| Hepatocellular carcinoma (HCC) is one of the most frequent and lethal malignancies worldwide, and the 5-year survival rate after hepatectomy is 30-50%. Given the high incidence and mortality of HCC, it's important to develop biomarkers for assessing onset or prediction of therapy outcome, as well as to identify targets for the development of novel therapies. Although there are multiple promising diagnostic biomarkers of HCC in the development phase, there are at the present time no definitive antibody-based serologic markers for its early diagnosis in clinic. Several experimental studies have led to the identification of autoantigens via recognition by autoantibodies present in HCC sera. Antigens that have been shown to induce a humoral response in HCC include p53 and diverse other nuclear proteins. Autoantibodies to cyclin B1 and to a novel cytoplasmic protein with RNA binding motifs have also been reported. However, efforts to consistently predict HCC based on autoimmunity to antigens have not resulted in serologic markers with definitive specificity and sensitivity.Serological proteome analysis (SERPA) is a powerful tool for the identification of a large group of candidate cancer biomarkers recognized as autoantigens by cancer patients' sera. This approach permits the transfer and immobilisation of proteins to a semi rigid support, allowing subsequent immunodetection of relevant antigens among thousands of individual proteins separated by 2-DE. To survey autoantibodies associated with HCC, we used this SERPA approach and identified thirteen proteins that induced a humoral response in HCC patients but not in healthy individuals, and subsequently investigated the prevalence of the autoantibodies in a large number of HCC patients and negative controls.We separated total proteins extracted from the two cell lines (HepG2 and HepG2.2.15) by 2-DE and transferred onto PVDF membranes. Initially, mixed serum samples from eight HBV-related HCC patients were incubated with HepG2 proteins. Normal sera were utilized as controls. Then mixed serum samples from ten HBV-related HCC patients and ten virus negative HCC patients were separately reacted with HepG2.2.15 proteins. We detected six antigenic protein spots of HepG2 cells that reacted to HCC sera but not reacted to normal controls. In the case of HepG2.2.15, we detected ten antigenic protein spots that reacted to HCC sera but not to normal controls. Among the ten proteins, five reacted to sera from all HCC patients with or without HBV infection, two only to sera without HBV infection, and three only to sera with HBV infection, reflecting a difference between the prevalence of autoantibodies in HBV-related HCC and virus negative HCC.We tried to identify the proteins detected by SERPA. The proteins of interest were excised from stained Colloidal Coomassie Brilliant G-250 gels and submitted to digest with trypsin. The peptide mixtures were analyzed by MALDI-TOF MS. The resulting peptide mass maps were used for protein database searches using Mascot software. Six protein spots from HepG2 cell were identified as four distinct proteins, including keratin 8, lamin A/C, DEAD (Asp-Glu-Ala-Asp) box polypeptide 3, and eukaryotic translation elongation factor 2 (eEF2). Of the ten protein spots from HepG2.2.15 cell, two reactive proteins with autoantibodies in virus negative HCC sera were identified as triosephosphate isomerase (TIM) and prostatic binding protein, three reactive proteins with autoantibodies in HBV-related HCC sera were identified as pyruvate kinase, phosphoglycerate kinase 1, and phosphoglycerate mutase isozyme B, and the rest which reacted to both of the HCC sera were identified as programmed cell death 8 (apoptosis-inducing factor, AIF), heterogeneous nuclear ribonucleoprotein (hnRNP) A2, cyclophilin A, and aspartate aminotransferase 1.Antigenicity of the identified protein was confirmed by WB. Four identified antigens from HepG2 cell and three from HepG2.2.15 cell were randomly selected for the expression of recombinant proteins. The protein TIM was purchased from Sigma-Aldrich Company. Keratin 8 cDNA was subcloned into pGEX-4T-l vector, producing a fusion protein with GST-tag. cDNA from lamin A/C, eEF2, and prostatic binding protein was subcloned into pET28a(+) vector respectively, and recombinant proteins were expressed as fusion proteins with His-tags. cDNA from DEAD box polypeptide 3, hnRNP A2, and AIF was separately subcloned into pET43.1a(+) vector for protein expression as fusion proteins with Nus-tags and His-tags. The purified fusion proteins were separated on 10% SDS-PAGE and stained with Coomassie blue. The purity of the recombinant proteins was over 90%, which was evaluated with the software Bandscan.Using the eight proteins as antigens, sera from 28 HCC patients and 10 normal controls were screened individually by WB analysis. The frequency of autoantibodies to individual protein ranged from 21.4% to 71.4% in HCC sera, but from 0 to 20% in normal sera. The antigen arrays contained seven recombinant proteins and TIM purchased from Sigma. These candidate antigens were robotically attached in ordered arrays to aldehyde surface of slides, on which the binding of serum autoantibodies was detected. Sera from 118 patients with HCC, 63 patients with CHB or CHC, 66 patients with other cancers and 43 healthy individuals were analyzed by protein microarrays. After the removal of local background, the mean signal intensity of 43 normal sera against each protein respectively plus 2 standard deviation was used to determine whether the target was considered to be positive or not. Two antigens, keratin 8 and lamin A/C, showed strong signals in chronic hepatitis with a high frequency, presumably due to cross-react with the antibodies in both HCC sera and chronic hepatitis sera. These two antigens were removed from subsequent analyses.The frequency of positive reactions to each antigen in HCC was significantly higher than that in normal controls (P<0.001). When comparing the frequency of autoantibodies between HCC and chronic hepatitis, it was observed that HCC patients had a higher percentage of positive reactions to DEAD box polypeptide 3, eEF2, AIF, and TIM at P<0.001 and to hnRNP A2 and prostatic binding protein at P<0.05. For alpha-fetoprotein (AFP), no significant difference was found between HCC and chronic hepatitis.In order to determine whether the autoantibodies could distinguish HCC with other cancers, the prevalence of autoantibodies to the six antigens in sera from other cancers was further studied. For DEAD box polypeptide 3, AIF, and prostatic binding protein, antibody frequency in any other individual cancer was significantly lower than that in HCC (P<0.05), ranging from 17.6% to 35.3%. In breast cancer, it was observed that high positive reactions to eEF2 and hnRNP A2 were 73.3% and 53.3%, respectively. In lung cancer, high positive reactions to hnRNP A2 and TIM were both 64.7%. Differences in the reactions of different cancers were observed for each of the six antigens.Using an AFP level of 20 ng/ml as a cut-off value for diagnosing HCC, the sensitivity and specificity were 72.9% and 75%, respectively. For the identified autoantibodies, the sensitivity of DEAD box polypeptide 3, eEF2, AIF, hnRNP A2, prostatic binding protein and TIM was 85.6%, 78.8%, 55.9%, 64.4%, 48.3% and 64.4%, respectively, and the specificity versus all non-HCC subjects ranged from 69.8-82.6%. In the case of HCC at TNM stage I, the sensitivity for AFP was only 50% (10/20), and the sensitivity for DEAD box polypeptide 3, eEF2, AIF, hnRNP A2, prostatic binding protein and TIM was 85% (17/20), 75% (15/20), 60% (12/20), 70% (14/20), 50% (10/20) and 60% (12/20), respectively. In particular, we observed that 8 of 20 stage I HCC patients with autoantibodies against any of the six antigens had a normal range of serum AFP levels. With the addition of antigens to a total of six, it was appeared that the sensitivity in diagnosing stage I HCC increased, from 50-85% when one antigen was used to 90% when six antigens were used.In order to identify markers associated with clinicopathological characteristics of patients, the relationship between tumor size, histological grade, metastasis, TNM classification, and antibody levels needed additional research. There was no statistical difference among each autoantibody expression and sex, histological grade or TNM classification. Tumors (>5 cm) were more frequently anti-eEF2 positive than small tumors (≤5 cm) (P<0.05). The positive rates of AIF and hnRNP A2 in HCC without regional lymph node metastasis were significantly higher than those with regional lymph node metastasis (P<0.05).Other markers such as lens culinaris agglutinin-reactive AFP (AFP-L3) and des-gamma carboxyprothrombin (DCP) have also been proposed as markers for HCC diagnosis. AFP-L3, known as fucosylated variant of AFP, appeared to be useful at improving the specificity for patients with AFP elevation between 20 and 200 ng/mL, but suffered from a poor sensitivity in detecting early HCC. In HCC (<2 cm), AFP-L3 had a sensitivity of only 35-45%. The efficacy of DCP in diagnosis of early HCC still remains unclear. It has been reported that the clinical utility of DPC might be dependent on tumor size, and the sensitivity of DPC (41%) was worse than that of AFP (55%) for tumors less than 3 cm. Compared with these two biomarkers, the sensitivity of autoantibodies we identified varied from 50-85%, enabling detection of HCC at stage I. Further analysis with large randomized controlled trials is needed to compare HCC-associated antigens with AFP, AFP-L3 and DCP. Since AFP, AFP-L3, DPC and HCC-associated antigens behave independently, they have the potential to complement each other for diagnosis of early HCC.In conclusion, we have identified several novel autoantibodies present in the sera of HCC patients with serological proteome analysis combined with protein microarray. Some of the autoantibodies against HCC-associated antigens we identified may have value on HCC early diagnosis. Additionally, the combination of proteomic technique and highly sensitive protein microarrays may provide a powerful diagnosis system in the near future. |
PDF Full Text Request