OPTIMIZATION OF DIAGNOSTIC ELISA-BASED TESTS FOR THE DETECTION OF AUTOANTIBODIES AGAINST TUMOR ANTIGENS IN HUMAN SERUM

Colorectal cancer is one of the most common cancer types worldwide and it continues to be a serious public health problem. Early detection and diagnosis are of great importance in cancer management. At present, diagnostic blood tests are based on the detection of tumor-associated markers such as carcinoembryonic antigen (CEA), the cancer antigen CA- for gastrointestinal cancer, CA- for breast cancer or CA for ovarian cancer. Th e lack of sensitivity and specifi city of these markers prevents their general use in cancer screening of an average risk population. Th erefore, new cancer biomarkers or better screening methods are necessary to improve the diagnostics of the disease. Th is study was directed to the optimization of a diagnostic, enzyme linked immunosorbent assay (ELISA) based test to identify and validate new serum markers, such as extracellular Protein Kinase A (ecPKA) and Nicotinamide N-Methyltransferase (NNMT). In this type of assay, the cancer antigens are quantifi ed indirectly by detecting the presence of auto-antibodies against tumor proteins in human serum. Th e result of the optimization and validation process was in the case of ecPKA a reproducible and stable assay. In case of NNMT the assay was probably not sensitive enough.


Introduction
Serum tumor markers have been the most widely used approach for cancer detection.Currently, the majority of the available tumor markers represent cancer antigens.For example, the PSA represents a marker for prostate cancer, the CEA for colorectal cancer, the cancer antigen CA- for breast cancer, the cancer antigen CA- for gastrointestinal cancer, the cancer antigen CA for the diagnosis of ovarian cancer, AFP is a liver cancer marker or human hCG, a breast cancer marker (, ).Quantifi cation of all markers is based on the antigen determination method and generally exhibits limited specifi city and sensitivity.Th erefore, new cancer biomarkers or better screening methods are necessary to improve the diagnostics of the disease.Th ere is increasing evidence that the immune system of patients with cancer responds to tumor antigens with the production of antibodies.Th e observed accumulation of such "auto-antibodies" in cancer patients suggests them to be of high diagnostic/prognostic value (, ).Th erefore, specifi cally developed auto-antibody detecting ELISAs could be used to quantify tumor markers, indirectly.As shown in Figure , the serum auto-antibodies that will be measured are bound between the coating antigen (tumor marker) and the enzyme labeled antibody, directed against human IgG.An increase in signal indicates the presence of auto-antibodies in the serum samples.It has been shown previously that the auto-antibody ELISAs exhibit enhanced sensitivity and specifi city for some of the tumor markers, compared to ELISAs detecting the tumor antigen.In general, the auto-antibody ELISA is superior to antigen-determining kits concerning rapidity and reproducibility, is easy to perform and represents a cost-saving method (, ).In this context, two interesting candidates for tumor markers are ecPKA and NNMT.In normal mammalian cells, the cAMP-dependent PKA is located strictly intracellular.In cancer cells of various cell types, however, PKA has been shown to be secreted into the medium.Th e speculation is that ecPKA excretion might elicit the induction of serum auto-antibodies against ecPKA, indicating that ecPKA might be a cancer antigen, and that the presence of such auto-antibodies could serve as a diagnostic marker.Th is has already led to the development of a novel enzyme immunoassay method that measures the level of IgG auto-antibodies generated against ecPKA (), which we intended to confirm and optimize with our antibodies (Figure ).
NNMT has recently attracted much interest in the fi eld of diff erent types of cancer.In humans, NNMT is predominantly expressed in the liver and is predicted to be located in the cytoplasm (, ).It could represent a new marker for colorectal cancer (CRC) that will further enhance detection of the disease and trigger a follow-up colonoscopy, as CRC patients showed elevated NNMT levels in serum.Furthermore, NNMT showed higher sensitivity than the established CRCtumor marker CEA, indicating that NNMT could be more suitable to discriminate between patients with CRC and apparently healthy individuals (, , , ).But further studies are needed to address the potential correlation of NNMT-serum levels with the presence of CRC.As NNMT was described to be secreted into serum, we tested whether it elicited the development of auto-antibodies as well.All these facts indicate the probable utility of this new auto-antibody assay as a cancer screening tool without the false positives often associated with conventional testing.It could also provide a novel technology for cancer detection ().

Materials and Methods
The serum samples used for the optimization of the detection method were obtained from subjects with CRC (n = , mean age  years ±  years).Tested cancer patients were of diff erent age and sex, with a wide range of stages of malignancy and diff erent therapies.Control serum was obtained from healthy volunteers (n = , mean age  years ±  years).Th e serum samples were kept at -°C until use and were thawed only once before use.Th e anti-ecPKA auto-antibodies and NNMT-levels were measured by solid phase ELI-SA.For this purpose, fl at bottom polystyrene -well

Results
Th e anti PKA auto-antibody assay.First, the coating antigen and HRP-labeled antibody concentrations of the published ecPKA auto-antibody assay were tested for their applicability, using positive (CRC) and negative (healthy) serum samples.Wells coated with antigen showed a mean absorbance of , ± , with coeffi cient of the variation (CV) of , for the positive samples and , ± ,, CV = , for the negative samples.When the plates were used uncoated (in absence of coating antigen), the mean value for the positive samples was , ± ,, CV = , and , ± ,, CV = , for the negative.In both cases the CV was below  and therefore acceptable.For purposes of optimization, four different dilutions of two diff erent HRP-labeled IgG antibody preparations, the whole molecule (H+L) and IgG-Fcγ (specifi c fragment), were tested (Jackson ImmunoReserach Laboratories Inc.,USA; Cat.Nr. -- and --).
When positive samples (patients with CRC) were tested, the .-fold dilution of the IgG-HRP (H+L) antibody showed the best performance with a mean OD value of , ± , (CV ,) which was signifi cantly higher compared to the IgG-HRP Fcγ specifi c fragment (OD = , ± ,, CV ,).In this case, the OD for the blank was , ± ,.
The NNMT auto-antibody assay.Analogue to the anti-ecPKA auto-antibody assay, the goal was to establish an ELISA that could detect potentially existing auto-antibodies against NNMT in the serum of patients with CRC.Several concentrations were tested for the STD, in this case the chicken α-human polyclonal NNMT IgY and four diff erent dilutions of anti-human IgG-HRP antibody.As human STD-preparations were not commercially available, the STD of different species (e.g.mouse, chicken) was used.Fortunately, the anti-human IgG exhibited the expected cross-reactivity towards these non human STD.These tests showed on the one hand whether the absorbance values were in the desired range (OD < ) and on the other hand which dilution of the enzyme conjugate was the most suitable one.STD concentration of  ng/cm  (and lower) in combination with the .-fold diluted IgG-HRP showed the best signal to noise (s/n) ratio and absorbance signals were in desired range (Figure ).Additionally, mouse α-human polyclonal NNMT was also tested as STD, but showed a lower s/n ratio indicating lower sensitivity.Th e mean s/n ratio of the mouse α-human polyclonal NNMT was . and the mean s/n ratio of the chicken α-human polyclonal NNMT IgY was . (determined with a low STD concentration of , ng/cm  ).Th e mouse universal negative control showed lower absorbance with a mean OD value of , ± ,; CV= ,.Serum samples (.-fold dilution) from a small set of patients with CRC showed the mean OD value of , ± ,, CV = , and the healthy controls , ± ,, CV = , which resulted in no signifi cant diff erences between the healthy controls and the patients with cancer.Th e OD value for the blank was , ± ,, CV = ,, indicating the possibility that auto-antibodies against NNMT don't exist or a too low sensitivity of the assay system.

Discussion
Th e diagnosis of cancer is one of the most important and critical steps in cancer management.It has been shown that in the early phases of carcinogenesis, an increase of ecPKA-and NNMT expression occurs and that the presence of auto-antibodies generated against ecPKA highly correlate with cancer.Th e anti-ecPKA auto-antibodies serum levels were shown to be markedly up-regulated in patients with cancer (, , ).For NNMT, auto-antibodies have not been analyzed so far.Nevertheless it is assumed that the detection of ecPKA-and potentially also of NNMT auto-antibodies could serve as a diagnostic method for cancers of various cell types.
Regarding auto-antibody measurement, the ecPKA and the NNMT protein could represent two new, interesting "indirect" biomarkers for early cancer detection.nize a single epitope and therefore, the use of MAbs results in a higher specifi city of the detecting method but at the costs of lower sensitivity.Polyclonal antibodies (PAb) show higher sensitivity because of the possibility that more antibodies bind to a single antigen molecule, but they have a higher risk for unspecifi c cross-reactivity ().Th ere is no empirically correct choice and therefore, all candidate antibodies/antigens have to be tested to determine the optimal components.In this study, PAbs were used for the ELISAs, because the adequate MAb was not commercially available at the time of ELISA optimization.Th erefore, a higher specifi city of the assays couldn't be established.Th e advantage of using PAbs for these ELISAs is that they can be generated in a variety of species (e.g.rabbit, goat, mouse, chicken, donkey, etc.) thus giving rise to more options in experimental design.Sometimes cross-reactivity is a desired phenomenon, as in our case, the anti-human IgG fortunately exhibited positive cross-reactivity towards the non human NNMT -STD.PAbs generally make the detection more robust and are more tolerant to small changes in the nature of the detected analyte.Sample type (e.g.serum or tissue extract) can also have a great eff ect on assay performance and on the choice of the assay components.Serum samples are complex mixtures and the autoantibodies which are to be measured can be present in various concentrations.Th ey can also be similar to other molecules which can lead to unwanted cross-reactivity with molecules that are not relevant for this test.All factors together can have an eff ect on the sensitivity and the specifi city of the method.Th e results of the NNMT auto-antibody ELISA suggested the absence of NNMT auto-antibodies in human serum.But the negative result could also have been due to a lack of assay sensitivity or due to the limited sample number.Th e assay parameters that could still be adapted for a better assay sensitivity are e.g. the coating antibody concentration, incubation time and sample dilution buff er.For the detection systems used in course of this study, a lower coating concentration was chosen.Th is should ensure that the antibody is the limiting factor.High concentrations of coating antigen can lead to less binding (the so-called "hook eff ect").Furthermore, the addition of detergents or salt to the reaction mixture should reduce low affi nity interactions indicating that the optimal composition and pH value of the fl uid phase has to be determined for better removal of unbound and excessive components.All these factors have an infl uence on the sensitivity as well as on the specifi city (, ).In course of this study several conditions were tested for the washing-and incubation buff ers.Lower concentration of the detergent Tween  and higher salt concentration was tested for the washing buff er.An increased detergent concentration was used in the case of the incubation buff er.After establishing the best (optimal) protocol, the respective buff ers were used for serum sample analysis.Still, the interassay absolute values were not completely identical, therefore a standard dilution series has to be included with every plate.Th e antibody/antigen and buff er optimization experiments performed in this pilot study, using a two and three dimensional serial dilution system (CTA), led to the development of an assay that was successful in distinguishing CRC patients from healthy volunteers using the anti-ecPKA auto-antibody ELISA.

Conclusion
Th e detection of auto-antibodies against tumor antigens in human serum certainly represents a promising approach in the fi eld of cancer diagnostics.In fact, each tumor antigen that is shed into the blood stream is capable of inducing autoantibody formation.Both developed tests successfully detect antibodies against the two tumor proteins that were used for the generation of the standard curve.In serum samples, however only PKA auto-antibodies were detectable.All together, both tests still need to be validated and have to be tested on a larger number of samples.In case of NNMT to fi nally prove the absence of NNMT auto-antibodies and in case of ecPKA to further confi rm the correlation with CRC.
DARIA ŠTEFATIĆ ET AL.: OPTIMIZATION OF DIAGNOSTIC ELISA  BASED TESTS FOR THE DETECTION OF AUTOANTIBODIES AGAINST TUMOR ANTIGENS IN HUMAN SERUM Immunolon- HBX microtiter plates were used (binding capacity - ng IgG/cm  ).Each single ELISA was repeated at least twice and standards, samples, blanks and/or controls were analyzed in duplicates.Th e microtiter plates were coated with  μg/ cm  of the purifi ed recombinant human PKA Cα subunit in  mm  coating buff er (phosphate buff ered saline (PBS).Th e plate was sealed with adhesive sealing fi lms for microplates (EXCEL Scientifi c, Inc.) and was incubated overnight, in the dark at room temperature (RT).Th e plates were than washed once with PKA-buff er I ( mmol/dm  Hepes, , NaCl,  mmol//dm  sucrose, plus , bovine serum albumin (BSA), pH ,), .-fold dilution) to determine the most suitable concentration.Mouse α-human polyclonal NNMT (Abnova Inc.; Cat.Nr.H-A)was also tested as a standard reagent in the optimization process but showed less sensitivity and specifi city compared to chicken α-human polyclonal NNMT IgY.
Detection system for anti-PKA auto-antibody ELI-SA.blocked for  h at RT with  μl of a four-fold dilution of BlockAce (AbD Serotec) and washed two times with PKA-buff er II ( mmol/dm  sodium citrate, , mol/ dm  NaCl, , Tween , pH ,-,).Th en  mm  of .-fold diluted serum samples were added and incubated for  h at °C.Sera were diluted in sample PKA-buffer III (PBS pH ,; , Tween  and plus , BSA).After three washes with PKA-buffer II,  mm  of .-fold diluted anti-human IgG-HRP (Jackson ImmunoReserach Laboratories Inc., USA; Cat.Nr. --) conjugate were added in buff er IV (PBS, , NaCl and plus  BSA), incubated for  h at RT. Th en the plate was washed fi ve times in washing PKA-buff er II, and incubated with  mm  of the prestained TMB PLUS substrate solution (BOTREND) for  minutes at RT in the dark.The enzyme reaction was stopped with  mm  of , mol/dm  HSO.Th e absorbance was read within  minutes at  nm with an ELISA reader (SPEKTRA MAX).fi city of the assay,  mm  of pure and ten-fold diluted mouse universal negative control (Dako Cytomation; Code Nr.N) were added.Anti-human IgG-HRP conjugate (Jackson ImmunoResearch Laboratories, Inc., USA; Cat.Nr. --) was (two-fold) serially diluted in buff er IV (PBS, , NaCl,  BSA; from .-fold dilution to Th e auto-antibody biomarker ELISA represents a good alternative to the antigen determining ELISA because it saves time and costs of the currently available antigen diagnostic kits.Critical factors in ELISA development are antibodies/antigens, liquid and solid phase and of course the samples.Th e choice of antibodies is particularly important.Monoclonal antibodies (MAb) recog-DARIA ŠTEFATIĆ ET AL.: OPTIMIZATION OF DIAGNOSTIC ELISA  BASED TESTS FOR THE DETECTION OF AUTOANTIBODIES AGAINST TUMOR ANTIGENS IN HUMAN SERUM