Identification and validation of novel candidate protein biomarkers for the detection of human gastric cancer

Identification and validation of novel candidate protein biomarkers for the detection of human gastric cancer

Biochimica et Biophysica Acta 1844 (2014) 1051–1058 Contents lists available at ScienceDirect Biochimica et Biophysica Acta journal homepage: www.el...

869KB Sizes 0 Downloads 55 Views

Biochimica et Biophysica Acta 1844 (2014) 1051–1058

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbapap

Identification and validation of novel candidate protein biomarkers for the detection of human gastric cancer☆ Julia M. Humphries a,2, Megan A.S. Penno a,1, Florian Weiland a, Manuela Klingler-Hoffmann a, Agnieszka Zuber b, Alex Boussioutas c,d, Matthias Ernst e, Peter Hoffmann a,⁎ a

Adelaide Proteomics Centre, School of Molecular Biomedical Science, The University of Adelaide, Adelaide, SA 5005, Australia Institute for Photonics & Advanced Sensing, The University of Adelaide, Adelaide, SA 5005, Australia Department of Medicine, Royal Melbourne Hospital, University of Melbourne, Parkville, VIC 3050, Australia d Cancer Genomics and Predictive Medicine, Peter MacCallum Cancer Centre, East Melbourne, VIC 8006, Australia e Walter and Eliza Hall Institute of Medical Research, Parkville, VIC 3052, Australia b c

a r t i c l e

i n f o

Article history: Received 9 May 2013 Received in revised form 13 January 2014 Accepted 30 January 2014 Available online 7 February 2014 Keywords: Biomarkers Intestinal-type gastric adenocarcinoma Serum proteomics Post-translational modifications Isoforms

a b s t r a c t The timely detection of gastric cancer will contribute significantly towards effective treatment and is aided by the availability and reliability of appropriate biomarkers. A combination of several biomarkers can improve the sensitivity and specificity of cancer detection and this work reports results from a panel of 4 proteins. By combining a validated preclinical mouse model with a proteomic approach we have recently discovered novel biomarkers for the detection of gastric cancer. Here, we investigate the specificity of four of those biomarkers (afamin, clusterin, VDBP and haptoglobin) for the detection of gastric cancer using two independent methods of validation: ELISA, and a non antibody based method: Multiple Reaction Monitoring with High Resolution Mass Spectrometry (MRM-HR). All four biomarkers reliably differentiated GC from benign patient serum, and also in a small cohort of 11 early stage cases. We also present a novel isoform specific biomarker alpha-1-antitrypsin (A1AT) that was identified using a mouse model for gastric cancer. This isoform is distinct in charge and mobility in a pH gradient and was validated using human samples by isoelectric focussing and Western-blot (IEF-WB). This article is part of a Special Issue entitled: Biomarkers: A Proteomic Challenge. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Gastric cancer (GC) is the fourth most common cancer in the world and the second leading cause of death due to cancer, reflecting the overall 5 year survival rate (5YSR) of 24% [1]. This poor outcome can be attributed to an extended asymptomatic period associated with this cancer, and difficulty in the detection of early stage gastric adenocarcinoma when treatment could improve long term survival of patients.

Abbreviations: VDBP, vitamin D binding protein; MRM, multiple reaction monitoring; PTM, post translational modification; A1AT, alpha-1-antitrypsin; ToF, time of flight; MS, mass spectrometry; DIGE, difference gel electrophoresis ☆ This article is part of a Special Issue entitled: Biomarkers: A Proteomic Challenge. ⁎ Corresponding author at: Adelaide Proteomics Centre, School of Molecular and Biomedical Science, via Gate 8 Victoria Drive, The University of Adelaide, North Terrace Campus, Adelaide, SA 5005, Australia. Tel.: +61 8 8303 5507; fax: +61 8 8303 4362. E-mail addresses: [email protected] (J.M. Humphries), [email protected] (M.A.S. Penno), fl[email protected] (F. Weiland), [email protected] (M. Klingler-Hoffmann), [email protected] (A. Zuber), [email protected] (P. Hoffmann). 1 Current address: Department of Paediatrics, The University of Adelaide, Women's and Children's Hospital, SA 5006, Australia. 2 Current address: SA Pathology, Surgical Pathology, Bone and Joint Research Laboratory, Adelaide, SA, 5000, Australia 1570-9639/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbapap.2014.01.018

Endoscopic investigations are invasive, and given the lack of early symptoms for GC, not generally conducted until an advanced stage where the tumour has significant invasion [2]. While routine screening by gastric photofluorography in Japan has led to an improvement in the 5YSR by over 50% [3], the costs involved cannot be justified in Western countries where the incidence is low, or in those countries with lower GDP. In addition, the sensitivity of barium X-ray reaches only 39% for early stage cancer [4], leading to this being combined with esophagogastroduodenoscopy (EGD), the combination of which is in turn limited by the number of highly experienced endoscopists and the variability of the luminous intensity and quality of endoscopic images depending on differences in endoscope diameter [5]. Unfortunately, EGD is known to be an unpleasant experience for patients invoking anxiety, pharyngeal discomfort, nausea, the gag-reflex and choking, and has been associated with adverse incidents such as cardiovascular responses [6–8]. The incidence of GC in Asia and South America is similar to that in Japan [9], and a cost effective and non-invasive test to identify individuals at risk of GC for further referral for endoscopic biopsy could contribute significantly to improving survival and reducing the health cost burden associated with this disease. Current non-invasive tests for GC include serum pepsinogen in combination with Helicobacter pylori infection testing, gastrin 17 [9], carcinoembryonic antigen (CEA), carbohydrate antigen

1052

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058

(CA) 19–9, CA72–4 [10], and alpha-fetoprotein (AFP) [11], however, all of the above tests provide relatively low sensitivity and specificity [12,13]. The advantages of multiple biomarkers over single biomarkers include: that protein biomarkers may be indicative of more than one disease and therefore not unique to a specific cancer e.g. Il6 is over expressed in oral, prostate, multiple myeloma and renal cell cancers [14]; single biomarkers often have inadequate predictive value e.g. about 70% for PSA [15]; a single biomarker for a given cancer may for a biochemical reason be poorly expressed in a particular patient but it is unlikely that an entire panel of protein biomarkers would fail to be expressed. It is acknowledged that at the early validation stage the use of multiple biomarkers is more expensive than a single biomarker. However, once clinical relevance is established, ELISA analysis can be integrated into routine clinical pathology platforms and multiplexed. Few studies have looked at a suite of protein biomarkers that are differentially regulated and can show congruence of late-stage and early-stage GC biomarkers [16]. This is possibly because serum biomarkers that are able to distinguish early-stage GC are difficult to identify and validate due to the lack of serum samples from corresponding patients, the inherent difficulties of human genetic diversity and the relatively low abundance of potential serum biomarkers in comparison with the complex serum proteome [17]. Given the incidence of GC in densely populated low-income countries, a diagnostic test based on a suite of differentially regulated serum biomarkers providing good sensitivity and specificity, and that can detect early-stage GC would have international application. Recently we reported the differential regulation of serum biomarkers in tumour-bearing versus tumour-free cohorts in a highly reproducible, preclinical validated mouse model for early stage GC [18]. We showed that the three candidate biomarker proteins afamin, clusterin and haptoglobin were individually superior to a current clinical marker CA72-4 in discriminating GC patients from healthy controls. Prior to this, and to the best of our knowledge, changes in serum levels of afamin and clusterin had not been implicated in the diagnosis of GC, thus potentially yielding novel gastric cancer biomarkers. Here we report differential regulation of 4 biomarkers, with the addition of vitamin D binding protein (VDBP) to those previously reported, in serum of an extended cohort of GC patients, including a small number (n = 11) of early gastric cancer patients. The biomarkers were validated using ELISA and MRM mass spectrometry. Isoform specific biomarkers have been described in a number of cancers including epithelial ovarian cancer [19], and their relevance in the detection of cancer has been recently reviewed [20]. The difference in physico-chemical properties of the different isoforms has been mostly attributed to posttranslational modifications such as glycosylation. Although the identification of the modification is not absolutely necessary, it might provide important insight into the regulation of the protein of interest. The reliable quantification of protein isoforms is technically challenging and therefore limits their applicability in a clinical or diagnostic setting. However, we were able to confirm the specificity of an isoform specific biomarker using human serum by IEF-WB. Future advances in technologies are of critical importance to allow the use of this novel set of biomarkers. The need for serum biomarkers is two-fold: firstly to reduce patient impact of screening techniques and the costs involved with such invasive measures; and secondly to enable highly specific detection of early stage GC such that treatments are highly effective thus improving %YSR's. The well-being of patients is of utmost concern during detection and treatment of gastric and other forms of cancer at the late and early stages, and the high incidence of late stage versus early stage GC detection further highlights the need for a non-invasive, and thus low patient impact and sensitive screening technique.

2. Materials and methods 2.1. Patient samples Samples were collected with approval from the Ethics Committees of the Peter MacCallum Cancer Centre (Melbourne, Australia), National University Hospital (Singapore) and the University of Adelaide (Adelaide, Australia). Serum samples were obtained from 37 (female n = 10, 69 ± 10 years; male n = 17, 66 ± 11 years) preoperative GC patients with intestinal type gastric adenocarcinoma (according to Lauren classification). Eleven of these were classified as early-stage gastric cancer (AJCC 6th Edition Stage I disease, with minimal depth of invasion into mucosa and no metastatic lymph nodes) (female n = 6, 67 ± 13 yrs; male n = 5, 60 ± 13 yrs). The benign gastrointestinal disease serum samples (female n = 16, 54 ± 3 yrs; male n = 4, 55 ± 5 yrs) were in the majority female, an acknowledged gap in the cohort. Demographic and clinical pathological details of patients from which serum for the ELISA analysis was obtained are shown in Tables 1A and 1B. The clinical and demographic information on the 10 serum samples used in determination of isoforms is shown in Table 2.

2.2. Enzyme-linked immunosorbent assays (ELISA) ELISAs were performed in accordance with the manufacturer's recommendations. ELISA kits were obtained for: afamin (E92284Hu), clusterin (E91180Hu), haptoglobin (E90817Hu) and VDBP (E91810Hu) from USCN Life Science Inc., China. Serum protein concentrations were interpolated from kit-specific standard curves generated in GraphPad Prism (GraphPad Software). Unpaired Student's t-test (significance threshold of p b 0.05) and receiver operating characteristics (ROC) were performed in GraphPad Prism.

2.3. Multiple reaction monitoring high resolution mass spectrometry (MRM-HR) Ten GC and 10 control serum samples were prepared for MRM analysis using QProteome spin columns, followed by an in-solution tryptic digestion for downstream MRM analysis following protocols provided by the Australian Proteome Analysis Facility (APAF) (for QProteome depletion) followed by digestion with trypsin (20 mg) at 37 °C for 6 h (enzyme/substrate ratio approximately 1:25) [21,22]. Samples were run on Triple TOF 5600 (AB Sciex) mass spectrometer with an Eksigent Ultra nano-LC system (Eksigent) with a SGE ProteCol C18, 300 Å, 3 μm, 150 μm × 10 cm column. Prior to analysis, samples were reconstituted in 100 μL of 2% acetonitrile, 0.1% formic acid and vortexed. Samples were then sonicated in a water-bath for 10 min and centrifuged at 12,000 g for 5 min prior to nano-LC for MRM-HR analysis. Samples were diluted 1:1 with 2% acetonitrile, 0.1% formic acid then injected (10 μL, full loop injection) onto a peptide trap (Michrom peptide Captrap) for pre-concentration and desalted with 0.1% formic acid, 2% acetonitrile at 5 μL/min for 10 min. The peptide trap was then switched into line with the analytical column. Peptides were eluted from the column using a linear solvent gradient from mobile phases A:B (98:2) to mobile phases A:B (60:40) at 600 nL/min over a 150 min period where mobile phase A was 0.1% formic acid and mobile phase B was 90% acetonitrile and 0.1% formic acid. The reverse phase nano-LC eluent was subject to positive ion nanoflow electrospray analysis. A TOF-MS scan was acquired (m/z 350–1500, 0.25 s) followed by product ion scans for 23 pre-selected precursor ions from 9 targeted proteins. The product ion scans were 100 ms in the mass range m/z 100–1500 with the total cycle time of 2.6 s. MRM-HR data were acquired twice for each sample with a blank run between samples.

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058 Table 1A Demographic and clinical pathological information on the patients from which the gastric cancer serum samples used for the ELISA analysis were obtained. Gender

Age

AJCC staging (6th ed.)

H. pylori infection

Source

Male Male Female Female Female Female Male Female Male Male Female Female Female Female Male Male Male Male Male Male Male Male Male Male Male Male Male Male Male Female Male Male Male Male Male Male Male

46 75 76 81 53 64 66 67 56 64 68 69 76 76 53 44 73 67 74 73 71 48 60 70 60 71 73 79 70 64 55 72 77 71 87 61 78

1b 1b 1b 1b 1a 1a 2 2 4 4 4 4 4 1a 1b 1b 1b 2 2 2 3a 3a 3a 3a 3a 3a 3a 3a 3a 3a 3b 3b 3b 3b 3b 4 4

Negative Positive Negative Negative Negative Negative Negative Negative Negative Positive Positive Negative Negative Positive Negative Negative Negative Positive Positive Positive Positive Positive Positive Positive Negative Negative Positive Negative Positive Negative Positive Positive Positive Positive Negative Positive Negative

SNG SNG SNG SNG SNG SNG SNG SNG SNG SNG SNG SNG SNG WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI WEHI

Table 1B Demographic and clinical pathological information on the patients from which the benign gastrointestinal disease serum samples used for the ELISA analysis were obtained. Gender Age Chronic gastritis H. pylori infection Intestinal metaplasia Atrophy Female Female Female Female Female Female Female Female Female Female Female Female Female Female Female Female Male Male Male Male

50 51 52 52 52 52 53 54 54 54 55 56 56 56 61 62 51 52 55 63

Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes

No Yes No No No No No No No No No No No No No No Yes Yes No No

No No No No No No No No No No No No No No No No No No No No

No No No No No No No No No No No No No No No No No Yes No No

1053

gastric cancer. Three different forms of A1AT corresponding to the products of three different genes were identified in the DIGE experiment involving murine serum: A1A1T1 (accession number P07758), A1AT2 (accession number P22599) and A1AT4 (accession number Q00897). Peptides unique to each murine isoform were identified by LC–MS/MS (not shown). A1AT is encoded by a single gene in humans (protein accession number P01009). The sequence similarity between human A1AT and the mouse proteins are 64% for both A1AT1 and A1AT2, and 63% for ATA14. 2.5. Serum preparation for isoform study 2.5.1. Multiple affinity removal system (MARS) The seven most abundant serum proteins were depleted using the MARS Human-7 system (Agilent Technologies), according to the manufacturer's protocol. Prior to depletion, serum samples were diluted 1:4 with MARS buffer A (Agilent) containing a protease inhibitor cocktail (SIGMA) at a concentration of 0.1 mg/mL, and filtered using spin columns (Costar Spin-X centrifuge tubes 0.22 μm cellulose acetate, Corning). The collected low abundance protein fraction of each sample was precipitated as published previously [23] and resuspended in TUC4% (7 M urea (Merck), 2 M thiourea (GE Healthcare), 4% 3-[(3-Cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS) (Roche), 30 mM 2-Amino-2-hydroxymethyl-propane-1,3-diol (Tris) (Merck). Protein concentration was estimated using the EZQ protein assay (Invitrogen), as per the manufacturer's instructions. 2.6. Isoelectric focussing IEF was performed using pH 3–7 IEF gels (Invitrogen) according to the manufacturer's protocol. Each well of the IEF gel was loaded with 6.5 μl human serum, which was diluted 1:15 in H2O beforehand. 2.7. SDS-PAGE and Western-blotting IEF gels were electro-blotted onto a nitrocellulose membrane (Pall) using a Criterion Blotter (Bio-Rad) in combination with a 25 mM Tris base (Merck), 25 mM glycine (GE Healthcare), 20% methanol (Merck) and 0.1% SDS (GE Healthcare), pH 9.2 buffer system. Transfer was carried out for 60 min at a maximum of 100 mA per gel, 60 V and 50 W at room-temperature. 2.8. Western blot detection Membranes were blocked in 5% skim milk in Tris buffered saline (TBS) (50 mM Tris, 150 mM NaCl, pH 7.6) overnight at 4 °C, washed three times in TBS-T (TBS, 0.1% Tween-20 (Sigma)) and incubated with the primary antibody against the target protein overnight at 4 °C. The primary antibody, diluted in 5% skim milk in TBS-T, used in this investigation was chicken anti-human α-A1AT (ab#21437, Abcam) at a dilution of 1:200. Following overnight incubation in the primary antibody, membranes were washed three times with TBS-T, then incubated for 1 h at room temperature with Alexa-fluor 555 goat-anti-chicken IgG (#A21437, Invitrogen), at a dilution of 1:1000 in the dark at room temperature. Membranes were washed a further three times with TBS-T then scanned using a Typhoon (GE Healthcare). 3. Results 3.1. Candidate gastric cancer biomarker validation by ELISA

2.4. Mouse models and experimental design for DIGE For a detailed description of the mouse models and experimental designs please see Penno et al., 2012 [18]. In brief, we used a genetically modified mouse model (gp130F/F) that develops a gastric adenoma phenotype which resembles primary lesions of human intestinal-type

We performed ELISAs for afamin, clusterin, haptoglobin and VDBP to quantitatively assess the abundance of these proteins in patient-derived serum samples in a clinically applicable manner. Of the 18 serum samples from patients with gastric cancer, 11 were classified by AJCC criteria 6th edition as Stage I early-stage gastric cancer with minimal depth of

1054

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058

Table 2 Demographic and clinical pathological information on patients from which the case and control samples used in the isoform study came.

Age (yr) Gender, n AJCC stagea, n a

Mean ± SD (range) Male Female 1b 2

Gastric cancer (n = 5)

Healthy control (n = 5)

66 ± 11 (54–79) 3 2 3 2

67 ± 10 (51–75) 3 2

Based on the American Joint Committee on Cancer 6th edition staging system.

invasion into mucosa and no metastatic lymph node involvement. All other gastric cancer patients displayed gastric cancer as Stages II–III. Compared to sera that were closely age and gender matched from patients with benign gastrointestinal disease without intestinal metaplasia, we found down regulation of clusterin (1.7 folds, p b 0.001, Fig. 1A), and VDBP (1.7 folds, p = 0.058, Fig. 1B), conversely upregulation of haptoglobin (1.7 folds, p = 0.024) in serum from GC versus benign patients was noted (Fig. 1C), we also found down regulation of afamin (1.2 folds, p = 0.315) in sera of gastric cancer patients in comparison with serum from patients with benign GI disease (Fig. 1D). Surprisingly, the levels of these four serum proteins in the 11 early-stage GC samples were indistinguishable from those in the advanced stage GC samples (Fig. 1 A–D, open circles). Collectively, these findings suggest that 4 proteins could serve as promising biomarkers for the detection of early-stage GC. Using the above information, we investigated the individual diagnostic potential of afamin, clusterin, haptoglobin and VDBP as measured by ELISA using receiver operating characteristic (ROC) analysis. This is created by plotting the fraction of true positives out of the positives (TPR = true positive rate) vs. the fraction of false positives out of the negatives (FPR = false positive rate). TPR is also known as sensitivity, and FPR is one minus the specificity or true negative rate. This analysis showed that the sensitivity and specificity values represented as Area Under the Curve (AUC) were: 0.94 for clusterin, 0.84 for VDBP, 0.58 for haptoglobin, and 0.52 for afamin (Fig. 1E–H). 3.2. MRM-HR To confirm the data from the antibody based ELISA described in Fig. 1, we performed a second independent validation using a non antibody based method: multiple reaction monitoring with high resolution mass spectrometry (MRM-HR). The targeted proteins were investigated by using three proteotypic peptides for each protein of interest. Their amino acid sequences, m/z values of the precursor ions, retention times and m/z values of the productions used for quantification are summarised in Table 3. We used ten GC serum samples and 10 control serum samples, and MRM-HR data were acquired twice for each sample. The data were analysed (Table 4) and revealed a trend for differential regulation of afamin (1.5 fold, p = 0.143), clusterin (1.5 fold, p = 0.11) and VDBP (1.3 fold, p = 0.353). Although larger datasets may be required to validate differences between cohorts, we detected changes that were similar to those observed in ELISA (Fig. 1). As notable exception, we observed a much lower ratio of VDBP between GC samples and controls. Meanwhile, MRM-HR analysis revealed a significant higher abundance of haptoglobin in GC (1.3 folds, p = 0.048), which was similar to that observed in ELISA (Fig. 1). It is possible that depletion had a negative effect on the quantification and that different amounts of quantifiable protein were removed during the depletion and influenced the accuracy of the measurement. In summary we have validated novel biomarkers for GC by MRM-HR in the serum of ten GC patients compared to 10 healthy individuals. Moreover, all four biomarkers showed similar regulation as shown by the ELISA measurements but with different outcomes in statistical significance.

3.3. Isoform specific regulation During our screen of genetically engineered mouse (GEM) models for potential novel serum biomarkers [18], we identified diseasespecific abundance of alpha-1-antitrypsin (A1AT) protein isoforms (Fig. 2). Specifically, among the 5 protein spots detected that corresponded to A1AT (Fig. 2A), we found two spots (spot 1021 and 1023) of less and three spots (spot 1143, 1003 and 1014) of higher abundance in GC samples. This data indicates that A1AT may undergo differential post-translational processing. A1AT is also expressed in humans and shares a 64% sequence identity with the murine protein. Box-and-whisker plots in Fig. 2B show the regulation of the A1AT isoforms across the various GEM strains for the two down-regulated isoforms (Fig. 2B upper panel) and the three up-regulated isoforms (Fig. 2B lower panel). These observations track well with wild-type and IL6KOmice being tumour free, and gp130F/F and gp130F/F; IL6KO mice bearing gastric tumours. Box-and-whisker plots in Fig. 2C depict a summary of the average spot volumes associated with the 5 different A1AT isoforms detected in sera from tumour-bearing relative to tumour-free cohorts for the 2 down regulated isoforms (Fig. 2C). 3.4. Confirmation of regulated protein isoforms In order to confirm isoform specific regulation of A1AT in humans, we first detected the total amount of A1AT in 10 human serum samples (5 control, 5 GC) using SDS-PAGE followed by Western-blot (Fig. 3A). Densitometry analysis revealed equal abundance of A1AT in both sera of both cohorts (Fig. 3B). To explore whether human sera would reveal similar differences observed in mice, we separated these samples on 1dimensional IEF gels and probed the corresponding Western-blot membrane with A1AT antibody. We could detect four distinct isoforms of A1AT, referred to as α, β, γ and δ according to their migration on the IEF gel (Fig. 3C). Using ImageQuant (GE Healthcare) each of the distinct protein bands was quantified and normalised against the total A1AT signal of the respective sample. The γ-isoform was significantly upregulated (1.4 folds, p = 0.02) in the GC samples when compared to the control samples (Fig. 3D). 4. Discussion 4.1. Ability of serum biomarkers to identify early stage gastric cancer Gastric cancer is most commonly detected at Stages III and IV, at which time prognosis is poor. Familial history (e.g. Lynch Syndrome and mismatch repair deficiencies) is known to play a large part in intestinal type GC with those at risk currently undergoing regular investigations. These investigations can take the form of endoscopy, or if in Japan gastric fluorography, both of which are invasive and expensive. Serum biomarkers, reporting protein profile changes as a reflection of pathological changes, would enable screening of at-risk individuals using a non-invasive and inexpensive test, thus reducing cost both in monetary and human terms. Given the poor outcome of GC identified at the late stages III and IV, it is essential, that to have a benefit, such a test would be able to detect GC at an early stage (Stage I).

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058

1055

Clusterin p< 0.05

A

E

Concentration (µg/ml)

Specificity % (FPR)

VDBP p< 0.05

F

VDBP

n.s.

Sensitivity % (TPR)

Concentration (µg/ml)

B

Clusterin

Sensitivity % (TPR)

n.s.

Specificity % (FPR)

Haptoglobin

Haptoglobin

n.s.

Sensitivity % (TPR)

G

p<0.05

Concentration (µg/ml)

C

Specificity % (FPR)

Afamin

Afamin

p< 0.05

H p< 0.05

Sensitivity % (TPR)

Concentration (µg/ml)

D

Specificity % (FPR) Fig. 1. Differential abundance of proteins in gastric cancer samples compared to control samples was confirmed by ELISA. Serum samples from intestinal gastric cancer patients, n = 37 (circles), containing 11 early stage gastric cancer samples (open circles) were compared to serum samples from benign gastrointestinal disease patients n = 20 (squares) and healthy controls n = 7 (triangles) using ELISA (A—clusterin, B—VDBP, C—haptoglobin, D—afamin) and the results were analysed using ROC curves (E—clusterin, F—VDBP, G—haptoglobin, H—afamin).

1056

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058

Table 3 Peptides used for MRM. Protein

Peptides

Precursor ions (m/z)

Afamin

FLVNLVK

416.7682

68.71

IAPQLSTEELVSLGEK

571.9805

82.79

SDVGFLPPFPTLDPEEK

944.4724

106.09

ASSIIDELFQDR [21,25]

697.3515

102.06

VTTVASHTSDSDVPSGVTEVVVK

772.064

57.99

THLPEVFLSK [25]

585.8282

60.1

HLSLLTTLSNR

627.8612

63.4

PAAQLPELPDVELPTNK

916.4921

88.41

DIAPTLTLYVGK

645.8688

80.34

TEGDGVYTLNNEK

720.3361

43.97

VTSIQDWVQK

602.322

62.12

Clusterin

VDBP

Haptoglobin

The use of a panel of protein biomarkers rather than a single protein in isolation to identify individuals requiring referral for endoscopy can further improve specificity and sensitivity of the testing regimen, resulting in fewer false positives/false negatives which adversely impact on patient outcome. The differential regulation of 4 serum proteins in gastric cancer patients compared with normal healthy individuals is shown here, with clusterin, VDBP and afamin down-regulated, and haptoglobin up-regulated in serum from GC versus benign GI disease cases (Fig. 1). Our observations in humans support those published previously in tumour-bearing mice [18], showing comparative direction of regulation and significance. Differences in the degree of regulation of VDBP between these results and those previously reported might be attributable to antibody affinity used in the ELISA kits from two different manufacturers. Of particular note is our finding that the abundance of these 4 serum proteins in early-stage GC can be superimposed on the corresponding findings in serum samples of advanced-stage GC, indicating the ability of these biomarkers to identify various pathologically relevant stages.

Table 4 Regulation of proteins as determined by MRM-HR. MRM-HR Protein

GC (intensity)

Control (intensity)

t-Test

GC/Control ratio

Afamin Clusterin Haptoglobin VDBP

2.87 9.02 71.89 5.25

4.2 13.1 57.89 6.95

0.143 0.110 0.048 0.353

−1.5 folds −1.5 folds +0.8 folds −1.3 folds

± ± ± ±

1.7 4.6 11.1 3.1

± ± ± ±

1.2 4.3 11.7 3.5

Retention time (min)

Product ions (Da) 685.4607 572.3766 473.3082 745.4454 632.3614 533.293 1269.6361 928.4622 1035.5106 922.4265 807.3995 1014.583 917.5302 830.4982 932.5451 819.4611 1117.6575 1004.5735 917.5415 1351.7104 1125.615 1012.531 991.5823 894.5295 793.4818 1209.5746 1152.5531 1037.5262 881.4363 718.373 1104.5684 1003.5207 916.4887 803.4046

4.2. MRM-HR MRM is a tandem MS scan mode unique to triple quadrupole MS instruments, and is capable of rapid sensitive and specific quantitation of analytes in highly complex matrices [24]. MRM follows the same principle but using a Q-TOF MS instrument and for that reason a high resolution of the fragment ions, on which the quantification is done on, can be achieved. Here the targeted proteins, corresponding peptides, retention times of the peptides and m/z values of the product ions used for quantification (as shown in Table 3) were used for determination of the ratio between GC and Control samples and to identify those proteins that were differentially regulated. The MRM-HR results confirmed the differential regulation shown by ELISA for afamin, clusterin, haptoglobin and VDBP.

4.3. Isoform specific regulation Isoforms, as a result of various post-translation modifications are ubiquitous, dynamic and are a true representation of the circulating expression of the proteome, relating information of disease status, amongst others. Thus, the isoform profile of a given individual can provide information on physiological changes that in this case might indicate a diversion from normal to gastric cancer. Here we report the isoform specific profile of human serum comparing GC and control patients (Fig. 3). Our data illustrate the potential of isoform specific biomarkers in the diagnostic of gastric cancer. Advances in technology development will make those biomarkers more accessible and their measurement more accurate.

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058

A

pI(NL)

4

1057

7

kDa 150 102

SDS-PAGE

76 1143 1003 1014

52

1023 1021

38 31 24

17 12

B

Spot numbers 1021, 1023

Spot numbers 1003, 1014, 1143

C

Spot numbers 1021, 1023

Spot numbers 1003, 1014, 1143

Fig. 2. Isoform specific regulation of A1AT in gastric cancer mouse model. A. Representative DIGE-image indicating isoform-specific biomarker candidates. Mouse serum was analysed by DIGE and protein spots of interest were identified using mass-spectrometry. Red indicates less acidic forms which were down-regulated in GC, and green indicates more acidic forms which were up-regulated in GC. B. Spot volumes of those indicated in Fig. 2A grouped according to genetically modified mouse genotype. Box-and-whisker plots representing the average spot volume associated with A1AT in mouse serum showing down-regulation (top panel, spots 1021 and 1023) and up-regulation (bottom panel, spot numbers 1003, 1014 and 1143) in tumour-bearing relative to the tumour-free cohorts. Note that the Stat3 mice represent an intermediate step between the tumour-bearing and tumour-free cohorts. Box = 1st quartile/ mean/3rd quartile; whiskers = 5/95 percentiles. C. Spot volumes of those indicated in Fig. 2A grouped according to classification as control or gastric cancer affected. Box-and-whisker plots representing the average spot volume associated with A1AT in mouse serum showing down-regulation (top panel, spots 1021 and 1023) and up-regulation (bottom panel, spot numbers 1003, 1014 and 1143) in tumour-bearing relative to the tumour-free cohorts. Box = 1st quartile/mean/3rd quartile; whiskers = 5/95 percentiles. Wild-type and IL6 which report neither spontaneous tumour formation nor systemic inflammation, and are grouped as Control, while FF and FF IL6 report a combination of these and are grouped as GC. Note that the Stat3 mice represent an intermediate step between the tumour-bearing and tumour-free cohorts and so are excluded from the data in this graph.

5. Conclusion In conclusion, the role of early diagnostics in the treatment of gastric cancer, as for all cancers, cannot be underestimated, thus the value in identifying an easily accessible serum biomarker. Here the ability of the circulating proteome profile to reflect the current physiological status of an individual was used to identify a suite of 4 proteins that were differentially regulated in GC versus Control cohorts. Given the

silent progression of the early and treatable stages of GC to late stage and poor prognosis, the clustering of a small cohort of early stage GC with that of late stage GC is of particular note. Obviously further statistical power needs to be attributed to the small number of early stage GC data presented here, however this work represents an exciting progression in the field of serum biomarkers for GC. We also show data for isoform specific regulation in GC versus Control and this presents an additional dimension to the identified biomarkers.

1058

J.M. Humphries et al. / Biochimica et Biophysica Acta 1844 (2014) 1051–1058

A 1

2 3

C

GC

Control 4 5

6 7

1

8 9 10

SDS-PAGE WB: anti-A1AT

2

3

4

5

6

7

8 9 10

IEF gel WB: anti-A1AT

B Arbitrary unit

GC

Control

D

*

2.0 1.0 0.0

Control

GC

Fig. 3. A1AT isoform γ is upregulated in gastric cancer serum. A. Five control serum samples and 5 gastric cancer samples were separated by SDS-PAGE. Western-blot membrane was probed with anti-A1AT antibody. B. The abundance of A1AT was quantified by densitometry. C. Five control serum samples and 5 gastric cancer samples were separated by IEF. After Western-blot the membrane was probed with anti-A1AT antibody. D. The abundance of A1AT isoforms α, β, γ and δ was quantified by densitometry. Control 2 showed one extra band between γ and δ which was not included in the analysis.

Improvement in technology to separate and detect such biomarkers in a high throughput fashion will be necessary to use them to their full potential. Acknowledgements The MRM-HR analysis was undertaken at APAF the infrastructure provided by the Australian Government through the National Collaborative Research Infrastructure Strategy (NCRIS). Matthias Ernst is a senior research fellow of the NHMRC. References [1] H. Brenner, D. Rothenbacher, V. Arndt, Epidemiology of stomach cancer, Methods Mol. Biol. 472 (2009) 467–477. [2] J.C. Layke, P.P. Lopez, Gastric cancer: diagnosis and treatment options, Am. Fam. Physician 69 (5) (2004) 1133–1140. [3] C. Hamashima, et al., The Japanese guidelines for gastric cancer screening, Jpn. J. Clin. Oncol. 38 (4) (2008) 259–267. [4] M. Nishizawa, Present status and prospect for cancer screening, J. Gastroenterol. Mass Survey 78 (1993) 100–103. [5] H. Nakata, et al., Transnasal and standard transoral endoscopies in the screening of gastric mucosal neoplasias, World J. Gastrointest. Endosc. 3 (8) (2011) 162–170. [6] L. Brandt, Patients' attitudes and apprehensions about endoscopy: how to calm troubled waters, Am. J. Gastroenterol. 96 (2001) 280–284. [7] R. Hart, M. Classen, Complications of diagnostic gastrointestinal endoscopy, Endoscopy 22 (1990) 229–233. [8] E. Bough, S. Meyers, Cardiovascular responses to upper gastrointestinal endoscopy, Am. J. Gastroenterol. 69 (1978) 655–661. [9] W.K. Leung, et al., Asia Pacific Working Group on Gastric Cancer. Screening for gastric cancer in Asia: current evidence and practice, Lancet Oncol. 9 (3) (2008) 279–287.

[10] D. Marrelli, et al., Prognostic significance of CEA, CA 19–9 and CA 72–4 preoperative serum levels in gastric carcinoma, Oncology 57 (1) (1999) 55–62. [11] K. Lam, S. Lo, Discovery of diagnostic serum biomarkers of gastric cancer using proteomics, Proteomics Clin. Appl. 2 (2) (2008) 219–228. [12] J. Norton, Surgery: Basic Science and Clinical Evidence, Springer, 2001. [13] M.P. Ebert, C. Rocken, Molecular screening of gastric cancer by proteome analysis, Eur. J. Gastroenterol. Hepatol. 18 (2006) 847–853. [14] F. Riedel, et al., Serum levels of interleukin 6 in patients with primary head and neck squamous cell carcinoma, Anticancer Res. 25 (2005) 2761–2766. [15] J.F. Rusling, et al., Measurement of biomarker proteins for point-of-care early detection and monitoring of cancer, Analyst 135 (2010) 2496–2511. [16] W. Liu, et al., Down-regulated expression of complement factor I: a potential suppressive protein for gastric cancer identified by serum proteome analysis, Clin. Chim. Acta 377 (1–2) (2007) 119–126. [17] S. Surinova, et al., On the development of plasma protein biomarkers, J. Proteome Res. 10 (2011) 5–16. [18] M. Penno, et al., 2D-DIGE analysis of sera from transgenic mouse models reveals novel candidate protein biomarkers for human gastric cancer, J. Proteomics 77 (2012) 40–58. [19] N. Ahmed, et al., Proteomic tracking of serum protein isoforms as screening biomarkers of ovarian cancer, Proteomics 5 (17) (2005) 4625–4636. [20] Z. Zhang, et al., Isoform-level expression profiles provide better cancer signatures than gene-level expression profiles, Genome Med. 5 (4) (2013) 33. [21] M. McKay, et al., The development of multiple reaction monitoring assays for liver-derived plasma proteins, Proteomics Clin. Appl. 1 (12) (2007) 1570–1581. [22] S. Randall, M. McKay, M. Molloy, Evaluation of blood collection tubes using selected reaction monitoring MS: implications for proteomic biomarker studies, Proteomics 10 (2010) 2050–2056. [23] M. Penno, M. Ernst, P. Hoffmann, Comparative 2D DIGE analysis of the depleted serum proteome for biomarker discovery, Methods Mol. Biol. 854 (2012) 207. [24] R. Kondrat, G. McClusky, R. Cooks, Multiple reaction monitoring in mass spectrometry/mass spectrometry for direct analysis of complex mixtures, Anal. Chem. 50 (1978) 2017–2021. [25] L. Anderson, C.L. Hunter, Quantitative mass spectrometric multiple reaction monitoring assays for major plasma proteins, Mol. Cell. Proteomics 5 (4) (2006) 573–588.