Highly selective Cooperative Primers for multiplex detection of rare mutations

We describe a real time PCR-based technique capable of detecting and quantifying rare somatic mutations in circulating tumor DNA reference materials. Our approach utilizes previously described Cooperative Primers, structurally modified to exhibit high allele-specificity and one-copy target sensitivity. Cooperative Primers are bi-functional molecules, consisting of a high affinity probe fragment that guarantees sensitivity, and a covalently attached lower affinity primer providing specificity. Additional optimization of Cooperative Primer structure generated molecules capable of reliable detection of allele changes as small as a single nucleotide. These highly selective Cooperative Primers maintain excellent discrimination properties in rare mutant allele scenarios, in both monoplex and multiplex assays. With synthetic DNA samples, Cooperative Primers can detect as little as 100 copies of mutant template amongst 1 000 000 copies of wild-type template (minor allele fraction of 0.01 %). Multiplex Cooperative Primer assay was validated with cell-free DNA reference materials and consistently detected the lowest minor allele fraction available (0.1 %) for EGFR L858R, G719S and V769-D770insASV mutations, while simultaneously providing qualitative and quantitative assessment of cell-free DNA with integrated β-Actin assay. Easy to design, rapid and inexpensive, Cooperative Primer - based real time PCR assays are a promising tool for evaluation of cancer therapy response, occurrence of resistance mutations and relapse monitoring.


44
In the past few years, considerable efforts have been directed toward the development and 45 validation of methods for gene mutation profiling of liquid biopsy samples [1][2][3][4][5][6][7][8][9][10][11][12][13]. Liquid biopsy, 46 a simple and non-invasive alternative to surgical biopsies, is a source of a variety of biomarkers 47 including circulating tumor cells, circulating tumor DNA (ctDNA), circulating miRNAs, and 48 tumor-derived extracellular vesicles [14][15][16][17]. Circulating tumor DNA originates from apoptotic 49 and necrotic tumor cells that shed their DNA content into the circulatory system [3,[18][19][20][21]. The 50 analysis of ctDNA from blood offers a range of clinical application such as detecting cancer in 51 early stages, monitoring tumor burden in response to treatment, identifying occurrence of 52 resistance mutations and detecting disease relapse [22]. 53 As promising as the ctDNA analysis in liquid biopsy is, it faces numerous challenges. Since normal 54 apoptotic cells shed their DNA into the bloodstream as well, the ctDNA originating from tumor 55 cells is very rare, often representing as little as 0.01% of the total cfDNA. Additionally, the cell-56 free DNA (cfDNA) is extensively fragmented, averaging about 160 bp in length [23,24] and 57 exhibiting a modest half-life ranging from 16 minutes to 2.5 hours [25]. 58 Several techniques capable of overcoming these challenges have been developed to detect and 59 analyze ctDNA [26]. Next generation sequencing (NGS) is a very powerful large-scale DNA 60 sequencing technology allowing researchers to sequence and analyze data at a rate previously not 61 possible. It can identify a variety of cancer-associated alterations such as point mutations, (genome 62 wide) copy number variations, small insertions and deletions, large genome rearrangements and 63 methylation changes, thus making it an excellent approach for cancer screening and early diagnosis 64 [27,28]. However, NGS library preparation, sequencing and data analysis are labor-intensive 6 116 1). The capture sequence, due to its higher affinity, hybridizes to the template first. As a 117 consequence, the priming sequence is brought into vicinity of its target sequence, thus increasing 118 the local concentration of the primer by a factor of ~1 500 [44]. Finally, the priming sequence 119 hybridizes to the template, allowing extension from its' 3' end. As the priming sequence is 120 designed to bind upstream of the capture, the latter can double (upon labeling with a fluorophore 121 and a quencher) as a hydrolysis probe. Because the capture sequence must first bind for the short 122 primer to extend, theoretically 100 % of primer extensions result in probe cleavage, leading to high 123 signal-to-noise ratio [43]. linker is blocked on both ends by Spacer 18 molecules to prevent DNA polymerase from extending 128 through the linker. The capture sequence is labeled with a fluorophore and a quencher. The priming 129 sequence functions as a primer and can be extended by DNA polymerase from its 3' end. During 130 primer extension, the capture sequence is hydrolyzed, and the fluorophore is released. 131 The priming sequence exhibits robust performance in the PCR despite its short length, because the 132 hig-affinity capture sequence enhances primer binding stability. However, random binding of the 133 priming sequence in the absence of pre-requisite capture hybridization is highly unlikely. Low 134 priming fragment affinity combined with polymerase propagation inhibition due to the presence 135 of the non-extendable linker renders CoPrimers largely immune to primer-dimer formation and 136 propagation [44]. As such, CoPrimers are not only amenable to multiplexing, but also eradicate 137 false-positive results due to primer-dimer formation.
Owing to its abbreviated length, the primer sequence provides an opportunity for design of highly exhibit not only excellent allele differentiation properties, but also possess great target sensitivity, 153 while their resistance to primer dimer formation further decreases the likelihood of false positive 154 result. Their amenability to multiplexing in rare allele scenarios is an added bonus, as multiplexed 155 assays are less labour intensive and more cost effective.

Materials and Methods
Technologies (Petaluma, CA). All CoPrimers were reconstituted in 10mM TrisHCl, 1mM EDTA, 161 pH 8.0. The CoPrimers responsible for allele differentiation (forward or reverse) were labeled with 162 a fluorophore and a quencher; the non-differentiating CoPrimer in each primer pair remained 163 unlabeled. All CoPrimers used in this study are listed in Tables 1 and 2  length. All synthetic templates were reconstituted in 10mM TrisHCl, 1mM EDTA, pH 8.0.

Reference materials
The EGFR Multiplex cfDNA Reference Standard Set was purchased from Horizon 177 Discovery (Cambridge, UK). This set is supplied at 5 %, 1 %, 0.1 % and 0 % minor allele 178 frequencies and covers ten EGFR variants associated with responsiveness to EGFR tyrosine kinase 179 inhibitors. While derived from normal and cancer cell lines and fragmented to an average size of 180~160 bp, it closely resembles cfDNA extracted from human plasma. CoPrimers with varying lengths of capture and priming fragments ( number at a constant value of 1 000 000 copies/µL in each sample, the mutant template copy 263 number was either 0 copies/µL, 10 copies/µL, 100 copies/µL, 1 000 copies/µL, 10 000 copies/µL, 264 100 000 copies/µL or 1 000 000 copies/µL. A negative PCR reaction containing no EGFR template 265 was included as well. All reactions were performed in duplicate. The goal of this experiment was 266 to identify the optimal CoPrimer design possessing high sensitivity, maximum selectivity and also template copy number at a constant value of 1 000 000 copies/µL in each sample, the mutant 281 template concentration was either 0 copies/ µL, 10 copies/µL, 100 copies/µL, 1 000 copies/µL, 10 282 000 copies/µL, 100 000 copies/µL and 1 000 000 copies/µL. Real-time amplification curves 283 generated with ultimate B (bottom, A) and penultimate C (bottom, B) CoPrimer designs are shown.

300
The influence of gap size on CoPrimer selectivity was investigated as well. The best performing 301 CoPrimers from both ultimate and penultimate design groups were those lacking the gap between in the sample (Fig 3). Again, the adverse effect of the gap size on selectivity was less pronounced 306 in penultimate CoPrimer designs (Fig 3B) when compared to ultimate designs ( Fig 3A).

358
The PCR reactions were initiated with samples containing mixture of wild-type templates for 359 EGFR L858R locus, G719S locus, V769-D770insASV locus and β-actin locus at a constant value 360 of 1 000 000 copies/µL (each), while the EGFR L858R, G719S and V769-D770insASV mutant 361 template copy number was either 0 copies/ µL, 10 copies/µL, 100 copies/µL, 1 000 copies/µL, 10 362 000 copies/µL, 100 000 copies/µL and 1 000 000 copies/µL (each). A reaction containing no  The ultimate C quadruplex assay showed great sensitivity and specificity for all three mutant 384 targets ( Fig 5A). Each individual assay within the multplex exhibited linear amplification response 385 for the three reference samples with MAF varying between 5 % and 0.1 %. The wild-type-only control for the L858R mutation amplified with a significant Cq delay (15.7 cycles on average) 387 when compared to that of the sample with the lowest concentration of the mutant DNA (MAF of 388 0.1 %). The wild-type-only control for the V769-D770insASV sample did not amplify at all, thus 389 producing zero wild-type background (Fig 5A). The wild-type-only sample for G719S amplified 390 with a significant Cq delay (9.1 cycles on average). However, according to the quality control   (Table 3 and Figs 2 and 3), but was also evident from the performance of L858R, 465 G719S, V769-D770insASV and β-actin quadruplex assays (Fig 4). The linearity of the dependence 466 of the Cq value on the logarithm of the mutant allele concentration in the sample could be 467 significantly improved by decreasing the CoPrimer gap size (Fig 3). However, it appears that this impediment of CoPrimer binding and extension initiation is aiding in suppression of unwanted annealing and extension (mutant-specific CoPrimer annealing to and 478 amplifying the wild-type template), which results in enhanced CoPrimer specificity.

479
Ultimate C CoPrimer designs and Penultimate B CoPrimer designs (Table 1) possess gap, priming, 480 and capture lengths that allows them to perform efficiently and robustly in multiplex assays (Fig   481   4). Both L858R assay and V769-D770insASV assay (a 9 bp insertion) exhibited excellent linearity 482 down to MAF of 0.001 %. However, the G719S assay only exhibited good linearity down to MAF 483 of 0.01 %. The less-than excellent performance of the G719S assay is not unexpected. The L858R 484 mutation is a T>G substitution, resulting in G:A mismatch in the assay. The G719S mutation is a 485 G>A substitution, resulting in G:T mismatch in the assay (in our G719S assay, the differentiating

486
CoPrimer is the reverse one). The G:A mismatches near the 3'-end of the PCR primer are known 487 to reduce polymerase activity by as much as 10 000-fold whereas G:T be easier to design and should have greater rates of success than the ultimate designs.

503
Allele-specific PCR CoPrimer assays, described in this publication, exhibit specificity and 504 sensitivity that is typically required for rare allele analysis in liquid biopsy samples. Thanks to the 505 elucidation of optimal CoPrimer design characteristics, custom assays can be generated easily and 506 in a very short time. As CoPrimers are amenable to multiplexing, status of several mutations can 507 be monitored simultaneously in one well. A quantification control assay, such as β-actin, RNaseP 508 or GAPDH can be included in the multiplex assay as well. Unique CoPrimer structure guarantees 509 very low rate of false positives originating from primer-dimer formation, while their specificity 510 minimizes non-specific amplification due to mutant-specific CoPrimer mispriming and amplifying 511 the abundant wild-type DNA.