Pooled CRISPR Inverse PCR sequencing (PCIP-seq): simultaneous sequencing of viral insertion points and the integrated viral genomes with long reads

TECHNICAL FIELD

The present invention relates to a method for detecting an integration pattern of a virus in a host genome, tools for performing the method and applications thereof.

BACKGROUND

The integration of viral DNA into the host genome is a defining feature of the retroviral life cycle, irreversibly linking provirus and cell. This intimate association facilitates viral persistence and replication in somatic cells, and with integration into germ cells bequeaths the provirus to subsequent generations. Considerable effort has been expended to understand patterns of proviral integration, both from a basic virology stand point, and due to the use of retroviral vectors in gene therapy ¹ . The application of next generation sequencing (NGS) over the last ~10 years has had a dramatic impact on our ability to explore the landscape of retroviral integration for both exogenous and endogenous retroviruses. Methods based on ligation mediated PCR and lllumina sequencing have facilitated the identification of hundreds of thousands of insertion sites in exogenous viruses such as Human T-cell leukemia virus-1 (HTLV-1) ² and Human immunodeficiency virus (HIV- 1) ^3'6 . These techniques have shown that in HTLV-1 ² , Bovine Leukemia Virus (BLV) ⁷ and Avian Leukosis Virus (ALV) ⁸ integration sites are not random, pointing to clonal selection. In HIV-1 it has also become apparent that provirus integration can drive clonal expansion ³ ' ⁴ ' ⁶ · ⁹ , magnifying the HIV-1 reservoir and placing a major road block in the way of a complete cure.

Current methods based on short-read (high throughput) sequencing identify the insertion point, but the provirus itself is largely unexplored. Whether variation in the provirus influences the fate of the clone remains difficult to investigate. Using long range PCR it has been shown that proviruses in HTLV-1 induced Adult T-cell leukemia (ATL) are frequently (-45%) defective ¹⁰ , although the abundance of defective proviruses within asymptomatic HTLV-1 carriers has not been systematically investigated. Recently, there has been a concerted effort to better understand the structure of HIV-1 proviruses in the latent reservoir. Methods such as Full-Length Individual Proviral Sequencing (FLIPS) have been developed to identify functional proviruses ¹¹ but without identifying the provirus integration site. More recently matched integration site and proviral sequencing (MIP-Seq) has allowed the sequence of individual proviruses to be linked to integration site in the genome ⁶ . However, this method relies on whole genome amplification of isolated HIV-1 genomes, with separate reactions to identify the integration site and sequence the associated provirus ⁶ . As a result, this method is quite labor intensive limiting the number of proviruses one can reasonably interrogate.

Retroviruses are primarily associated with the diseases they provoke through the infection of somatic cells. Over the course of evolutionary time they have also played a major role in shaping the genome. Retroviral invasion of the germ line has occurred multiple times, resulting in the remarkable fact that endogenous retrovirus (ERV)-like elements comprise a larger proportion of the human genome (8%) than protein coding sequences (~1.5 %) ¹² . With the availability of multiple vertebrate genome assemblies, much of the focus has been on comparison of ERVs between species. However, single genomes represent a fraction of the variation within a species, prompting some to take a population approach to investigate ERV-host genome variation ¹³ . While capable of identifying polymorphic ERVs in the population, approaches relying on conventional paired-end libraries and short reads cannot capture the sequence of the provirus beyond the first few hundred bases of the proviral long terminal repeat (LTR), leaving the variation within uncharted.

In contrast to retroviruses, papillomaviruses do not integrate into the host genome as part of their lifecycle. Human papillomavirus (HPV) is usually present in the cell as a multi copy circular episome (~8kb in size), however in a small fraction of infections, it can integrate into the host genome leading to the dysregulation of the viral oncogenes E6 and E7 ¹⁴ . Genome wide profiling of HPV integration sites via capture probes and lllumina sequencing has also identified hotspots of integration indicating that disruption of host genes may also play a role in driving clonal expansion ¹⁵ . As a consequence, HPV integration is a risk factor for the development of cervical carcinoma ¹⁶ .

HPV accounts for >95% of cervical carcinoma and -70% of oropharyngeal carcinoma ⁵² . While infection with a high-risk HPV strain (HPV16 & HPV18) is generally necessary for the development of cervical cancer, it is not sufficient ⁴¹ . The progression towards cancer is driven by a combination of both viral and host factors, as a result, a greater understanding of both is required to identify high risk infections ⁴¹ .

The HPV vaccine will cut the rate of cervical cancer in vaccinated women by -75%, however it will take 20 to 30 years for the full impact of vaccination to become apparent ⁶⁴ . Additionally, vaccination uptake varies widely, with the Belgian French speaking community only having a 36% uptake in 2018 ⁶⁵ . As consequence HPV induced cervical cancer will remain a major health issue in the medium term and the cause of a nontrivial number of cancers into the foreseeable future.

The centrality of HPV integration in carcinogenesis makes a deeper understanding of the process a priority, both to understand the basic biology behind HPV induced cervical cancer, but also because of its potential as a biomarker to identify high risk cases sooner. The study of HPV integration is hampered by the unpredictability of the breakpoint sites in the integrated HPV genome. This limits the applicability of approaches based on ligation mediated PCR and short read sequencing. Techniques such as real-time PCR can identify HPV infections, but cannot identify integrations associated with clonal expansion. Biotin capture probes and lllumina sequencing have provided an unbiased genome wide picture of integration sites in cervical carcinomas, hinting at potential hot spots of integration ¹⁵ . However, this technique is not suited to exploring precancerous stages, where only a small fraction of the cells carries integrated virus. Looking beyond integration sites, work on HPV16 using a targeted sequencing approach has shown that conservation of the HPV E7 gene is critical for carcinogenesis ⁶⁶ .

The application of NGS as well as Sanger sequencing before, has had a large impact on our understanding of both exogenous and endogenous proviruses. The development of long-read sequencing, linked-read technologies and associated computational tools ¹⁷ have the potential to explore questions inaccessible to short reads. Groups investigating Long interspersed nuclear elements-1 (LINE-1) insertions ¹⁸ and the koala retrovirus, KoRV ¹⁹ have highlighted this potential and described techniques utilizing the Oxford Nanopore and PacBio platforms, to investigate insertion sites and retroelement structure.

SUMMARY OF THE INVENTION

To more fully exploit the potential of long reads we developed Pooled CRISPR Inverse PCR sequencing (PCIP-seq), a method that leverages selective cleavage of circularized DNA fragments carrying proviral DNA / integrated viral DNA with CRISPR guide RNAs or a pool of CRISPR guide RNAs, followed by inverse long-range PCR and multiplexed sequencing, such as on the Oxford Nanopore MinlON platform. Using this approach, we can now simultaneously identify the integration site and track clone abundance while also sequencing the provirus / viral DNA inserted at that position. We have successfully applied the technique to the retroviruses HTLV-1, HIV-1 and BLV, endogenous retroviruses in cattle and sheep as well as HPV18 and HPV16. In an aspect, the invention provides a method for detecting an integration pattern of human papillomavirus (HPV) in genomic DNA of a subject, said method comprising:

(a) fragmenting genomic DNA isolated from a sample of the subject;

(b) circularizing the DNA fragments to generate circular DNA;

(c) removing non-circularized DNA fragments;

(d) linearizing the circular DNA using an RNA-guided DNA endonuclease and at least one guide RNA or at least one pool of guide RNAs, which target a region in the viral genome, to generate linearized DNA molecules;

(e) amplifying the linearized DNA molecules by an inverse amplification reaction using a pair of primers arranged about and oriented outwardly with respect to the linearization site;

(f) sequencing the amplified DNA;

(g) mapping the sequenced DNA to human genomic DNA sequence; and

(h) optionally mapping the sequenced DNA to the HPV genome.

The invention also provides for a kit for detecting an integration pattern of human papillomavirus (HPV) in genomic DNA of a subject according to the method of of the invention, said kit comprising:

- at least one first guide RNA or at least one first pool of guide RNAs, which target a first region in the viral genome, preferably wherein said first region of the viral DNA comprises E6 gene and/or E7 gene; and/or, preferably and,

-a pair of primers arranged about and oriented outwardly with respect to a first linearization site in the viral genome defined by said at least one first guide RNA or at least first one pool of guide RNAs.

A further aspect relates to a method for monitoring the progression of a human papillomavirus (HPV) infection in a subject comprising:

- detecting an integration pattern of human papillomavirus (HPV) in genomic DNA isolated from a sample of the subject according to the method of the invention; and

- comparing said integration pattern with an integration pattern of HPV in genomic DNA isolated from a sample of the subject at an earlier point in time.

A further aspect relates to a method for assessing a risk of having or developing a cancer in a subject comprising:

- detecting an integration pattern of human papillomavirus (HPV) in genomic DNA of the subject according to the method of the invention; and

- determining whether the integration pattern predisposes the subject to cancer or cancer development. These and further aspects and preferred embodiments of the invention are described in the following sections and in the appended claims. The subject-matter of the appended claims is hereby specifically incorporated in this specification.

BRIEF DESCRIPTION OF THE FIGURES

The teaching of the application is illustrated by the following Figures which are to be considered as illustrative only and do not in any way limit the scope of the claims.

Figure 1. Overview of the PCIP-seq method (a) Simplified outline of method (b) A pool of CRISPR guide-RNAs targets each region, the region is flanked by PCR primers. Guides and primers adjacent to 5' & 3' LTRs are multiplexed, (c) As the region between the PCR primers is not sequenced we created two sets of guides and primers. Following circularization, the sample is split, with CRISPR mediated cleavage and PCR occurring separately for each set. After PCR the products of the two sets of guides and primers are combined for sequencing, (d) Screen shot from the Integrative Genomics Viewer (IGV) showing a small fraction of the resultant reads (grey bars) mapped to the provirus, coverage is shown on top, coverage drops close to the 5’ and 3’ ends are regions flanked by primers.

Figure 2. PCIP-seq applied to ATL (a) In ATL100 both lllumina and Nanopore based methods show a single predominant insertion site (b) Screen shot from IGV shows a ~16kb window with the provirus insertion site in the tumor clone identified via PCIP-seq and ligation mediated PCR with lllumina sequencing (c) PCIP-seq reads in IGV show a ~3,600bp deletion in the provirus, confirmed via long range PCR and lllumina sequencing, (d) The ATL2 tumor clone contains three proviruses (named according to chromosome inserted into), the provirus on chr1 inserted into a repetitive element (LTR) and short reads generated from host DNA flanking the insertion site map to multiple positions in the genome. Filtering out multi-mapping reads causes an underestimation of the abundance of this insertion site (13.6 %), this can be partially corrected by retaining multi-mapping reads at this position (25.4 %). However, that approach can cause the potentially spurious inflation of other integration sites (red slice 9%). The long PCIP-seq reads can span repetitive elements and produce even coverage for each provirus without correction, (e) Screen shot from IGV shows representative reads coming from the three proviruses at positions where four de novo mutations were observed.

Figure 3. (a) Screen shot from IGV shows representative reads from a subset of the clones from each BLV-infected animal with a mutation in the first base of codon 303 in the viral protein Tax. (b) Structural variants observed in the BLV provirus. BLV sense and antisense transcripts are shown on top. Deletions (blue bars) and duplications (red bars) observed in the BLV provirus from both ovine and bovine samples are shown below. Figure 5. HPV ‘looping’ integration in an expanded clone (a) PCIP-seq reads mapping to a ~87 kb region on chr3 revealed three HPV-host breakpoints. The large number of reads suggests expansion of the clone carrying these integrations, (b) PCR was carried out with primer pairs matching regions a and b, as well as a and g. Both primer pairs produced a ~9kb PCR product. Nanopore sequencing of the PCR products show the HPV genome connects these breakpoints, (c) Schematic of the breakpoints with the integrated HPV genome. This conformation indicates that this dramatic structural rearrangement in the host genome was generated via ‘looping’ integration of the HPV genome.

Figure 6. (a) Reads from four HPV16 samples mapped to the HPV16 subtype A1 genome. Vertical lines identify position where the base differs from the reference genome, (b) Consensus sequences were generated for 12 HPV16 samples and a phylogenetic tree with the HPV16 subtype reference genomes (highlighted) was generated. The 12 samples cluster with the HPV16A1 and A2, both are European isolates.

Figure 7. Clone persistence was observed in two patients. The first patient had an integration in the LAPTM4B gene (histology= ASC-H), a second sampling from 7 months later (upgraded to HSIL) showed the same integration sites (a) The discordant breakpoints again points to 'looping' integration in an expanded clone, (b) When the reads are mapped to the HPV genome the sample from July 2019 has reads originating from episomal copies of HPV as well as reads from the integrated copy of HPV. All the HPV reads from the December 2019 sample contain the deletion associated with the integrated copy of HPV indicating that the infection has cleared but the clonally expanded cell remains. PCR with primer pairs matching regions a and 13 produced a ~9kb PCR product, again indicating that the integration has caused a structural rearrangement in this region, (c) In the second patient (a 71 year old, histology= ASC-US at both time points) HPV was found to be integrated at three positions in the genome (within exons of the genes TMEM177, IL20RB and ARMH3 ), introducing at least three copies of HPV (E6 and E7 are intact in thep integrated HPV genomes). It is not possible to tell at this point if all are in the same or separate clones, (d) For both time points the integrated HPV reads represent -10% of the total HPV reads, although the greater number of unique shear sites in the second time point (especially for the chr2 integration) suggest the clone may be expanding.

Figure 8. Use of Cas-9 mediated cleavage in the PCIP-seq method. 8 μg of DNA from a BLV infected sheep with a proviral load of 82.6% was circularized and linear DNA was eliminated. One quarter of the resultant DNA was subject to CRISPR-cas9 cleavage using the Pool A guides (CRISPR+, PA), the second quarter was cleaved using the Pool B guides (CRISPR+, PB), the remaining half was kept aside. The linearized DNA was cleaned and used as template in 2x 50mI PCR reactions using the appropriate primer pairs for Pool A (PA) or Pool B (PB). For the uncut DNA half was used as template for2x50pl PCR reactions using the PA primers (CRISPR-, PA) and the other half was used for 2x 50mI PCR reactions using the PB primers (CRISPR-, PB). Following 25 PCR cycles, 10mI of each reaction were loaded on a 1% agarose gel. A=unshared genomic DNA, B=genomic DNA sheared to 8kb.

Figure 9. Coverage of the pure viral reads as well as the chimeric reads produced by the libraries shown in Figure 8 on the BLV proviral genome. BC refers to the barcode used for each library.

Figure 10. Pie charts showing the relative abundance of the 200 largest clones in the four sheep (top) and three cattle (bottom) infected with BLV, each slice of the pie represents a single insertion site, the % below indicated what fraction of the overall reads these 200 clones represent.

DESCRIPTION

As used herein, the singular forms “a”, “an”, and “the” include both singular and plural referents unless the context clearly dictates otherwise.

The terms “comprising”, “comprises” and “comprised of” as used herein are synonymous with “including”, “includes” or “containing”, “contains”, and are inclusive or open-ended and do not exclude additional, non-recited members, elements or method steps. The terms also encompass “consisting of” and “consisting essentially of”, which enjoy well-established meanings in patent terminology.

The recitation of numerical ranges by endpoints includes all numbers and fractions subsumed within the respective ranges, as well as the recited endpoints. This applies to numerical ranges irrespective of whether they are introduced by the expression “from... to...” or the expression “between... and...” or another expression.

The terms “about” or “approximately” as used herein when referring to a measurable value such as a parameter, an amount, a temporal duration, and the like, are meant to encompass variations of and from the specified value, such as variations of +/-10% or less, preferably +1-5% or less, more preferably +/- 1% or less, and still more preferably +/-0.1% or less of and from the specified value, insofar such variations are appropriate to perform in the disclosed invention. It is to be understood that the value to which the modifier “about” or “approximately” refers is itself also specifically, and preferably, disclosed.

Whereas the terms “one or more” or “at least one”, such as one or more members or at least one member of a group of members, is clear per se, by means of further exemplification, the term encompasses inter alia a reference to any one of said members, or to any two or more of said members, such as, e.g., any ³3, ³4, ³5, ³6 or ³7 etc. of said members, and up to all said members. In another example, “one or more” or “at least one” may refer to 1, 2, 3, 4, 5, 6, 7 or more.

The discussion of the background to the invention herein is included to explain the context of the invention. This is not to be taken as an admission that any of the material referred to was published, known, or part of the common general knowledge in any country as of the priority date of any of the claims.

Throughout this disclosure, various publications, patents and published patent specifications are referenced by an identifying citation. All documents cited in the present specification are hereby incorporated by reference in their entirety. In particular, the teachings or sections of such documents herein specifically referred to are incorporated by reference.

Unless otherwise defined, all terms used in disclosing the invention, including technical and scientific terms, have the meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. By means of further guidance, term definitions are included to better appreciate the teaching of the invention. When specific terms are defined in connection with a particular aspect of the invention or a particular embodiment of the invention, such connotation or meaning is meant to apply throughout this specification, i.e., also in the context of other aspects or embodiments of the invention, unless otherwise defined.

In the following passages, different aspects or embodiments of the invention are defined in more detail. Each aspect or embodiment so defined may be combined with any other aspect(s) orembodiment(s) unless clearly indicated to the contrary. In particular, any feature indicated as being preferred or advantageous may be combined with any other feature or features indicated as being preferred or advantageous.

Reference throughout this specification to “one embodiment”, “an embodiment” means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrases “in one embodiment” or “in an embodiment” in various places throughout this specification are not necessarily all referring to the same embodiment, but may. Furthermore, the particular features, structures or characteristics may be combined in any suitable manner, as would be apparent to a person skilled in the art from this disclosure, in one or more embodiments. Furthermore, while some embodiments described herein include some but not other features included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the invention, and form different embodiments, as would be understood by those in the art. For example, in the appended claims, any of the claimed embodiments can be used in any combination.

For general methods relating to the invention, reference is made inter alia to well-known textbooks, including, e.g., “Molecular Cloning: A Laboratory Manual, 4th Ed.” (Green and Sambrook, 2012, Cold Spring Harbor Laboratory Press), “Current Protocols in Molecular Biology” (Ausubel et al., 1987).

Provided herein is a method for detecting an integration pattern of a virus in genomic DNA of a subject, said method comprising:

(a) fragmenting genomic DNA isolated from a sample of the subject;

(b) circularizing the DNA fragments to generate circular DNA;

(c) removing non-circularized DNA fragments;

(d) optionally linearizing the circular DNA using an RNA-guided DNA endonuclease and at least one guide RNA or at least one pool of guide RNAs, which target a region in the viral genome to generate linearized DNA molecules;

(e) amplifying the circular DNA or the linearized DNA molecules by an inverse amplification reaction using a pair of primers arranged about and oriented outwardly with respect to the linearization site;

(f) sequencing the amplified DNA;

(g) mapping the sequenced DNA to genomic DNA sequence of the subject; and

(h) optionally mapping the sequenced DNA to the viral genome.

As used herein, the terms “integration pattern” or “viral integration pattern” refer to the pattern of viral DNA that is integrated in host genomic DNA. The term may refer to a visualized DNA pattern comprising viral DNA and host genomic DNA, as well as to information quantified by or correlated with such DNA pattern. Non-limiting examples of information quantified by, or correlated with an integration pattern include the presence of absence of integrated viral DNA; the number of viral integration sites in host genomic DNA or the average number of such integrations; the insertion site(s) of viral DNA in the host genome; mutations (e.g. deletions, duplications, SNPs, etc.) in the viral DNA integrations; the size in kb of viral DNA integrations into host genomic DNA; the number of viral genomes integrated at each integration site; the number of viral integration sites per cellular genome; the mean number of viral genomes integrated per integration site (or the mean size of integration sites); maximum number of viral genomes integrated per integration site (or the maximum size of integration sites); minimum number of viral genomes integrated per integration site (or minimum size of integration sites), number of viral genomes integrated per cellular genome, and any combinations thereof.

The method of the invention allows to detect integration of viruses such as retroviruses that integrate into a host cell genome as part of their lifecycle, as well as viruses such as papillomaviruses that do not integrate into a host cell genome as part of their lifecycle. The virus may be a DNA virus or an RNA virus. DNA viruses include, for example, human papillomavirus (HPV); RNA viruses include, for example, human T lymphophilic virus (HTLV, particularly HTLV-1), human immunodeficiency virus (HIV), bovine leukemia virus (BLV). In embodiments, the virus is a retrovirus. In further embodiments, the retrovirus is an exogenous retrovirus such as HTLV, in particular HTLV-1 , HIV or BLV. In further embodiments, the retrovirus is an endogenous retrovirus. In other embodiments, the virus is HPV. In further embodiments, said HPV is a high risk HPV such as a HPV strain 16, 18, 31, 33, 35, 39, 45, 51 , 55, 56, 58, 59 or 66, preferably a HPV strain 18 or a HPV strain 16.

"Integrated viral DNA" refers to a complete or partial genome of a virus that is integrated into a host cell chromosome. "Episomal viral DNA" refers to non-integrated viral DNA, i.e. , viral DNA that has not integrated into a host cell chromosome. “Provirus” refers to viral DNA, in particular retroviral DNA, that is integrated into the DNA of a host cell as a stage of virus replication, or a state that persists over longer periods of time as either inactive viral infections or an endogenous viral element.

The terms "subject" and “host” and "patient" are used interchangeably and refer to a human or non-human animal that is tested for the presence of integrated viral DNA. The host is not particularly limited as long as the virus infects and viral nucleic acid is integrated into the genome. Preferably, the host is a mammal, most preferably a human. Hosts may be domestic animals such as cows, horses, pigs, sheep, goats and chickens. In preferred embodiments, the subject is a human. In embodiments, the subject is an ovine. In embodiments, the subject is a bovine.

The term "sample" generally refers to a material of biological origin that includes cells. Samples can include, e.g., an in vitro cell culture ortissue obtained from a subject as defined herein. Samples can be purified or semi-purified to remove certain constituents (e.g., extracellular constituents or non-target cell populations). In embodiments, the sample comprises cervical or vaginal epithelial cells, such as wherein the sample is a pap smear. In embodiments, the sample comprises oropharyngeal epithelial cells, such as wherein the sample is an oropharyngeal swab. In embodiments, the sample comprises peripheral blood mononuclear cells (PBMC), in particular CD4+ T cells, such as wherein the sample is a blood sample, e.g. a whole blood sample. In embodiments, the sample is a sperm sample. Isolation of DNA from the samples can be carried out by standard methods.

In step (a) genomic DNA of the subject is fragmented. In embodiments, fragmenting the genomic DNA of the subject comprises shearing the genomic DNA, thereby producing (sheared) DNA fragments. Shearing of the genomic DNA may occur e.g. by acoustic or mechanical means as known to the skilled person. In further embodiments, shearing of the genomic DNA of the subject is followed by end-repair of the sheared DNA fragments.

In embodiments, the (sheared) DNA fragments have an average size of about the size of the viral genome. In particular embodiments, the (sheared) DNA fragments have an average size of between 6000 and 10000 basepairs (bp), preferably between 7000 and 9000 bp, more preferably about 8000 bp.

In step (b) of the method (sheared) DNA fragments are circularized. Circularization or intramolecular ligation of the DNA fragments may be achieved by incubation of the DNA fragments in the presence of a DNA ligase, e.g. T4 DNA ligase, as known to the skilled person, thereby generating circular DNA.

Step (c) of the method encompasses removal of remaining linear DNA. In embodiments, non-circularized DNA is removed by digestion. Selective digestion of non-circularized or linear DNA may be achieved using an appropriate selective DNase as commercially available (e.g. Plasmid-Safe™ ATP-Dependent DNase (Epicentre).

Preferably, the circular DNA is linearized in step (d) before the amplification step (e), which improves the efficiency of the amplification reactions. Linearization of the circular DNA can be achieved using an RNA-guided DNA endonuclease, such as a CRISPR-Cas system as known to the skilled person, and corresponding guide RNAs. In particular embodiments, the RNA-guided DNA endonuclease is a Cas-9 endonuclease.

In order to achieve selective linearization of circular DNA that comprises integrated viral DNA and host DNA, guide RNA(s) are used that target a region of the viral DNA. Preferably, the “linearization site”, i.e. the region in the viral DNA that is targeted by a guide RNA or a pool of guide RNAs, comprises a region of the viral genome that is prone to integration in host DNA. For example, for HPV, a linearization site may comprise E6 gene and/or E7 gene. For retroviruses, a linearization site may be adjacent to a 5’LTR or adjacent to a 3’LTR.

Particular guide RNA targeting domains and pools of guide RNA targeting domains are provided in Table 1. The sequences set forth in SEQ ID NO:7-79 refer to oligonucleotide sequences used for synthesizing the guide RNAs. These sequences comprise a “targeting domain” as well as accessory sequences required by the kit, in particular the EnGen® sgRNA Synthesis Kit (New England Biolabs), for synthesizing the guide RNA, which elements can be identified by the skilled person. By way of example, oligonucleotide sequences encoding HPV18 and HPV16 gRNAs and their corresponding targeting domain and flanking PAM site (underlined) are summarized in the below table. With “targeting domain” is meant herein a sequence that is capable of hybridizing to a sequence in the region of the viral DNA that is targeted by the guide RNA (i.e. in the linearization site of the viral DNA). Wth “PAM site” is meant herein a protospacer adjacent sequence as is known in the art. When reference is made to a guide RNA comprising a sequence set forth in any one of SEQ ID NO:7-79, a guide RNA comprising the targeting domain of said sequence is envisaged, i.e. the sequence without the sequence TTCTAATACGACTCACTATA (SEQ ID NO:244) 5 prime and without the sequence GTTTTAGAGCTAGA (SEQ I D NO:245) 3 prime. When reference is made to a guide RNA comprising a sequence set forth in any one of SEQ ID NO:232-243, a guide RNA comprising the targeting of said sequence is envisaged, i.e. the sequence without the NGG sequence 3 prime. As will be appreciated by the skilled person, the guide RNA comprises in addition to a targeting domain, a tracer and a tracer mate as known in the art, wherein the tracer and tracer mate may be provided chimeric. The guide RNA is an RNA molecule and will therefore comprise the base uracil (U), while the oligonucleotide encoding the gRNA molecule comprises the base thymine (T).

To improve cleavage of a linearization site, more than one guide RNA targeting said linearization site can be used. As used herein, a “pool of guide RNAs” refers to a set of guide RNAs that target a defined region of the viral DNA, i.e. the linearization site. It is to be understood that each guide RNA within a pool of guide RNAs may be capable of hybridizing to different, non-overlapping or partially overlapping, sequences within said linearization site. A pool of guide RNAs may comprise at least 2 or at least 3 guide RNAs, preferably at least 3 guide RNAs, more preferably between 3 and 10 or between 3 and 8 guide RNAs, such as 3, 4, 5, 6, 7 or 8 guide RNAs.

The circular DNA may be linearized using a first guide RNA or a first pool of guide RNAs, which target a first region of the viral DNA, and at least one other guide RNA or at least one other pool of guide RNAs, which target a non-overlapping region(s) of the viral RNA. When targeting more than one linearization site, a more complete integration pattern may be obtained (e.g. more integration sites may be detected).

Accordingly, in embodiments, a first portion of the circular DNA is linearized using a first guide RNA or a first pool of guide RNAs that target a first region of the viral DNA to generate a first set of linearized DNA molecules; and a second portion of the circular DNA is linearized using a second guide RNA or a second pool of guide RNAs that target a second region of the viral DNA to generate a second set of linearized DNA molecules, wherein the first region and the second region of the viral DNA do not overlap.

In embodiments of the method for detecting an integration pattern of a retrovirus in genomic DNA of a subject, a first portion of the circular DNA is linearized using a first guide RNA or a first pool of guide RNAs that target a region of the viral DNA adjacent to the 5’ long terminal repeat (LTR) to generate a first set of linearized DNA molecules; and a second portion of the circular DNA is linearized using a second guide RNA or a second pool of guide RNAs that target a region of the viral DNA adjacent to the 3’LTR to generate a second set of linearized DNA molecules.

In embodiments of the method for detecting an integration pattern of a HPV in genomic DNA of a subject, a first portion of the circular DNA is linearized using a first guide RNA or a first pool of guide RNAs that target a first region of the viral DNA comprising E6 gene and/or E7 gene to generate a first set of linearized DNA molecules; and a second portion of the circular DNA is linearized using a second guide RNA or a second pool of guide RNAs that target a second region of the viral DNA to generate a second set of linearized DNA molecules, wherein said first and second regions of the viral DNA do not overlap.

In the amplification step (e), the circular DNA or preferably the linearized DNA molecules are amplified by an inverse amplification reaction using a pair of primers arranged about and oriented outwardly with respect to the linearization site. In particular, a primer pair is used comprising a forward primer capable of hybridizing to a viral DNA sequence in a 3’ flanking region of the viral DNA region targeted by the guide RNA or the pool of guide RNAs and a reverse primer capable of hybridizing to a viral DNA sequence in a 5’ flanking region of the viral DNA region targeted by the guide RNA or the pool of guide RNAs.

Particular primer pairs corresponding to the guide RNA targeting domains or pools of guide RNA targeting domains of Table 1 are provided in Table 2. The primers in Table 2 may comprise a tail, in particular a tail consisting of the sequence (SEQ ID NO:246) or the sequence (SEQ ID NO:247). When reference is made herein to a primer comprising a sequence set forth in any one of SEQ ID NO:80-127, the tailed primer as well as a corresponding primer without the tail or with another tail are envisaged herein.

Preferably, each set of linearized DNA molecules (i.e. linearized DNA molecules generated by one guide RNA or one pool of guide RNAs as described herein and thus characterized by cleavage in a defined linearization site) is amplified in a separate amplification reaction using an appropriate pair of primers arranged about and oriented outwardly with respect to the linearization site.

In further embodiments, the linearization step and the amplification step may be carried out in a single solution, wherein a guide RNA or a pool of guide RNAs and a corresponding pair of primers are multiplexed.

In preferred embodiments, said amplification reaction comprises a long range amplification reaction such as a long range PCR. As used herein, “long range PCR” refers to a method to amplify DNA fragments of increased size, typically of more than 3-5 kb, using a modified DNA polymerase or high-fidelity DNA polymerase. DNA polymerases for long range PCR are known to the skilled person and are commercially available.

In further embodiments, tailed primers are used in the amplification reaction and the amplicons are subjected to a second amplification reaction using a set of indexing primers, thereby generating indexed amplification products. This facilitates multiplexed sequencing of the amplified DNA.

Particular methods are provided herein for detecting an integration pattern of a retrovirus in genomic DNA of a subject, said method comprising:

(a) fragmenting genomic DNA isolated from a sample of the subject;

(b) circularizing the DNA fragments to generate circular DNA;

(c) removing non-circularized DNA fragments;

(d) linearizing the circular DNA using an RNA-guided DNA endonuclease and at least one guide RNA or at least one pool of guide RNAs, which target a region in the viral genome adjacent to the 5’ long terminal repeat (LTR) or adjacent to the 3’LTR to generate linearized DNA molecules;

(e) amplifying the linearized DNA molecules by an inverse amplification reaction using a pair of primers arranged about and oriented outwardly with respect to the linearization site;

(f) sequencing the amplified DNA;

(g) mapping the sequenced DNA to genomic DNA sequence of the subject; and

(h) optionally mapping the sequenced DNA to the viral genome.

In further embodiments of the method for detecting an integration pattern of a retrovirus in genomic DNA of a subject, the linearization of the circular DNA comprises linearizing a first portion of the circular DNA using a first guide RNA or a first pool of guide RNAs, preferably a first pool of guide RNAs, which target a region of the viral DNA adjacent to the 5’ long terminal repeat (LTR) to generate a first set of linearized DNA molecules, and linearizing a second portion of the circular DNA using a second guide RNA or a second pool of guide RNAs, preferably a second pool of guide RNAs, which target a region of the viral DNA adjacent to the 3’LTR to generate a second set of linearized DNA molecules; and the amplification of the linearized DNA molecules comprises amplifying the first set of linearized DNA molecules using a first pair of primers arranged about and oriented outwardly with respect to the viral DNA region adjacent to the 5’ LTR targeted by the first guide RNA or the first pool of guide RNAs, and amplifying the second set of linearized DNA molecules using a second pair of primers arranged about and oriented outwardly with respect to the viral DNA region adjacent to the 3’ LTR targeted by the second guide RNA or the second pool of guide RNAs.

A further aspect relates to a kit for performing the method described herein, said kit comprising: - at least one first guide RNA or at least one first pool of guide RNAs, which target a first region of the viral DNA; and/or, preferably and,

- a pair of primers arranged about and oriented outwardly with respect to a first linearization site in the viral DNA defined by said at least one first guide RNA or at least one first pool of guide RNAs.

In further embodiments, the kit comprises:

- a first guide RNA or a first pool of guide RNAs, which target a first region of the viral DNA;

- a second guide RNA or a second pool of guide RNAs, which target a second region of the viral DNA, wherein the first and the second regions of the viral DNA do not overlap;

- a first pair of primers arranged about and oriented outwardly with respect to a first linearization site in the viral DNA defined by said first guide RNA or said first pool of guide RNAs; and/or, preferably and, a second pair of primers arranged about and oriented outwardly with respect to a second linearization site in the viral DNA defined by said second guide RNA or said second pool of guide RNAs.

Particular kits are provided herein for the detection of an integration pattern of a HPV in genomic DNA of a subject according to the method disclosed herein, said kit comprising:

- at least one guide RNA or at least one pool of guide RNAs, which target a region of the viral DNA comprising E6 gene and/or E7 gene; and/or, preferably and

- a pair of primers arranged about and oriented outwardly with respect to a linearization site in the viral DNA defined by said at least one guide RNA or at least one pool of guide RNAs.

In other embodiments, said kit comprises:

- at least one guide RNA or at least one pool of guide RNAs, which target a region of the viral DNA comprising or adjacent to L1 gene; and

- a pair of primers arranged about and oriented outwardly with respect to a linearization site in the viral DNA defined by said at least one guide RNA or at least one pool of guide RNAs.

In further embodiments, said kit for the detection of an integration pattern of a HPV comprises:

- a first guide RNA or a first pool of guide RNAs, which target a first region of the viral DNA comprising E6 gene and/or E7 gene;

- a first pair of primers arranged about and oriented outwardly with respect to a linearization site in the viral DNA defined by said first guide RNA or said first pool of guide RNAs; - a second guide RNA or a second pool of guide RNAs, which target a second region of the viral DNA, wherein said first and second regions of the viral DNA do not overlap; and

- a second pair of primers arranged about and oriented outwardly with respect to a linearization site in the viral DNA defined by said second guide RNA or said second pool of guide RNAs.

In particular embodiments, said second region of the viral DNA comprises a region of the viral DNA comprising L1 gene or a region of the viral DNA adjacent to L1 gene.

Particular embodiments for the guide RNAs, pools of guide RNAs and primer pairs are as described above for the method. Particular combinations of guide RNA targetind domains or pools of guide RNA targeting domains and primer pairs are described in Tables 1 and 2.

The kit may also contain reagents, e.g., buffers, enzymes and other necessary reagents, for performing the method described above. In particular embodiments, the kit further comprises an RNA-guided DNA endonuclease. In particular embodiments, the kit further comprises a DNA polymerase, preferably a DNA polymerase for long range PCR.

The various components of the kit may be present in separate containers or certain compatible components may be pre-combined into a single container, as desired.

The herein disclosed aspects and embodiments of the invention are further supported by the following non-limiting examples.

EXAMPLES

Example 1 : Materials and methods

Samples

Both the BLV infected sheep ⁷ and HTLV-1 samples ⁷ · ²⁰ have been previously described. Briefly, the sheep were infected with the molecular clone pBLV344 ²¹ , following the experimental procedures approved by the University of Saskatchewan Animal Care Committee based on the Canadian Council on Animal Care Guidelines (Protocol #19940212). The HTLV-1 samples ⁷ · ²⁰ were obtained with informed consent following the institutional review board-approved protocol at the Necker Hospital, University of Paris, France, in accordance with the Declaration of Helsinki. The BLV bovine samples were natural infections, obtained from commercially kept adult dairy cows in Alberta, Canada. Sampling was approved by VSACC (Veterinary Sciences Animal care Committee) of the University of Calgary: protocol number: AC15-0159. The bovine 571 used for ERV identification was collected as part of this cohort. The two sheep samples used for Jaagsiekte sheep retrovirus (enJSRV) identification were the BLV infected ovine samples (220 & 221 (032014)), with a PVL of 3.8 and 16% respectively. PBMCs were isolated using standard Ficoll-Hypaque separation. The DNA for the bovine Mannequin was extracted from sperm, while the DNA for bovine 10201e6 was extracted from whole blood using standard procedures. The HIV-1 U1 cell line DNA sequenced without dilution was provided by Dr. Carine Van Lint, IBMM, Gosselies, Belgium. The HIV-1 U1 cell line dilutions in Jurkat were generated at Ghent University Hospital.

HPV material was prepared from PAP smears obtained from HPV-infected patients at the CHU Liege University hospital. Both patients were PCR positive for HPV18, HPV18_PY was classified as having Atypical Squamous Cell of Undetermined Significance (ASC-US), while HPV18_PX was classified as having Atypical Glandular Cells (AGC). Patients provided written informed consent and the study was approved by the Comite d’Ethique Hospitalo-Facultaire Universitaire de Liege (Reference number: 2019/139). No statistical test was used to determine adequate sample size and the study did not use blinding.

PCIP-seq

Total genomic DNA isolation was carried out using the Qiagen AllPrep DNA/RNA/miRNA kit (BLV, HTLV-1 and HPV infected individuals) or the Qiagen DNeasy Blood & Tissue Kit (HIV-1 patients) according to manufacturer’s protocol. High molecular weight DNA was sheared to ~8kb using Covaris g-tubesTM (Woburn, MA) or a Megaruptor (Diagenode), followed by end-repair using the NEBNext EndRepair Module (New England Biolabs). Intramolecular circularization was achieved by overnight incubation at 16°C with T4 DNA Ligase. Remaining linear DNA was removed with Plasmid-Safe-ATP-Dependent DNAse (Epicentre, Madison Wl). Guide RNAs were designed using chopchop (http://chopchop.cbu.uib.no/index.php). The EnGen™ sgRNA Template Oligo Designer (http://nebiocalculator.neb. com/# !/sgrna) provided the final oligo sequence. Oligos were synthesized by Integrated DNA Technologies (IDT). Oligos were pooled and guide RNAs synthesized with the EnGen sgRNA Synthesis kit, S. pyogenes (New England Biolabs). Selective linearization reactions were performed with the Cas-9 nuclease, S. pyogenes (New England Biolabs). (See Example 3 for the rationale behind using of CRISPR-cas9 to cleave the circular DNA). PCR primers flanking the cut sites were designed using primer3 (http://bioinfo.ut.ee/primer3/). Primers were tailed to facilitate the addition of Oxford Nanopore indexes in a subsequent PCR reaction. The linearized fragments were PCR amplified with LongAmp Taq DNA Polymerase (New England Biolabs) and purified using 1x AmpureXP beads, (Beckman Coulter). A second PCR added the appropriate Oxford Nanopore index. PCR products were visualized on a 1% agarose gel, purified using 1x AmpureXP beads and quantified on a Nanodrop spectrophotometer. Indexed PCR products were multiplexed and Oxford Nanopore libraries prepared with either the Ligation Sequencing Kit 1D (SQK-LSK108) or 1D ^A 2 Sequencing Kit (SQK-LSK308) (only the 1 D were used) The resulting libraries were sequenced on Oxford Nanopore MinlON R9.4 or R9.5 flow cells respectively. The endogenous retrovirus libraries were base called using albacore 2.3.1, all other PCIP-seq libraries were base called with Guppy 3.1.5 (https://nanoporetech.com) using the "high accuracy" base calling model. For the endogenous retrovirus libraries, demultiplexing was carried out via porechop (https://github.com/rrwick/Porechop) using the default setting. The HIV, HTLV-1, BLV and HPV PCIP-seq libraries were subjected to a more stringent demultiplexing with the guppy_barcoder (https://nanoporetech.com) tool using the --require_barcodes_both_ends option. The output was also passed through porechop, again barcodes were required on both ends, adapter sequence was trimmed and reads with middle adapters were discarded. Oligos used can be found in Tables 1 and 2.

Table 1 : Guide RNA oligo’s.

Table 2: Primers used for amplification of linearized DNA molecules

Identification of proviral integration sites in PCIP-seq

Reads were mapped with Minimap2 ⁵⁵ to the host genome with the proviral genome as a separate chromosome. In-house R-scripts were used to identify integration sites (IS). Briefly, chimeric reads that partially mapped to at least one extremity of the proviral genome were used to extract virus-host junctions and shear sites. Junctions within a 200bp window were clustered together to form an “IS cluster”, compensating for sequencing/mapping errors. The IS retained corresponded to the position supported by the highest number of virus-host junctions in each IS cluster. Clone abundance was estimated based on the number of reads supporting each IS cluster. Reads sharing the same integration site and same shear site were considered PCR duplicates. Custom software, code description and detailed outline of the workflow are available on Github: https://github.com/GIGA- AnimalGenomics-BLV/PCI R

Measure of proviral load (PVL) and identification of proviral integration sites (lllumina)

PVLs and integration sites of HTLV-1- and BLV-positive individuals were determined as previously described in Rosewick et al 2017 ⁷ and Artesi et al 2017 ²⁰ . PVL represents the percentage of infected cells, considering a single proviral integration per cell. Total HIV-1 DNA content of CD4 T-cell DNA isolates was measured by digital droplet PCR (ddPCR, QX200 platform, Bio-Rad, Temse, Belgium), as described by Rutsaertet al. ⁵⁶ The DNA was subjected to a restriction digest with EcoRI (Promega, Leiden, The Netherlands) for one hour, and diluted 1:2 in nuclease free water. HIV-1 DNA was measured in triplicate using 4 pL of the diluted DNA as input into a 20pL reaction, while the RPP30 reference gene was measured in duplicate using 1 pL as input. Primers and probes are summarized in Table 3. Thermocycling conditions were as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 30 s and 56°C for 60 s, followed by 98°C for 10 min. Data was analyzed with the ddpcRquant analysis software ⁵⁷ . Table 3

Variant Calling

After PCR duplicate removal, proviruses with an IS supported by more than 10 reads were retained for further processing. SNPs were identified using LoFreq ²² with default parameters, only SNPs with an allele frequency of >0.6 in the provirus associated with the insertion site were considered. We also called variants on proviruses supported by more than 10 reads without PCR duplicate removal (this greatly increased the number of proviruses examined). This data was used to explore the number of proviruses carrying the Tax 303 variant. Deletions were called on proviruses supported by more than 10 reads without PCR duplicate removal using an in house R-scripts. Briefly, samtools pileup ⁵⁸ was used to calculate/compute coverage and deletions at base resolution. We used the changepoint detection algorithm PELT ⁵⁹ to identify genomic windows showing an abrupt change in coverage. Windows that showed at least a 4-fold increase in the frequency of deletions (absence of a nucleotide for that position within a read) were flagged as deletions and visually confirmed in IGV ⁶⁰ .

HIV-1 proviral sequences

Sequences of the two major proviruses integrated in chr2 (SEQ ID NO:5) and chrX (SEQ ID NO:4) of the U1 cell line were generated by initially mapping the reads from both platforms to the HIV-1 provirus, isolate NY5 (GenBank: M38431.1), where the 5’LTR sequence is appended to the end of the sequence to produce a full-length HIV-1 proviral genome reference. The sequence was then manually curated to produce the sequence for each provirus. To check for recombination, reads of selected clones were mapped to the sequence from the chrX provirus and the patterns of SNPs examined to determine if the variants matched the chrX or chr2 proviruses. Endogenous retroviruses

The sequence of bovine APOB ERV (SEQ ID NO:6) was generated by PCR amplifying the full length ERV with LongAmp Taq DNA Polymerase (New England Biolabs) from a Holstein suffering from cholesterol deficiency. The resultant PCR product was sequenced on the lllumina platform as described below. It was also sequenced with an Oxford Nanopore MinlON R7 flow cell as previously described ²⁹ . Full length sequence of the element was generated via manual curation. Guide RNAs and primer pairs were designed using this ERV reference. For the Ovine ERV we used the previously published enJSRV-7 sequence ⁴⁰ as a reference to design PCIP-seq guide RNAs and PCR primers.

As the ovine and bovine genome contains sequences matching the ERV, mapping ERV PCIP-seq reads back to the reference genome creates a large pileup of reads in these regions. To avoid this, prior to mapping to the reference we first used BLAST ⁶¹ to identify the regions in the reference genome containing sequences matching the ERV, we then used BEDtools ⁶² to mask those regions. The appropriate ERV reference was then added as an additional chromosome in the reference.

PCR validation and lllumina sequencing

Clone specific PCR products were generated by placing primers in the flanking DNA as well as inside the provirus. LongAmp Taq DNA Polymerase (New England Biolabs) was used for amplification following the manufacturers guidelines. Resultant PCR products were sheared to ~400bp using the Bioruptor Pico (Diagenode) and Nextera XT indexes added as previously described ²⁹ , lllumina PCIP-seq libraries were generated in the same manner. Sequencing was carried out on either an lllumina MiSeq or NextSeq 500. Clone specific PCR products sequenced on Nanopore were indexed by PCR, multiplexed and libraries prepared using the Ligation Sequencing Kit 1 D (SQK-LSK108) and sequenced on a MinlON R9.4 flow cell. Oligos used can be found in Tables 4-7. Table 4: Primers used for clone specific validation of SNPs

Ovine 220 122013

Ovine 221_022016 &

221_032014 to

Bovine 14 ³ 9

Table 5: Primers for clone specific validation of SV

Bovine 1439

Ovine 221

Ovine 223

Table 6: Primers for long range PCR to validate ERVs in the Bovine

Table 7: Primers for long range PCR to validate ERVs in the Ovine

BLV references

The sequence (SEQ ID NO:1) of the pBLV344 provirus was generated via a combination of Sanger and lllumina based sequencing with manual curation of the sequence to produce a full length proviral sequence. The consensus BLV sequences for the bovine samples 1439 & 1053 (SEQ ID NO:3,2) were generated by first mapping the PCIP-seq Nanopore reads to the pBLV344 provirus. We then used Nanopolish ⁶³ to create an improved consensus. PCIP-seq libraries sequenced on the lllumina and Nanopore platform were mapped to this improved consensus visualized in IGV and manually corrected.

Genome references used Sheep=OAR3.1 ; Cattle=UMD3.1 ; Human=hg38; For HTLV-1 integration sites hg19 was used; HPV18=GenBank: AY262282.1 ; Sequences of the exogenous and endogenous proviruses can be found in SEQ ID NO:1-SEQ ID NO:6.

Data availability

Sequence data that support the findings of this study have been deposited in the European Nucleotide Archive (ENA) hosted by the European Bioinformatics Institute (EMBL-EBI) and are accessible through study accession number PRJEB34495. All other relevant data are available within the article and its Supplementary Information files or from the corresponding authors upon reasonable request.

Code availability

The code and a detailed outline of the PCIP-seq analysis workflow are publicly available on Github: https://github.com/GIGA-AnimalGenomics-BLV/PCIP

Example 2: Overview of PCIP-seq (Pooled CRISPR Inverse PCR-sequencing)

The genome size of the viruses targeted ranged from 6.8 to 9.7kb, therefore we chose to shear the DNA to ~8kb in length. In most cases this creates two fragments for each provirus, one containing the 5’ end with host DNA upstream of the insertion site and the second with the 3’ end and downstream host DNA. Depending on the shear site the amount of host and proviral DNA in each fragment will vary (Fig. 1a). To facilitate identification of the provirus insertion site via inverse PCR we carry out intramolecular ligation, followed by digestion of the remaining linear DNA. To selectively linearize the circular DNA containing proviral sequences (this helps increase PCR efficiency), regions adjacent to the 5’ and 3’ LTRs in the provirus are targeted for CRISPR mediated cleavage. We sought a balance between ensuring that the majority of the reads contained part of the flanking DNA (for clone identification) while also generating sufficient reads extending into the midpoint of the provirus. We found that using a pool of CRISPR guides for each region increased the efficiency and by multiplexing the guide pools and PCR primers for the 5’ and 3’ ends we could generate coverage for the majority of a clonally expanded provirus in a single reaction (Fig. 1b). The multiplexed pool of guides and primers leaves coverage gaps in the regions flanked by the primers. To address these coverage gaps we designed a second set of guides and primers. Following separate CRISPR cleavage and PCR amplification the products of these two sets of guides and primers were combined for sequencing (Fig. 1c). This approach ensured that the complete provirus was sequenced (Fig. 1d). Pooled CRISPR Inverse PCR sequencing (PCIP-seq) leverages long reads on the Oxford Nanopore MinlON platform to sequence the insertion site and its associated provirus. The technique was applied to natural infections produced by three exogenous retroviruses, HTLV-1, BLV and HIV-1 as well as endogenous retroviruses in both cattle and sheep. The high efficiency of the method facilitated the identification of tens of thousands of insertion sites in a single sample. Thousands of SNPs and dozens of structural variants within proviruses were observed. While initially developed for retroviruses the method has also been successfully extended to DNA extracted from HPV positive PAP smears, where it could assist in identifying viral integrations associated with clonal expansion. An overview of the applications tested herein is provided in Table 8.

Table 8: Number of insertion sites (IS) identified via PCIP-seq. Chimeric reads = reads containing host and viral DNA. Largest clone % = insertion site with highest number of reads in that sample. PVL = Proviral Load. (Percentage cells carrying a single copy of integrated provirus or number proviral copies per 100 cells).

Example 3: Rationale behind the use of CRISPR-cas9 to cleave circular DNA

It is established practice to linearize plasmids (generally via cutting with a restriction enzyme) prior to their use as template in PCR. It is believed that this avoids supercoiling and thereby increases PCR efficiency ⁶⁷ . Following the same logic, we speculated that linearizing our circularized DNA could also increase PCR efficiency. Figure 8 shows an experiment carried out using 8 μg of DNA from a BLV infected sheep with a proviral load of 82.6%. The DNA was circularized and linear DNA was eliminated (to prevent PCR amplification/recombination involving the remaining linear fragments) using plasmid safe DNase (see Example 1 for a complete description). One quarter of the resultant DNA was subject to CRISPR-cas9 cleavage using the Pool A guides, the second quarter was cleaved using the Pool B guides, the remaining half was kept aside. The linearized DNA was cleaned and used as template in 2x 50mI PCR reactions using the appropriate primer pairs for Pool A (PA) or Pool B (PB). For the uncut DNA half was used as template for 2x 50mI PCR reactions using the PA primers and the other half was used for 2x 50mI PCR reactions using the PB primers. Following 25 PCR cycles, 10mI of each reaction were loaded on a 1% agarose gel. As can be seen in Figure 8, the band intensity for the CRISPR-cas9 cut samples is higher. It should be noted that in lane 3 the PCR smear is shifted down, we generally discard these types of products as the fraction of host-virus fragments is low. (A=unshared genomic DNA, B=genomic DNA sheared to 8kb). Following clean up and elution in ~40 mI of H2O we took an equal volume (3mI) of each library and indexed them via PCR, in a 50 mI reaction volume and using 8 cycles. Again, following clean up, an equal volume of library was pooled and a nanopore library (LSK- 109) was prepared and sequenced on a r9.4 flow cell. Base calling and demultiplexing was carried out as described in Example 1. The results are outlined in Table 9. In addition the coverage of the resultant reads is shown in Figure 9.

Table 9

Table 9 shows that libraries prepared with the CRISPR cut generally produced more raw reads and a much larger fraction of them is composed of the desired chimeric reads containing proviral and host DNA. The CRISPR cut libraries also identified a large number of integration sites. The comparison with an lllumina based library prepared from the same timepoint, using ~4ug of template, shows that PCIP can identify more integration sites. This experiment also shows that only libraries with a size distribution that mirrors that observed in the sheared DNA should be sequenced, libraries with a preponderance of shorter fragments mainly represent nonspecific amplification.

Example 4: Identifying genomic insertions and internal variants in HTLV-1

Adult T-cell leukemia (ATL) is an aggressive cancer induced by HTLV-1. It is generally characterized by the presence of a single dominant malignant clone, identifiable by a unique proviral integration site. We and others have developed methods based on ligation mediated PCR and lllumina sequencing to simultaneously identify integration sites and determine the abundance of the corresponding clones ²¹ . We initially applied PCIP-seq to two HTLV-1 induced cases of ATL, both previously analyzed with our lllumina based method (ATL2 ⁷ & ATL100 ²⁰ ). In ATL100 both methods identify a single dominant clone, with >95% of the reads mapping to a single insertion site on chr18 (Fig. 2a, 2b & Table 8). Using the integration site information, we extracted the PCIP-seq hybrid reads spanning the provirus/host insertion site, uncovering a ~3,600bp deletion within the provirus (Fig. 2c). In the case of ATL2, PCIP-seq showed three major proviruses located on chr5, chr16 and chr1, each responsible for -33% of the HTLV-1/host hybrid reads. We had previously established that these three proviruses are in a single clone via examination of the T-cell receptor gene rearrangement ⁷ . However, it is interesting to note that this was not initially obvious using our lllumina based method as the proviral insertion site on chr1 falls within a repetitive element (LTR) causing many of the reads to map to multiple regions in the genome. If multi mapping reads are filtered out, the chr1 insertion site accounted for 13.7% of the remaining reads, while retaining multi mapping produces values closer to reality (25.4%). In contrast the long reads from PCIP-seq allow unambiguous mapping and closely matched the expected 33% for each insertion site (Fig. 2d), highlighting the advantage long reads have in repetitive regions. Looking at the three proviruses, proviral reads revealed all to be full length. Three de novo mutations were observed in one provirus and a single de novo mutation was identified in the second (Fig. 2e).

Example 5: Insertion sites identified in samples with multiple clones of low abundance

The samples utilized above represent a best-case scenario, with -100% of cells infected and a small number of major clones. We next applied PCIP-seq to four samples from BLV infected sheep (experimental infection ²¹ ) and three cattle (natural infection) to explore its performance on polyclonal and low proviral load (PVL) samples and compared PCIP-seq to our previously published lllumina method ⁷ . PCIP-seq revealed all samples to be highly polyclonal (Figure 10 and Table 8) with the number of unique insertion sites identified varying from 172 in the bovine sample 560 (1 μg template, PVL 0.644%) to 17,903 in bovine sample 1053 (6 μg template, PVL 23.5%). In general, PCIP- seq identified more insertion sites, using less input DNA than our lllumina based method (Table 10).

Table 10. Comparing PCIP-seq to ligation mediated PCR and lllumina sequencing. For the lllumina libraries the template DNA used was 4 μg. For the PCIP-seq it varied between libraries (233=7 μg, 221 (022016)=4 μg, 221(032014)=4 μg, 220=2 μg, 1439=3 μg, 560=1 μg, 1053=6 μg). >3 signifies insertion sites supported by more than 3 reads after PCR duplicate removal. ILLUMINA = Ligation mediated PCR with lllumina sequencing. U-IS ILL. in PCIP = Unique insertion sites (%) identified in ILLUMINA and also found in PCIP-seq. Correlation Abundance Overlapping IS. Pearson’s correlation Abundance = correlation of abundances from proviruses detected in both lllumina and PCIP-seq.

Comparison of the results showed a significant overlap between the two methods. When we consider insertion sites supported by more than three reads in both methods (larger clones, more likely to be present in both samples), in the majority of cases >50% of the insertion sites identified in the lllumina data were also observed via PCIP-seq (Table 10). These results show the utility of PCIP-seq for insertion site identification, especially considering the advantages long reads have in repetitive regions of the genome.

Example 6: Identifying SNPs in BLV proviruses

Portions of the proviruses with more than ten supporting reads (PCR duplicates removed) were examined for SNPs with LoFreq ²² . For the four sheep samples, the variants were called relative to the pBLV344 provirus (used to infect the animals). For the bovine samples 1439 and 1053 custom consensus BLV sequences were generated for each and the variants were called in relation to the appropriate reference (SNPs were not called in 560). Across all the samples 3,209 proviruses were examined, 934 SNPs were called and 680 (21%) of the proviruses carried one or more SNPs (Table 11). Table 11. Numbers of SNPs identified in each sample. We validated 10 BLV SNPs in the ovine samples and 15 in the bovine via clone specific long-range PCR and lllumina sequencing. For Ovine 221, which was sequenced twice over a two-year interval, we identified and validated three instances where the same SNP and provirus were observed at both time points. We noted a small number of positions in the BLV provirus prone to erroneous SNP calls. By comparing allele frequencies from bulk lllumina and Nanopore data these problematic positions could be identified. For example, we observed a number of BLV proviruses in all the samples that had an apparent SNP at position 8213. When we looked at this position in reads mapped to the provirus without first sorting based on insertion site (referred to as bulk) we saw a C called 36 and 38% of the time respectively in the Nanopore data. In the bulk lllumina data, generated from the same sample, we saw the C is called 0% of the time indicating a technical artifact. As a consequence, SNPs from this position were excluded.

Approximately half of the SNPs (47.1% sheep, 51.6% cattle) were found in multiple proviruses. Generally, SNPs found at the same position in multiple proviruses were concentrated in a single individual, indicating their presence in a founder provirus or via a mutation in the very early rounds of viral replication. For example, in animal 233 we found 16 proviruses (provirus inclusion was based on the less stringent criteria of >10 reads covering the position, not filtered for PCR duplicates) carrying a T-to-C transition within the TaxORF at position 8154, this variant does not change the amino acid, lllumina and Nanopore bulk sequencing from the same sample show C is called at a 2% frequency in Nanopore, while with lllumina C is called at a 1% frequency. This indicates that the SNPs observed in these proviruses are not a technical artifact. Alternatively, a variant may also rise in frequency due to increased fitness of clones carrying a mutation in that position. In this instance, we would expect to see the same position mutated in multiple individuals. One potential example is found in the first base of codon 303 (position 8155) of the viral protein Tax, a potent viral transactivator, stimulator of cellular proliferation and highly immunogenic ²³ . A variant was observed at this position in five proviruses for sheep 233 and three for sheep 221 as well as one provirus from bovine 1439 (Fig. 3 a). Using less stringent criteria for the inclusion of a proviral region (>10 reads, not filtered for PCR duplicates) we found 34 proviruses in the ovine and 3 in the bovine carrying a variant in this position. The majority of the variants observed were G- to-A transitions (results in E-to-K amino acid change), however we also observed G-to- T (E-to-STOP) and G-to-C (E-to-Q) transversions. It has been previously shown that the G-to-A mutation abolishes the Tax proteins transactivator activity ²³ · ²⁴ . The repeated selection of variants at this specific position suggests that they reduce viral protein recognition by the immune system, while preserving the Tax proteins other proliferative properties.

Patterns of provirus-wide APOBEC3G ²⁵ induced hypermutation (G-to-A) were not observed in BLV. However, three proviruses (two from sheep 233 and one in bovine 1053) showed seven or more A-to-G transitions, confined to a ~70bp window in the first half of the U3 portion of the 3’LTR. The pattern of mutation, as well as their location in the provirus suggests the action of RNA adenosine deaminases 1 (ADAR1) ²⁶ · ²⁷ .

Example 7: PCIP-seq identifies BLV structural variants in multiple clones

Proviruses were also examined for structural variants (SVs) using a custom script and via visualization in IGV (see Example 1). Between the sheep and bovine samples, we identified 66 deletions and 3 tandem duplications, with sizes ranging from 15bp to 4,152bp, with a median of 113bp (Table 12).

Table 12: BLV structural variants identified via PCIP-seq

221 (022016 & 032014) 221 (032014)

221 (022016) 233

We validated 14 of these via clone specific PCR. As seen in Fig. 3b SVs were found throughout the majority of the provirus, encompassing the highly expressed microRNAs ²⁸ as well as the second exon of the constitutively expressed antisense transcript AS1 ²⁹ . Only two small regions at the 3’ end lacked any SVs. More proviruses will need to be examined to see if this pattern holds, but these results again suggest the importance of the 3’LTR and its previously reported interactions with adjacent host genes ⁷ .

Example 8: Identifying HIV-1 integration sites and the associated provirus Despite the effectiveness of combination antiretroviral therapy (ART) in suppressing HIV- 1 replication, cART is not capable of eliminating latently infected cells, ensuring a viral rebound if cART is suspended ³⁰ . This HIV-1 reservoir represents a major obstacle to a HIV cure ³¹ making its exploration a priority. However, this task is complicated by its elusiveness, with only -0.1% of CD4 ⁺ T cells carrying integrated HIV-1 DNA ³² . To see if PCIP-seq could be applied to these extremely low proviral loads we initially carried out dilution experiments using U1 ³³ , a HIV-1 cell line containing replication competent proviruses ³⁴ . PCIP-seq on undiluted U1 DNA found the major insertion sites on chr2 and chrX (accounting for 47% & 41% of the hybrid reads respectively) and identified the previously reported variants that disrupt Tat function ³⁵ in both proviruses. In the chr2 provirus a T-to-C changes ATG to ACG and the first methionine to a threonine. In the chrX provirus an A-to-T changes CAT to CTT replacing a histidine at position 13 with a leucine. In addition to the two major proviruses we identified an additional -700 low abundance insertion sites (Table 8) including one on chr19 (0.8%) reported by Symons et al 2017 ³⁴ that is actually a product of recombination between the major chrX and chr2 proviruses, and one on chr7 (chr7: 100.5). Identification of the chr7: 100.5 & chr19: 34.9 proviruses as the products of recombination between major chrX and chr2 proviruses was shown by mapping proviral reads from all four proviruses to a full length proviral genome (the sequence (SEQ ID NO:4) of the chrX provirus was used as the reference). This allowed to identify SNPs and sequences derived from respectively, the chr2 and chrX provides. We then serially diluted U1 DNA in Jurkat cell line DNA. PCIP-seq was carried out with 5 μg of template DNA where U1 represents 0.1% and 0.01% of the total DNA. We also processed 5 μg of Jurkat DNA in parallel as a negative control. The three PCIP-seq libraries were prepared using the same guides and primers. Following sequencing and demultiplexing the Jurkat negative control produced 12,137 reads, Jurkat + U1 0.01% produced 234,421 reads and Jurkat + U1 0.1% 252,913 reads. The resultant reads were mapped to the human genome. We were able to detect the major proviruses on chr2 and chrX in both dilutions (Table 8). The reads were also mapped the HIV-1 genome. No reads of pure HIV-1 or chimeric HIV-1/host reads mapping to HIV-1 were observed in the Jurkat negative control (Table 14). In Jurkat + U1 0.01% samples 12.6% of the reads were chimeric HIV-1/host, in Jurkat + U1 0.1% this rose to 43.2%.

Example 9: Identifying full-length and polymorphic endogenous retroviruses in cattle and sheep

ERVs in the genome can be present as full length, complete provirus, or more commonly as solo-LTRs, the products of non-allelic recombination ³⁷ . At the current time conventional short read sequencing, using targeted or whole genome approaches, cannot distinguish between the two classes. Examining full length ERVs would provide a more complete picture of ERV variation, while also revealing which elements can produce de novo ERV insertions. As PCIP-seq targets inside the provirus we can preferentially amplify full length ERVs, opening this type of ERV to study in larger numbers of individuals. As a proof of concept we targeted the class 11 bovine endogenous retrovirus BERVK2, known to be transcribed in the bovine placenta ³⁸ . We applied the technique to three cattle, of which one (10201e6) was a Holstein suffering from cholesterol deficiency, an autosomal recessive genetic defect recently ascribed to the insertion of a 1.3kb LTR in the APOB gene ³⁹ . PCIP-seq clearly identified the APOB ERV insertion in 10201e6, whereas no reads were seen mapping to this position in libraries from the other two cattle (Mannequin & 571). In contrast to previous reports ³⁹ PCIP-seq shows it to be a full-length element. We identified a total of 67 ERVs, with 8 present in all three samples (Table 15). Table 15: Endogenous retroviruses (BERVK2) identified in cattle via PCIP-seq. *LTR matches APOB ERV (BTA11_77.9); #ERV inserted into APOB; Full = Full length ERV; Partial = ERV with large deletion.

We validated three ERVs via long range PCR and lllumina sequencing. We did not find any with an identical sequence to the APOB ERV, although the ERV BTA3_115.3 has an identical LTR sequence, highlighting that the sequence of the LTR cannot be used to infer the complete sequence of the ERV. We also adapted PCIP-seq to amplify the Ovine endogenous retrovirus Jaagsiekte sheep retrovirus (enJSRV), a model for retrovirus-host co-evolution ⁴⁰ . The PCIP-seq reads were mapped to the reference genome (OAR3) where sequences matching enJSRV had been masked out, this preventing reads from multiple proviruses mapping to these positions. Hybrid reads in the unique flanking sequence allowed us to determine the sequence of the proviruses present at these locations. Using two sheep (220 & 221) as template we identified a total of 48 enJSRV proviruses, (33 in 220 and 38 in 221 , with 22 common to both) and of these -54% were full length (Table 16).

Table 16: Endogenous retroviruses (enJSRV) identified in sheep via PCIP-seq. Full =

Full length ERV; Partial = ERV with large deletion.

We validated seven proviruses via long-range PCR and lllumina sequencing.

Example 10: Extending PCIP-seq to human papillomaviruses (HPV)

The majority of HPV infections clear or are suppressed within 1-2 years ⁴¹ , however a minority evolve into cancer, and these are generally associated with integration of the virus into the host genome. This integration into the host genome is not part of the viral lifecycle and the breakpoint in the viral genome can occur at any point across is 8kb circular genome ¹⁶ . As a consequence the part of the viral genome found at the virus host breakpoint varies considerably, making the identifying of integration sites difficult using existing approaches ¹⁶ . The long reads employed by PCIP-seq mean that even when the breakpoint is a number of kb away from the position targeted by primers we should still capture the integration site. As a proof of concept, we applied PCIP-seq to two HPV18 positive cases, (HPV18_PX and HPV18_PY) using 4 μg of DNA extracted from Pap smear material. We identified 55 integration sites in HPV18_PX and 19 integration sites in HPV18_PY (Table 17). Table 17: HPV integration sites identified in patients HPV18_PX and HPV18_PY. Estimated read count refers to number of reads after PCR duplicates have been removed, see https://github.com/GIGA-AnimalGenomics-

BLV/PCI P/blob/master/READM E. md

In HPV18_PY the vast majority of the reads only contained HPV sequences, the integration sites identified were defined by single reads, suggesting little or no clonal expansion (Table 8). In HPV18_PX most integration sites were again defined by a single read, however there were some exceptions (Table 17). HPV18_PX had integrated copies of HPV18 on chr21 and chr3 (Figure 5). Both integration sites contained multiple copies of the HPV genome. The most striking of these was a cluster of what appeared to be three integration sites located within the region chr3:52477576-52564190 (Fig. 5a). The unusual pattern of read coverage combined with the close proximity of the virus-host breakpoints indicated that these three integration sites were connected. Long range PCR with primers spanning positions a-b and a-g, showed that a genomic rearrangement had occurred in this clonally expanded cell (Fig. 5a). Regions a and b are adjacent to one another with HPV integrated between, however PCR also showed regions a and g to be adjacent to one another, again with the HPV genome integrated between (Fig. 5b). The sequence of the virus found between a-b looks to be derived from the a-g virus as it shares a breakpoint and is slightly shorter (Fig. 5b). This complex arrangement suggests that this rearrangement was generated via the recently described ‘looping’ integration mechanism ¹⁶ · ⁴² . The a and b breakpoints fall within exons of the NISCH gene while the Y breakpoint falls within exon 27 of PBRM1 (Fig. 5c), a gene previously shown to be a cancer driver in renal carcinoma ⁴³ and intrahepatic cholangiocarcinomas ⁴⁴ . This patient was classified by histology as having atypical glandular cells and a follow up three months later was classified as a high grade CIN3.The PCIP-seq method was applied to DNA from leftover Pap smears, assaying 29 HPV18 and 42 HPV16 positive cases. The majority of the samples had been classified by cytology as Atypical squamous cells of undetermined significance (ASC-US). In both, episomal HPV was the most common finding. We found that the reads generated from episomal HPV can be used to generate a consensus sequence for HPV and as shown in Figure 6 it is possible to examine the phylogenetic relationships between the isolates.

As regards HPV integrations, we identified six patients where integration is associated with a pronounced clonal expansion, four, including HPV18_PX, were infected with HPV18 and two with HPV16.

The second patient had an integration of HPV18 within an intron of LRRC49 (histology = low grade squamous intraepithelial lesion). From the next two clonally expanded integrations (both HPV18), samples from two time points were available. The first had an integration in the LAPTM4B gene, the integration was found in both samplings and in the second it appears that episomal HPV18 has been cleared (Figure 7a, b). (Histology, 1st sample = atypical squamous cells cannot exclude HSIL, 2nd sampling upgraded to High Grade Squamous Intraepithelial Lesion, HSIL).

The last clonally expanded integrations were found in a seventy-one-year-old patient, integration was observed in three different positions in the genome, all were observed in two samplings 5 months apart (Figure 7c, d) (Both time points, histology = atypical squamous cell of undetermined significance). All the integrated copies of HPV18 had intact E6 and E7 genes (both are cancer driver genes and are deregulated when HPV integrates). As regards HPV16, we identified two samples with clonally expanded integrations. The first was observed in a 53-year-old with a low-grade squamous intraepithelial lesion, the HPV16 genome had integrated ~2.5kb upstream of the KRT5 gene. No episomal HPV16 DNA was observed in this sample. The integrated HPV genome contains a ~3kb deletion that does not overlap with the E6 and E7 genes. The second HPV16 sample has an integration in intron 4 of the POFUT1 gene. Again, the inserted viral genome contains a large deletion (~5.5kb) that does not overlap with E6 and E7. In contrast to the other HPV16 sample the majority (-75%) of the HPV16 reads in this patient were still derived from episomal HPV16.

Discussion

In the present report we describe how PCIP-seq can be utilized to identify insertion sites while also sequencing parts of, and in some cases the entire associated provirus, and confirm this methodology is effective with a number of different retroviruses as well as HPV. For insertion site identification, the method was capable of identifying more than ten thousand BLV insertion sites in a single sample, using ~4 μg of template DNA. Even in samples with a PVL of 0.66%, it was possible to identify hundreds of insertion sites with only 1 μg of DNA as template. The improved performance of PCIP-seq in repetitive regions further highlights its utility, strictly from the standpoint of insertion site identification. In addition to its application in research, high throughput sequencing of retrovirus insertion sites has shown promise as a clinical tool to monitor ATL progression ²⁰ . Illumina based techniques require access to a number of capital-intensive instruments. In contrast PCIP-seq libraries can be generated, sequenced and analyzed with the basics found in most molecular biology labs, moreover, preliminary results are available just minutes after sequencing begins ⁴⁵ . As a consequence, the method may have use in a clinical context to track clonal evolutions in HTLV-1 infected individuals, especially as the majority of HTLV-1 infected individuals live in regions of the world with poor biomedical infrastructure ⁴⁶ .

One of the common issues raised regarding Oxford Nanopore data is read accuracy. Early versions of the MinlON had read identities of less than 60% ⁴⁷ , however the development of new pores and base calling algorithms make read identities of -90% achievable ⁴⁸ . Accuracy can be further improved by generating a consensus from multiple reads, making accuracies of -99.4% ⁴⁸ possible. Recently Greig et al ⁴⁹ compared the performance of Illumina and Oxford Nanopore technologies for SNP identification in two isolates of Escherichia coii. They found that after accounting for variants observed at 5- methylcytosine motif sequences only ~7 discrepancies remained between the platforms. It should be noted that as PCIP-seq sequences PCR amplified DNA, errors generated by base modifications will be avoided. Despite these improvements in accuracy, Nanopore specific errors can be an issue at some positions. Comparison with lllumina data is helpful in the identification of problematic regions and custom base calling models may be a way to improve accuracy in such regions ⁴⁸ . Additionally, PCIP-seq libraries could equally be sequenced using long reads on the Pacific Biosciences platform or via 10X Genomics linked reads on lllumina if high single molecule accuracy is required ¹⁷ . In the current study we focused on SNPs observed in clonally expanded BLV proviruses. For viruses such as HIV-1, which have much lower proviral loads, more caution will be requited as the majority of proviral sequences will be generated from single provirus, making errors introduced by PCR more of an issue.

When analyzing SNPs from BLV the most striking result was the presence of the recurrent mutations at the first base of codon 303 in the viral protein Tax, a central player in the biology of both HTLV-1 ⁴⁶ and BLV ⁵⁰ . It has previously been reported that this mutation causes an E-to-K amino acid substitution which ablates the transactivator activity of the Tax protein ²³ . Collectively, these observations suggest this mutation confers an advantage to clones carrying it, possibly contributing to immune evasion, while retaining Tax protein functions that contribute to clonal expansion. However, there is a cost to the virus as this mutation prevents infection of new cells due to the loss of Tax mediated transactivation of the proviral 5’LTR making it an evolutionary dead end. It will be interesting to see if PCIP-seq can provide a tool to identify other examples of variants that increase the fitness of the provirus in the context of an infected individual but hinder viral spread to new hosts. Additionally, the technique could be used to explore the demographic features of the proviral population within and between hosts, how these populations evolve over time and how they vary.

A second notable observation is the cluster of A-to-G transitions observed within a ~70bp window in the 3’LTR. Similar patterns have been ascribed to ADAR1 hypermutation in a number of viruses ²⁶ , including the close BLV relatives HTLV-2 and simian T-cell leukemia virus type 3 (STLV-3) ⁵¹ . Given the small number of hypermutated proviruses observed, it appears to be a minor source of variation in BLV, although it will be interesting to see it this holds for different retroviruses and at different time points during infection. In the current study we focused our analysis on retroviruses and ERVs. However, as this methodology is potentially applicable to a number of different targets we extended its use to HPV as a proof of concept. It is estimated that HPV is responsible for >95% of cervical carcinoma and -70% of oropharyngeal carcinoma ⁵² . While infection with a high- risk HPV strain (HPV16 & HPV18) is generally necessary for the development of cervical cancer, it is not sufficient and the majority of infections resolve without adverse consequences ⁴¹ . The use of next-generation sequencing has highlighted the central role HPV integration plays in driving the development of cervical cancer ¹⁶ . Our results show that PCIP-seq can be applied to identify HPV integration sites in early precancerous samples. This opens up the possibility of generating a more detailed map of HPV integrations as well as potentially providing a biomarker to identify HPV integrations on the road to cervical cancer.

Other potential applications include determining the insertion sites and integrity of retroviral vectors ⁵⁴ and detecting transgenes in genetically modified organisms. We envision that in addition to the potential applications outlined above many other novel targets/questions could be addressed using this method.

References

1. Bushman, F. et al. Genome-wide analysis of retroviral DNA integration. Nat Rev Micro 3, 848-858 (2005).

2. Gillet, N. A. et al. The host genomic environment of the provirus determines the abundance of HTLV-1 -infected T-cell clones. B/ood 117, 3113-3122 (2011).

3. Maldarelli, F. et al. Specific HIV integration sites are linked to clonal expansion and persistence of infected cells. Science (2014). doi: 10.1126/science.1254194

4. Wagner, T. A. et al. HIV latency. Proliferation of cells with HIV integrated into cancer genes contributes to persistent infection. Science 345, 570-573 (2014).

5. Bruner, K. M. et al. A quantitative approach for measuring the reservoir of latent HIV-1 proviruses. Nature 566, 1-19 (2019).

6. Einkauf, K. B. et al. Intact HIV-1 proviruses accumulate at distinct chromosomal positions during prolonged antiretroviral therapy. J. Clin. Invest. 129, 988-998 (2019). 7. Rosewick, N. et al. Cis-perturbation of cancer drivers by the HTLV-1/BLV proviruses is an early determinant of leukemogenesis. Nature Communications 8, 15264 (2017).

8. Malhotra, S. et al. Selection for avian leukosis virus integration sites determines the clonal progression of B-cell lymphomas. PLoS Pathog 13, e1006708-25 (2017).

9. Simonetti, F. R. etal. Clonally expanded CD4 +T cells can produce infectious HIV- 1 in vivo. Proceedings of the National Academy of Sciences 113, 1883-1888 (2016).

10. Miyazaki, M. et al. Preferential selection of human T-cell leukemia virus type 1 provirus lacking the 5' long terminal repeat during oncogenesis. Journal of Virology 81, 5714-5723 (2007).

11. Hiener, B. et al. Identification of Genetically Intact HIV-1 Proviruses in Specific CD4+ T Cells from Effectively Treated Participants. CellReports 21, 813-822 (2017).

12. Lander, E. S. etal. Initial sequencing and analysis of the human genome. Nature 409, 860-921 (2001).

13. Rivas-Carrillo, S. D., Pettersson, M. E., Rubin, C.-J. & Jern, P. Whole-genome comparison of endogenous retrovirus segregation across wild and domestic host species populations. PNAS 115, 11012-11017 (2018).

14. Pett, M. & Coleman, N. Integration of high-risk human papillomavirus: a key event in cervical carcinogenesis? The Journal of Pathology 212, 356-367 (2007).

15. Hu, Z. et al. Genome-wide profiling of HPV integration in cervical cancer identifies clustered genomic hot spots and a potential microhomology-mediated integration mechanism. Nat Genet 47, 158-163 (2015).

16. Groves, I. J. & Coleman, N. Human papillomavirus genome integration in squamous carcinogenesis: what have next-generation sequencing studies taught us? The Journal of Pathology 245, 9-18 (2018).

17. Sedlazeck, F. J., Lee, H., Darby, C. A. & Schatz, M. C. Piercing the dark matter: bioinformatics of long-range sequencing and mapping. Nature Reviews Genetics 17, 1-18 (2018). 18. Pradhan, B. etal. Detection of subclonal L1 transductions in colorectal cancer by long-distance inverse-PCR and Nanopore sequencing. Scientific Reports 7, 1-12 (2017).

19. Lober, U. et ai. Degradation and remobilization of endogenous retroviruses by recombination during the earliest stages of a germ-line invasion. Proceedings of the National Academy of Sciences 5, 201807598-15 (2018).

20. Artesi, M. et ai. Monitoring molecular response in adult T-cell leukemia by high- throughput sequencing analysis of HTLV-1 clonality. Leukemia 31, 2532-2535 (2017).

21. Willems, L. et ai. In vivo infection of sheep by bovine leukemia virus mutants. Journal of Virology 67, 4078-4085 (1993).

22. Wilm, A. et ai. LoFreq: a sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets. Nucleic Acids Research 40, 11189-11201 (2012).

23. Van den Broeke, A. et ai. In vivo rescue of a silent tax-deficient bovine leukemia virus from a tumor-derived ovine B-cell line by recombination with a retrovirally transduced wild-type tax gene. Journal of Virology 73, 1054-1065 (1999).

24. Merimi, M. etal. Complete suppression of viral gene expression is associated with the onset and progression of lymphoid malignancy: observations in Bovine Leukemia Virus-infected sheep. Retrovirology 4, 51 (2007).

25. Armitage, A. E. et ai. APOBEC3G-lnduced Hypermutation of Human Immunodeficiency Virus Type-1 Is Typically a Discrete ‘All or Nothing’ Phenomenon. PLoS Genet 8, e1002550-12 (2012).

26. Samuel, C. E. Adenosine deaminases acting on RNA (ADARs) are both antiviral and proviral. Virology 411, 180-193 (2011).

27. Cachat, A. et ai. ADAR1 enhances HTLV-1 and HTLV-2 replication through inhibition of PKR activity. Retrovirology 11 , 7415-15 (2014).

28. Rosewick, N. et ai. Deep sequencing reveals abundant noncanonical retroviral microRNAs in B-cell leukemia/lymphoma. Proceedings of the National Academy of Sciences 110, 2306-2311 (2013). 29. Durkin, K. et al. Characterization of novel Bovine Leukemia Virus (BLV) antisense transcripts by deep sequencing reveals constitutive expression in tumors and transcriptional interaction with viral microRNAs. Retrovirology 13, 1-16 (2016).

30. Finzi, D. etal. Latent infection of CD4+ T cells provides a mechanism for lifelong persistence of HIV-1, even in patients on effective combination therapy. Nat Med 5, 512-517 (1999).

31. Anderson, E. M. & Maldarelli, F. The role of integration and clonal expansion in HIV infection: live long and prosper. Retrovirology 15, 1-22 (2018).

32. Kiselinova, M. et al. Integrated and Total HIV-1 DNA Predict Ex Vivo Viral Outgrowth. PLoS Pathog 12, e1005472-17 (2016).

33. Folks, T. M., Justement, J., Kinter, A., Dinarello, C. A. & Fauci, A. S. Cytokine- induced expression of HIV-1 in a chronically infected promonocyte cell line. Science 238, 800-802 (1987).

34. Symons, J. et al. HIV integration sites in latently infected cell lines: evidence of ongoing replication. Retrovirology 14, 1-11 (2017).

35. Emiliani, S. et al. Mutations in the tat Gene Are Responsible for Human Immunodeficiency Virus Type 1 Postintegration Latency in the U1 Cell Line. Journal of Virology 72, 1666-1670 (1998).

37. Hughes, J. F. & Coffin, J. M. Human endogenous retrovirus K solo-LTR formation and insertional polymorphisms: implications for human and viral evolution. Proceedings of the National Academy of Sciences 101, 1668-1672 (2004).

38. Cornells, G. etal. Captured retroviral envelope syncytin gene associated with the unique placental structure of higher ruminants. Proceedings of the National Academy of Sciences 110, E828-E837 (2013).

39. Menzi, F. et al. A transposable element insertion in APOB causes cholesterol deficiency in Holstein cattle. Animal Genetics 47, 253-257 (2016).

40. Arnaud, F. et al. A Paradigm for Virus-Host Coevolution: Sequential Counter- Adaptations between Endogenous and Exogenous Retroviruses. PLoS Pathog 3, e 170-14 (2007).

41. Schiffman, M., Castle, P. E., Jeronimo, J., Rodriguez, A. C. & Wacholder, S. Human papillomavirus and cervical cancer. The Lancet 370, 890-907 (2007). 42. Akagi, K. et al. Genome-wide analysis of HPV integration in human cancers reveals recurrent, focal genomic instability. Genome Research 24, 185-199 (2014).

43. Varela, I. et al. Exome sequencing identifies frequent mutation of the SWI/SNF complex gene PBRM1 in renal carcinoma. Nature 469, 1-5 (2011).

44. Jiao, Y. et al. Exome sequencing identifies frequent inactivating mutations in BAP1, ARID1A and PBRM1 in intrahepatic cholangiocarcinomas. Nat Genet 45, 1470-1473 (2013).

45. Quick, J. et al. Real-time, portable genome sequencing for Ebola surveillance. Nature 530, 228-232 (2016).

46. Bangham, C. R. M. Human T Cell Leukemia Virus Type 1: Persistence and Pathogenesis. Annu. Rev. Immunol. 36, annurev-immunol-042617-053222-29 (2017).

47. Goodwin, S. et al. Oxford Nanopore sequencing, hybrid error correction, and de novo assembly of a eukaryotic genome. Genome Research 25, 1750-1756 (2015).

48. Wick, R. Performance of neural network basecalling tools for Oxford Nanopore sequencing. Genome Biol 20, 1-10 (2019).

49. Greig, D. R., Jenkins, C., Gharbia, S. & Dallman, T. J. Comparison of singlenucleotide variants identified by lllumina and Oxford Nanopore technologies in the context of a potential outbreak of Shiga toxin-producing Escherichia coli. GigaScience 8, 822-12 (2019).

50. Gillet, N. et al. Mechanisms of leukemogenesis induced by bovine leukemia virus: prospects for novel anti-retroviral therapies in human. Retrovirology 4, 18 (2007).

51. Ko, N. L., Birlouez, E., Wain-Hobson, S., Mahieux, R. & Vartanian, J. P. Hyperediting of human T-cell leukemia virus type 2 and simian T-cell leukemia virus type 3 by the dsRNA adenosine deaminase ADAR-1. Journal of General Virology 93, 2646-2651 (2012).

52. Schiffman, M. etal. Carcinogenic human papillomavirus infection. Nature reviews Disease primers 2, 16086 (2016). 54. Goodwin, L. O. et al. Large-scale discovery of mouse transgenic integration sites reveals frequent structural variation and insertional mutagenesis. Genome Research 29, gr.233866.117-505 (2019).

55. Li, H. Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics 34, 3094-3100 (2018).

56. Rutsaert, S., De Spiegelaere, W., De Clercq, L. & Vandekerckhove, L. Evaluation of HIV-1 reservoir levels as possible markers for virological failure during boosted darunavir monotherapy. Journal of Antimicrobial Chemotherapy (2019).

57. Trypsteen, W. et al. ddpcRquant: threshold determination for single channel droplet digital PCR experiments. Analytical and bioanalytical chemistry 407, 5827-5834 (2015).

58. Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078-2079 (2009).

59. Killick, R., Fearnhead, P. & Eckley, I. A. Optimal detection of changepoints with a linear computational cost. Journal of the American Statistical Association 107, 1590-1598 (2012).

60. Thorvaldsdottir, H., Robinson, J. T. & Mesirov, J. P. Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Briefings in Bioinformatics 14, 178-192 (2013).

61. Camacho, C. et al. BLAST+: architecture and applications. BMC Bioinformatics 10, 421 (2009).

62. Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841-842 (2010).

63. Loman, N. J., Quick, J. & Simpson, J. T. A complete bacterial genome assembled de novo using only nanopore sequencing data. Nature Methods 12, 733-735 (2015).

64. Tjalma WAA, Kim E, Vandeweyer K. The impact on women's health and the cervical cancer screening budget of primary HPV screening with dual-stain cytology triage in Belgium. Eur J Obstet Gynecol Reprod Biol. 2017;212: 171-181. Tjalma W, Brasseur C, Top G, Ribesse N, Morales I, Van Damme PA. HPV vaccination coverage in the federal state of Belgium according to regions and their impact. Facts Views Vis Obgyn. 2018;10: 101-105. Mirabello L, Yeager M, Yu K, Clifford GM, Xiao Y, Bin Zhu, et al. HPV16 E7 Genetic Conservation Is Critical to Carcinogenesis. Cell. Elsevier Inc; 2017;170: 1164-

1172. e6. doi:10.1016/j. cell.2017.08.001 Chen, J., Kadlubar, F. F. & Chen, J. Z. DNA supercoiling suppresses real-time PCR: a new approach to the quantification of mitochondrial DNA damage and repair. Nucleic Acids Res 35, 1377-1388 (2007).