Downloaded from journal.pda.org on December 8, 2014
Whole Genome: Next-Generation Sequencing as a Virus Safety Test for Biotechnological Products Eric Cabannes, Charles Hébert and Marc Eloit
PDA J Pharm Sci and Tech 2014, 68 631-638 Access the most recent version at doi:10.5731/pdajpst.2014.01015
Downloaded from journal.pda.org on December 8, 2014
CONFERENCE PROCEEDING: Proceedings of the PDA/FDA Advanced Technologies for Virus Detection in the Evaluation of Biologicals Conference: Applications and Challenges Workshop in Bethesda, MD, USA; November 13-14, 2013 Guest Editors: Arifa S. Khan (Rockville, MD), Dominick Vacante (Malvern, PA)
APPLICATIONS OF NEW ANALYTICAL TECHNOLOGIES
Whole Genome: Next-Generation Sequencing as a Virus Safety Test for Biotechnological Products ERIC CABANNES1, CHARLES HE´BERT1, and MARC ELOIT1,2* 1
PathoQuest, Baˆtiment Franc¸ois Jacob, Paris, France; and 2Institut Pasteur, Laboratory of Pathogen Discovery, Department of Virology, Paris, France ©PDA, Inc. 2014 ABSTRACT: The availability of next-generation sequencing techniques is about to strongly modify the strategies of control of the viral safety of biologicals products. It is now possible to use the tools of metagenomics, which is the study of the microbial genetic sequences recovered directly from a given sample. In this setting, the sequence of all the nucleic acids species of the sample are determined and compared with those in databases. We describe the pipeline we are currently using and show comparison of its analytical sensitivity with that of polymerase chain reaction together with results of time course experiments in infected cells. We propose to test cell supernatants for viral genomes associated to capsids and testing cells for viral RNA transcripts as the hallmark for virus replication, and we suggest rules of interpretation of the results. We also discuss the applicability of next-generation sequencing for the control of raw materials. 1. Introduction The risks of contamination of cell culture by endogenous viruses or through the use of biological reagents are well known. Direct testing of master cell banks, late cell banks and unprocessed bulk for animal viruses, associated with the control and/or inactivation of biological raw materials, are means of mitigating the risks of contamination. Biosafety testing batteries must demonstrate a high negative predictive value (NPV) (1), meaning that negative results must correspond to virus-free samples in a reliable manner, which is the primary objective of the test. NPV increases with the diagnostic sensitivity (i.e., the proportion of infected samples detected), which rely per se on both the analytical sensitivity (the smallest amount of virus detected) and the range of detection (the range of virus species and strains that are detectable). It depends also on the prevalence (the lowest the prevalence of infected samples the better the NPV, but while this parameter cannot be controlled the prevalence of truly infected test samples should in practice be very low). Additional criteria are time to results or good predictive positive value (PPV) (1). False-positive results would necessitate carrying out lengthy and costly use-
*Corresponding Author: Marc [email protected] doi: 10.5731/pdajpst.2014.01015
Eloit,
Vol. 68, No. 6, November–December 2014
e-mail:
less investigations. PPV increases with the diagnostic specificity and the prevalence of infected samples. As this prevalence is generally very low, a highly specific test is required to avoid false-positive results. Ideally, this test should also be unique, replacing rather than completing current tests, and should be cost-effective. Biosafety assays have been driven by the use of cell culture and, more recently, by (reverse-transcriptase)polymerase chain reaction (PCR)((RT)PCR) [for a review see (2)]. Isolation in cell culture is unsuitable for viruses for which no permissive cell line is known (such as noroviruses). (RT)PCRs, for practical reasons, can generally be conducted only for a selected number of known viruses, and indeed cannot detect unknown viruses. The range of detection of (RT)PCR can be increased by multiplex PCR using primers targeting numerous pathogens and multiple loci within pathogens and resolving these amplicons using electrospray ionization-mass spectrometry (3) or NGS (4). Nevertheless, the spectrum of potential viral contamination is very high as it encompasses the viruses of several animal species, including numerous unknown viruses for many animal species, which limits the use of multiplex PCRs. An alternative strategy takes advantage of whole genome–NGS (WG-NGS) technology and related tools of metagenomics, which is the study of the microbial genetic sequences recovered directly from a given sample. In this setting, all of the nucleic acid sequences in a sample are determined and compared 631
Downloaded from journal.pda.org on December 8, 2014
with those in databases. De novo assembly of large parts or full length genomes of pathogens can sometimes be achieved directly from the samples. WGNGS has become a very useful technology for detecting known or unknown viruses in various biological specimens [review in (2)], including such biological products as human paediatric live vaccines (5). We have developed a pipeline (6), from sample extraction to bioinformatics analysis, leading to the discovery of several new human viruses (7–9). We show here some elements of validation regarding the capacity of the pipeline to detect adventitious viruses and discuss its use for the screening of biotechnological products.
or viral genomes, respectively. For acellular samples (raw materials, cell culture supernatants, virus seeds, among others) and with the objective of detecting bona fide viral particles, free nucleic acids are subjected to a controlled hydrolysis followed by the extraction of encapsidated and/or protected nucleic acids. As regards viruses, all types of nucleic acids are extracted: DNA and RNA, single- and double-stranded, linear and circular genomes. The next stage involves the conversion of RNA into complementary DNA (cDNA) followed by random amplification (from the pico- to nanogram range depending on the starting material) in order to match the specifications provided in the sequencing protocol (usually about a few hundred nanograms, but may vary with the protocol used).
2. Description of the Pipeline 2.1. Sample Preparation WG-NGS presents the unprecedented advantage of being a non-hypothesis-driven technology able to detect any virus species within a given sample. However, to fully benefit from this particular leverage the sample preparation procedure is constrained by some difficulties. For instance, the non-selective and random amplification of extracted nucleic acids may cause sensitivity losses if the sample contains significant amounts of host nucleic acids (virus seeds, crude cell extracts) and/or is contaminated by exogenous (environmental contamination) nucleic acids. Therefore, one must ensure that both the design and the experimental aspects of the pipeline leading to the construction of the sequencing library are optimised in regard to the targeted objective. In order to avoid/limit contaminations, the laboratory is organised according to the go-forward principle where low-concentration (extracts) and high-concentration (amplified) nucleic acid samples with associated reagents are clearly separated. Samples are manipulated under UV and/or safety cabinets, one tube at the time, avoiding cross-contamination as much as possible. Prior and after each manipulation, working areas are carefully decontaminated with appropriate reagents (DNA Away威, DNA-ExitusPlus威, among others). These protection measures against cross-contamination are of particular importance on sequencing platforms where NGS is performed on a daily basis. Depending on the cellular or acellular nature of the biologicals to be tested, we developed two robust pipelines dedicated to the detection of viral transcripts 632
Regarding cellular samples (master and working cell banks) and with the objective of detecting signs of viral replication–presence of viral transcripts within the cells– solely RNAs are extracted. Due to the complexity of the entire cellular transcriptome it may be tempting to apply some selection steps such as polyA⫹ RNA selection and/or ribosomal RNA (rRNA) depletion to enrich for the viral RNA fraction. However, one must keep in mind that some viruses do not polyadenylate their transcripts—most arboviruses for instance—and commercially available rRNA depletion systems based on hybridization do not provide the probe(s) sequence(s) used, making it an uncertain approach for viral metagenomic studies. Like acellular samples, the extracted RNAs are converted into cDNA prior amplification. Once available in sufficient quantities, the amplified DNAs are then mechanically or enzymatically sheared according to the size of the sequencing chemistry. Though enzymatic shearing (Nextera威 from Illumina and Ion Xpress威 from Life Technologies) represents a simple and straightforward procedure with no investment in specialized equipment (sonicator or ultrasonicator), and has been validated for genomic studies, the impact of the very slight bias introduced by the enzyme(s) has not as yet been evaluated in metagenomics studies and especially on small genomes such as viruses. Of note, in some protocols (Illumina RNA-Seq protocol), the polyA⫹ RNA selection, the fragmentation into an appropriate size for library construction, and the reverse transcription are performed in few successive steps. Following fragmentation, the library construction pursues with an end-repair step (mechanical shearing only) prior to the insertion of sequencing adapters at both extremities of sheared fragments. At this stage, PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on December 8, 2014
and if samples are multiplexed during the sequencing run, each library must carry a barcoded-adapter. It is good practice to introduce this tag even when samples are run on their own, as it may avoid biological and/or bioinformatic cross-contamination between samples. Samples are then ready for sequencing using the manufacturer’s dedicated protocol. 2.2. Bioinformatics Pipeline We have developed a bioinformatics pipeline that is sequencing technology agnostic and deals with the standardized sequence “fastq” file format. If the library carries a barcoded-adapter, an initial demultiplexing step should be performed using the manufacturer dedicated software. Data analysis consists of two steps: raw data filtering and taxonomic assignment. WG-NGS runs produce a large amount (more than 100 million reads) of raw sequences that requires— before assignment—a series of prefiltering steps including adapter trimming, library PCR artefact removal, and Phred score– based quality filtering. A large number of sequences belongs to the host genome and shouldn’t be considered for the analysis. In order to eliminate these sequences, we built a comprehensive reference dataset by combining all available sequences related to the host: reference genome, draft genome, scaffold, etc. This reference is used in conjunction with stringent local alignment algorithms (Smith & Waterman and BurrowsWheeler transformation) to perform a negative filtering step. Unaligned reads are considered as highly relevant and selected for a de novo assembly step. De novo assembly produces longer sequences (called contigs) and a remaining set of unassembled reads sequences (called singletons). De novo assembly improves taxonomic assignment by increasing the accuracy of sequence alignments. Then, both contigs and singletons are aligned against homologous sequences. The major limitation in analysis of viral metagenomes remains the high genetic variability of many virus families. To overcome this issue, we use comprehensive databases of nucleic acid and protein sequences that are continuously implemented. These databases are versioned and archived for a long term period. Virus detection requires a highly sensitive method of identification of sequence homology. A local sequence alignment algorithm with relaxed parameters (identity, sequence match length) is used to ensure a high sensitivity of detection. As a result, we are able to detect Vol. 68, No. 6, November–December 2014
even short fragments (i.e., ⬎100 nt) that share a very low level of similarity against a known reference genome. Distant alignments ensure that unknown viruses that share a minimal homology at the level of nucleotide or protein may be detected. The key advantage of the pipeline is that we use a two-step alignment strategy. Reads and singletons are aligned, following the same scheme, first against the viral nucleotidic subpart then against the proteic subpart of the database. Alignments are produced using the Smith and Waterman algorithm. In order to decrease the false positive rate, significant alignments are invalidated against the whole sequence database. Remaining nonassigned sequences are screened for protein signature using a probabilistic identification algorithm based on profile Hidden Markov models (HMMs). HMMs are a highly effective means of identifying a common motif within a set of unaligned sequences (10). Validated viral hits give rise to a taxonomic assignment that are classified into two categories (close or distant to known virus) by a decision tree. To achieve an industrial grade of confidence, this bioinformatics pipeline is fully automated, tested and versioned. A dedicated job dispatcher ensures that there is no deviation during the analysis and logs all performed actions. 3. Use of Wg-ngs for Testing Cell Lines 3.1. Virus Cycle and Targets of WG-NGS Virus-infected cells harbor different types of viral nucleic acids that have different meanings as regard the status of the cells. First, nucleic acids from viruses replicating in cells can be found. In the supernatant, viral genomes protected by viral capsids are present. In addition viral DNA or RNA genomes and RNA transcripts released from lysed cells can be found either free or associated with subcellular components. Within the cells, the same type of nucleic acids can be found. Second, nucleic acids from non-replicating viruses originating from raw materials used in the development of the cell banks or in manufacturing might be present in the form of viral genomes protected by capsids. In this case, no RNA transcripts are found in the cells. It should be noted that lack of replication can be due to previous virus inactivation (for example, due to gamma irradiation of raw materials), but can also be 633
Downloaded from journal.pda.org on December 8, 2014
TABLE I Interpretation of the WG-NGS Analysis of Combined Analysis of Cell-associated RNA and Cell Supernatant CASE 1
CASE 2
CASE 3
CASE 4
DNA/RNA in cell supernatant
ⴚ
ⴙ
ⴙ
ⴚ
Cell-associated RNA
ⴚ
ⴙ
ⴚ
INTERPRETATION
Virus-free cells
Virus-infected cells
COMMENTS
due to a lack of permissiveness of the cell line to an otherwise living virus. Thus it is proposed to test the cell supernatant for the presence of DNA/RNA associated with virus particles, and the cells for the presence of RNA transcripts, which are a common hallmark of virus replication irrespective of the type of viral genome (DNA or RNA, ss or ds). The interpretation of the results is presented in Table I. 3.2. Analytical Sensitivity of Detection of Viral Genomes in Supernatants Determining the analytical sensitivity of virus detection by WG-NGS is not only important from a technical point of view but mandatory for acceptance of the technology by regulatory agencies as a screening tool for biologicals. Due to the diversity of acellular samples regarding their nucleic acid content (cell culture supernatants, raw materials, virus seeds to name a few), it is hardly possible to define an absolute limit of detection (LOD) for a given virus as is commonly accepted for quantitative PCR (qPCR) assays (from tens to a few hundreds of copies per milliliter for the most sensitive diagnostic assays on the market). Indeed, if highly variable nucleic acid content (“background noise”) in a given type of sample (human plasma, for instance) may introduce some slight variability in the qPCR analytical sensitivity, this is of much greater impact in WG-NGS metagenomics experiments where the whole nucleic acid content of the sample—non-biased approach—is amplified prior to sequencing. At present, the best approach to evaluate the analytical sensitivity of WG-NGS seems to be a direct comparison with qPCR on well-characterized samples. For this purpose we spiked cell culture supernatants and 634
ⴙ
Non-infected cells
Need complementary analysis
Non-replicative virus (trace of nucleic acids, or inactivated virus or live virus but nonpermissive cells)
Non-productive cycle, or cell-associated virus
human plasma samples with titrated virus stocks and compared the results between WG-NGS and qPCR. Using referenced qPCR assays published in the literature, we were able to detect the presence of viral nucleic acids by both technologies in a comparable qualitative manner assuming that optimally about a hundred million reads were obtained by NGS, in line with the output of massive parallel sequencers (Illumina HiSeq, Life Techologies Proton) but too high for desktop sequencers (Illumina MiSeq, Life Technologies PGM). Nevertheless, a lower number of reads might also work (see an example in Table II). The optimal number of reads also depends on the residual amount of host DNA, and using a high depth of sequencing allows to address worst cases of high host DNA residual loads. However, we have seen that in a particular case (Illumina HiSeq run of 91 million reads) we were unable to detect the presence of HIV-1 in a human plasma sample that was detected and titrated at 125 copies/mL (i.e., around 2 copies in the sample size submitted to WG-NGS) using the COBAS威 diagnostic kit. In contrast to qPCR, which focuses on a small defined part of the viral genome— the amplicon—WG-NGS is a whole genome approach that for a given virus concentration increases the analytical sensitivity for large-genome viruses— herpesvirus, adenovirus— compared with their smaller counterparts, for example, parvovirus and calicivirus. In addition, we have identified in several clinical cases viruses that were supposed to be absent based on the results of (RT)PCRs. Such discrepancy was mainly related to variants of RNA or single-stranded circular DNA (ss DNA) viruses not detected by the cognate (RT)PCRs due to mismatches of target sequences with the primers. Finally, data collected on infection time course experiments did show a similar increase in the virus production within the cell culture supernatant over time when monitored by both qPCR and WGNGS. Although it seems too premature at this stage to PDA Journal of Pharmaceutical Science and Technology
Two tubes were spiked with different viruses at a concentration 1x or 0.1x and subjected to WG-NGS with an output of 42-47 million reads. * Log number of copies in 12 L, which corresponds the volume equivalent of the original sample subjected to the first step of the pipeline. ** Number of positive PCRs among two replicates. *** CT ⬎ 45 were scored positive. **** Number of reads directed against the corresponding virus.
190
187
35/35
34/33
2/2 3,1 5
4 528
11,054 33/34
33/33 2/2
2/2 4,1
3.1 5
4381
3,600
Feline calicivirus***
Control B
6
2,1
2/2
753 38/⬎45 15,69 Distemper disease virus (canine paramyxovirus)
4,4
2,5
2/2
34/35
19,26?
3,4
1,5
1/2
23
907
33/34
38/-
2/2
1/2
2,7
1,4 3,3
4,6 713
35,062 36/37
30/29 2/2
2/2 2,4
3,7 5,6 5,323
Parainfluenza type 5 ss RNA
15,246
Canine parvovirus** ss DNA
4,3
0
4,63 37/36 2/2 3,5 337,322 32/32 2/2 2,6 135,797 Feline herpesvirus*
4,5
2 1,129 ⬎45/⬎45 0/2 1,1 31,323 Canine adenovirus 2
3
1,6
⬎45/⬎45 0/2
2,633 34/35
0,1
Cycles
2/2 98,984 21,000 Control A dsDNA
5
3,1
2/2
31/31
4
2,1
PCR (run1/run2) log copy/12 L* NGS #reads****s CT***
Concentration 1x
PCR (ⴙve/total)**) log copy/12L* log copies/mL Genome Size Genome Type
Spike
TABLE II Comparison of (RT)PCR and WG-NGS on Plasma Samples Spiked with Different Viruses
log copy/mL
Spike
Concentration 0.1x
NGS Reads
Downloaded from journal.pda.org on December 8, 2014
Vol. 68, No. 6, November–December 2014
expect quantitative measures and therefore to define an absolute analytical sensitivity value from WG-NGS experiments as routinely provided by qPCR assays— mainly due to the “background noise”—preliminary data raise hope for this technology to provide a semiquantitative (“relative”) answer regarding the virus concentration within a given sample. 3.3. Detection of Virus Replication though Transcriptomic Analysis of Cells The detection of viral replication within a master or working cell bank is with no doubt the major criterion/ parameter to be assessed by WG-NGS as far as safety of biologicals is concerned. Although WG-NGS may in some instances (no polyA⫹ selection to broaden the spectrum of viral detection, no rRNA depletion, no host transcriptional shutoff by the virus) suffer from a lower analytical sensitivity than qPCR mainly due to the massive “background noise” represented by the cellular transcriptome, the whole sample approach per se offers a critical advantage over qPCR. This whole sample approach means that both compartments (cells for the presence of viral transcripts, and supernatant for the presence of viral genomes associated to virus particles) should be tested independently to document the viral contamination status. Capability of detecting infected cells does not seem to be affected thanks to the use of a high depth of sequencing (see Figure 1). In contrast to PCR, which focuses essentially on a single or few viral transcripts or genome loci, WG-NGS can provide a global picture— the viral transcriptome—allowing discrimination between non-productive and productive viral cycles. This is of importance when one considers that many cell lines were obtained through viral infection of primary cells followed by integration of viral genes/genome fragments into the host genome and leading eventually to immortalization (like SV40 and 324k cells and human papillomavirus and HeLa cells). Moreover, the presence of “naturally occurring” viral elements within cellular genomes— endogenous retroviruses—is also another concern when dealing with viral safety issues. It is clear that in such situations a limited screening focusing on a few viral genes by qPCR will potentially lead to erroneous conclusions. On the other hand, extensive analysis by WG-NGS of both cell and culture supernatant compartments will reliably identify the replicative or nonreplicative nature of the infection. We have shown the benefit of NGS over qPCR in few experimental settings using tumor biopsies (see Figure 2) and infection time course experiments (Figure 1) and reproducibly obtained the same qualitative result by the two approaches. The 635
Downloaded from journal.pda.org on December 8, 2014
Figure 1 Transcriptome analysis of productively infected cells. Vero cells were infected with Borna disease virus at a very low multiplicity of infection (0.025 pfu/cell) and analyzed by immunofluorescence. RNAs were extracted and analyzed by RT-PCR (top panel) and NGS (bottom panel). semi-quantitative result provided by WG-NGS is particularly well suited for viral safety applied to master and working cell banks as the expected result is primarily of qualitative nature. 3.4. Implications for the Control of Cell Lines by WG-NGS Based on the previous data, it is likely that if a cell line is chronically infected (i.e., by a virus developing a 636
productive cycle at low or high level), the analytical sensitivity of WG-NGS would be low enough to provide a positive response for the transcriptomic analysis (cells). This is also likely for the supernatant (virus particles) but may vary as a function of the degree of spontaneous release of mature particles (or due to cell lysis). A negative result for the cells and the supernatant is highly presumptive of a non-infected cell line. For these reasons, control of this cell bank by specific PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on December 8, 2014
Figure 2 Transcriptome analysis of non-productively infected cells. RNas were extracted from an equine sarcoid, a tumor due to Bovine Papillomavirus type 1. Blind analysis of the cells identified BPV1 sequences, and mapping of the reads on the BPV1 genomes shows that the virus is blocked in the early phase of the cycle. PCRs would be of low utility (independently of regulatory requirements) (Table I). The same considerations apply also for the control of the bulks each time the adventitious virus could only be an infectious virus originating from the production cells (meaning in all the cases where no biological reagent has been used during the post–Working Cell Bank production steps). In case of a positive result (cell or supernatant), the result should be discussed (in particular with regard to the number and distribution of reads along the genome), and may be if necessary more deeply investigated by other tools like specific (RT)PCRs designed on the sequence evidenced. 4. Use of WG-NGS for Testing Raw Materials As discussed above, the analytical sensitivity of some PCRs is better than that of WG-NGS, while some others have the same range of detection as WG-NGS if the primer sequences fit exactly to the virus sequence but could be unable to detect distant strains, and indeed PCRs are unable to detect unknown viruses. Also, many of the members of the increasingly growing list of known viruses are currently not screened by PCR. For example, among the 12 human polyomaviruses, two (JC and BK) are classically screened (but not the ten other ones including at least two human Vol. 68, No. 6, November–December 2014
pathogens– Merkel and Trichodysplasia virus). One should bear in mind that, in contrast to chronically infected cell lines, raw materials can be contaminated by very low amount of adventitious viruses coming from contaminated animal or human tissues or biological fluids diluted in huge pools. These low viral loads might nevertheless be of concern regarding virus safety if they are plated on a permissive cell line or even inoculated in humans. Thus, the use of WG-NGS should be discussed case by case. Some very efficient dedicated PCRs cannot currently be replaced by WG-NGS (for example for all major human pathogens for the control of plasma donations), even if it could be argued that low virus loads not detected by WG-HTS would be cleared by manufacturing processes. WG-NGS might appear currently as a very broad test that can be used regularly to check the viral burden of some batches to refine routine PCR controls and inactivation procedures (11). Alternatively, it might also be used as a unique virus biosafety assay in all cases where there is a strong inactivation process downstream (e.g., gamma irradiation), meaning that the main objective would be to avoid a high load of virus in the raw material. 5. General Conclusion WG-NGS appears as a method of major interest for the testing of cell lines and for the analysis of the viral 637
Downloaded from journal.pda.org on December 8, 2014
burden in animal or human raw materials upstream of inactivation steps. There is now an urgent need to discuss a common strategy for testing cell substrate and propose common concepts and/or reagents for validation. Optimization of sample preparation and bioinformatic pipelines should progressively decrease the need for a very high depth of sequencing. Also, the place of WG-NGS within the set of different in vitro, in vivo, and molecular techniques can now be better defined based on technical and scientific grounds. Acknowledgements This work received financial support from the foundation Biosecure, Paris and Labex IBEID. We thank Jennifer Richardson for critical reading of the manuscript.
tant Klebsiella pneumoniae by targeted next-generation sequencing. J. Clin. Microbiol. 2014, 52 (3), 987–990. 5. Victoria, J. G.; Wang, C.; Jones, M. S.; Jaing, C.; McLoughlin, K.; Gardner, S.; Delwart, E. L. Viral nucleic acids in live-attenuated vaccines: detection of minority variants and an adventitious virus. J. Virol. 2010, 84 (12), 6033– 6040. 6. Cheval, J.; Sauvage, V.; Frangeul, L.; Dacheux, L.; Guigon, G.; Dumey, N.; Pariente, K.; Rousseaux, C.; Dorange, F.; Berthet, N.; Brisse, S.; Moszer, I.; Bourhy, H.; Manuguerra, C. J.; Lecuit, M.; Burguiere, A.; Caro, V.; Eloit, M. Evaluation of high-throughput sequencing for identifying known and unknown viruses in biological samples. J. Clin. Microbiol. 2011, 49 (9), 3268 –3275.
Conflict of Interests Declaration EC and CH are employees of PathoQuest, a spinoff of Institut Pasteur dedicated to the identification of Pathogens by NGS. ME is chairman of PathoQuest. References 1. Banoo, S.; Bell, D.; Bossuyt, P.; Herring, A.; Mabey, D.; Poole, F.; Smith, P. G.; Sriram, N.; Wongsrichanalai, C.; Linke, R.; O’Brien, R.; Perkins, M.; Cunningham, J.; Matsoso, P.; Nathanson, C. M.; Olliaro, P.; Peeling, R. W.; Ramsay, A.; The TDR Diagnostics Evaluation Expert Panel. Evaluation of diagnostic tests for infectious diseases: general principles. Nat. Rev. Microbiol. 2010, 8, S15–28. 2. Mokili, J. L.; Rohwer, F.; Dutilh, B. E. Metagenomics and future perspectives in virus discovery. Curr. Opin. Virol. 2012, 2 (1), 63–77. 3. Wolk, D. M.; Kaleta, E. J.; Wysocki, V. H. PCRelectrospray ionization mass spectrometry: the potential to change infectious disease diagnostics in clinical and public health laboratories. J. Mol. Diagn. 2012, 14 (4), 295–304. 4. Arena, F.; Rolfe, P. A.; Doran, G.; Conte, V.; Gruszka, S.; Clarke, T.; Adesokan, Y.; Giani, T.; Rossolini, G. M. Rapid resistome fingerprinting and clonal lineage profiling of carbapenem-resis-
638
7. Sauvage, V.; Foulongne, V.; Cheval, J.; Ar Gouilh, M.; Pariente, K.; Dereure, O.; Manuguerra, J. C.; Richardson, J.; Lecuit, M.; Burguie` re, A.; Caro, V.; Eloit, M. Human polyomavirus related to African green monkey lymphotropic polyomavirus. Emerg. Infect. Dis. 2011, 17 (8),1364 –1370. 8. Sauvage, V.; Cheval, J.; Foulongne, V.; Gouilh, M. A.; Pariente K.; Manuguerra J. C.; Richardson J.; Dereure O.; Lecuit M.; Burguiere A.; Caro V.; Eloit M. Identification of the first human gyrovirus, a virus related to chicken anemia virus. 2011, J. Virol. 85 (15), 7948 –7950. 9. Sauvage, V.; Ar Gouilh, M.; Cheval, J.; Muth, E.; Pariente, K.; Burguiere, A.; Caro, V.; Manuguerra, J-C.; Eloit, M. A member of a new Picornaviridae Genus is shed in pig feces. J. Virol. 2012, 86 (18), 10036 –10046. 10. Yoon, B.-J. Hidden Markov models and their applications in biological sequence analysis. Curr. Genomics 2009, 10 (6), 402– 415. 11. Gagnieur, L.; Cheval, J.; Gratigny, M.; He´bert, C.; Muth, E.; Dumarest, M.; Eloit, M. Unbiased analysis by high throughput sequencing of the viral diversity in fetal bovine serum and trypsin used in cell culture. Biologicals 2014, 42 (3), 1045–152.
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on December 8, 2014
An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal
Whole Genome: Next-Generation Sequencing as a Virus Safety Test for Biotechnological Products Eric Cabannes, Charles Hébert and Marc Eloit
PDA J Pharm Sci and Tech 2014, 68 631-638 Access the most recent version at doi:10.5731/pdajpst.2014.01015
Downloaded from journal.pda.org on December 8, 2014
CONFERENCE PROCEEDING: Proceedings of the PDA/FDA Advanced Technologies for Virus Detection in the Evaluation of Biologicals Conference: Applications and Challenges Workshop in Bethesda, MD, USA; November 13-14, 2013 Guest Editors: Arifa S. Khan (Rockville, MD), Dominick Vacante (Malvern, PA)
APPLICATIONS OF NEW ANALYTICAL TECHNOLOGIES
Whole Genome: Next-Generation Sequencing as a Virus Safety Test for Biotechnological Products ERIC CABANNES1, CHARLES HE´BERT1, and MARC ELOIT1,2* 1
PathoQuest, Baˆtiment Franc¸ois Jacob, Paris, France; and 2Institut Pasteur, Laboratory of Pathogen Discovery, Department of Virology, Paris, France ©PDA, Inc. 2014 ABSTRACT: The availability of next-generation sequencing techniques is about to strongly modify the strategies of control of the viral safety of biologicals products. It is now possible to use the tools of metagenomics, which is the study of the microbial genetic sequences recovered directly from a given sample. In this setting, the sequence of all the nucleic acids species of the sample are determined and compared with those in databases. We describe the pipeline we are currently using and show comparison of its analytical sensitivity with that of polymerase chain reaction together with results of time course experiments in infected cells. We propose to test cell supernatants for viral genomes associated to capsids and testing cells for viral RNA transcripts as the hallmark for virus replication, and we suggest rules of interpretation of the results. We also discuss the applicability of next-generation sequencing for the control of raw materials. 1. Introduction The risks of contamination of cell culture by endogenous viruses or through the use of biological reagents are well known. Direct testing of master cell banks, late cell banks and unprocessed bulk for animal viruses, associated with the control and/or inactivation of biological raw materials, are means of mitigating the risks of contamination. Biosafety testing batteries must demonstrate a high negative predictive value (NPV) (1), meaning that negative results must correspond to virus-free samples in a reliable manner, which is the primary objective of the test. NPV increases with the diagnostic sensitivity (i.e., the proportion of infected samples detected), which rely per se on both the analytical sensitivity (the smallest amount of virus detected) and the range of detection (the range of virus species and strains that are detectable). It depends also on the prevalence (the lowest the prevalence of infected samples the better the NPV, but while this parameter cannot be controlled the prevalence of truly infected test samples should in practice be very low). Additional criteria are time to results or good predictive positive value (PPV) (1). False-positive results would necessitate carrying out lengthy and costly use-
*Corresponding Author: Marc [email protected] doi: 10.5731/pdajpst.2014.01015
Eloit,
Vol. 68, No. 6, November–December 2014
e-mail:
less investigations. PPV increases with the diagnostic specificity and the prevalence of infected samples. As this prevalence is generally very low, a highly specific test is required to avoid false-positive results. Ideally, this test should also be unique, replacing rather than completing current tests, and should be cost-effective. Biosafety assays have been driven by the use of cell culture and, more recently, by (reverse-transcriptase)polymerase chain reaction (PCR)((RT)PCR) [for a review see (2)]. Isolation in cell culture is unsuitable for viruses for which no permissive cell line is known (such as noroviruses). (RT)PCRs, for practical reasons, can generally be conducted only for a selected number of known viruses, and indeed cannot detect unknown viruses. The range of detection of (RT)PCR can be increased by multiplex PCR using primers targeting numerous pathogens and multiple loci within pathogens and resolving these amplicons using electrospray ionization-mass spectrometry (3) or NGS (4). Nevertheless, the spectrum of potential viral contamination is very high as it encompasses the viruses of several animal species, including numerous unknown viruses for many animal species, which limits the use of multiplex PCRs. An alternative strategy takes advantage of whole genome–NGS (WG-NGS) technology and related tools of metagenomics, which is the study of the microbial genetic sequences recovered directly from a given sample. In this setting, all of the nucleic acid sequences in a sample are determined and compared 631
Downloaded from journal.pda.org on December 8, 2014
with those in databases. De novo assembly of large parts or full length genomes of pathogens can sometimes be achieved directly from the samples. WGNGS has become a very useful technology for detecting known or unknown viruses in various biological specimens [review in (2)], including such biological products as human paediatric live vaccines (5). We have developed a pipeline (6), from sample extraction to bioinformatics analysis, leading to the discovery of several new human viruses (7–9). We show here some elements of validation regarding the capacity of the pipeline to detect adventitious viruses and discuss its use for the screening of biotechnological products.
or viral genomes, respectively. For acellular samples (raw materials, cell culture supernatants, virus seeds, among others) and with the objective of detecting bona fide viral particles, free nucleic acids are subjected to a controlled hydrolysis followed by the extraction of encapsidated and/or protected nucleic acids. As regards viruses, all types of nucleic acids are extracted: DNA and RNA, single- and double-stranded, linear and circular genomes. The next stage involves the conversion of RNA into complementary DNA (cDNA) followed by random amplification (from the pico- to nanogram range depending on the starting material) in order to match the specifications provided in the sequencing protocol (usually about a few hundred nanograms, but may vary with the protocol used).
2. Description of the Pipeline 2.1. Sample Preparation WG-NGS presents the unprecedented advantage of being a non-hypothesis-driven technology able to detect any virus species within a given sample. However, to fully benefit from this particular leverage the sample preparation procedure is constrained by some difficulties. For instance, the non-selective and random amplification of extracted nucleic acids may cause sensitivity losses if the sample contains significant amounts of host nucleic acids (virus seeds, crude cell extracts) and/or is contaminated by exogenous (environmental contamination) nucleic acids. Therefore, one must ensure that both the design and the experimental aspects of the pipeline leading to the construction of the sequencing library are optimised in regard to the targeted objective. In order to avoid/limit contaminations, the laboratory is organised according to the go-forward principle where low-concentration (extracts) and high-concentration (amplified) nucleic acid samples with associated reagents are clearly separated. Samples are manipulated under UV and/or safety cabinets, one tube at the time, avoiding cross-contamination as much as possible. Prior and after each manipulation, working areas are carefully decontaminated with appropriate reagents (DNA Away威, DNA-ExitusPlus威, among others). These protection measures against cross-contamination are of particular importance on sequencing platforms where NGS is performed on a daily basis. Depending on the cellular or acellular nature of the biologicals to be tested, we developed two robust pipelines dedicated to the detection of viral transcripts 632
Regarding cellular samples (master and working cell banks) and with the objective of detecting signs of viral replication–presence of viral transcripts within the cells– solely RNAs are extracted. Due to the complexity of the entire cellular transcriptome it may be tempting to apply some selection steps such as polyA⫹ RNA selection and/or ribosomal RNA (rRNA) depletion to enrich for the viral RNA fraction. However, one must keep in mind that some viruses do not polyadenylate their transcripts—most arboviruses for instance—and commercially available rRNA depletion systems based on hybridization do not provide the probe(s) sequence(s) used, making it an uncertain approach for viral metagenomic studies. Like acellular samples, the extracted RNAs are converted into cDNA prior amplification. Once available in sufficient quantities, the amplified DNAs are then mechanically or enzymatically sheared according to the size of the sequencing chemistry. Though enzymatic shearing (Nextera威 from Illumina and Ion Xpress威 from Life Technologies) represents a simple and straightforward procedure with no investment in specialized equipment (sonicator or ultrasonicator), and has been validated for genomic studies, the impact of the very slight bias introduced by the enzyme(s) has not as yet been evaluated in metagenomics studies and especially on small genomes such as viruses. Of note, in some protocols (Illumina RNA-Seq protocol), the polyA⫹ RNA selection, the fragmentation into an appropriate size for library construction, and the reverse transcription are performed in few successive steps. Following fragmentation, the library construction pursues with an end-repair step (mechanical shearing only) prior to the insertion of sequencing adapters at both extremities of sheared fragments. At this stage, PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on December 8, 2014
and if samples are multiplexed during the sequencing run, each library must carry a barcoded-adapter. It is good practice to introduce this tag even when samples are run on their own, as it may avoid biological and/or bioinformatic cross-contamination between samples. Samples are then ready for sequencing using the manufacturer’s dedicated protocol. 2.2. Bioinformatics Pipeline We have developed a bioinformatics pipeline that is sequencing technology agnostic and deals with the standardized sequence “fastq” file format. If the library carries a barcoded-adapter, an initial demultiplexing step should be performed using the manufacturer dedicated software. Data analysis consists of two steps: raw data filtering and taxonomic assignment. WG-NGS runs produce a large amount (more than 100 million reads) of raw sequences that requires— before assignment—a series of prefiltering steps including adapter trimming, library PCR artefact removal, and Phred score– based quality filtering. A large number of sequences belongs to the host genome and shouldn’t be considered for the analysis. In order to eliminate these sequences, we built a comprehensive reference dataset by combining all available sequences related to the host: reference genome, draft genome, scaffold, etc. This reference is used in conjunction with stringent local alignment algorithms (Smith & Waterman and BurrowsWheeler transformation) to perform a negative filtering step. Unaligned reads are considered as highly relevant and selected for a de novo assembly step. De novo assembly produces longer sequences (called contigs) and a remaining set of unassembled reads sequences (called singletons). De novo assembly improves taxonomic assignment by increasing the accuracy of sequence alignments. Then, both contigs and singletons are aligned against homologous sequences. The major limitation in analysis of viral metagenomes remains the high genetic variability of many virus families. To overcome this issue, we use comprehensive databases of nucleic acid and protein sequences that are continuously implemented. These databases are versioned and archived for a long term period. Virus detection requires a highly sensitive method of identification of sequence homology. A local sequence alignment algorithm with relaxed parameters (identity, sequence match length) is used to ensure a high sensitivity of detection. As a result, we are able to detect Vol. 68, No. 6, November–December 2014
even short fragments (i.e., ⬎100 nt) that share a very low level of similarity against a known reference genome. Distant alignments ensure that unknown viruses that share a minimal homology at the level of nucleotide or protein may be detected. The key advantage of the pipeline is that we use a two-step alignment strategy. Reads and singletons are aligned, following the same scheme, first against the viral nucleotidic subpart then against the proteic subpart of the database. Alignments are produced using the Smith and Waterman algorithm. In order to decrease the false positive rate, significant alignments are invalidated against the whole sequence database. Remaining nonassigned sequences are screened for protein signature using a probabilistic identification algorithm based on profile Hidden Markov models (HMMs). HMMs are a highly effective means of identifying a common motif within a set of unaligned sequences (10). Validated viral hits give rise to a taxonomic assignment that are classified into two categories (close or distant to known virus) by a decision tree. To achieve an industrial grade of confidence, this bioinformatics pipeline is fully automated, tested and versioned. A dedicated job dispatcher ensures that there is no deviation during the analysis and logs all performed actions. 3. Use of Wg-ngs for Testing Cell Lines 3.1. Virus Cycle and Targets of WG-NGS Virus-infected cells harbor different types of viral nucleic acids that have different meanings as regard the status of the cells. First, nucleic acids from viruses replicating in cells can be found. In the supernatant, viral genomes protected by viral capsids are present. In addition viral DNA or RNA genomes and RNA transcripts released from lysed cells can be found either free or associated with subcellular components. Within the cells, the same type of nucleic acids can be found. Second, nucleic acids from non-replicating viruses originating from raw materials used in the development of the cell banks or in manufacturing might be present in the form of viral genomes protected by capsids. In this case, no RNA transcripts are found in the cells. It should be noted that lack of replication can be due to previous virus inactivation (for example, due to gamma irradiation of raw materials), but can also be 633
Downloaded from journal.pda.org on December 8, 2014
TABLE I Interpretation of the WG-NGS Analysis of Combined Analysis of Cell-associated RNA and Cell Supernatant CASE 1
CASE 2
CASE 3
CASE 4
DNA/RNA in cell supernatant
ⴚ
ⴙ
ⴙ
ⴚ
Cell-associated RNA
ⴚ
ⴙ
ⴚ
INTERPRETATION
Virus-free cells
Virus-infected cells
COMMENTS
due to a lack of permissiveness of the cell line to an otherwise living virus. Thus it is proposed to test the cell supernatant for the presence of DNA/RNA associated with virus particles, and the cells for the presence of RNA transcripts, which are a common hallmark of virus replication irrespective of the type of viral genome (DNA or RNA, ss or ds). The interpretation of the results is presented in Table I. 3.2. Analytical Sensitivity of Detection of Viral Genomes in Supernatants Determining the analytical sensitivity of virus detection by WG-NGS is not only important from a technical point of view but mandatory for acceptance of the technology by regulatory agencies as a screening tool for biologicals. Due to the diversity of acellular samples regarding their nucleic acid content (cell culture supernatants, raw materials, virus seeds to name a few), it is hardly possible to define an absolute limit of detection (LOD) for a given virus as is commonly accepted for quantitative PCR (qPCR) assays (from tens to a few hundreds of copies per milliliter for the most sensitive diagnostic assays on the market). Indeed, if highly variable nucleic acid content (“background noise”) in a given type of sample (human plasma, for instance) may introduce some slight variability in the qPCR analytical sensitivity, this is of much greater impact in WG-NGS metagenomics experiments where the whole nucleic acid content of the sample—non-biased approach—is amplified prior to sequencing. At present, the best approach to evaluate the analytical sensitivity of WG-NGS seems to be a direct comparison with qPCR on well-characterized samples. For this purpose we spiked cell culture supernatants and 634
ⴙ
Non-infected cells
Need complementary analysis
Non-replicative virus (trace of nucleic acids, or inactivated virus or live virus but nonpermissive cells)
Non-productive cycle, or cell-associated virus
human plasma samples with titrated virus stocks and compared the results between WG-NGS and qPCR. Using referenced qPCR assays published in the literature, we were able to detect the presence of viral nucleic acids by both technologies in a comparable qualitative manner assuming that optimally about a hundred million reads were obtained by NGS, in line with the output of massive parallel sequencers (Illumina HiSeq, Life Techologies Proton) but too high for desktop sequencers (Illumina MiSeq, Life Technologies PGM). Nevertheless, a lower number of reads might also work (see an example in Table II). The optimal number of reads also depends on the residual amount of host DNA, and using a high depth of sequencing allows to address worst cases of high host DNA residual loads. However, we have seen that in a particular case (Illumina HiSeq run of 91 million reads) we were unable to detect the presence of HIV-1 in a human plasma sample that was detected and titrated at 125 copies/mL (i.e., around 2 copies in the sample size submitted to WG-NGS) using the COBAS威 diagnostic kit. In contrast to qPCR, which focuses on a small defined part of the viral genome— the amplicon—WG-NGS is a whole genome approach that for a given virus concentration increases the analytical sensitivity for large-genome viruses— herpesvirus, adenovirus— compared with their smaller counterparts, for example, parvovirus and calicivirus. In addition, we have identified in several clinical cases viruses that were supposed to be absent based on the results of (RT)PCRs. Such discrepancy was mainly related to variants of RNA or single-stranded circular DNA (ss DNA) viruses not detected by the cognate (RT)PCRs due to mismatches of target sequences with the primers. Finally, data collected on infection time course experiments did show a similar increase in the virus production within the cell culture supernatant over time when monitored by both qPCR and WGNGS. Although it seems too premature at this stage to PDA Journal of Pharmaceutical Science and Technology
Two tubes were spiked with different viruses at a concentration 1x or 0.1x and subjected to WG-NGS with an output of 42-47 million reads. * Log number of copies in 12 L, which corresponds the volume equivalent of the original sample subjected to the first step of the pipeline. ** Number of positive PCRs among two replicates. *** CT ⬎ 45 were scored positive. **** Number of reads directed against the corresponding virus.
190
187
35/35
34/33
2/2 3,1 5
4 528
11,054 33/34
33/33 2/2
2/2 4,1
3.1 5
4381
3,600
Feline calicivirus***
Control B
6
2,1
2/2
753 38/⬎45 15,69 Distemper disease virus (canine paramyxovirus)
4,4
2,5
2/2
34/35
19,26?
3,4
1,5
1/2
23
907
33/34
38/-
2/2
1/2
2,7
1,4 3,3
4,6 713
35,062 36/37
30/29 2/2
2/2 2,4
3,7 5,6 5,323
Parainfluenza type 5 ss RNA
15,246
Canine parvovirus** ss DNA
4,3
0
4,63 37/36 2/2 3,5 337,322 32/32 2/2 2,6 135,797 Feline herpesvirus*
4,5
2 1,129 ⬎45/⬎45 0/2 1,1 31,323 Canine adenovirus 2
3
1,6
⬎45/⬎45 0/2
2,633 34/35
0,1
Cycles
2/2 98,984 21,000 Control A dsDNA
5
3,1
2/2
31/31
4
2,1
PCR (run1/run2) log copy/12 L* NGS #reads****s CT***
Concentration 1x
PCR (ⴙve/total)**) log copy/12L* log copies/mL Genome Size Genome Type
Spike
TABLE II Comparison of (RT)PCR and WG-NGS on Plasma Samples Spiked with Different Viruses
log copy/mL
Spike
Concentration 0.1x
NGS Reads
Downloaded from journal.pda.org on December 8, 2014
Vol. 68, No. 6, November–December 2014
expect quantitative measures and therefore to define an absolute analytical sensitivity value from WG-NGS experiments as routinely provided by qPCR assays— mainly due to the “background noise”—preliminary data raise hope for this technology to provide a semiquantitative (“relative”) answer regarding the virus concentration within a given sample. 3.3. Detection of Virus Replication though Transcriptomic Analysis of Cells The detection of viral replication within a master or working cell bank is with no doubt the major criterion/ parameter to be assessed by WG-NGS as far as safety of biologicals is concerned. Although WG-NGS may in some instances (no polyA⫹ selection to broaden the spectrum of viral detection, no rRNA depletion, no host transcriptional shutoff by the virus) suffer from a lower analytical sensitivity than qPCR mainly due to the massive “background noise” represented by the cellular transcriptome, the whole sample approach per se offers a critical advantage over qPCR. This whole sample approach means that both compartments (cells for the presence of viral transcripts, and supernatant for the presence of viral genomes associated to virus particles) should be tested independently to document the viral contamination status. Capability of detecting infected cells does not seem to be affected thanks to the use of a high depth of sequencing (see Figure 1). In contrast to PCR, which focuses essentially on a single or few viral transcripts or genome loci, WG-NGS can provide a global picture— the viral transcriptome—allowing discrimination between non-productive and productive viral cycles. This is of importance when one considers that many cell lines were obtained through viral infection of primary cells followed by integration of viral genes/genome fragments into the host genome and leading eventually to immortalization (like SV40 and 324k cells and human papillomavirus and HeLa cells). Moreover, the presence of “naturally occurring” viral elements within cellular genomes— endogenous retroviruses—is also another concern when dealing with viral safety issues. It is clear that in such situations a limited screening focusing on a few viral genes by qPCR will potentially lead to erroneous conclusions. On the other hand, extensive analysis by WG-NGS of both cell and culture supernatant compartments will reliably identify the replicative or nonreplicative nature of the infection. We have shown the benefit of NGS over qPCR in few experimental settings using tumor biopsies (see Figure 2) and infection time course experiments (Figure 1) and reproducibly obtained the same qualitative result by the two approaches. The 635
Downloaded from journal.pda.org on December 8, 2014
Figure 1 Transcriptome analysis of productively infected cells. Vero cells were infected with Borna disease virus at a very low multiplicity of infection (0.025 pfu/cell) and analyzed by immunofluorescence. RNAs were extracted and analyzed by RT-PCR (top panel) and NGS (bottom panel). semi-quantitative result provided by WG-NGS is particularly well suited for viral safety applied to master and working cell banks as the expected result is primarily of qualitative nature. 3.4. Implications for the Control of Cell Lines by WG-NGS Based on the previous data, it is likely that if a cell line is chronically infected (i.e., by a virus developing a 636
productive cycle at low or high level), the analytical sensitivity of WG-NGS would be low enough to provide a positive response for the transcriptomic analysis (cells). This is also likely for the supernatant (virus particles) but may vary as a function of the degree of spontaneous release of mature particles (or due to cell lysis). A negative result for the cells and the supernatant is highly presumptive of a non-infected cell line. For these reasons, control of this cell bank by specific PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on December 8, 2014
Figure 2 Transcriptome analysis of non-productively infected cells. RNas were extracted from an equine sarcoid, a tumor due to Bovine Papillomavirus type 1. Blind analysis of the cells identified BPV1 sequences, and mapping of the reads on the BPV1 genomes shows that the virus is blocked in the early phase of the cycle. PCRs would be of low utility (independently of regulatory requirements) (Table I). The same considerations apply also for the control of the bulks each time the adventitious virus could only be an infectious virus originating from the production cells (meaning in all the cases where no biological reagent has been used during the post–Working Cell Bank production steps). In case of a positive result (cell or supernatant), the result should be discussed (in particular with regard to the number and distribution of reads along the genome), and may be if necessary more deeply investigated by other tools like specific (RT)PCRs designed on the sequence evidenced. 4. Use of WG-NGS for Testing Raw Materials As discussed above, the analytical sensitivity of some PCRs is better than that of WG-NGS, while some others have the same range of detection as WG-NGS if the primer sequences fit exactly to the virus sequence but could be unable to detect distant strains, and indeed PCRs are unable to detect unknown viruses. Also, many of the members of the increasingly growing list of known viruses are currently not screened by PCR. For example, among the 12 human polyomaviruses, two (JC and BK) are classically screened (but not the ten other ones including at least two human Vol. 68, No. 6, November–December 2014
pathogens– Merkel and Trichodysplasia virus). One should bear in mind that, in contrast to chronically infected cell lines, raw materials can be contaminated by very low amount of adventitious viruses coming from contaminated animal or human tissues or biological fluids diluted in huge pools. These low viral loads might nevertheless be of concern regarding virus safety if they are plated on a permissive cell line or even inoculated in humans. Thus, the use of WG-NGS should be discussed case by case. Some very efficient dedicated PCRs cannot currently be replaced by WG-NGS (for example for all major human pathogens for the control of plasma donations), even if it could be argued that low virus loads not detected by WG-HTS would be cleared by manufacturing processes. WG-NGS might appear currently as a very broad test that can be used regularly to check the viral burden of some batches to refine routine PCR controls and inactivation procedures (11). Alternatively, it might also be used as a unique virus biosafety assay in all cases where there is a strong inactivation process downstream (e.g., gamma irradiation), meaning that the main objective would be to avoid a high load of virus in the raw material. 5. General Conclusion WG-NGS appears as a method of major interest for the testing of cell lines and for the analysis of the viral 637
Downloaded from journal.pda.org on December 8, 2014
burden in animal or human raw materials upstream of inactivation steps. There is now an urgent need to discuss a common strategy for testing cell substrate and propose common concepts and/or reagents for validation. Optimization of sample preparation and bioinformatic pipelines should progressively decrease the need for a very high depth of sequencing. Also, the place of WG-NGS within the set of different in vitro, in vivo, and molecular techniques can now be better defined based on technical and scientific grounds. Acknowledgements This work received financial support from the foundation Biosecure, Paris and Labex IBEID. We thank Jennifer Richardson for critical reading of the manuscript.
tant Klebsiella pneumoniae by targeted next-generation sequencing. J. Clin. Microbiol. 2014, 52 (3), 987–990. 5. Victoria, J. G.; Wang, C.; Jones, M. S.; Jaing, C.; McLoughlin, K.; Gardner, S.; Delwart, E. L. Viral nucleic acids in live-attenuated vaccines: detection of minority variants and an adventitious virus. J. Virol. 2010, 84 (12), 6033– 6040. 6. Cheval, J.; Sauvage, V.; Frangeul, L.; Dacheux, L.; Guigon, G.; Dumey, N.; Pariente, K.; Rousseaux, C.; Dorange, F.; Berthet, N.; Brisse, S.; Moszer, I.; Bourhy, H.; Manuguerra, C. J.; Lecuit, M.; Burguiere, A.; Caro, V.; Eloit, M. Evaluation of high-throughput sequencing for identifying known and unknown viruses in biological samples. J. Clin. Microbiol. 2011, 49 (9), 3268 –3275.
Conflict of Interests Declaration EC and CH are employees of PathoQuest, a spinoff of Institut Pasteur dedicated to the identification of Pathogens by NGS. ME is chairman of PathoQuest. References 1. Banoo, S.; Bell, D.; Bossuyt, P.; Herring, A.; Mabey, D.; Poole, F.; Smith, P. G.; Sriram, N.; Wongsrichanalai, C.; Linke, R.; O’Brien, R.; Perkins, M.; Cunningham, J.; Matsoso, P.; Nathanson, C. M.; Olliaro, P.; Peeling, R. W.; Ramsay, A.; The TDR Diagnostics Evaluation Expert Panel. Evaluation of diagnostic tests for infectious diseases: general principles. Nat. Rev. Microbiol. 2010, 8, S15–28. 2. Mokili, J. L.; Rohwer, F.; Dutilh, B. E. Metagenomics and future perspectives in virus discovery. Curr. Opin. Virol. 2012, 2 (1), 63–77. 3. Wolk, D. M.; Kaleta, E. J.; Wysocki, V. H. PCRelectrospray ionization mass spectrometry: the potential to change infectious disease diagnostics in clinical and public health laboratories. J. Mol. Diagn. 2012, 14 (4), 295–304. 4. Arena, F.; Rolfe, P. A.; Doran, G.; Conte, V.; Gruszka, S.; Clarke, T.; Adesokan, Y.; Giani, T.; Rossolini, G. M. Rapid resistome fingerprinting and clonal lineage profiling of carbapenem-resis-
638
7. Sauvage, V.; Foulongne, V.; Cheval, J.; Ar Gouilh, M.; Pariente, K.; Dereure, O.; Manuguerra, J. C.; Richardson, J.; Lecuit, M.; Burguie` re, A.; Caro, V.; Eloit, M. Human polyomavirus related to African green monkey lymphotropic polyomavirus. Emerg. Infect. Dis. 2011, 17 (8),1364 –1370. 8. Sauvage, V.; Cheval, J.; Foulongne, V.; Gouilh, M. A.; Pariente K.; Manuguerra J. C.; Richardson J.; Dereure O.; Lecuit M.; Burguiere A.; Caro V.; Eloit M. Identification of the first human gyrovirus, a virus related to chicken anemia virus. 2011, J. Virol. 85 (15), 7948 –7950. 9. Sauvage, V.; Ar Gouilh, M.; Cheval, J.; Muth, E.; Pariente, K.; Burguiere, A.; Caro, V.; Manuguerra, J-C.; Eloit, M. A member of a new Picornaviridae Genus is shed in pig feces. J. Virol. 2012, 86 (18), 10036 –10046. 10. Yoon, B.-J. Hidden Markov models and their applications in biological sequence analysis. Curr. Genomics 2009, 10 (6), 402– 415. 11. Gagnieur, L.; Cheval, J.; Gratigny, M.; He´bert, C.; Muth, E.; Dumarest, M.; Eloit, M. Unbiased analysis by high throughput sequencing of the viral diversity in fetal bovine serum and trypsin used in cell culture. Biologicals 2014, 42 (3), 1045–152.
PDA Journal of Pharmaceutical Science and Technology
Downloaded from journal.pda.org on December 8, 2014
An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal
