Application of nonsense-mediated primer exclusion (NOPE) for preparation of unique molecular barcoded libraries

Design of NOPE oligo

NOPE oligo should be modified at its 3′-end, resulting in full inhibition of its enzymatic elongation. Modification must be stable and resistant to degradation or enzymatic removing after synthesis. Earlier we tried several variants of the 3′ end modification (Ref. [20] and unpublished data) and finally opted for black hole quencher 1 (BHQ1), which prevents elongation without noticeably influencing the oligos annealing properties.

NOPE oligo may overlap completely with the undesired primer or only partially with its 3′-end. The length and content of a “nonsense” template (5′-end of a NOPE oligo that serves for the elongation and neutralization of the unwanted primer) may also vary.

To get better understanding of how efficiency of a NOPE oligo depends on its structure, we tested different possibilities for 3′-overlap and 5′-extension. We designed four different NOPE oligos for neutralization of UMI-containing EGFR-ex20_NNN primer: NOPE-R1, NOPE-R2, NOPE-R3, and NOPE-R0. The latter oligo completely lacked the part for the “nonsense” elongation of the EGFR-ex20_NNN primer. Region of human EGFR gene sequence showing the location of UMI-containing primer and four different NOPE oligos is presented on the Fig. 4.

EGFR gene fragment was amplified as described on Fig. 2. EGFR-ex20_NNN primer neutralization step was performed using one of the NOPE oligos. To detect the minimal NOPE oligos concentration required for efficient elimination of the EGFR-ex20_NNN primer present at concentration 0.1 μM, we carried out a set of neutralisation PCRs with NOPE oligos concentrations ranging from 0 to 1.5 μM. After neutralization step, all reactions were divided into two halves (Fig. 3). One half was subsequently amplified (1st PCR) using step-out TruSeq-short primer and reverse primer EGFR-ex20_R1 and showed the overall efficiency of amplification (control). The other half was amplified using only one reverse EGFR-ex20_R1 primer, and reflected the completeness of EGFR-ex20_NNN elimination by a NOPE oligo (test).

Real-time PCR analysis performed with EGFR exon 20 specific TaqMan probe showed that primer exclusion by NOPE-R0 oligo which completely lacks the 5′-nonsense sequence was inefficient (Figs. 5a,b and 6). The difference in the number of amplification cycles for the PCR with and without TruSeq-short primer was only about 2–3 cycles. This demonstrated that effect of neutralisation based solely on the competition for the EGFR-ex20_NNN annealing between the template and blocking oligo is minimal.

At the same time, all three blocking oligos carrying the 5′-extension efficiently excluded EGFR-ex20_NNN primer from reaction (Fig. 5a,b), demonstrating robustness of the NOPE approach (see Fig. 5a,b). In the presence of TruSeq-short primer, inhibition of the target PCR amplification by NOPE-R1 and NOPE-R3 oligos was negligible, demonstrating that these oligos do not influence the efficiency of target PCR reaction (Fig. 5a, control, Fig. 5c, Fig. 6a). At the same time, NOPE-R2 notably inhibited target PCR, presumably interfering with the reaction in a sequence-specific manner. This indicates that newly designed NOPE oligo should be tested for side effects, as well as any new primer.

Simultaneously with the different NOPE oligos comparison we have determined their effective concentrations. The concentration of NOPE oligo should be sufficient to elongate nearly all molecules of the undesirable primer. At the same time, too high concentrations of the NOPE oligo could have an inhibiting effect on further amplification of target DNA. Increase of the NOPE oligos concentration from 0.1 μM to 0.5 μM consistently improved the efficiency of EGFR-ex20_NNN primer neutralisation (Fig. 5), while further increase of NOPE oligo concentration yielded poor returns.

To further verify universality of NOPE approach, we have additionally designed and tested NOPE-oligos for fragments of EGFR exons 18, 19, 21, and 22. All NOPE oligos demonstrated efficient exclusion of UMI-introducing primer at concentration of 0.5–1 μM (Fig. 5d,e), with minimal inhibition of target PCR (Fig. 5f), thus confirming the general robustness of NOPE method.

EGFR gene fragments library analysis by HTS

Designed pipeline was applied to the EGFR-ex20 gene fragment library preparation procedure for the four circulating cell-free DNA samples of patients with lung cancer together with the control for assessment of NOPE-R3 efficiency. NOPE-R3 oligowas used at concentration 0.8 μM for suppression of 0.1 μM EGFR-ex20_NNN primer. Bands of anticipated size were observed for all samples. No PCR product was detected in control lane (without TruSeq-short primer in the 1st PCR). The final difference between the number of amplification cycles after three rounds of amplification for target samples and the control was 20–22. This internal control confirmed that addition of NOPE oligo removes the UMI-containing primer from the reaction very effectively, if not completely. For comparison, we prepared the same EGFR-ex20 libraries using alternative method of EGFR-ex20_NNN primer elimination with E. coli Exonuclease I treatment followed by column purification (according to Ref. [3], with minor modifications, details on the method to be published elsewhere). 20 ng of circulating cell-free DNA was used per each sample. 1/5 of the linear PCR product was used in case of NOPE. 1/2 of the column-purified linear PCR product was used in case of ExoI treatment. The resulting libraries of EGFR gene fragment were analyzed using paired-end Hiseq2500 Illumina sequencing.

Data analysis was performed using our MAGERI software designed to analyze UMI-tagged targeted genome re-sequencing data [23]. In this back-to-back comparison starting with identical amounts of DNA, NOPE method outperformed ExoI method in respect of the number of successfully labeled and amplified molecules (Fig. 7a), even in spite of the lower amount of the linear PCR product used in the 1st PCR. This probably results from damage of template molecules by ExoI and material loss during purification step used after ExoI treatment.

At the same time, narrow shape of the UMI coverage peaks (Fig. 7b) confirms absence of notable inclusion of EGFR-ex20_NNN primer in PCR amplification after initial incorporation during linear PCR.

Next, we prepared EGFR-ex20 gene fragment libraries from the healthy PBMC DNA sample, standard reference genomic DNA (Tru-Q 7 1% Tier reference mutation panel, Horizon Dx, USA; Cat. ID HD734), and the same reference DNA mixed with PBMC DNA in 1:9 proportion. Libraries were prepared by ExoI method and NOPE method, sequenced on Illumina HiSeq2500 and analyzed using MAGERI software. Both methods successfully identified mutation EGFR_T790 M present at frequency 0.01 in Tru-Q 7 1% Tier reference mutation panel and at 0.001 frequency in reference panel mixed 1:9 with a healthy donor PBMC (Fig. 7c). These results demonstrate applicability and efficiency of NOPE method for targeted re-sequencing with UMI.

UMI-barcoded library preparation using NOPE oligo

Specimens collection and DNA extraction. Samples of healthy donors PMBC DNA, circulating cell-free DNA, and reference genomic DNA (Tru-Q 7 (1.3% Tier) Reference Standard, Horizon) were used in experiments. Circulating cell-free DNA was isolated from the blood plasma of patients with adenocarcinoma at Molecular Biology & Cytogenetics Lab, Russian Center for Roentgenology & Radiology (Moscow, Russian Federation). DNA extraction was performed with the use of QIAamp Circulating Nucleic Acid Kit (Qiagen).

Introduction of UMI via linear PCR. 25 ng of gDNA was used for the reaction. Besides DNA, linear PCR (50 μL) contained also: 1× Tersus Buffer (Evrogen), 200 μM of each dNTP, 5 U of Tersus polymerase (Evrogen) and 0.1 μM of the forward primer EGFR-ex20_NNN. Thermal cycling conditions were: 1 min at 95 °C; 10 cycles of 15 s at 95 °C, 20 s at 60 °C, 30 s at 72 °C; and finally 2 min at 72 °C. Nontemplate control was included.

Library amplification

1st PCR. After neutralization of UMI-containing primer, reactions were divided into two halves. For further libraries amplification, 0.2 μM of the forward primer TruSeq-short and 0.2 μM of the reverse primer EGFR-ex20_R1 were added to the positive control half of the reaction. Only reverse primer EGFR-ex20_R1 was added to the test half of the reaction (Fig. 3). The following cycle settings were used: 30 cycles of 15 s at 95 °C, 20 s at 62 °C and 30 s at 72 °C, final extension for 2 min at 72 °C.

2nd PCR. The second amplification was carried out using 1 μL of unpurified 1000-fold diluted product of the 1st PCR. 25 μL reaction contained: 1× Tersus Buffer, 200 μM of each dNTP, 5 U of Tersus polymerase, 0.2 μM of the forward primer TruSeq-long and 0.2 μM of the reverse primer EGFR-ex20_R2. The following cycle settings were used: denaturation, 1 min at 95 °C, followed by 11 cycles of 15 s at 95 °C, 20 s at 63 °C and 30 s at 72 °C. Final extension was 2 min at 72 °C.

3rd PCR. During the third amplification Illumina indexes were incorporated into DNA library. 3rd amplification was carried out using 2 μL of unpurified 1000-fold diluted product of the 2nd PCR. 25 μL reaction contained: 1× qPCRmix-HS (Evrogen), 0.2 μM of the Illumina-Dir primer, 0.2 μM of the TruSeq-Index 1 primer and 0.2 μM of the TaqMan probe. Real-time PCR was performed using ABI PRISM 7500 Sequence Detection System (Applied Biosystems). Reactions were set up in duplicate in two independent experiments. The following cycle settings were used: 3 min of polymerase activation at 95 °C, followed by 45 cycles of 15 s at 95 °C, 30 s at 63 °C and 1 min at 72 °C. Fluorescence signal was collected at every annealing step.

Testing of NOPE oligos for EGFR exons 18, 19, 21, 22 was performed under the same reaction conditions that have been described above for EGFR exon 20.