Abstract
Background:
Tackling HIV drug resistance is one of major challenges for ending AIDS epidemic, but the elevated expense of cutting-edge genomics hampers the advancement of HIV genotype testing for clinical care.
Methods:
We developed a HIV genotype testing pipeline that centers on a cost-efficient portable Nanopore sequencer. Accuracy verification was conducted through comparison with parallel data obtained via fixed-site Pacbio sequencing. Our complete pol-gene sequencing strategy coupled with portable high-throughput sequencing was applied to identify drug resistance mutations across 58 samples sourced from the ART-treated Los Angeles General Medical Center Rand Schrader Clinic (LARSC) cohort (7 samples from 7 individuals) and the ART-naïve Center for HIV/AIDS Vaccine Immunology (CHAVI) cohort (51 samples from 38 individuals).
Results:
A total of 472 HIV consensus sequences, each tagged with a unique molecular identifier, were produced from over 1.4 million bases acquired through portable Nanopore sequencing, which matched those obtained independently via Pacbio sequencing. With this desirable accuracy, we first documented the linkage of multidrug cross-resistance mutations across Integrase Strand Transfer inhibitors (INSTIs) and Non-Nucleoside Reverse Transcriptase Inhibitors (NNRTIs) from an individual failing a second-generation INSTI regimen. By producing more than 500 full-length HIV pol gene sequences in a single portable sequencing run, we detected Protease Inhibitor (PI), Nucleoside Reverse Transcriptase Inhibitor (NRTI), NNRTI and INSTI resistance mutations. All drug resistance mutations identified through portable sequencing were cross-validated using fixed-site Pacbio sequencing.
Conclusions:
Our accurate and affordable HIV drug resistance testing solution is adaptable for both individual patient care and large-scale surveillance initiatives.
INTRODUCTION
HIV drug resistance mutations stand as a predominant driver of virologic failures, posing a global obstacle to sustained lifelong therapy [1, 2]. Even the World Health Organization (WHO) recommends HIV drug resistance surveillance in those initiating or reinitiating antiretroviral therapy (ART) [3], the pragmatic implementation of HIV genotype testing for clinical care continues to be a challenge in low-income and middle-income countries, primarily due to infrastructure limitations and cost-related hurdles [2, 4]. In high-income settings, despite the recent advances in next-generation sequencing technologies, HIV drug resistance mutation testing still relies heavily on traditional Sanger sequencing methods for clinical care [4]. Next-generation sequencing genotypic resistance assays have improved the drug resistance mutation detection capacity, however, its high per-sample cost (>$400), restricts their routine use in patient care [5, 6]. Moreover, all the assays currently employed in clinical practice sequence the HIV genome in multiple segments, making it challenging to identify linked cross-class drug resistance mutations that can pose an increased risk of virological failure in combination therapy [7].
One of the major barriers that deter a transition to next-generation sequencing based HIV genotype testing is the expense associated with high-priced next-generation sequencers, impeding the wide adoption of high-throughput, massively parallel sequencing [8]. In this study, we have developed a robust HIV genotype testing workflow using a cost-effective portable Nanopore sequencer. We have employed a full-length pol-gene sequencing strategy with a unique molecular identifier (UMI) scheme to detect the linkage of cross-class multi-drug resistance mutations [9, 10]. Herein, we present the first documented linkage of multidrug cross-resistance mutations across Integrase Strand Transfer inhibitors (INSTIs) and Non-Nucleoside Reverse Transcriptase Inhibitors (NNRTIs). We offer a cutting-edge yet accessible solution for HIV drug resistance mutation screening not only in high-income but also in low-income and middle-income countries.
METHODS
Study population
A total of 58 plasma specimens were sequenced in this study, consisting of 7 samples from 7 ART-treated individuals within the Los Angeles General Medical Center Rand Schrader Clinic (LARSC) cohort (USC IRB protocol #HS-12-00121) and 51 samples from 38 ART-naïve HIV infected individuals in the Center for HIV/AIDS Vaccine Immunology (CHAVI) cohort (DUHS IRB #CR6_Pro00007562) (Tables 1 and 2 and Supplementary Table 1). All study participants provided written informed consent upon enrollment.
Table 1.
Drug resistance mutation profiles of 58 plasma specimens.
| Specimena | Number of complete pol gene consensus sequences (Number of raw reads) by Nanopore sequencing |
Number of complete pol gene consensus sequences (Number of raw reads) by Pacbio sequencing |
Drug resistance mutations by Nanopore sequencing (Number of sequences ) |
Drug resistance mutations by Pacbio sequencing (Number of sequences ) |
Gender | Sample collection date |
Viral Load (copies/ml ) |
CD4+ T cell count (cells/mm 3) |
Diversity (%) |
|---|---|---|---|---|---|---|---|---|---|
| United States | |||||||||
| EM2045b | 8 (2.687) | 9 (752) | NNRTI: K103N (8/8) and K238T (8/8) INSTI: S153F (5/8) and R263K (3/8) |
NNRTI: K103N (9/9) and K238T (9/9) INSTI: S153F (6/9), S230R (2/9), and R263K (3/9) |
M | 3/29/22 | 501 | 289 | 0.22 |
| 5 (1,315) | NNRTI: K103N (5/5) and K238T (5/5) INSTI: S153F (5/5) and S230R (3/5) |
0.11 | |||||||
| IE0099 | 25 (8,248) | 2 (187) | none | none | M | 3/7/22 | 10,058 | 143 | 0.7 |
| QE0095 | 17 (4,907) | 1 (170) | none | none | M | 2/24/22 | 59,013 | 197 | 0.63 |
| FI3720 | 35 (16,990) | 2 (235) | none | none | M | 12/14/21 | 181,992 | 1,272 | 0.05 |
| KI6633 | 33 (5,994) | 57 (73,311) | NRTI: K70E (2/33) NNRTI: E138A (5/33) |
NRTI: K70E (4/57) NNRTI: E138A (8/57) |
M | 9/11/12 | 460,004 | 115 | 0.94 |
| ZI9923b | 45 (4,420) | 69 (57,798) | NNRTI: E138A (29/45) | NNRTI: E138A (50/69) | M | 10/16/12 | 85,900 | 70 | 0.69 |
| 19 (14,462) | NNRTI: E138A (10/19) | 0.89 | |||||||
| MP7985b | 6 (870) | 6 (14,748) | NRTI: K70R (1/6) | NRTI: K70R (1/6) | F | 6/3/13 | 30,262 | 400 | 0.92 |
| 6 (5,297) | none | 0.36 | |||||||
| 716010054-1 | 22 (3,348) | 33 (40,271) | none | none | M | 7/16/09 | 65300 | 693 | 0.06 |
| 716010054-2 | 12 (1,983) | 23 (22,062) | none | none | M | 12/9/10 | 71263 | 677 | 0.28 |
| 701010222-1b | 6 (1,367) | 14 (12,504) | INSTI: E157Q (6/6) | INSTI: E157Q (14/14) | M | 1/4/08 | 53507 | 541 | 0.1 |
| 10 (3,022) | INSTI: E157Q (10/10) | 0.03 | |||||||
| 701010222-2 | 3 (1,581) | 3 (17,330) | INSTI: E157Q (3/3) | INSTI: E157Q (3/3) | M | 6/17/09 | 2119 | 564 | 0.47 |
| 701010248-1 | 6 (1,444) | 10 (19,110) | none | none | M | 2/13/08 | 2485 | 873 | 0.05 |
| Malawi | |||||||||
| 702010157-2 | 12 (1,467) | 20 (23,094) | none | none | M | 8/26/08 | 27580 | 256 | 0.25 |
| 702010176-2 | 4 (1,271) | 41 (14,817) | none | none | M | 7/10/08 | 15631 | 277 | 0.52 |
| 702010374-1 | 3 (2,330) | 58 (21,431) | none | none | F | 3/25/08 | 367133 | 774 | 1.37 |
| 702010374-2 | 5 (1,617) | 5 (15,939) | none | none | F | 5/18/09 | 6849 | 930 | 0.81 |
| 702010600-2b | 22 (1,633) | 26 (14,599) | none | none | M | 11/10/09 | 33542 | 304 | 0.69 |
| 11 (6,854) | none | 0.43 | |||||||
| 703010677-2 | 11 (1,170) | 25 (14,258) | none | none | M | 6/15/09 | 57949 | 369 | 0.3 |
| 702010773-2b | 22 (1,947) | 30 (17,721) | none | none | M | 1/26/10 | 31523 | 313 | 0.32 |
| 19 (10,016) | none | 0.29 | |||||||
| 702010809-1 | 1 (304) | 14 (4,607) | none | none | M | 12/22/08 | 9876 | N/A | N/A |
| 702010809-2 | 20 (2,222) | 28 (20,520) | none | none | M | 11/23/09 | 37386 | 594 | 0.29 |
| 703010863-1 | 14 (979) | 14 (20,033) | none | none | F | 8/12/08 | 103456 | 547 | 0.13 |
| 703010863-2b | 1 (649) | 26 (7,270) | NRTI: M184V (1/1) NNRTI: K103N (1/1) |
NRTI: M184V (26/26) NNRTI: K103N (26/26) |
F | 6/28/10 | 29056 | 410 | N/A |
| 9 (11,080) | NRTI: M184V (9/9) NNRTI: K103N (9/9) |
0.33 | |||||||
| 702010895-1 | 13 (6,704) | 1 (13,033) | none | none | M | 3/5/09 | 306676 | 567 | 0.01 |
| 702010895-2 | 6 (1,975) | 10 (19,546) | none | none | M | 2/4/10 | 28905 | 714 | 0.28 |
| 703010976-2 | 11 (1,222) | 22 (15,418) | none | none | M | 3/11/10 | 154543 | 232 | 0.7 |
| 703011129-1 | 32 (3,257) | 62 (36,000) | none | none | M | 4/2/09 | 123865 | 369 | 0.01 |
| 703011129-2 | 15 (1,535) | 18 (13,491) | none | none | M | 9/27/10 | 18065 | 220 | 0.29 |
| 703011165-1b | 3 (1,749) | 44 (18,757) | none | none | F | 4/17/09 | >750000 | 409 | 0.02 |
| 24 (12,707) | none | 0.01 | |||||||
| 703011231-1 | 11 (900) | 17 (19,645) | none | none | M | 6/2/09 | 10819 | 732 | 0.01 |
| 703011231-2 | 17 (6,251) | 25 (19,924) | INSTI: E158Q (10/17) | INSTI: E158Q (17/25) | M | 1/11/11 | 219506 | 707 | 0.11 |
| 703011278-2 | 10 (2,723) | 17 (24,373) | none | none | M | 8/19/10 | 78943 | 382 | 0.17 |
| 703011280-2 | 6 (1,190) | 44 (12,192) | none | none | M | 11/8/10 | 91778 | 506 | 0.34 |
| 703011306-1 | 5 (2,319) | 9 (25,050) | none | none | F | 7/20/09 | 2765 | 646 | 0.05 |
| 703011306-2 | 3 (296) | 11 (6,731) | none | none | F | 9/14/10 | 17339 | 269 | 0.69 |
| 703011350-2b | 8 (1,227) | 17 (14,071) | none | none | M | 1/6/11 | 31268 | 318 | 0.33 |
| 9 (2,474) | none | 0.23 | |||||||
| 703011634-2b | 2 (1,695) | 7 (23,752) | none | INSTI: P145S (2/7) | M | 3/22/11 | 14260 | 372 | 0.03 |
| 4 (2,118) | none | 0.44 | |||||||
| 703011642-2 | 17 (5,556) | 63 (33,300) | none | none | M | 3/14/11 | >750000 | 103 | 0.06 |
| 703011736-2 | 28 (2,884) | 47 (27,043) | none | none | F | 4/23/12 | 267291 | 164 | 0.31 |
| 703011762-1 | 2 (3,090) | 7 (42,604) | none | none | F | 8/17/10 | 14189 | N/A | 0 |
| 703011762-2 | 1 (580) | 1 (9,641) | none | none | F | 9/10/12 | 3084 | 708 | N/A |
| 703011773-2 | 5 (708) | 21 (12,662) | none | none | M | 10/12/11 | 117674 | 305 | 0.44 |
| 703011798-2 | 4 (1,367) | 34 (11,421) | none | none | M | 1/4/12 | 561506 | 279 | 0.11 |
| 703011809-2 | 25 (2,352) | 35 (22,853) | none | none | M | 4/22/12 | 44190 | 273 | 0.33 |
| 703011813-1b | 7 (1,837) | 10 (18,136) | none | none | M | 12/6/10 | 97160 | 663 | 0.03 |
| 7 (8,930) | none | 0.06 | |||||||
| 703011813-2b | 4 (788) | 21 (15,792) | PI: V32A (1/4) | PI: V32A (1/21) | M | 4/25/12 | 53229 | 390 | 0.31 |
| 4 (2,306) | none | 0.33 | |||||||
| 703011824-2 | 11 (1,413) | 16 (19,920) | none | none | M | 2/21/12 | 53229 | 302 | 0.4 |
| 703011860-2 | 28 (4,627) | 54 (11,519) | INSTI: A128T (27/28) | INSTI: A128T (53/54) | F | 3/23/12 | 53229 | 366 | 0.1 |
| 703011895-2b | 10 (1,209) | 24 (11,037) | none | none | M | 6/4/12 | 53229 | 492 | 0.32 |
| 24 (11,439) | none | 0.14 | |||||||
| 703011927-2 | 12 (6,040) | 37 (15,833) | none | none | M | 9/3/12 | 443565 | 224 | 0.07 |
| 703011955-2 | 2 (762) | 20 (6,197) | none | none | M | 3/8/13 | 6345 | 280 | 0.43 |
| 703011992-2 | 6 (797) | 9 (16,370) | none | none | M | 12/17/12 | 25257 | 878 | 0.28 |
| South Africa | |||||||||
| 705010136-1 | 3 (838) | 3 (19,614) | none | none | F | 5/16/07 | 1202 | 622 | 0 |
| 705010136-2b | 4 (1,790) | 4 (7,647) | INSTI: L74M (3/4) | INSTI: L74M (3/4) | F | 4/16/08 | 3091 | 402 | 0.13 |
| 2 (4,077) | INSTI: L74M (2/2) | 0.07 | |||||||
| 706010139-1 | 2 (379) | 5 (5,421) | none | none | F | 7/18/07 | 19000 | 1032 | 0.07 |
| 706010139-2b | 4 (845) | 36 (11,653) | none | none | F | 9/10/08 | 199000 | 587 | 0.24 |
| 27 (12,654) | none | 0.18 | |||||||
| 705010411-2 | 17 (2,021) | 13 (20,676) | none | none | F | 3/26/09 | 10587 | 385 | 0.13 |
| 705010450-1 | 10 (1,452) | 40 (14,694) | none | none | M | 4/3/08 | 116275 | 388 | 0.03 |
The six-character ID represents for ART-positive samples from the Los Angeles General Medical Center Rand Schrader Clinic (LARSC) cohort and IDs beginning with “7” correspond to ART-naïve samples from the Center for HIV/AIDS Vaccine Immunology (CHAVI) cohort. Two different samples collected from the same individual were labeled as either “−1” or “−2”.
Two separate portable Nanopore sequencing runs were conducted, and the results from each run were presented separately.
Table 2.
Predicted drug resistance by portable Nanopore sequencing.
| Specimen | Predicted drug resistancea | Linked cross-resistanceb | ART |
|---|---|---|---|
| EM2045 | NNRTI: EFV (h), NVP (h) INSTI: BIC (i), CAB (i), DTG (i), EVG (i), RAL (l) |
3: NNRTI: EFV (h), NVP (h) and INSTI: BIC (i), CAB (i), DTG (i), EVG (i), RAL (l) 5: NNRTI: EFV (h), NVP (h) and INSTI: BIC (l), CAB (l), DTG (l), EVG (l), RAL (p) |
NRTI: FTC and TAF INSTI: BIC and DTG |
| NNRTI: EFV (h) and NVP (h) INSTI: BIC (l), CAB (i), DTG (i), EVG (i), RAL (i) |
3: NNRTI: EFV (h), NVP (h) and INSTI: BIC (l), CAB (i), DTG (i), EVG (i), RAL (i) 2: NNRTI: EFV (h), NVP (h) and INSTI: BIC (l), CAB (l), DTG (l), EVG (l), RAL (p) |
||
| KI6633 | NRTI: ABC (l), 3TC (p), D4T (l), DDI (l), FTC (p), TDF (l) NNRTI: ETR (p), RPV (l) |
2: NRTI: ABC (l), 3TC (p), D4T (l), DDI (l), FTC (p), TDF (l) 5: NNRTI: ETR (p), RPV (l) |
PI: ATV NRTI: 3TC, FTC, TDF, ZDV (AZT) NNRTI: EFV |
| ZI9923 | NNRTI: ETR (p), RPV (l) | 29: NNRTI: ETR (p), RPV (l) | PI: DRV RTV NRTI: FTC, TDF |
| NNRTI: ETR (p), RPV (l) | 10: NNRTI: ETR (p), RPV (l) | ||
| MP7985 | NRTI: AZT (i), D4T (l), DDI (p) | 1: NRTI: AZT (i), D4T (l), DDI (p) | NRTI: 3TC, ABC, ZDV |
| none | none | ||
| 701010222-1 | INSTI: EVG (p), RAL (p) | 6: INSTI: EVG (p), RAL (p) | naive |
| INSTI: EVG (p), RAL (p) | 10: INSTI: EVG (p), RAL (p) | ||
| 701010222-2 | INSTI: EVG (p), RAL (p) | 3: INSTI: EVG (p), RAL (p) | naive |
| 703010863-2 | NRTI: 3TC (h), DDI (p), FTC (h), ABC (l) NNRTI: NVP (h), EFV (h) |
1: NRTI: 3TC (h), DDI (p), FTC (h), ABC (l) and NNRTI: NVP (h), EFV (h) | naive |
| NRTI: 3TC (h), DDI (p), FTC (h), ABC (l) NNRTI: NVP (h), EFV (h) |
9: NRTI: 3TC (h), DDI (p), FTC (h), ABC (l) and NNRTI: NVP (h), EFV (h) | ||
| 703011231-2 | INSTI: EVG (p), RAL (p) | 10: INSTI: EVG (p), RAL (p) | naive |
The resistance levels are denoted by high-level resistance (h), intermediate-level resistance (i), low-level resistance (l), and potential low-level resistance (p).
Number of consensus sequences with a particular linkage is presented.
Library preparation
HIV RNA was extracted from plasma specimens, followed by the synthesis of cDNA labeled by a unique molecular identifier (UMI) [9, 10]. The UMI-labeled HIV cDNA was subsequently amplified by either droplet PCR or conventional bulk PCR (see Supplementary Materials) [9, 10]. Samples were then purified with Ampure beads (Beckman Coulter, CA), quantified, and pooled, which was size-selected.
Nanopore MinION sequencing and Pacbio Revio sequencing
The size selected library was sequenced by Oxford Nanopore MinION with the Ligation Sequencing kit V14 (Oxford Nanopore Technologies). The library was then diluted to 39 ng in 12 μl and sequenced via the MinION Mk1b sequencer using the R10.4.1 flow cell (Oxford Nanopore Technologies) for 72 hours. The GPU version of the Guppy basecaller (version 6.5.7) was used with the AI network model dna_r10.4.1_e8.2_400bps_sup, operating in Super Accuracy (SUP) mode [11]. In parallel, the library was sequenced by Pacbio Revio system with a 24-hour run time at DNA Technologies Core, UC Davis Genome Center.
RESULTS
Whole-length pol gene sequencing provides a unique opportunity for detecting linked cross-class resistance mutations, the emergence of which holds pivotal importance in understanding virologic failure within contemporary combination therapy regimens [7]. We assessed the feasibility of acquiring complete pol gene sequences through portable Nanopore sequencing by conducting a direct comparison alongside fixed-site Pacbio sequencing. In the process of cDNA synthesis, each HIV RNA template was labeled with a unique molecular identifier (UMI) [12-14]. Each sample’s UMI-labeled HIV RNAs were then subsequently indexed during the PCR amplification phase targeting the entire pol gene region (HXB2 2085-5096) (Figure 1A). We used both microdroplet amplification and conventional bulk-PCR amplification (Figure 1B) [9, 10]. The pooled multiplexed samples were subjected to long-read high-throughput sequencing, employing both portable Oxford Nanopore sequencing or fixed-site Pacbio sequencing (Figure 1C).
Figure 1. Potable Nanopore sequencing for cross-resistance mutation detection.
A. The template for our full-length pol gene sequencing for HIV drug resistance screening comprises the complete pol gene (HXB2 2085-5096) region (in grey), a unique molecular identifier (UMI) (in red), an insert for PCR amplification (in blue) and forward and reverse sample indexes (in yellow and green). B. UMI-labeled HIV cDNAs were amplified using both microdroplet-PCR and conventional bulk-PCR. C. Both portable Nanopore sequencing (MinION sequencer) and fixed-site Pacbio sequencing (Revio sequencer) were employed for long-read high-throughput sequencing. D. Linked cross-resistance mutations across NNRTIs and INSTIs were detected from participant EM2045 through both portable and fixed-site sequencings. Three distinct types of linkages were identified, predicting cross-resistance to five INSTI drugs, bictegravir (BIC), cabotegravir (CAB), dolutegravir (DTG), elvitegravir (EVG), and raltegravir (RAL), alongside two NNRTI drugs, efavirenz (EFV) and nevirapine (NVP). The resistance levels are denoted by high-level resistance (in red), intermediate-level resistance (in orange), low-level resistance (in brown), and potential low-level resistance (in purple). The number of full-length pol gene sequences with each linkage type is given to the left of the sequence and to the right of the sequence for fixed-site Pacbio sequencing.
Linked cross-resistance mutations between INSTI and NNRTI
Our portable sequencing assay detected linked mutations resistant to multiple INSTIs and NNRTIs within participant EM2045 in the LARSC cohort. A major INSTI resistance mutation, R263K, was linked with two NNRTI resistance mutations, K103N and K238T (Figure 1D). This linkage was identified in three out of the 13 full-length pol gene sequences obtained through portable Nanopore sequencing. Collectively, these linked mutations predicted cross-class resistance across five INSTI drugs – bictegravir (BIC), cabotegravir (CAB), dolutegravir (DTG), elvitegravir (EVG), and raltegravir (RAL) – as well as two NNRTI drugs –efavirenz (EFV) and nevirapine (NVP) (Figure 1D). The two NNRTI mutations were also found to be linked with an additional INSTI mutation, S153F, observed in seven sequences (Figure 1D). Another three sequences showed a connection between two INSTI resistance mutations (S153F and S230R) and two NNRTI resistance mutations (K103N and K238T). The presence of these three distinct types of linkages was confirmed through independent fixed-site Pacbio sequencing (Figure 1D).
Participant EM2045 was predicted to be resistant to both BIC and DTG, which aligned with this patient’s prior and ongoing ART regimens that included these drugs. Conversely, clinical HIV genotype testing conducted on the same sample, reliant on Sanger sequencing, conducted on the same sample, only detected two NNRTI mutations (K103N and K238T) and failed to identify any INSTI-resistant mutations, despite screening the entire IN gene (QuestDiagnostics HIV-1 Genotype and HIV-1 Integrase Genotype). This inability to predict resistance to BIC and DTG underscores the limitations of current HIV genotype testing in clinical care. This case clearly indicates that our portable sequencing solution can enhance the resolution of HIV drug resistance testing in clinical practice.
High-throughput capacity of portable Nanopore sequencing
We demonstrated the high-throughput capacity of portable Nanopore sequencing by pooling 49 specimens from 37 individuals for a single MinION run. This yielded a total of 81,796 raw reads, from which 505 consensus sequences were derived by error filtration and consensus sequence building of raw reads with the same UMI. The observed capacity suggested the viability of cost-effective full-length pol gene sequencing, enabling comprehensive screening of HIV drug resistance mutations across the key ART classes currently assessed in clinical practices – PI, NRTI, NNRTI and INSTI.
We produced drug resistant mutation reports from 58 specimens we screened, as presented in Table 1. False positive drug resistance mutations can arise when mixed bases are observed at drug resistance mutation sites among reads sharing the same UMI as these observations may indicate errors linked to DNA polymerase mutations [15] or heteroduplex formation [16]. Consequently, sequences with minor drug resistance mutations were excluded when mixed bases were detected at those positions, aiming to minimize the likelihood of false-positive drug resistance mutations. Our portable Nanopore sequencing detected drug resistance mutations in 11 specimens from the 58 specimens we screened, and these findings were subsequently corroborated through fixed-site Pacbio sequencing (Table 1). All drug resistance mutations identified through portable sequencing were cross-validated using fixed-site Pacbio sequencing, and reciprocally, all drug resistance mutations identified through Pacbio sequencing were detected by Nanopore sequencing, except for one minority mutation P145S with a frequency of 29% from participant 703011634 (Table 1). We also identified linked cross-resistance mutations between NRTI and NNRTI from ART-naïve participant 703010863 in the CHAVI cohort (Table 1).
Accuracy validation of portable Nanopore sequencing
HIV drug resistance mutation screening demands highly precise sequencing, as it necessitates the detection of substitutions at the single-base level. To establish the viability of portable Nanopore sequencing for detecting drug resistance mutations, we used the most up-to-date Nanopore sequencing flow cell (R10.4.1), the Ligation Sequencing kit (v14), and the artificial-intelligence (AI)-assisted Guppy basecaller in the Super Accuracy (SUP) mode (version 6.5.7) [11]. In addition, we collected raw reads sharing the same UMI and built the consensus sequence from these to further enhance the accuracy.
We validated the accuracy of our portable Nanopore sequencing pipeline by producing highly accurate sequencing data via an independent fixed-site Pacbio sequencing. The precision of Pacbio sequencing was previously corroborated through Sanger reference sequencing [10, 17]. For the purpose of this comparative analysis, we conducted parallel side-by-side sequencing of our pooled samples: one stream for portable Nanopore sequencing and another for fixed-site Pacbio sequencing. Among the 505 consensus sequences acquired through portable Nanopore sequencing after error filtering, a total of 472 sequences were matched with their respective UMIs from sequences obtained through Pacbio sequencing, spanning a cumulative 1,421,402 bases (Figure 2A). In consensus sequences obtained through Pacbio sequencing, a mixed base was marked when the frequency of the second most abundant base exceeded 10%. Only three base errors – one mutation and two insertions – were observed, resulting in an error rate of 2.1× 10−6 errors per base (Figure 2B). All of these errors were then corrected by our local realignment algorithms, resulting in a sequencing outcome free from errors (Figure 2C).
Figure 2. Accuracy validation of portable Nanopore sequencing.
A. Consensus sequence of raw reads with the same UMI, obtained by portable Nanopore sequencing, was compared with the consensus sequence having the corresponding UMI obtained from fixed-site Pacbio sequencing. B. One mutation and two insertions were observed across 474 consensus sequences with a total of 1,421,402 bases (equating to an error rate of 2.1× 10−6 errors per base) before implementing our local realignment algorithms. After the application of these algorithms, all of these errors were corrected. C. Local realignment of raw reads from participant KI6633 removed an insertion error, fixing the consensus sequence from ‘AAAAAAGGAA’ to ‘AAAAAAGAA’.
Reproducible drug resistance mutation reports
We conducted additional independent portable sequencing runs to assess the reliability of drug resistance mutation reports. We reprocessed 15 samples commencing with HIV RNA extraction with UMI labeling, followed by both micro-droplet and regular PCR amplifications. The resulting pol gene sequences from each of 15 samples were then acquired through additional portable Nanopore sequencing runs. Among the samples subjected to reprocessing, consistent resistance profiles were observed in 12 samples (Table 1). Specifically, eight samples displayed no detected resistance mutations across different sequencing runs. Notably, for participant ZI9923, two separate runs detected the same resistance mutation, E138A (NNRTI), with frequencies of 64% and 53%, respectively, indicative of resistance to etravirine (ETR) and rilpivirine (RPV). From participant 701010222, an INSTI mutation, E157Q, was identified with 100% prevalence across both independent portable sequencing runs. Likewise, for the second sample from study participant 705010136, L74M was detected with 75% and 100% prevalence, respectively. In cases where discrepancies arose between the two portable sequencing runs, the resistant mutations were not present at 100% frequency. These results validate the reliability and repeatability of our potable sequencing approach.
Maximum likelihood tree analysis
A phylogenetic tree of 857 full-length pol gene sequences that were obtained via portable Nanopore sequencing was presented in Figure 3, along with the HXB2 reference sequence. The tree’s topology indicated that subtype-B pol gene sequences from samples collected in the United States showed proximity to each other and to the HXB2 sequence. Conversely, pol gene sequences of subtypes A, C, and D from samples collected in Malawi and South Africa displayed greater divergence from the HXB2 sequence, as anticipated. Additionally, the tree highlighted the clustering of pol gene sequences from the same study participant, underscoring the reliability of portable sequencing. We also quantified the diversity of completed pol gene sequences (calculated as the average pairwise nucleotide difference per length) obtained from each individual sample, as presented in Table 1. The range of intrahost diversity spanned from 0% to 1.37%, consistent with earlier findings of pol gene diversities measured from single genome sequences [18]. This congruence in quantitative outcomes also attests the accuracy of Nanopore sequencing with the UMI labeling strategy.
Figure 3. Maximum likelihood tree.
Maximum likelihood tree of 857 full-length pol gene sequences, generated through portable Nanopore sequencing. These sequences were obtained from 51 samples of 38 ART-naïve individuals in the CHAVI cohort (9-character ID) and 7 samples of 7 ART-positive individuals from the LARSC cohort (6-character ID) with the HXB2-pol sequence. The pol gene sequences were aligned by MAFFT (version 7.392) [20], and the tree was constructed using FastTree (version 2.1.8) [21], and visualized with FigTree (version 1.4.4). The samples originated from three different countries: Malawi (M), South Africa (SA), and the United States (US), as denoted within parenthesis. Starting from the HXB2 position (center) and proceeding clockwise, each study participant’s sequences were highlighted in a uniquely colored box. The subtype of each participant’s sequences was denoted by the first character within parentheses, as annotated by the Stanford University HIV Drug Resistance Database. All sequences from the United States were classified as subtype B, while most sequences from Malawi and South Africa sequences were subtype C, except for subtype D for participant 703010677’s sequences and subtype A for participant 703011165’s sequences.
Meta-analysis on drug resistance profiles
Our two cohorts showed contrasting drug resistance mutation profiles associated with ART status, which we compared with publicly accessible pol gene sequences that were collected from the same time periods and geographic locations. These GenBank data consisted of 14,109 Protease (PR), 14,446 Reverse Transcriptase (RT), and 3,471 Integrase (IN) gene sequences (Supplementary Tables 2 and 3). Figure 4A showed the prevalence of each drug resistance mutation among sequences collected from either ART naïve or ART positive individuals in Malawi (2007-2013), South Africa (2007-2013), the United States (2012-2013), and the United States (2018-2022) (Supplementary Table 4).
Figure 4. Meta-analysis on drug resistance mutation profiles.
A. The prevalence of drug resistance mutations within the PI, NRTI, NNRTI, and INSTI classes was evaluated across seven distinct groups (Supplementary Table 4). All the sequences used in this meta-analysis are accessible on GenBank, and their accession numbers were provided in Supplementary Table 3. The initial two groups were sequences from Malawi, collected between 2007 and 2013, originating from ART-negative and ART-positive individuals, respectively. The subsequent two groups comprised sequences from South Africa from the same timeframe, categorized into ART-negative and ART-positive individuals. The fifth and sixth groups were sequences from the United States, obtained from individuals who were ART-negative and ART-positive between 2012 and 2013. The final group consisted of sequences from ART positive individuals, collected during the period from 2018 to 2022. Sequences from ART negative individuals in the United States between 2018-2022 were unavailable on GenBank. Integrase gene sequences from the ART positive groups in both Malawi and South Africa were unavailable and were represented as white. A total of 3,326 Protease (PR), 3,251 Reverse Transcriptase (RT), 3,203 Integrase (IN) gene sequences originated from ART naïve individuals and 10,783 PR, 11,195 RT, 268 IN gene sequences were sourced from ART positive individuals (as detailed in Supplementary Table 2). B. The frequency of sequences within each of the seven groups, showing intermediate or high-level resistance to individual drugs within the PI, NRTI, NNRTI, and INSTI classes (Supplementary Table 5). Drug abbreviations include: atazanavir/ritonavir (ATV/r), darunavir/ritonavir (DRV/r), fosamprenavir/ritonavir (FPV/r), indinavir/ritonavir (IDV/r), lopinavir/ritonavir (LPV/r), nelfinavir (NFV), saquinavir/ritonavir (SQV/r), tipranavir/ritonavir (TPV/r), lamivudine (3TC), abacavir (ABC), zidovudine (AZT), stavudine (D4T), didanosine (DDI), emtricitabine (FTC), tenofovir disoproxil fumarate (TDF), doravirine (DOR), efavirenz (EFV), etravirine (ETR), nevirapine (NVP), rilpivirine (RPV), bictegravir (BIC), cabotegravir (CAB), dolutegravir (DTG), elvitegravir (EVG), and raltegravir (RAL).
The prevalence of drug resistance mutations in our Malawi ART-naïve cohort was relatively low, amounting to less than 5%. This observation aligned with publicly available data, wherein the prevalence of most pretreatment drug resistance mutations was below 2% in Malawi between 2007 and 2013. However, the E138A (NNRTI) mutation showed a higher prevalence of 23.8% among 357 public RT gene sequences. This mutation was previously reported to have a high transmission rate in Malawi between 2012 and 2014 [19] (Supplementary Table 4).
We identified only L74M (INSTI) mutation from four ART-naive individuals in South Africa. This mutation’s prevalence was 2.0% among 1,314 IN gene sequences that were collected from ART naïve individuals in South Africa between 2007 and 2013 (Figure 4A and Supplementary Table 4). A higher prevalence of pretreatment NNRTI resistant mutations, including K103N (4.1%), V106M (4.0%), and Y181C (2.1%), were observed in 1,898 RT gene sequences in South Africa, compared to the prevalence in Malawi.
We identified an intermediate-level resistance mutation, R263K, to second-generation INSTIs, in an individual from the Unites States. The prevalence of this mutation was 1.5% among 268 IN gene sequences collected from ART-treated individuals between 2012 and 2013 and 0.3% among 2,686 sequences between 2018 and 2022 in the United States (Figure 4A). The prevalence of several other INSTI resistance mutations, E138K, G140S, E157Q and G163R, showed an increase from years 2012-2013 to 2018-2022 in the United States (Figure 4A and Supplementary Table 4).
DISCUSSION
We demonstrated that portable Nanopore sequencing presents a viable solution for conducing comprehensive and efficient HIV drug resistance mutation testing. We detected linked multidrug cross-resistance mutations between INSTI and NNRTI within a participant receiving bictegravir (BIC) treatment. To our knowledge, this marks the first observation of co-existing mutations resistant to both INSTIs and NNRTIs within a single virus within an individual failing an INSTI regimen.
Our portable sequencing assay showed the capacity for extensive parallel sequencing of plasma specimens. This high-throughput capability, devoid of the necessity for costly next-generation sequencer, holds the potential to substantially reduce testing costs, thereby facilitating the incorporation of next-generation genomics into HIV clinical care. The total per-specimen supply cost for our portable sequencing assay, including reagents and consumables, is less than $40, increasing its viability for routine implementation in clinical practice [9]. Furthermore, the high-throughput functionality can advance large-scale HIV drug resistance surveillance operations.
We validated the accuracy of portable Nanopore sequencing by two steps: i) a direct comparison of UMI-labeled full-length pol gene sequences with those produced by independent Pacbio sequencing and ii) reprocessing the same samples. However, beyond sequencing errors, we observed aligned base substitutions among reads sharing the same UMI, which may indicate errors linked to DNA polymerase mutations [15] or heteroduplex formation [16]. Consequently, we needed to exclude minor sequences containing mixed bases at locations associated with drug resistance, in order to minimize the potential for false positive drug resistance mutations. This exclusion could potentially lead to reduced sensitivity in detecting minority mutations. Another limitation of our study is that our assay was not comprehensively tested with samples with low viral loads. Despite the lowest viral load of 501 copies/ml, the median viral load among the 58 samples we examined was 48,710 copies/ml. In this study, we conducted both microdroplet amplification and bulk amplification. While microdroplet amplification showed a greater capacity to recover more unique sequences than bulk PCR [9], it is important to systematically compare these two methods across a broad panel of samples with diverse viral loads.
In summary, we developed a portable sequencing tool for HIV genotype testing and validated its detection power and accuracy using both ART-naïve and ART-positive cohorts. This tool can advance HIV drug resistance screening both for individual care and population surveys.
Supplementary Material
Highlights.
We developed a HIV genotype testing pipeline that centers on a cost-efficient portable Nanopore sequencer.
A total of 472 HIV consensus sequences, each tagged with a unique molecular identifier, were produced from over 1.4 million bases acquired through portable Nanopore sequencing, which matched those obtained independently via Pacbio sequencing.
We first documented the linkage of multidrug cross-resistance mutations across Integrase Strand Transfer inhibitors (INSTIs) and Non-Nucleoside Reverse Transcriptase Inhibitors (NNRTIs) from an individual failing a second-generation INSTI regimen.
By generating over 500 full-length HIV pol gene sequences in a single portable sequencing run, we lowered the total per-specimen supply cost for our portable sequencing assay to below $40.
Acknowledgements
We thank Dr. Alan Perelson for reviewing the manuscript and providing helpful comments. We thank Dr. Barton Haynes for providing specimens from the Center for HIV/AIDS Vaccine Immunology (CHAVI) cohort. We thank the study participants of the Los Angeles General Medical Center Rand Schrader Clinic (LARSC) cohort and the Center for HIV/AIDS Vaccine Immunology (CHAVI) cohort. We thank Mr. Luis Mendez for support in the recruitment of study participants at the Los Angeles General Medical Center Rand Schrader Clinic (LARSC). We thank the DNA Technologies and Expression Analysis Core at the UC Davis Genome Center, supported by NIH Shared Instrumentation Grant 1S10OD010786-01. We acknowledge the Center for HIV/AIDS Vaccine Immunology (CHAVI) grant AI067854 from the Division of AIDS, NIAID, NIH. This study was supported by NIH grant R01-AI095066.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Competing Interests
All authors declare no competing interests.
Supplementary Materials
Supplementary materials of this article can be found in the online version.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data Availability
The HIV pol gene sequences reported in this study are available in the GenBank (accession numbers: OR572515 - OR573371).
REFERENCES
- 1.Tang MW, Shafer RW. HIV-1 antiretroviral resistance: scientific principles and clinical applications. Drugs. 2012;72(9):e1–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kantor R, Gupta RK. We should not stop considering HIV drug resistance testing at failure of first-line antiretroviral therapy. Lancet HIV. 2023;10(3):e202–e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. https://www.who.int/news-room/fact-sheets/detail/hiv-drug-resistance.
- 4.Bonsall D, Golubchik T, de Cesare M, Limbada M, Kosloff B, MacIntyre-Cockett G, et al. A Comprehensive Genomics Solution for HIV Surveillance and Clinical Monitoring in Low-Income Settings. J Clin Microbiol. 2020;58(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ellis KE, Nawas GT, Chan C, York L, Fisher J, Connick E, et al. Clinical Outcomes Following the Use of Archived Proviral HIV-1 DNA Genotype to Guide Antiretroviral Therapy Adjustment. Open Forum Infect Dis. 2020;7(1):ofz533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ji H, Sandstrom P, Paredes R, Harrigan PR, Brumme CJ, Avila Rios S, et al. Are We Ready for NGS HIV Drug Resistance Testing? The Second "Winnipeg Consensus" Symposium. Viruses. 2020;12(6). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Boltz VF, Shao W, Bale MJ, Halvas EK, Luke B, McIntyre JA, et al. Linked dual-class HIV resistance mutations are associated with treatment failure. JCI Insight. 2019;4(19). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Siedner MJ, Moorhouse MA, Simmons B, de Oliveira T, Lessells R, Giandhari J, et al. Reduced efficacy of HIV-1 integrase inhibitors in patients with drug resistance mutations in reverse transcriptase. Nature communications. 2020;11(1):5922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Park SY, Faraci G, Murphy G, Pilcher C, Busch MP, Lee HY. Microdrop Human Immunodeficiency Virus Sequencing for Incidence and Drug Resistance Surveillance. J Infect Dis. 2021;224(6):1048–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Faraci G, Park SY, Dube MP, Lee HY. Full-spectrum HIV drug resistance mutation detection by high-resolution complete pol gene sequencing. J Clin Virol. 2023;164:105491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. https://nanoporetech.com/how-it-works/basecalling.
- 12.Boltz VF, Rausch J, Shao W, Hattori J, Luke B, Maldarelli F, et al. Ultrasensitive single-genome sequencing: accurate, targeted, next generation sequencing of HIV-1 RNA. Retrovirology. 2016;13(1):87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Jabara CB, Jones CD, Roach J, Anderson JA, Swanstrom R. Accurate sampling and deep sequencing of the HIV-1 protease gene using a Primer ID. Proc Natl Acad Sci U S A. 2011;108(50):20166–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dennis AM, Zhou S, Sellers CJ, Learner E, Potempa M, Cohen MS, et al. Using Primer-ID Deep Sequencing to Detect Recent Human Immunodeficiency Virus Type 1 Infection. J Infect Dis. 2018;218(11):1777–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cline J, Braman JC, Hogrefe HH. PCR fidelity of pfu DNA polymerase and other thermostable DNA polymerases. Nucleic Acids Res. 1996;24(18):3546–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Liu J, Song HS, Liu DL, Zuo T, Lu FM, Zhuang H, et al. Extensive Recombination Due to Heteroduplexes Generates Large Amounts of Artificial Gene Fragments during PCR. Plos One. 2014;9(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Park SY, Faraci G, Ward PM, Emerson JF, Lee HY. High-precision and cost-efficient sequencing for real-time COVID-19 surveillance. Sci Rep. 2021;11(1):13669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Maldarelli F, Kearney M, Palmer S, Stephens R, Mican J, Polis MA, et al. HIV populations are large and accumulate high genetic diversity in a nonlinear fashion. J Virol. 2013;87(18):10313–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Rutstein SE, Chen JS, Nelson JAE, Phiri S, Miller WC, Hosseinipour MC. High rates of transmitted NNRTI resistance among persons with acute HIV infection in Malawi: implications for first-line dolutegravir scale-up. AIDS research and therapy. 2019;16(1):5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Katoh K, Misawa K, Kuma K, Miyata T. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 2002;30(14):3059–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Price MN, Dehal PS, Arkin AP. FastTree 2--approximately maximum-likelihood trees for large alignments. PLoS One. 2010;5(3):e9490. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The HIV pol gene sequences reported in this study are available in the GenBank (accession numbers: OR572515 - OR573371).