Mass spectrometry (MS) has today become an indispensable tool for protein quantitative detection and identification in systems biology, biomedical research, and clinical proteomics [1
]. However, its application to the detection of proteins present at low and ultralow concentrations in blood plasma or serum, which is of great importance for the search and verification of novel protein disease markers, is limited by the immense dynamic range of protein abundance levels that can span 12 orders of magnitude [2
]. While the concentration of high-abundance proteins ranges in plasma from micrograms to tens of milligrams per milliliter, the upper concentration boundary for low-abundance proteins (LAPs) is usually set as 100 ng/mL [6
]. LAPs include tissue-leakage proteins, which are of the most interest as putative biomarkers [9
]. For example, the well-known plasma tumor markers are LAPs present at concentrations well below 100 ng/mL [10
]. To enhance the sensitivity of MS detection, the plasma (serum) protein dynamic range can be reduced prior to MS analysis either by the removal of the most abundant proteins by means of immunoaffinity columns (immunoaffinity depletion) [11
] or by the immunoaffinity enrichment of target proteins with monoclonal antibodies (mAbs) immobilized on a solid phase carrier [12
]. As an alternative to antibodies, affinity-based reagents, such as DNA and RNA aptamers, have been suggested for a selective enrichment of target proteins [14
]. Aptamers have a number of technological advantages over antibodies. Exhibiting (at average) affinities to proteins comparable to those of antibodies, aptamers show no batch-to-batch variability in affinities to a target and are easily immobilized onto various solid phase supports without a loss of affinity via different chemical linkers, which can be attached to the 3'- or 5'-ends of aptamer oligonucleotides during chemical synthesis [15
The utility of aptamer-based affinity enrichment of targeted proteins for the purpose of MS analysis was shown in a number of publications [16
]. It is worth noting that the use of aptamers instead of mAbs was reported to result in a remarkable reduction of the complexity of tryptic hydrolysates due to the absence of peptides originating from mAbs [17
]. A coupling of the aptamer-based affinity enrichment of protein targets with different modes of MS was demonstrated, including liquid chromatography–mass spectrometry [23
] and liquid chromatography–tandem mass spectrometry [17
], as well as matrix-assisted laser desorption/ionization–time of flight–mass spectrometry (MALDI-ToF-MS) [16
]. Among the MS modes used, a particular method known as ‘selected reaction monitoring’ (SRM) is characterized by the exceptionally high selectivity and sensitivity of protein detection [27
], allowing the detection of proteins in cell lysates as low as in a few copies per cell for selected molecules [28
]. When coupled with immunoaffinity depletion, SRM is able to detect proteins down to picomolar concentrations in blood plasma [4
]. However, immunodepletion can result in a complete or partial loss of target proteins due to either their non-specific binding with resin and mAbs or their specific interaction with the removed proteins [29
]. While it has been shown that the detection sensitivity can benefit from the coupling of aptamer-based affinity enrichment with SRM for the detection of target proteins in cell lysates [18
], the question of how the amount of target protein in the enriched fraction, measured by SRM, relates to its initial amount in blood plasma in a wide range of protein concentrations is not answered yet.
In the present work, using recombinant protein SMAD4 (mothers against decapentaplegic homolog 4, the neXtProt accession number NX_Q13485) as a model target added at known concentrations to undepleted human blood plasma samples, we investigated the relationship between the initial amount of the target protein and its amount in the enriched fraction, as determined by SRM. The DNA aptamer constructed on the basis of the consensus sequence recognized by SMAD4 [30
] and immobilized on magnetic beads was used for the aptamer-based enrichment of the target protein. It was found that aptamer-based enrichment can enhance the sensitivity of SRM detection of recombinant SMAD4 (rSMAD4) by a factor of 18 to 26, depending on the protein concentration in blood plasma, that results in about a 30-fold decrease of the limit of detection.
3. Results and Discussion
SMAD4 is a common partner for proteins of the RSMAD (receptor-regulated SMADs) group involved in the formation of transcriptionally active complexes induced by transforming growth factor beta [39
]. Since SMAD4 is considered as a potential target for targeted cancer therapies [39
], we recently tried to select DNA aptamers against recombinant SMAD4 protein (rSMAD4) by the SELEX (Systematic Evolution of Ligands by EXponential enrichment) method [30
]. As a result, oligonucleotides containing 5'-GTCT-3’ and 5’-AGAC-3' sequences known as SMAD-binding elements (SBE) have evolved by the end of selection, thus indicating that the SMAD4–SBE interaction dominated the aptamer selection process. The SBE sequences were used to construct the anti-SMAD4 aptamer [30
], as shown in Table 1
. The equilibrium dissociation constant, Kd
, of the aptamer/SMAD4 complexes, measured by microscale thermophoresis, was found to equal (45 ± 7) nM.
Though SMAD4 is not considered as a potential circulating biomarker for cancer diagnostics, this is an intracellular protein that is not normally present in blood plasma at a detectable level and therefore will not interfere with the detection of rSMAD4 spiked in plasma. Moreover, the immobilization of the anti-SMAD4 aptamer via the biotin moiety (Table 1
) on streptavidin-coated magnetic beads was shown to allow the selective pull down of rSMAD4 from E. coli
cell lysates as demonstrated by electrophoretic analysis [30
]. Thus, rSMAD4 and the anti-SMAD4 aptamer were employed here as a convenient model for evaluating aptamers as affinity reagents for the enhancement of SRM-based detection of low-abundance proteins in blood plasma.
In SRM, the mass spectrometric analysis is carried out on a triple quadrupole mass spectrometer in which the first mass analyzer is set to filtering mode with a narrow isolation window (with a particular value of mass-to-charge ratio, m/z, and a width, as a rule, of ±1) to isolate a specific peptide (precursor ion) of interest. Then, the isolated peptide is fragmented, and one of the resulting fragment ions is monitored with another mass analyzer set to filter a certain mass-to-charge ratio. The double selection of a peptide precursor ion and a ‘daughter’ fragment ion is called a ‘transition’ and determines the high specificity of the SRM method, resulting in low chemical noise and, consequently, high sensitivity of the protein detection in complex samples [27
]. Peptides whose m/z ratios together with m/z ratios for daughter fragments uniquely identify a given protein are called ‘proteotypic peptides’ [40
]. For the absolute quantification of a target protein, the pertinent proteotyping peptides are synthesized using stable isotope-labeled amino acid precursors and spiked at known concentrations into analyzed samples as internal standards [41
Two peptides, GWGPDYPR and IYPSAYIK, were selected in silico as proteotypic for the SMAD4 protein, using bioinformatics tools, and chemically synthesized. Their isotope-labeled analogues were also chemically synthesized to serve as internal standards. In Table 2
, a set of corresponding transitions for both native and isotope-labeled (“heavy”) synthetic peptides is shown. Figure 1
presents the SRM spectra for these peptides. As seen, the relations between the signal intensities for the daughter fragment ions differ for the peptides: The domination of y6 ions in a spectrum is much more pronounced for IYPSAYIK, compared to GWGPDYPR, that makes the later peptide look more promising for confident SRM detection. However, when these two peptides were used for the SRM detection of rSMAD4 spiked into plasma, GWGPDYPR turned out to have a worse signal-to-noise ratio and a much greater variability in the detected amount of the protein at low rSMAD4 concentrations, compared to those for IYPSAYIK (data not shown). Apparently, a strong interference from a particular proteolytic matrix peptide (or peptides) is the reason behind the poor performance of GWGPDYPR as a proteotypic peptide for rSMAD4 detection. In contrast, no interference from matrix peptides was observed for peptide IYPSAYIK down to nanomolar concentrations (Figure 2
). Consequently, peptide IYPSAYIK and its “heavy” analogue were employed to quantify spiked rSMAD4 in samples of undepleted human plasma in the wide range of rSMAD4 concentrations, from approximately 40 pM up to about 30 µM, and in samples derived from the tested plasma by the anti-SMAD4 aptamer-based enrichment (further referred to as ‘pull-downs’ for simplicity).
shows the molar amount of peptide IYPSAYIK per 1 µg of the total tryptic peptides, measured by SRM in plasma samples and pull-downs, as a function of its expected molar amount calculated based on the known molar concentrations of rSMAD4 and the weight concentration of the total plasma protein (measured as 73 g/L). To calculate the expected molar amount of peptide IYPSAYIK per 1 µg of the total peptide, we assumed that the ratio of the SMAD4 molar concentration to the weight concentration of the total plasma protein holds throughout all sample preparation procedures. As seen from Figure 3
(curve 1), the measured amount of peptide IYPSAYIK is practically directly proportional to that expected from the SMAD4 concentration in plasma within the concentration range of ≈ 10 nM to ≈ 30 µM. Indeed, the linear regression analysis gives in this case the following equation: Y = (0.799 ± 0.005)·X + (0.7 ± 0.8), r = 0.999. The slope is not equal to unity, probably due to differences in the estimation of the protein amount by the BSA assay and the SRM quantification with an internal standard. Below ≈ 10 nM, the measured amount starts to deviate from direct proportionality to the expected values, which may be attributed to an increasing interference from matrix peptides. Finally, the dependence flattens out at SMAD4 concentrations of 40 pM to 1 nM since the signal becomes indistinguishable from the background noise. For the pull-downs, the corresponding dependence (curve 2 in Figure 3
) is shifted towards lower SMAD4 concentrations due to the aptamer-based enrichment of the target protein. The dependence can be approximated by two linear functions with different slopes for the concentration ranges from 1 nM up to ≈100 nM and from 1 nM down to ≈40 pM, viz.
Y = (18.5 ± 0.8)·X + (0.8 ± 0.5), r = 0.996, and Y = (26.0 ± 2.6)·X + (0.09 ± 0.02), r = 0.990. Thus, the aptamer-based enrichment allowed for about a 20-fold increase in the concentration of the target protein over the concentration range tested. The obtained data also clearly demonstrate that the coupling of SRM with the aptamer-based enrichment of SMAD4 allows one to reliably detect this target protein at concentrations that become ‘undetectable’ for the SRM method applied directly to undepleted plasma samples.
To estimate the analytical performance of SRM alone and SRM coupled with the aptamer-based enrichment, we compared the limits of detection (LOD) for both cases. By definition, LOD can be determined as follows [42
]: LOD = LOB + 1.645·SD, where LOB is a so-called ‘limit of blank’ and SD is a standard deviation for the measured values of the analyte. LOB is the highest apparent
analyte concentration expected to be found when replicates of a blank sample containing no analyte are tested. LOB can be calculated as LOB = meanblank
]. We took the corresponding averages from four experimental points on the flat section of the curve 1 in Figure 3
(background noise) as meanblank
that gives 86 amole (of the measured amount of peptide IYPSAYIK) as an estimate for LOB in our case. For the direct application of SRM to plasma, this corresponds to 2.2 nM of SMAD4 in plasma (curve 1 in Figure 3
) while for SRM coupled with the aptamer-based enrichment, 44 pM. To estimate the SD values, we averaged the standard deviations from two experimental points on each curve in Figure 3
, which lies right above the LOB. This gave us 6 nM and 0.2 nM of the SMAD4 concentration as estimates for the LOD in the case of SRM directly applied to plasma and coupled with the aptamer-based enrichment, respectively. Thus, the aptamer-based enrichment of SMAD4 provides a 30-fold analytical improvement for SRM detection of this target protein in undepleted plasma. Taking into account that the molecular weight of SMAD4 is around 60 kDa, LODs for its direct and aptamer-assisted SRM detection can be calculated as weight concentrations of 12 and 360 ng/mL. The LOD of 12 ng/mL matches the requirement for LAP detection, in contrast to that for SRM applied directly to plasma.
It appears interesting to compare the LOD achieved with the aptamer-based enrichment to LODs for SRM detection coupled with an antibody-based enrichment. For instance, the antibody-based enrichment of carcinoembryonic antigen allowed an LOD of 15 ng/mL [43
] to be achieved while for cardiac troponin, the antibody-based enrichment allowed its quantification down to 11 ng/mL [44
]. Recently, Säll et al. [45
] reported LODs of 0.05 to 12 ng/mL for 11 proteins detected in undepleted plasma by SRM coupled with the antibody-assisted enrichment. They used the oriented immobilization of different recombinant single-chain antibodies on magnetic beads. Apparently, an LOD for the affinity-based enhancement of SRM detection depends on a range of experimental parameters, including the affinity of a particular antibody (or aptamer) to a given target protein. The anti-SMAD4 aptamer characterized by the Kd
value of 45 nM provides the LOD of 12 ng/mL, which is in general comparable with LODs for the antibody-assisted SRM detection [43
]. It is thought that the use of more potent aptamers might further decrease the LOD of the aptamer-assisted SRM detection.
In summary, we demonstrated that aptamers immobilized on a solid carrier, such as magnetic beads, can be successfully employed as affinity reagents for an enhancement of SRM-based detection of LAPs directly in blood plasma. For rSMAD4 and the anti-SMAD4 DNA-aptamer constructed based on the SBE sequence, the LOD is decreased by a factor of 30. The growing number of aptamers developed for human proteins of diagnostic interest [46
] and improvements in aptamer selection methodologies, including the use of modified nucleobases to increase aptamer affinity and selectivity [47
], make the aptamer-based affinity enrichment of target proteins for the purpose of mass spectrometric analysis a very promising alternative to that based on antibodies.