Developing nanotechnology-based protein detection

[5]

Figure 1. (click to enlarge) [5] Biobarcode assay detection formats: (a) Coloaded probe format. (b) Biotin-streptavidin format.

Recent clinical studies have indicated that earlier diagnosis of certain diseases will require more-sensitive detection of related protein markers that are present at concentrations below the detection limit of current diagnostic immunoassay technologies.^1,2 For example, a recent study examining cardiac troponin I (cTnI) levels, the gold standard test for diagnosing acute myocardial infarction, demonstrated that the presence of any detectable cTnI is associated with increased inpatient mortality.³ A more sensitive cTnI assay may enable a more rapid diagnosis of acute myocardial infarction and differentiation of unstable angina from less-dangerous forms of chest pain.² Prior studies also demonstrated that diagnosing prostate cancer recurrence after radical prostatectomy may be facilitated by detecting lower levels of prostate specific antigen (PSA).^4–6

With earlier diagnosis, clinicians could initiate appropriate therapies for patients earlier, potentially benefiting patient outcomes. Furthermore, the recent evaluation of subpicomolar levels of amyloid-B-derived diffusible ligands (ADDL) in cerebrospinal fluid (CSF) using an ultrasensitive protein detection assay points to the potential for developing new IVD tests for diseases such as Alzheimer's.⁷ Because of the low ADDL concentration in CSF (less than 1 pM), detecting ADDL in CSF using traditional immunoassay detection formats was not possible.⁷ This advance in assay sensitivity provides further opportunities to study new disease markers in CSF and human serum.

The Biobarcode Assay

A recent study reported the de­velopment of the biobarcode assay to detect protein targets at levels six orders of magnitude below conventional diagnostic immunoassays (see Figure 1a).⁸ This assay utilizes antibody-coated magnetic beads to capture and concentrate the protein targets. The captured protein targets are labeled with gold nanoparticle probes that are coloaded with target-specific secondary antibodies and single-stranded DNA (ssDNA) or double-stranded DNA (dsDNA) barcodes. The resulting magnetic bead-target-gold nanoparticle complexes are separated magnetically and washed to remove excess probes. The DNA barcodes are released from the complex and detected via hybridization to a surface-immobilized DNA probe and an oligonucleotide functionalized gold nanoparticle (see Figure 1a [5]).⁹

The gold particles are enlarged by depositing silver, and the light scattered from the particles is detected with the Verigene Reader optical detection system by Nanosphere Inc. (Northbrook, IL).¹⁰ The detection of a specific DNA barcode sequence indicates the presence of a specific protein. The increased sensitivity is derived partly from the release of multiple barcodes per captured protein target.⁸ The number of barcodes released depends on the size of the gold particle, with a 15-nm-diameter gold particle containing up to 100 DNA barcodes.¹¹ The sensitivity is further enhanced by detecting light scattered from silver-amplified gold particles, which enables detection of 10,000 barcode DNA copies per assay. A theoretical assay detection limit is approximately 100 captured protein targets.

Using antibodies directed against PSA in a model system, this study reported a detection limit of 180 PSA copies in 10 ㎕ (30 AM) of a buffered saline solution.⁸ The biobarcode assay has also been examined for detecting other protein targets including ADDL in CSF, a potentially soluble pathogenic marker for Alzheimer's disease.⁷ In a study of 30 patient samples, clinically relevant concentrations of ADDL were detected in CSF using the biobarcode assay. Furthermore, multiplexed assays have been developed for protein cancer markers, including PSA, human chorionic gonadotropin (a testicular cancer marker), and alpha-fetoprotein (a hepatocellular carcinoma marker).¹² The multiplexed assays use different DNA barcode sequences coloaded with marker-specific antibodies on each gold probe as indicators for different protein markers.

A more recent study described an alternative approach to labeling the protein targets after capture onto antibody-coated magnetic beads.¹³ In place of a coloaded gold particle, this approach utilized a biotinylated secondary antibody that is labeled with streptavidin-coated gold nanoparticles and biotinylated ssDNA barcodes (see Figure 1b [5]). Once the ssDNA barcodes are released from the complex, DNA is detected as described above for the coloaded probe format (see Figure 1 [5]). Using the biotin-streptavidin approach, an absolute detection limit of 0.01 pg/ml PSA (10 fg/ml, approximately a 300 aM concentration) was achieved with an excellent dose response in phosphate-buffered saline.¹³ Given antibodies of equivalent binding affinity, this detection format can be used in other clinically relevant protein biomarkers.

The previously reported biobarcode assay results for PSA used mock samples that consisted of free PSA spiked into phosphate-buffered saline solutions containing detergent and/or bovine serum albumin. A goal of the study­ was to measure the detection limit of the biotin-streptavidin biobarcode assay using total PSA (tPSA) spiked into human serum samples, and compare it with commercially available detection platforms. In patient sera with prostate cancer, approximately 90% of the PSA is associated with antichymotrypsin (ACT) to form a PSA-ACT complex, and the other 10% is free PSA.

To aid in standardizing test results, the World Health Organization (WHO; Geneva) developed a total PSA standard (NIBSC, code 96/670) that consists of 90% PSA-ACT complex and 10% free PSA to mimic the disease state. This standard was used for the study discussed above. Both forms of PSA are equally detectable with the selected PSA antibody pair used in the study.¹³ In addition, endogenous PSA levels were measured in female sera and in a set of serial samples from men that had undergone radical prostatectomy and/or salvage radiation therapy in order to evaluate the utility of the biobarcode assay using clinical samples.

Results and Discussion

[6]

Figure 2. (click to enlarge) [6] Biobarcode assay platform for research use. This instrument consists of an automated 96-well pipetting platform modified with a core module which enables magnetic bead separation, mixing, and heating.

PSA tests using the biobarcode assays were conducted with a research platform that enabled automated pipetting in a 96-well plate format, magnetic bead separation, heating, and sample mixing (see Figure 2 [6]). This platform performed all of the steps of the immunoassay with an output of barcode DNA. Previously developed PSA assays used female serum as a reference human serum for assay development since females lack a prostate gland.¹⁴ However, PSA has been detected in female sera using an assay with a limit of detection (LOD) of less than 10 pg/ml.¹⁵ Since the biobarcode assay exhibits an LOD for PSA below this range in buffer, several female sera were screened for PSA prior to performing LOD measurements in human serum (see Figure 3 [7]).

[7]

Figure 3. (click to enlarge) [7] Detection of PSA in female sera. A control experiment was performed (female serum + anti-PSA Mab) to demonstrate PSA immunoreactivity.

PSA was detected in two of the four female sera tested. To verify that the signal was PSA dependent, a competitive binding of PSA was performed by adding anti-PSA antibody to each female sera. The addition of anti-PSA antibody decreased the signal intensities of the two female sera with PSA to the baseline levels observed for the other two female sera. The two female sera that were unaffected by adding anti-PSA antibody were characterized as having undetectable PSA levels in the assay and used as reference serum. An LOD of 0.16 pg/ml was established by performing a serial four-fold dilution of tPSA in reference serum starting at 10 pg/ml (see Figure 4 [8]).

By comparison, these samples did not exhibit any dose response in PSA value when measured with the Access Immunoassay system by Beckman Coulter (Fullerton, CA), which reported an LOD of 8 pg/ml for PSA. A higher concentration PSA sample (40 pg/ml) was detected using Beckman's Access system, verifying functionality above the reported LOD of this platform. A recent study detailed the LODs of six different PSA assays as reported by the manufacturers, including Roche Diagnostics (Indianapolis), Abbott Laboratories (Abbott Park, IL), Bayer Diagnostics, Beckman, and Diagnostic Products Corp.¹⁶ The lowest reported LOD is 3 pg/ml of the WHO's 96/670 PSA standard using the E170 analyzer by Roche.

[8]

Figure 4. (click to enlarge) [8] Detection of total PSA in human serum using the Biobarcode assay. An LOD of 0.16 pg/ml was established based on detectable signal that is two standard deviations above the no target control (four replicates total for each concentration).

Five serum samples that registered undetectable PSA levels (reported as less than 10 pg/ml) using Bayer's Centaur assay were obtained and tested using the PSA biobarcode assay. Three of the five samples came from male patients who had undergone radical prosta­tectomy and were being monitored for biochemical PSA recurrence; the other two came from female serum donors. A calibration curve consisting of tPSA spiked into reference serum was generated to assign PSA values to the unknown samples following linear regression analysis of the calibration data (see Table I [9]).

[9]

Table I. (click to enlarge) [9] Sera tested for PSA using the Biobarcode assay and the Bayer Centaur assay. A control experiment was performed with the Biobarcode assay + anti-PSA Mab to demonstrate PSA immunoreactivity.

PSA was detected in four of the five samples tested, with values ranging from 2 to 14 pg/ml. The sample with the highest measured PSA level (14.5 pg/ml) came from a male patient who had undergone radical prostatectomy. As described above, adding anti-PSA antibody to the sera with PSA eliminates the signal confirming the test's specificity. These data also underscore the difficulty of detecting PSA reproducibly near the limit of detection using commercially available assays. With both the Beckman and Bayer analyzers, PSA samples around 10–15 pg/ml were undetectable, even though the reported LODs were 8 and 10 pg/ml, respectively.

Serum PSA is widely used to identify prostate cancer recurrence following radical prostatectomy. Recent studies have demonstrated that the time from surgery to biochemical recurrence and the rate of increase of PSA levels over time are both significant risk factors for prostate cancer specific mortality.⁵ Such risk factors were established using a clinical cutoff of more than 200 pg/ml for PSA.

A recent study compared the ability of an ultrasensitive PSA assay and a regular PSA assay to identify biochemical relapse in prostate cancer patients after radical prostatectomy during a five-year follow-up period.⁴ This study demonstrated that an ultrasensitive PSA assay provides much earlier detection of biochemical relapse and identifies three distinct groups of patients based on changes in PSA levels over time. The three distinct groups were classified as no change in PSA over time (no relapse), slow increases in PSA (slow rise), and fast increases in PSA (fast rise).

[10]

Figure 5. (click to enlarge) [10] Detection of PSA in clinical samples post radical prostatectomy. (a) No relapse patients. (b) Patients exhibiting biochemical recurrence. Red line indicates current clinical threshold for recurrence (200 pg/ml). (c) Patient treated with salvage radiation therapy following biochemical recurrence. Arrow indicates time that salvage radiation therapy (SRT) was performed.

The PSA levels of a subset of these previously tested samples were measured using the PSA Biobarcode assay (see Figure 5 [10]). The PSA levels of patients with no biochemical relapse remained constant at 0–4 pg/ml for PSA during a period of more than four years. These values agree with the prior testing done in the study discussed above. Other patients exhibited increases in PSA at levels below the 200 pg/ml clinical cutoff (see Figure 5b [10]). For example, patient 38's PSA level increased from 6 to 42 pg/ml during the first year following radical prostatectomy, eventually rising above 200 pg/ml more than two years following surgery.

For comparison, the PSA levels of a unique patient who was also treated with salvage radiation therapy were measured (see Figure 5c [10]). After radical prostatectomy, this patient exhibited a high PSA level on the initial measurement (17 pg/ml at day 49) when compared with the “no relapse” patients, followed by a fast rise in PSA to levels above the current 200 pg/ml clinical cutoff. Even though a large decrease in PSA was observed after salvage radiation therapy, the PSA level did not reach the level of the “no relapse” patients (less than 5 pg/ml) (see Figure 5a [10]). The patient exhibited a second fast rise in PSA, suggesting that the therapy may have been unsuccessful at eradicating all of the cancerous tissue. More importantly, the PSA level was measurable following radical prostatectomy at day 49 and salvage radiation therapy, even though it was below the 200 pg/ml clinical cutoff and the detection limit of most commercially available PSA assays.

Conclusion

These data demonstrate that an automated Biobarcode assay can detect tPSA in human serum samples at concentrations of 0.160 pg/ml (160 fg/ml) or greater. By comparison, the reported LODs of commercially available assays are more than 3 pg/ml. During the development of such assays, spiked serum samples and clinical samples should be tested against a comparison system since the reported LODs of different assays are determined using different methods. In this regard, a direct comparison of assays with the same set of samples enables an assay developer or end-user to determine accurately the assays' relative LODs.

Current studies are focused on detecting other known clinically relevant biomarkers and potential new biomarkers, including cardiovascular, respiratory, neurodegenerative, oncology, and renal disease states. Such efforts require screening and selection of antibody pairs with high target binding affinity and low cross reactivity to achieve the high sensitivity reported using PSA as a target. For any ultrasensitive protein detection format, the criteria of low antibody cross reactivity is crucial since assay sensitivity will be limited by background signal generated from antibody-antibody cross reactivity and cross reactivity of other components in the assay (e.g., gold particles or DNA barcodes with a magnetic bead surface). Other potential assay development hurdles include nonspecific signal generation from components in the sample matrix.

For example, human serum contains high concentrations of background proteins and antibodies, and other potential interferents such as heterophilic antibodies that have been shown to cause false positives in PSA and other assays. As assay sensitivity increases, such potential sources of background signal can become a bigger problem and must be accounted for through rigorous testing of a wide range of clinical samples and optimizing assay conditions. In the studies discussed in this article, a serum PSA assay detects total PSA at 0.160 pg/ml or greater, while the same assay in buffer detects 0.01 pg/ml. This LOD difference is attributed to the complex serum matrix. Further improvements in assay sensitivity will require optimization of assay conditions to minimize nonspecific binding of serum components that generate background signal or inhibit signal.

In addition, clinical studies must be conducted to validate that earlier biomarker detection will correlate with effective risk stratification and improved outcomes after earlier therapies. For example, one application of an ultrasensitive PSA assay is to monitor PSA levels in patients who have undergone radical prostatectomy or radiation therapy. This assay currently enables sample measurements at levels that are approximately 1000-fold below the current threshold for biochemical recurrence following radical prostatectomy. Prior clinical studies demonstrate that the time to prostate cancer specific mortality is linked to the rate at which PSA increases when measured above the current threshold for biochemical recurrence.⁵ The key question is whether measurements of PSA levels below this threshold using an ultrasensitive assay will enable earlier risk stratification.

	[5]	[5]
James J. Storhoff, PhD, is a senior scientist at Nanosphere Inc. (Northbrook, IL) He can be reached at jstorhoff@ nanosphere.us [11].	Michael J. Senical is an assay development scientist at Nanosphere Inc. He can be reached at msenical@ nanosphere.us [12].	William H. Cork is the chief technology officer and vice president of research and development at Nanosphere Inc. He can be reached at wcork@ nanosphere.us [13].