Detection of Tumor DNA in Human Plasma with a Functional PLL-Based Surface Layer and Plasmonic Biosensing

Abstract

Standard protocols for the analysis of circulating tumor DNA (ctDNA) include the isolation of DNA from the patient's plasma and its amplification and analysis in buffered solutions. The application of such protocols is hampered by several factors, including the complexity and time-constrained preanalytical procedures, risks for sample contamination, extended analysis time, and assay costs. A recently introduced nanoparticle-enhanced surface plasmon resonance imaging-based assay has been shown to simplify procedures for the direct detection of tumor DNA in the patient's plasma, greatly simplifying the cumbersome preanalytical phase. To further simplify the protocol, a new dual-functional low-fouling poly-l-lysine (PLL)-based surface layer has been introduced that is described herein. The new PLL-based layer includes a densely immobilized CEEEEE oligopeptide to create a charge-balanced system preventing the nonspecific adsorption of plasma components on the sensor surface. The layer also comprises sparsely attached peptide nucleic acid probes complementary to the sequence of circulating DNA, e.g., the analyte that has to be captured in the plasma from cancer patients. We thoroughly investigated the contribution of each component of the dual-functional polymer to the antifouling properties of the surface layer. The low-fouling property of the new surface layer allowed us to detect wild-type and KRAS p.G12D-mutated DNA in human plasma at the attomolar level (∼2.5 aM) and KRAS p.G13D-mutated tumor DNA in liquid biopsy from a cancer patient with almost no preanalytical treatment of the patient's plasma, no need to isolate DNA from plasma, and without PCR amplification of the target sequence.

Keywords: cancer diagnosis; peptide nucleic acids; plasmonics; poly-l-lysine; surface plasmon resonance.

Figure 1

Pictorial representation of the ultrasensitive detection of the p.G12D KRAS-mutated ctDNA sequence in human plasma samples using (a) the dual-functional PLL-based polymer and the nanoparticle-enhanced SPR-based sandwich assay. To simplify the representation, only specifically adsorbed ctDNA is shown. (b) We adsorbed the plasma sample on both PNA-WT and PNA-G12D probes (c) to recognize the mutated DNA sequence and discriminate it from wild-type DNA.

Figure 2

Pictorial representation of (a) PLL-mal(26%), (b) PNA probe, and (c) CEEEEE immobilization on the surface of the SPR gold sensor for the fabrication of the dual-functional PLL-based surface layer. The exposure of the PLL-mal(26%)-PNA-CEEEEE final surface to plasma samples allows both the hybridization of the circulating tumor DNA (ctDNA) target with the complementary PNA probe and the repulsion of plasma proteins.

Figure 3

Percent reflectivity (Δ%R) over time detected for PLL-mal(26%) deposition, PNA-WT or PNA-G12D immobilization (0.1 μM, 30 min), and CEEEEE (1.0 mM, 40 min) anchoring to the PLL-mal(26%)-PNA layer.

Figure 4

SPRI data for the adsorption of 10% diluted human plasma (30 min) on (a) PLL/PNA/CEEEEE, (b) PLL/PNA, (c) PLL-mal(26%), and (d) PLL-mal(26%)-PNA-CEEEEE layers. PNA-WT was used as the probe.

Figure 5

Representative time-dependent SPRI curves for the adsorption of AuNP@KRAS on (a) wild-type and (b) p.G12D-mutated gDNA in plasma previously adsorbed on surface-immobilized PNA-WT and PNA-G12D probes.

Figure 6

Δ%R_PNA-G12D/Δ%R_PNA-WT ratio values obtained from replicated experiments aimed at detecting wild-type (WT) and p.G12D gDNAs in 10% diluted human plasma samples (5 pg μL^–1). Ratios were obtained by considering Δ%R values after 1000 s of adsorption of AuNP@KRAS. The ratio considers SPRI responses (Δ%R) referred to PNA-G12D (Δ%R_PNA-G12D) and PNA-WT (Δ%R_PNA-WT) probes when the same plasma sample was detected. Wild-type (Δ%R_PNA-G12D/Δ%R_PNA-WT ratio population mean confidence interval (CI) at the 95% level = 0.86 ± 0.09, replicate measurements n = 8) and p.G12D (Δ%R_PNA-G12D/Δ%R_PNA-WT ratio population mean CI = 1.20 ± 0.16, replicate measurements n = 10) samples generated significantly different Δ%R_PNA-G12D/Δ%R_PNA-WT ratios (t-test, level 95%, two-tailed, p-value = 7.15 × 10^–5). A dotted line is shown to highlight better the values of the Δ%R_PNA-G12D/Δ%R_PNA-WT ratio below and beyond 1.

Figure 7

(a) SPRI calibration curve of different p.G12D-mutated gDNA concentrations spiked in 10% diluted plasma of an individual healthy donor (sample #4). The average Δ%R ratio generated by blank samples is shown (gray star, n = 5). (b) Four-parameter logistic function for experimental data fitting of the SPRI calibration curve is shown (gray line). p.G12D gDNA concentration values are reported on a log-scale axis. We added 2 to the actual concentration to include the negative control (p.G12D concentration = 0) in the fitting procedure, as described in ref (73). The same number was subtracted after the end of the process. The best fit (adj. R² = 0.987) was obtained using the equation y = d + (a – d)/(1 + ([p.G12D] + 2/c)∧b) with the following parameters: a = −51.3582; b = 1.10261; c = 0.04994; d = 1.83694. By the data fitting, the minimum detectable concentration (MDC = 0.58 pg μL^–1) and the reliable detection limit (RDL = 1.45 pg μL^–1) were estimated as DNA concentrations corresponding to the interpolated intersections of the lower asymptote of the upper 95% prediction limit with the four-parameter logistic fit curve and the lower 95% prediction limit, respectively.

Figure 8

Box plot of the Δ%R_PNA-G13D/Δ%R_PNA-WT ratio calculated after the adsorption of AuNP@KRAS. Plasma samples from the CRC patient with p.G13D KRAS-mutated ctDNA (sample pt#34) provided values greater than 1, whereas samples from the healthy donor (sample #4, wild-type cfDNA) provided significantly different Δ%R_PNA-G13D/Δ%R_PNA-WT ratios (t-test, level 95%, two-tailed, p-value = 1.5 × 10^–3).