2.1 Samples
The pfhrp2 MinION assay was developed using well-characterized reference isolates cultured at the US Centers for Disease Control and Prevention (CDC). 7G8 (Brazil), 3D7 (suspected origin Africa), HB3 (Honduras), D6 (Sierra Leone), and FC27 (Papua New Guinea), which express pfhrp2, were used as proxies for positive samples, and Dd2 (Indo-China), which lacks pfhrp2, was used as a negative control.
The assay was evaluated using three additional sample sets. The first set was a collection of cultured control isolates from the Malaria Research and Reference Reagent Resource Center (MR4) (Wu, et al., 2001) including 7G8, 3D7, HB3, FC27, and Dd2. The MR4 set was utilized for final quality control experiments prior to sequencing field samples. Remaining clinical isolates from Sub-Saharan Africa therapeutic efficacy studies (TES) that were used for opportunistic pfhrp2 surveillance previously (McCaffery, et al., 2021; Rogier, et al., 2022) were also used. Additionally, numerous domestic and international clinical isolates were utilized. Domestic samples represent suspected cases of imported malaria from travelers seeking care at a U.S. medical facility. Whole blood samples were obtained from U.S. domestic, imported malaria cases from the CDC domestic malaria surveillance network and tested in accordance with protocol 2017 − 309 approved by the Office of the Associate Director of Science, Center for Global Health, Centers for Disease Control and Prevention as a surveillance activity. Informed consent was obtained from all participants. These samples were previously used to evaluate the one-step pfhrp2 PCR protocol (Jones, et al., 2020). A subset of the clinical samples was selected to perform PacBio sequencing against which to compare the resulting pfhrp2 types of this assay. This subset of field samples will be referred to as the PacBio set from this point forward. See Supplementary Table 1 for field sample metadata.
2.2 One-step PCR for pfhrp2
Here, we utilized the one-step PCR protocol for pfhrp2 (Jones, et al., 2020). A 50 µL reaction was prepared consisting of 0.126 µM each of the one-step pfhrp2 forward and reverse primers (Supplementary Table 2), 1 x Q5 reaction buffer (New England Biolabs, MA, USA), 0.2 mM dNTPs (New England Biolabs, MA, USA), 0.02 U/µL Q5 high-fidelity polymerase (New England Biolabs, MA, USA), and 5 µL of DNA template. Each reaction batch included a positive control (7G8 or 3D7), negative control (Dd2), and nuclease-free water negative control. PCR conditions were an initial denaturation at 98°C for 3 minutes (mins), a cycle of (98°C for 30 seconds (s), 60°C for 90 s, and 68° for 2 mins) repeated 30 times, and a final elongation at 68°C for 5 mins. The PCR product was visualized on a 1.0% TBE gel.
2.3 Barcoding primer design
We designed barcoding primers based on the PCR Barcoding (96) Amplicons (PBC096) kit and protocol from Oxford Nanopore Technologies (ONT) (Oxford, UK), customized for pfhrp2 (Supplementary Table 2). Primers for barcode sequences 1–60 were assessed in silico for self- and cross-dimers. Twenty-eight barcoding primers were suitable (Supplementary Table 2). We utilized this target-specific barcoding primer method instead of ONT’s formal PBC096 protocol due to pfhrp2-specific issues we encountered with the kit (see Supplementary Information, Supplementary Figs. 1 and 2).
2.4 Barcoding PCR
Barcodes were assigned to samples such that no single sequencing pool would share two samples with the same barcode. The barcoding reaction was like the one-step pfhrp2 PCR reaction, except each sample received 0.126 µM each of the forward and reverse barcoding primer assigned to it and 10 µL of pfhrp2 amplicon from the one-step reaction were added. PCR conditions were an initial denaturation at 98°C for 3 mins, a cycle of (98°C for 30 s, 69°C for 90 s, 68°C for 2 mins) repeated 30 times, and a final elongation at 68°C for 5 mins. Each reaction batch included the positive control (7G8 or 3D7), negative control (Dd2), and nuclease-free water negative control from the previous reaction. The PCR product was visualized on a 1.0% TBE gel.
2.5 Sample pooling and library preparation
Samples were quantified on a Qubit 4 Fluorometer with the dsDNA High Sensitivity assay (Thermo Fisher Scientific, MA, USA) following the barcoding PCR and prior to pooling, using 1 µL of product. These measurements were used for experiments with normalized pools (Supplementary Information).
Pools were then purified using AMPure XP beads (Beckman Coulter, IN, USA). The beads were added to each pool at a ratio of 0.5. Library preparation for MinION sequencing was performed according to the PCR Barcoding amplicons (96) ONT protocol section “DNA repair and end-prep” followed by “Adapter ligation and sequencing cleanup” with reagents from the ONT ligation sequencing kit (SQK-LSK109) (Oxford, UK), NEBNext FFPE DNA Repair Mix (New England Biolabs, MA, USA), NEBNext Ultra II End repair / dA-tailing Module (New England Biolabs, MA, USA), and NEBNext Quick Ligation Module (New England Biolabs, MA, USA). One microliter from the final eluate of each of these two final steps was taken for quantitation on the Qubit (Thermo Fisher Scientific, MA, USA).
2.6 MinION Sequencing
Sequencing was performed on MinION Mk1B and Mk1C devices on either standard or Flongle R.9 flow cells (Oxford Nanopore Technologies, Oxford, UK). For standard flow cells the Flow Cell Priming Kit was used, and for Flongle flow cells the Flongle Sequencing Expansion Kit was used. Flow cell priming and loading was performed according to manufacturer instructions. In ONT’s MinKNOW software, barcoding was turned on, the kit was set to EXP-PBC096, and the minimum barcoding quality was set to 60. The minimum read quality score was set to 7, and no minimum read length filter was set. Information on sequencing run lengths can be found in the Supplementary Information (Supplementary Table 3).
2.7 PacBio sequencing
Amplicons from the one-step pfhrp2 reaction (section 2.2) were purified with AMPure XP beads (Beckman Coulter, IN, USA). Libraries were prepared using the 5kb PacBio protocol with the DNA preparation kit (Pacific Biosciences, CA, USA). Samples were pooled together with 100 ng per sample and 8 samples per pool. Finished libraries were bound to P6 polymerase and sequenced on the PacBio RSII with C4 chemistry with one pool per SMRTcell.
2.8 Analytical pipeline
Quality filtered reads were aligned to exon 1 and the repeat region of exon 2 of the pfhrp2 reference sequence (PF3D7_0831800) with minimap2 (v. 2.21-r1071) using its “map-ont” option (Li, 2018; Li, 2021). The SAM file was sorted and converted to a BAM file with samtools sort (v. 1.13), and the BAM file was converted to a FASTQ file using samtools fastq (Li, et al., 2009). Mapping statistics were calculated using Qualimap (v. 2.2.2) (Okonechnikov, et al., 2016) and compiled into reports with MultiQC (v. 1.10.1) (Ewels, et al., 2016). The FASTQ containing successfully aligned sequencing reads was then used as input for de novo assembly with canu (v. 2.1.1) (Koren, et al., 2017) using its “-nanopore” option and “genomeSize” 1.3 kb. We used aligned reads for assembly to eliminate the inclusion of off target reads. De novo assembly was also performed in Geneious 2022.0.2. The canu and Geneious contigs were all converted to consensus sequences, visually explored, and manually edited to fit in the correct translation frame if necessary, based on AA marker sequences (Table 1, Supplementary Information). Up to ten contig consensus sequences per sample were curated based on their having > 10 reads and few low-quality bases. The nucleotide sequences were then translated to AA sequences and exported to FASTA format. Repeat type counts were calculated by a Python (v. 3.8.5) script (Data and Code Availability). Visualizations of repeat-type counts were created using the Python library seaborn (v. 0.11.2) (Waskom, 2021). For PacBio-sequenced data, an analogous analytical pipeline was used, with minimap2’s “map-pb” option and otherwise identical treatment.
Table 1
Amino acid sequence markers for pfhrp2. These markers were used as guide for correctly framing the pfhrp2 sequences for translation. Not all repeat types are present in all positive samples. The region between amino acid sequences VDD and CLRH was considered the repeat region of exon 2 and inspected for repeat types.
Marker
|
Amino acid sequence
|
Exon 1
|
MVSFSKNKVLSAAVFASVLLDN
|
Exon2 start
|
NNSAFNNNLCSKNA
|
Beginning of repeat region of exon 2
|
VDD
|
End of repeat region of exon 2
|
CLRH
|
Repeat types:
|
Type 1
|
AHHAHHVAD
|
Type 2
|
AHHAHHAAD
|
Type 3
|
AHHAHHAAY
|
Type 4
|
AHH
|
Type 5
|
AHHAHHASD
|
Type 6
|
AHHATD
|
Type 7
|
AHHAAD
|
Type 8
|
AHHAAY
|
Type 9
|
AAY
|
Type 10
|
AHHAAAHHATD
|
Type 11
|
AHN
|
Type 12
|
AHHAAAHHEAATH
|
Type 13
|
AHHASD
|
Type 14
|
AHHAHHATD
|
Type 30
|
AHHAVD
|
Type 31
|
SHHAAY
|
2.9 Kelch 13 control experiment for barcoding method
To determine if uneven performance between barcodes with normalized DNA input was an issue specific to pfhrp2, we designed a version of our assay for k13 amplicons. Five barcoding primer pairs were designed for k13 and checked for self- and cross-dimers in silico prior to being ordered (Supplementary Table 4). k13 was amplified from a panel of reference strains including 7G8, 3D7, HB3, D6, and Dd2 according to the Malaria Resistance Surveillance Project (MaRS) protocol (Talundzic, et al., 2018). The amplicons were then barcoded with k13-specific barcoding primers under the same PCR conditions described in section 2.4. Two normalized sequencing pools were made, the first containing the reference panel and the second containing 7G8 replicates. Bead purification, library preparation, and sequencing were performed as described in section 2.6.
2.10 Barcode primer in vitro experiment
Given uneven performance between barcodes, we assessed the performance of each on normalized sequencing pools of 7G8 replicates to determine which performed better than others and should therefore be more heavily utilized. The 28 barcodes were spread across 6 sequencing pools with 3–5 samples each and sequenced on standard flow cells for 3 hours. The following ten barcodes were selected for the evaluation experiments: bc01, bc02, bc12, bc13, bc17, bc18, bc49, bc30, bc52, and bc60 (Supplementary Information, Supplementary Table 5).
2.11 Evaluation
The MR4 and PacBio sets were normalized, and the remaining field samples were not, given the expectation in field-based use of this assay that there will be pfhrp2 positive and negative samples. Otherwise, samples used for evaluation were treated identically according to the methods described in sections 2.2 to 2.6. To confirm pfhrp2 negativity and to assess the viability of a sample for genotyping, we implemented a coverage-based positivity threshold (Supplementary Information, Supplementary Table 6). In a field application, the results of this threshold test are the first of two possible outcomes from this assay (Fig. 1). Putatively positive samples from this step were processed for repeat-typing as described in section 2.8.