Evaluation of antibody level against Fusobacterium nucleatum in the serological diagnosis of colorectal cancer

Fusobacterium nucleatum (F. nucleatum, Fn) is associated with the colorectal cancer (CRC). Fn-infection could induce significant levels of serum Fn-specific antibodies in human and mice. The objective of this study was to identify Fn-infection that elicit a humoral response in patients with CRC and evaluate the diagnostic performance of serum anti-Fn antibodies. In this work, we showed the mean absorbance value of anti-Fn-IgA and -IgG in the CRC group were significantly higher than those in the benign colon disease group and healthy control group (P < 0.001). The sensitivity and specificity of ELISA for the detection of anti-Fn-IgA were 36.43% and 92.71% based on the optimal cut-off. The combination of anti-Fn-IgA and carcino-embryonic antigen (CEA) was better for diagnosing CRC (Sen: 53.10%, Spe: 96.41%; AUC = 0.848). Furthermore, combining anti-Fn-IgA with CEA and carbohydrate antigen 19-9 (CA19-9) (Sen: 40.00%, Spe: 94.22%; AUC = 0.743) had the better ability to classify CRC patients with stages I-II. These results suggested that Fn-infection elicited high level of serum anti-Fn antibodies in CRC patients, and serum anti-Fn-IgA level may be a potential diagnosing biomarker for CRC. Serum anti-Fn-IgA in combination with CEA and CA19-9 increases the sensitivity of detecting early CRC.

Commensal bacteria in the colon may play a role in colorectal cancer (CRC) development 1 . Accumulated studies show that several bacterial species seem to be involved in pathogenesis of CRC 2 . Fusobacterium nucleatum (F. nucleatum, Fn) is an opportunistic commensal anaerobe in the oral cavity which is also prevalent in human CRC tissues 3 , and is over-represented in disease tissue versus matched normal tissue in CRC patients 4,5 . Moreover, an overabundance of Fn has been reported to be associated with colorectal adenomas 6 , and show the potential as a risk factor for disease progression from adenoma to cancer 7 . Recently, multiple studies further identified the increased carriage of Fn in mucosal or fecal samples of CRC patients by 16 s RNA sequencing and quantitative PCR (qPCR) analysis 1, 8,9 . The amount of Fn in CRC tissue has been proved to be associated with proximal tumor location and CpG island methylator phenotype (CIMP) status, microsatellite instability (MSI), and gene mutations in BRAF, KRAS and so forth [10][11][12] . Furthermore, a study has demonstrated that combining qPCR measurements of Fn with other several bacterial species had the ability to accurately diagnose patients with CRC 13 . In addition, it was demonstrated that Fn DNA was enriched in early-stage (I-II) patient and had the potential as a non-invasive early diagnostic biomarker for CRC from faecal samples 13 .
In general, microbe antigens can elicit a humoral immune response in vivo. In infection-associated cancers, a promising approach for the early detection of cancer is the assessment of immune response to antigens of tumor-associated microbe. Serological testing of antibodies against cancer-associated microorganisms including Epstein-Barr Virus (EBV), human papillomavirus (HPV) and Helicobacter pylori (Hp) has been used in diagnosis of the infection and tumor screening 14,15 . Previous studies reported that Fn induced significant humoral antibody response in human and mice with chronic oral Fn-infection [16][17][18][19] . The elevated antibodies level of Fn is also a risk factor for Alzheimer's disease and rheumatoid arthritis [20][21][22] . Moreover, F. nucleatum immunization prior to that Scientific RepoRts | 6:33440 | DOI: 10.1038/srep33440 of P. gingivalis inhibited the production of anti-P. gingivalis antibodies, suggest that F. nucleatum do not allow the production of cross-reactive antibodies to other similar oral microorganisms 23 . CRC incidence has increased at an alarming rate over the last twenty years 24 . Most cases of CRC are curable if diagnosed early enough, survival rates for early stage detection is about five times that of late stage cancers 25 . As a consequence, there is an urgent need to explore valuable early diagnosis markers for CRC patients. In the present study, we measured the preoperative anti-Fn levels in CRC patients to evaluate the clinical value of anti-Fn as a diagnostic parameter in those patients with colon cancer.

Anti-Fn antibodies in sera of CRC patients with Fn infection.
To investigate the presence of antibodies against Fn in sera of CRC patients, we first screened Fn infection by PCR from the stool samples of 10 CRC patients and 10 matched healthy controls (Fig. 1A). 6 Fn-positive samples from CRC patients and 1 positive sample from healthy controls were detected. Sera of 6 Fn-positive patients and equal numbers of Fn-negative healthy controls were used to detect the specificity reactive antigens with the serum by western blotting. Several strong reactive antigen bands were observed with all Fn-positive sera samples when the serum was diluted 1:10000 and incubated with HRP-IgA, but no obvious bands were observed to react with sera of 6 healthy controls (Fig. 1B). However, it presented some nonspecific reactivity bands when followed with HRP-IgG (Fig. 1C). Additionally, to determine whether the serum antibodies were specific to Fn, we conducted the parallel experiments to detect the antibodies to the four control microorganisms. No obvious band was detected in B. fragilis, E. coli, E. cloacae or E. Faecalis with Fn positive or negative serum when diluted at 1:10000 and incubated with HRP-IgA (data not shown). 7 obviously reactive antigen bands in a molecular mass range from 15 kDa to 75 kDa were chosen as the interest proteins which triggered a strong anti-Fn-IgA response. Those corresponding proteins were extracted from the gels following SDS-PAGE and coomassie brilliant blue R250 staining (Fig. 1D). These proteins were digested with trypsin, and the resulting peptides were analyzed by MALDI-TOF/TOF analyzer. The corresponding spectra were used for a protein search in the NCBI and ExPASy databases. 7 major proteins identified by bioinformatics are shown in Table 1. The two corresponding proteins which present the strongest reactive bands are Alkyl hydrogen peroxide C (21.2 kDa) and major outer membrane protein FomA (42.3 kDa).
Those results suggested that Fn protein can elicit a strong humoral response and produce high levels of specific antibodies in CRC patients. The serum anti-Fn-IgA and anti-Fn-IgG showed an ability to diagnose the Fn infection. Evaluation of serum anti-Fn levels in patients with CRC. Furthermore, anti-Fn level was investigated in sera samples by indirect whole-cell ELISA. The coated bacteria whole cells served as an antigen to react with sera of patients and controls for detecting potential antibodies present in the sera. A total of 258 patients with CRC (55 in stage I-II) and two control groups (150 benign colon disease and 200 healthy controls) were recruited. CRC patients infected with Fn produce higher level of IgA than IgG. The average absorbance ± SD was observed using the 1:1000 diluted sera in IgA and 1:4000 diluted in IgG. As shown in Fig. 2A, the average absorbance ± SD for anti-Fn-IgA in CRC groups and its stage I-II of CRC groups was 0.390 ± 0.215, 0.367 ± 0.163, respectively; while in healthy controls and benign colon disease group was 0.246 ± 0.132, 0.268 ± 0.158 respectively. The circulating levels of anti-Fn-IgA in patients with CRC group and its stage I-II groups were significantly higher than those of the two control groups (P < 0.001), and there was no significant difference in the benign colon disease group and healthy control group. In addition, the circulating levels of anti-Fn-IgG in patients with CRC group (0.362 ± 0.194) were significantly higher than those of two control groups (0.262 ± 0.152; 0.270 ± 0.162), but there was no significant difference in stage I-II and control group (Fig. 2B). Although the highest levels were observed in later stage CRC (stage III-IV), the serum levels of anti-Fn-IgA and anti-Fn-IgG of CRC patients in stage I-II were not significantly difference than stage III-IV (P = 0.890; P = 0.200). To determine whether the serum antibodies were specific to Fn, we conducted the parallel-controlled assays by using whole cell ELISA. Fig. S3A-C showed that the OD value of anti-B. fragilis, anti-E. coli and anti-E. cloacae-IgA or -IgG were too low to be detected when applied diluted sera (1:1000 for anti-IgA or 1:4000 for anti-IgG level detection) from both patients and controls. Though the level of anti-E. Faecalis-IgA or -IgG were much higher than the other three bacterial strains, it showed no significant difference between CRC and two control groups (Fig. S3D). Table 2 shows the relationship between anti-Fn and clinicopathological variables in CRC patients. The anti-Fn-IgA and anti-Fn-IgG were not obviously correlated with age, gender, tumor volume, T classification, N classification, metastasis and clinical stage. However, there was a significant association between the presence of anti-Fn-IgG and CRC histological differentiation (P = 0.032). The level of anti-Fn-IgG was higher in CRC patients with poorly differentiated adenoma than those with good differentiation. These results indicated that anti-Fn level is similar in CRC patients with advanced stage and in patients with early stage.

Association between antibodies against Fn and clinicopathological, serological characteristics of CRC patients. The data presented in
Furthermore, Pearson's correlation coefficient and linear regression analysis were applied to analyze the correlation among the levels of anti-Fn-IgA, anti-Fn-IgG and the tumor markers CEA and CA19-9. Anti-Fn-IgA had a positively correlation with Anti-Fn-IgG (R = 0.365, P < 0.001). Neither CEA nor CA19-9 was associated with anti-Fn-IgA or -IgG levels (data not shown). In addition, anti-Fn-IgA and -IgG levels were similar in both CEA-positive and CEA-negative CRC patients, or in CA19-9-positive and CA19-9-neagtive CRC patients. These results suggested that anti-Fn-IgA and -IgG level were not associated with the progression of the CRC, but anti-Fn-IgA and -IgG level may provide an additional diagnostic value for those patients with CEA-negative or in early stage of CRC.

Anti-Fn antibody levels have a diagnostic value for CRC.
To determine whether levels of antibodies against Fn had diagnostic value for CRC, the ROC curve was plotted to identify a cut-off value that would distinguish CRC from both of nonmalignant colon diseases and health controls ( Fig. 3; Table 3). As shown in Fig. 3, Protein credibility. g) The total ion score. h) The total ion credibility. i) The number of indentified amino acids against the total number of amino acids in the target protein.
the area under the ROC curve (AUC) for anti-Fn-IgA and anti-Fn-IgG was 0.704 (95% CI: 0.663-0.746) and 0.645 (95% CI: 0.601-0.688), with an optimal cut-off value 0.420 and 0.230 respectively, whereas the AUCs for CEA and CA19-9 were 0.796 (95% CI: 0.758-0.834) and 0.635 (95% CI: 0.587-0.683), respectively. As shown in Table 3, the sensitivity of anti-Fn-IgA was 36.43%, which is higher than that for CA19-9 (20.16%), but lower than CEA (41.47%), whereas the specificity of anti-Fn-IgA was slightly lower (92.71%) based on the optimal cut-off (0.42) according to the Youden Index. The sensitivity of anti-Fn-IgA (30.62%) is reduced with the further adjustment of cut-off value (0.45) based on the similar specificity with CEA (about 96%) in this study. Anti-Fn-IgG demonstrated lower sensitivity and specificity, similar to that of CA19-9. Moreover, Table 3 demonstrated that the sensitivity of the combination of anti-Fn-IgA and CEA was superior to that of the combination of CEA and CA19-9 (53.10% vs. 45.74%) without compromising the specificity (96.41% vs. 96.26%) by using a higher adjustment cut-off score (0.70). In addition, the combination of anti-Fn-IgA and CEA exhibited a similar PPV (positive predictive value, 92.6% vs. 91.5%) and higher NPV (negative predictive value, 70.9% vs. 66.9%) compared with the combination of CEA and CA19-9. The combination of anti-Fn-IgA with CEA and CA19-9 slightly increased the diagnosis value compared with the combination of anti-Fn-IgA and CEA. These results suggest that IgA, but not IgG against Fn possess diagnostic capabilities for CRC, and anti-Fn-IgA increases the sensitivity in CRC detection and offers additional diagnosis value in CRC.
Diagnostic value of individual serum anti-Fn, CEA and/or CA19-9 levels in the detection of early stage CRC. The level of anti-Fn-IgA was similar in CRC patients with advanced stage and early disease. Due to the lack of biological markers for detecting early stage colorectal cancer, to investigate the early diagnostic value of individual serum anti-Fn is valuable.
The performance of anti-Fn and CEA or CA19-9 in detecting early stages (stage I, stage II) of CRC was performed in 55 patients with CRC at early stages. As shown in Fig. 4, the AUC of anti-Fn-IgA and anti-Fn-IgG reached 0.709 (95% CI: 0.635-0.784), and 0.612 (95% CI: 0.543-0.681), respectively, whereas the AUCs for CEA and CA19-9 were only 0.624 and 0.492. Table 4 demonstrated the sensitivity and specificity of anti-Fn-IgA was 69.09% and 63.27% based on the optimal cut-off (0.28), and the sensitivity was 27.27% with the higher adjustment cut-off score (0.45) without compromising the specificity (96.21%), but CEA and CA19-9 were detected in 9 patients and the sensitivity was only 16.36% (AUC = 0.605). By combining with the anti-Fn-IgA, CEA and CA19-9, the sensitivity and specificity reached 40.0% and 94.22% based on the optimal cut-off (AUC = 0.743), and exhibited a higher PPV (56.4% vs. 29.7%) and NPV (89.4% vs. 87.4%) than combining CEA with CA19-9 alone. The findings validate the performance of anti-Fn-IgA as a plasma marker to increase the diagnostic sensitivity for   early detection of CRC, and indicate that the serum levels of anti-Fn-IgA, CEA and CA19-9 combined have a better diagnosis values for screening early stage of CRC than CEA and CA19-9 combined diagnosis.

Discussion
In this study, we demonstrated higher levels of polyclonal antibodies against Fn present in the blood of CRC patients with Fn-infections compared to healthy controls. This finding indicated that Fn-infection induced a specific and stronger humoral antibody in patients with CRC. In mice, vaccination of heat-killed whole bacteria sonic extracts with Fn induced a specific humoral antibody response and produced specific higher titer antibodies than other oral bacteria 26 . Recently, the FomA porin from Fn was identified as a toll-like receptor 2 (TLR2) agonist with immune adjuvant activity. FomA induces TLR2-dependent antigen-specific antibody secretion by mouse B cell activation in vivo 27 , which was in line with our findings that Fn-infection patients drive stronger immune responses than healthy controls.
Antibody levels to Fn were also found to be significantly increased in patients with Alzheimer's disease compared with controls by whole Fn cell ELISA assay 22 . The commensal intestinal microbes play a crucial role in the bidirectional gut-brain axis that integrates gut and central nervous system activities. A dysfunction of this axis has been associated with the pathogenesis of peripheral and central nervous system diseases 28,29 . The result further shows that digestive tract infection induced the higher titer of Fn antibodies in humans.
Fn has historically been viewed as an oral pathogen, multiple studies have shown that the antibody levels of whole Fn cell is able to diagnose Fn-infection oral patients. Although it could be suggested that the antibody to these pathogens may have been cross-reactive with antigens from other sources, the previously published data are replete with oral pathogen studies supporting the specificity of these antibodies for oral infections [30][31][32][33][34] , and that successful treatment and maintenance of periodontitis significantly lowers these antibody levels 35 . In addition, H. pylori Whole-Cell antigens ELISA was used to detect the anti-Hp level previously and the results do not show much difference with its anti-CagA ELISA in multiple studies 36,37 .
Specific antibodies have been widely used to diagnosis infection. Some cancer-associated microorganisms, which enter through mucosal infection sites, present specific IgA antibodies for diagnosis and screening. CagA antibodies were significantly more prevalent among individuals with elevated H. pylori antibody titers of the IgA class than in those with IgG antibodies 38 . EBV-IgA related and HPV-IgA related serological testing has been used  Table 4. Diagnostic values of CEA, CA19-9 and antibodies against F. nucleatum in early stage CRC.
for nasopharyngeal carcinoma and cervical carcinoma diagnosis 39,40 . Our study also shows that IgA has higher titer than IgG, suggesting the IgA class may be more valuable for diagnosis CRC than the IgG class. Furthermore, we demonstrated significantly higher anti-Fn-IgA and anti-Fn-IgG levels were present in the serum of CRC patients with Fn infection compare to the healthy controls and or patients with benign colon disease. Recently, numerous studies have indicated that Fn is closely associated with the development and progression of CRC [4][5][6][7] . Our data were consistent with those studies in which Fn-DNA is overabundant in CRC patients 1, 8,9 . Moreover, we showed that anti-Fn-IgA and anti-Fn-IgG were significantly elevated in early stage CRC compared with normal controls or those with benign disease, but only slightly elevated when compared early stage CRC to advanced CRC. These data are in line with the study in which Fn-DNA was enriched in early-stage CRC patients 13 , corresponding to a greater lag time in the production of antibodies in response to infection.
Early diagnosis and treatment of CRC is of great value to improve survival of CRC patients. Currently, CEA and CA19-9 are the two most commonly used diagnosis markers for CRC. However, serum tests such as CEA or CA19-9 have poor sensitivity for detection of early CRC. In our study, Fn-IgA showed higher sensitivity (36.43%) than CA19-9 (20.16%), and anti-Fn-IgA combined with CEA and CA19-9 improved the sensitivity (54.65%) without compromising specificity (96.60%) for CRC detection. These results show the diagnostic value of CRC with anti-Fn-IgA. Furthermore, anti-Fn-IgA exhibited much higher sensitivity (27.27%) than CEA (9.09%) or a combination of CEA and CA19-9 (16.36%) for early CRC detection without compromising specificity (96.21%). The combination of anti-Fn-IgA, CEA and CA19-9 exhibited highest sensitivity (40.0%) and specificity (94.22%) for early CRC detection with highest PPV (56.4%) and good NPV (89.4%). These results demonstrated that the combination of anti-Fn-IgA, CEA and CA19-9 could detect about 40% early CRC, and the improvement in sensitivity compared to CEA was accomplished without obviously compromising specificity. Therefore, anti-Fn-IgA is a valuable serological diagnosis marker for screening early CRC.
Fn DNA load quantified from stool is inconvenient as a large scale population screening assay. Serologic testing for Fn antibodies will be a simpler diagnosis method and may be useful to monitor population exposure to Fn. The identification of microbial antigens that elicit an immune response is important for clinical applications, and microbial antigens may be used for early diagnosis, prognosis, and immunotherapy against the disease with cancer-associated microorganisms. The development of antibody detection for Fn is valuable for prospective epidemiological surveillance and large-scale screening of early CRC. In recent years, mSeptin 9 blood test has been a promising biomarker for detection of colorectal cancer due to its high sensitivity (about 70%), high specificity (above 90%) and its cost-effectiveness [41][42][43] . To improve the sensitivity and specificity for the diagnosis of early CRC, the combination of single or multivalence Fn antigens with mSeptin 9 is worthy of further study.
In summary, Fn infection can elicit a strong humoral immune response. The levels of anti-Fn-IgA and anti-Fn-IgG in sera from CRC patients were significantly higher than those from healthy subjects and benign colon disease. Moreover, our study demonstrated that serum levels of anti-Fn-IgA combined with CEA and CA19-9 have a superior sensitivity than CEA and CA19-9 combined alone in detecting early stage of CRC. Therefore, Fn-IgA antibodies become a potential and useful screen marker for early CRC.

Materials and Methods
Patients, blood and stool samples. Sera   less than 12%. All the OD value were required between 0.0-2.0 to adjust the optimal linear range. Parallel assays using B. fragilis, E. coli, E. cloacae, and E. Faecalis coating the ELISA plates were kept as controls to determine the specificity of the titers.
CEA and CA19-9 assay. The concentrations of carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) in the serum were assessed using electrochemiluminescence immunoassay (ECLIA) kits on a Roche E170 fully automatic electrochemistry luminescence immunity analyzer (Roche, Germany). Each test included a standard control (CV < 5%). The cut-off levels were 5.0 ng/ml and 35.0 U/ml for CEA and CA19-9, as recommended by the manufacturer.
Statistical analysis. All statistical analyses were carried out using the SPSS 16.0 statistical software package (SPSS Inc., Chicago, IL). The relationships between the specific anti-Fn-IgG or -IgA antibodies and the clinicopathologic features were analyzed by the Mann-Whitney U test. And the comparisons of specific anti-Fn-IgG or -IgA antibodies among different groups were assessed using the Kruskal-Wallis test. The efficacy of specific anti-Fn-IgG or -IgA antibodies for diagnosis was evaluated by the area under receiver operating characteristic (ROC) curve (AUC). Cut-off value for anti-Fn-IgG or -IgA antibodies was defined as the value with the maximization of the Yuden index. Furthermore, sensitivity (Sen), specificity (Spe), positive predictive value (PPV) and negative predictive value (NPV) were used to compare the efficiency of diagnosis among anti-Fn-IgG or -IgA antibodies, CEA and CA19-9. All statistical tests were two-sided, and P < 0.05 was considered statistically significant.