Deep Sequencing: Becoming a Critical Tool in Clinical Virology

Abstract

Population (Sanger) sequencing has been the standard method in basic and clinical DNA sequencing for almost 40 years; however, next-generation (deep) sequencing methodologies are now revolutionizing the field of genomics, and clinical virology is no exception. Deep sequencing is highly efficient, producing an enormous amount of information at low cost in a relatively short period of time. High-throughput sequencing techniques have enabled significant contributions to multiples areas in virology, including virus discovery and metagenomics (viromes), molecular epidemiology, pathogenesis, and studies of how viruses to escape the host immune system and antiviral pressures. In addition, new and more affordable deep sequencing-based assays are now being implemented in clinical laboratories. Here we review the use of the current deep sequencing platforms in virology, focusing on three of the most studied viruses: human immunodeficiency virus (HIV), hepatitis C virus (HCV), and influenza virus.

1. Introduction

DNA sequencing has evolved considerably during the last 50 years. The first sequence of a tRNA molecule in 1965 [1] led to the development of more robust DNA sequencing methodologies in 1977 by Walter Gilbert and Allan Maxam (chemical degradation) [2] and Frederick Sanger (chain-termination) [3]. Sanger et al. determined the first complete genome sequence of any organism: the bacteriophage ϕ [3] which was followed by the DNA sequencing of a multitude of organisms, including DNA and RNA viruses, such as Epstein-Barr virus in 1984 [4] and Human Immunodeficiency Virus in 1985 [5]. Thirty years later, more than 2.5 million viral nucleotide sequences have been deposited in GenBank (https://www.ncbi.nlm.nih.gov/genbank/). This information has been instrumental to understand the structure of viral genomes, their biology, evolution, diversity, transmission, pathogenesis, as well evasion from the host immune response, antiviral drugs, and vaccines. Despite the impressive success of “Sanger” or “population” sequencing, and the development of automated DNA sequencing instruments in the mid 1980s, these first-generation sequencing methods are not high-throughput, have limited scalability and are not cost-effective for sequencing a multitude of samples and/or large genomes. A second generation of sequencing technologies has now addressed these limitations.

Following the development of several novel methods for DNA sequencing in the late 1990s, early 2000s [6, 7], different technologies have become commercially available for so-called next generation sequencing (NGS), also known as “second generation”, “massive parallel” or “deep” sequencing. Deep sequencing technologies are able to generate three to four orders of magnitude more information than Sanger sequencing and are significantly less expensive, considering the cost per nucleotide sequenced [8, 9]. Several deep sequencing systems have been developed during the last 10 years, each of them with their own intrinsic performance metrics such as number of reads obtained, read length, accuracy, time to run, cost, etc. [9, 10]. In 2004 454 Life Sciences (Branford, CT) introduced the first instrument based on pyrosequencing [11], prior to being acquired by Roche (Basel. Switzerland). This was followed by the release of the Genome Analyzer (GA, Solexa, Chesterford, UK) in 2005 [12], now Illumina (San Diego, CA). In 2007 the first SOLiD sequencing system was released by Applied Biosystems (Foster City, CA) [13], while the Helicos sequencer [14] and the Ion Torrent Personal Genome Machine [15], were commercialized by Life Technologies (Carlsbad, CA) in 2009 and 2011, respectively. Pacific Biosciences (Menlo Park, CA) introduced the single molecule real-time sequencer in 2011 [16] and more recently Oxford Nanopore Technologies (Oxford, UK) released an ultra-long single molecule sequencer [17]. All these methodologies have continued to evolve, with new chemistries and more potent instruments, resulting in impressive levels of sequencing throughput at constantly lower costs. As a result, the use of deep sequencing continues to expand both in research and clinical settings, and the virology field is no exception [18–20]. In this review we provide an overview of the use of deep sequencing in virology and how these techniques have impacted, and will continue to influence, the study of three of the most important human viral agents that top the list of viral nucleotide sequences in GenBank, i.e., human immunodeficiency virus (HIV), hepatitis C virus (HCV), and influenza virus.

2. Current deep sequencing platforms

Many excellent articles have compared and described on detail the different deep sequencing methodologies and instruments, including methods used in template preparation, sequencing, and data analysis [9, 22]. Four platforms dominate the deep sequencing field: 454™ (454 Life Sciences/Roche, Branford, CT) [11], Illumina® (Illumina, Inc. San Diego, CA) [12], Ion Torrent™ (Ion Torrent/Life Technologies, South San Francisco, CA) [15], and PacBio® (Pacific Biosciences, Menlo Park, CA)[16]. In general, all four technologies are able to generate valuable sequence information [26]; however, there are key and significant differences between the amount and quality of the data, and the applications that each system could support (Fig. 1). Selecting a deep sequencing platform depends on the potential application(s) and resources available, including cost of the instrument and reagents, existing infrastructure and personal experience. Although to date most of the published studies in virology have used 454™ or Illumina® systems, perhaps due to the fact that these were the first two methodologies available (Fig. 1), all deep sequencing technologies continue to improve and are being used in a multitude of virological studies [18–20, 24].

Figure 1.

Principal characteristics of the four most used deep sequencing platforms to date: 454™ (GS Junior and GS FLX+ systems), Illumina® (MiSeq v2 and HiSeq 2500 systems), Ion Torrent™ (Ion Personal Genome Machine®, 318 v2 chip and Ion Proton™, I chip systems), and Pacific Biosciences® (PacBio RS II SMRT®). Information about virus-related publications was obtained from the respective companies (personal communications and/or websites) and PubMed (http://www.ncbi28nlm.nih.gov/pubmed/) search as of March 27^th, 2014.

3. Applications in general virology

Deep sequencing has been a “shot in the arm” to all genomic studies and virologists have taken particularly advantage of this methodology. First of all, now it is simpler and more affordable than ever to sequence full viral genomes. Likewise, identification and classification of novel and known viruses, unbiased characterization of viral populations without the need for virus culturing (viromes), molecular epidemiology, viral diversity and evolution, transmission and pathogenesis, and in particular medical virology have greatly benefited from the use of deep sequencing. As shown in Figure 2A, the number of publications using deep sequencing in virology has skyrocketed since 2008, particularly those associated with HIV. By allowing the cost-effective study of a greater number of viral variants, including complex viral populations, deep sequencing is the perfect tool for a broad number of applications for studies of HIV, HCV and influenza virus (Fig. 2B).

Figure 2.

Scientific publications based on deep sequencing. (A) Cumulative number of publications associated with HIV, HCV, or influenza virus and deep sequencing during the last 13 years. The size of the circles corresponds to the number of publications in each given year, e.g., 57 HIV studies using deep sequencing were published in 2013. The total number of publications per virus is indicated. (B) Percentage of scientific publications per research area. Search in PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) as of March 27^th, 2014 using “HIV”, “HCV”, “influenza”, “deep sequencing”, “pyrosequencing”, “epidemiology”, “quasispecies”, “pathogenesis”, “transmission”, “drug resistance”, “tropism”, and/or “vaccine” as keywords.

3.1. Virus discovery

The field of virus discovery has flourished with the advent of deep sequencing methodologies. Coupled with bioinformatics tools, high throughput sequencing has revolutionized the field by allowing the identification and characterization of novel viruses [27] including a novel rhabdovirus associated with acute hemorrhagic fever identified in Central Africa [30] and a new cyclovirus detected in cerebrospinal fluid of patients with central nervous system infections [31]. Together with classical methods of diagnostics such as virus isolation, immunochemistry, and PCR, deep sequencing will continue to play a significant role in the identification of novel viruses, particularly in the face of outbreaks of known and/or new diseases.

3.2. Human, animal, plant, and environmental viromes

Viral metagenomic studies, i.e., the characterization of viral genomes directly from samples, have become extremely popular with the arrival of deep sequencing [28] and numerous viral metagenomic analyses have been published (reviewed in [18, 27–29]). The viral metagenome or “virome” refers to the collection of viruses found in a particular sample from humans, animals, plants or from a specific environmental sample. Virome studies can lead to the discovery of new viruses and/or to their association with known or novel diseases. In the case of the human virome, its composition and impact on human health has been the subject of multiple studies [37, 38]. The healthy human gut virome seems to consist mainly of bacteriophages [39], and can vary among individuals and in response to diet [40]. On the other hand, the human skin virome comprises a high variety of DNA viruses [41]. Interestingly, the human gut, nasopharyngeal, or plasma viromes are highly susceptible to changes associated with diseases such as HIV/AIDS [42, 43], antiviral and immunosuppression therapy [44], or respiratory infections [45, 46]. Similar studies have analyzed the gut virome in different healthy animals, such as pigs [47], wild rodents [48] and pigeons [49], and turkeys [50] and, like in humans; certain diseases seem to affect the animal virome [51, 52]. Numerous viruses have been identified as part of the bat virome, including influenza A viruses [53], hepaciviruses and pegiviruses [54], hepadnaviruses [55], hantaviruses [56] and a series of novel viruses such as cedar virus [57] and a new SARS-like betacoronavirus [58]. These animals have been intensely studied since they are believed to be the reservoir of many potential zoonotic infections [59].

3.3. Molecular epidemiology

Sanger-based DNA sequencing and phylogenetic analyses have been instrumental to characterize the distribution of viruses in a population [63]; however, deep sequencing now allows more complex analysis of huge datasets and even full-length viral genomes. Following multiple zoonotic transmissions, which resulted in distinct lineages in humans, HIV has diversified over time [64, 65]. Accordingly, HIV has been classified into types, groups, subtypes or clades, circulating recombinant forms (CRF) and unique recombinant forms (URF) mainly based on the diversity of the viral env and pol genes [65–67]. Although to date most of the deep sequencing studies of HIV have focused on drug resistance and tropism, the methodology has been used to characterize novel HIV recombinants [69] and to determine HIV incidence [70]. Interestingly, deep sequencing has been particularly useful in the detection and characterization of HIV superinfection events, which seem to occur at higher rates than previously identified [71].

Seven phylogenetic clades -designated as genotypes- have been identified in HCV, with close to 70 subtypes distributed among the different genotypes [72]. HCV subtypes are epidemiologically distinct, with differences in risk group targeting and geographical distributions that are associated with substantial genetic diversity that reflects their recent epidemic spread [73]. Thus far the reconstruction of the HCV epidemic has been based on Sanger sequences, used to model evolutionary histories of currently circulating variants and to identify historical factors such as widespread use of blood transfusion and other parenterally delivered treatments and vaccination as the facilitators of HCV transmission [74]. However, deep sequencing approaches have not only accelerated the rate at which HCV sequences are generated, but also represent a substantial advance in sensitivity and molecular resolution to distinguish closely and distantly related HCV genomes. For example, deep sequencing has been used to identify new HCV subtypes and recombinants [75] and to study the history of the HCV epidemic in remote communities [76].

Given the importance of transmission and adaptation of avian influenza viruses, and more recently swine strains for epidemics and pandemics in humans, an important number of studies based on deep sequencing techniques have described avian [77–81] and porcine [82–85] influenza virus evolution. For example, the selection of minority variants with a deletion in the neuraminidase (NA) gene that allows the adaptation of avian influenza viruses from waterfowl to domestic poultry [77], suggests a high frequency of mixed infections and genetic reassortment within these viruses [78, 79, 85]. Other studies described the prevalence and spread patterns of different human influenza viruses in specific geographic areas, based not only on the hemagglutinin (HA) and NA genes but by analyzing all viral segments [86–89].

3.4. Viral diversity, transmission, and pathogenesis

All RNA viruses including HIV, HCV, and influenza virus replicate as a multitude of related but nonidentical genetic variants known as quasispecies [90]. These highly diverse viral populations provide numerous advantages to the virus, including escaping the pressure from the host immune system, and resistance to antiviral agents [91]. Prior to the arrival of deep sequencing, viral quasispecies studies relied on the labor-intensive Sanger sequencing of numerous molecular clones [92, 93]. Today, deep sequencing methodologies are capable of generating an extraordinary number of sequences (reads), which make them the ideal tool to study intra- and inter-host viral diversity, virus transmission and adaptation dynamics, and disease progression (Fig. 3).

Figure 3.

Comparison of phylogenetic analyses based on Sanger or deep sequencing. Neighbor-joining phylogenetic trees were constructed using (A) Sanger sequencing of 105-bp fragments corresponding to the HIV-1 V3 region of gp120 (env gene) from 12 HIV-infected individuals or (B) deep sequencing reads with a frequency >1 corresponding to the same 105-bp fragments (Gibson and Quiñones-Mateu, unpublished results). Each color-coded dot represents a unique variant, frequency is not depicted. Bootstrap resampling (1,000 data sets) of the multiple alignments tested the statistical robustness of the trees, with percentage values above 75% indicated by an asterisk. s/nt, substitutions per nucleotide. Phylogenetic trees were constructed using MEGA 5.05 [261].

Recent strategies to study HIV variability have involved deep sequencing of nearly complete viral genomes [94–96] or specific genomic regions [26, 97–107]. Other studies have focused on the analysis of HIV intra-patient evolution, such as the characterization of transmitted HIV and persistence of minority variants [108], estimation of primary infection dates [109], description of intra-host evolution dynamics during the course of infection with or without antiretroviral treatment [95, 98, 103, 107, 110, 111]. Others have analyzed hypermutation patterns [112], viral evolution in different host compartments [113] or in response to the host immune system [114–116], and simultaneous assessment of replication fitness of different drug resistant variants [117]. In the case of HCV, deep sequencing has allowed the study of minority viral variants and intra-host quasispecies diversity [118, 119], the identification of transmitted viral variants [120, 121], and factors associated with HCV pathogenesis. In addition, deep sequencing-based studies allowed the identification of a significant number of non-coding RNAs (ncRNAs) associated with hepatocellular carcinoma caused by HCV infection [123].

Similar to HIV and HCV, deep sequencing provides a better alternative to dissect genetically complex populations of influenza viruses and to detect minority variants with clinical or epidemiological relevance. Deep sequencing-based methods have recently been proposed for the assessment of influenza A viruses diversity and rates of evolution [124–126], as well as antigenic stability [127] using complete influenza A genomes and exploiting the ability to detect and quantify mutations in heterogeneous viral populations. Others have focused on the evolution of avian influenza strains with potential to become pandemic in humans [128–132], as well as the detection of virulence signatures [133] and reassortment patterns [134]. Many of these studies have aimed at understanding the transmission and adaptation of avian viruses to humans, which could lead to a better preparedness for a potential influenza pandemic.

4. Deep sequencing in the clinical setting

4.1. Drug resistance

HIV, HCV, or influenza virus susceptibility to antiviral drugs has been traditionally assessed using phenotypic assays (i.e., evaluate the ability of the virus to replicate in the presence of the drug) or detecting drug resistance mutations using Sanger sequencing [24, 151]. Despite longer turn-around-time and higher cost, phenotypic assays can assess susceptibility to any antiviral drug without any prior knowledge of the viral sequence. On the other hand, many HIV- mutations associated with reduced susceptibility to antiviral drugs have been identified [152]. Several Sanger-sequencing based HIV genotypic assays have been developed, while a few are commercially available [24]. Unfortunately, Sanger sequencing can only detect minority variants present at frequencies above 15% to 20% of the viral quasispecies [153–157] and thus fail to quantify low-level HIV drug resistant variants. Using deep sequencing, a series of studies have demonstrated the ability to detect minority HIV drug resistant variants at levels as low as 0.1% to 1% of the virus population [26, 106, 158–162]. Most of these studies used the 454™ platform [106, 158, 159, 163, 164], but recently other deep sequencing technologies such as Illumina® [165] and Ion Torrent™ [162, 166] have proved to be useful in the identification of low-level drug resistant HIV variants. Deep sequencing has been used in HIV drug resistance surveillance [167, 168], to study transmission of HIV drug resistant viruses [101, 160, 169–175], and to evaluate the impact of minority variants on treatment efficacy [160, 176–183].

During the last decade treatment of chronic hepatitis C has been based exclusively on the combination of pegylated(peg)-IFN-α and ribavirin (RBV), however, the characterization the HCV lifecycle has led to the identification of a number of potential new targets for direct-acting antiviral (DAA) drugs [184, 185]. Several DAA drugs have been approved for the treatment of HCV infection and many more are in preclinical or clinical development phases [184]. Similar to HIV, reduced susceptibility to DAA drugs has been reported [186–188] and needs to be monitored. Deep sequencing analysis of the HCV NS3 protease in DAA-naïve patients showed that over 80% of the patients carried viruses with at least one DAA resistance mutation [189, 190]. Other studies have shown minority variants that are resistant to inhibitors targeting other HCV genomic regions [75, 190, 191] or have used deep sequencing to characterize the dynamics of drug-resistant variants during and after treatment [192, 193].

Multiple studies have focused on the development of deep sequencing-based assays to determine drug resistance in different influenza virus subtypes, including highly pathogenic avian viruses like H7N9 and H5N1, that could be used in the clinical setting [194–210]. Some have described the role of minority variants resistant to treatment with oseltamivir [211–213]. Transmission of oseltamivir-resistant minority variants prior to the onset of therapy allowed the rapid development of full-blown resistance once treatment started [212], showing the importance of the use of highly sensitive methods to detect drug resistance-associated polymorphisms in the clinical setting. Other studies have focused on surveillance of drug resistance-associated mutations for both NA inhibitors [201, 214–230] and adamantanes [231–234].

4.2. HIV tropism

The discovery that HIV requires a coreceptor to enter target cells, mainly the chemokine receptors CCR5 and CXCR4 [235–237], prompted the development of novel anti-HIV molecules targeting these receptors, as represented by the CCR5-receptor antagonist maraviroc (Selzentry/Celsentri, Pfizer, NY)[238]. Treatment with CCR5 antagonists requires the prior knowledge of the HIV coreceptor tropism in the patient, which led to the development of numerous phenotypic and genotypic approaches to determine HIV coreceptor usage or tropism [24, 239–241]. While a phenotypic assay, a Trofile (Monogram Biosciences)[242, 243] is currently the standard method to determine HIV-1 coreceptor tropism in the U.S., in-house genotypic HIV-1 tropism tests are largely used in Europe [244, 245]. Both approaches have advantages and disadvantages; however, the lack of sensitivity of Sanger sequencing-based assays has led to the use of deep sequencing to increase the ability of genotypic assays to detect minority CXCR4-tropic variants [26, 162, 246–258]. Thus, more sensitive genotypic HIV tropism assays based on deep sequencing have been developed to detect non-R5 variants below 20% of the population in the clinical setting, which have been shown to correlate well with other HIV-1 tropism tests [162, 246, 247, 250]. Finally, using deep sequencing researchers have been able to detect CXCR4-tropic viruses during early HIV infection [257], to study the mutational pathway of the V3 region in HIV during the transition from CCR5 to CXCR4 usage [259], and to identify low-levels of CXCR4-tropic provirus [260].

4.3. Implications of deep sequencing in clinical management

As described above, sequencing-based assays are being increasingly used in the clinical setting and clinical virology laboratories are not the exception. Is deep sequencing going to eventually replace Sanger/population sequencing in clinical laboratories? The short answer may be yes; however, a series of barriers need to be cleared before these assays become fully established. Although historically high costs have been a major hurdle with these methodologies, this can be addressed by (i) batching and multiplexing samples in single sequencing runs and (ii) the continuous decrease in prices of both reagents and instruments. Access to relatively sophisticated technical expertise to perform and analyze deep sequencing data, including bioinformatics solutions, is another key requirement. All this, together with robust implementation and validation processes, is essential to guarantee the success of laboratory-developed tests (LDT) based on deep sequencing in a CLIA-certified environment (to our knowledge, no deep sequencing-based assay has received IVD approval from the U.S. Food and Drug Administration to be used in clinical virology).

Finally, assay sensitivity, error rate, and data reporting and interpretation by clinicians are critical features of any deep sequencing-based test. The intrinsic error (amplification and sequencing) rate of deep sequencing LDTs needs to be calculated, which defines the analytical sensitivity of the assay. In the case of HIV, the two commercially available tests are able to reliable detect minority variants at frequencies as low as 1% of the virus population [162, 250]. Still, what is the clinical significance of these low-abundance variants? HIV genotyping tests based on Sanger sequencing have been used during the last 25 years to make clinical decisions; however, these tests can only detect minority variants present above approximately 20% of the virus population [153, 154]. Thus, all the current knowledge related to antiretroviral treatment monitoring has been based on the detection of variants at this threshold. It is reasonable to assume that under the appropriate selection (i.e., drug pressure), minority variants present at low frequencies (1% to 20%, detected only by deep sequencing) will eventually outcompete other members of the viral population; however, the clinical relevance of these minority members of the viral population is still under debate [176–181]. Additional studies will needed to elucidate if early detection of these minority variants could influence strategies to control the growth of HIV and other viruses in the clinical setting.

5. Conclusions

There is no doubt that the use of deep sequencing in virology will continue to grow exponentially. We may not have the perfect deep sequencing platform yet, that is, one that combines low error rates, with long reads, and relatively low cost. In addition, the methodology still requires considerable technical expertise, including strong bioinformatics capability. Nevertheless, deep sequencing is poised to revolutionize clinical virology. This technology has already been critical in the discovery of many new viruses, the characterization of virus populations in humans and the potential association of these viromes with the pathogenesis of several diseases. As described here for HIV, HCV and influenza virus, deep sequencing will facilitate the study of virus diversity in many other DNA and RNA viral pathogens. Deep sequencing-based assays are already being used in the clinical setting to monitor HIV-infected individuals [162, 247, 250] and it is only a matter of time before it will be applied routinely in clinical virology laboratories for virtually all viral pathogens.

Highlights.

• Deep sequencing is revolutionizing the clinical virology field.

• More affordable deep sequencing methodologies allow their use in clinical laboratories.

• Most deep sequencing-based studies are related to HIV, HCV, and influenza virus.

• Deep sequencing assays permit early identification of minority drug resistant variants

• Clinical significance of minority viral variants is still under debate.