using an OmniTech TH tissue homogeniser fitted with a polycarbonate probe (Kelly Scientific, North Sydney, NSW, Australia). Six ca. 250 mg aliquots of each homogenised tissue standard were microwave digested and analysed using solution ICP-MS to confirm the concentration and homogeneity of each element in the tissue standards. The concentrations and standard deviations of each m/z are presented in Supplementary Table 1†. The spiked tissue was frozen and sectioned into 30 μm sections and placed onto glass microscope slides for analysis.

Determining limiting signal

The limiting signal was calculated equivalent to a limit of quantification (LOQ) using a gas blank for each analysis. The limiting signal was calculated as:

Limiting Signal = y₀ + 10s₀

where y₀ is the average of the gas blank signal and s₀ is the standard deviation of the gas blank signal.¹⁴

Mass spectrometer dwell times

The shortest dwell times of the mass spectrometer that maintained sufficient signal intensity, i.e. raw counts greater than the limiting signal, were determined by analysing the lowest concentration tissue standard for 14 isotopes. Limiting signal experiments were performed using 11 dwell times for each measured mass (0.001, 0.005, 0.01, 0.015, 0.02, 0.025, 0.03, 0.035, 0.04, 0.045 and 0.05 s). A 30 μm spot diameter was used for each experiment run at a scan speed (v_l) of 3, 4 and 5 times the spot diameter (x_s) per second. With each increasing v_l, the laser fluence was increased to ensure sufficient tissue ablation across all scan speeds. Each of the experiments was repeated five times for statistical purposes.

Brain section analysis

Brain sections were sourced from adult wild-type mice prepared by the Mental Health Research Institute (MHRI), Victoria, Australia. All methods conformed to the Australian National Health and Medical Research Council published code of practice for animal research and was approved by the University of Melbourne Animal Ethics Committee. One hemisphere of each brain was sectioned at 30 μm and mounted on microscope slides. Sections taken at similar stereotaxic coordinates were analysed for ease of comparison.

Three brain sections were ablated using x_s at 15, 65 and 100 μm. Each brain section was divided into five evenly defined segments with each segment imaged at different speeds, namely v_l of 1, 2, 3, 4 and 5 times x_s per second. The scan cycles (t_sc) were 1.33, 0.665, 0.443, 0.333 and 0.266 s, respectively. Quantification was performed using matrix-matched tissue standards described earlier. A new calibration curve was constructed for every change in v_l and x_s allowing comparison of images on the same scale. A fourth brain section prepared from a wild-type mouse was imaged using t_sc and v_l which maintained the samples relative dimensions.

Image processing was performed with a macro in Microsoft Excel 2003/2007 which normalised each m/z to m/z = 13, representing ¹³C, i.e. the signal at each m/z was divided by the ¹³C signal.¹⁶ The macro arranged the data in a format suitable to be read by the hyperspectral imaging software ITT Visual ENVI 4.2 (Research Systems Inc., Boulder, CO, USA). Background signals for each m/z obtained from the gas blank were subtracted to improve image clarity and account for daily variation in the background levels of m/z = 55 and 56 in particular. Linear regression analysis of the average background-corrected signal intensity of each ablated standard was performed to produce calibration data, which was then used to convert all pixels from signal intensity to μg g⁻¹.

Signal carry-over experiments

Experiments were performed to determine the number of data points required for adequate washout after ablation. Using the highest concentration tissue standard, a line was ablated half on tissue and half on the glass mount for x_s of 15, 30, 65 and 100 μm and v_l equal to 1, 2, 3, 4, and 5 times x_s per second. Laser fluence was altered for each increase in v_l to ensure sufficient ablation of the tissue as the laser spent less time at each point on the tissue with increasing velocity.

Results and discussion

Image construction

Typical bio-imaging experiments normally employ scan speeds where the distance traversed in one second is equal to or less than the diameter of the laser beam. For example, a X μm laser spot diameter moving at a scan speed of X μm s⁻¹ would move X μm in one second. If the t_sc is equal to one second, a square pixel representative of X μm is recorded every scan cycle. If t_sc is the only variable altered and is either greater or less than one second, a true representation of the relative dimensions is lost. As all pixels are squares, the resulting image is either compressed or expanded. In the case of the former, a loss of resolution occurs.

This is best described by consideration of Fig. 1. The distance taken for one sampling cycle to complete based on an X μm spot diameter and t_sc equal to one second and varying only scan speed of X μm s⁻¹ is schematically represented. Each box represents one data point that is converted to a single pixel when the image is processed. Box A provides a true representation of the dimensions of the original tissue section. Box B consists of two times the distance for a sample cycle to complete and fewer data points will be collected on a line scan equidistant to that of Box A. Therefore the image will be compressed in the direction of the line scan and results in a loss of resolution. Box C is the opposite of this. More data points are collected and the resulting image has more pixels and therefore higher resolution in the X direction.

Fig. 1 The effect of varying the laser scan speed on image dimensions.

The benefit of using t_sc less than one second is that scan speed may be increased whilst maintaining the relative dimensions of the original tissue section through image post-processing. If t_sc is known, the scan speed can be calculated in order to maintain a resolution equivalent to X μm². The original relative dimensions of the sample is maintained if the scan speed is equal to the spot diameter divided by t_sc as described by eqn (1), where v_l is the laser scan speed and x_s is the laser spot diameter.

v_l = x_s/t_sc (1)

Limitations to t_sc

In order to estimate theoretical maximum laser scan speeds when using eqn (1), the lowest dwell times for biologically significant isotopes were determined, based on determining the limiting signal. The raw counts per data point and limiting signal (averaged at v_l of 3, 4 and 5 times x_s per second) were plotted against increasing dwell time (selected examples are shown in Fig. 2). A two-tailed t-test (ρ = 0.01) confirmed that the minimum dwell time required to exceed the limiting signal was applicable to each tested laser scan speed for each m/z.

Fig. 2 Counts per data point versus the dwell time of the quadrupole MS compared with the limiting signal at each dwell time for selected m/z. (v_l = 3 x_s μms⁻¹).

The minimum dwell time cut-off for each m/z was determined as the minimum dwell time at which the raw counts per point was at least 1000 counts per second and above the limiting signal. These dwell times and maximum v_l for each experiment for each measured m/z are available as Supplementary Table 2†. For example, an analysis of m/z 13, 31, 55, 56, 57, 63 and 66, produced a t_sc of 0.146 s, not including the time the quadrupole took to move between masses. Using eqn (1) and a x_s of 30 μm, the maximum v_l is 205 μms⁻¹. The maximum v_l for these seven m/z is more than 6.8 times a v_l of 30μms⁻¹. This would reduce an 8.5 h runtime for a 5 mm² sample to a less than 2 h. Higher maximum v_l may be employed when fewer masses are measured as the minimum dwell times per m/z may be increased whilst maintaining tissue dimensions. These values are instrument-specific and must be determined for each quadrupole ICP-MS system.

Washout times

Analysis of the signal carry-over from preceding data points was performed to determine the effect of increasing v_l on washout times and signal intensity. Increasing washout time increases the interference of the signal from a previous data point on following data points, resulting in image “blurring”. At v_l of 1, 2 and 3 times x_s per second (x_s = 15, 30 and 65 μm), between two and three data points after completion of the ablation of tissue is required to reach background signal levels (see Supplementary Figure 1†). Four to six data points were required to reach background for v_l of 4 and 5 times. At x_s of 100 μm up to 8 data points were required to reach background. In general the washout times increased with increasing scan speed. This was expected as the amount of material ablated per unit time increased as v_l increased. The longer washout times were not of sufficient magnitude to significantly affect the quality of images acquired at higher speed.

Laser fluence

In all experiments, the laser's fluence was adjusted to ablate sufficient material without penetrating to the glass substrate. This was to ensure that variations in tissue density and thickness could be normalised by dividing the signal by m/z = 13, i.e.¹³C. Increasing scan speeds required increasing the laser's energy as the laser spent less time at each point, reducing the total energy transferred to the tissue. Higher scan speeds also improved the normalised signal (Supplementary Figure 2†). This was due to an increasing sample load on the plasma.

Imaging with increased laser scan speed

Representative brain sections at similar stereotaxic coordinates were imaged to test the applicability of eqn (1). In all cases the images were acquired with t_sc values approximately 30% greater than equivalent dimensions. These values were calculated relative to v_l in order to allow whole number multiples of v_l. The software allows only two significant figures up to three decimal places for each dwell time. Consequently the t_sc of all experiments was approximately 30% more than that required for image equivalent dimensions, resulting in image compression of 30%.

Nevertheless, Fig. 3 shows representative images of copper as the scan speed was increased from 1 to 5 times x_s per second at resolutions of 15, 65 and 100 μm². As the scan speed increased the t_sc were proportionally decreased. Visual inspection of Fig. 3 indicated that there were no significant variations in calculated concentrations between scan speeds or image resolutions. The relative dimensions of each segment (i.e 30% compression relative to sample) were also maintained at each of the scan speeds, indicating that eqn (1) was applicable.

Fig. 3 Laser scan speed and resolution comparison for Copper-63 using spot diameters of 15, 65 and 100μm.

Supplementary Figure 3† shows a representative image of copper of a coronal slice of a mouse brain with t_sc selected to maintain the sample dimensions. This section was ablated with x_s of 30μm, t_sc of 0.2372 s and v_l of 127μms⁻¹. This resulted in a total acquisition time of 6.5 h, representing a 4.2 times reduction of acquisition time (cf. typical data acquisition parameters), whilst maintaining the sample's relative dimensions and similar sensitivities.

Conclusions

Decreasing the scan cycle of the quadrupole and increasing the laser scan speed was a viable means for reducing the acquisition times of EBI. By increasing v_l and decreasing t_sc it was possible to construct images of tissue sections in a shorter period of time for faster turnaround of samples. An equation was deduced to calculate the v_l required to maintain sample dimensions when using a specific x_s and t_sc of each measured m/z.

The minimum mass spectrometer dwell times for each m/z were based on the signal intensity and limiting signal for the sampling period. A maximum v_l may be calculated from dwell times, the number of m/z measured, and the detection limits required. These dwell times are specific to the ICP-MS system and experiments must be repeated for each system. Limiting signal experiments must be performed for each set of standards.

The number of data points for the signal to return to baseline levels was generally equivalent for v_l equal to 1, 2 and 3 times x_s per second. The number of data points increased for scan speeds of 4 and 5 times the spot diameter, however this increase was not significant. The upper limits of increased speed are yet to be determined, and likely to be limited by the scan speed of the quadrupole. Washout of ablated material from the sample cell and transfer lines also plays a significant role in blurring of images and carry-over.

References

1. D. Hare, in Chemistry and Forensic Science, University of Technology, Sydney, 2009.
2. D. Hare, S. Tolmachev, A. James, D. Bishop, C. Austin, F. Fryer and P. Doble, Anal. Chem., 2010, 82, 3176–3182.
3. D. Hare, F. Burger, C. Austin, F. Fryer, R. Grimm, B. Reedy, A. Scolyer Richard, F. Thompson John and P. Doble, Analyst, 2009, 134, 450–453.
4. J. S. Becker, R. C. Dietrich, A. Matusch, D. Pozebon and V. L. Dressler, Spectrochim. Acta, Part B, 2008, 63, 1248–1252.
5. J. Dobrowolska, M. Dehnhardt, A. Matusch, M. Zoriy, N. Palomero-Gallagher, P. Koscielniak, K. Zilles and J. S. Becker, Talanta, 2008, 74, 717–723.
6. B. Wu, M. Zoriy, Y. Chen and J. S. Becker, Talanta, 2009, 78, 132–137.
7. M. Zoriy, A. Matusch, T. Spruss and J. S. Becker, Int. J. Mass Spectrom., 2007, 260, 102–106.
8. M. V. Zoriy and J. S. Becker, Int. J. Mass Spectrom., 2007, 264, 175–180.
9. M. V. Zoriy, M. Dehnhardt, A. Matusch and J. S. Becker, Spectrochim. Acta, Part B, 2008, 63, 375–382.
10. M. V. Zoriy, M. Dehnhardt, G. Reifenberger, K. Zilles and J. S. Becker, Int. J. Mass Spectrom., 2006, 257, 27–33.
11. D. Hare, B. Reedy, R. Grimm, S. Wilkins, I. Volitakis, J. George, R. A. Cherny, A. I. Bush, I. Finkelstein David and P. Doble, Metallomics, 2009, 1, 53.
12. D. Potter, J. Anal. At. Spectrom., 2008, 23, 690–693.
13. R. Thomas, Practical Guide to ICP-MS, Marcel Dekker, Inc., New York, 2004.
14. D. C. Harris, Quantitative Chemical Analysis, W.H. Freeman and Co., 2002.
15. J. Triglav, J. T. van Elteren and V. S. Selih, Anal. Chem., 2010, 82, 8153–8160.
16. C. Austin, F. Fryer, J. Lear, D. Bishop, D. Hare, T. Rawling, L. Kirkup, A. McDonagh and P. Doble, J. Anal. At. Spectrom., 2011, 26, 1494–1501.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c1ja10301f

‡ Equal senior author.

---