New, fast and cheap prediction tests for BRCA1 gene mutations identification in clinical samples

The need to make quick therapeutic decisions requires procedures that allow reliable results to be obtained quickly and unambiguously. Such an ideal genetic test for detecting predictive mutations, i.e. those that already have a defined therapeutic path, should be characterized above all by the reliability of their results, simplicity, a short implementation time, low cost, and the use of only a small amount of diagnostic material. Most of the tests available on the market are based on the polymerase chain reaction, but unfortunately, such tests are expensive, time-consuming and must be carried out by qualified personnel. There is still a need, then, for fast and cheap mutation identification tools. In this work, we have proved that SPR and QCM-D perfectly fill the gap in what is expected from diagnostic tests for gene mutations.

SPR analysis

Surface plasmon resonance is an optical effect that can be used to study the bonding of molecules in real time, without labeling. Using this technique, we can obtain information about the amount of the substance sought in the analyzed sample, and about the specificity and selectivity of the complex formed. In this work, SPR measurements were performed on samples with and without the BRCA1 mutation. The resulting sensorgrams are shown in Fig. 1.

Figure 1
figure 1

SPR sensorgrams recorded during the hybridization process with target DNA sequences: synthetic (dashed lines), obtained from patients with (green lines) and without (red lines) appropriate mutations (AE) and for wild type DNA sequence (F). Experimental conditions: 0.01 M PBS with the addition of 1 mM EDTA and 1 M NaCl; Cprobe DNA = 100 nM (contact time: 600 s; flow rate 5 μL min−1); CMCH = 1.0 μM (contact time: 200 s; flow rate 5 μL min−1); Csynthetic target DNA = 100 nM (0.15 ng μL−1) or 200-fold diluted clinical sample (CDNA in clinical samples = 0.13 ng μL−1), contact time: 300 s, dissociation time: 600 s, flow rate 5 μL min−1.

The procedure for identifying mutations consisted of three stages. In the first stage, as a result of the self-assembly process, single probe DNA fragments characteristic of each type of BRCA1 mutation were attached to the surface of the gold chip. The next step was to seal the receptor layer: the sealing thiol should effectively protect the probe DNA fragments against becoming adsorbed on the gold substrate, but should not in any way impede the hybridization process. For this purpose, mercaptothiol (MCH) was used; its length was consistent with the length of the carbon linker present in the probe DNA. The last step concerned the detection of the DNA sequence complementary to the probe DNA. The sensorgrams recorded at each of these stages contain two components that describe the association and dissociation processes. In the association phase, the DNA fragments present in the analyzed sample undergo a hybridization process with the probe DNA immobilized on the gold chip surface. The effectiveness of this process depends on the strength of the interactions between the target DNA and the probe DNA (their degree of complementarity). As a result of the formation of a DNA double helix, the signal increases until equilibrium is reached (if this is achieved), when the maximum number of analyte molecules has bound to the receptor layer. Replacing the analyte solution with the buffer results in a reverse process (the dissociation phase), in which the complex formed degrades. The intensity of the SPR signal then begins to decrease. The relationships describing the phases of association and dissociation should be exponential. Analyzing the obtained sensorgrams, we see that such an exponential character was obtained only when the analyte (target DNA) was fully complementary to the DNA probe (the curves marked in Fig. 1A–E with black dashed lines (synthetic sequences) and green lines (real samples)). Even a partial lack of complementarity with the target DNA (the red curves in Fig. 1A–E) resulted in a drastic change in the nature of the curve—the increase of the SPR signal was much smaller and slower. It should be emphasized that, even in the case of a point mutation (5370C > T, see in Fig. 1A), the SPR relationship obtained was significantly different from that recorded for the samples containing a wild type (WT) DNA sequence. This observation confirms the specificity of the developed test operation. Control experiments with the probe DNA corresponding sequence without mutation were also performed. The character of the relationships obtained, SPR response = f(t) was identical; a significant exponential increase in the SPR signal was only observed when there was a fully complementary DNA sequence (Fig. 1F). It should be noted that in the kinetic studies the synthetic target DNA concentration as well as the DNA concentration in clinical samples was very similar and equaled 0.15 and 0.13 ng μL−1, respectively. To obtain detailed information from the SPR measurements, a simple model of 1:1 binding \(\left( {{\text{probe DNA + target DNA}}\underset{{\text{kd, Kd}}}{\overset{{\text{ka, Ka}}}{\leftrightarrows}}{\text{ double stranded DNA}}} \right)\) was selected for fitting the experimental results to determine the values of the association (ka) and dissociation (kd) rate constants. The values of dissociation (Kd) and association (Ka) equilibrium constants were determined from following equations:

$${{K}}_{\text{d}}\text{=}\frac{{k}_{\text{d}}}{{k}_{\text{a}}}$$

(1)

$$ {{K}}_{\text{a}}\text{=}\frac{1}{{{K}}_{\text{d}}}$$

(2)

Table 3 presents the kinetic parameters (ka: association rate constant; kd: dissociation rate constant; Ka: association equilibrium constant; Kd: dissociation equilibrium constant) determined from SPR experiments. Evidently, the association rate constants for DNA duplex formation step were at least 103–104 times higher, when DNA fragments present in the analyzed solution were fully complementary to the oligonucleotide sequence immobilized on the gold surface than for non-fully-complementary DNA fragments. Whereas, the values of the association equilibrium constants clearly show that a strong affinity between the DNA fragments (probe DNA : target DNA) was observed only when the DNA strands were completely complementary to each other. Where there was a partial lack of complementarity, the values of the dissociation constants were at least 3–4 orders of magnitude lower. Taking into account the above information, it can be concluded that the threshold value of the association equilibrium constant (Ka) allowing for the classification of a sample to a group of samples containing a given mutation or to a group of mutation-free samples should be a maximum of an order of magnitude lower than the value determined for fully complementary target synthetic DNA.

Table 3 Kinetic parameters (ka: association rate constant; kd: dissociation rate constant; Ka: association equilibrium constant; Kd: dissociation equilibrium constant) obtained from SPR data for the hybridization process with target DNA sequences: synthetic, and obtained from patients with and without appropriate mutations.

QCM-D analysis

Another technique that allows for real-time control over the interactions between molecules without labeling is quartz crystal microbalance with dissipation. This technique, like SPR, makes it possible to obtain information about the specificity of the complex formed. A typical relationship recorded using this technique describes the changes in the frequency energy dissipation coefficient of the quartz crystal and as a function of time, see Fig. 1S in Supporting Information. The synthetic target DNA concentration as well as the DNA concentration in clinical samples applied during the hybridization process was very similar and equaled 0.15 and 0.13 ng μL−1, respectively. Determining the parameters of the analyte-receptor interaction on the basis of the frequency changes of the quartz crystal is possible only when the analyte is precisely defined (its structure, molecular weight, etc. are known). In the case of the clinical samples, the exact length of the DNA fragments present in the solution is unknown; such DNA fragments, in addition to the target sequence, is also enriched in other nucleotides that the synthetic target sequence (standard) does not contain. In such situation, the recognition of the target DNA with QCM-D detector should be based on the ΔD = f(Δf) relationship, which is a much better diagnostic parameter. In the case of the presence of a correct (without mutation) DNA sequence in the analyzed solution, the DNA strands formed as a result of the recognition process were only partially hybridized. The as-formed hybrid: probe DNA-target DNA is only fragmentarily a double helix. This situation must be reflected in the morphology of the formed DNA layer—in its organization, packing density and regularity. The degree to which the DNA fragments match each other in accordance with the complementarity principle is very well illustrated by the relationships ΔD = f(Δf), presented in Fig. 2. These relationships clearly show the organization of the layer, where the change in slope indicates changes in the regularity and packing density of the layer formed on the quartz crystal surface. As can be seen in Fig. 2A–E, only when the analyzed sample contained fragments that were fully complementary to the probe DNA fragments modifying the quartz crystal surface was the nature of these changes identical to those obtained for the synthetic complementary DNA strand. In the presence in the analyzed solution of not fully complementary DNA fragments, the slope of the relationship was significantly lower than the reference curve (obtained for the fully complementary synthetic target DNA). Unfortunately, gravimetric detection does not make it possible to detect a point mutation (5370C > T, see Fig. 2A). The relationships ΔD = f(Δf) recorded for the mutated and non-mutated samples were very similar to each other. Similar relationships were observed when a wild type DNA sequence of the BRCA1 gene was used as the probe DNA (see Fig. 2F). In this situation, agreement between the reference curve (the black curve in Fig. 2F) and the obtained result was found only when the DNA sequence from the patient without the mutated gene (the red curve in Fig. 2F) was present in the solution.

Figure 2
figure 2

Relationships ΔD = f(Δf) recorded during the hybridization process with target DNA sequences: synthetic mutated BRCA1 gene sequence (black lines), sequences obtained from patients with (green lines) and without (red lines) appropriate mutations (AE) and wild type DNA sequence (F). Experimental conditions: 0.01 M PBS with the addition of 1 mM EDTA and 1 M NaCl; Cprobe DNA = 100 nM (Vdroplet = 100 μL, t = 2 h); CMCH = 1.0 μM (Vdroplet = 100 μL, t = 1 h); Csynthetic target DNA = 100 nM (0.15 ng μL−1) or 200-fold diluted clinical sample. (CDNA in clinical samples = 0.13 ng μL−1).

AFM analysis

The morphology of the films deposited on the gold surface was analyzed based on AFM images obtained in an aqueous solution of 0.01 M PBS with the addition of 1 mM EDTA and 1 M NaCl. Sample images are presented in Fig. 3. After initial modification with the thiolated probe DNA (see Fig. 3A), irregularly shaped islands of the deposited material appeared on the gold surface. Our interpretation of these images assumes that they are formed by chemically adsorbed probe DNA molecules, which under these conditions most likely assume a random/coiled conformation. Similar topographic features were reported by Holmberg et al. for single-stranded DNA immobilized on a gold surface31. Sealing the probe DNA layer with MCH thiol (see Fig. 3B) did not significantly change the surface morphology; irregular islets are still visible. However, definite changes were observed after exposing the test (Au/probe DNA/MCH) for 2 h to the sample containing the synthetic target DNA fully complementary to the probe DNA (see Fig. 3C). It should be stressed that, for all the probe DNA sequences and fully complementary synthetic target DNA sequences tested, the AFM response was the same. In this case, stretched strands are visible on the surface, which may be the result of hybridization between the probe DNA and the target DNA, which are fully complementary. Hybridization may force randomly oriented probe DNA strands to unfold and facilitate base pairing, thereby causing the molecules to change their conformation from randomly coiled to partially or fully stretched. As a result, fragments of hybridized DNA are exposed on the surface of the film. This scenario seems to be reasonable if one analyzes the morphology of the surface film after exposure to samples that contained DNA fragments with BRCA1 mutation capable of hybridizing with the probe DNA dedicated to them. As can be seen in Fig. 3D–G, in these cases as well fragments of stretched DNA strands are visible on the surface of the film. Nevertheless, the frequency of their occurrence is lower than with the target DNA, and the observed fragments are usually shorter, which may be due to the fact that the DNA fragments present in the clinical samples differed in length from the probe DNA, making the base match only partial. On the other hand, after exposure to the samples containing DNA fragments without the BRCA1 mutation, which had a limited ability to hybridize with the probe DNA, stretched strands did not appear. The AFM images presented in Fig. 3H, I clearly show that in this case we are dealing with irregular aggregates. Obviously, the above analysis of the AFM images provides only a purely qualitative comparison. Nevertheless, it clearly shows a noticeable difference in the morphology of the surface films, depending on the degree of complementarity and the ability to hybridize individual samples to the probe DNA. Moreover, the AFM data seem to correlate with the results obtained from the SPR and QCM-D.

Figure 3
figure 3

AFM images recorded for an Au(111) substrate subsequently modified with the probe DNA (A), MCH (B) and further exposed for 2 h to the target DNA (synthetic fully complementary to the probe DNA) (C); and clinical samples containing: variant 5382insC (D); variant 3819del5GTAAA (E); variant 5370C > T (F); variant c.4035delA (G); wild type (H); wild type (I). The size of the images is 400 × 200  nm2.

Identification of BRCA1 gene mutations in clinical samples

To verify the functionality of the procedure developed for identifying BRCA1 gene mutations, studies were performed using clinical samples containing DNA. In this part of the studies, 22 different clinical samples collected from patients were used: 17 samples from patients with various BRCA1 gene mutation variants (5370C > T: × 1 sample; 5382insC: × 10 samples; c.4035delA: × 2 samples; 185delAG: × 1 sample and 3819del5GTAAA: × 3 samples) and 5 samples from patients without any BRCA1 mutation. The experiments were performed using BRCA1 mutated gene sequences as the probe DNA. As can be seen from Table 4, the genotests with SPR and QCM detection performed perfectly as predictive tests to confirm or exclude the presence of specific mutations in an analyzed sample. The compatibility of the results obtained by these procedures and the gold standard-PCR is very good; ambiguous results were obtained in just two cases. In addition, the testing method presented detects missense single-nucleotide (SNP) mutations (class I (C/T conversions)), which are challenging for various screening technologies, including HRM. In contrast, several-nucleotide deletions and insertions are very easily detected by screening methods such as HRM, as they clearly destabilize the DNA structure32. There are four classes of SNPs in the human genome33,34,35. In the current study, we were able to distinguish samples in which there were deletions and insertions on the basis of the two measurement methods (QCM-D, SPR). And in the case of SNP 5370C > T it was possible to distinguish between the two genotypes. Importantly, polymorphisms of this type account for 64% of all SNPs in the human genome33,34,35. The specificity of the developed BRCA1 mutation tests was determined on the basis of the negative samples analysis and was equalled 80% and 100% (4/5 and 5/5) for QCM-D and SPR tests, respectively. In turn, the selectivity calculated from the analysis the positive samples as a ratio of correct result/total no. of positive samples was equalled 94.1% and 100% (16/17 and 17/17) for QCM-D and SPR tests, respectively. The conventional methods applied in BRCA1 gene mutation identification are characterized by relatively broad range of the selectivity: 100% for enzymatic mutation detection (EMD), 50–96% for single-strand conformation polymorphism (SSCP), 88–91% for two-dimensional gene scanning (TDGS), 76% for conformation-sensitive gel electrophoresis (CSGE), 75% for protein truncation test (PTT), and 58% for micronucleus test (MNT)26. Whilst, the specificities of these methods are close to 100%, except for MNT. Taking into account the above information, it can be concluded that the developed SPR and QCM-D tests for BRCA1 mutation identification can be a very good alternative and can be successfully used in diagnostics.

Table 4 Results of genotyping by PCR; SPR and QCM-D performed using BRCA1 mutated gene sequences as the probe DNA.

Read more here: Source link