Cellular metabolic profiling using ToF-SIMS

Abstract

Proteomic and metabolomic changes associated with cellular metabolism are important metabolic indicators that can give insight into cellular function. Metabolites are small molecular weight cellular components of mass typically <1500 Da and well suited for detection using ToF-SIMS. The advent of cluster ion beams, such as bismuth (Bi_n) and C₆₀, as primary ion sources, has improved the sputtering efficiencies, with higher secondary ion yields and lower damage cross sections, enabling better detection of these components. However, the spectral information is dominated by fragment peaks, matrix interference and preferential ionizations making it challenging to apply time-of-flight secondary ion mass spectrometry for non-targeted analyses. In this investigation, the challenges in detecting metabolic changes in cells and tissues are discussed, and areas that need further development are highlighted. Copyright © 2012 John Wiley & Sons, Ltd.

Introduction

The genomic revolution led to the sequencing of organisms, which currently approaches 3000 completely sequenced genomes and over 7000 ongoing projects (http://genomesonline.org/cgi-bin/GOLD/bin/gold.cgi). These efforts enable the genetic makeup of a biological system to be catalogued, but are not sufficient on their own in explaining how biological systems function. Consequently, there has been a greater emphasis on technologies that would elucidate the functional aspects of genes and gene products, in the ‘post-genome’ era. In this regard, ‘-omic’ technologies that enable large-scale assessments at the level of the mRNA (transcriptomics), proteins (proteomics) and metabolites (metabolomics) have emerged and are continually being developed. Proteomic and metabolomic changes associated with cellular metabolism are important metabolic indicators that can give insight into cellular function. The objective in these approaches is to characterise the system in as ‘holistic’ and comprehensive a manner as possible, necessitating non-targeted analyses, so an unbiased perspective can be obtained. However, in some cases, it may be possible, and in fact more desirable, to derive sufficient information from a targeted approach, where the fate of a few known proteins or metabolites can be tracked and studied.

Metabolites are low molecular weight cellular components (typically <1500 Da in mass) that are intermediates of metabolism and play key roles in mediating biochemical processes within the cell. Techniques that enable metabolic differences to be delineated between different physiological states of an organism or its exposure to different environmental conditions can enable a better understanding of the biological system and its functioning, even when a complete knowledge of the genetic makeup of the organism is not available.[1] Time-of-flight secondary ion mass spectrometry (ToF-SIMS) appears to be ideally suited for the detection of metabolites, in principle, as these are molecular species whose size range is within the detection capabilities of the technique. If cellular metabolic profiles can be generated using ToF-SIMS, such that discriminatory information can be gathered about the cells and/or its environment, it will be useful in understanding cellular function. Changes in cellular metabolite compositions have the potential to explain cellular function or dysfunction. The attractiveness of ToF-SIMS for generating cellular metabolic profiles is in its ability to detect intrinsic chemical changes at high data resolutions and the ability to analyse surface layers with minimal damage to the sample surface. This is particularly useful in imaging spatial distributions.

The advent of cluster ion beams, such as bismuth (Bi_n) and C₆₀, as primary ion sources, has improved the sputtering efficiencies, with higher secondary ion yields and lower damage cross sections possible, and their application in ToF-SIMS has enabled better detection of molecular species.[2-5] The employment of cluster ion beams in generating metabolic profiles should therefore yield potentially useful results. We are studying the application of the technique to generate metabolic profiles from different biological systems. Here, we discuss some of the challenges involved in the process of generating cellular metabolic profiles using ToF-SIMS.

Diagnostic ions

Metabolites are chemically diverse, and their composition in the cells can include a few hundreds to several thousand species that have concentrations ranging over four orders of magnitude or greater. In order to capture changes in metabolic profiles that are reflective of the biological phenomenon being investigated, it is essential that the technique used has the capability to detect changes in a reliable, reproducible and a sufficiently representative manner. ToF-SIMS typically results in spectral information that is dominated by fragment ion peaks, particularly those at very low masses (m/z <100). Since these can have multiple molecular origins, they are of lower diagnostic value in metabolic profiling. Nevertheless, it is possible to detect ions at higher masses, whose origins can be traced, as is well known with the detection of lipid species, for example the phosphocholine head group at m/z 184 [M + H]⁺, and cholesterol at m/z 369 [M-H₂O + H]⁺. To be of value in metabolic profiling, it is desirable to generate spectral information that contains molecular, protonated and deprotonated molecular ions from metabolites.

It is possible to generate molecular, and/or protonated and deprotonated molecular ions from metabolites in sufficient yields to render imaging, and differences in the spatial distribution to be picked up, as is seen in Fig. 1, for deprotonated metabolite species in a synthetic metabolite cocktail, in the negative ion mode. Here, a metabolite cocktail consisting of 29 representative metabolites including aspartate, glutamate, citrate, glucosamine-6-phosphate, fructose-1,6-bisphosphate and adenosine monophosphate (all from Sigma Aldrich, UK) was prepared and an aliquot (2 uL) was spotted onto a piranha solution cleaned silicon shard, air dried and imaged (300 µm², 128 × 128 pixels) using a SIMS V instrument (Ion-ToF, Germany), equipped with a 25 kV bismuth primary ion source (Bi₃⁺⁺). Molecular, protonated or deprotonated molecular ions can be more diagnostic, especially in non-targeted analyses. However, fragment ions and adducts still dominate the spectral information, even in synthetic cocktails, making spectral interpretations difficult. Ways to minimise fragmentation would improve yields of the more diagnostic species that can help interpretations. Deconvolution of metabolic information in non-targeted analyses in a biological matrix is still a considerable challenge.

Figure 1. Spatial distribution of metabolites detected in a synthetic cocktail on silicon. The negative ion ToF-SIMS spectrum is shown with ion images for representative deprotonated molecular ion signals [M-H]^-. Asp = aspartic acid; Glu – glutamic acid; Cit – citric acid; Gln6P – glucosamine 6 phosphate; FBP – fructose 1,6 bisphosphate; AMP – adenosine monophosphate.

Simultaneous multianalyte determinations

An advantage of ToF-SIMS in metabolic profiling is the ability of the technique to profile multiple analytes. In principle, it is possible to detect several metabolites in a given profiling experiment. It is also practical within reasonable limits to detect multiple components in the same sample, as is demonstrated in Fig. 1. However, the signal yields of the components detected vary significantly, even in simple mixtures containing equimolar concentrations of the metabolites (Fig. 1). Whilst this is expected, as different compounds ionize differently and hence result in different ion yields, it makes it difficult to apply the technique for non-targeted analyses, such as will be required for metabolic profiling. Developments in methodologies are required to enable non-targeted analyses to be performed to detect as many metabolites as is possible in a reliable robust manner. Current technology enables us to detect known metabolites in a targeted manner and monitor associated changes in signal intensities in cells or tissues in different physiological or pathological conditions. However, even here, matrix interferences can contribute to the changes in intensities of a known metabolite signal in different matrices,[6-8] necessitating the need to validate the observations with other modalities or corroborate evidences with further experimentations. Unless these challenges are met, the promise of using ToF-SIMS for cellular metabolic profiling will remain a promise and a bridge too far to cross.

Sample integrity

A major concern with the analysis of biological samples is in maintaining the chemical and biological integrity of the sample, especially when imaging and interpreting spatial distribution of metabolites. In this regard, growing cells on substrates that can be directly examined in the SIMS chamber would minimise sample intervention. It is desirable to keep the sample processing steps to a minimum, in order to avoid artifacts originating from these steps, and observe the system in as integral a state as is possible. Even when profiles are generated from ToF-SIMS spectra, the fast reaction rates in metabolic processes will necessitate appropriate steps be taken to quench metabolism rapidly, before metabolite extractions and/or analysis, so the metabolic profiles generated are representative of the biological phenomenon investigated.

For imaging spatial distributions, maintenance of low temperatures and preventing the exposure of the sample surface to oxygen and water vapour would minimise changes to the prepared surface before and during SIMS analysis. Cryo-techniques offer the required conditions to preserve sample integrity, in this regard. For example, direct analysis of cryo-fixed cells grown on silicon are likely to preserve the integrity of cell surfaces, so changes in the surface of the cell can be monitored (Fig. 2). Here, the cells (human dermal endothelial cells)[9] were grown directly on silicon, media components diluted by dip washing briefly in 150 mM ammonium acetate (three 3s dips), before snap freezing in liquid nitrogen, and freeze drying. This, in principle, quenches metabolism and removes the water from the cells, preventing changes from taking place in the cells immediately before and during analysis, so the chemical composition of the surface can be studied in an integral state. In such procedures, since the growth medium typically contains salts, it is required to wash the medium off the cell surface, using a mass spectrometry friendly buffer that does not disrupt the cells, as salts tend to suppress ion signals. Although retention of sample morphology before and after cryo-fixation can be taken as a sign of sample preservation, there is no guarantee that the chemical integrity of the sample is preserved, and ancillary techniques will be required to ascertain this fact, before interpretations can be made.

Figure 2. Endothelial cells (EC) can be cultivated in vitro on silicon surfaces that can be directly imaged using ToF-SIMS – the cells on silicon were washed with 150 mM ammonium acetate (3 dips of 3s each), flash frozen in liquid nitrogen and freeze dried, before imaging in the positive ion mode; the optical image of the imaged surface (300 µm²,128 × 128 pixels) is shown as well as an RGB overlay of three ion images (R – m/z 205, G – m/z 146 and B – m/z 184).

In order to access the intracellular space and study intracellular metabolite distributions, freeze-fracture[10-12] can be adopted for cells grown on silicon, as is shown for breast cancer cells (MCF-7) grown[13] on silicon (Fig. 3). In such procedures, it is essential to make sure that the cells are not squeezed between the silicon shards and that they do not loose integrity during the fracturing process. The use of spacer beads, as suggested by Chandra,[10] would minimise damage to the cells during the fracturing process. Fractured cells from both the silicon shards will be available for examination. However, the fracture plane may not always be reproducible, necessitating replicate examinations. Nevertheless, the use of beads of appropriate sizes, taking into consideration the size of the cells on the silicon shard, may help in defining the fracture plane. Techniques such as AFM and electron microscopy need to be used to ascertain that other than the fracture plane, the morphology of the cells and hence the chemical integrity of the sample is minimally disrupted.

Figure 3. Freeze-fractured breast cancer cells (MCF) – the cells were sandwiched between two silicon shards, a bottom shard on which the cells were grown and a top shard, which is shown – the shards were separated by spacer beads; the optical image is shown in the top and an RGB overlay of three ion images (R – m/z 255, G – m/z 265 and B – m/z 281), acquired in the negative ion mode (500 µm²,128 × 128 pixels) is shown at the bottom.

Alternatively, when the sample is large, as is the case with tissues, cryo-sections[14, 15] can be prepared and imaged (Fig. 4). Here, fresh frozen tissue (wheat seed at 4dpa) was cryo-sectioned to 15 µm thickness in a cryomicrotome (Leica CM1900) maintained at −15 to −20 °C, after embedding in optimum cutting temperature liquid and snap freezing in liquid nitrogen. The sections were cold mounted on to silicon shards and freeze dried. Creation of thin sections will minimises the influence of surface topography[16] on SIMS metabolic profiles, especially when imaging spatial distributions. In all cases, artifacts need to be minimised, and the technique optimised to preserve sample integrity. The method adopted and technique used in preparing the sample is sample specific and needs to be optimized for the respective sample, before meaningful interpretations can be made.

Figure 4. A freeze-dried wheat seed cross section (at 4 dpa) – the seed was snap frozen in liquid nitrogen, cryo-sectioned and freeze dried; the total ion image (5000 µm², 128 × 128 pixels) in the negative ion mode is shown above and an RGB overlay of ion images (R – m/z 145, G – m/z 133 and B – m/z 132) is shown below.

Metabolite identifications and spectral deconvolutions

A key challenge in the use of ToF-SIMS to cellular metabolic profiling is in the correlation of the ToF-SIMS signal with the metabolite identities that reflect the metabolic status of the system investigated. Currently, ToF-SIMS instrumentation allows only the generation of MS¹ data. This can at best result in the generation of exact mass information. Although putative identifications are possible with exact mass data, for example, the signals Figs. 2-4 can be assigned to phosphocholine (m/z 184), aspartate (m/z 132), malate (m/z 133), glutamine (m/z 145), palmitate (m/z 255) and oleate (m/z 281), further characterisations will be needed before interpretations are possible. The use of Q-ToF has been demonstrated[17] in generating MS² data that can be very useful in deriving additional information on the signals and help in correlations with metabolite identities. However, this facility needs to be more widely available. The combination of SIMS with other mass spectrometers, such as ion traps will be needed to help generate data that enables the metabolic origins of the SIMS signals to be identified. We have noticed that spectral information from metabolic extracts that show variations to discriminate cancer and non-cancer lines can be generated. However, correlation of spectral information with metabolic changes is not straightforward and requires development of appropriate deconvolution methodologies. The metabolic origins of the signals are contextual to the system investigated and need to be mined from existing information about the system. There are several databases that are increasingly becoming available, and these need to be availed and robust identification strategies devised. Although signals presumably from metabolites can be detected from biological cells and tissue sections under appropriate sample preparation conditions, as is seen in this investigation, the data density currently achievable with ToF-SIMS leaves much to be desired before the technique can be applied for metabolic profiling and its attractive features can be fully exploited.

Conclusions

ToF-SIMS can yield high-resolution data that can be used to derive cellular metabolic profiles and image spatial distribution of metabolites on the cell surface, as well as inside and outside the cell. However, there are considerable challenges that need to be addressed before in generating metabolic profiles using ToF-SIMS that can be reliable and of value in making biological interpretations. The challenges include: (i) increasing diagnostic ion yields, so a significant portion of the metabolome is detectable and identifiable; (ii) ensuring reproducible and reliable techniques are available that preserve sample integrity, both with respect to the chemical and biological status of the cells examined; (iii) devising strategies to deconvolve and correlate ToF-SIMS signals to metabolite identities so meaningful interpretations are possible, with respect to metabolic changes in the biological system investigated.

Acknowledgements

The author is grateful to EPSRC (grant # EP/I03453X/1) for support. The author provision of endothelial and breast cancer cell lines by Prof. Nicola Brown is gratefully acknowledged as is the help provided by Dr. Carolyn Staton and Mr. Ahtasham Raza for growth and processing of these cells. The author is also grateful to Prof. Mike Burrell and Mr. Fawaz Alzahari for the provision of wheat tissue sections. Assistance provided by Dr. Malinda Salim and Dr. Claire Hurley for help with sample processing and SIMS analysis is also gratefully acknowledged.