A simple urine test by 3D‐plus‐3D immunoassay guides precise in vitro cancer diagnosis

Abstract Although a variety of urinary cancer markers are available for in vitro diagnosis, inherent problems of urine environment—containing various inorganic/organic ions/molecules that vary in concentration over a 20‐fold range or more and significantly attenuate antibody avidity for markers—render conventional immunoassays unsuitable, remaining unresolved and a major challenge. Here we developed a 3D‐plus‐3D (3p3) immunoassay method, based on a single‐step urinary marker detection by 3D‐antibody probes, which are free of steric hindrance and capable of omnidirectional capture of markers in a 3D solution. The 3p3 immunoassay showed an excellent performance in the diagnosis of prostate cancer (PCa) through detecting PCa‐specific urinary engrailed‐2 protein, demonstrating 100% sensitivity and 100% specificity with the urine specimens of PCa‐related and other related disease patients and healthy individuals. This innovative approach holds a great potential in opening up a novel clinical route for precise in vitro cancer diagnosis and also pushing urine immunoassay closer to more widespread adoption.

disease-specific urinary markers and their cognate antibodies (capture probes) for immunoassay and thereby render conventional immunoassays not suitable any more, [21][22][23] requiring an advanced methodology that can resolve the urine-associated problems.
A variety of immunoassay methods have been introduced for in vitro diagnosis, including molecularly imprinted polymer (MIP)-based, electrochemiluminescence immunoassay-based (ECLIA), and singlemolecule array (SIMOA)-based detection methods. [24][25][26][27] However, the following main limitations remain unresolved yet. First, most immunoassays depend on two-dimensional (2D) mode detection, that is, antibody probes-that are already immobilized on a 2D solid surfacecapture disease markers in viscous serum, and thus the access of disease markers to the immobilized probes depends on only marker molecules' diffusional mobility. Furthermore, the immobilized probes are in general randomly oriented and thus sterically hindered on the 2D surface, often leading to poor diagnostic performance. 28 For example, immunoglobulin G (IgG) probes need to be immobilized with such an orientation that Fc region is always perpendicular to the surface; however, which is not actually implementable. Although various chemical methods have been introduced to resolve the random orientation problem, they have been unsuccessful yet except for several complicated and costly methods. 29 Finally, conventional immunoassays require tedious multiple steps for washing out nonspecifically, weakly bound false molecules and/or for adding prelabeled reporter probes, which significantly lowers diagnostic speed and accuracy and precludes convenient POCT. (Lateral flow assays are currently used in POCT but only for preliminary diagnostic assay due to low sensitivity and specificity.) Due to these limitations, conventional 2D-immunodetection cannot resolve the aforementioned urine-associated problems as proven in this study. Here, we developed an innovative 3p3 immunoassay method, which is based on target marker detection by 3D-IgG probes-which do not have a steric hindrance problem-in a 3D solution where both 3D-IgG probes and disease markers move freely through omnidirectional diffusion, enhancing target accessibility to probes and thus leading to sensitive detection of urinary disease markers. The 3D-IgG probe was constructed by attaching multiple IgG molecules on a single 3D scaffold-genetically modified capsid particle ($ 36 nm) of hepatitis B virus (HBV), consisting of 240 subunit proteins. That is, 240 copies of B domain of Staphylococcal protein A (SPA B ) was genetically presented on the surface of a single HBV capsid particle, followed by attaching multiple IgG probes to the SPA B -presenting capsid particle through specific and strong interaction between SPA B and Fc region of IgG. 30 Thereby, all antigen-binding domains of IgG probes are subject to become fully available in capturing disease markers without steric hindrance. Furthermore, the 3p3 immunoassay is performed in a single step in an assay solution, requiring neither washing steps nor prelabeled reporter probes. Consequently, this 3p3 immunoassay is likely to have significant advantages in resolving the aforementioned problems of 2D immunoassays and thus overcoming the hurdles in urine immunoassay.
Here we applied the 3p3 immunoassay to the detection of urinary engrailed-2 protein (EN2), a prostate cancer (PCa)-specific marker. PCa has recently become a major health threat to men, surpassing the incidence rate of lung cancer but can be hardly prevented due to numerous risk factors such as aging, family history of prostate-related diseases and BRCA 1 or 2 gene mutations, and so on. The immunoassay of serum PSA that is excessively released from the prostate tissue damaged by carcinogenesis into the bloodstream through sloppy vascular tissues has significantly increased the detection rate of PCa over the past 30 years. [31][32][33][34][35][36] However, PSA is prostate tissue-specific, not PCa-specific, meaning that the PSA level also increases in case of other prostate diseases such as infectious and chronic inflammation, benign prostatic hyperplasia (BPH), and other tissue-associated diseases near prostate. 37,38 Reportedly, only 20% of subjects with high PSA levels are finally diagnosed with PCa, [39][40][41][42] causing many unnecessary biopsies or treatment of individuals who do not have prostate malignancies, 43,44 and thus precise diagnosis of PCa is critical but yet remains unaccomplished. EN2 is a transcription factor involved in early embryo development of PCa and thus is only expressed in PCa cells and secreted into urine. 5,[45][46][47][48][49][50][51][52] The detection of urinary EN2 is currently based on ELISA, but the sensitivity and specificity are only 66%-80% and 80%-88%, respectively. 5,[45][46][47]  also with fluorescein isothiocyanate (FITC)-labeled anti-EN2 antibody (Cat. No. TA32493; Origene, Rockville, MD, USA), and the fluorescence from the cell cultures was analyzed using confocal fluorescence microscopy. It was found that PSA was expressed in both cancer and normal cell lines, and notably even a more amount of PSA was expressed in normal RWPE-1 cells than in cancerous DU-145 and PC-3 cells (Figure 1a,b), indicating that PSA is not PCa-specific at all. Western blot analysis of all cell lysates shows the same results as the confocal fluorescence microscopy ( Figure 1c). These results exactly correspond to the previous reports that PSA is expressed in normal epithelial cells lining prostate wall and readily secreted into a blood vessels or urethra, depending on age and prostate size. [53][54][55] The blood PSA level increases even under non-malignant conditions such as prostatitis, 43 and on the other hand, no PSA or a very little amount of PSA is often detected in the blood of PCa patients (Table S1). On the contrary to PSA, EN2 is expressed only in cancer cells (Figure 1d-f), indicating that EN2 is strictly PCa-specific as previously reported, 5,[45][46][47] and thus we confirm here that EN2 is a very promising biomarker for PCa diagnosis.

| Construction of 3D-IgG probes
The tandem repeat of SPA B that has strong and specific affinity for Fc region of IgG (reportedly, K d [dissociation constant] = 6.2 Â 10 À10 M) (named [SPA B ] 2 ) was genetically inserted into spike loop region of each subunit protein of HBV capsid (HBVC). Hexa-histidine tag (H 6 ) for both Ni +2 -affinity purification and Au +3 adsorption was also fused to the Nterminus of each subunit protein. Considering that HBVC is comprised of 240 subunits, 240 copies of (SPA B ) 2 were subject to be presented on the outermost surface of HBVC, and the SPA B -presenting HBVC-which was used as a 3D scaffold for constructing 3D-IgG probes-was synthesized in recombinant Escherichia coli and subsequently purified through Ni +2affinity chromatography ( Figure S1a (Figure 2a). Uric acid that lowers pH of urine affects both hydrogen bonds between and charge properties of amino acid residues and thereby weakens the interaction between protein markers and IgG probes. For example, at low urine pH (often below 6.0), histidine (pK R ≈ 6.0) charge is altered, and overall net charge of glutamate residues (pK R ≈ 4.3) could be less negative in urine than in serum. In addition, a large amount of urea not only denatures the conformation of IgG and protein markers but also interrupts hydrogen bonds. Therefore, this unique urine environment renders conventional 2D immunoassays (e.g. ELISA, ECLIA, etc.) not any more suitable  Figure 3a, a dark blue color appeared more rapidly as the EN2 concentration increases from 0.01 to 20 ng/ml. Also, it is notable that for the EN2-free healthy control, any discernable color change was not observed even at 80 min after the 3p3 immunoassay began. The analyses of TEM and energy-dispersive X-ray (EDX) spectroscopy confirm that the dark blue color comes from large Au particles, that is, the larger the size of Au particles is, the darker the blue color becomes (Figure S1e,f). Interestingly, light absorption spectra shows that as the color gradually darkens the maximum absorbance always occurs at a wavelength of 620 nm (λ max ) (Figure 3a (c) Time-course change of OD 620 measured for the optical assay signals of (a). (d) OD 620 measured at 80 min after the EN2 detection by 3p3 immunoassay of (a) begins. 3p3, 3D-plus-3D; EN2, engrailed-2 protein point, 80 min after the assay began. It seems that the difference in λ max in the EN2 and PSA detection arises in part from the difference in the content of lysine, which exerts a strong influence on λ max of AuNPs. Reportedly, when attached on protein surface, AuNPs are stabilized by the electrostatic interaction between negatively charged AuNP surface and protonated ε-amine group of lysine residues, leading to the red-shift of absorption spectra of AuNPs. [58][59][60] Thus, upon the close interaction of EN2 or PSA with AuNPs on the 3D-IgG probe-marker complexes, the lysine-rich EN2-having 22 lysine residues as DNA-binding motif 61 -is subject to causing more red-shift of λ max compared to PSA with 11 lysine residues.

| Superiority of 3p3 immunoassay over 2D immunoassay in urinary EN2 detection
The performance of 3p3 and 2D immunoassays was compared in the detection of EN2 (0.01-20 ng/ml) in serum and urine. The EN2 detection signals (OD 620 ) of 3p3 immunoassay were measured at postassay 80 min as described above, and a commercial ELISA kit for EN2 (ELISA-EN2)   Figure 4c shows that 3D-IgG probe has much higher avidity for both urinary and serum EN2 than 2D-IgG probe. In particular, it is worthy of noting that the EN2-binding avidity of both probes reduces when serum is switched with urine, as seen in the following K d values: 0.013 (serum EN2) and 0.025 nM (urinary EN2) for 3D-IgG probe and 0.260 (serum EN2) and 1.010 nM (urinary EN2) for 2D-IgG probe, indicating that the avidity reduction fold is much smaller in 3D-IgG probe (1.7-fold) than in 2D-IgG probe (5-fold). This is because of the aforementioned advantage of 3D-IgG probe (Figure 2b) over 2D-IgG probes-all IgG molecules in 3D-IgG probe bind to antigens without a steric hindrance, thus keeping high antigen-capturing capacity even under urine environment. Of course, in this context, 2D immunoassay (ELISA-EN2) shows a significant limitation in the urinary EN2 detection (Figure 4d).
2.4 | Performance of 3p3 immunoassay in PCa diagnosis using patient urine specimens 2.4.1 | PCa diagnosis: "3p3vs. 2D immunoassay" and "urinary EN2-vs. PSA detection" As per the aforementioned procedure of 3p3 immunoassay using anti-EN2 3D-IgG probe, PCa diagnosis was performed with 30 patients' and 30 healthy urine specimens (negative controls) ( Table 1), and only a small volume of urine (4 μl) was used in a 384-well plate. Optical signals appeared clearly from all the urine samples of 30 PCa patients, while any discernible signals were not observed for 30 healthy controls ( Figure S4a,b). Time-course absorbance (OD 620 ) change demonstrates that the assay signals between all PCa patients and healthy controls were apparently differentiated at 80 min after the diagnosis began ( Figure 5a). (Using a predetermined linear correlation between OD 620 and EN2 concentration [ Figure 3d], OD 620 at the 80 min was quantitatively converted to EN2 concentration for all the urine samples [- Table S2a,b].). Surprisingly, this simple urine test shows superior sensitivity and specificity-generation of no false-negative signals from all PCa patients (100% sensitivity) and no false-positive signals from all healthy controls (100% specificity). On the other hand, the PCa diagnosis by ELISA-EN2 using the same patient and healthy urine specimens shows seven false-negative signals (76.7% sensitivity) and three false-positive signals (90.0% specificity) (Figure 5b and Table S2c,d).
Furthermore, with the same patient and healthy urine specimens, PCa diagnosis was performed through the 3p3 immunoassay using anti-PSA 3D-IgG probe ( Figure 5c, Table S3a,b). Time-course absorbance (OD 585 ) signals were never clearly differentiated between PCa patients and healthy controls during the entire period of assay (80 min), which is because healthy urine also has a detectable amount of PSA (Table S3b).
The concentrations of EN2 and PSA in the urine samples of 30 PCa patients (Tables S2a and S3a)-which were all estimated through the 3p3 immunoassay-are plotted in Figure 5d, showing a very weak correlation (i.e., a Spearman's rank correlation coefficient of 0.176) between urinary EN2 and PSA and thus indicating again that PSA is not a reliable PCa marker. The PCa diagnosis was also performed using ELISA-PSA (Cat. No. MBS355412; MyBiosource, San Diego, CA, USA) with the same patient and healthy urine specimens, resulting in nine false-negative signals and two false-positive signals ( Figure S4c and Table S3c,d). Consequently, all the results above indicate that the 3p3 immunoassay can detect urinary EN2 with superior sensitivity and specificity and thereby enables precise PCa diagnosis. Furthermore, as shown in Figure S5, another important reason for the superiority of 3p3 immunoassay over ELISA in the urinary marker detection is that the use of much less amount of urine sample (4 μl) makes the 3p3 immunoassay less influenced by urea and uric acid and thus more sensitively detect the urinary marker compared to conventional 2D immunoassay like ELISA that uses in general much larger amount of sample (mostly 100 μl).

| Diagnostic selectivity: PCa vs. other related diseases
A precise diagnosis requires clear discrimination not only between disease and normal conditions but also between disease and related other diseases with similar symptoms. Reportedly, PCa diagnosis based on the detection of serum PSA frequently gives PCa-positive signals even for other related disorders such as BPH and BCa that have common PCa-like symptoms. Western blot analysis shows that EN2 is found only in the urine of PCa patient, not in BPH and BCa patients (Figure 6a), confirming again that EN2 is a PCa-specific biomarker. Here we applied the 3p3 immunoassay to detect urinary EN2 for the diagnosis of BPH and BCa using the urine samples collected T A B L E 1 Demographics of urine donors and summary of EN2 and PSA concentration in the donated urine samples, estimated through 3p3 immunoassay.

| DISCUSSION
Although urine is an excellent specimen for in vitro cancer diagnosis, the urine complexity and variability interfere with the interaction between disease-specific urinary markers and their cognate antibody Here, we developed an innovative assay methodology, that is, 3p3 immunoassay that is based on cancer marker capture by 3D-IgG probes in a 3D solution where both 3D-IgG probes and cancer markers move through omnidirectional diffusion, which remarkably enhances diagnostic sensitivity. The 3D-IgG probes were constructed by attaching multiple IgG molecules on a single 3D scaffoldgenetically modified HBV capsid particle. That is, the surface of HBV The LOD in the urinary EN2 detection by 3p3 immunoassay was about 10 pg EN2/ml urine, low enough to detect urinary EN2 of PCa patients.
In this study, it has been evidently proven that the 3p3 immunoassay shows superior sensitivity and specificity in the urinary EN2 detection-no false-negative signals for PCa patients (i.e., 100% sensitivity) and no false-positive signals for other prostate-related diseases (BPH and BCa) and healthy controls (i.e., 100% specificity), indicating a precise in vitro diagnosis of prostate cancer. On the contrary, the 2D immunoassay using commercially available ELISA kit shows far lower sensitivity and specificity. The 3p3 immunoassay also detected PSA in urine, but PSA was also found in the urine of some patients who suffer from other related disorders such as BPH and BCa-exactly corresponding to the previous findings-confirming that PSA is not a reliable marker in PCa diagnosis. Although EN2 is detected as a cancer marker for proof-of-concept in this study, the 3p3 immunoassay can

| Immunoblotting analysis
To determine whether EN2 or PSA is expressed in the aforemen-

| Estimation of dissociation constants (K d ) in the binding between anti-EN2 IgG probes and EN2 in urine and serum
The dissociation constants (K d ) of anti-EN2 3D-IgG probe and anti-

| Preparation of 3D-IgG probes
For the 3p3 immunoassay of EN2 in patient urine, anti-EN2 3D-IgG probes were prepared as shown in Figure S1. Assembly PCR with the appropriate primers was used to prepare two clones coding for NH 2 -