HIV-1 molecular epidemiology and drug resistance-associated mutations among treatment-naïve blood donors in China

Surveillance of human immunodeficiency virus (HIV) molecular diversity and drug resistance-associated mutations (DRMs) among treatment-naïve blood donors is critical for monitoring viral evolution and blood safety. From 2016-2017, 199 plasma samples were collected from 24 blood centers and confirmed as HIV viral load positive or serologically reactive in National Centers for Clinical Laboratories (NCCL), of which 179 were sequenced and subtyped in the gag, protease (PR)-reverse transcriptase (RT), integrase (IN) and/or envelope (env) regions. DRMs in PR-RT and IN regions were analyzed in Stanford HIVdb Program. The majority of subtypes were circulating recombinant form (CRF) 07_BC (34.6%) and CRF01_AE (32.4%); many unique recombinant forms (URFs) (39, 21.8%) and other rare CRFs were observed in the study. Notably, CRF02_AG and CRF06_cpx strains typically found in Africa were firstly identified amongst Chinese blood donors. DRMs were common, with 28 of 179 (15.6%) specimens carrying DRMs, including the PR N88S and RT K103N mutations, which have been implicated in elevated resistance to antiretroviral drugs. Furthermore, 4 HIV-1 isolates (2.4%, 4/168) had surveillance drug-resistance mutation (SDRM), including 3 nonnucleosidereverse transcriptase inhibitors (NNRTI) SDRMs (1 K101E, 2 K103N) and 1 protease inhibitor (PI) SDRM (M46I). The HIV viral diversity among blood donors observed in this study suggest that ongoing HIV-1 recombination is becoming progressively complex in China, and lots of DRMs found in the study exacerbate the primary drug resistance landscape, which highlight the necessity of timely genotypic drug resistance monitoring and molecular surveillance of HIV-1 among blood donors.


Results
Demographic characteristics of blood donors. A total of 199 blood donors confirmed as HIV-1 seropositive or viral load positive were enrolled in this study, and 179 donations were successfully sequenced. From the 179 plasma samples, 168 gag, 170 IN, 168 PR-RT and 166 env sequences were generated by Sanger methods. Demographic information from all participants is summarized in Table 1. Notably, the majority of study participants were male (92.2%, 165/179), first-time donors (

HIV-1 subtype classification
After initial HIV-1 genotyping by the HIV BLAST tool, REGA HIV-1 Subtyping Tool-Version 3.0 and jpHMM program, the specimen classifications were performed using phylogenetic inference of the gag, PR-RT, IN and env regions ( Fig. 1). HIV-1 subtype was confirmed by the consistent results from the subtyping tools above, between different gene regions. All potential unique recombinant sequences (the sequences with inconsistent subtyping results from the tools above) were further analyzed by SimPlot 3.5.1 software to determine recombination breakpoints (Fig. S1) and subtypes. Recombinant composition of URFs were displayed in Table 2. It is noted that a limitation of the recombinant HIV-1 drawing tool used to generate Fig. S1 does not allow CRF labels other than CRF01_AE or CRF02_AG. Therefore, regions that were classified as a CRF with a strong bootstrap value and branching pattern are labeled as the parental strains for that CRF, including regions where no recombinant breakpoints are present. In addition to subtype B and C sequences, a diverse set of CRFs were identified amongst the sequenced regions, including CRF01_AE, CRF02_AG, CRF06_cpx, CRF07_BC, CRF08_BC, CRF15_01B, CRF52_01B, CRF55_01B, CRF59_01B, CRF65_cpx, CRF67_01B, CRF77_cpx, CRF78_cpx, CRF79_0107, CRF83_cpx, and CRF85_BC. The majority of specimens were classified as CRF07_BC (34.6%, 62/179) or CRF01_ AE (32.4%, 58/179), with URFs being nearly as common as these CRFs (21.8%, 39/179). Although the relative prevalence of each classification varied between geographic regions in China (Number of samples ≥10), URFs were present in all locations (Fig. 2). Notably, several sequences (Shaanxi-001, Shaanxi-015, Shaanxi-017) were identified with rare classifications, CRF02_AG and CRF06_cpx, which are uncommon in China (Figs. 3 and 4). Of these, CRF06_cpx regions were present within two different URF strains, whereas the CRF02_AG sequences did not display evidence of recombination ( Figs. 3 and 4). The three rare recombinant partial-genome maps were shown in Fig. 5. Furthermore, the details of bootscan and similarity analyses among the other 7 rare CRF strains including CRF55_01B, CRF59_01B, CRF65_cpx, CRF67_01B, CRF79_0107 and CRF85_BC subtypes were described in Figs. S2 and S3 respectively.

Discussion
During the past two decades, the HIV-1 epidemic has expanded from high risk groups (injection drug users, men who have sex with men, female sex workers etc.) to the general population, including blood donors 16 . Surveillance of the molecular epidemiology and diversity of HIV amongst blood donors is critical to determining the origin and evolution of HIV-1 variants in China and to prevention of transfusion-transmitted HIV-1 infections. Our study is the most geographically comprehensive epidemiological investigation of HIV among blood donors to date, encompassing 17 provinces and municipalities. The majority of the HIV-1 positive blood donors in the study were males with low educated level aged between 18-35 years. Prevention and screening strategies targeted towards these populations may have a greater impact towards ending the HIV pandemic.
Molecular characterization of the HIV-1 strains circulating within the blood donor population revealed unique patterns in comparison to other groups. Notably, subtype B and CRF08_BC accounted for 5.0% and 1.7% of blood donor infections, respectively, which was lower than the national prevalence reported in the 2006 survey 15 . Furthermore, the majority of subtype B strains (6/9) identified in this study were from Henan province, and 28.6% (6/21) of the HIV-1 strains from Henan blood centers were subtype B. During the mid-1990s, commercial plasma collection in central China (Henan and Shanxi provinces) led to an outbreak of subtype B among blood donors 17 . Since the prohibition of commercial blood collection, the prevalence of subtype B transmitted via blood transfusion has significantly decreased, which is reflected by the low proportion of subtype B among  www.nature.com/scientificreports www.nature.com/scientificreports/ enrolled samples in this study. The absence of donor samples from Yunnan, Guizhou and Sichuan provinces in our study may have contributed to the observed low percentage of CRF08_BC, since CRF08_BC predominates in these provinces 17 . The geographic subtype distribution in Fig. 2 (excluding Qinghai-Tibet region: number of samples ≥10) indicated that CRF01_AE and CRF07_BC were the two main genotypes in each geographic region, which was consistent with previous studies amongst high-risk populations 15,[18][19][20][21][22][23] . The presence of these same strains amongst the local blood donor population was evidence of the expansion of HIV-1 from high-risk groups to blood donor groups. This was consistent with the NHLBI Retrovirus Epidemiology Donor Study-II 9 , which also found many URFs in the Chinese blood donor population, indicating that ongoing HIV-1 recombination is becoming progressively complex in China. Due to the limited sampling of some blood centers and lack of samples from Qinghai-Tibet region in this study, we cannot make conclusions on the genotype distribution among blood donors throughout China. Future research must focus on expanded geographical coverage to get a more comprehensive dataset to improve blood safety, ART strategy and HIV control and prevention in China.
Our study was also consistent with previous work demonstrating that most Chinese URFs consist of CRF01_ AE, B, and C regions ( Table 2) 24,25 . However, the identification of two URFs containing CRF06_cpx regions and an ostensibly pure CRF02_AG infection are novel observations in a Chinese blood donor population. CRF02_AG is a subtype A/G recombinant form endemic to Africa 26 . Since the first identification of CRF02_AG in neighboring Taiwan in 1998 27 , several CRF02_AG variants have been reported in China 28,29 , but CRF02_AG has not previously been found amongst volunteer blood donors in China. Likewise, CRF06_cpx was first reported in Burkina Faso in 1998 and had circulated widely in West African countries 30 . Although a few CRF06_cpx isolates have been found in Beijing, Shenzhen and Hong Kong, this strain has not been reported in volunteer blood donors in China [31][32][33] . It was noteworthy that the CRF06_cpx regions found in Shaanxi blood donors in our study had recombined with CRF01_AE, A3, B and C strains. Therefore, it was likely that the recombination events that gave rise to these CRF06_cpx-containing URFs occurred in China. Complete genome sequencing of these strains and the CRF02_AG isolate will be required to identify all recombination breakpoints and estimate when these strains may have entered the Chinese blood donor population.
Generally, all HIV-1 positive blood donors are presumed to be treatment-naïve and the presence of ART resistant strains in this group is a reflection of the rate of transmitted drug resistance in a population 34 . DRM determinations were analyzed by Stanford HIVdb Program, which was based on subtype B and had biased the results for non-B strains 35 . In the present study, 15.6% (28/179) of HIV-1 infected blood donors had accessory or major DRMs. Notably, Q58E and V179D/E were the most common DRMs to PI and NNRTI respectively in the study, which is consistent with a previous study focused on five blood centers in China 9 . Q58E was identified as the potential low-level resistance (PLLR)-related muation to Nelfinavir (NFV) and low-level resistance (LLR)-related mutation to Tipranavir (TPV) 36 .
In particular, the Q58E DRM may be more common in CRF07_BC strains, which was also consistent with a recent study 9 . The PI major DRMs included M46L, M46I and N88S. M46I/L caused PLLR to many INSTIs among HIV-1 positive individuals, while N88S could result in HLR to Atazanavir (ATV) and NFV, LLR to Indinavir (IDV) and Saquinavir (SQV) [37][38][39] . Q58E and other DRMs (M46L/I, N88S) that confer resistance to PIs were present in our study. About PIs, only Lopinavir (LPV) was included in the Free AIDS Antiretroviral Therapy Manual, and not included in the first-line ART in China 40 , suggesting that either these DRMs were imported or they did not arise from selective pressure during treatment. In contrast, the most common NNRTI DRMs in our   www.nature.com/scientificreports www.nature.com/scientificreports/ study, V179D/E mutations, were observed within a variety of strains, consistent with selective pressure from use of NNRTI in China 40,41 . A combination of RT V179D and K103R found in two samples with HIV-1 infection may synergistically reduce EFV and NVP susceptibility about 10-fold 42 , RT mutations with combination of V179D and K103R were also observed in treatment-naïve individuals in China 43 . The RT K103N mutation found in two strains can reduce EFV and NVP susceptibility by about 20-and 50-fold, respectively 44 .
Drug resistance analysis demonstrated that 2.4% of HIV-1 isolates contained at least one NNRTI (K101E, K103N) or PI (M46I) SDRMs, the overall prevalence of TDR was lower than previous reports in Zhejiang (11.1%) and Shijiazhuang (6.1%) among treatment-naïve HIV-infected individuals 45,46 , but similar to a nationwide cross-sectional survey about prevalence of TDR (3.6%) in 2015 in China 47 . Although the rate of TDR remained relatively low in Chinese blood donors in this study, the detection of 3 major NNRTI mutations and 1 PI mutation underlined the importance of a continuous surveillance of resistance mutations.
Overall, the prevalence of DRMs in South China was higher than other regions. The distribution of HIV-1 with DRMs in this study (Table S1) suggests that the HIV-1 strains isolated from positive blood donors in urban centers such as Beijing, Zhengzhou (Henan provincial capital) and Shenzhen had higher rates of DRMs. Moreover, 75% (3/4) HIV-1 isolates with SDRMs in the study were from Beijing Tongzhou district blood bank. It is possible that increased international travel and immigration in these populations may have contributed to the observed higher rates of DRMs.
In summary, our study characterized increasing HIV-1 diversity and high rates of drug resistance in the Chinese blood donor population, with unique province-level trends observed therein. The main HIV-1 subtypes of blood donors in most provinces were consistent with the local high-risk populations, suggesting that the HIV-1 epidemic has expanded from high risk groups to the general population. Most importantly, the integration of imported CRF02_AG and CRF06_cpx strains into the Chinese blood donor population is further evidence of the newly emerging migration patterns of the global HIV-1 pandemic. Furthermore, lots of DRMs and several TDR were found in treatment-naïve blood donors, underscoring the need for continued molecular surveillance to monitor and appropriately respond to expanding local HIV-1 diversity with diagnostic tests and therapeutics that are effective for circulating strains.

Limitations
Since the prevalence of HIV-1 among Chinese blood donors remains low and not all the HIV-1 positive donations in blood screening laboratories were enrolled in our study, the limited sample sizes from several blood screening laboratories may bias the molecular epidemiological results. Furthermore, the socio-demographic data of the HIV-1 infected blood donor lack possible mode of transmission which would be used for analysis of the risks of HIV-1 transmission. For HIV-1 subtype analysis, HIV-1 genome sequencing including env, pol and gag genes is most reliable for subtype classification, but it's hard to get HIV-1 genome sequence, due to the long length of genome sequence, low viral load in several samples.
Future research must focus on expanded geographical coverage and HIV-1 genome sequences to get a more comprehensive dataset to improve blood safety, ART strategy and HIV control and prevention in China.

Materials and Methods
Study samples. From January 2016 to December 2017, a total of 199 blood donations collected from 24 blood screening laboratories were confirmed as HIV viral load positive by the Abbott RealTime HIV-1 (Abbott Molecular Diagnostics, Des Plaines, IL, USA) test or serologically reactive by the Abbott ARCHITECT HIV Ag/ Ab Combo test (Abbott Diagnostics, Weisbaden, Germany). These samples were tested non-reactive for Hepatitis B surface antigen (HBsAg), antibody to Hepatitis C Virus (anti-HCV) and antibody to treponema pallidum (anti-TP) in blood screening laboratories. Of these, at least two HIV regions were successfully sequenced for 179 plasma samples (Supplementary materials Table S1). Geographical localizationof blood screening laboratories in the study and the number of blood donations in these laboratories from January 2016 to December 2017 were shown in Fig. S4. The other three regions were amplified following the same PCR conditions, except with an annealing temperature of 55 °C. The fist-round PCR products for four regions (5 µL) along with their respective inner primers were used in the nested PCR 48 . Nested PCR for pol-PR-RT was conducted with one cycle at 94 °C for 2 min, followed by 40 cycles at 94 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min, and finally an extension of 10 min at 72 °C. Nested PCR for gag, IN and env genes followed the same procedure but with an annealing temperature of 50 °C. The nested PCR products were purified and sequenced by Sangon Biotech (Shanghai) Co., Ltd using Sanger methods. Details of the primers used in the study are described in Supplementary materials Table S2. HIV-1 genotype and phylogenetic analysis. The sample sequences were edited and aligned by Geneious 9.1.2 (https://www.geneious.com/products/prime/resources/download/previous-versions). Gag and two pol gene sequences covering PR and part of RT and the entire IN were submitted to the Los Alamos HIV BLAST tool (2020) 10:7571 | https://doi.org/10.1038/s41598-020-64463-w www.nature.com/scientificreports www.nature.com/scientificreports/ for initial HIV-1 subtyping (https://www.hiv.lanl.gov/content/sequence/BASIC_BLAST/basic_blast.html) 49 and then analyzed by REGA HIV-1 Subtyping Tool-Version 3.0 (http://dbpartners.stanford.edu:8080/RegaSubtyping/ stanford-hiv/typingtool) 50 and jpHMM program (http://jphmm.gobics.de/submission_hiv.html) 51 . Previous reports support utilizing the env, gag and pol regions for reliable subtype assignment [52][53][54] . The sequences were aligned with HIV-1 reference sequences (Accession numbers: Table S3) obtained from the Los Alamos database (https://www.hiv.lanl.gov) and then the nucleotide alignments were used to build phylogenetic tree for further HIV-1 subtyping by MEGA 7.0.2 (https://www.megasoftware.net) using the neighbor-joining algorithm based on Kimura 2-parameter model in 1000 bootstrap replicates 55 . The final subtype of HIV-1 isolate was confirmed by the consistent results from all the subtyping tools above. Boot-scanning and intra-genomic breakpoints analyses were conducted on sequences with possibly unidentified recombinant strains (the sequences with inconsistent results from all the subtyping tools above) through SimPlot 3.5.1 (https://www.softpedia.com/get/Science-CAD/ SimPlot.shtml) 56 . Sequences with recombinant patterns that did not match established CRFs were classified as URFs 6,57 . The genome maps of URFs were generated through Recombinant HIV-1 Drawing Tool (https://www. hiv.lanl.gov/content/sequence/DRAW_CRF/recom_mapper.html). For finalization of phylogenetic classifications, neighbor-joining phylogenetic trees were prepared using PHYLIP 3.5 as previously described 58

Ethics approval and consent to participate. The study was approved by the Ethics Committee in Beijing
Hospital (Ethics board approval number: 2016BJYYEC-118-01). Ethics statement in Chinese shown in supplementary material. The methods in the study were in accordance with the guidelines of the Declaration of Helsinki. Written informed consent was obtained from all subjects before blood donation.

Data availability
The data for this study is available from the corresponding author on reasonable request.