WP 3: Methods for the Traceable Quantification of Metal Containing Biomolecules as Potential Biomarkers
The aim of this WP is to develop primary analytical methods such as species specific IDMS for accurate quantification of several metalloproteins related to AD (i.e. Fe-albumin (ALB), Cu-ALB, Fe-transferrin (TRF), Cu/Zn-SOD, CER, FER) in small sample volumes in the µL range (i.e. in serum, CFS, brain homogenate). The permeation of metals through neuronal barriers (from serum to CSF) may vary between healthy and diseased individuals so that the role of such metalloproteins as biomarkers for the diagnosis of AD should be investigated. The determination of total metal/metallospecies (e.g. total Cu/Cu-CER) ratios in serum versus CSF (measurements in both matrices) could also shed some light in this respect.
Sensitive and reliable methods are urgently needed, which are able to identify and quantify metalloproteins involved in AD, as this will help to better understand the origin and progress of the disease and, therefore, to prevent or treat AD.
When IDMS is applied to speciation studies it is called species specific IDMS (SS-IDMS) and in this case, it makes use of a spike compound identical to the compound under investigation but isotopically labelled in the target element. This concept, described previously for covalently bound organic compounds (e.g. tributyltin), was extended to the following metalloproteins: transferrin (TRF), ceruloplasmin (CER), Cu, Zn superoxide dismutase (SOD1) and haemoglobin (HGB) in the previous EMRP project HLT05 Metalloproteins, where candidate reference measurement procedures were developed for TRF and CER in serum and for SOD1 and HGB in erythrocytes. However, the concentration of these proteins in CSF and brain tissues (the matrices under study here) are expected to be lower.
Due to their potential involvement in the progression of AD, two extra proteins (FER and ALB) were included in this project, which have required the development of new separation and quantification methods. As it was mentioned before, the metalloprotein is quantified via its metal or heteroatom by ICP-MS (e.g. Fe in TRF, Cu in ALB or CER). And only if stoichiometric ratio of metal-protein and the molecular weight of the protein are known, the protein mass fraction can be calculated (e.g. 2 moles Fe bound to 1 mol TRF). Therefore, different measurands have been pursued within the project: metal bound to the protein (e.g. Fe bound to FER or Cu bound to ALB) and the protein content (e.g. FER via S and CER via Cu).
Model samples of the three different matrices, selected by the consortium at the beginning of the project, were used for method development. Examples of commercially available sample sources used here are given below:
- Serum: ERM DA470k/IFCC, quality control (QC) Seronorm Trace Level 1 (Seron, Norway), QC Human Seronorm (Sero), fetal calf serum (Merck), etc.
- CSF: QC Liquid CSF Control 2 Randox (Randox Laboratories Ltd), artificial CSF (Tocris, Harvard Apparatus), etc.
- Brain: locally purchased animal brains from different sources such as pig and bovine.
A standard operating procedure (SOP) about the preparation of artificial CSF as model sample was created and shared between the partners to be used for method development. Two different procedures were used: a protein free solution based on glucose and different electrolytes in a tris(hydroxymethyl)aminomethane (Tris) buffer solution and the second one contaning bovine serum ALB at different concentrations in order to simulate protein content in CSF.
The preservation of the biomolecules under study (metal bound to the protein, protein activity and monomer form, among others) is a requirement and a special challenge in any analytical method and particularly for speciation analysis. Special attention was paid to sample handling and storage conditions of biomolecules for which no sample handling SOPs were previously developed or published. In this sense, protein stability in the matrices under study was evaluated under different storage conditions for SOD1, TRF, FER and Cu-ALB.
In general, samples should be stored cooled or frozen in low protein binding tubes. Freeze/thaw cycles should be avoided, so it is always recommended to prepare aliquots of the protein solutions and samples and store them at -80 °C for single use. In the particular case of SOD1, for short-term storage (up to two weeks) a storage temperature of 4 °C was found suitable, whereas for longer periods (up to six weeks) the samples should be stored at -20 °C. For long-term storage (several months) a storage temperature of -80 °C is recommended. Brain tissue samples should be stored at -80 °C. Futhermore, it was found that human SOD1 is less stable in pig brain matrix compared to the artificial CSF samples.
Moreover, the storage temperature was found to have a significant effect on the formation of FER oligomers in commercially available FER and fetal calf serum (FCS). Freezing of FCS led to the highest FER oligomer formation. On the other hand, lowest formation was observed when the serum sample was stored at 4 – 8 °C. Therefore, a sample storage temperature in this range is recommended when analysing FER to ensure it is in its monomeric form and to achieve a good separation from other Fe containing biomolecules.
In order to carry out metalloprotein speciation studies in tissue samples, the first required step is the extraction of the metallo-species into a liquid phase without any alteration of its chemical form. Sodium dodecyl sulfate (SDS), an ionic detergent widely used for the rapid disruption of biological membranes, should be avoided for speciation analysis of non-covalently bound metalloproteins. It breaks the non-covalent bonds and S-S bridges of the proteins, denaturing them and the protein loses its structure. Therefore, the use of SDS is not recommended when the structure and the activity of the proteins have to be preserved.
A dedicated SOP for the extraction of cytosolic proteins from brain tissues using a Tris buffer saline (TBS-buffer) was also developed and sucessfully applied for the extraction of SOD1 and FER from pig brain tissue. The extract can be used for the determination of either proteins per se or for further enzymatic digestion of proteins. This buffer is especially useful for tau-protein and determination of native proteins or their tryptic digests by ICP-MS, since it is relative low in S or P containing compounds. Moerover, a heat denaturation protocol of the extracted proteins was used for semi-purification of heat-stable proteins such as SOD1, FER and tau-protein.
The development of SS IDMS methods requires several steps:
· the separation of the analyte protein from interfering components in the biological matrix
· the production and characterisation of the species-specific spike material (see section 4.2 for more detail) and calibrant (reference material)
· both analyte and spike isotopes should be free of spectral interferences. These are most likely to occur in the sample blend due to the sample matrix
Therefore, isotopically labelled spike and standard reference materials were produced and characterised for the target metalloproteins (see section 4.2 for more details). After establishing appropriate separation and detection methods in the different biological matrices under study, the whole procedure of SS ICP-IDMS was carefully validated according to metrological principles and a comprehensive uncertainty budget was estimated. Potential reference measurement procedures based on SS-IDMS for the determination of TRF, Cu-ALB and SOD1 at low concentrations in body fluids (serum, CSF and brain homogenates) were developed achieving the target uncertainties of < 15%.
Transferrin
The method previously developed in the EMRP project HLT05 for the quantification of TRF in serum was adapted to lower protein concentrations expected in CSF (around 100 times lower) and in brain tissue samples. The separation of TRF from other biomolecules in the three matrices under study (human serum, a pooled CSF from a clinical laboratory and bovine brain homogenates) was achieved using a strong anion exchange column (MonoQ GL 5/50 HPLC column, Pharmacia) with an ammonium acetate gradient for elution. In the case of tissue samples, TRF was spiked at the concentration expected in CSF. During sample preparation, neither lipid precipitation nor ALB removal were applied. The determination of the TRF via its Fe content was conducted by collision cell ICP-MS/MS using hydrogen. Triple SS-IDMS was applied and TRF mass fraction values with target expanded uncertainties < 15 % were achieved: ~ 6 % for ERM-DA470k/IFCC and CSF (TRF value 100 times lower than in serum) and ~ 12 % for the spiked bovine brain matrix.
Copper bound to albumin
First experiments were focused on the separation of Cu-containing proteins (main target Cu-ALB) and Fe-TRF by complementary monolithic liquid chromatography and FFF coupled to ICP-MS/MS using standards and/or matrix matched standards. When the complementary monolithic liquid chromatography separation method was applied to serum samples, a poor resolution of the main Cu-containing proteins (ALB and CER) was obtained, hampering the accurate quantification of Cu bound to ALB. Finally, a good separation was achieved by strong anion exchange HPLC in serum and CSF (QC Liquid CSF Control 2). The method was later applied to Wilson’s disease serum samples to get a better understanding of the distribution of Cu species in patient samples. The validation of the developed double SS-IDMS approach for the determination of Cu-ALB in a serum matrix was performed on three independent days obtaining an expanded uncertainty of ~5 %.
Cu, Zn-Superoxide dismutase
The quantification method previously developed for SOD1 using strong anion exchange chromatography was successfully transferred from erythrocytes to CSF and brain matrix. The QC Liquid CSF and pig brain were used as CSF and brain model samples, respectively. Since no native human protein was present, native SOD1 was spiked at the beginning of the sample preparation for proof of concept. Double SS-IDMS was used in this candidate measurement procedure for the quantification of SOD1 via its Cu content. By measuring the Cu containing part of the SOD1, only the amount of active protein is quantified, which is the important quantity. The quantification of SOD1 in CSF was successfully applied to the stability measurements mentioned previously. The method could not be applied to pig brain matrix since the human SOD1 calibrant and spike materials were different from the porcine SOD1 present, eluting at different retention times due to different isoelectric points (6.8 and 5.0, respectively).
Ferritin
Two different quantification approaches were pursued in this project for the quantification of FER: via its a) Fe content and b) S content. Size exclusion chromatography (SEC) is a state-of-the-art separation technique, which has the potential to fractionate FER from other Fe containing biomolecules according to their molecular weight. The hyphenation of SEC to ICP-MS, and particularly to a ICP-MS/MS using oxygen as reaction gas, has allowed the simultaneous determination of Fe and S content in FER.
a) FER bound Fe
The Fe content is directly related to the amount of Fe atoms stored in the FER cage and S present in its backbone. Consequently, FER bound Fe can be determined via the Fe to S ratio of FER. A SS-IDMS method for the quantification of FER bound Fe in pig brain homogenates was developed using the isotopically enriched 57Fe-FER spike (see section 4.2). Measurements were performed using an on-line fractionation method for FER from other metal containing proteins such as TRF and ALB using ultra performance liquid chromatography (UPLC) SEC-ICP-MS. The quantification of FER was not possible when the samples were frozen or lyophilised due to changes in the native form of FER (oligomerisation) or the binding to other proteins.
b) FER via S
A complementary approach is the quantification of FER at the whole protein level via the S content in the protein backbone without the need of peptide digestion using HPLC-ICP-MS/MS. Strategies including SEC, heat treatment, ALB depletion, desalting, ultrafiltration and immune affinity chromatography were tested for the separation of FER from interfering matrix compounds in serum without success due to the binding of FER to other proteins. As alternative, a quantification of FER via specific peptides is now attempted.
The determination of FER presented the biggest challenge among all the metalloproteins. FER oligomerisation was observed to increase over time and depending on the storage temperature conditions. Although special attention was paid to convert FER oligomers into monomers, a decrease in the protein content was observed. Moreover, the presence of other proteins, mainly ALB, in the serum matrix hampered the quantification of FER. Therefore, development of ALB depletion protocols was required for the determination of FER. This challenge was also common in the quantification of CER in CSF. Furthermore, it was observed that FER is bound to other FER binding proteins in serum and CSF, the more heavier chains the protein contains (FER is a complex of 24 chains containing a varying amount of light and heavy chains). This leads to a different behaviour of FER spike/calibration material and FER originally contained in serum /CSF.
Ceruloplasmin
The method for the quantification of CER in CSF was adapted from the previous one developed in serum in the project HLT05. Different kits were tested to achieve a complete separation of CER from ALB on a SEC column. The use of an ALB depletion kit from Agilent turned out to be the most successful. However, the CER spike production method previously developed was not reproducible. Therefore, its procedure had to be modified by replacing potassium thiocyanide (KSCN) with potassium cyanide (KCN) as demetallation agent. First SS-IDMS measurements of CER in CSF proved the feasibility of the developed method in this type of matrix.
A new CER immunoprecipitation method was also developed, based on protein-coated nanodiamonds, for the production of enriched CER fractions from human serum and CSF samples for Cu isotope ratio analysis. The proposed immunoprecipitation method was applied to human serum and CSF for the final Cu IR ratio by multicollector ICP-MS (see section 4.5).
Haemoglobin
The determination of HGB is used for the control of blood contamination in CSF. Therefore, PTB tried to transfer the method previously developed for its determination in blood. However, the quantification of HGB was not possible since the protein is not present in a free form in CSF but bound to haptoglobin which made its quantification with the developed species spike material for HGB impossible.
In order to show the capabilities of the methods developed for metals and metal containing proteins using isotopically labelled spike materials, an interlaboratory comparison was conducted at the end of the project. The sample consisted of a lyophilised human serum (Seronorm Human, Sero) used as a QC material to be used to monitor precision and trueness of laboratory measurement procedures, providing reference values of the following analytes under study: TRF, Fe, Zn and Cu. The material was distributed between the partners together with a preparation and stability protocol based on manufacter’s instructions.
Various partners contributed to the comparison with the measurement of different parameters. A good agreement between partners’ data and manufacture’s certificate was found for TRF and for Fe, Cu and Zn. The only common parameter among the participants was Cu and Zn isotope ratio in bulk serum. For the isotope ratio measurements, Cu values were comparable, however quite large uncertainties reported for Zn renders a comparison difficult. Mass fraction of Cu bound to ALB, S and Cu isotope ratio in the CER immunoprecipated fraction of the serum were also measured in this material.
Collaboration:
The cooperation within this consortium enabled the development of methods for a lot more metalloproteins than one institution alone would be able to deliver within this timeframe and with available resources. So was the method for TRF developed by TUBITAK, Cu bound to ALB by LGC, SOD1, CER and HGB by PTB and FER by both PTB and UNIVIE. Methods for the latter were investigated for the quantification via Fe by UNIVIE and via S by PTB, thus looking at the protein from different perspectives.
Key outputs and conclusions:
This objective was successfully completed, potential reference measurement procedures based on SS-IDMS were developed for metals and metal containing biomolecules. Target uncertainties < 15 % were achieved in clinical samples (serum, CSF and brain tissue homogenates) for the determination of SOD1, TRF and Cu bound to ALB. Feasibility of the developed SS-IDMS was proven also for CER and Fe bound to FER. Their validation could not be carried out within the lifetime of the project. In general, the determination of FER via S or Fe presented the biggest challenge. The preservation of FER in its native form and the presence of other interfering proteins, mainly ALB, hampered its quantification in serum and CSF. To address this, PTB is currently investigating a completely different approach using FER specific peptides after tryptic digestion.