Biomarkers for an individual biological condition can be of outstanding value for mapping a normal or pathophysiological process or for studying the response to a therapeutic intervention. Specific proteins are of particular interest as biomarkers, as their concentrations can usually be measured cheaply and reliably. Moreover, the enormous progress in the development of proteomic techniques provides highly-specific and sensitive technologies for the identification of individual proteins, including potential biomarkers 1 .

Consequently, the protein profiling of body fluids is of great interest for the identification of biomarker candidates. Albumin and immunoglobulins (Igs) are the main protein constituents of plasma or serum and other body fluids. In particular, it has been shown that albumin contributes up to 60% of the protein content of cerebrospinal fluid (CSF), which is considerably more than in plasma 2 . However, potential biomarkers are usually proteins of lower abundance and these are often easier to detect after depletion of highly-abundant species. Various procedures for depleting albumin and/or Igs have been described in the literature (e.g., 3 4 5 6 7 8 9 ), and diverse “depletion kits” can be obtained from different commercial sources which, in turn, have been tested for several parameters (e.g., 4 10 11 ).

Unfortunately, these depletion kits have been tested for only a limited number of donor species (usually human beings and rodents). On the other hand, there is increasing interest in the development of dog models for diseases, such as the Hurler syndrome 12 , the Sanfilippo syndrome 13 or Duchenne muscular dystrophy 14 . This interest, together with the first proteomic studies directed at the identification of CSF biomarkers for canine diseases 15 16 , prompted us to investigate the applicability of commercial depletion kits for dog CSF.

CSF samples from eight healthy adult Beagle dogs, four male and four females, obtained by lumbar puncture, were purchased from Seralab (Haywards Heath, UK). The provider guarantees that the specimens were collected at USDA inspected and approved facilities. These samples (pooled pairwise - one male, one female) were centrifuged (2,000 × g for 10 min) to remove cells or other insoluble substances. Proteins were precipitated overnight at 20°C by adding four sample volumes of ice-cold acetone. The proteins were pelleted by centrifugation (10,000 × g, 30 min, 4°C). The supernatant was removed and the pellet was washed with ice-cold acetone (90%). The samples were vortexed and centrifuged again (10,000 × g, 30 min, 4°C). The supernatant was discarded and the pellets were dried at room temperature (30 min). Control experiments using dog serum were performed (see Additional file 1) to confirm the compatibility of this procedure, which has been recommended for proteomic investigations using CSF 17 . The ProteoSeek™ Albumin/Ig Removal Kit (Thermo-Fisher, Bonn, Germany) was applied for albumin and Ig depletion, as based on an affinity resin column. According to the manufacturer, this kit has been positively tested with human, monkey, swine and rabbit samples. Depletion experiments directed at albumin and Ig were also carried out using the ProteoExtract Albumin Removal Kit (Merck Millipore, Darmstadt, Germany, tested by the producer for human, rabbit, rat, mouse, pig and bovine samples, with decreasing activity) and by the antibody-based Vivapure Anti-HSA/Ig Kit (Sartorius, Göttingen, Germany), designed for humans, but also reactive in rat and mouse samples (according to the product information). Proteins obtained from at least three different pairwise pooled CSF samples were applied to each kit according to the manufacturer’s instructions after acetone precipitation. Optimisation procedures suggested by the manufacturers (such as prolonged and/or repeated sample application, extended mixing) did not clearly improve the depletion results for the kits when tested in dog serum. The protein quantities subjected to depletion (100 μg each) were always lower than the upper limits specified by the respective instructions, by a factor of at least 5. Elution from the depletion columns was performed as suggested by manufacturer’s protocols (if available) or by using Laemmli buffer. Flow-through (FT) and elution fractions (E) of CSF were reconstituted in Laemmli buffer. Cysteine residues were reduced using dithiothreitol (10 mM, 15 min, 60°C) and alkylated (55 mM iodoacetamide, 15 min, 20°C) in the dark. All fractions (FT and E of each kit) were separated electrophoretically on precast Novex Tris/Glycine 4%-20% gels (Life Technologies, Darmstadt, Germany). Proteins were visualised using Coomassie Brilliant Blue G-250, and the resulting band patterns obtained from at least three different samples were qualitatively equivalent for the respective kit. All clearly stained bands of eluate fractions (originating from not less than two independent samples for each kit) were excised for subsequent mass spectrometry (MS). After in-gel digestion of proteins and peptide extraction 18 , MS was conducted on a LTQ-Orbitrap XL mass spectrometer (Thermo Scientific) equipped with a reverse-phase capillary liquid chromatography system (Eksigent 2D nanoflow LC, Axel Semrau GmbH, Sprockhövel, Germany). The generated peak lists and the MASCOT server (version 2.2, Matrix Science Ltd., London/UK) were used to search against the non-redundant NCBI database (version 131022) restricted to mammals (1,974,848 sequences). A maximum of two missed cleavages was allowed, and the mass tolerance of precursor and sequence ions was set to 10 ppm and 0.35 Da, respectively. The program Scaffold (version 4.2.1, Proteome Software Inc., Portland OR, USA) was used to validate MS/MS-based peptide and protein identifications, which were accepted if established at greater than 99.0% probability with at least two identified peptides.

Different depletion kits for albumin and Igs were tested using dog CSF samples. Numerous other products for depletion are commercially available 19 . However, data providing an assessment of depletion kits is sparse even for human CSF, and several authors have found that these depletion kits give relatively low reproducibility. Previous work has shown that more than 95% albumin was depleted from human CSF with the ProteoSeek kit, resulting in a 20% increase in the number of detected proteins (from 135 to 163) 20 . The Vivapure anti-HSA/IgG kit was applied to human CSF and depleted albumin less effectively than the Sigma anti-HSA/IgG column or the ProteaPrep HSA/IgG column 21 . The Vivapure kit was found to be superior to the ProteoExtract kit for the analysis of the recovery of the prostate-specific antigen in human serum samples 22 . On the other hand, in studies on human serum, ProteoExtract has been found to give satisfactory reproducibility and specificity 23 . Other authors (e.g., 24 ) used the ProteoExtract kit for depleting human CSF, without reporting problems.

The Coomassie-stained gels shown in Figure 1 demonstrate that none of the kits gave satisfactory depletion of albumin and Igs from dog CSF samples. The ProteoSeek kit clearly exhibited the highest efficacy with respect to albumin; the amounts depleted from CSF by Vivapure and ProteoExtract kits were lower or negligible, respectively. Proteins identified in bands of the eluate gel lanes are given in Table 1 which lists proteins with substantial abundance, i.e., exponentially modified protein abundance index (emPAI) > 1.0. For approximately 56% of the canine proteins found, at least one isoform coincides with the human CSF proteome previously reported 25 . Albumin was enriched in the eluate fractions obtained from the ProteoSeek and Vivapure kits, but only one low abundant fragment was found in the eluate from the ProteoExtract kit (see Additional file 2 showing all eluted proteins identified by mass spectrometry). In agreement with the protein staining, immunoglobulin fragments were detected with lower emPAI values in the eluate fractions obtained by application of the depletion kits. However, additional proteins were identified by MS in all these fractions. Significant quantities of haemoglobin (obviously a contamination of the CSF samples) were found in the eluate fractions of all kits tested. As expected from the gel staining, the greatest co-depletion was observed for the ProteoSeek kit (115 proteins detected in total), but numerous non-targeted proteins have been also identified in samples obtained with the Vivapure kit (total proteins, 64), whereas only 3 detectable proteins were eluted from the columns of the ProteoExtract kit (see Additional file 2. Table showing all eluted proteins identified by mass spectrometry).

¹The complete list of identified proteins is available at Additional file 2.

²Accession number in NCBI database.

³M.W., Protein molecular weight in kDa.

⁴The emPAI value expresses mass spectrometry-based quantitation of protein abundance and is calculated according to emPAI = 10 N observed N observable − 1 , (N _observed , N _observable , number of observed/observable peptides); cf. http://www.matrixscience.com/help/quant_empai_help.html for more details.

Unspecific binding, either to the matrix or to the bait material, is a well-known phenomenon occurring in affinity purification approaches. In addition, one has to expect that proteins bound to albumin will recover in the eluate fractions. All of the proteins identified in the eluate were also present in the recently characterised CSF “albuminome” 21 for ProteoExtract, compared to 44% for Vivapure and 30% for ProteoSeek, although this comparison does not consider different experimental conditions and species. Since albuminome data were obtained in human CSF samples, it is likely that the co-depletion observed in our experiments results from both albumin binding and unspecific association.

In conclusion, our results do not support the use of these kits for depletion of albumin and Igs from dog CSF samples. Even with the most active kit tested, albumin and Igs were only incompletely removed. This was accompanied by co-depletion of numerous other proteins, a particular drawback for the identification of potential biomarkers.