A novel hotspot specific isothermal amplification method for detection of the common PIK3CA p.H1047R breast cancer mutation

Breast cancer (BC) is a common cancer in women worldwide. Despite advances in treatment, up to 30% of women eventually relapse and die of metastatic breast cancer. Liquid biopsy analysis of circulating cell-free DNA fragments in the patients’ blood can monitor clonality and evolving mutations as a surrogate for tumour biopsy. Next generation sequencing platforms and digital droplet PCR can be used to profile circulating tumour DNA from liquid biopsies; however, they are expensive and time consuming for clinical use. Here, we report a novel strategy with proof-of-concept data that supports the usage of loop-mediated isothermal amplification (LAMP) to detect PIK3CA c.3140 A > G (H1047R), a prevalent BC missense mutation that is attributed to BC tumour growth. Allele-specific primers were designed and optimized to detect the p.H1047R variant following the USS-sbLAMP method. The assay was developed with synthetic DNA templates and validated with DNA from two breast cancer cell-lines and two patient tumour tissue samples through a qPCR instrument and finally piloted on an ISFET enabled microchip. This work sets a foundation for BC mutational profiling on a Lab-on-Chip device, to help the early detection of patient relapse and to monitor efficacy of systemic therapies for personalised cancer patient management.

indicated that the PIK3CA p.H1047R missense mutation causes a constitutively active form of the PI3K protein and is associated with poor prognosis and BC disease progression 8 . Detecting BC mutations from ctDNA has been successful with next-generation sequencing workflows like BEAMing [9][10][11] or digital droplet PCR (ddPCR) but cannot be easily translated into the clinic due to multifactorial reasons such as requirement of clear prognostic value, expensive costs and specialised labour per sample 5,12,13 .
To circumvent this requirement of expensive thermal-cycling machinery of PCR technology and sequencing workflows, isothermal amplification was investigated as an alternative method of detection. Loop-mediated isothermal amplification (LAMP) developed by Notomi et al. 14 uses 6 template specific primers to amplify DNA with high specificity and sensitivity at a constant temperature. LAMP has often been used in the study of pathogens and infectious diseases but has recently started to be used for mutational detection. A variation of the LAMP protocol using USS-sbLAMP primers 15 can help discriminate single base mutational detection allowing cancer variants to be investigated by LAMP, as shown in Fig. 1a. These properties of LAMP coupled with its higher amplification yield allows it to be utilised in tandem with standard microchip technology used for chemical sensing such as the Ion-Sensitive Field-Effect Transistors (ISFETs) 16 . ISFETs fabricated in unmodified CMOS technology have been already described to detect DNA targets via the pH change induced during positive amplification results [17][18][19][20][21][22][23] . ISFET-based DNA detection in combination with LAMP's ability to run isothermally allows the technology to be applied at the point-of-care as a Lab-on-Chip (LoC) device.
In this study, novel primer sets for the detection of PIK3CA p.H1047R variant in isothermal conditions have been developed, enabling fast, sensitive and specific detection of wild type (WT) and mutant (MT) p.H1047R alleles, while differentiating the detectable signals in an environment of interfering DNA templates. Time-to-results below 25 minutes with a limit of detection of 10 3 copies per reaction show the potential of adapting this assay to the clinic for rapid and sensitive profiling of common BC mutations. Besides lab-based validation, we demonstrate on-chip specific detection of both WT and MT alleles using an ISFET-based Lab-on-Chip platform. Using this fully-electronic chemical sensing technology in combination with the molecular methods developed here, we are in a position to create a lab-free, cost-effective sample-to-result system for routine monitoring of a panel of breast cancer mutations through liquid biopsy profiling of ctDNA, to be used as a diagnostic companion to stratify patient treatment. Future adoption would lead to significant advantages, such as offering continuous monitoring of tumour progression while assisting imaging intervention and tumour localisation, aiming to adaptively individualise breast cancer treatment for patient benefit. The technology would also be primed to monitor MRD post-surgery and help proactively search for therapy resistance mutations during a patients' treatment regime. The proposed workflow of the system and the Lab-on-Chip platform used to demonstrate feasibility of the SNV-based LAMP method are demonstrated in Fig. 1(b,c).

Results
SNV-specific LAMP for detection of p.H1047R. Specific amplification of the PIK3CA p.H1047R gene mutation in isothermal conditions was achieved following the USS-sbLAMP method described by Malpartida-Cardenas et al. 15 . This method consisted of (i) Single Nucleotide Variant (SNV) LAMP (sbLAMP) primers in charge of allele-specific amplification, and (ii) unmodified self-stabilising (USS) competitive primers responsible for preventing unspecific amplification. The sbLAMP primers included the LAMP priming regions F3, B3, LF, LB and the allele-specific primers sbFIP and sbBIP. These last two primers contained the regions F1c-F2 and B1c-B2, respectively. Considering the GC% content of the target sequence, different lengths of F1c and B1c priming regions were designed for optimal performance at 63 °C and screened experimentally; due to the higher GC% content at the region where B1c is located, the final design consisted of B1c being shorter than F1c to enhance allele-specificity without hampering the analytical sensitivity.
Specifically, the sbLAMP primer set targeting the mutant (MT) allele, named as sbLAMP MT, included sbFIP 17 _ MT and sbBIP 14 _MT primers. On the other hand, the sbLAMP primer set targeting the wild type (WT) allele, named as sbLAMP WT, consisted of sbFIP 19 _WT and sbBIP 14 _WT primers. USS primers were designed manually based on the aforementioned sbFIP and sbBIP primers. USS primers were incorporated in the reaction as follows: MT reaction included sbLAMP MT with USS WT , and WT reaction included sbLAMP MT with USS WT . Different lengths and concentrations were tested following the guidelines described 15 . The final primer sets consisted of USS MT FB2 21 _MT/BB2 24 _MT and USS WT FB2 19 _WT/BB1 23 _WT, both at a concentration of 3 µM per reaction. Primer sequences are shown in Table 1.
WT and MT templates were tested in both reactions to assess the analytical sensitivity and specificity of the designed USS-sbLAMP primer sets. Both templates were uniquely amplified within their corresponding specific reactions at concentrations below 1 × 10 6 copies/reaction with a limit of detection of 1 × 10 3 copies/reaction within 25 min. In Fig. 2a, standard curves are presented showing coefficient of determinations (R 2 ) of 0.992 and 0.972 for the specific amplification of the WT and MT template, respectively. These results showed the capability of the assay for sample quantification. Both WT and MT reactions presented a linear working range from 1 × 10 7 to 1 × 10 3 copies/reaction. A complex and more realistic genomic environment was created by incorporating Salmon testes sonicated DNA in the reaction mixture as background DNA. Results presented in Fig. 2b show standard curves of specific WT and MT reactions. Remarkably, the limit of detection (1 × 10 3 copies/reaction) and the linear working range (1 × 10 7 to 1 × 10 3 copies/reaction) were preserved in this scenario without any significant variation in the TTP values (correlation of 0.98 and 0.99 for each specific reaction, according to statistical linear regression), as shown in Table 2.
Accuracy of detection in a mixture of allele ratios. Allele discrimination within allelic ratios (100/0, 80/20, 50/50, 20/80, and 0/100, in percentages) in spiked mixed populations was studied. This scenario included Salmon testes sonicated background DNA with the results shown in Fig. 3. Both alleles were discriminated by combining the results of the WT and MT reactions, denoting the high specificity of the assay to distinguish each allele within mixed populations. Reactions targeting the allele that was not present in the mixed population, such as the ratios 100/0 and 0/100, did not show any amplification. The TTP values obtained clearly fitted in the corresponding standard curves previously built for each allele (Fig. 2b), and therefore, the capability of DNA quantification within a mixed population was demonstrated. Calculated concentrations in copies/reaction and the calculated ratios are presented in Fig. 3b. Overall, these results show the feasibility of the developed assay to estimate copy number and ratios with high accuracy.  TTT TGA TGA CA   H1047R_B3  TGG AAT CCA GAG TGA GCT   H1047R_LF  AAT ACT CCA AAG CCT CTT GCT C   H1047R_LB  TGG ATC TTC CAC ACA ATT AAA CAG C   H1047R_sbFIP 19 _WT  TGT GCA TCA TTC ATT TGT TGC ATA CAT TCG AAA GAC CCT   H1047R_sbBIP 14 _WT  ATC ATG GTG GCT GGC TCA GTT ATC TTT TCA GTT CAA TG   H1047R_sbFIP 17 _MT  CGT GCA TCA TTC ATT TGG CAT ACA TTC GAA AGA CCC T   H1047R_sbBIP 14 _MT  GTC ATG GTG GCT GGC TCA GTT ATC TTT TCA GTT CAA TG   H1047R_FB2 21 _MT  ACG TCA TGG TGG CTG GAC AAC AA   H1047R_BB2 24 _MT  GAC GTG CAT CAT TCA TTT GTT TCA TG   H1047R_FB2 19 _WT  ACA TCA TGG TGG CTG GAC AAC   H1047R_BB1 23 _WT  ATG TGC ATC ATT CAT TTG TTT CAT   Table 1. USS-sbLAMP primer sequences for allele-specific amplification of SNV p.H1047R.
www.nature.com/scientificreports www.nature.com/scientificreports/ patient tumour tissue cores ( Fig. 4d-f) were assessed as positive and negative controls for the p.H1047R mutation using the developed USS-sbLAMP assay. The T47D cell line DNA was analysed by Sanger sequencing (Fig. 4a), showing the high predominance of the nucleotide "G" over nucleotide "A". Amplification curves are shown in Fig. 4b,c, corroborating the specificity and accuracy of the developed assay to amplify the corresponding target; the T47D cell line was uniquely amplified with the MT reaction indicating that the presence of the WT allele was out of the limit of detection to be amplified by the WT reaction. The MCF7 cell line was uniquely amplified with the WT reaction as expected. The obtained TTP values corresponded to approximately 1.90 × 10 3 copies/reaction (MCF7 cell line) and 1.53 × 10 4 copies/reaction (T47D cell line) according to the standard curves previously built for the WT and MT allele respectively.   www.nature.com/scientificreports www.nature.com/scientificreports/ DNA extracted from two patient FFPE tissue cores were used as controls for wild type (healthy tissue core) and heterozygous mutant (tumour tissue core) status respectively. Presence or absence of the p.H1047R mutation was confirmed by ddPCR (Fig. 4d). FFPE tissue DNA negative for the p.H1047R mutation (left hand panels in Fig. 4d) showed an allele fraction of 0% c.3140 A > G MT allele. The heterozygous sample (right hand panels in Fig. 4d) showed an allele fraction of 38.2% c.3140 A > G MT allele and 61.8% c.3140 A WT allele. The WT sample (FFPE tissue DNA negative for p.H1047R) uniquely amplified with the WT reaction (Fig. 4e), whilst the heterozygous sample (FFPE tissue DNA positive for p.H1047R) amplified with both reactions obtaining similar TTP values (Fig. 4f). The TTP values obtained with the FFPE tissue DNA WT sample corresponded to approximately 1.70 × 10 2 copies/reaction. In the case of the heterozygous sample, TTP values obtained with the WT reaction corresponded to 6.06 × 10 4 copies/reaction of the WT allele and TTP values obtained with the MT reaction corresponded to 1.04 × 10 5 copies/reaction of the MT allele. These results reinforce the capability of the developed assay to discriminate the SNV p.H1047R at isothermal conditions with high accuracy in FFPE tumour tissues, even in the presence of mixed populations.

Adaptation of SNV-specific LAMP on a Lab-on-Chip platform. The developed WT and MT reactions
were tested using an ISFET-based Lab-on-Chip platform 16,18 with synthetic DNA harbouring either the WT or MT allele to prove the feasible translation of the lab-based assay at a point-of-care setting. The reaction conditions were modified such that pH changes could occur during DNA amplification. This consisted of changing the buffering conditions in LAMP to those described in pH-LAMP 20 . Each synthetic sample was tested with each of the WT and MT reactions independently for 25 min at a constant temperature of 63 °C (provided by a thermal controller). As shown in Fig. 5, each reaction amplified the sample harbouring the targeted allele, as previously shown with the qPCR instrument. The amplification curves obtained on-chip are shown in Fig. 5b, and the TTP values were in agreement with the values obtained with the qPCR instrument. Besides the correlation of the TTP values, pH measurements of the reactions carried out on-chip and with the bulky instrument were also similar, as described in Fig. 5a. Consequently, we demonstrate a proof-of-concept that the developed assays can be safely transferred to the LoC platform for lab-free SNV detection of the mutation p.H1047R related to BC (Fig. 5).

Discussion
In this study, we report two novel primer sets using an allele-specific LAMP assay to detect the missense p.H1047R, PIK3CA somatic mutation, which is a common driver mutation in breast cancer. PIK3CA is also one of the most frequently mutated genes in human breast cancer and in numerous other malignancies 24 due to the cellular importance of the PI3K signalling cascade. The potential clinical utility of being able to easily and rapidly determine PIK3CA mutations is yet to be fully explored. The SOLAR-1 trial recently presented that using combination therapy with Alpelisib for patients with an oncogenic PIK3CA mutation showed nearly double progression free survival 25 . Therefore, this assay could potentially be used to analyse diagnostic core biopsy tissue to identify a potential option for targeted monotherapy or combination therapies, with the recently approved small molecule inhibitor Alpelisib. Moreover, PIK3CA mutation is also associated with decreased effectivity of other common drugs such as Trastuzumab 26 . Therefore, detection of PIK3CA mutation could enable a precision medicine approach to determine the most appropriate therapy. These are two examples where a single mutation being identified can aid clinicians in their decision-making. Multiplexing this system where the LoC platform can detect panels of mutations would thus be a highly beneficial aid to clinicians in determining course of action as well as potentially monitoring MRD and relapse in BC patients.
The variant specific LAMP primer sets were designed to discriminate each allele by delaying or preventing unspecific amplification. At DNA concentrations below 1 × 10 6 copies/reaction of synthetic template, each www.nature.com/scientificreports www.nature.com/scientificreports/ primer set uniquely amplified the targeted allele. Analytical specificity and sensitivity were preserved in the presence of background Salmon DNA with a limit of detection of 1 × 10 3 copies/reaction, which equates to 0.003 pg/reaction for the synthetic template. Furthermore, allele discrimination within allelic ratios in spiked mixed populations was demonstrated enabling quantification with values within 8% from the reference ones. As a proof-of-concept, DNA samples from two breast cancer cell lines and FFPE tumour tissues were tested with the developed assay. The results obtained, reinforced the robustness, sensitivity and specificity of the developed  www.nature.com/scientificreports www.nature.com/scientificreports/ assay for the discrimination of p.H1047R in cell lines and patient tissue samples. In addition, we show allelic discrimination using a pH sensitive variation of the proposed assay with a CMOS microchip (Fig. 5) 16 . The microchip consisted of an array of ISFET sensors which detected both alleles with a positive change in pH of around 1-1.48 pH units. Overall, we show novel sbLAMP primer sets with the potential to discriminate BC tissue samples carrying the mutation p.H1047R as well as detection with CMOS-based ISFET sensors. This may provide the basis to develop an alternative to blood-based mutation testing by NGS and ddPCR technology in a quicker and more affordable way for use with plasma-derived cfDNA.
Other studies have reported the capability of detecting the mutation p.H1047R. Alvarez-Garcia et al. 27 genotyped PIK3CA status with an analytical sensitivity of 56 pg (9 genome equivalents) using qPCR technology and chemically modified primers. In a similar way, Wang et al. 28 used non-extendable primer blockers for allele-specific PCR detection of three mutations in cancer K-RAS, B-RAF, and EGFR with a limit of detection of a single copy. A different approach reported by Gyanchandani et al. 29 , has been used to demonstrate amplification of cfDNA from liquid biopsies linked to metastatic BC without compromising allelic balance. This step enriched the cfDNA sample, allowing ddPCR and sequencing to become more feasible due to the higher amount of DNA template required. Numerous NGS technologies such as BEAMing and Safe-SeqS have been used to detect known breast cancer mutations using ctDNA with sensitivity higher than 99% 30 . Our own studies have shown for primary patients that recurred, patient specific ctDNA profiling detected molecular relapse up to 2 years ahead of clinical relapse with 89% sensitivity and 100% specificity 31 . However, despite its high sensitivity and specificity, NGS platforms are still costly and require specialist time for data analysis and experimentation. As such, NGS can be useful to discover tumour mutations that can aid in treatment decisions and cancer classifications but this technique is not currently used routinely to monitor cancer response to treatment. On the other hand, our isothermal assay and Lab-on-Chip platform may allow for rapid, affordable and portable monitoring of individual breast cancer mutations without the need of thermal cycling, chemically modified primers or bulky and expensive equipment. Going forward, the next step will be to develop a multiplex assay to survey several common breast cancer mutations. The cost-effectiveness of such a test will allow breast cancer to be monitored routinely during patient's therapeutic journey.
LAMP has been previously used to detect mutations in tumour tissue such as for KRAS mutations in colorectal colon cancer 32 using LAMP in tandem with ligation substrates. Two studies specific to the detection of PIK3CA mutations published recently involved a colorimetric assay using strand displacement amplification 33 , and recombinase polymerase amplification (RPA), which is another isothermal amplification method 34 . These other studies highlight the demand for a cost-effective and rapid method for mutational tracking of DNA markers in cancer. In this paper, we demonstrate the genotyping of breast cancer variants with isothermal methods and CMOS technology, specifically detecting the PIK3CA p.H1047R variant. The combination of these principles indicates the feasibility of a label free Lab-on-Chip platform that can genotype breast cancer variants rapidly to assist tumour progression surveillance and individuality of BC in the clinic as an affordable alternative to NGS and ddPCR.

Materials and Methods
Samples and DNA extraction methods. Two plasmids, purchased from ThermoFisher Scientific (United States), were utilised as the synthetic DNA material for validation of the developed primer sets. Both plasmids consisted of PUC 18 cloning vectors containing a fragment of interest of 531 bp harbouring the SNV p.H1047R. The plasmid harbouring the WT allele p.H1047R is named as WT template, and the plasmid harbouring the MT allele p.H1047R is named as MT template. On the other hand, DNA was extracted from patient FFPE tissue cores using the Qiagen GeneRead Kit according to manufacturer's instructions and quantified using Qubit fluorometer. DNA was extracted from a healthy control tissue core as a WT control (H1047R negative) and from a tumour patient core (H1047R heterozygous sample). Patient samples used in this study were from our recent paper Coombes et al. 31  Droplet Digital PCR. Validation of the PIK3CA p.H1047R mutation was performed using a Bio-Rad QX200 droplet digital PCR system as described previously 36 . Primer sequences included: forward, 5′-AGAGGCTTTGGAGTATTTCATG-3′; reverse, 5′-TGCATGCTGTTTAATTGTGTG-3′; probe sequences were wild-type VIC-MGB 5′-CCACCATGATGTGCA-3′; and mutant FAM-MGB 5′-CCACCATGACGTGCA-3′. The optimum melting temperature for this assay was 62 °C. The assay was designed using OligoArchitect. For each assay 10 ng of FFPE tissue DNA was run alongside 10 ng of hgDNA (negative control), 10 ng TD47 cell line DNA (heterozygous control) and NTC (no template control). SNV-specific primer design for detecting p.H1047R. Consensus reference genomic sequence from human gene PIK3CA (Gene ID: 5290) was retrieved from National Centre for Biotechnology Information (NCBI) 37 and analysed for specific target regions among biological sequences using BLAST 38 . Following the guidelines provided by Malpartida-Cardenas et al. 15 for the design of USS-sbLAMP primers, two primer sets were designed targeting the SNV p.H1047R within the gene PIK3CA. Each primer set consisted of 8 primers, being 6 of them responsible for SNV-based loop mediated isothermal amplification (sbLAMP) including sbFIP and sbBIP which are in charge of the allele-specificity, and 2 unmodified self-stabilising (USS) primers, FB and www.nature.com/scientificreports www.nature.com/scientificreports/ BB, responsible for delaying or preventing unspecific sbLAMP amplification. The sbLAMP primers were designed using Primer Explorer V5 (Eiken Chemical Co. Ltd., Tokyo, Japan; http://primerexplorer.jp/lampv5e/index.html) and LAMP guidelines 14 to obtain F3, B3, F2 and B2 priming regions, and optimized manually to locate both loop primers LF and LB, and the SNV at the 5′ end of F1c and B1c within sbFIP and sbBIP. USS primers were designed following the method described by Malpartida-Cardenas et al. 15 based on the most optimal sbFIP and sbBIP tested experimentally. The primer set specifically detecting the WT allele p.1047 H, named as USS MT -sbLAMP WT (WT reaction), consisted of F1c 19 -B1c 14 + FB2 21 /BB2 24 at 3 µM; the primer set specifically detecting the MT allele, named as USS WT  Analytical sensitivity and specificity of USS-sbLAMP primer sets for detection of SNV p.H1047R. Analytical sensitivity was tested using ten-fold serial dilutions (1 × 10 7 , 1 × 10 6 , 1 × 10 5 , 1 × 10 4 and 1 × 10 3 copies/reaction which are equivalent to 31.356, 3.136, 0.314, 0.031 and 0.003 pg/reaction) of WT and MT templates, independently. Standard curves were obtained by plotting the TTP values with errors at one standard deviation against their corresponding DNA concentrations. Sensitivity including background DNA was also assessed by including in the reaction mixture Salmon testes sonicated DNA (Sigma-Aldrich) at 13 ng/reaction. Sensitivity of mixed populations was evaluated by spiking WT and MT templates at different ratios (100/0, 80/20, 50/50, 20/80 and 0/100, in percentage) to a final concentration of 1 × 10 7 copies/reaction.

SNV-specific LAMP on a microchip-based Lab-on-Chip platform.
Electrochemical sensing for the detection of the pH changes induced during pH-LAMP is facilitated using ISFETs fabricated in unmodified complementary metal-oxide-semiconductor (CMOS) technology. ISFETs detect changes in the concentration of hydrogen ions at the passivation layer of the microchip through the modulation of an induced voltage across that layer 22,39 . This way, owing to the compatibility of fabrication in standard CMOS technology allows for the mass manufacturing, low cost, and miniaturisation of sensors while ensuring non-optical and fully-electronic chemical (pH) detection.
The Lab-on-Chip platform used here comprises an array of 64 × 64 ISFET sensors fabricated in the AMS 0.35 µm CMOS process using silicon nitride Si N ( ) 3 4 as the passivation (sensing) layer. The total sensing area spans 0.56 mm 2 with an input-output pH sensitivity of 9.23 mV/pH. A detailed description of the circuit characteristics and additional specification is provided in N. Miscourides et al. 16 . Furthermore, a custom microfluidic manifold was added on top of the microchip to host the reaction mix during LAMP, laser cut from plastic acrylic sheets. Thermal control for DNA amplification is facilitated using an external Verity Thermal Cycler instrument.

Data availability
All data generated or analysed during this study are included in this published article. Data generated and analysed throughout the study can be made available from the corresponding author on reasonable request.