Epidemiology of mutant Plasmodium falciparum parasites lacking histidine-rich protein 2/3 genes in Eritrea 2 years after switching from HRP2-based RDTs

Eritrea was the first African country to complete a nationwide switch in 2016 away from HRP2-based RDTs due to high rates of false-negative RDT results caused by Plasmodium falciparum parasites lacking hrp2/hrp3 genes. A cross-sectional survey was conducted during 2019 enrolling symptomatic malaria patients from nine health facilities across three zones consecutively to investigate the epidemiology of P. falciparum lacking hrp2/3 after the RDT switch. Molecular analyses of 715 samples revealed the overall prevalence of hrp2-, hrp3-, and dual hrp2/3-deleted parasites as 9.4% (95%CI 7.4–11.7%), 41.7% (95% CI 38.1–45.3%) and 7.6% (95% CI 5.8–9.7%), respectively. The prevalence of hrp2- and hrp3-deletion is heterogeneous within and between zones: highest in Anseba (27.1% and 57.9%), followed by Gash Barka (6.4% and 37.9%) and Debub zone (5.2% and 43.8%). hrp2/3-deleted parasites have multiple diverse haplotypes, with many shared or connected among parasites of different hrp2/3 status, indicating mutant parasites have likely evolved from multiple and local parasite genetic backgrounds. The findings show although prevalence of hrp2/3-deleted parasites is lower 2 years after RDT switching, HRP2-based RDTs remain unsuitable for malaria diagnosis in Eritrea. Continued surveillance of hrp2/3-deleted parasites in Eritrea and neighbouring countries is required to monitor the trend.

Dual hrp2/3-deleted parasites were detected in six of the nine health facilities in the three zones: 3/5 facilities in Gash Barka zone, 2/2 facilities in Anseba zone and 1/2 facilities in the Debub zone. The prevalence of dual hrp2/3-deleted parasites at Keren Hospital (21.8%, 95% CI 14.0-32.2%) and Hagaz Health Centre (17.2%, 95% CI 7.1-35.0%), the two facilities with the highest prevalence in this study, were significantly lower than that previously observed in patients from Ghindae Hospital (80.8%, 95% CI 61.7-92.0%; Fishers exact test, p < 0.001). Although the prevalence at both sites was also lower than that previously observed in Massawa Hospital (41.7%, 95% CI 24.4-61.2%), the difference failed to reach statistical significance (Fisher's exact test, p > 0.065). HRP2 levels and hrp2/3 status. HRP2 protein levels were measured for all samples classified by PCR as single hrp2-, single hrp3-and dual hrp2/3-deleted parasites, as well as a randomly selected set of samples without gene deletions using ELISA (Fig. 4). The mean ΔOD for 105 samples without hrp2 and hrp3 deletions was 3.032 (95% CI 2.783-3.282). The mean ΔOD for 244 samples with single hrp3-deletions was 2.576 (95% CI 2.373-2.779), not significantly different to samples without gene deletions (P = 0.053), but significantly higher than single hrp2-or dual hrp2/3-deleted parasites (P < 0.0001). The mean ΔOD for 13 single hrp2-deleted samples was 0.167 (95% CI 0.053-0.388), not significantly different to dual hrp2/3-deleted samples (n = 53) which had a mean ΔOD of 0.058 (95%CI 0.043-0.072) (P > 0.999). Hence, this ELISA assay detected all samples with dual hrp2/3 deletions and all single hrp2 deletions except one sample, and showed little cross reactivity with HRP3. Sequencing revealed that the exceptional parasite had deleted the 3' end of exon2 but retained exon1 and the 5' end of the exon 2 of the hrp2 gene which encoded 37 amino acids (excluding primer sequences) including three type 2 repeats (AHHAHHAADAHHAHHAADAHHAHHAADAHHAAAHH). It is likely that the presence of a small number of type 2 repeats (contains a major epitope recognised by monoclonal antibodies used in malaria RDTs 5 ) has contributed to the positive ELISA results although persisting HRP2 from previous infections could not be ruled out.
Microsatellite typing. 180    Genetic relatedness of parasites. Genetic relatedness of parasites with different hrp2/3 status is analysed by connecting parasites sharing identical alleles on at least five of the seven microsatellite markers using Phyloviz. A major cluster was revealed connecting 59 parasite haplotypes: 49 obtained from this study and 10 obtained previously from Ghindae and Massawa hospitals in the Northern Red Sea zone (Fig. 5). 14/19 (73.7%) of haplotypes of dual hrp2/3 deletions, 5/6 (83.3%) of single hrp2 deletion, and 28/64 (43.8%) of single hrp3 deletions, as well as 28/48 (58.3%) of intact hrp2/3 genes from the current study were connected in this major cluster. This indicates that parasites with different gene deletion status are genetically related among themselves, and to parasites without gene deletions. Unlike previous observations from Ghindae and Massawa hospital where all (E1, E2, E3, E5, E13) but one (E4) haplotype with dual hrp2/3 deletions were connected, 5/19 haplotypes with dual hrp2/3 deletions (8, 33, 57, 63 and 113) were not connected to the major cluster. One haplotype with single hrp2 deletion and 36 haplotypes with single hrp3 deletions were also outside of the major cluster. Combined, these results demonstrate that hrp2-and hrp3-deleted parasites in Eritrea have evolved from local parasite populations and from parasites of different genetic back grounds.
While parasites from the Northern Red Sea zone tend to cluster on one branch (Fig. 7a), there is no major clustering of parasites between other zones indicating no geographical restrictions on parasite movement. There is no major clustering of parasites within or between different health facilities although several parasite haplotypes from Ghindae Hospital appear to be closely related to some from Keren Hospital (Fig. 7b), possibly indicating frequent human movement between these two areas. The overall lack of parasite clustering observed in different zones may reflect frequent human movement between zones for purpose of trading, schooling and other business, particularly among the youth.

Discussion
Eritrea was the first African country to switch away from HRP2-based RDTs nationwide due to high rates of false negative RDT results caused by hrp2/3-deleted parasites. It is important for the global response strategy to assess the impact of removing HRP2-based RDTs for case management on prevalence trends and epidemiology of hrp2/3 deletions. This study shows the overall prevalence of hrp2, hrp3 and dual hrp2/3 gene deletions estimated across three zones in Eritrea is 9.4% (95% CI 7.4-11.7%), 41.7% (95% CI 38.1-45.3%) and 7. 6% (95% CI 5.8-9.7%) respectively. Because HRP2-based RDTs are no longer used in Eritrea it is not possible to describe the prevalence of hrp2/3 deletions causing false negative RDT results; however, a conservative estimate can be made based on the prevalence of dual hrp2/3 deletions, which exceeds the WHO 5% prevalence criterion to switch away from HRP2-based RDTs for case management 6 . Therefore, non-HRP2 based RDTs or microscopy remain the most suitable point-of-care malaria diagnostic tools for Eritrea. Currently, malaria diagnosis in Eritrea involves two steps: a pan-pLDH RDT for malaria suspects, and if positive, a Pf (HRP2)/Pv-pLDH RDT is followed to distinguish P. falciparum from P. vivax. Prevalence of gene deletions is highly heterogeneous within and between zones. In general, North-west areas of the country have a higher prevalence of hrp2/3 deletions than South, South-West areas. Keren hospital in the Anseba zone had the highest prevalence of samples containing hrp2-deleted (29.5%) and dual hrp2/3-deleted (21.8%) parasites, followed by Hagaz Health Centre from the same zone, and then by Agordat Hospital from the neighbouring Gash Barka zone. MaiMine Health Centre in the Debub zone had the lowest hrp2/3 deletion prevalence, followed by Barentu Hospital in the Gash Barka zone. Prevalence of hrp2/3-deleted parasites at any health facilities surveyed in this study was markedly lower than that observed in patients from Ghindae Hospital (81%) in the Northern Red Sea zone in the 2016 survey, while health facilities in the Gash Barka and Debub zones also had lower prevalence than that previously reported for Massawa Hospital (42%) 3 .
Ghindae and Massawa hospitals in the Northern Red Sea zone were included in this survey to enable a direct comparison of hrp2/3-deletion prevalence estimates before and 2 years after switching RDTs. However, low patient numbers in these hospitals during the current survey resulted in no samples collected, and prevented such a direct comparison. Nevertheless, the prevalence of hrp2-deleted parasites determined in the current survey for the Gash Barka (6.4%), Anseba (27.1%) and Debub zone (5.2%) was much lower than the false negative RDT rate reported for these zones in 2016, 64.7%, 92.9% and 71.4% respectively 2 , although the sample number was relatively small and gene deletion status was not confirmed by molecular analysis in the 2016 RDT study. The prevalence estimates of hrp2 and hrp3 deletions determined for Gash Barka zone in the current survey, 6.4% and 37.9%, was also lower than the corresponding prevalence of 10.8% and 47.7% reported for the same zone in a set of samples collected in 2013-2014 7 . Furthermore, the current prevalence estimates of hrp2-deleted parasites in Eritrea (9.4%) appears to be lower than that recently reported from areas in neighbouring countries of Ethiopia (50%, 100%) 8,9 , Djibouti (86.5%) 10 , and amongst samples tested in returning travellers from Sudan (11.2%) and South Sudan (17.7%) 11 . This reduction in prevalence may reflect a genuine downward trend following the removal of HRP2-based RDTs for case management in Eritrea, but may also result from different sampling, different levels of genetic diversity and different laboratory methods used. Importantly, data generated from the current survey will provide baseline information for future surveys to be conducted in these areas to determine the trend of gene deletion prevalence.
In this study we performed ELISA to confirm the HRP protein expression status. The ELISA results showed that parasites with dual hrp2/3 deletions or single hrp2 deletions had negligible levels of HRP antigens and are www.nature.com/scientificreports/ expected to be undetectable by HRP2-based RDTs, thus confirming gene deletion status determined by PCRs. While the number of single hrp2-deleted parasites was small (n = 13), it was expected that most of these parasites would have produced positive ELISA results due to cross reactivity between HRP3 and anti-HRP2 antibody used in the ELISA kit. However, all but one sample had negligible levels of HRP. This likely resulted from little cross reactivity of antibodies against HRP3 in the ELISA kit, not from insufficient quantity of HRP3 due to low parasitemia since all patients were enrolled based on positive microscopy results.
In contrast to single hrp2-deleted parasites, single hrp3-deleted parasites showed high levels of HRP and are expected to be detected by HRP2-based RDTs, albeit possibly with slightly lower detection sensitivity than parasites without gene deletions. Therefore, unlike hrp2-deleted parasites 12 , single hrp3-deleted parasites are not www.nature.com/scientificreports/ expected to have a selection advantage exerted by use of HRP2-based RDTs. The driving force behind the high prevalence of hrp3-deleted parasites in Eritrea is unclear. In addition to determining gene deletion prevalence, the survey provided a unique opportunity to compare parasite genetic diversity and genetic relatedness between the two surveys. While these analyses were limited to only the dominant haplotypes present in a subset of samples from this study, our data provided strong evidence that that hrp2/3-deleted parasites have likely emerged de novo from multiple genetic backgrounds. Firstly, parasites with different hrp2/3 gene status shared identical haplotypes or showed close genetic connectivity, suggesting they have evolved from same genetic backgrounds; secondly, dual hrp2/hrp3-deleted parasites from the current survey have a markedly higher heterogeneity (H E = 0.48) than those from the 2016 survey (H E = 0.11), and processed a wider number of haplotypes (19 haplotypes in 27 parasite isolates) than those from the 2016 survey (6 haplotypes in 31 parasite isolates) suggesting they have evolved from multiple genetic backgrounds. This implies that besides surveillance and switching RDTs, control as opposed to containment strategies may be most effective for combatting these mutant parasites.
Interestingly, the patterns of hrp2/3 gene deletion are also much more diverse in the current survey compared to the 2016 survey. In the current survey, we observed parasites deleting hrp2 only, hrp3 only and dual hrp2/3 gene, while in the 2016 survey only dual hrp2/3-deleted and single hrp3-deleted parasites were observed. In addition, we observed significant proportions of parasites having deleted either exon1 or exon2 or both exons of hrp2 and hrp3, while in the 2016 survey we only observed parasites deleting both exons of hrp2 and hrp3. This suggests that there likely have been an outbreak of dual hrp2/3-deleted P. falciparum in the Northern Red Sea zone during the 2015-2016 transmission season. Importantly, our data, combined with several recent reports of significant proportions of single exon deletion of hrp2/3 genes in other areas 8,13 , suggest molecular assays focusing on only one of the exons in hrp2/3 genes 14-16 could underestimate a large proportions of hrp2/3-deleted parasites in some areas.
In conclusion, this study demonstrates that although lower, the prevalence of hrp2/3 deleted parasites remains high in Eritrea 2 years after the country switched away from HRP2-based RDTs. While the prevalence is highly heterogeneous within and between different zones, HRP2-based RDTs remain unsuitable for malaria diagnosis in the country. Results also demonstrate hrp2/3-deleted parasites have multiple haplotypes and many are shared with or connected with parasites without gene deletions, indicating that these mutant parasites have highly likely evolved from multiple and local parasite genetic backgrounds. Therefore, although preventing spread of these mutants is important, it is likely that all areas using HRP2-based RDTs are at risk of the same phenomena. A better understanding of the factors driving selection beyond use of HRP2-based RDTs is needed. Continued surveillance on the prevalence of hrp2/3-deleted parasites in the affected zones in Eritrea and neighbouring countries using the WHO standard survey protocol 17 is required to better understand the prevalence trend and to continuously inform malaria diagnosis and case management policies.

Methods
All methods were performed in accordance with the relevant guidelines and regulations.
Survey sites, patient recruitment and sample collection. A cross-sectional surveillance involving key malaria endemic sites in Eritrea was conducted during the transmission season 2019-2020 (Jan 2019-Jan 2020) in 11 health facilities across the 4 zones where hrp2/3 deletions or high false negativity rate have been previously reported 2,3 (Table 1). Suspected malaria cases seeking treatment at these facilities were tested by microscopy. Patients meeting inclusion criteria (confirmed P. falciparum infection by microscopy; both sexes with ages ≥ 1) were informed about the survey and invited to give consent for participation. Enrolment was consecutive until the required sample number was reached or at the end of transmission season. Sample collection. Eligible patients were finger pricked to produce 3-5 blood spots (~ 75-80µL each) on Whatman Protein Saver Cards. Cards were air-dried for 24 h, placed in sealed plastic bag with desiccant and stored at room temperature. DNA extraction and Plasmodium speciation. Genomic DNA was extracted from 3 punches of a dried blood spot (DBS) using QIAamp DNA Mini Kit and a QIAcube (QIAGEN) following the manufacturer's instructions 3 . Plasmodium spp was detected using an 18S rRNA gene-based multiplex PCR assay 18 . P. vivax infections were confirmed by PCR using P. vivax specific primers alone.
Characterization of hrp2/3 gene deletions. PCR assays were used to detect the presence of exon1 across exon2 (exon1-2) and exon2 of hrp2 and hrp3 genes in each sample as previously described 19 . PCR assays were also used to detect the presence of msp1 and msp2 genes in the same sample. As previously described, samples were classified as infected with parasites having deleted hrp2/3 genes if they were PCR negative for exon1-2 and/or exon2 of the hrp2 and/or hrp3 gene, but were PCR positive for both msp1 and msp2 genes 20 .
HRP2 levels. Levels of HRP2 in DBS samples were measured for all samples classified as having gene deletions (n = 311) and a randomly selected set of samples without gene deletions (n = 105) using a Quantimal Celisa Pf HRP2 Assay kit (Cellabs, Cat number: KM 810). The assay involves two steps: (1) Elution of parasite proteins from a second DBS following a published method 21  www.nature.com/scientificreports/ Tween 20). The tube was vortexed, spun to keep DBS submerged and followed by shaking in Thermomixer (12 °C, 350 rpm) for 12 h. During the first three hours of shaking, each tube was vortexed for 30 s, spun and returned to the Thermomixer once every hour. After shaking, the tubes were centrifuged at 13, 000 rpm for 3 min. Eluate (200 µl) was transferred to a new tube and 100 µl used for the CELISA HRP2 assay. If not immediately assayed, the eluate was stored at − 20 °C for 2 days and assayed. (2) ELISA following the manufacturer's instructions. Briefly, 100 µl of each serial dilution of positive control, test sample, HRP positive control sample (Pf 3D7), HRP negative control sample (Pf 3BD5), normal RBC control was pipetted into individual wells. The contents were mixed by pipetting up and down 4 times. The plate was covered, placed in a humid chamber and incubated for one hour at 37 °C. The plate was then washed 4 times with wash buffer (PBS- Tween 20). Conjugate (100 µl/well) was added and the plate was incubated in a humid chamber for one hour at 37 °C. This was followed by 4 washes with wash buffer. The plate was inverted and dried on a sheet of paper. Freshly prepared Substrate Solution (TMB, 100 µl/well) was added, the plate was covered with foil and incubated for 15 min at room temperature. Stop Solution (50 µl/well) was added and results were read in a spectrophotometer (TECAN) at 450 nm wavelength.
Final absorbance OD 450nm for each sample was normalised with the average of OD 450nm from normal human blood (n = 40). ΔOD 450nm (OD 450nm sample-OD 450nm normal human blood) was used to represent the quantity of HRP2 in the DBS sample.
Microsatellite analysis. Seven neutral microsatellite markers (TA1, PolyA, PfPK2, TA109, 2490, 313, and 383) were analysed for a total of 180 samples as previously published [22][23][24] . These included 20 samples selected from each health facility to represent local parasites with different hrp2/3 status. Five of the seven microsatellite markers (TA1, PolyA, PfPK2, TA109 and 2490) were amplified by semi-nested PCRs 22 while the remaining two microsatellite markers (313 and 383) by a single round PCR 23 using published primers and PCR conditions. Sizes of fluorescent labelled PCR products were analysed on an ABI 3100 Genetic Analyzer sequencer (Applied Biosystems). The microsatellite fragments/alleles were scored manually using Peak Scanner Software version 1.0 (Applied Biosystems, https:// peak-scann er-softw are. softw are. infor mer. com/1. 0/) and a peak height > 300 relative fluorescence units (rfu) was considered as a positive peak 3 . This criteria is more stringent than that (> 200 rfu) used in earlier studies 22,25 . A laboratory line (3D7) was included in each run for size calibration.
Haplotype construction, genetic diversity-expected heterogeneity (H E ) and genetic relatedness. Seven-microsatellite-marker-haplotypes were constructed from the most dominant allele at each marker for samples with positive peaks at all seven markers. Haplotypes in relation to health facilities and gene deletion status were imported into the FSTAT software (V2.9.4) and calculated. Genetic relatedness analysis was conducted using seven-microsatellite-marker-haplotypes and PHYLOViZ online software (version 1.1) following instructions.
Statistical analysis. Prevalence of deletions was compared between selected health facilities using Fisher's exact test, and between zones using an Independent Samples Chi-square test in Graphpad Prism (version 9). HRP2 levels between samples of different hrp2/3 gene status and genetic diversity between samples of different health facilities and hrp2/3 status, were compared using Kruskal-Wallis test with Dwass-Critchlow-Fligner pairwise comparison in Jamovi (version 1.8.1). Confidence intervals for all proportions were calculated using the modified Wald method. Sequencing hrp2 exon2. DNA sequencing was performed for a single sample that was determined as single hrp2-deletion but had a detectable HRP antigen level. PCR product of hrp2 exon2 was excised from an agarose gel, purified using a Zymoclean™ gel DNA recovery kit and sequenced using Big-dye 26 . DNA sequence was analysed and translated into protein sequence using MEGA4.02 software 27 . Amino acid repeat type was determined as previously described 26 .
Ethical consideration. The study was approved by the Eritrean MOH Research and Ethical Committee.
Patients were enrolled after providing informed consent following a detailed explanation about the investigation. For participants under 18, informed consent was obtained from a parent and/or legal guardian. All patient specimens were given a unique identifying (ID) number after collection and only this ID was used for data linkage. Laboratory analyses on hrp2/hrp3 status, genetic diversity and relatedness conducted was approved by the Australian Defence Joint Health Command Low Risk Ethics Panel (LREP 15-004).

Data availability
Data generated from this study are included in the published article or as supplementary information. Data are also available from corresponding author on request. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.