Proteomics‐based identification of α‐enolase as a tumor antigen in non‐small lung cancer

Abstract

Autoantibodies against tumor antigens represent one type of biomarker that may be assayed in serum for detection of cancer and monitoring of disease progression. In the present study, we used a proteomics‐based approach to identify novel tumor antigens in non‐small cell lung cancer (NSCLC). By combining two‐dimensional electrophoresis, western blotting, mass spectrometry and enzyme‐linked immunosorbent assay technology, we detected autoantibodies against α‐enolase in a subset of NSCLC patients’ sera. When ‘Mean OD_{healthy control sera} + 3 SD_{healthy control sera}’ was used as the cut‐off point, the prevalence of this autoantibody was 27.7% in patients with NSCLC (26 of 94), 1.7% in healthy control subjects (1 of 60), and not detectable in sera from 15 patients with small cell lung cancer, 18 patients with gastrointestinal cancer and nine patients with Mycobacterium avium complex infection of lung. Immunohistochemical staining showed that expression of α‐enolase was increased in cancer tissues of NSCLC patients, and flow cytometric analysis confirmed the expression of α‐enolase at the surface of cancer cells. The combined detection of autoantibodies against α‐enolase, carcinoembryonic antigen and cytokeratin 19 fragment (CYFRA21‐1) enhanced sensitivity for the diagnosis of NSCLC. Therefore, autoantibodies against α‐enolase may constitute a promising biomarker for NSCLC. (Cancer Sci 2007; 98: 1234–1240)

Abbreviations: 1‐DE

one‐dimensional electrophoresis

2‐DE

two‐dimensional electrophoresis

a.a.

amino acids

CA

cancer antigen

CBB

Coomassie brilliant blue

CEA

carcinoembryonic antigen

CYFRA21‐1

cytokeratin 19 fragment

ELISA

enzyme‐linked immunosorbent assay

HRP

horseradish peroxidase

IEF

isoelectric focusing

IHC

immunohistochemical

LC

liquid chromotography

MAC

Mycobacterium avium complex

MALDI‐TOF

matrix‐assisted laser desorption–ionization time‐of‐flight

MS

mass spectrometry

NSCLC

non‐small cell lung cancer

NSE

neuronspecific enolase

OD

optical density

PCR

polymerase chain reaction

PMF

peptide mass fingerprinting

PVDF

polyvinylidene difluoride

SCC

squamous cell carcinoma

SCLC

small cell lung cancer; TNM, tumour–node–metastasis.

Lung cancer is the leading cause of cancer death,^{( 1 )} and NSCLC accounts for nearly 80% of lung cancer cases. There is an urgent need for a better understanding of the biological mechanisms of NSCLC as well as the identification of reliable biomarkers for its diagnosis and prognosis. To date, a number of NSCLC markers have been evaluated, including CEA, CYFRA 21‐1, SCC antigen, CA125 and NSE.^{( 2} , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ^{8 )} Autoantibodies against several tumor antigens such as L‐myc and c‐myc, p53 and antineural/antinuclear antigens have also been investigated.^{( 9} , ¹⁰ , ¹¹ , ^{12 )} Recently, autoantibodies against PGP9.5, peroxiredoxin‐I, annexin‐I and annexin‐II were identified in the sera of lung cancer patients using a proteomic approach.^{( 13} , ¹⁴ , ^{15 )} However, the sensitivity and specificity of these biomarkers are not yet satisfactory and there are currently no data to support any particular method for screening for lung cancer.^{( 16 )}

Autoantibodies against tumor antigens represent one type of biomarker that may be assayed in serum for detection of cancer and monitoring of disease progression. In spite of the fact that the quantity of any tumor antigen in cancer cells or in the circulation is usually very small, especially in the early stages of cancer, the body's immune response to such antigens represents a remarkable phenomenon of biological amplification of these weak signals from tumor antigens.^{( 17 )} The identification of panels of tumor antigens that elicit an immune response may thus be useful for detecting potential specific biomarkers as well as for the initiation of immunotherapy against NSCLC. The aim of the present study was to identify novel candidate tumor antigens in NSCLC by means of a proteomics‐based approach. One of these antigens was identified as α‐enolase, and its immunogenicity was confirmed by western blotting using recombinant protein. The results obtained with enzyme‐linked immunosorbent assay (ELISA) demonstrated that when ‘Mean OD_{healthy control sera} + 3 SD_{healthy control sera}’ was used as the cut‐off point, a humoral immune response directed against α‐enolase occurred in 27.7% of NSCLC patients, but in only 1.7% of healthy control subjects. Immunohistochemical staining showed that α‐enolase was overexpressed in cancer tissues of NSCLC patients. The combined detection of autoantibodies against α‐enolase, CEA and CYFRA 21‐1 enhanced sensitivity for NSCLC diagnosis. Therefore, autoantibodies against α‐enolase may constitute a promising biomarker for NSCLC.

Materials and Methods

Subjects.  Sera and tumor tissue were obtained at the time of diagnosis after informed consent had been given by the subjects. The experimental protocol was approved by the ethics committee of Osaka University. Sera from 94 patients with NSCLC, 15 patients with SCLC, 18 patients with gastrointestinal cancer (10 patients with gastric cancer, 8 patients with colon cancer) and nine patients with MAC were analyzed. In terms of TNM stages, the NSCLC patients comprised 17 cases of stage I, 14 cases of stage II, 34 cases of stage III and 29 cases of stage IV. The histological distribution of NSCLC was 73 adenocarcinoma cases and 21 SCC cases. Clinical data for the serum tumor marker CEA and CYFRA 21‐1 were also collected for investigation. Sera from 60 asymptomatic healthy subjects, whose average age and sex were comparable to those of the NSCLC patient group, were used as controls.

1‐DE and 2‐DE.  Proteins were extracted from NSCLC tumor tissues using the Complete Mammalian Proteome Extraction Kit (Calbiochem, Darmstadt, Germany). For 1‐DE, extracted proteins were resolved by using 10% Bis‐Tris Criterion XT Precast gels (Bio‐Rad Laboratories, Hercules, CA, USA), transferred to PVDF membranes or stained with CBB. For 2‐DE, IEF was carried out using the PROTEAN IEF cell (Bio‐Rad Laboratories) according to the manufacturer's instructions. Extracted proteins were reconstituted in a rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 2 mM tributyl phosphine (TBP), 0.0002% bromophenol blue (BPB), 0.2% bio‐lyte ampholyte 3‐10) and applied to ReadyStrip IPG strips (11 cm, pH 3–10). IEF was run for 45 000 Vh, and 2‐DE was carried out using 10% Bis‐Tris Criterion XT Precast gels. The gels were then stained with the Silver Stain MS Kit (Wako Pure Chemical Industries, Osaka, Japan) or used for protein transfer to PVDF membranes.

Western blotting.  After blocking with 5% skim milk, the PVDF membranes were incubated with serum at a 1:100 dilution or rabbit anti‐enolase antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at a 1:1500 dilution. The membranes were then incubated with sheep anti‐human IgG or donkey anti‐rabbit IgG (Amersham Biosciences UK, Buckinghamshire, UK). Membranes incubated with sheep anti‐human IgG only were used as negative controls. Finally, the signals were visualized with an enhanced chemiluminescence reaction system (Perkin Elmer Life Sciences, Boston, MA, USA).

Identification of protein bands or spots.  Protein bands on gels stained with CBB or protein spots on gels stained with silver, which corresponded to positive bands or spots on western blot membranes, were excised from the gel and digested with trypsin (Promega, Madison, WI, USA) according to published procedures.^{( 18 )} For protein bands, the LC‐MS/MS analysis was carried out using an LCQ ion trap mass spectrometer (ThermoElectron, San Jose, CA, USA) coupled on‐line with Magic 2002 capillary high‐performance liquid chromatography (Michrom BioResources, Auburn, CA, USA). For protein spots, all PMF spectra were obtained by using an ultraflex TOF/TOF MALDI‐TOF mass spectrometer (Bruker Daltonics, Bremen, Germany). MS/MS or PMF data were then searched with Mascot software (Matrix Science, London, UK) against the NCBInr or swiss‐prot databases. Protein database searching was carried out with following parameters for PMF: Homo sapiens, maximum of one missed, cleavage by trypsin, monoisotopic mass value, charge state of 1+, allowing a mass tolerance of 100 p.p.m., and carbamidomethyl modification of cysteine.

Preparation of recombinant protein.  To prepare recombinant proteins, the human full‐length α‐enolase complementary DNA (1–434 a.a.) was amplified by PCR from the Hep3B cell line cDNA library using the primers: sense 5′‐GTGGCTAGAAGTTCACCATG‐3′, antisense 5′‐TTACTTGGCCAAGGGGTTTC‐3′. To map the autoepitope on α‐enolase, three cDNA fragments that encode C‐terminal deletion mutant proteins (α‐Eno1, α‐Eno2, α‐Eno3) were similarly amplified. The nucleotide sequences of the primers for PCR were: α‐Eno1 (1–334 a.a.) sense 5′‐TGTCTATTCTCAAGATCCATGCC‐3′, antisense 5′‐TTACTCGTTCACGGCCTTGGC‐3′; α‐Eno2 (1–234 a.a.), sense 5′‐TGTCTATTCTCAAGATCCATGCC‐3′, antisense 5′‐TTAAGCTTTCCCAATAGCAGTC‐3′; and α‐Eno3 (1–134 a.a.) sense 5′‐TGTCTATTCTCAAGATCCATGCC‐3′, antisense 5′‐TTAGATGTGGCGGTACAGGGG‐3′. These cDNA fragments were then subcloned into the pET‐28a(+) vector (Novagen, Madison, WI, USA), resulting in expression of α‐enolase or its fragments with a 6 × His tag. Recombinant proteins were produced in Escherischia coli BL21‐CodonPlus (DE3)‐RIL cells (Stratagene, La Jolla, CA, USA) and purified by affinity chromatography using Ni‐NTA resin (QIAGEN, Tokyo, Japan). Recombinant human full‐length or C‐terminal deletion mutant α‐enolase, rabbit β‐enolase (Sigma, St Louis, MO, USA) and human γ‐enolase (Calbiochem) were subjected to sodium dodecylsulfate–polyacrylamide gel electrophoresis, using a 4–20% precast gel, then stained with CBB or transferred to PVDF membrane and probed with anti‐enolase antibody, anti‐6 × His monoclonal antibody or sera as described above.

Flow cytometry.  Human lung adenocarcinoma cell line A549 was maintained in RPMI‐1640 medium supplemented with 10% heat‐inactivated fetal bovine serum, 100 U/mL penicillin and 100 µg/mL streptomycin. Cells (10⁶) were incubated with rabbit anti‐enolase antibody at a 1:100 dilution and labeled with fluorescein isothiocyanate‐conjugated goat anti‐rabbit immunoglobulin (BD Biosciences, San Jose, CA, USA). Normal rabbit IgG was used as a control. Stained cells were analyzed using a FACS Canto cytometer (Becton‐Dickinson, Mountain View, CA, USA) and the results were analyzed using FlowJo software (Tree Star, Stanford, CA, USA).

ELISA.  To assess the potential of these autoantibodies as a diagnostic marker, their frequencies in the sera were determined by means of ELISA using recombinant human full‐length α‐enolase protein. The ELISA was carried out as published elsewhere, with modifications.^{( 19 )} Briefly, each well of a Microtiter plate (MaxiSorp; Nunc A/S, Roskilde, Denmark) was coated with 1 µg of recombinant human full‐length α‐enolase. After blocking with 1% bovine serum albumin, all wells were incubated with human serum at a 1:500 dilution at room temperature for 1 h. To reduce the background level originating from the non‐specific reactivity of sera with bacterial proteins, the sera were diluted and incubated with 100 µg/mL E. coli BL21‐CodonPlus (DE3)‐RIL cell lysate for 2 h at room temperature before incubation with coated recombinant human α‐enolase. The antigen–antibody complexes were detected with 1:5000‐diluted HRP‐conjugate sheep anti‐human IgG with TMB (Dako, Carpinteria, CA, USA) as the substrate. OD was read at 450 nm. The antibody titer was expressed by using arbitrary binding units calculated according to the formula:

binding units of sample = (OD_sample/[Mean OD_{healthy control sera} + 3 SD_{healthy control sera}]) × 100.

Based on this formula, 100 binding units was used as the cut‐off point.

IHC staining.  After deparaffinization, tissue sections were treated with 100% cold methanol containing freshly prepared 0.3% hydrogen peroxide for 30 min, blocked in 10% normal goat serum for 20 min and incubated with rabbit anti‐enolase antibody (Santa Cruz Biotechnology) at a 1:250 dilution overnight. Incubation of parallel sections omitting the first antibody was done to generate negative controls. Staining of sections was completed with a biotin‐conjugated secondary antibody, HRP‐conjugated streptavidin and diaminobenzidine.

Statistical analysis.  Significant differences between groups were assessed with the χ²‐test and Fisher's exact test. P < 0.05 was considered significant.

Results

Detection of autoantigens associated with NSCLC by 1‐DE western blotting and LC‐MS/MS.  In order to screen for autoantibodies against cancer cells in patients with NSCLC, proteins extracted from a given patient's tumor tissue were subjected to 1‐DE, transferred to membranes, and incubated with sera from the same patient or from healthy control subjects. Membranes incubated with only the secondary antibody were used as negative controls. An approximately 47‐kDa band was recognized only by a subset of NSCLC patient sera, whereas no such reaction was observed with healthy control sera or negative controls (Fig. 1a). To identify this 47‐kDa protein, the corresponding band on the gel stained with CBB was digested and analyzed using LC‐MS/MS. The eight proteins, including α‐enolase and elongation factor 1‐α 1, which were identified by database searching through Mascot software, are listed in Table 1. Many of these proteins are of similar molecular weight and one of them may be the autoantigen associated with NSCLC. We used western blotting with rabbit anti‐enolase antibodies to confirm that the expression of α‐enolase occurred at the same position as that of the 47‐kDa positive band (Fig. 1a).

Figure 1.

(a) Screening by means of one‐dimensional electrophoresis (1‐DE) western blotting analysis for autoantigen associated with non‐small cell lung cancer (NSCLC). Lane 1: ~47‐kDa positive band (arrow), which was recognized by anti‐enolase antibodies. Lanes 2, 3: 47‐kDa positive band, which was recognized by NSCLC sera. Lane 4: no 47‐kDa positive band was observed in a negative NSCLC case. Lane 5: no positive reaction was observed with healthy control sera. Lane 6: no positive reaction was observed in negative controls. (b) Detection by means of two‐dimensional electrophoresis (2‐DE) western blotting analysis of autoantigen associated with NSCLC. Left panel: Representative 2‐DE western blotting analysis. Right panel: corresponding 2‐DE silver‐stained image. Protein spots recognized only by NSCLC sera are marked with arrows and numbers. (c) Peptide mass fingerprinting spectra of positive spot 5. For spot 5, 14 peptide masses were matched with human α‐enolase by executing an NCBInr database search, yielding 52% protein sequence coverage. The matched mass peaks are marked with arrow heads.

Table 1.

Mascot search results of the liquid chromatography‐mass spectrometry/mass spectrometry (LC‐MS/MS) data

Protein name	Swiss Prot Accession number	Molecular weight (Da)	pI	Score*	Peptide matched	Protein coverage (%)
Elongation factor 1‐α 1	P68104	50 451	9.10	123	4	7
Cytokeratin 17	Q04695	48 230	4.97	271	13	25
α‐Enolase	P06733	47 037	6.99	96	3	5
Elongation factor Tu	P49411	49 852	7.26	56	1	3
α‐1‐acid glycoprotein 1 precursor	P02763	23 725	4.93	160	4	18
Vimentin	P08670	53 545	5.06	115	5	9
Albumin precursor	P02768	71 317	5.92	106	4	6
Actin‐like protein 3	P61158	47 797	5.61	46	2	7

*Scores > 39 indicate identity or extensive homology (P < 0.05). To identify the 47‐kDa protein recognized only by non‐small cell lung cancer patient sera, the corresponding band stained with Coomassie brilliant blue was digested and analyzed by LC‐MS/MS. The eight proteins identified are listed.

Autoantibodies against α‐enolase present in NSCLC patient sera.  To characterize autoantibodies in NSCLC sera, proteins extracted from a given patient's tumor tissue were separated by 2‐DE, transferred to membranes, and incubated with sera from the same patient or from healthy control subjects. Compared with the sera of healthy control subjects, 2‐DE western blotting with NSCLC patient sera showed five positive protein spots (Fig. 1b, left panel), including one spot (spot 5) that also had a molecular weight of approximately 47 kDa and a pI value of approximately 7.0. The corresponding spots on the silver‐stained gel (Fig. 1b, right panel) were identified by MALDI‐TOF/MS and database search. The identified proteins are summarized in Table 2. Spot 5 was recognized as α‐enolase, as in the previous LC‐MS/MS analysis. Its PMF spectrum is shown in Fig. 1c and the database search produced 14 peptide masses that coincided with human α‐enolase, thus yielding 52% protein sequence coverage.

Table 2.

 Mascot search results of matrix‐assisted laser desorption–ionization time‐of‐flight (MALDI‐TOF) data

Spot no.	Protein name	Sequence coverage (%)	Molecular weight (Da)	pI
1	Chain D, myeloperoxidase	22	53 806	9.43
2	Tumor rejection antigen‐1 (gp96)	17	92 696	4.76
3	Not identified
4	α glucosidase II α subunit	15	86 236	5.71
5	α‐enolase	52	47 037	6.99

To identify the immunoreactive spots in two‐dimensional electrophoresis western blotting analysis recognized by non‐small cell lung cancer patient sera, corresponding silver‐stained spots were digested and analyzed by MALDI‐TOF/mass spectrometry. ‘Spot no.’ corresponds to spots marked in Fig. 1b.

Next, western blotting with full‐length recombinant human α‐enolase protein was used to confirm and analyze the immunogenicity of α‐enolase. Correct expression of the recombinant protein was verified by western blotting using rabbit anti‐enolase antibodies (Fig. 2a). The recombinant proteins were then probed with sera from NSCLC patients or healthy control subjects, and positive bands were detected only in sera from the former, not from the latter (Fig. 2a). In addition, western blotting was used to determine the reactivity of NSCLC patient sera to enolase isoforms, which contain α, β and γ‐enolase. The sera that were positive for autoantibodies against α‐enolase reacted with neither β‐enolase nor γ‐enolase (Fig. 2b), whereas healthy control sera did not react with any of the enolase isoforms. This indicates the specificity of autoantibodies against α‐enolase in NSCLC patient sera, and the overall results suggest that a subset of NSCLC patient sera contains autoantibodies against α‐enolase.

Figure 2.

(a) Western blotting analysis of recombinant human full‐length α‐enolase protein. Recombinant human full‐length α‐enolase protein was probed with rabbit anti‐enolase antibodies (lane 1), with sera from non‐small cell lung cancer (NSCLC) patients (lane 2–5), and with sera from healthy control subjects (lane 6–7). (b) Western blotting analysis of α, β and γ‐enolase. α, β and γ‐enolase protein were stained with Coomassie brilliant blue (left panel) or probed with sera from NSCLC patients (middle panel) and from healthy control subjects (right panel).

Frequencies of autoantibodies against α‐enolase in the sera were determined by means of ELISA using recombinant human full‐length α‐enolase protein. We tested 94 sera from patients with NSCLC, 15 from patients with SCLC, 18 from patients with gastrointestinal cancer (10 patients with gastric cancer, 8 patients with colon cancer), nine from patients with MAC infection of lung and 60 from healthy control subjects. When ‘Mean OD_{healthy control sera} + 3 SD_{healthy control sera}’ was used as the cut‐off point, the prevalence of autoantibodies against α‐enolase was 27.7% in patients with NSCLC (26 of 94), 1.7% in healthy control subjects (1 of 60), and not detectable in SCLC, gastrointestinal cancer or MAC infection of lung (Fig. 3a). These results showed that titers of autoantibodies against α‐enolase are increased in a subset of NSCLC patients. Next, we examined the correlation between the prevalence of autoantibodies against α‐enolase and clinicopathological features in NSCLC patients. Positive reactivity was detected in 21 of the 73 sera from adenocarcinoma patients (28.8%) and in five of the 21 sera from SCC patients (23.8%). There was no significant correlation between the occurrence of autoantibodies against α‐enolase and pathological types (P = 0.654). In addition, there was a tendency for autoantibodies against α‐enolase to be more prevalent in patients with advanced NSCLC cases (stage III/IV, 33.3%, 21 of 63) than in stage I/II cases (16.1%, 5 of 31), although the results of the statistical analysis suggest that the prevalence has no significant correlation with disease stage (P = 0.08). We also investigated the relationship between autoantibodies against α‐enolase and other tumor markers (CEA, CYFRA 21‐1) that have been applied to clinical practice in NSCLC patients. In a total of 94 NSCLC patients, clinical data of both CEA and CYFRA 21‐1 were available for 75 patients. In these patients, 25.3% (19/75) were positive for autoantibodies against α‐enolase, 42.7% (32/75) were positive for CEA, and 30.7% (23/75) were positive for CYFRA 21‐1. The occurrence of autoantibodies against α‐enolase didn't show a significant correlation with CEA (P = 0.63) or with CYFRA 21‐1(P = 0.92). Detection of CYFRA 21‐1 and CEA in combination increased the positive detection rate to 58.7% (44/75). Furthermore, positive detection rate was enhanced up to 69.3% (52/75) when combined detection of autoantibodies against α‐enolase, CEA and CYFRA 21‐1 was used (Fig. 3b).

Figure 3.

(a) Prevalence of autoantibodies against α‐enolase determined by enzyme‐linked immunosorbent assay (ELISA). The y‐axis denotes binding units. The solid horizontal line represents the positive cut‐off limit. The prevalence of autoantibodies against α‐enolase was 27.7% in patients with non‐small cell lung cancer (NSCLC) (26 of 94), 1.7% in healthy control subjects (1 of 60), and not detectable in small cell lung cancer gastrointestinal cancer (10 patients with gastric cancer, 8 patients with colon cancer) and Mycobacterium avium complex infection of lung. (b) Positive detection rate of autoantibodies against α‐enolase, cytokeratin 19 fragment (CYFRA21‐1) and carcinoembryonic antigen (CEA) in NSCLC patients. Detection of CYFRA 21‐1 and CEA in combination increased the positive detection rate to 58.7%. Furthermore, combined detection of autoantibodies against α‐enolase, CEA and CYFRA 21‐1 achieved a positive detection rate of up to 69.3%.

IHC and flow cytometric analysis of α‐enolase.  We used IHC staining to compare α‐enolase expression in non‐malignant and malignant lung tissues from 20 NSCLC patients, including 10 patients with autoantibodies against α‐enolase and 10 patients without autoantibodies against α‐enolase. The staining results showed that expression of α‐enolase was increased in malignant lung tissue of NSCLC patients (Fig. 4a–c). Additionally, IHC staining showed not only cytoplasmic but also membranous immunoreactivity in cancer cells (Fig. 4c). Flow cytometric analysis of human lung adenocarcinoma cell line A549 also confirmed the expression of α‐enolase at the surface of lung cancer cells (Fig. 4d).

Figure 4.

Immunohistochemical and flow cytometric analysis of α‐enolase. (a) Normal lung tissue (×100). (b) Lung adenocarcinoma (×100). (c) α‐Enolase staining showing a mixture of cytoplasmic, and membranous immunoreactivity (×400). (d) A549 cells were stained with anti‐enolase antibody, labeled with fluorescein‐isothiocyanate‐conjugated goat antirabbit immunoglobulin, and analyzed on a FACS Canto (open histogram). Shaded histogram indicates staining with control IgG.

Epitopes located at the N‐terminal region (1–134 a.a.) of α‐enolase that are recognized by autoantibodies.  To locate the serological epitopes of α‐enolase, full‐length and C‐terminal deletion mutant proteins (α‐Eno1, α‐Eno2, α‐Eno3) were prepared. The full‐length α‐enolase (1–434 a.a.), α‐Eno1 (1–334 a.a.), α‐Eno2 (1–234 a.a.) and α‐Eno3 (1–134 a.a.) recombinant proteins were clearly shown by an anti‐6 × His antibody or stained with CBB, which verified their expression (Fig. 5). A commercially available rabbit anti‐enolase antibody reacted only with the full‐length α‐enolase, α‐Eno1 and α‐Eno2 recombinant proteins (Fig. 5). However, sera from NSCLC patients who showed the presence of autoantibodies against α‐enolase reacted with the α‐Eno1, α‐Eno2, and α‐Eno3 recombinant proteins (Fig. 5), indicating that in NSCLC patients at least the N‐terminal region of α‐enolase contains epitopes.

Figure 5.

Upper panel: preparation of recombinant human full‐length α‐enolase protein: α‐Eno (1–434 a.a.), C‐terminal deletion mutant α‐enolase proteins: α‐Eno1 (1–334 a.a.), α‐Eno2 (1–234 a.a.) and α‐Eno3 (1–134 a.a.). Lower panel: recombinant proteins were stained with Coomassie brilliant blue or probed with anti‐6 × His antibody, anti‐enolase antibody or non‐small cell lung cancer sera.

Discussion

In the present study we used a proteomics‐based screen test to identify proteins such as α‐enolase and gp96 that may elicit a humoral immune response in NSCLC patients. We then confirmed that some NSCLC patients’ sera contained autoantibodies against α‐enolase by means of western blotting using recombinant protein. Furthermore, the results obtained with ELISA demonstrated that when ‘Mean OD_{healthy control sera} + 3 SD_{healthy control sera}’ was used as the cut‐off point, the humoral immune response directed against α‐enolase occurred in 27.7% of NSCLC patients, but in only 1.7% of healthy control subjects. α‐enolase is an isoenzyme of enolase, a key protein that catalyzes the conversion of 2‐phosphoglycerate to phosphoenolpyruvate, which is the second of the two high‐energy intermediates that generate ATP in glycolysis.^{( 20 )} Three isoforms of enolase have been identified and are known as α, β and γ‐enolase. α‐enolase is present in most tissues and is predominant in early embryonic tissue, β‐enolase is expressed in muscle tissue, and γ‐enolase, also known as NSE, is found only in neuronal tissues.

Autoantibody responses to tumors are generally thought to be elicited in three ways. These are overexpression of specific proteins, especially on the cell surface, gene mutation or post‐translational modification of proteins, which shows new epitopes as immunogens, and other types of protein processing in tumor tissue.^{( 21 )} In the present study we used normal recombinant proteins to confirm the immunogenicity of α‐enolase in NSCLC patients. In addition, although a past study reported that expression of α‐enolase was downregulated in NSCLC tissues,^{( 22 )} IHC staining used in our study showed that α‐enolase expression is commonly increased in malignant lung tissue from NSCLC patients compared to the expression in non‐malignant lung tissue, which is consistent with other previous reports.^{( 23} , ²⁴ , ²⁵ , ²⁶ , ^{27 )} Interestingly, IHC staining also showed not only cytoplasmic but also membranous immunoreactivity. Moreover, flow cytometric analysis demonstrated expression of α‐enolase on the surface of lung cancer cells. We think that enhanced expression of α‐enolase on cancer‐cell surface might be one reason for autoantigenicity, and might be required for induction of autoantibody responses. However, α‐enolase expression alone is insufficient for autoantibody production as increased expression was also found in these patients without autoantibodies against α‐enolase. Further study is necessary to investigate the detailed mechanisms involved in this autoantibody response.

It is widely accepted that the propensity for glycolysis is enhanced in cancer cells because of increased cell proliferation. In fact, α‐enolase, a key enzyme in the glycolysis pathway, was found to be overexpressed in 18 cancers.^{( 25 )} Furthermore, although the mechanism of its surface expression and orientation on the membrane are not yet clearly understood, it is known that the C‐terminal a.a. of α‐enolase, lysine, is exposed at the cell surface and is involved in binding to plasminogen, which is then activated and converted to plasmin.^{( 28 )} Once plasmin is stabilized at the cell surface, it in turn induces fibrinolysis.^{( 20 )} In response to the upregulation of α‐enolase expression, progression of the fibrinolytic system is markedly accelerated, and the resultant increase in local fibrinolysis may contribute to cancer cell invasion and metastasis. This is consistent with our finding that there was a tendency for autoantibodies against α‐enolase to be more prevalent in patients with advanced NSCLC (stage III/IV) than in stage I/II cases.

Tumor markers for NSCLC are potentially useful for both diagnostic and therapeutic practice. To date, a variety of NSCLC tumor markers have been identified and the most extensively investigated circulating protein markers include CEA, CYFRA 21‐1, SCC antigen, NSE and CA125.^{( 5 )} The percentages of NSCLC patients who have elevated serum protein levels of the above markers are 26%–42% for CEA, 51%–74% for CYFRA 21‐1, 20%–32% for SCC, 28–32% for NSE, and 46–55% for CA125, with variations depending on histology and stage.^{( 5} , ⁶ , ⁷ , ^{8 )} However, because elevated serum protein levels of these markers have been observed in tumors other than NSCLC, their sensitivity and specificity are not satisfactory and their clinical applicability is limited. Our study demonstrated that when ‘Mean OD_{healthy control sera}+ 3 SD_{healthy control sera}’ was used as the cut‐off point, autoantibodies against α‐enolase occurred in 27.7% of NSCLC patients but not in those with SCLC, gastrointestinal cancer or MAC infection of the lungs. Moreover, combined detection of autoantibodies against α‐enolase, CEA and CYFRA 21‐1 enhanced sensitivity for the diagnosis of NSCLC. These results suggest that using autoantibodies against α‐enolase has potential as a clinical biomarker for serological screening of NSCLC. Further large‐scale validation studies will be needed to determine the sensitivity, specificity and positive prognostic value of this marker in real‐world screening scenarios.

Autoantibodies against α‐enolase have been detected in some autoimmune diseases,^{( 20 )} and Fujii et al. report that in Hashimoto's encephalopathy one of these autoantibodies recognizes the N‐terminal region of α‐enolase.^{( 29 )} Because our results show this same recognition in NSCLC, further experiments are warranted to compare epitopes in α‐enolase as detected in NSCLC and in autoimmune diseases to determine its utility as a biomarker for NSCLC.