In this study we discovered serum BTD as a potential breast cancer biomarker through the biomarker development pipeline encompassing mass spectrometry based screening and independent downstream immunoblot assays. Biomarker candidates discovered by ICAT analysis of plasmas from 6 breast cancer patients and 6 age-matched normal healthy controls were examined by Western blot in the same sample set. The two candidates, BTD and GPX3, confirmed by this approach were next tested with immunoblot assay in a blinded set of breast cancer and control to ascertain the markers ability to differentiate the two groups.
The ICAT method applied here for the screening of differentially expressed proteins has low-throughput and is not suitable for a large number of samples. Therefore, a sample pooling strategy was employed to overcome this drawback. Although pooling reduces the expense of costly assays, nevertheless their still remains a possibility to obtain biased quantification result stemming from individual variations, which necessitates an independent downstream assay in individual samples. In the immunoblot assays after the ICAT discovery phase, we had to drop some candidate markers like GPX3. Our results suggest that the sample pooling strategy has both advantages and disadvantages. We adapted immunodepletion of top six high-abundance proteins to dig deep into low abundance proteins since plasma proteins are present over a wide dynamic range in concentration. This antibody-based separation system has demonstrated high efficiency to remove the specifically targeted proteins as well as both reproducibility and selectivity [16–19]. Actually, it was effective in our study enough to detect protein that exist at about a few μg/ml in plasma such as L-selectin .
Differential profiling by ICAT method enabled us to identify and quantify a total of 155 plasma proteins. The number was much smaller than that of proteins identified by the same method with tissue extract proteome , which was accounted for by the small number of the proteins identified based on single peptide match. On the contrary, spectral counts for each peptide count were distributed more widely: for example, spectral counts ranged from 2 to 23 for the proteins identified with two unique peptides. This is clearly due to the concentration depth of plasma proteome. Our result indeed covered 4 orders of magnitude from the lowest L-selectin at 0.67 μg/ml to the highest α-2-macroglobulin at 1.8 mg/ml . The lower detection limit was similar to another study performed on plasma proteome , which indicated that ICAT analysis for immunodepleted plasma samples is effective for biomarker discovery medium-abundance protein. Comparing current plasma proteome list with those of our previous studies [11, 12] and a recent report related to multiplex serum biomarker , several proteins, such as transthyretin, vitamin D-binding protein, and endorepellin were commonly identified. However, in the current study, these proteins showed little change among breast cancer patients and normal healthy controls (Additional file 1), and hence were excluded from further analysis. This is due to the difference in sample sources and in the screening method. 2DE or SELDI-TOF was used in the previous studies while we adapted ICAT strategy in this study. As mentioned earlier, protein quantification by ICAT is based only on cysteine containing peptide whereas 2DE reflects whole protein features including proteolytic processing. In addition, the previous study also could not draw decisive conclusion about discriminatory power of transthyretin and vitamin D-binding protein . In case of endorepellin, we discovered LG3 fragment of the protein as a biomarker in the previous study . In contrast, the single ICAT peptide matched to endorepellin covered other part than the LG3 fragment in the current study. Moreover, common proteins such as heptoglobin, serum albumin and transferrin were detected not due to their differential regulation in breast cancer but due to incomplete depletion of these proteins from plasma during immunoaffinity chromatography (in this study) or incomplete removal of blood vessels in the preparation of breast tissues (in the previous study ). Therefore, such proteins were not put into the next qualification phase of this study, while the proteins chosen for qualification were seemingly novel proteins discovered in relation to breast cancer.
Among the proteins showing differential level between breast cancer patients and normal healthy women, we chose five proteins for further verification. In this experimental flow, 4 biomarker candidates including ORM2, CD14, BTD and GPX3 showed relatively similar quantification results both in ICAT and Western blot. In case of CHL1, we observed an inconsistency between two quantification results (Table 2). According to ICAT result, CHL1 was quantified based on a single peptide (LHMLELHCESK; Additional file 2) in which the methionine residue had formerly been oxidized. Methionine oxidation occurs frequently during sample preparation and handling. Therefore, it seems inappropriate to quantify proteins based on a single oxidized peptide. Increased level of ORM2  and CD14  or down regulation of GPX3  appears to be reasonable when interpreted in relation to their biological functions. However, we could not confirm their ICAT fold changes in the subsequent verification steps. It is likely that expression levels of ORM2, CD14 and GPX3 vary with age or with other factors that are currently unknown. Other possibility like stochastic variation cannot be excluded. Interestingly, the mRNA level of BTD in breast cancer tissue also changed compared to normal breast tissues. There is no evidence and it is unlikely that expression change of BTD in breast cancer tissues affect their levels in plasma. Therefore, the abundance change of BTD observed in the plasmas of breast cancer patients need not necessarily be the same as the abundance change of tissue mRNA. Nevertheless, it is worthy investing BTD for their potential use as tissue biomarkers in future study.
Among five candidates, BTD subsisted until the last step of verification of the biomarker pipeline. BTD is known to catalyze the release and recycling of endogenous biotin . It is known to be secreted into blood circulation , and is highly active in the serum, liver, kidney and adrenal glands. Its deficiency results in various diseases such as seizures, hypotonia, hyperammonemia, and so forth . We first reported here that its level was consistently down-regulated in breast cancer plasma. It is, yet, difficult to explain how BTD is down-regulated in breast cancer plasma. Interestingly, transcriptional levels of BTD in breast cancer tissues were also down-regulated. So, it merits further studies to elucidate down-regulation of BTD in relation to the behavior of breast cancer cells. Despite many uncertainties in the cellular and molecular mechanism of BTD, it is apparent from our results that BTD is down-regulated in breast cancer plasma. It will be useful and applicable for clinical use alone or in combination with other biomarkers in detecting breast cancer by less invasive techniques using plasma samples.