Purpose: Toca 511 is a gammaretroviral replicating vector encoding cytosine deaminase that selectively infects tumor cells and converts the antifungal drug 5-fluorocytosine into the antineoplastic drug 5-fluorouracil, which directly kills tumor cells and stimulates antitumor immune responses. As part of clinical monitoring of phase I clinical trials in recurrent high-grade glioma, we have performed extensive molecular analyses of patient specimens to track vector fate.

Patients and Methods: Toca 511 and Toca FC (extended-release 5-fluorocytosine) have been administered to 127 high-grade glioma patients across three phase I studies. We measured Toca 511 RNA and DNA levels in available body fluids and tumor samples from patients to assess tumor specificity. We mapped Toca 511 integration sites and sequenced integrated Toca 511 genomes from patient samples with detectable virus. We measured Toca 511 levels in a diverse set of tissue samples from one patient.

Results: Integrated Toca 511 is commonly detected in tumor samples and is only transiently detected in blood in a small fraction of patients. There was no believable evidence for clonal expansion of cells with integrated Toca 511 DNA, or preferential retrieval of integration sites near oncogenes. Toca 511 sequence profiles suggest most mutations are caused by APOBEC cytidine deaminases acting during reverse transcription. Tissue samples from a single whole-body autopsy affirm Toca 511 tumor selectivity.

Conclusions: Toca 511 and Toca FC treatment was not associated with inappropriate integration sites and clonal expansion. The vector is tumor-selective and persistent in patients who received Toca 511 injections. Clin Cancer Res; 24(19); 4680–93. ©2018 AACR.

Translational Relevance

Retroviral replicating vectors (RRVs) are potential therapies for a wide range of oncologic malignancies, with ongoing durable responses and multiyear survival outcomes for some recurrent high-grade glioma patients in a phase I trial. Detailed characterization including viral presence in biological fluids and cancer cells, viral insertion site location and clonality, as well as viral sequence stability are important to support future approval of biologic therapies built on RRVs. Previous treatment-related adverse events, including lymphomas, in some clinical trials with distinct nonreplicating retroviral vectors were preceded by clonal expansion of infected cells due to integration near a proto-oncogene. As part of our ongoing development of RRV Toca 511 in recurrent high-grade glioma, we performed extensive monitoring of the virus in patient tumors and body fluids, including Toca 511 quantitation, integration site identification, and mutation profile characterization. Our results support the favorable safety profile of Toca 511 and expansion of clinical trials into other indications. Moreover, this work provides a broad framework for molecular monitoring of RRV therapies during clinical development.

Given the limits of conventional cancer treatments, retroviral replicating vectors (RRVs) have emerged as a potential backbone for useful therapies across a wide range of oncologic malignancies. RRVs selectively infect tumor cells without directly lysing them. This differentiates them from directly oncolytic and highly inflammatory viruses such as adenovirus and herpes viruses (1–3). Thus, RRVs provide a platform for therapies based on tumor-specific gene delivery strategies without the inherent limitation of rapidly killing infected cells. RRV are selective for tumor cells partially due to virus-selective advantages in the tumor microenvironment from blunted innate immune responses as well as suppressed adaptive immune responses relative to normal dividing cells (3–6). Viral dependency on mitosis for integration contributes to cancer cell selectivity (7), and the noninflammatory nature of the infection (and replication competency) allows subsequent spread (8).

Toca 511 (vocimagene amiretrorepvec) is a RRV derived from a murine gammaretrovirus (gRV), with an amphotropic envelope to allow infection of human cells. Such viruses are known to be nonpathogenic to humans (9), and all adult humans so far tested appear to be seropositive (10, 11). The modified Toca 511 virus encodes a transgene for an optimized yeast cytosine deaminase (yCD2) that converts the orally available prodrug 5-fluorocytosine (5-FC) into cytotoxic 5-fluorouracil (5-FU; refs. 12–14).

Three phase I clinical trials, NCT01156584 (Study 8), NCT01470794 (Study 11) and NCT01985256 (Study 13), evaluated safety of Toca 511 and Toca FC (an extended-release formulation of 5-FC) therapy in patients with recurrent high-grade glioma (HGG) by different routes of Toca 511 administration, followed by oral Toca FC dosing. In Study 8, Toca 511 was administered by injection into the tumor, without resection. In Study 11, tumor resection was performed, followed by multiple injections of Toca 511 into the tissue walls of the resection cavity (9). In Study 13, intravenous administration of Toca 511 (bolus injections for one, three or five consecutive days) was followed by surgical resection plus additional injections of Toca 511 into the tissues of the resection cavity 8 to 14 days later (clinical results from Study 13 will be presented elsewhere). For all trials, 4 to 6 weeks after the final Toca 511 administration, patients were treated with Toca FC for approximately one week, in repeated cycles, every 4 to 8 weeks, for 6 months, or until radiologic evidence of tumor progression, clinical progression, or termination by treating physicians (Fig. 1A).

Figure 1.

Toca 511 detection frequency and timing in DNA isolated from patient blood and tumor samples. A, Timeline of resection Study 11 and IV Study 13 treatment regimes. The timelines start at day 1 and are not to scale. The gray boxes above the timelines show when Toca FC was taken. The vertical red lines below the timelines indicate regularly scheduled blood draws for Toca 511 monitoring. B, Heatmap representation of Toca 511 DNA signal in blood as a function of time for the eleven patients of 56 in Study 11 with detectable signal. Visit number is shown above the heatmap. Visit 3a is 1 week after visit 3. Visit 4a is 1 week after visit 4. C, Barplot showing the percentage of patients with detectable Toca 511 signal in DNA: blood = red, reresected tumor after Toca 511/FC treatment = magenta, resected tumor after intravenous treatment = blue. The numbers on top of the bars show the fraction of patients with detectable signal. D, Boxplot showing the estimated number of copies of Toca 511 per μg of genomic DNA for samples with detectable signal in C. The horizontal bar shows the median, the lower and upper hinges correspond to the first and third quartiles respectively, and the whiskers extend from the hinge to the largest value no further than 1.5× the interquartile range from the hinges. There are 15 samples from 11 patients with quantifiable signal in blood (B).

Figure 1.

Toca 511 detection frequency and timing in DNA isolated from patient blood and tumor samples. A, Timeline of resection Study 11 and IV Study 13 treatment regimes. The timelines start at day 1 and are not to scale. The gray boxes above the timelines show when Toca FC was taken. The vertical red lines below the timelines indicate regularly scheduled blood draws for Toca 511 monitoring. B, Heatmap representation of Toca 511 DNA signal in blood as a function of time for the eleven patients of 56 in Study 11 with detectable signal. Visit number is shown above the heatmap. Visit 3a is 1 week after visit 3. Visit 4a is 1 week after visit 4. C, Barplot showing the percentage of patients with detectable Toca 511 signal in DNA: blood = red, reresected tumor after Toca 511/FC treatment = magenta, resected tumor after intravenous treatment = blue. The numbers on top of the bars show the fraction of patients with detectable signal. D, Boxplot showing the estimated number of copies of Toca 511 per μg of genomic DNA for samples with detectable signal in C. The horizontal bar shows the median, the lower and upper hinges correspond to the first and third quartiles respectively, and the whiskers extend from the hinge to the largest value no further than 1.5× the interquartile range from the hinges. There are 15 samples from 11 patients with quantifiable signal in blood (B).

Close modal

Historical safety concerns are associated with different retroviral nonreplicative vectors (RNV). For instance, RNV preparations grossly contaminated with replicative murine leukemia virus (MLV) derived from recombination during serial passage through ecotropic and amphotropic murine cell–derived packaging cell lines were employed for ex vivo transduction of hematopoietic stem cells (HSC) followed by autologous transplant into monkeys (15, 16). Three of 10 monkeys so treated, who were unable to develop antiviral immune responses, developed lymphomas containing multiple copies of an array of MLV related sequences in the tumor genomes. A second source of concern was the observation, following transplant of autologous cytokine-stimulated HSCs transduced ex vivo with RNV, of delayed occurrence of lymphoma in some clinical trial patients several years after treatment (17–19). T-cell lymphoma was linked to treatment of X-linked Severe Combined Immune Deficiency Syndrome (X-SCIDS) with the common gamma chain interleukin receptor and Wisckott–Aldrich Syndrome (WAS) gene replacement, which in turn were linked to insertions near the promoter for LMO2, leading to its ectopic expression. Myelodysplastic syndrome was linked to treatment of chronic granulomatous disease and insertion and activation of MDS-EVI-1 oncogene (19, 20). However, a therapy based on RNV transduction and reimplantation of autologous HSCs for ADA deficient SCIDS has not shown such side effects, and was recently approved for sale in Europe (Strimvelis, GSK Ltd.; ref. 18). In addition, long-term outcomes with persistent RNV transduced T cells have also failed to show such side effects (21). These observations suggest that such adverse outcomes are influenced by factors including multiplicity of infection, nature of the transgene, indication, clinical condition, and age of the recipient, expected mechanism of action of the transgene, and target tissue (17, 22).

Several methodologies have been developed to track the fate of viral vectors in patients. Viral RNA and DNA levels can be measured over a broad dynamic range with qPCR. Viral integration sites, originally identified by Sanger sequencing of cloned PCR products (23), can be systematically mapped via next-generation sequencing (24–26).

While previous gRV trials used RNVs, often transducing autologous cells ex vivo, Toca 511 is a RRV used to infect cancer cells in vivo. Thus, host immunity to viral proteins and virus–host restriction factor interactions may influence Toca 511 spread and activity (27). For instance, APOBEC proteins target a number of retroviruses in humans, causing G to A hypermutation via processive cytidine deamination during reverse transcription (28, 29). The influence of APOBEC proteins on Toca 511 in patient samples can be measured via next-generation sequencing. In addition, RNVs (unlike RRVs) usually lack genes to produce viral proteins that could stimulate immune responses against infected cells. Thus, Toca 511 therapy requires an overlapping but distinct set of assays for molecular monitoring than those used in RNV trials.

In this study, we report results of Toca 511 monitoring in blood and tumor samples from patients who received Toca 511 through direct injection into the resection cavity (Study 11) and intravenous delivery (Study 13) 8 to 14 days prior to tumor removal (additional Toca 511 is delivered via direct injection into the resection cavity), as well as analyses of tissue from a full autopsy of a patient that participated in a phase III clinical study (NCT02414165) and received Toca 511 injection into the HGG resection cavity. Our monitoring methods include Toca 511 quantification, identification of genome integration sites, and mutation profiles of Toca 511 genomes from available Toca 511–positive tumor tissue and blood cells. This work provides molecular and genomic corollaries to traditional safety assessments and presents a framework for clinical monitoring of RRV therapies during initial development, allowing for continued confidence in using such vectors in patients.

Toca 511 quantification

DNA and RNA purification.

DNA and RNA were extracted from the same source (tumor or whole blood) using a modified version of Promega's Maxwell 16 Purification System as communicated by the manufacturer. Tumor sections were minced in petri dishes on dry ice and transferred into 1.5 mL microfuge tubes. A total of 500 μL of homogenization/1-thioglycerol solution were added into each tube and the test articles were agitated with pestles (VWR: catalog no. 47747-366) followed by vortexing at full speed for 30 seconds. For whole-blood DNA and RNA purification, up to 200 μL of whole blood (brought to 200 μL with PBS if necessary) was transferred into 1.5-mL microfuge tubes. A total of 300 μL of homogenization/1-thioglycerol solution was added to the tube and vortexed at full speed for 30 seconds. In each case, 300 μL of the homogenate was transferred into Maxwell 16 DNA Purification cartridges and subjected to automated DNA purification. Purified DNA was eluted in 300 μL of nuclease-free water. For RNA purification, 200 μL of lysis buffer was added into the 1.5-mL microfuge tubes containing the remaining 200 μL of the original 500-μL homogenate. Test articles were vortexed for 30 seconds and the full volume was transferred into Promega's Maxwell 16 simplyRNA Purification cartridges. Automated RNA purification was carried out and purified RNA was eluted in 50 μL of nuclease-free water. RNA purification from plasma, urine, and saliva and IHC of CD gene were described previously (4).

qPCR.

TaqMan probe-based qPCR was performed as single-targeted 20 μL reactions for Study 11 test articles and as multiplex reactions for Study 13 test articles. The reactions were prepared in triplicate, on a CFX96 or CFX384 Real-Time System (Bio-Rad) using primers and probes annealing to sequences in the long terminal repeat region of Toca 511 or in the yCD2 transgene (monoplex) or in sequences in the amphotropic MLV4070A ENV, POL, and yCD2 transgene (multiplex). Primers and probes were designed with PrimerQuest software and synthesized by Integrated DNA Technologies. For qPCR detection of gRV-specific sequences, primers were used at final concentration of 300 nmol/L each of MLV-F (5′- AGC CCA CAA CCC CTC ACT C-3′) and MLV-R (5′- TCT CCC GAT CCC GGA CGA-3′), and 100 nmol/L of MLV hydrolysis probe (5′- FAM-CCC CAA ATG AAA GAC CCC CGC TGA CG-3′BHQ_1) with iQ PCR Supermix (Bio-Rad). For qPCR detection of yCD2, primers were used at 600 nmol/L each of yCD2-F (5′-ATC ATC ATG TAC GGC ATC CCT AG-3′) and yCD2-R (5′-TGA ACT GCT TCA TCA GCT TCT TAC-3′), and 100 nmol/L of yCD2 hydrolysis probe (5′-FAM/TCA TCG TCA ACA ACC ACC ACC TCG T/3′BHQ_1). For simultaneous qPCR detection of POL, ENV, and yCD2, the following concentrations were used: 300 nmol/L each of Pol2-F (5′-CAA GGG GCT ACT GGA GGA AAG-3′) and Pol2-R (5′-CAG TCT GGT ACA TGG AGG AAA G-3′); 100 nmol/L of Pol2 hydrolysis probe (5′-HEX/TAT CGC TGG ACC ACG GAT CGC AA/3′BHQ_1); 300 nmol/L each of Pol3-F (5′-CGA CAC CAG ACT AAG AAC CTA G-3′) and Pol3-R (5′-CGA TGC CGT CTA CTT TGA GG-3′); 100 nmol/L of Pol3 hydrolysis probe (5′-HEX/CCT CGC TGG AAA GGA CCT TAC ACA/3′IABkFQ); 300 nmol/L each of Env2-F (5′-ACC CTC AAC CTC CCC TAC AAG T-3′) and Env2-R (5′-GTT AAG CGC CTG ATA GGC TC-3′); 100 nmol/L of Env2 hydrolysis probe (5′-TEX615/AGC CAC CCC CAG GAA CTG GAG ATA GA/3′BHQ_2); 300 nmol/L each of yCD2-F (5′-ATC ATC ATG TAC GGC ATC CCT AG-3′) and yCD2-R (5′-TGA ACT GCT TCA TCA GCT TCT TAC-3′); and 100 nmol/L of yCD2 hydrolysis probe (5′-FAM/TCA TCG TCA ACA ACC ACC ACC TCG T/3′BHQ_1). Either the Pol2 or Pol3 primer/probe set was used as a component in the triplex reaction.

Thermal cycling conditions consisted of 95°C for 5 minutes, followed by three cycles of 95°C for 15 seconds and 65°C for 10 seconds, followed by 38 cycles of 95°C for 15 seconds and 65°C for 30 seconds. CFX Manager 3.0 software (Bio-Rad) was used to calculate threshold cycle (Ct) values. Technical replicates were averaged and absolute quantification was determined from linear regression using a six-log serial dilution standard curve (25 to 2.5e6 copies/reaction) from a Toca 511–containing plasmid, pAZ3-yCD2.

qRT-PCR.

TaqMan probe-based qRT-PCR was performed as single-targeted 20-μL reactions for Study 11 test articles and as multiplex reactions for Study 13 test articles. The reactions were prepared in triplicate using primers and probes annealing to sequences in POL (monoplex) or to sequences in the amphotropic MLV4070A ENV, POL, and yCD2 (multiplex).

For simultaneous qRT-PCR detection of POL, ENV, and yCD2 in a single reaction, primers were used at 300 nmol/L each of Pol2-F (5′-CAA GGG GCT ACT GGA GGA AAG-3′) and Pol2-R (5′-CAG TCT GGT ACA TGG AGG AAA G-3′); 100 nmol/L of Pol2 hydrolysis probe (5′-HEX/TAT CGC TGG ACC ACG GAT CGC AA/3′BHQ_1); 300 nmol/L each of Pol3-F (5′-CGA CAC CAG ACT AAG AAC CTA G-3′) and Pol3-R (5′-CGA TGC CGT CTA CTT TGA GG-3′); 100 nmol/L of Pol3 hydrolysis probe (5′-HEX/CCT CGC TGG AAA GGA CCT TAC ACA/3′IABkFQ); 300 nmol/L each of Env2-F (5′-ACC CTC AAC CTC CCC TAC AAG T-3′) and Env2-R (5′-GTT AAG CGC CTG ATA GGC TC-3′); 100 nmol/L of Env2 hydrolysis probe (5′-TEX615/AGC CAC CCC CAG GAA CTG GAG ATA GA/3′BHQ_2); 300 nmol/L each of yCD2-F (5′-ATC ATC ATG TAC GGC ATC CCT AG-3′) and yCD2-R (5′-TGA ACT GCT TCA TCA GCT TCT TAC-3′); and 100 nmol/L of yCD2 hydrolysis probe (5′-FAM/TCA TCG TCA ACA ACC ACC ACC TCG T/3′BHQ_1) were used with AgPath-ID One-Step RT-PCR Reagent (Life Technologies; either the Pol2 or Pol3 primer/probe set was used as a component in the triplex reaction).

Thermal cycling conditions consisted of 46°C for 20 minutes, followed by 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 55°C for 45 seconds. Absolute quantification was determined from linear regression on a seven-log serial dilution RNA standard curve (2.55e2 to 2.55e8 copies/mL) purified from Toca 511 viral vector containing the corresponding targets that underwent qRT-PCR in parallel with the test articles. The Toca 511 standard was originally qualified by comparison with an in vitro–synthesized RNA standard quantified by absorbance at 260 nm.

Identification of Toca 511 integration sites

Preparation of sequencing libraries.

A total of 100–1,000 ng of genomic DNA (isolated as described above) was sheared using Covaris E220 to 300 bp peak size. Following cleanup with Ampure XP beads (Beckman Coulter), DNA was end repaired (End-It DNA End-Repair Kit, Epicentre), a 3′ overhang A was added (NEBNext dA-Tailing Module, New England Biolabs), and partially double stranded adaptors (Adaptor1 + Adaptor 2) were ligated with T4 ligase (Quick Ligation Kit, NEB). Two rounds of PCR were performed using NEBNext High-Fidelity 2X PCR Master Mix. The second round of PCR included primers with barcoded Illumina adaptors. Following clean-up with Ampure XP beads, libraries were quantified by qPCR and Qubit (Thermo Fisher Scientific) and pooled for paired-end sequencing on Miseq or Hiseq 2000.

Oligonucleotides (ordered from IDT):

Adaptor1: 5′-GTAATACGACTCACTATAGGCTTTCAGACGTGTGCTCTTCCGATCT(NNNNNNN)GCTCCGCTTAAGGGA CT-3′

Adaptor2: 5′-/5Phos/GTC CCT TAA GCG GAG/3AmMO/-3′

Adaptors 1 and 2 were mixed at equimolar concentrations, annealed, and used in ligation reactions at approximately 10× molar excess to sheared DNA.

Toca511_PCR1: 5′-CCTTGGGAGGGTCTCCTCTGAGT-3′

Linker_PCR1: 5′-GTAATACGACTCACTATAGGCT-3′

Toca511_PCR2:5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCTTGACCATGACTACCCGTCAGCGGGGGTC-3′

Linker_PCR2: 5′-CAAGCAGAAGACGGCATACGAGAT(6mer barcode) GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT-3′

Processing of sequencing data.

Paired fastq files were processed using Cutadapt 1.1 (30):

cutadapt -e 0.05 -g TGACCATGACTACCCGTCAGCGGGGGTCTTTCATTAGTCCCTTAAGCGGAGC –discard-trimmed -O 17 -o tmp1.1.fastq -p tmp1.2.fastq fastq1 fastq2

cutadapt -e 0.05 -g TGACCATGACTACCCGTCAGCGGGGGTCTTTCATTTGGGGG –discard-trimmed -O 17 -o tmp2.1.fastq -p tmp2.2.fastq tmp1.1.fastq tmp1.2.fastq

cutadapt -e 0.05 -g TGACCATGACTACCCGTCAGCGGGGGTC –discard-untrimmed -O 17 -o tmp3.1.fastq -p tmp3.2.fastq tmp2.1.fastq tmp2.2.fastq

cutadapt -e 0.00 -g ⁁TTTCA –no-indels –discard-untrimmed -o tmp4.1.fastq -p tmp4.2.fastq tmp3.1.fastq tmp3.2.fastq

cutadapt -e 0.05 -a AGTCCCTTAAGCGGAGC -O 10 -m 11 -o tmp5.1.fastq -p tmp5.2.fastq tmp4.1.fastq tmp4.2.fastq

cutadapt -e 0.05 -g GCTCCGCTTAAGGGACT –discard-untrimmed -O 10 -m 11 -o tmp6.2.fastq -p tmp6.1.fastq tmp5.2.fastq tmp5.1.fastq

cutadapt -e 0.05 -a TGAAAGACCCCCGCTGACGGGTAGTCATGGTCA -O 10 -m 11 -o tmp7.2.fastq -p tmp7.1.fastq tmp6.2.fastq tmp6.1.fastq

Trimmed and filtered fastq files were mapped to the human (hg19) and Toca 511 genomes using Bowtie2 (for human genome: -q -X 2000 –no-mixed –no-discordant –no-unal –score-min L,0,-0.4; ref. 31). Sorted BAM files were filtered to remove read pairs in which at least one of the reads contained at least two mismatches to the reference genome in the first ten bases using RSamtools (32). Filtered BAM files were converted to BED files (Bedtools; ref. 33) and read pairs with the same start and end positions (± 2 bp) were collapsed. We removed read pairs that shared the same integration site but that had a different fragmentation site in cases in which the fragmentation read site accounted for less than 1% of total integration events at this site. We did this because we found that most (or all) cases were likely due to spurious PCR products created from PCR duplicates of a single integration event. Integration sites that contained sequence motifs enriched in sequences adjacent to integration sites from negative controls, identified using HOMER2, were removed (34). These motifs contained matches to the 3′ end of Toca511 PCR_2 primer and/or the Toca 511 integration footprint (TTTCA). Finally, we filtered out integration events in which another sample had the same integration site and fragmentation site in at least 100× greater abundance, which was likely due to index primer misidentification during sequencing. The locations of integration sites can be found in Supplementary Dataset S2.

Other integration analyses.

We used Bedtools to identify EMSEMBL transcripts adjacent to integration sites. Gene functional enrichment analyses were performed with Metascape (35). Cancer gene lists were downloaded from Supplementary Material associated with Vogelstein and colleagues (36) and from Bushman lab website: http://www.bushmanlab.org/links/genelists. The hypergeometric density distribution was used to test for the significance of overlap between gene lists.

Toca 511 sequencing

Full-length PCR for sequencing.

PCR for full-length amplification of Toca 511 was performed as primary reactions followed by nested reactions using LongRange PCR Kit (Qiagen). Primary PCRs (9197 bp) were executed using primers targeting the flanking long terminal repeat regions of Toca 511 at 400 nmol/L each: Tri Set 1-forward (5′- GACTTGTGGTCTCGCTGTTCCTT-3′) and Tri Set 3-reverse (5′- GAGTGAGGGGTTGTGGGCTCT-3′). Primers for nested PCRs (8349 bp) were designed to anneal internally to the primary amplicons and were used at 400 nmol/L each: 5′ long PCR sequencing primer-forward (5-TGGTAGGAGACGAGAACCTAAA-3′) and 3′ UCLA 3-37 IRES-reverse (5′-CCCCTTTTTCTGGAGACTAAATAA-3′). Genomic DNA (up to 400 ng) for the primary PCR or primary PCR reaction (1 μL) for the nested PCR, long mix buffer (containing 25 mmol/L MgCl2), dNTPs (500 μmol/L each), DMSO (2%), primers (400 nmol/L), and LongRange Enzyme Mix (1 U) were combined and PCRs initiated. Primary PCR: thermal cycling conditions consisted of one cycle of 93°C for 3 minutes, 20 cycles of 93°C for 15 seconds, 55°C for 30 seconds, 68°C for 9 minutes, followed by one cycle of 93°C for 15 seconds, 55°C for 30 seconds, 68°C with 20-second incremental increases from previous time, for an additional 15 cycles. For nested PCR, thermal cycling conditions consisted of one cycle of 93°C for 3 minutes, 35 cycles of 93°C for 15 seconds, 60°C for 30 seconds, and 68°C for 8 minutes.

Tripartite PCR for sequencing.

PCR for tripartite amplification of Toca 511 was performed with three sets of overlapping primers. Primers were utilized at 150 or 400 nmol/L each: upstream sequence (2356 bp), 713_LTR Set 2-forward (5′-CGGGGGTCTTTCATTTGGGG-3′) and Tri Set 1-Reverse (5′ACAGTCTGGTACATGGAGGAAAG-3′); middle sequence (4842 bp), 5′ 2F2569-forward (5′- GGACAGAGGATGAGCAGAAAGA-3′) and TriSet2-reverse (5′- GCGGTGGAATGATTGGTATAAGTG-3′); and downstream sequence (3797 bp), Set 14-forward (5′- AGC CTT CCC AAC CAA GAA AGA-3′) and Set 14-reverse (5′- AGC TAG CTT GCC AAA CCT ACA-3′). Reactions were prepared as 25-μL aliquots with LongRange PCR Kit as described above. Thermal cycling conditions consisted of one cycle of 93°C for 3 minutes, followed by 30 cycles of 93°C for 15 seconds, 60°C for 30 seconds, and 68°C for 4 minutes.

Preparation for sequencing.

Following the nested PCR, products were separated on 1% TAE ethidium bromide agarose gels and bands corresponding to the expected size were excised and purified according to the manufacturer's instructions (Gel PCR Purification Kit: Qiagen). Quantification of purified PCR products was performed using QuantiFluor dsDNA (Promega) on an Infinite M200 Plate Reader (Tecan).

Purified PCR products were prepared for sequencing using Nextera XT kit (Illumina) according to manufacturer's instructions. Pooled libraries were sequenced on Illumina MiSeq (2 × 75 base reads) at University of California San Diego IGM facility. Fastq files are available from NCBI SRA database (SRP149280).

Toca 511 sequence analyses

PCR duplicates were removed with PRINSEQ-lite 0.20.4 (37). Read pairs were quality trimmed with Cutadapt (-q 30,30 –minimum-length 30) and mapped to the Toca 511 genome using bwa mem 0.7.15-r1140 using default parameters (38). SNVs and indels were identified with Varscan 2.3 (P = 0.01; min-coverage 500; min-avg-qual 25; min-var-freq 0.01; ref. 39) from Samtools (40) generated mpileup files (-d 20000). The resulting VCF files were parsed, combined, and analyzed in R using custom scripts. Plots for figures were made with ggplot2.

yCD2 sequencing

Primers used to amplify the yCD2 coding region included Illumina Nextera XT universal primer sequences at their 5′ ends: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG CACGGGGACGTGGTTTTCCTT-3′ and 5′- GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGTACAGGTGGGGTCTTTCATTCC-3′. Primary PCR was performed with 400 nmol/L of each primer using Qiagen Long Range PCR kit components. PCR products were gel purified and a secondary PCR was performed with Nextera XT dual-indexed primers to prepare samples for sequencing. Samples were pooled and sequenced on Illumina MiSeq (2 × 300 base reads). Adaptors were removed from sequencing reads with Cutadapt. Reads were quality trimmed with Trimmomatic v0.32 (SLIDINGWINDOW:30:30; ref. 41) and then mapped to Toca 511 genome with bwa mem as described above. SNVs were identified with Varscan2 as described above. Fastq files are available from NCBI SRA database (SRP149351).

Patient autopsy samples

The whole-body autopsy patient was a 65-year-old male with a past medical history of left temporal anaplastic astrocytoma with transformation to glioblastoma (GBM) status postresection. The patient completed Toca 511 dosing and a single cycle of Toca FC. Main study informed consent was reviewed and received approval from Western Institutional Review Board, which allows Tocagen to request tissues from an autopsy case. Full autopsy consent was obtained by the investigators at UTHealth. Samples were collected approximately 40 days after the last dose of Toca FC. Tissue and biofluids samples were collected as follows: brain -adjacent, tumor – noninjected site A, tumor – noninjected site B, tumor – noninjected site C, tumor injection site A, tumor injection site B, tumor injection site C, 1 mL blood, 1 mL urine, skin, lower GI, testes, liver, spleen, lymph nodes, bone marrow, spinal cord, lung - left lower lobe, lung - Lobe Consolidation, CSF.

The brain autopsy was from an 80-year-old male with a past medical history of GBM and recurrent GBM postresection. Samples from the following locations were collected: contralateral temp lobe (amygdala), ipsilateral frontal cortex (sagittal), anterior half (putamen and internal capsule), posterior half, posterior tumor, posterior tumor (slightly more anterior), anterior tumor (presumed injection site), anterior tumor (presumed injection site, slightly more posterior).

Toca 511 is transiently detected in blood in a small subset of patients in Study 11

In Study 11, surgical resection was followed by multiple injections of Toca 511 into the walls of the resection cavity; 4–6 weeks later, Toca FC dosing started (Supplementary Table S1; ref. 9). To quantitatively measure Toca 511 DNA and RNA levels over a broad dynamic range, robust [reverse-transcription (RT)]-qPCR assays that utilized a six-log standard curve for absolute quantification were developed (see Materials and Methods). The estimated lower limit of quantification (LLOQ) for the qPCR assay is 25 viral genome copies per reaction, whereas the LLOQ for the RT-PCR assay is approximately 7,300 viral RNA copies per mL of plasma. As part of Study 11, we performed Toca 511 qPCR on DNA isolated from whole blood as well as qRT-PCR on total RNA isolated from plasma, urine, and saliva, sampled longitudinally starting just prior to Toca 511 treatment and continuing at intervals over the course of treatment, including > 500 DNA and RNA test samples across 56 patients (Fig. 1A – red lines; ref. 9). Samples that were within the quantitative range of the assay, as defined above, were considered detected.

Quantitative Toca 511 RNA signal was rarely detected in urine or saliva samples but was detected in 32 plasma samples collected from 22 of 56 patients tested (Supplementary Fig. S1; Supplementary Dataset S1) and rapidly cleared. Sixteen of the positive RNA samples from plasma were detected from samples collected one day after surgery (visit 2). This is probably too early for infection and viral replication to occur and likely represents leakage of the injected vector into the blood stream following surgery. Thirteen patients had positive RNA signal at times after visit 2, occurring between day ∼10 (visit 3) and week 6(visit 5).

We detected Toca 511 in DNA isolated from blood in 15 samples collected from 11 patients (Fig. 1B; Supplementary Dataset S1). Detection of Toca 511 DNA in whole blood did not correlate with patients’ Toca 511 dose level (Supplementary Dataset S1). In all cases, quantitative signal occurred between week 4 (visit 4) and week 10 (visit 7; Fig. 1B). Patients begin taking Toca FC at week 6 (visit 5) and it is likely this helped clear virus from the blood. Even in cases in which integrated Toca 511 was detected, overall levels were low; the maximum DNA signal was 3,400 copies per μg of genomic DNA (∼2 copies per 100 diploid genome equivalents; Fig. 1B and D). In summary, virus was well controlled as it was infrequently and only transiently detected in the patients’ blood with quantitative Toca 511 signal in 6% of plasma samples and 3% of whole blood samples.

Toca 511 infects patients’ HGG tumors

While it was not feasible to systematically measure Toca 511 levels in residual tumors from patients in Study 11, we obtained tumor samples from 8 patients in Study 11 for whom their tumor recurred following Toca 511 and Toca FC treatment and the tumor was subsequently resected (re tumor samples in the figures). We detected Toca 511 in DNA isolated from three of seven tumors tested. Two of these patients’ tumors had both quantifiable Toca 511 DNA and RNA signal (Fig. 1C; Supplementary Dataset S1), It is likely viral levels in these tumors are not representative of the initial uptake in the patient population as Toca FC treatment is expected to deplete Toca 511-expressing cells; moreover, one reason for tumor recurrence would be from a paucity of Toca 511-infected cancer cells.

In Study 13, recurrent HGG tumor was resected 8 to 14 days after the initial day of intravenous administration of Toca 511, additional Toca 511 was injected into the resected cavity followed by Toca FC treatment 4 to 6 weeks later (Fig. 1A; Supplementary Table S2). Appropriate qPCR was performed on DNA and RNA isolated from multiple tumor pieces (referred to, herein, as intravenous tumor samples; Fig. 1A). For 8 of the 17 patients given intravenous Toca 511, integrated Toca 511 was detected in at least one tumor piece (Fig. 1C), at levels up to 47,000 copies per μg of DNA (∼31 copies per 100 diploid genome equivalents; Fig. 1D; Supplementary Dataset S1). We detected Toca 511 RNA isolated from tumor samples in 9 patients, including 6 patients whose tumors also had quantifiable levels of integrated Toca 511 (Supplementary Dataset S1). The first two low-dose patients showed no quantifiable signal for DNA or RNA in their tumors. At the third, fourth, and fifth dose levels with 1-, 3-, or 5-day administration respectively, Toca 511 was detected in 11 of 15 tumors. These results suggest that Toca 511 gains access to the brain tumor after virus infusion into peripheral blood and can successfully deliver therapeutic transgenes.

Toca 511 integration patterns are tissue specific and consistent with previous analyses of gammaretroviral integration

Identification of integration sites in samples with few integration events per hundreds of genomes is challenging and we optimized an integration site enrichment procedure and sequencing analysis workflow to reliably do so (Supplementary Text, Supplementary Figs. S2–S4). We distinguished clonal events from PCR duplicates by virtue of having the same integration site, but different fragmentation site, as is standard (e.g., refs. 42–44). Nine blood samples from Study 11, and 10 tumor samples from study 13 with quantifiable Toca 511 signal were analyzed along with three negative controls (DNA isolated from human cells not exposed to Toca 511; Fig. 2A). The three negative controls yielded zero, three, and eight “integration events,” respectively (from millions of sequencing reads); these events were likely created from spurious priming by the Toca 511-specific nested primer in locations of the human genome that happened to be immediately followed by the four bp Toca 511 integration sequence. Patient samples generally had orders of magnitude more read pairs that mapped to the human genome than the negative control samples, yielding on average 34 integration events per blood sample and 165 integration events per tumor sample for a total of 1,984 measured integration events across all tumor samples tested (Fig. 2A; Supplementary Fig. S5).

Figure 2.

Toca 511 integration patterns are tissue specific and consistent with previous analyses of gammaretroviral integration. A, Boxplot showing the number of identified integration events obtained from negative controls (gray), blood samples (red), and intravenous tumor samples (blue). B, Distribution of Toca 511 integration sites in blood samples (red) and tumor samples (blue) around the nearest annotated transcription start site (TSS). Preferential Toca 511 integration near TSSs is consistent with established integration preferences of gRVs. C, Genes that harbor Toca 511 integration sites immediately upstream (within 10 kb) or within the gene encode proteins whose expression is linked to the site from which samples were taken. (top-red) Integration sites in blood tend to occur within or near genes encoding proteins linked to blood-cell functions. (bottom-blue) In contrast, brain tumor integration sites are enriched for genes linked to neuronal stem cell proliferation and differentiation and neuronal functions. The significance of enrichment of integration sites near genes annotated in each category is represented as a bar plot of the P value, calculated using the hypergeometric density distribution function. GO, Gene Ontology term; PID, Pathway Interaction Database.

Figure 2.

Toca 511 integration patterns are tissue specific and consistent with previous analyses of gammaretroviral integration. A, Boxplot showing the number of identified integration events obtained from negative controls (gray), blood samples (red), and intravenous tumor samples (blue). B, Distribution of Toca 511 integration sites in blood samples (red) and tumor samples (blue) around the nearest annotated transcription start site (TSS). Preferential Toca 511 integration near TSSs is consistent with established integration preferences of gRVs. C, Genes that harbor Toca 511 integration sites immediately upstream (within 10 kb) or within the gene encode proteins whose expression is linked to the site from which samples were taken. (top-red) Integration sites in blood tend to occur within or near genes encoding proteins linked to blood-cell functions. (bottom-blue) In contrast, brain tumor integration sites are enriched for genes linked to neuronal stem cell proliferation and differentiation and neuronal functions. The significance of enrichment of integration sites near genes annotated in each category is represented as a bar plot of the P value, calculated using the hypergeometric density distribution function. GO, Gene Ontology term; PID, Pathway Interaction Database.

Close modal

Given gRVs’ preference to integrate near the 5′ends of transcriptionally active units, as also observed for Toca 511 in cell culture (Supplementary Fig. S3D), we predicted there would be enrichment of Toca 511 integration sites near annotated mRNA transcription start sites (TSS) from patient samples. For these analyses we combined integration sites from blood and integration sites from tumor into two pools of data. We plotted the distribution of integration sites adjacent to annotated TSSs and observed a characteristic enrichment centered around TSSs in both blood and tumor samples, albeit with a shallower peak than observed in cell culture (Fig. 2B). For both blood and tumor, 14% of integrations occurred within 5 kb of the nearest annotated TSS (vs. 5% expected on the basis of random chance; P = 0.03 and P < 1e−5 for blood and tumor, respectively based on proportions test). We asked whether genes proximal to integration sites [immediately upstream (within 10 kb) or within the gene] encode proteins that are functionally linked to the site from which samples were taken; blood or brain tumor. Utilizing the hundreds of GO and REACTOME gene sets (45, 46), we found that brain tumor–derived integration sites preferentially occurred near genes involved in neuronal functions and growth/differentiation, while blood-derived integration sites preferentially occurred near genes linked to blood functions (Fig. 2C). Thus, Toca 511 integration site preferences from patient tumor and blood transduced in vivo are congruent with previous work on gRVs showing preference for sites of active transcription ex vivo and in mice (25, 26).

Absence of compelling evidence for clonal expansion of Toca 511-integrated cells

The degree to which integration sites are unique versus subclonal expansion is key to assess the putative risk of insertion mutagenesis leading to hyperproliferation. An abundance of unique integration events would also indicate viral spread within the tumor versus expansion of one to a few tumor cells with integrated genomes. For each sample, we determined the number of sites with two or more integration events and plotted the results as a fraction of all integration events (Fig. 3A). For samples with at least 20 identified integration events, multiple events from the same site comprised fewer than 5% of the total, with one exception, from patient 11_33 visit 4a blood (not shown in Fig. 3A). Follow-up analyses suggest this potential clonal expansion, which occurred in an Alu element, was a technical artifact caused by recombination with other Alu elements during PCR (Supplementary Text; Supplementary Fig. S6).

Figure 3.

Absence of compelling evidence for clonal expansion of Toca 511-integrated cells. A, The number of total integration events for each sample are color-coded by the proportion of sites represented once (gray) versus the number of sites represented multiple times (black). Samples in red are from blood: for example, 11_05_v7 indicates Study 11 patient 05 visit 7 blood draw. Samples in blue are from brain tumors: for example, 13_03_p5 indicates Study 13 patient 03 tumor piece 5. B, Neither blood (red) nor tumor (blue) integration sites preferentially occur near genes linked to tumorigenesis. The significance of enrichment of integration sites near genes annotated in each category is represented as a bar plot of the P value, calculated using the hypergeometric density distribution function.

Figure 3.

Absence of compelling evidence for clonal expansion of Toca 511-integrated cells. A, The number of total integration events for each sample are color-coded by the proportion of sites represented once (gray) versus the number of sites represented multiple times (black). Samples in red are from blood: for example, 11_05_v7 indicates Study 11 patient 05 visit 7 blood draw. Samples in blue are from brain tumors: for example, 13_03_p5 indicates Study 13 patient 03 tumor piece 5. B, Neither blood (red) nor tumor (blue) integration sites preferentially occur near genes linked to tumorigenesis. The significance of enrichment of integration sites near genes annotated in each category is represented as a bar plot of the P value, calculated using the hypergeometric density distribution function.

Close modal

Given the precedent for clonal expansion due to integration of gRVs adjacent to oncogenes, particularly LMO2, we determined whether there was a preponderance of sites adjacent to or within various classes of cancer-related genes, in both blood and tumor. We found some integration events within 10 kb of oncogenic drivers or tumor suppressor genes in blood or tumor (Fig. 3B). There was one integration site in blood (11_05 visit 7) near a lymphoma-linked gene, NKAIN2/TCBA1, which encodes a transmembrane protein that interacts with the beta subunit of a sodium/potassium-transporting ATPase (47). There was no observed clonal expansion from this integration event and like all other patients, detectable Toca 511 signal was undetectable after Toca FC treatment. These results are consistent with the lack of development of lymphoma-like clinical features in surviving patients after Toca 511 delivery to date and the paucity of Toca 511-infected cells in all patients’ blood samples (9).

Toca 511 genomes are mutated by restriction factors in patients

We previously reported Toca 511 gross genome stability in vitro across multiple passages of the virus in cell culture (13). In the clinical setting, Toca 511 was PCR-amplified from patient DNA samples using a single nested PCR that spanned 9 kb of the genome including all coding regions, or three overlapping PCR products that spanned the same region (Supplementary Fig. S7). PCR products were gel purified and those with sufficient material were prepared for Illumina sequencing. We obtained quality sequencing results (> 1,000× coverage across at least 6,000 bp) from three blood samples, 6 samples from re-resected tumors, 6 tumor samples following intravenous treatment, and 4 nonclinical samples, including two cell lines (HT-1080 and U-87MG) infected with Toca 511 and two plasmids containing the parental Toca 511 genome.

Following quality filtering and read mapping to the Toca 511 reference genome, we characterized the mutation profiles of the Toca 511 genomes from each sample using Varscan2 (39), which identifies both single nucleotide variants (SNV) and short insertions and deletions (indels; Fig. 4; Supplementary Fig. S8). At a mutation frequency threshold of 3%, there were four SNVs and zero indels in the two plasmid controls and U-87MG–infected cells. These mutations were previously identified shared silent point mutations from the initial cloned MLV genome from which Toca 511 was derived (48) and were removed for subsequent analyses. There were two additional SNVs in the HT-1080 control sample, both occurring at less than 5% frequency. In contrast to the plasmid and cell culture samples, there was a wide spectrum of mutations among patient samples. Total SNVs per sample ranged from 31 (11_33 blood visit 7) to 742 (11_02 re-tumor section 8; mean = 204; Fig. 4A and B). Indels were much less frequent than SNVs (Supplementary Fig. S8). There were generally more SNVs and indels in Toca 511 genomes from re-resected tumors versus tumors resected prior to Toca FC treatment and blood, which could reflect more rounds of infection and replication and/or enrichment of nonfunctional Toca 511 following Toca FC treatment (patient 11_02 took one cycle of Toca FC while patient 11_31 took three cycles of Toca FC).

Figure 4.

Toca 511 genomes are mutated by restriction factors in patients. A, Boxplot showing the total number of single nucleotide variants (SNV) from different classes of samples at a mutation frequency threshold of 3%: Toca 511-containing plasmid and transduced cell lines (black, n = 4), blood (red, n = 3), brain tumors re-resected after further progression of disease after Toca 511 delivery (magenta, n = 4), brain tumors following intravenous dosing prior to Toca FC treatment (blue, n = 6). B, Mutations occur at a wide range of frequencies in patient samples. Barplots show the fraction of mutations across all patients from whom data could be obtained, with frequency 3%–10% (light green), 10%–30% (green) and greater than 30% (dark green). C, Boxplot showing the percent totals of all possible point mutations from different classes of samples. For instance, A>C indicates A to C transitions. Nearly all mutations are transitions. D, Barplot showing the percentage frequency of the four possible transitions in each patient sample. G to A mutations predominate for all samples.

Figure 4.

Toca 511 genomes are mutated by restriction factors in patients. A, Boxplot showing the total number of single nucleotide variants (SNV) from different classes of samples at a mutation frequency threshold of 3%: Toca 511-containing plasmid and transduced cell lines (black, n = 4), blood (red, n = 3), brain tumors re-resected after further progression of disease after Toca 511 delivery (magenta, n = 4), brain tumors following intravenous dosing prior to Toca FC treatment (blue, n = 6). B, Mutations occur at a wide range of frequencies in patient samples. Barplots show the fraction of mutations across all patients from whom data could be obtained, with frequency 3%–10% (light green), 10%–30% (green) and greater than 30% (dark green). C, Boxplot showing the percent totals of all possible point mutations from different classes of samples. For instance, A>C indicates A to C transitions. Nearly all mutations are transitions. D, Barplot showing the percentage frequency of the four possible transitions in each patient sample. G to A mutations predominate for all samples.

Close modal

Next we asked whether there was a bias in the mutation patterns among SNVs, which may suggest underlying mechanisms of mutation. For each sample, we calculated the fraction of all SNVs corresponding to each of the twelve possible pairwise combinations. As shown in Fig. 4C, an overwhelming majority of SNVs corresponded to three of four possible transitions: G to A, A to G, and T to C. For 11 of 13 samples, the majority of SNVs were G to A transitions, which is commonly seen in retroviral mutation profiles due to APOBEC cytidine deaminase-mediated C to U transitions during reverse transcription (Fig. 4D; refs. 28, 29). G to A transitions were more likely to occur at higher frequency than other transitions in nine samples (Dunn test, P < 0.001). These results suggest that APOBEC-mediated cytidine deamination during reverse transcription contributes to Toca 511 mutation spectrum in both patient blood and tumors, but that its influence varies among samples.

G to A mutations can create premature stop codons in Toca 511

The abundance of SNVs in integrated Toca 511 genomes from patient samples raises the question as to what degree these mutations inactivate the vector. We focused our attention on nonsynonymous mutations caused by G to A transitions due their abundance. What stood out was the potential for the codon encoding tryptophan (TGG) to be converted to stop codons (STOP - TGA, TAG, and TAA) by any combination of G to A at the second or third position. Toca 511 contains 52 tryptophan codons: 12 in gag, 24 in pol, 14 in env, and 2 in yCD2. We calculated the frequency of G to A mutations in tryptophan codons and plotted the results for each sample as a heatmap (Fig. 5A). There were a number of sites in the genome with high frequency tryptophan to STOP that occurred in multiple samples. There were 10 samples with at least one tryptophan to STOP at 30% frequency or greater (Fig. 5B). Coinciding with increased mutation accumulation and/or selection for nonfunctional Toca 511 genomes, tryptophan to STOP was more abundant in samples from tumors re-resected following repeated cycles of Toca FC treatment. The most common tryptophan to STOP mutation was in amino acid 10 (W10) in yCD2, which was mutated in 8 samples at > 30% frequency. This mutation, occurring near the 5′ end of yCD2 coding sequence presumably eliminates functional yCD2 protein unless an alternative start codon rescues function; for instance, the closest downstream methionine is at amino acid 15. W10 is immediately adjacent to the first alpha helix and is conserved in orthologs from closely related fungal species (Supplementary Fig. S9; ref. 49). The next two most frequent mutations affected W804 and W1016 in pol. W804 to STOP occurred in three re-resected tumors and five tumors resected prior to Toca FC treatment, while W1016 to STOP occurred in one blood, three re-resected tumors and two tumors resected prior to Toca FC treatment.

Figure 5.

G to A mutations can create premature stop codons in Toca 511. A, The tryptophan (W) codon is converted to stop codons (STOP) by G to A transitions. Each row corresponds to the Toca 511 mutation spectrum from a single DNA sample. The color bar on the left side indicates each sample's tissue of origin (colors as in Fig. 4A). The heatmap shows the percentage frequency of conversion of W to STOP along the Toca 511 genome. High-frequency inactivating mutations were largely restricted to tumor re-resected (magenta) after multiple Toca FC treatment cycles. B, The number of W to STOP changes (on an amino acid basis) across Toca 511 as a function of their frequency in the sample. C, Boxplot representation of the estimated percentage of “functional” yCD2 based on targeted yCD2 sequencing. “Functional” is defined as 100 minus the maximum mutation frequency of the three observed G to A mutations that are expected to inactivate yCD2 (Supplementary Fig. S10).

Figure 5.

G to A mutations can create premature stop codons in Toca 511. A, The tryptophan (W) codon is converted to stop codons (STOP) by G to A transitions. Each row corresponds to the Toca 511 mutation spectrum from a single DNA sample. The color bar on the left side indicates each sample's tissue of origin (colors as in Fig. 4A). The heatmap shows the percentage frequency of conversion of W to STOP along the Toca 511 genome. High-frequency inactivating mutations were largely restricted to tumor re-resected (magenta) after multiple Toca FC treatment cycles. B, The number of W to STOP changes (on an amino acid basis) across Toca 511 as a function of their frequency in the sample. C, Boxplot representation of the estimated percentage of “functional” yCD2 based on targeted yCD2 sequencing. “Functional” is defined as 100 minus the maximum mutation frequency of the three observed G to A mutations that are expected to inactivate yCD2 (Supplementary Fig. S10).

Close modal

To corroborate these results we performed targeted sequencing on the yCD2 coding region (Supplementary Text; Supplementary Fig. S10). According to the targeted yCD2 sequencing results at least 30% of yCD2-coding sequences were functional in 10 of 16 samples isolated from tumors following intravenous treatment, 1 of 5 samples isolated from blood (3 of 5 were from one patient) and 1 of 3 samples from re-resected tumors (Fig. 5C). While technical replicates for targeted sequencing of yCD2 were highly concordant (Supplementary Fig. S11), the concordance between Toca 511 sequencing results and targeted yCD2 sequencing results varied (Supplementary Text; Supplementary Fig. S10). Thus, we must be cautious about quantitative interpretation of specific mutation frequencies, which could be skewed by a small number of viral copies going into PCR reactions prior to sequencing preparation (50). For instance, even in tumors resected after Toca 511 delivery and multiple rounds of Toca FC, where deleterious mutations are observed at high frequency, we still detect yCD2 expression, suggesting a reservoir of functional virus persists (Supplementary Fig. S12; ref. 9).

Toca 511 preferentially targets tumors in a recurrent GBM patient

While we have characterized the tissue and biofluid distribution of Toca 511 in mice (12), analogous analyses have not been reported in humans. Twenty-one tissue samples throughout the body were obtained after the death of a male patient with recurrent GBM who was treated with Toca 511 via injection into the newly resected tumor cavity, as in Study 11. The patient died approximately 3 months after Toca 511 injection and approximately 6 weeks after a single cycle of Toca FC as the patient declined further treatment due to poor quality of life with global aphasia. We obtained three pieces of recurrent tumor at the Toca 511–injection site, three from a separate noninjected GBM tumor, two from nonneoplastic brain, one from spinal cord, one from cerebral spinal fluid (CSF) as well as from sites that are considered potential repositories for gRVs or important organs to test, including spleen, lymph node, bone marrow, lung, liver, lower GI, testes, and skin as well as whole blood and urine.

We measured Toca 511 levels in DNA and total RNA isolated from each autopsy tissue sample. Toca 511 was detected in RNA isolated from two of three samples from the injected tumor site and from one of three samples from neighboring noninjected tumor, but not elsewhere (Fig. 6A, right). Toca 511 was detected in DNA isolated from all three injected-site tumor pieces, two of three noninjected tumor pieces, immediately adjacent nonneoplastic brain region, spinal cord, and cerebrospinal fluid (CSF, Fig. 6A, left). Outside of the central nervous system, we found observable but not quantifiable (< 100 copies/μg, < 1 copy/1,500 diploid cell equivalents) Toca 511 DNA in blood, liver, spleen, lymph node, lung, and bone marrow. No Toca 511 signal was observed in DNA isolated from testes, urine, skin, or lower GI. No Toca 511 RNA was detected outside of the tumor samples, reinforcing the low probability of live virus shedding by patients. Toca 511 tumor specificity is corroborated by whole brain autopsy samples isolated from a patient in Study 8 who underwent one cycle of Toca FC (Supplementary Table S3).

Figure 6.

Toca 511 preferentially targets tumors in a GBM patient. A, (left) Average Toca 511 signal (copies/μg) from qPCR replicates with quantifiable signal. *, indicates samples for which signal was below the LLOQ – the bars are arbitrarily set at 50% of the LLOQ; (right) Average Toca 511 RNA signal (copies/μg) from qRT-PCR replicates with quantifiable signal. Note the break in the x-axis. B, Barplot showing the fraction of SNVs with percentage frequency 3%–10% (light green), 10%–30% (green), and greater than 30% (dark green). The source of DNA is indicated below the plots. C, Barplot showing mutation frequency for the four possible transitions. D, Heatmap representation of percentage frequency of SNVs across the 3′ end of Toca 511 genome from the four patient samples. Only positions in Toca 511 in which at least one sample had a SNV were included. Samples were clustered by Euclidean distance.

Figure 6.

Toca 511 preferentially targets tumors in a GBM patient. A, (left) Average Toca 511 signal (copies/μg) from qPCR replicates with quantifiable signal. *, indicates samples for which signal was below the LLOQ – the bars are arbitrarily set at 50% of the LLOQ; (right) Average Toca 511 RNA signal (copies/μg) from qRT-PCR replicates with quantifiable signal. Note the break in the x-axis. B, Barplot showing the fraction of SNVs with percentage frequency 3%–10% (light green), 10%–30% (green), and greater than 30% (dark green). The source of DNA is indicated below the plots. C, Barplot showing mutation frequency for the four possible transitions. D, Heatmap representation of percentage frequency of SNVs across the 3′ end of Toca 511 genome from the four patient samples. Only positions in Toca 511 in which at least one sample had a SNV were included. Samples were clustered by Euclidean distance.

Close modal

We amplified and sequenced integrated Toca 511 from DNA isolated from the injection site tumor, the noninjected tumor, CSF, and blood. For all four samples, we obtained quality results from the 3′ PCR product covering 3,750 bp including yCD2 (Supplementary Fig. S7). There were more than twice as many SNVs from the two tumor samples and CSF relative to blood (Fig. 6B). Most mutations in tumor and CSF were G to A (Fig. 6C). Nonetheless the specific mutation pattern in each sample was unique (Fig. 6D). Thus while the general mutation biases were similar among the tumor and CSF samples, the locations and frequencies of specific mutations varied considerably.

The logic behind gammaretroviral replicating vector gene therapy for oncology

Gammaretroviral replicating vector gene therapy holds promise for a range of oncologic therapeutic indications due its ability to selectively infect cancer cells without direct cell lysis and deliver a therapeutic transgene. In animal models, Toca 511–infected tumor cells are killed by 5-FU converted from 5- FC. The diffusible 5-FU also kills susceptible neighboring cells, including immune suppressor myeloid cells that contribute to the immune-suppressed tumor microenvironment (51, 52). After several cycles of Toca FC, treated immunocompetent animals that clear tumor are resistant to tumor rechallenge (12) and this resistance is T-cell mediated (51, 52).

The potential advantages of a RRV, Toca 511, over RNV center around the premise that there will be multiple rounds of infection and spread in the tumor over time, leading to a greater proportion of tumor cells and geographical regions infected, including noninjected tumors. The replication of retroviruses, including Toca 511, includes stable integration of viral genomes into infected cancer cell genomes creating a reservoir of Toca 511. This reservoir persists during Toca FC treatment in part because noncycling cancer cells are more resistant to 5-FU killing than replicating cells (53). The reservoir provides a source of Toca 511 production and spread to newly formed cancer cells between Toca FC doses. This allows for continued killing of tumor cells over multiple cycles of Toca FC and subsequent breaking of immune tolerance and reactivation of the immune system against the tumor. This is seen in preclinical models where several cycles of Toca FC are required to produce durable response in immunocompetent animals (12).

Toca 511 selectively replicates and spreads in patient tumors

Several lines of circumstantial evidence suggest Toca 511 does indeed undergo multiple rounds of infection in patient samples. The strongest evidence comes from mutational profiles in which the frequencies of specific mutations vary over 30-fold (Fig. 4). Samples from tumors isolated after Toca 511 and Toca FC treatment categorically displayed more mutations than samples from tumors taken before Toca FC treatment, likely due to depletion of functional Toca 511 genomes from yCD2-dependent 5-FU–mediated cell death. Detection of integrated Toca 511 in blood samples emerges weeks after dosing suggesting ongoing viral replication and spread of productive virus over time (Fig. 1). Toca 511 was quantitatively detected in both the tumor from the injection site and in a separate noninjected tumor in the brain (Fig. 6).

Multiple lines of evidence presented herein support Toca 511′s selectivity for tumor cells over normal cells and tissues in humans. Integrated Toca 511 was detected transiently in blood but only in 11 of 56 patients treated in Study 11 (Fig. 1). Toca 511 RNA was only detected in 2 of the 15 blood samples with detectable levels of integrated Toca 511 DNA (Supplementary Dataset S1). While it was not feasible to measure Toca 511 levels in tumor cells in the resection study prior to Toca FC treatment, we did measure Toca 511 levels in the context of intravenous delivery as well as in tumors that recurred post Toca FC treatment. In these contexts, Toca 511 DNA and Toca 511 RNA were detected in >40% of tumors, at levels generally higher than seen in blood (Fig. 1), arguing for active virus production in the tumor (Supplementary Dataset S1). These results are extended in a more expansive analysis of nontumor samples from a patient's whole body autopsy, where we detected integrated virus and viral RNA in the injected tumor site as well as in a noninjected brain tumor, but we only detected nonquantifiable levels of virus DNA in a subset of bodily sites outside of the central nervous system (Fig. 6). All patients cleared detectable virus signal in blood within the first couple cycles of Toca FC suggesting that human patients are generally able to control the virus systemically, even if the yCD2 gene is inactivated (Figs. 4 and 5). We hypothesize blood clearance is due to a combination of yCD2-dependent apoptosis following 5-FC administration, Toca 511 independent cell turnover and multiple innate and adaptive defense mechanisms that act naturally to clear MLV, which is not zoonotic.

APOBEC-mediated cytidine deamination is likely the dominant source of Toca 511 mutations in patients

Most mutations in Toca 511 genomes isolated from patient samples were G to A transitions (Fig. 4). The human genome encodes a repertoire of APOBEC cytidine deaminases, some of which are incorporated into retroviral particles via interactions with the core viral protein GAG and viral RNA and catalyze C to U transitions on single-stranded DNA during reverse transcription in the cytoplasm (54), leading to G to A mutations in the viral coding strand. The most parsimonious explanation for the mutation profiles is that Toca 511 particles encapsulate one or more APOBEC molecules, which then processively mutate C's to U's during reverse transcription in the subsequent rounds of infection, leading to mutated and often inactivated integrated virus (Fig. 4). Among G to A mutations, those resulting in conversion of the tryptophan codon (TGG) to premature stop codons (TGA, TAG, and TAA) are expected to be particularly deleterious to viral functions. The complement of the tryptophan codon is the preferred sequence context for APOBEC3G, suggesting this paralog as a likely culprit. Generally though, the sources and identities of APOBECs incorporated into Toca 511 particles in patients are not known. While in some samples we estimate >90% frequency of specific G to A stop mutations, such as mutations leading to conversion of tryptophan at position 10 in yCD2 (Figs. 4 and 5C; Supplementary Fig. S10), we are cautious about quantitative interpretation of the mutation frequencies given the observed intrasample discordance (Supplementary Text and Supplementary Fig. S10). Indeed, we detected yCD2 protein via IHC in some tumor samples with high frequency of conversion of tryptophan at position 10 to stop codon (Supplementary Fig. S12), suggesting APOBECs inactivate some, but not all, Toca 511 genomes. However, we are unable to make quantitative connections between observed yCD2 expression by IHC from one piece of tumor to Toca 511 DNA and RNA levels in another piece of tumor due to intratumor heterogeneity.

As Toca 511 is derived from a mouse gammaretrovirus, it did not specifically evolve mechanisms to minimize inactivation by human APOBECs (or other human restriction factors). In principle, future generations of gRVs could be engineered to minimize tryptophan residues within the transgene, which could lead to enhanced efficacy in tumors while maintaining the safety profile.

No evidence for pathology or molecular abnormalities from Toca 511 insertions

A primary safety concern using RRVs has been that infection of blood progenitor cells leading to oncogenic transformation could occur in some clinical situations as has been observed with RNVs. Toca 511 integration profiles from blood and tumor samples described herein, after administration both into the tumor resection cavity and intravenously, show no compelling evidence for clonal expansion of infected cells within the time frame of these trials (Study 11 started in 2011; Fig. 3). These results are consistent with the absence of direct clinical evidence for this kind of adverse event so far in patients with recurrent HGG treated with Toca 511 and Toca FC in a resection setting (9). However, given the precedence for subsequent malignancies in patients with HGG and temozolomide-related hematologic adverse events (55, 56), it is likely that we will eventually encounter patients treated with Toca 511 and Toca FC that develop hematologic adverse events, including lymphoma. The assays presented herein would enable us to gauge the contribution, if any, of Toca 511 insertional mutagenesis.

Gammaretrovirus integration is stimulated by physical interactions between integrase with BET proteins which orchestrate assembly of transcription initiation complexes, leading to so-called pseudo-random insertion preferentially near transcription start sites, including active enhancers and promoters (57). Therefore, many gRV integration sites are within gene-regulatory regions, which in turn could influence the transcriptional regulation of linked genes, as seen for LMO2 in HSCs in some treated X-SCID patients, resulting in the delayed onset of leukemia in some of these severely immune compromised patients (58). Our analyses of Toca 511 integration sites in blood and brain tumor suggest that while Toca 511 follows pseudo-random preferences of other gRVs, there was no compelling evidence for clonal expansion as judged by overrepresentation of specific integration sites, nor was there preferential insertion near oncogenes (Fig. 3). We did find that insertion sites in blood and brain tumor occurred preferentially near genes that function in their respective tissue type (Fig. 2).

Conclusions

Toca 511 and Toca FC treatment represents a general novel anticancer modality (9). This study fills in crucial gaps in our understanding of the therapeutic use of Toca 511 and replicating retroviral vectors in general. The data provided herein provide molecular rationales for the previously reported lack of Toca 511–related excess tumorigenicity observed in patients treated with Toca 511 and Toca FC (9) as well as the continued investigation of replicating retrovirus-based immunotherapies. A randomized phase III trial (NCT02414165) in patients with recurrent GBM and anaplastic astrocytoma and a phase Ib trial investigating treatment in solid tumors (NCT02576665) are ongoing.

J. Zhu reports receiving commercial research support from Tocagen Inc. D.J. Hogan, O.R. Diago, D. Gammon, A. Haghighi, A. Das, H.E. Gruber, D.J. Jolly and D. Ostertag are employees of Tocagen Inc., and own stock. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D.J. Hogan, J.-J. Zhu, H.E. Gruber, D.J. Jolly, D. Ostertag

Development of methodology: D.J. Hogan, D.K. Gammon, D. Ostertag

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.J. Hogan, J.-J. Zhu, D.K. Gammon, A. Haghighi, G. Lu, A. Das, D. Ostertag

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.J. Hogan, A. Haghighi, A. Das, D. Ostertag

Writing, review, and/or revision of the manuscript: D.J. Hogan, J.-J. Zhu, G. Lu, A. Das, H.E. Gruber, D.J. Jolly, D. Ostertag

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.J. Hogan, O. Diago, D. Ostertag

Study supervision: D.J. Hogan, D. Ostertag

We would like to thank John Wood MBA RAC (Tocagen Inc.) and John M. Coffin Ph.D. (Tufts University) for critically reading the manuscript. We wish to thank the patients and their families for participating in Tocagen's Clinical Trials. We are also deeply grateful to those patients and families who consented to the autopsy procedures. The authors also thank the ABC2 Foundation (Washington, DC), the National Brain Tumor Society (Watertown, MA), the American Brain Tumor Association (Chicago, IL), the Musella Foundation (Hewlett, NY), and Voices Against Brain Cancer (New York, NY) for their support and collaborations.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Kaufman
HL
,
Kohlhapp
FJ
,
Zloza
A
. 
Oncolytic viruses: a new class of immunotherapy drugs
.
Nat Rev Drug Discovery
2015
;
14
:
642
62
.
2.
Kohlhapp
FJ
,
Kaufman
HL
. 
Molecular pathways: mechanism of action for talimogene laherparepvec, a new oncolytic virus immunotherapy
.
Clin Cancer Res
2016
;
22
:
1048
54
.
3.
Logg
CR
,
Robbins
JM
,
Jolly
DJ
,
Gruber
HE
,
Kasahara
N
. 
Retroviral replicating vectors in cancer
.
Methods Enzymol
2012
;
507
:
199
228
.
4.
Critchley-Thorne
RJ
,
Simons
DL
,
Yan
N
,
Miyahira
AK
,
Dirbas
FM
,
Johnson
DL
, et al
Impaired interferon signaling is a common immune defect in human cancer
.
PNAS
2009
;
106
:
9010
5
.
5.
Rabinovich
GA
,
Gabrilovich
D
,
Sotomayor
EM
. 
Immunosuppressive strategies that are mediated by tumor cells
.
Annu Rev Immunol
2007
;
25
:
267
96
.
6.
Zitvogel
L
,
Tesniere
A
,
Kroemer
G
. 
Cancer despite immunosurveillance: immunoselection and immunosubversion
.
Nat Rev Immunol
2006
;
6
:
715
27
.
7.
Roe
T
,
Reynolds
TC
,
Yu
G
,
Brown
PO
. 
Integration of murine leukemia virus DNA depends on mitosis
.
EMBO J
1993
;
12
:
2099
108
.
8.
Lin
AH
,
Burrascano
C
,
Pettersson
PL
,
Ibanez
CE
,
Gruber
HE
,
Jolly
DJ
. 
Blockade of type I interferon (IFN) production by retroviral replicating vectors and reduced tumor cell responses to IFN likely contribute to tumor selectivity
.
J Virol
2014
;
88
:
10066
77
.
9.
Cloughesy
TF
,
Landolfi
J
,
Hogan
DJ
,
Bloomfield
S
,
Carter
B
,
Chen
CC
, et al
Phase 1 trial of vocimagene amiretrorepvec and 5-fluorocytosine for recurrent high-grade glioma
.
Sci Transl Med
2016
;
8
:
341ra75
.
10.
Kim
S
,
Park
EJ
,
Yu
SS
,
Kim
S
. 
Development of enzyme-linked immunosorbent assay for detecting antibodies to replication-competent murine leukemia virus
.
J Virol Methods
2004
;
118
:
1
7
.
11.
Martineau
D
,
Klump
WM
,
McCormack
JE
,
DePolo
NJ
,
Kamantigue
E
,
Petrowski
M
, et al
Evaluation of PCR and ELISA assays for screening clinical trial subjects for replication-competent retrovirus
.
Hum Gene Ther
1997
;
8
:
1231
41
.
12.
Ostertag
D
,
Amundson
KK
,
Lopez Espinoza
F
,
Martin
B
,
Buckley
T
,
Galvao da Silva
AP
, et al
Brain tumor eradication and prolonged survival from intratumoral conversion of 5-fluorocytosine to 5-fluorouracil using a nonlytic retroviral replicating vector
.
Neuro Oncol
2012
;
14
:
145
59
.
13.
Perez
OD
,
Logg
CR
,
Hiraoka
K
,
Diago
O
,
Burnett
R
,
Inagaki
A
, et al
Design and selection of Toca 511 for clinical use: modified retroviral
.
Mol Ther
2012
;
20
:
1689
98
.
14.
Twitty
CG
,
Diago
OR
,
Hogan
DJ
,
Burrascano
C
,
Ibanez
CE
,
Jolly
DJ
, et al
Retroviral replicating vectors deliver cytosine deaminase leading to targeted 5-Fluorouracil-mediated cytotoxicity in multiple human cancer types
.
Human Gene Ther Methods
2016
;
27
:
17
31
.
15.
Vanin
EF
,
Kaloss
M
,
Broscius
C
,
Nienhuis
AW
. 
Characterization of replication-competent retroviruses from nonhuman primates with virus-induced T-cell lymphomas and observations regarding the mechanism of oncogenesis
.
J Virol
1994
;
68
:
4241
50
.
16.
Purcell
DF
,
Broscius
CM
,
Vanin
EF
,
Buckler
CE
,
Nienhuis
AW
,
Martin
MA
. 
An array of murine leukemia virus-related elements is transmitted and expressed in a primate recipient of retroviral gene transfer
.
J Virol
1996
;
70
:
887
97
.
17.
Yi
Y
,
Hahm
SH
,
Lee
KH
. 
Retroviral gene therapy: safety issues and possible solutions
.
Curr Gene Ther
2005
;
5
:
25
35
.
18.
Cicalese
MP
,
Ferrua
F
,
Castagnaro
L
,
Pajno
R
,
Barzaghi
F
,
Giannelli
S
, et al
Update on the safety and efficacy of retroviral gene therapy for immunodeficiency due to adenosine deaminase deficiency
.
Blood
2016
;
128
:
45
54
.
19.
Baum
C
,
Dullmann
J
,
Li
Z
,
Fehse
B
,
Meyer
J
,
Williams
DA
, et al
Side effects of retroviral gene transfer into hematopoietic stem cells
.
Blood
2003
;
101
:
2099
114
.
20.
Yi
Y
,
Noh
MJ
,
Lee
KH
. 
Current advances in retroviral gene therapy
.
Curr Gene Ther
2011
;
11
:
218
28
.
21.
Muul
LM
,
Tuschong
LM
,
Soenen
SL
,
Jagadeesh
GJ
,
Ramsey
WJ
,
Long
Z
, et al
Persistence and expression of the adenosine deaminase gene for 12 years and immune reaction to gene transfer components: long-term results of the first clinical gene therapy trial
.
Blood
2003
;
101
:
2563
9
.
22.
Haviernik
P
,
Bunting
KD
. 
Safety concerns related to hematopoietic stem cell gene transfer using retroviral vectors
.
Curr Gene Ther
2004
;
4
:
263
76
.
23.
Deichmann
A
,
Hacein-Bey-Abina
S
,
Schmidt
M
,
Garrigue
A
,
Brugman
MH
,
Hu
J
, et al
Vector integration is nonrandom and clustered and influences the fate of lymphopoiesis in SCID-X1 gene therapy
.
J Clin Invest
2007
;
117
:
2225
32
.
24.
Scholler
J
,
Brady
TL
,
Binder-Scholl
G
,
Hwang
WT
,
Plesa
G
,
Hege
KM
, et al
Decade-long safety and function of retroviral-modified chimeric antigen receptor T cells
.
Sci Transl Med
2012
;
4
:
132ra53
.
25.
LaFave
MC
,
Varshney
GK
,
Gildea
DE
,
Wolfsberg
TG
,
Baxevanis
AD
,
Burgess
SM
. 
MLV integration site selection is driven by strong enhancers and active promoters
.
Nucleic Acids Res
2014
;
42
:
4257
69
.
26.
De Ravin
SS
,
Su
L
,
Theobald
N
,
Choi
U
,
Macpherson
JL
,
Poidinger
M
, et al
Enhancers are major targets for murine leukemia virus vector integration
.
J Virol
2014
;
88
:
4504
13
.
27.
Wolf
D
,
Goff
SP
. 
Host restriction factors blocking retroviral replication
.
Annu Rev Genet
2008
;
42
:
143
63
.
28.
Yu
Q
,
Konig
R
,
Pillai
S
,
Chiles
K
,
Kearney
M
,
Palmer
S
, et al
Single-strand specificity of APOBEC3G accounts for minus-strand deamination of the HIV genome
.
Nat Struct Mol Biol
2004
;
11
:
435
42
.
29.
Armitage
AE
,
Katzourakis
A
,
de Oliveira
T
,
Welch
JJ
,
Belshaw
R
,
Bishop
KN
, et al
Conserved footprints of APOBEC3G on Hypermutated human immunodeficiency virus type 1 and human endogenous retrovirus HERV-K(HML2) sequences
.
J Virol
2008
;
82
:
8743
61
.
30.
Martin
M
. 
Cutadapt removes adapter sequences from high-throughput sequencing reads
.
doi cutadapt removes adapter sequences from high-throughput sequencing reads
.
31.
Langmead
B
,
Salzberg
SL
. 
Fast gapped-read alignment with Bowtie 2
.
Nat Methods
2012
;
9
:
357
9
.
32.
Morgan M
PH
,
Obenchain
V
,
Hayden
N
. 
Rsamtools: binary alignment (BAM), FASTA, variant call (BCF), and tabix file import R package version 1261
. http://bioconductor.org/packages/release/bioc/html/Rsamtools.html.2016.
33.
Quinlan
AR
,
Hall
IM
. 
BEDTools: a flexible suite of utilities for comparing genomic features
.
Bioinformatics
2010
;
26
:
841
2
.
34.
Heinz
S
,
Benner
C
,
Spann
N
,
Bertolino
E
,
Lin
YC
,
Laslo
P
, et al
Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities
.
Mol Cell
2010
;
38
:
576
89
.
35.
Tripathi
S
,
Pohl
MO
,
Zhou
Y
,
Rodriguez-Frandsen
A
,
Wang
G
,
Stein
DA
, et al
Meta- and orthogonal integration of influenza "OMICs" data defines a role for UBR4 in virus budding
.
Cell Host Microbe
2015
;
18
:
723
35
.
36.
Vogelstein
B
,
Papadopoulos
N
,
Velculescu
VE
,
Zhou
S
,
Diaz
LA
 Jr
,
Kinzler
KW
. 
Cancer genome landscapes
.
Science
2013
;
339
:
1546
58
.
37.
Schmieder
R
,
Edwards
R
. 
Quality control and preprocessing of metagenomic datasets
.
Bioinformatics
2011
;
27
:
863
4
.
38.
Li
H
,
Durbin
R
. 
Fast and accurate short read alignment with Burrows-Wheeler transform
.
Bioinformatics
2009
;
25
:
1754
60
.
39.
Koboldt
DC
,
Zhang
Q
,
Larson
DE
,
Shen
D
,
McLellan
MD
,
Lin
L
, et al
VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing
.
Genome Res
2012
;
22
:
568
76
.
40.
Li
H
,
Handsaker
B
,
Wysoker
A
,
Fennell
T
,
Ruan
J
,
Homer
N
, et al
The sequence alignment/map format and SAMtools
.
Bioinformatics
2009
;
25
:
2078
9
.
41.
Bolger
AM
,
Lohse
M
,
Usadel
B
. 
Trimmomatic: a flexible trimmer for Illumina sequence data
.
Bioinformatics
2014
;
30
:
2114
20
.
42.
Hacein-Bey-Abina
S
,
Pai
SY
,
Gaspar
HB
,
Armant
M
,
Berry
CC
,
Blanche
S
, et al
A modified gamma-retrovirus vector for X-linked severe combined immunodeficiency
.
N Engl J Med
2014
;
371
:
1407
17
.
43.
Sherman
E
,
Nobles
C
,
Berry
CC
,
Six
E
,
Wu
Y
,
Dryga
A
, et al
INSPIIRED: a pipeline for quantitative analysis of sites of new DNA integration in cellular genomes
.
Mol Ther Methods Clin Develop
2017
;
4
:
39
49
.
44.
Gillet
NA
,
Melamed
A
,
Bangham
CR
. 
High-throughput mapping and clonal quantification of retroviral integration sites
.
Methods Mol Biol
2017
;
1582
:
127
41
.
45.
Fabregat
A
,
Sidiropoulos
K
,
Garapati
P
,
Gillespie
M
,
Hausmann
K
,
Haw
R
, et al
The Reactome pathway Knowledgebase
.
Nucleic Acids Res
2016
;
44
:
D481
7
.
46.
The Gene Ontology Consortium
. 
Expansion of the Gene Ontology knowledgebase and resources
.
Nucleic Acids Res
2017
;
45
:
D331
D8
.
47.
Tagawa
H
,
Miura
I
,
Suzuki
R
,
Suzuki
H
,
Hosokawa
Y
,
Seto
M
. 
Molecular cytogenetic analysis of the breakpoint region at 6q21–22 in T-cell lymphoma/leukemia cell lines
.
Genes Chromosome Cancer
2002
;
34
:
175
85
.
48.
Shoemaker
C
,
Hoffman
J
,
Goff
SP
,
Baltimore
D
. 
Intramolecular integration within Moloney murine leukemia virus DNA
.
J Virol
1981
;
40
:
164
72
.
49.
Stolworthy
TS
,
Korkegian
AM
,
Willmon
CL
,
Ardiani
A
,
Cundiff
J
,
Stoddard
BL
, et al
Yeast cytosine deaminase mutants with increased thermostability impart sensitivity to 5-fluorocytosine
.
J Mol Biol
2008
;
377
:
854
69
.
50.
Welkers
MR
,
Jonges
M
,
Jeeninga
RE
,
Koopmans
MP
,
de Jong
MD
. 
Improved detection of artifactual viral minority variants in high-throughput sequencing data
.
Front Microbiol
2014
;
5
:
804
.
51.
Mitchell
LA
,
Lopez Espinoza
F
,
Mendoza
D
,
Kato
Y
,
Inagaki
A
,
Hiraoka
K
, et al
Toca 511 gene transfer and treatment with the prodrug, 5-fluorocytosine, promotes durable antitumor immunity in a mouse glioma model
.
Neuro Oncol
2017
;
19
:
930
9
.
52.
Hiraoka
K
,
Inagaki
A
,
Kato
Y
,
Huang
TT
,
Mitchell
LA
,
Kamijima
S
, et al
Retroviral replicating vector-mediated gene therapy achieves long-term control of tumor recurrence and leads to durable anticancer immunity
.
Neuro Oncol
2017
;
19
:
918
29
.
53.
Richard
C
,
Duivenvoorden
W
,
Bourbeau
D
,
Massie
B
,
Roa
W
,
Yau
J
, et al
Sensitivity of 5-fluorouracil-resistant cancer cells to adenovirus suicide gene therapy
.
Cancer Gene Ther
2007
;
14
:
57
65
.
54.
Cullen
BR
. 
Role and mechanism of action of the APOBEC3 family of antiretroviral resistance factors
.
J Virol
2006
;
80
:
1067
76
.
55.
Villano
JL
,
Letarte
N
,
Yu
JM
,
Abdur
S
,
Bressler
LR
. 
Hematologic adverse events associated with temozolomide
.
Cancer Chemother Pharmacol
2012
;
69
:
107
13
.
56.
Li
X
,
Li
Y
,
Cao
Y
,
Li
P
,
Liang
B
,
Sun
J
, et al
Risk of subsequent cancer among pediatric, adult and elderly patients following a primary diagnosis of glioblastoma multiforme: a population-based study of the SEER database
.
Int J Neurosci
2017
;
127
:
1
7
.
57.
Sharma
A
,
Larue
RC
,
Plumb
MR
,
Malani
N
,
Male
F
,
Slaughter
A
, et al
BET proteins promote efficient murine leukemia virus integration at transcription start sites
.
Proc Natl Acad Sci U S A
2013
;
110
:
12036
41
.
58.
Hacein-Bey-Abina
S
,
Von Kalle
C
,
Schmidt
M
,
McCormack
MP
,
Wulffraat
N
,
Leboulch
P
, et al
LMO2-associated clonal T cell proliferation in two patients after gene therapy for SCID-X1
.
Science
2003
;
302
:
415
9
.