|Home | About | Journals | Submit | Contact Us | Français|
Detection of multiple cancer biomarker proteins in human serum and tissue at point-of-care is a viable approach for early cancer detection, but presents a major challenge to bioanalytical device development. This article reviews recent approaches developed in our laboratories combining nanoparticle decorated electrodes and multi-labeled secondary antibody labeled particles to achieve high sensitivity for the detection of cancer biomarker proteins. Two nanomaterials-based sensor platforms were used: (a) upright single wall carbon nanotube forests and (b) layers of densely packed 5 nm gold nanoparticles. Both platforms feature pendant carboxylate groups for easy attachment of enzymes or antibodies by amidization. In quality performance tests, the biocatalytic responses for determination of hydrogen peroxide of AuNP layers with attached horseradish peroxidase (HRP) on electrodes gave somewhat better detection limit and sensitivity than single wall carbon nanotube (SWNT) forest platforms with HRP attached. Evaluation of these sensors as platforms for sandwich immunoassays for cancer biomarker prostate specific antigen (PSA) in serum showed that both approaches gave accurate results for human serum samples from cancer patients. The best detection limit (0.5 pg mL−1) and sensitivity were obtained by combining the AuNP immunosensors with binding of 1 μm diameter magnetic particles decorated with secondary antibodies and 7500 HRP labels.
Early detection coupled with new targeted drug delivery therapies is one of the best immediate hopes to decrease deaths from cancer and improve therapeutic outcomes [1,2]. Biomarkers are molecules (proteins in this review) that increase in concentration during the onset of cancer, and they can be measured in serum and tissue for early cancer detection [3–6]. Accurate, sensitive measurement of proteins is of general importance to a broad range of applications in addition to cancer detection and monitoring, including systems biology, medical diagnostics for other diseases, proteomics, and drug development [7–9]. Measurement of collections of protein cancer biomarkers is the most promising for reliable early cancer detection [3–6]. For example, detecting 5 or more biomarkers for a given cancer by liquid chromatography-mass spectrometry (LC-MS) has provided nearly 100% reliable diagnostics [5,10–13] In contrast, single biomarkers often have low positive predictive value.
To achieve highly reliable cancer detection or prediction, it is thus necessary to measure a number of relevant biomarker proteins in serum or tissue lysates for each cancer. While modern LC-MS proteomics methods can achieve this, the technology is currently too expensive and technically complex for routine diagnostic point-of-care applications. However, array formats for protein detection employing optical or electrochemical detection have current promise to achieve relatively simple point-of-care devices at relatively low cost. These arrays can feature a series of analytical spots with each spot made specific for a different biomarker. Unlike gene arrays, accurate, simultaneous measurement of multiple proteins with arrays is at an early stage . Alternatively, bioelectronic and protein microarrays employing capture antibodies and enzyme-labeled secondary antibodies in sandwich assays (Scheme 1) hold the promise of high selectivity and sensitivity, ease of use, and reasonable cost. The bioelectronic approach involves relatively simple instrumentation with electronic detection, and is amenable to development into multiprotein immunoarrays .
The classical enzyme-linked immunosorbent assay (ELISA) based on optical absorbance is an important commercial bioanalysis method with detection limits (DL) as low as 3 pg/mL for protein biomarkers [17–19]. However, ELISA suffers limitations in analysis time, sensitivity for biomarkers whose normal range is in the low pg/mL range, relatively large sample size, and difficulty in measuring collections of proteins. Several commercial immunoassays provide very good DL for proteins but usually only determine a single protein per sample, although in several cases assays for 2 or 3 specific proteins are available. We have undertaken a program in our laboratory to develop general electrochemical immunosensor array platforms to measure collections of protein biomarkers simultaneously, without compromising analysis time, sensitivity, or sample size. Progress toward this goal, mainly in developing highly sensitive assay strategies, is the main topic of this review.
Heineman et al. pioneered electrochemical enzyme-linked immunoassays beginning in the 1980s . In their approach, antibody-antigen binding events and enzyme-substrate reactions are separated in space and time from detection of electroactive alkaline phosphatase products. Electroactive product is transported by a chromatographic or flow injection system to the detector. Excellent DLs in the pg mL−1 to ng mL−1 range have been obtained for small molecules and proteins depending on the design characteristics of the system used and the nature of the electrochemical detector [21,22]. For automation and miniaturization, the approach requires microfluidics and reagent reservoirs resulting in viable but somewhat complex systems, especially for multi-analyte determinations . However, such complexity may be unavoidable for fully automated systems designed for point-of-care applications.
Alternatively, significant efforts have been directed toward single probe, self-contained electrochemical immunosensors [24–28]. These approaches involve antibodies (Ab) attached to a sensor surface, so that antigen binding, enzyme reaction, and detection are all done on the same surface, reducing system complexity. The most selective and sensitive immunoassays employ 2 antibodies in a so-called sandwich format. In this scenario as we have used it, antigen from the sample binds to the capture antibody (Ab1) attached to the sensor electrode, then a horseradish peroxidase (HRP)-labeled secondary antibody (HRP-Ab2) binds to the antigen, similar to Scheme 1a. Using peroxide activated HRP, biocatalytic electrochemical reduction produces the signal via reduction of the activated enzyme . HRP-Ab2 is ideal for arrays since immobilization of the electroactive enzyme label on the electrode eliminates electrochemical crosstalk between array elements, which can occur when detecting a soluble electroactive product . Sensitivity and detection limits can be greatly improved by using multi-labeled nanostructured materials that have the secondary antibody attached [31,32]. Despite these advances, a general biocatalytic array platform for proteins has yet to be developed.
Nanoparticles and high surface area nanoparticle electrodes coupled with biomolecules that can capture biomarker proteins from complex samples like serum and tissue lysates offer unprecedented new opportunities to develop high sensitivity biosensors [32–33,34,35]. Here we review recent progress and strategies in our laboratories towards the development of multiprotein electrochemical immunoarrays for cancer biomarkers. In particular, this article addresses strategies to achieve high sensitivity and ultralow detection limits. We have used nanoparticles in two ways in these sensor strategies: (1) to provide highly conductive, high area surfaces, possibly with electrocatalytic capabilities, for attachment of a high surface concentration of capture antibodies; and (2) as detection particles carrying large cargoes of enzyme labels and several secondary antibodies for amplification of amperometric signals, a variation of an approach first reported by Munge et. al. . We began our work using pyrolytic graphite sensors surfaces with upright nano-constructs of single wall carbon nanotubes (SWNT) called SWNT forests. More recently we have considered alternatives involving gold nanoparticle electrodes with magnetic beads for the multilabel amplification. The sections below describe these approaches and several applications. For biosensors described below, full experimental details can be found in the cited references.
Carbon nanotubes (CNTs) are novel molecular nanowires with metallic or semi-conducting properties . CNTs can be obtained commercially as mixtures of metallic and semi-conducting nanotubes. In their pristine state, they consist solely of carbon atoms, and can be obtained in single wall (SWNTs) or nested, multiwalled (MWNTs) versions. Their unprecedented electrical properties have fostered their use in electrochemical biosensors in many applications [31,34,35,38–40 ].
Surface area of conductive SWNTs, estimated up to 1,600 m2 g−1, can provide a platform to attach high densities of capture antibodies to achieve high sensitivity. Also, ballistic electronic conduction along the length axis of SWNTs with current density approaching 109 A cm−2 can enhance sensor performance . To take advantage of these two nanotube properties, iron-assisted electrostatic assembly of upright, dense SWNT forests were utilized due to their superb mechanical stability . In this process, pristine SWNTs are carboxyl-functionalized and shortened by oxidation in 3:1 HNO3/H2SO4 under sonication. The resulting nanotube dispersions are filtered, washed, dried, and re-dispersed in slightly basic DMF by sonication. These dispersions give best results for forest fabrication when aged by standing at room temperature for one week. Unfortunately, the natural tendency of SWNT to slowly aggregate requires that this stock solution is used within 3 to 4 months for optimum performance [40,42].
Dense SWNT forests can be assembled onto any properly prepared surface from these DMF dispersions. Surfaces are prepared for forest assembly by first adsorbing a thin layer of Nafion. The substrates are then sequentially dipped into aqueous acidic (pH 1.7–1.9) FeCl3 to precipitate iron hydroxide onto the Nafion, washed with DMF, and then dipped into the SWNT-DMF dispersion for 30 min [32,40]. The left hand corner of Scheme 1 shows typical AFM images that confirm that SWNT forests made from aged dispersions provide nearly complete surface coverage .
Amperometric biocatalysis can be used to assess the electrochemical quality of SWNT forests, or indeed any conductive high surface area biosensor electrode platform. This involves linking peroxidase enzymes onto the carboxylate ends of the SWNT forests by amidization, and measuring sensitivity and detection limit for the biocatalytic determination of hydrogen peroxide. Peroxide converts iron heme peroxidase enzymes to a ferryloxy form that can be reduced at relatively low applied voltage. When amperometry was used to measure the response of HRP/SWNT forest electrode to injections of hydrogen peroxide in the nM range, detection limit of ~40 nM and sensitivity of 1.12 μA cm−2 μM−1 were obtained [40,42].
The studies described above confirmed the high efficiency of electronic communication between SWNT forests and enzymes . We then used the SWNT forests as platforms to fabricate sandwich immunoassays for proteins, using HRP labels associated with secondary antibodies to biocatalyze reduction of H2O2 and provide electrical signals. Our first biomarker target was the prostate cancer biomarker PSA. Two detection protocols were evaluated. In Scheme 1(a) conventional HRP-conjugated secondary antibodies (Ab2) were used for signal development. Scheme 1(b) illustrates the highest sensitivity protocol using carbon nanotubes (CNT) conjugated with HRP and secondary antibody (CNT-HRP-Ab2) at high HRP/Ab2 ratios . in this approach, independent CNT-HRP-Ab2 particles with 170 HRPs per 100 nm length of CNT43 were used to replace the conventional HRP-Ab2 in the sandwich immunoassay procedure. Using this approach, an ultra-low PSA detection limit (DL) of 4 pg mL−1 (mass DL 100 amol) was achieved . Figure 1 shows SWNT immunosensor responses for PSA standards in calf serum using the CNT-HRP-Ab2 conjugates. The procedure also involves the very important step of effective blocking of non-specific binding (NSB) of the labeled bioconjugates using protein-detergent blocking agents.
Controls showed that SWNT forests provided a significant gain in sensitivity over immunosensors without nanotubes. We also demonstrated excellent accuracy for PSA in human serum and tissue lysates from cancer patients . The approach is general and can be applied to any protein for which the appropriate antibodies are available.
While nanotube forest sensors gave excellent detection limits for PSA, problems to be overcome include the nanotube tendency to bundle, which in turn increases heterogeneity and compromises the long-term stability of SWNT dispersions. . While these limitations can be overcome, we have also evaluated alternatives featuring electrodes coated with dense layers of gold nanoparticles. We fabricated of nanoparticle electrodes by electrostatic layer-by-layer deposition . Using this technique, an ultrathin layer of cationic polyion was first adsorbed onto pyrolytic graphite electrode, followed by a layer of negatively charged gold nanoparticles (AuNPs), as shown on left hand side of Scheme 2. the goal here was to show that the AuNP platform was amenable to high sensitivity electrochemical immunosensing before inclusion into array formats.
To make immunosensors, we prepared electrodes from glutathione-decorated gold nanoparticles (GSH-AuNP) made by the reduction of a gold salt in the presence of glutathione solution (Scheme 3) . Particle size is kept very small by GHS binding on the gold nanoparticle surface as the particle grows.
The AuNPs particles were characterized by transmission electron microscopy and found to be 5±2 nm in diameter. Gold nanoparticle (AuNP) electrodes were then prepared by assembly of a layer of cation polymer poly(diallydimethylammonium chloride) (PDDA) on pyrolytic graphite, followed by a layer of the AuNPs . The underlayer of PDDA was about 0.5 nm. Subsequently, a dispersion of GSH-AuNPs was used to adsorb a layer of the negatively charged GSH-AuNPs onto the cationic surface. Quartz crystal microbalance (data not shown) and atomic force microscope studies (Figure 2) of these electrodes were consistent with a dense layer of the GSH-AuNPs on the sensor surface .
As with the SWNT forest platforms, we evaluated the bioelectronic sensing properties of the AuNP electrodes by covalently linking HRP via amidization to the glutathione carboxylate groups on the AuNP electrode surface. We then measured direct voltammetry of the HRP, as well as its biocatalytic activity for reduction of hydrogen peroxide. The resulting catalytic cyclic voltammograms were similar to those obtained on SWNT forest electrodes. The HRP-AuNP electrodes gave a sensitivity of 0.28 μA μM−1 hydrogen peroxide, an increase of ~40% compared to the 0.18 μA μM−1 found on our SWNT forest platform . The detection limit (DL) for hydrogen peroxide as 3 times the amperometric noise was 20 nM compared to 40 nM for SWNT forests .
We then used the AuNP electrodes to build immunosensors for the amperometric detection of PSA in undiluted calf serum. While several Ab2 amplification schemes were tested, we found the best combination of DL, sensitivity, and linearity was obtained by using a magnetic bead-Ab2-HRP bioconjugate in place of the traditional Ab2-HRP. The best results were obtained by using commercial carboxylate-coated iron oxide magnetic beads of ~1 μm diameter reacted with a high ratio of about HRP/Ab2 to give 7500 HRPs per bead. The magnetic beads also provide an easier synthesis of the bioconjugates since they can be separated from unreacted antibody and HRP by simple application of a magnet at the end of a test tube while washing away the unreacted biomolecules. These bioconjugate particles can be made and stored at 5 oC for up to a week with no noticeable decrease in performance. Very high sensitivity results for PSA in serum immunodetection on the AuNP platform using the magnetic bead-Ab2-HRP bioconjugate was obtained (Figure 3).
Sensitivity and detection limit were improved significantly compared to conventional Ab2-HRP, and sensitivity as the slope of the calibration curve from Figure 3b was 31.5 nA mL pg−1, ~1700 times higher than for PSA using the conventional Ab2-HRP. Detection limit as 3 times the average noise above zero PSA control was 0.5 pg mL−1 compared to 1 ng mL−1 when using conventional Ab2-HRP with the AuNP platform. These analytical figures of merit for PSA detection are significantly better than on SWNT electrodes, albeit the amplification step in this case was also improved compared to the SWNT study on PSA . The DL of 0.5 ng mL−1 represents a mass DL of 5 fg PSA. This DL for the AuNP sensors was eight fold better and sensitivity was 4-fold better than SWNT forest immunosensors featuring multiple labels on carbon nanotubes (CNT-HRP170- Ab2). Like the SWNT forest immunosensors, the AuNP-based immunosensors gave good correlations with a standard ELISA method for the determination of PSA in serum of human cancer patients and in lysates of cancer tissue .
The combination of a highly conductive, high surface area electrodes decorated with functional groups for attachment of capture antibodies with particles featuring multiple enzyme labels along with secondary antibodies provides high sensitivity and ultralow detection limits for cancer biomarker proteins in serum. Thus far, all of our evidence suggests that AuNP-decorated electrodes perform as good or better than the SWNT forests that we used earlier in bioelectronic immunosensor applications. Results summarized above along with potential ease of patterning suggest that AuNPs could prove an excellent choice for the automated production of nanostructured array electrodes, e.g. with ink-jet printing. Similar to nanotube forests, the AuNPs decorated with glutathione and arranged in dense monolayers and are ready for attachment of capture antibodies onto outward facing carboxylate groups. Future decoration of nanotube forests and the gold nanoparticles with polymer layers that will minimize non-specific binding of labeled particles could possibly impart additional electroactivity as well as suppression of non-specific binding. A typical multi-protein chip assay could be envisioned as suggested by Scheme 2. Progress toward these goals is currently underway in our laboratories.
Research described in this review was supported by PHS grant ES013557 from NIEHS/NIH. The authors thank Drs. Silvio Gutkind and Vyomesh Patel, National Institute of Dental and Craniofacial Research, NIH, and graduate students at the University of Connecticut named in joint publications for their diligent efforts without which progress in this research would not have been possible.
In honor of 60th birthday of Professor Lo Gorton
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.